Latin America Beyond the Crisis – Impacts, Policies and

LATIN‐AMERICA BEYOND THE CRISIS IMPACTS, POLICIES AND OPPORTUNITIES

AMÉRICA LATINA MÁS ALLÁ DE LA CRISIS IMPACTOS, POLÍTICAS Y OPORTUNIDADES

Table of Contents

LATIN‐AMERICA BEYOND THE CRISIS—IMPACTS, POLICIES AND OPPORTUNITIES— A SYNTHESIS Marcelo M. Giugale ..............................................................................................................................1

AMÉRICA LATINA MÁS ALLÁ DE LA CRISIS—IMPACTOS, POLÍTICAS Y OPORTUNIDADES—UNA SÍNTESIS Marcelo M. Giugale ..............................................................................................................................7

PART I ‐ WORKING PAPERS 1. THE GLOBAL FINANCIAL AND ECONOMIC STORM: HOW BAD IS THE WEATHER IN LATIN AMERICA AND THE

CARIBBEAN? Augusto de la Torre.............................................................................................................................13

2. REGULATORY REFORM: INTEGRATING PARADIGMS Augusto de la Torre and Alain Ize .......................................................................................................23

3. HOW HAS POVERTY EVOLVED IN LATIN AMERICA AND HOW IS IT LIKELY TO BE AFFECTED BY THE ECONOMIC CRISIS? Joao Pedro Azevedo, Ezequiel Molina, John Newman, Eliana Rubiano and Jaime Saavedra.............55

4. LABOR MARKETS AND THE CRISIS IN LATIN AMERICA AND THE CARIBBEAN (A PRELIMINARY REVIEW FOR SELECTED

COUNTRIES) Samuel Freije‐Rodríguez and Edmundo Murrugarra ..........................................................................81

PART II ‐ TECHNICAL NOTES 5. HOW MUCH ROOM DOES LATIN AMERICA AND THE CARIBBEAN HAVE FOR IMPLEMENTING COUNTER‐CYCLICAL FISCAL

POLICIES? Cesar Calderón and Pablo Fajnzylber..................................................................................................97

6. CRISIS IN LAC: INFRASTRUCTURE INVESTMENT AND THE POTENTIAL FOR EMPLOYMENT GENERATION Laura Tuck, Jordan Schwartz and Luis Andres ..................................................................................108

7. HOW WILL LABOR MARKETS ADJUST TO THE CRISIS? A DYNAMIC VIEW William Maloney ............................................................................................................................... 115

8. WHAT IS THE LIKELY IMPACT OF THE 2009 CRISIS ON REMITTANCES AND POVERTY IN LATIN AMERICA AND THE

CARIBBEAN? Gabriel Demombynes, Hector Valdés Conroy, Ezequiel Molina and Amparo Ballivián ....................123

9. WILL FDI BE RESILIENT IN THIS CRISIS? Cesar Calderon and Tatiana Didier ...................................................................................................129

10. PATTERNS OF FINANCING DURING PERIODS OF HIGH RISK AVERSION: HOW HAVE LATIN FIRMS FARED IN THIS CRISIS SO FAR? Tatiana Didier ...................................................................................................................................135

Additional References: World Bank Documents on the Current Global Crisis

World Bank Reports

• Global Development Finance 2009: Charting a Global Recovery. The World Bank, Washington DC, 2009. http://web.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTDECPROSPECTS/EXTGDF/EXTGDF2009/0,,contentMDK:22218327~menuPK:5924239~pagePK:64168445~piPK:64168309~theSitePK:5924232,00.html

• Global Monitoring Report 2009: A Development Emergency. The World Bank, Washington DC, 2009. http://web.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTGLOBALMONITOR/EXTGLOMONREP2009/0,,contentMDK:22149019~pagePK:64168445~piPK:64168309~theSitePK:5924405,00.html

Macroeconomics and Finance

• “Global Financial Crisis and Implications for Developing Countries.” G20 Finance’s Ministers Meeting. Sao Paulo, Brazil. November, 2008. World Bank. http://www.worldbank.org/html/extdr/financialcrisis/pdf/G20FinBackgroundpaper.pdf

• “Weathering the Storm: Economic Policy Responses to the Financial Crisis.” Poverty Reduction and Economic Management Network, World Bank, Washington DC. November, 2008. http://siteresources.worldbank.org/NEWS/Resources/weatheringstorm.pdf

• Borchert, Ingo and Mattoo, Aaditya. “The Crisis‐Resilience of Services Trade.” Policy Research Working Paper 4910. November, 2008. World Bank, Washington DC. http://econ.worldbank.org/external/default/main?ImgPagePK=64202990&entityID=000158349_20090428090316&menuPK=64210521&pagePK=64210502&theSitePK=544849&piPK=64210520

• Caprio Jr, Gerard, Demirgüç‐Kunt, Asli and Kane, Edward J. “The 2007 Meltdown in Structured Securitization: Searching for Lessons, Not Scapegoats.” Policy Research Working Paper 4756. October, 2008. Finance and Private Sector Team, World Bank, Washington DC. http://econ.worldbank.org/external/default/main?pagePK=64165259&piPK=64165421&theSitePK=469372&menuPK=64216926&entityID=000158349_20081125132435

• Development Research Group. “Lessons from World Bank Research on Financial Crises.” Policy Research Working Paper 4779. November, 2008. World Bank, Washington DC. http://econ.worldbank.org/external/default/main?pagePK=64165259&theSitePK=469372&piPK=64165421&menuPK=64166093&entityID=000158349_20081216093241

• Financial Crisis Website. The World Bank, Washington DC, 2009. http://www.worldbank.org/html/extdr/financialcrisis/

• Lin, Justin Yifu. “The Impact of the Financial Crisis on Developing Countries.” Korea Development Institute. October, 2008. World Bank, Washington DC. http://siteresources.worldbank.org/DEC/Resources/Oct_31_JustinLin_KDI_remarks.pdf

• World Bank Financial Systems and Development Economics Department. “The Unfolding Crisis: Implications for Financial Systems and Their Oversight.” Financial Systems and Development Economics Departments, World Bank, Washington DC. October, 2008. http://www.worldbank.org/html/extdr/financialcrisis/pdf/UnfoldingCrisis.pdf

Social Impact

• “How Should Labor Market Policy Respond to the Financial Crisis?” Human Development and Poverty Reduction and Economic Management Networks, World Bank, Washington DC. 2009. http://siteresources.worldbank.org/INTLM/Resources/Note‐LM_Crisis_Response_26April.pdf

• “The Financial Crisis and Mandatory Pension Systems in Developing Countries.” Human Development Network, World Bank, Washington DC. 2008. http://siteresources.worldbank.org/INTPENSIONS/Resources/395443‐1121194657824/PRPNote‐Financial_Crisis_12‐10‐2008.pdf

• Ravallion, Martin. “Bailing out the World’s Poorest.” Policy Research Working Paper 4763. October, 2008. Development

Research Group, World Bank, Washington DC. http://www-wds.worldbank.org/external/default/WDSContentServer/IW3P/IB/2008/12/16/000158349_20081216092058/Rendered/PDF/WPS4763.pdf http://econ.worldbank.org/external/default/main?pagePK=64165259&piPK=64165421&theSitePK=469372&menuPK=64216926&entityID=000158349_20081216092058

1

LATIN‐AMERICA BEYOND THE CRISIS—IMPACTS, POLICIES AND OPPORTUNITIES—A SYNTHESIS

Marcelo M. Giugale1

Introduction and Summary

Over the past five years, good policies and good luck had put Latin‐America on a path to prosperity.2 Slowly the mass of its poor was shrinking. In most countries, out went inflation, default, isolation, exclusion, uncertainty. In came budget surpluses, investment grades, free‐trade agreements, cash transfers, institutions. There was still a long way to go, but progress was real. But just when things seemed on track, the first global financial crisis in almost a century reaches the region—and will hit it hard. From fast growth, its economy will suddenly go in reverse. During this difficult period, the World Bank has sought to assist its Latin‐American clients with a package of rapid financial assistance (it tripled its lending) and a large body of crisis‐related policy advice.

This paper synthesizes that body of advice.3 It is organized around three core questions:

(i) How will the crisis impact the region? Slowly and harshly, but without catastrophe; (ii) How should Latin governments respond? With focused social assistance, tailored macro

stimuli, support for the unemployed, and securing debt roll‐overs; (iii) What issues will dominate the post‐crisis regional agenda? The rebalancing of the world’s

economy, short‐term growth management, the middle‐class, a new contract between people and the state, the regulation of finance, and global synergies.

Impacts

The global crisis has entered Latin‐America through four contractions—in external financing (notably, private trade finance), demand for exports, commodity prices and remittances. Different from previous, home‐grown episodes, there have been no massive currency devaluations, bank collapses, debt defaults, inflationary spikes or capital flights. In fact, most countries in the region had, and continue to have, liquid and solvent banking systems, primary fiscal surpluses, and manageable debt burdens. Half a dozen of them also have central banks that have successfully committed to inflation targets, and now find themselves able to allow for flexibility in their foreign exchange rates.

So, given the quality of its macroeconomic framework, what will be the main consequences of the crisis in Latin‐America? There will be five. First, recession. The region’s average growth will shift from over 4 percent p.a. in 2008 to minus 2‐2.5 percent in 2009. These averages hide important differences across

1 The author is the World Bank’s Director of Poverty Reduction and Economic Management for the Latin‐American and Caribbean region. The views expressed in this note are his own, and do not necessarily represent those of the World Bank Group, its Board of Executive Directors, or its member countries. 2 This paper uses the term “Latin‐America” as a short‐hand for Latin‐American and the Caribbean region. 3 The technical notes presented in this compilation can be found in the LCR Crisis Briefs Series at: http://go.worldbank.org/2IWPN6MH20.

2

countries, but very few of them will escape the downturn (Panama, Peru). Growth will return in 2010, but it is likely to be slow (1‐2 percent p.a.) and uneven.

Second, poverty will increase. The World Bank estimates that the crisis will push eight million Latin‐Americans into poverty.4 To put that in perspective, sixty million of them had left poverty during 2002‐2008, thanks to faster growth, smarter social policies, and larger remittances. But the crisis is expected to be unusually harsh on the region’s middle class—mostly because of the fall in the demand for non‐traditional exports that employ formal, urban, technologically‐more‐advanced workers.

Third, unemployment will also increase. All countries for which timely data is available, show short‐term rises in unemployment rates—so far between half and a full percentage point. But the reasons for the rise vary. In some cases (Brazil, Chile, Mexico), it is primarily salaried workers who either lost their jobs or saw job openings shrink; in others (Colombia), it is the self‐employed who are suffering the brunt of the recession. Wages are falling in some sectors in real terms. Informality is expected to expand, and productivity may suffer as a result.

Fourth, there will be less foreign financing. By the time the global crisis broke out (last quarter of 2008), Latin‐American sovereign borrowers had by and large secured the foreign financing they needed for 2009. Corporations, on the other hand, face a much tighter financial outlook. This is not surprising as net private capital flows to emerging markets for the year are projected to dive to less than US$200 billion—a fraction of its almost one trillion dollars peak in 2007. More importantly, foreign direct investment towards Latin‐America may no longer show the resilience of previous crisis, for the flow of mergers and acquisitions that sustained it then (“fire‐sale FDI”) will not be forthcoming now.

Fifth, there will be less remittances. In 2008, the 20 million Latin‐Americans living abroad managed to send home some US$60 billion (a third to Mexico). This made remittances one of the region’s largest foreign currency earners. Those flows will decline between 4 and 8 percent in 2009, and may continue to decline as long as the housing market in the G7 does not recover. And, if the global recovery does not materialize in 2010, a significant number of Latino migrants might return to their countries of origin.

Policy Responses – Today’s Priorities

Latin‐American governments reacted swiftly to the crisis and, on the whole, appropriately. This has defined a short‐term policy agenda that is not entirely similar to the one observed among G7 countries—and rightly so. For the region, the first priority continues to be to avoid a permanent loss of human capital. This is because its countries have a fairly advanced system of social assistance (thirteen of them make direct cash transfers to their poor) but lack a comprehensive system of social insurance (both unemployment and pensions benefits cover only a fraction of the population). The latter automatically reacts to falls in income; the former doesn’t. As a result, past crises in Latin‐America quickly translated into increases in malnutrition, high‐school drop‐outs, and interruptions in the provision of preventive and primary health care. In other words, the crises translated into losses of cognitive capacity amongst young children, a life of informal work for more teenagers, and jumps in mortality rates among adults—even in otherwise middle‐income countries. The mechanisms to avoid those impacts are in place (from school feeding programs to decentralized health budgets), and the costs involved are relatively small (possibly less than a tenth of one percent of GDP).

At the same time, Latin‐American governments will need to see through the various stimulus packages that they have put in place. By necessity, those packages have had a limited scope. On the fiscal side,

4 Poverty is defined here as US$4 PPP per day.

3

room for counter‐cyclical policies is small in all but a few countries (Chile, Brazil, Colombia). This is due to a combination of traditionally low tax collection, insufficient institutional capacity to implement additional public investment quickly enough, and a dearth of lenders willing to finance enlarged fiscal deficits at a time of global crisis. Put differently, fiscal stimulus has been easier for those that saved during the years of bonanza. Things look better on the monetary side. Several countries in the region have built their inflation‐fighter credentials during the period of fast growth and now find themselves able to cut interest rates and let their currencies depreciate to stimulate domestic and external demand, without risking a rise in inflationary expectations.5 More fundamentally, Latin‐America as a whole has not resorted to state ownership as a means of stimulus: governments have not had to take over private corporations, and central banks have not had to open their balance sheets to fund either of them directly. There have been no “bail‐outs” and no “quantitative easing”. The institutional equilibrium built over the past decade has been largely preserved.

However, the crisis is beginning to seriously challenge Latin‐America on two fronts—unemployment and debt. As mentioned before, job losses have so far not been massive. But, as global export demand continues to stagnate through 2010 and, perhaps, beyond, non‐extractive export industries will shed labor at accelerating speed. This will raise political pressure for government action, not least because the shedding will disproportionally affect the middle class. And few countries in the region have in place unemployment insurance systems with sufficient coverage. Some have worked on expanding that coverage (Brazil, Mexico). But, for the most part, efforts have focused on labor intermediation services, training, tax relief for small enterprises, subsidies to youth employment, state‐led temporary employment, and larger budgets for cash transfer programs. Whether these interventions work will depend on what form the recession takes. A short‐and‐sharp “V” shape downturn argues for transitory transfers to smooth the temporary fall in income, while a longer “U” or “L” shape contraction that causes changes in the productive structure calls for programs that facilitate inter‐sectoral adjustments—like retraining.

Independently of the shape of the recession (as of now, an unknown), the region’s governments have turned to public investment as an employment generation tool. They have pledged some US$ 25 billion in additional public works over 2009; data on actual execution is not available yet. The World Bank estimates that, on average, implementing US$ 1 billion of additional infrastructure outlays in Latin‐America employs 40,000 people, depending on the mix of sectors, technology, wages and leakage to imports.6 And the number of permanent jobs created in the economy as a result of those outlays can reach several times that figure.

For all the difficulties involved in employment generation, they may be dwarfed by, and will be framed in, the region’s financing needs. The World Bank estimates that, in 2010, Latin‐American governments will need to borrow between US$350 and 400 billion dollars. That assumes no major fiscal deterioration. It is mostly driven by amortizations coming due. For their part, private corporations will need an estimated US$200 billion next year. Little or no funding has so far been secured by either sovereign or private borrowers—unlike what happened during 2008 with 2009 obligations. At the same time, the international supply of finance will be constrained, even for investment‐grade borrowers, by the crowding‐out effect of the borrowing done by developed nations to pay for their own stimulus packages. And many of the traditional intermediaries of Latin‐American debt (notably investment banks) are currently out of commission or out of business. All this will shift some, perhaps as much as

5 Brasil, Chile, Colombia, Guatemala, México, Perú and Uruguay have set up formal inflation‐targeting arrangements. 6 Rural road maintenance appears to be the outlay with the largest employment impact: 200,000 to 500,000 jobs per billion dollars.

4

half, of the borrowing towards domestic sources—in the few countries that can count on those sources. But it still leaves a large gap. And the ability to roll‐over debt remains the largest single risk in the region’s short‐term horizon.

Policy Responses – Tomorrow’s Opportunities

For all the problems the crisis will cause to Latin‐America, it can also become the event that finally unleashes the region’s enormous potential. A broad agenda of reforms may now, or soon, be possible, based on an unprecedented constellation of new economic realities, political will and technical advances.

The first of those realities is that world growth will no longer be driven by G7 countries consuming beyond their means. At the margin, emerging markets will need to rebalance their export‐led growth models towards domestic absorption. In Latin‐America, this will be easier for the larger countries, but will put smaller economies to the test. Many, large and small, will see their currencies uncomfortably appreciate. All of which will put a new premium on trade competitiveness—even to preserve the same slice of a smaller trade cake. Many of the long‐delayed reforms that make integration worthy, from infrastructure and logistics to tertiary education and property rights, will now become inescapable.

At the same time, the crisis has brought about a new faith in the power of public investment to affect growth in the short term. This may transform that investment, for it will cease to be a de facto source of funding—cut whenever revenues fall or current expenditures rise. Much as in the early 1990s concerns about inflation forced the region’s governments to surrender money printing as a source of fiscal deficit financing, concerns about recession may now force them to formally link public investment to short‐term growth prospects—systematically doing more in the downturn and saving during upturns. It may also lead the marginal dollar of public investment towards projects that bring bigger private contributions, as they will have the largest total impact on growth. And it may usher a much‐needed improvement in implementation capacity. Of course, the technical and institutional issues around giving public investment a growth stabilization role are not minor. But the core principle of saving in good times to spend in bad ones made its debut in Latin‐America during this crisis (in Chile), and it has proven a success that many will seek to replicate.

The crisis may also transform social policy in Latin‐America, making it much more about equity than equality, that is, more about giving all the same opportunities rather than the same rewards. This will help the region progress beyond a debate that has for far too long been politically divisive and strategically paralyzing—a debate about whether the very purpose of the state is to protect private property or to pursue wealth redistribution. A combination of factors will account for the transformation. On the one hand, the crisis revealed that the region’s social assistance systems are insufficient to respond to sudden economy‐wide contractions in income, especially among the middle classes. On the other, the technology to measure inequality of opportunities has recently become available and, more critically, operational.7 Both realities will unlock a long overdue effort to focalize universal subsidies—why should the state continue to pay for, say, the heating, gasoline or college education consumed by the rich? The end result will be a pattern of social policy more focused on giving the same chances to all.

7 See Barros, R. P. de; F. H. G. Ferreira, J. R. Molinas Vega, and J. Saavedra Chanduvi, 2009, Measuring Inequality of Opportunity in Latin American and the Caribbean. New York: Palgrave Macmillian and Washington, DC: The World Bank.

5

More broadly, the role of the state will change worldwide, and Latin‐America will be no exception. What is different in the region is that the relationship between its states and its people has long been one of mistrust—a manifestation of which is Latin‐Americans’ idiosyncratic reluctance to pay taxes. The crisis may become an opportunity to change that relationship, to reach a new contract. At a time when less resources will be available to the state but more will be expected of it—from regulating finance to facilitating job creation—the door opened to begin to base public sector management on results. The technology is now available to connect public action, and more particularly public expenditures, to specific outcomes—in education, in health, in infrastructure, in public services. Several countries of the region were moving that way before the crisis, at both federal and sub‐national levels. That move is now likely to become the norm.

Result‐based management of the state will put a framework to its interventions in many sectors where it had been less active in the past. Nowhere will that be clearer than in finance. By and large, Latin‐America avoided many of the mistakes that led to, and triggered, the implosion of financial markets in the developed world—no subprime lending, no off‐balance‐sheet risks, no exotic instruments. Much of this is due to over a decade of laborious improvements in regulatory and supervisory institutions. Those institutions will now be challenged by the sweeping reforms that the global financial industry is about undergo. Systemic risk regulation, capital requirements, the use of credit ratings, accounting norms and consumer protection are just some of the parameters of the industry that will change worldwide. How Latin‐America adopts and adapts those parameters may prove critical for a region that will increasingly have to rely on its own savings to develop.8

Finally, the crisis has revealed the breadth of global interconnectedness—witness the viral speed at which financial and trade flows collapsed across the world. The externalities created by the actions of individual countries have become so patent that quickly triggered the appearance of new or renewed mechanisms for global coordination and support. Many of those mechanisms are critical for post‐crisis Latin‐America, from the G20 (where Argentina, Brazil and Mexico participate) to the increases in lending capacity of multilaterals to a trade regime that remains open, fair and clean. Making the most of them is the opportunity of a generation.

Conclusion ‐ The Day After Tomorrow, Latin‐America May Be Better

So, as thresholds go, 2009 may be remembered as the year in which Latin‐America’s latest growth run abruptly ended. Or as the year in which an unprecedented global crisis shook the region onto a faster development path. Which way it goes will depend on how its policy‐makers respond, whether they tailor their reactions to their reality, see the opportunity behind the crisis, and proactively take on the issues that were holding Latinos back well before subprime became a household term. Clearly, Latin‐American problems are not solely economic. The institutions that underpin politics are not fully cemented. Violence and the drugs trade that fuels it have their own dynamics. And nobody knows what development policies will work best in the post‐crisis world. But it remains true, and somewhat ironic, that a region that could not quite take off when the world was booming, could now make it on its own terms when the world tumbled.

8 For a comprehensive framework of ideas on the new regulation of finance see De la Torre, A. and A. Ize, 2009, Regulatory Reform: Integrating Paradigms, World Bank Policy Research Working Paper 4842, February.

6

7

AMÉRICA LATINA MÁS ALLÁ DE LA CRISIS—IMPACTOS, POLÍTICAS Y OPORTUNIDADES— SÍNTESIS

Marcelo M. Giugale9

Introducción y Resumen

Durante los últimos cinco años, buenas políticas y buena suerte pusieron a América Latina en un sendero de prosperidad.10 Lentamente la masa de pobreza se redujo. La mayoría de los países comenzaron a dejar atrás la inflación, la bancarrota, el aislamiento, la exclusión, y la incertidumbre. Aparecíeron los superávit fiscales, los grados de inversión, los tratados de libre comercio, las transferencias directas a los pobres, las instituciones. Había todavía mucho camino por recorrer, pero el progreso era real y tangible. Pero cuando las cosas finalmente parecían estar en el camino correcto, la primera crisis financiera global en casi un siglo llega a la región—y la va golpeará con fuerza. Su economía pasará del crecimiento rápido a la recesión. Durante este periodo indudablemente difícil, el Banco Mundial ha buscado apoyar a sus clientes Latino‐americanos con un paquete de asistencia financiera rápida (triplicó su volumen de préstamos) y un compendio de análisis de políticas públicas para responder a la crisis.

Este ensayo sintetiza ese compendio.11 Esta organizado alrededor de tres preguntas centrales:

(i) ¿Cómo afectará la crisis a la región? Lenta y duramente, pero sin catástrofe. (ii) ¿Cómo deberían responder los gobiernos latino‐americanos? Con asistencia social

focalizada, estímulos macroeconómicos a medida, apoyo a los desempleados, y asegurando el refinanciamiento de deudas.

(iii) ¿Cuáles son los temas que dominarán la agenda regional después de la crisis? El rebalanceo de la economía mundial, el manejo del crecimiento de corto plazo, la clase media, un nuevo contrato entre el estado y la gente, la regulación financiera, y las sinergias globales.

Impactos

La crisis global ha entrado en América Latina a través de cuatro contracciones—en financiamiento externo (en especial para el comercio internacional privado), demanda por exportaciones, precios de las materias primas, y remesas. A diferencia de episodios nacionales anteriores, no ha habido devaluaciones masivas de la moneda, colapsos bancarios, bancarrotas, picos inflacionarios o fuga de capitales. De hecho, la mayoría de los países de la región tenían, y siguen teniendo, sistemas bancarios líquidos y

9 El autor es el Director de Política Económica y Programas de Reducción de Pobreza del Banco Mundial para América Latina. Las opiniones expresadas en este documento le pertenecen, y no necesariamente reflejan las del Grupo Banco Mundial, su Directorio, o las de sus países miembros. 10 En este documento, el término «América Latina» se usa como abreviación de « América Latina y el Caribe ». . 11 La colección completa de notas técnicas contenidas en este compendio son parte del LCR Crisis Briefs Series y pueden encontrarse en: http://go.worldbank.org/2IWPN6MH20.

8

solventes, superávit fiscales primarios, y deudas manejables. Media docena de ellos también tienen bancos centrales que se han comprometido exitosamente con metas de inflación, y como consecuencia pueden ahora permitirse flexibilidad en sus tasas de cambio.

Dada esa calidad en el marco macroeconómico, ¿cuáles serán las consecuencias principales de la crisis para América Latina? Serán cinco. Primero, recesión. El crecimiento promedio de la región pasara de más de 4 por ciento en el 2008 a menos 2‐2.5 en el 2009. Estos promedios disimulan grandes diferencias entre países, pero muy pocos escaparan la caída en el producto (Panamá, Perú). El crecimiento volverá en el 2010, pero es probable que sea lento (1‐2 por ciento anual) y no uniforme.

Segundo, la pobreza se incrementará. El Banco Mundial estima que la crisis empujará a ocho millones de latino‐americanos a la pobreza.12 Para poner ese número en perspectiva, sesenta millones de ellos habían salido de la pobreza en el periodo 2002‐2008, gracias al crecimiento más rápido, a las mejores políticas sociales, y a las mayores remesas. Pero se espera que la crisis sea inusualmente dura con la clase media—por la caída en la demanda por exportaciones no tradicionales que tienden a emplear a trabajadores formales, urbanos y tecnológicamente más avanzados.

Tercero, el desempleo también se incrementará. Todos los países para los que existen datos puntuales, muestran un aumento de corto plazo en las tasas de desempleo—hasta ahora, de entre medio y un punto porcentual. Pero las razones detrás del aumento varían. En algunos casos (Brasil, Chile, México), son mayormente los trabajadores en relación de dependencia (“asalariados”) los que han perdido su empleo o encuentran menos oportunidades de empleo; en otros (Colombia), son los trabajadores independientes los que parecen estar sintiendo más el impacto de la recesión. Los salarios están cayendo en algunos sectores en términos reales. Se espera que la informalidad se expanda, y que la productividad sufra como resultado.

Cuarto, habrá menos financiamiento externo. Al momento en que se detonó la crisis global (último trimestre del 2008), los deudores Latino‐americanos soberanos, en su mayoría se habían ya asegurado el financiamiento externo que necesitaban para el 2009. Las corporaciones, en cambio, enfrentan un panorama financiero mucho más difícil. Esto no es muy sorprendente, pues las proyecciones del flujo neto de capital privado hacia los países emergentes para este año muestran un verdadero derrumbe—pasarán de un pico de casi un millón de millones de dólares en el 2007, a menos de 200,000 millones. Aun mas importante, la inversión extranjera directa hacia América Latina tal vez no siga mostrando la estabilidad que tuvo durante crisis anteriores, porque el flujo de fusiones y adquisiciones que la sostenía (“compras de remate”) ya no se harán presentes.

Quinto, habrá menos remesas. En el 2008, los 20 millones de latino‐americanos que viven en el exterior enviaron unos 60,000 millones de dólares (un tercio de ese dinero fue a México). Esto convirtió a las remesas en una de las más grandes fuentes de divisas de la región. Esos flujos se reducirán entre un 4 y un 8 por ciento en el 2009, y pueden continuar cayendo mientras no se recupere la industria de la vivienda en los países G7. Y, si la recuperación global no se materializa en el 2010, un número significativo de migrantes latinos podría volver a casa.

12 La pobreza se define aquí como US$4 PPP por día o menos.

9

Respuestas de Política Pública – Las Prioridades de Hoy.

Los gobiernos latino‐americanos reaccionaron rápidamente a la crisis y, en general, lo hicieron en forma adecuada. Esto ha definido una agenda de políticas públicas de corto plazo que no es exactamente igual a la que implementaron los países G7—y es acertado que así sea. Para la región, la primera prioridad continua siendo evitar una pérdida permanente de capital humano. La razón es que sus países tienen sistemas de asistencia social bastante desarrollados (trece de ellos hacen transferencias directas en efectivo hacia sus ciudadanos pobres) pero carecen de un sistema de seguridad social comprensivo (los beneficios por desempleo y por pensión solo cubren una pequeña porción de la población). Este último reacciona automáticamente cuando el ingreso familiar cae; el anterior no lo hace. No es casualidad entonces que, en el pasado, las crisis latino‐americanas se hayan traducido en aumentos en desnutrición, deserción en escolaridad secundaria, e interrupciones en los servicios de medicina básica y preventiva. En otras palabras, las crisis se traducían en pérdida de capacidad cognitiva entre los niños, una vida de trabajo informal para mas adolescentes, y un salto en la tasa de mortalidad entre adultos—aun en países que se consideran de ingreso medio. Los mecanismos para evitar esos impactos están ya en su lugar (desde comedores escolares hasta presupuestos de salud descentralizados), y el costo es relativamente bajo (posiblemente menos de una décima de uno por ciento del PIB).

Al mismo tiempo, los gobiernos latino‐americanos tendrán que implementar los paquetes de estimulo que han puesto en marcha. Por necesidad, el alcance de esos paquetes es limitado. Del lado fiscal, el espacio para hacer políticas “contra‐cíclicas” es pequeño en casi todos los países (Chile, Brasil, Colombia parecen ser la excepción). Esto se debe a una combinación de recaudación impositiva tradicionalmente baja, insuficiente capacidad institucional para ejecutar inversión pública adicional con rapidez, y ausencia de acreedores dispuestos a financiar expansiones en el déficit fiscal durante un tiempo de crisis global. Dicho de otra forma, el estimulo fiscal ha sido más fácil para aquellos que ahorraron durante los tiempos de bonanza. Las cosas lucen mejor del lado monetario. Varios países de la región han ganado credibilidad en su lucha contra la inflación durante el periodo de crecimiento rápido y son ahora capaces de bajar sus tasas de interés y devaluar sus monedas para estimular la demanda domestica y externa sin arriesgar un repunte en expectativas inflacionarias.13 Mas fundamentalmente, América Latina en su conjunto no ha recurrido a la propiedad pública como instrumento de estimulo: los gobiernos no han tomado control de empresas privadas, y los bancos centrales no han abierto sus balances a fondear directamente a ninguno de los dos. No ha habido ni rescates (“bail‐outs”) ni emisiones monetarias directas (“quantitative easing”). El equilibrio institucional laboriosamente construido durante la última década ha sido preservado.

Sin embargo, la crisis comienza a presentar dos serios desafíos para América Latina—el desempleo y la deuda. Como se mencionó antes, la pérdida de puestos de trabajo aún no ha sido masiva. Pero, en la medida en que la demanda mundial por exportaciones continúe estancada durante el 2010 y, tal vez, más allá del 2010, las industrias de exportación no extractivas acelerarán su tasa de despidos. Esto incrementara la presión política para que los gobiernos actúen, especialmente porque los despidos afectaran desproporcionalmente a la clase media. Y pocos países de la región tienen sistemas de seguro de desempleo con suficiente cobertura. Algunos han estado trabajando en expandir esa cobertura (Brasil, México). Pero, en general, las intervenciones se han concentrado en los servicios de intermediación de trabajo, entrenamiento, exenciones impositivas para pequeñas empresas, subsidios al empleo de jóvenes, programas de empleo temporal, y mayores presupuestos para los programas de transferencias directas en efectivo. El éxito que tengan esas intervenciones dependerá de la forma que tome la recesión. Una recesión profunda pero corta (en “V”) apunta a transferencias transitorias para 13 Brasil, Chile, Colombia, Guatemala, México, Perú y Uruguay han establecido sistemas formales de metas de inflación.

10

suavizar la caída temporaria en el ingreso como la mejor opción, mientras que una caída más prolongada en el producto (en “U” o en “L”) con cambios permanentes en la estructura productiva necesitará de políticas que faciliten el ajuste inter‐sectorial‐‐‐como los programas de re‐entrenamiento.

Independientemente de la forma que tome la recesión (una variable hasta ahora desconocida), los gobiernos de la región ha buscado usar a la inversión publica como instrumento de generación de empleo. En total, han anunciado unos 25,000 millones de dólares en obras publicas adicionales para el 2009; los datos de ejecución efectiva todavía no están disponibles. El Banco Mundial estima que, en promedio, implementar 1,000 millones de dólares en gastos de infraestructura en América Latina requiere emplear 40,000 trabajadores, dependiendo de la mezcla de sectores, tecnologías, salarios y necesidades de importación. 14 Y el número de puestos de trabajo permanentes creados en la economía como resultado de esos gastos puede llegar a varias veces esa cifra.

Las dificultades que significa crear empleo se comparan con, y estarán enmarcadas en, las dificultades para asegurar el financiamiento de la región. El Banco Mundial ha estimado que, en el 2010, los gobiernos de América Latina necesitaran pedir prestados entre 350,000 y 400,000 millones de dólares. Esto supone que no habrá mayores deterioros fiscales. Esta primeramente basado en los vencimientos de deuda que ocurrirán ese año. Por su parte, las corporaciones privadas necesitarán aproximadamente 200,000 millones de dólares. Pocos de esos fondos, públicos y privados, han sido asegurados hasta el momento—a diferencia de lo ocurrido en el 2008 con respecto a las obligaciones del 2009. Al mismo tiempo, la oferta internacional de fondos estará limitada, aún para deudores con grado de inversión, por el efecto de desplazo (“crowding‐out”) que causará el endeudamiento en el que incurrirán los países desarrollados para pagar por sus propios paquetes de estimulo. Y muchos de los intermediarios tradicionales de la deuda latino‐americana (en particular, bancos de inversión) están al momento fuera de actividad o en bancarrota. Todo esto conducirá parte, tal vez la mitad, del las necesidades de financiamiento hacia fuentes domésticas—en los países que cuentan con esas fuentes. Pero aún así habrá una amplia brecha. Y la capacidad de refinanciar deuda (“debt roll‐over”) se constituye como el riesgo individual más grande que existe en el horizonte de corto plazo de la región.

Respuestas de Política Pública – Las Oportunidades de Mañana

Por todos los problemas que la crisis causará a América Latina, puede también convertirse en el evento que finalmente libera el enorme potencial de la región. Una amplia agenda de reforma podría ahora, o muy pronto, hacerse viable en base a una constelación sin precedentes de nuevas realidades económicas, voluntad política y avances técnicos.

La primera de esas realidades es que el crecimiento del mundo ya no estará motorizado por países G7 consumiendo más allá de sus recursos. Al margen, los países emergentes necesitarán balancear sus modelos de crecimiento exportador con mayor participación de la absorción doméstica. En América Latina, eso será más fácil de hacer para países grandes, pero pondrá a prueba a las economías más pequeñas. Muchas, grandes y pequeñas, verán sus monedas apreciarse incómodamente. Todo lo cual dará un valor adicional a la competitividad comercial—aún para preservar la misma porción de un pastel más chico. Muchas de las postergadas reformas que hacen que la globalización rinda frutos—desde la infraestructura y la logística hasta la educación terciaria y los derechos de propiedad—se harán ahora inevitables.

14 El mantenimiento de caminos rurales parece ser el gasto en infraestructura que conlleva la mayor necesidad de empleo : entre 200,000 y 500,000 trabajadores por cada 1000 millones de dólares de gasto implementado.

11

Al mismo tiempo, la crisis ha generado una nueva fe en el poder de la inversión pública para afectar el crecimiento de corto plazo. Esto puede transformar a esa inversión, pues dejará de ser, de facto, una fuente de financiamiento—que se corta cuando caen los ingresos o aumentan los gastos corrientes. Así como al principio de los 90s la preocupación por la inflación forzó a los gobiernos de la región a abandonar la impresión de moneda como fuente de financiamiento del déficit fiscal, preocupación por la recesión puede ahora forzarlos a ligar formalmente la inversión pública con el panorama de crecimiento de corto plazo—sistemáticamente invirtiendo más cuando el ciclo productivo cae y ahorrando cuando sube. Esto también llevaría el dólar marginal de inversión pública hacia proyectos que movilicen mayores contribuciones privadas, pues tendrían mayor impacto en el crecimiento. Y detonaría las mejoras necesarias en la capacidad de implementación. Por supuesto, los problemas técnicos e institucionales de dar a la inversión pública el rol de estabilizador del crecimiento no son menores. Pero el principio central de ahorrar en los tiempos buenos para gastar en los malos hizo su debut en América Latina durante esta crisis (en Chile), y ha probado ser un éxito que muchos buscarán replicar.

La crisis podría también transformar la política social de América Latina, dirigiéndola más hacia la equidad que hacia la igualdad, esto es, más en dar a todos las mismas oportunidades que en dar a todos los mismos premios. Esto ayudaría a la región a dejar atrás un debate que, por décadas, ha sido políticamente divisivo y estratégicamente paralizante—un debate sobre si el propósito mismo del estado es proteger la propiedad privada o redistribuir la riqueza. Una combinación de factores dará cuenta de la transformación. Por un lado, la crisis reveló que los sistemas de asistencia social de la región no son suficientes para responder una contracción súbita del ingreso a través de la economía, especialmente en las clases medias. Por otro, la tecnología para medir la desigualdad de oportunidades ha sido recientemente desarrollada, está disponible, y es operacional.15 Ambas realidades destrabarán los esfuerzos para focalizar los subsidios universales—¿porque debe el estado continuar pagando por, digamos, la calefacción, la gasolina o la educación universitaria que consumen los ricos? El resultado final será una matriz de política social más enfocada en dar a todos las mismas chances.

Más ampliamente, el rol del estado cambiara en el mundo entero, y América Latina no será la excepción. Lo que es diferente en la región es que la relación entre sus estados y sus pueblos ha por mucho tiempo sido una de desconfianza—una manifestación de lo cual es la resistencia idiosincrática de los latino‐americanos a pagar impuestos. La crisis podría tornarse en una oportunidad para cambiar esa relación, para llegar a un nuevo contrato. En un momento en que habrá menos recursos para el estado, más se esperará de él—desde regular más las finanzas a crear más empleo—y la puerta se abrirá para comenzar a basar la gestión del estado en resultados. La tecnología ya está disponible para conectar la acción pública, y más particularmente el gasto público, con resultados específicos—en educación, en salud, en infraestructura, en servicios públicos. Varios países de la región se estaban moviendo en esa dirección antes de la crisis, tanto a nivel federal como sub‐nacional. Ese movimiento probablemente se convertirá ahora en la norma.

La gestión del estado por resultados pondrá un marco a sus intervenciones en sectores donde era menos activo en el pasado. El caso más claro es el sector financiero. En general, América Latina evitó muchos de los errores que llevaron a, y detonaron, la implosión de los mercados financieros en el mundo desarrollado—no hubo endeudamiento “subprime”, ni acumulación riesgos fuera de balance, ni instrumentos exóticos. Mucho de eso se debe a más de una década de meticulosas mejoras en las instituciones regulatorias y de supervisión. Esas instituciones enfrentarán ahora el desafío de las grandes

15 Ver Barros, R. P. de; F. H. G. Ferreira, J. R. Molinas Vega, and J. Saavedra Chanduvi, 2009, Measuring Inequality of Opportunity in Latin American and the Caribbean. (Midiendo la Desigualdad de Oportunidades en América Latina y el Caribe). New York: Palgrave Macmillian and Washington, DC: The World Bank.

12

reformas de las que será sujeto la industria financiera global. La regulación del riesgo sistémico, los requerimientos de capital, el uso de calificaciones de deuda, las normas contables, y la protección al consumidor son solo algunos de los parámetros de la industria que cambiaran alrededor del mundo. Como adopta y adapta América Latina esos parámetros será critico para una región que, con mayor frecuencia, deberá recurrir al ahorro domestico para desarrollarse.16

Finalmente, la crisis ha revelado el alcance de la interconexión global—basta con ver la velocidad viral a la que colapsaron los flujos financieros y comerciales alrededor del mundo. Las externalidades creadas por las acciones de países individuales han sido tan patentes que han llevaron a la aparición de nuevos o renovados mecanismos globales de coordinación y apoyo. Muchos de esos mecanismos son esenciales para la America‐Latina de post‐crisis, desde el grupo G20 (donde Argentina, Brasil y México participan) a los incrementos en la capacidad de préstamo de los organismos multilaterales a un régimen de comercio internacional abierto, justo y sustentable. Aprovecharlos al máximo es la oportunidad de esta generación.

Conclusiones – Pasado Mañana, América Latina Puede Estar Mejor

Como hito, el 2009 podría ser recordado como el año en el cual otro brote de crecimiento en América Latina llegó a un abrupto fin. O como el año en el cual una crisis global sin precedentes puso a la región sobre un sendero de desarrollo mucho más rápido, y más duradero. Cuál de los dos resultados se haga realidad, dependerá de cómo respondan sus líderes, si adaptan sus respuestas a las capacidades de sus economías, si ven la oportunidad detrás de la crisis, y si pro‐activamente abordan los problemas que frenaban a los latino‐americanos mucho antes que “subprime” fuera una palabra de uso común. Claramente, los problemas de América Latina no son solo económicos. Las instituciones sobre las que se basan sus sistemas políticos todavía no están completamente consolidadas. La violencia y el narco‐tráfico que la alimenta tienen una dinámica propia. Y nadie realmente sabe que políticas de desarrollo funcionaran mejor en el mundo post‐crisis. Pero por eso no deja de ser cierto, y un tanto irónico, que la región que no había podido despegar cuando el mundo estaba en auge, podría hacerlo ahora que el mundo tambalea.

16 Para acceder a un marco conceptual comprensivo de ideas sobre la nueva regulación financiera, ver De la Torre, A. and A. Ize, 2009, Regulatory Reform: Integrating Paradigms (Reforma Regulatoria: Integrando Paradigmas), World Bank Policy Research Working Paper 4842, February.

13

1. THE GLOBAL FINANCIAL AND ECONOMIC STORM: How Bad is the Weather in Latin America and the Caribbean?

Augusto de la Torre

April 2009*

Executive Summary

The current crisis, originated in the advanced financial markets of the center, has generated alarming ripple effects throughout the periphery. No emerging economy has remained immune to its destructive power, which intensified dramatically in the 4th quarter of 2008 after the failure of Lehman Brothers. The crisis is far from over and its rapid spread to the Latin America and the Caribbean (LAC) is occurring through mutually reinforcing channels (financial, remittances, terms of trade, export demand), leading to a sharp downturn in economic activity. Nevertheless, LAC is better prepared than in the past to withstand the global storm. Its traditional sources of vulnerability and magnification of external shocks—local currency, fiscal stance, financial system, external sector—are this time around, and by and large, not part of the problem. As a result, LAC may be able to avert a systemic financial crisis at home and a number of LAC countries enjoy some space for counter‐cyclical policy, particularly in the monetary field. However, the magnitude of the shock is such that LAC will inevitably endure an economic recession so long as the global crisis lasts. Moreover, LAC remains vulnerable to a recession‐induced reversal in social gains. Arguably, such a reversal would be a more difficult affair to manage in a period of electoral contests and considering that social indicators in the region, while having registered significant improvements in recent years, remain generally well below those of middle‐income countries in other regions. This puts a premium on preventing an undue contraction in vital public spending in health, education, basic infrastructure, and social programs. The recovery path for LAC depends crucially on the ability of rich countries and key emerging economies to successfully contain and recover from the current crisis. It also hinges on the availability of substantial financial support from multilateral institutions and on the prudency and effectiveness of LAC’s own policy responses. The Big Storm that originated in the center…

The current crisis affecting the world economy is of historical dimensions and is re‐shaping the international economic and financial landscape, including the traditional dividing lines between center and periphery. The eye of the storm is in the advanced economies, where the crisis has resulted in colossal failures of financial institutions, massive deleveraging, and a staggering collapse in asset values (some US$ 18 trillion in G‐7 stock market capitalization has vanished relative to the admittedly overvalued pre‐crisis peaks). The turmoil has also produced enormous job losses (total employment in the US and Euro area has shrunk by about 6.5 million jobs since the beginning of the recession) and a sharp contraction in economic activity (in the 4th quarter of 2008, GDP fell at an annualized quarterly rate of 6.3% in the U.S. and the Euro Area, and 12.1% in Japan). In an effort to restore confidence in financial markets and pull the economy out of the hole, governments in rich countries have resorted to large‐scale financial rescue and fiscal stimulus packages—raising the degree of state intervention in private markets to levels not seen since the Great Depression.

*Prepared for the IMf/World Bank Spring Meeting of April 2009.

14

… has reached deeply into the periphery

The ripple effects of the crisis on the periphery are in full swing and acting through multiple transmission channels. These include the sharply reduced availability of international market finance (debt and equity), the deterioration of terms of trade for net commodity exporters, the decline in remittances, and the pronounced contraction of external demand for emerging economies’ goods and services. As a result, the periphery is being forced into painful adjustments and the entire world is in crisis. Global trade is declining for the first time in 25 years and world GDP is expected to fall by nearly 2 percent in 2009. The International Labor Organization predicts that global unemployment could reach 38 million workers in 2009, up from 14 million in 2008. Furthermore, the globalization of this crisis has accentuated the feedback loops between the mentioned transmission channels. For example, the world recession keeps commodity prices and exports down, causing loan quality to decay. In turn, this threatens employment, weakens profit expectations, and undermines credit flows, all of which further undercuts private investment and consumption, and so on.

Timely policy responses are called for, but constraints vary across emerging countries

The recessionary implications of the external shock call in principle for timely responses, including counter‐cyclical macroeconomic policies, scaling up of social protection and basic infrastructure programs, and significant real exchange realignments to dampen the output and employment sacrifices involved in the adjustment. However, the capacity of emerging economies to respond in practice along these dimensions depends not just on the availability of financial resources from multilateral institutions but also on key policy and structural factors that determine the degree of vulnerability to the shocks as well as the scope for policy maneuver. These factors include:

a. The extent of pre‐existing macroeconomic and financial policy weaknesses; b. The extent of poverty and inequality, and the degree of social conflict; c. Structural features, such as the diversification of trade, the degree of trade and financial

openness, the extent of integration of the local economy to the global production chain, and the allocation of labor between tradable and non‐tradable sectors.

LAC is better prepared in the macro‐financial area, compared to its own past…

LAC’s history has been marked by frequent and devastating financial crises. In previous episodes (such as the debt crisis in the early 80s, the Tequila crisis in 1995, and the Asian and Russian crises of the late 90s), LAC countries were usually caught with substantial, home‐grown macroeconomic and financial vulnerabilities—reflected in high inflation, overvalued currencies, ample fiscal and current account deficits, and widespread maturity and currency mismatches (Figures 1 and 2). These conditions sapped LAC’s ability to undertake counter‐cyclical policies. LAC was instead compelled to raise interest rates or deeply cut fiscal spending in the midst of the crises in order to keep investors from fleeing, but exacerbating output and employment losses. In several past episodes, such desperate measures were unable to prevent financial meltdowns.

Fortunately, the pains from past crises have led to significant institutional and policy improvements in LAC’s macroeconomic and financial areas. More specifically, LAC’s vulnerability to shocks has fallen in tandem with the emergence of: (i) sounder and more flexible currencies; (ii) more resilient financial systems; (iii) better fiscal and public debt management; and (iv) stronger external positions. These

15

improvements are perhaps most noticeable in countries that have been able to build credible inflation targeting regimes.

Regarding local currencies, an increased number of countries in the region have moved to flexible exchange rate arrangements (Figure 3). The effectiveness of these regimes has risen in line with a reduction in the pass‐through —implying that exchange rate movements now coexist with low and stable inflation—and the deepening of local currency debt markets—implying that exchange rate movements now generate much less adverse balance sheet effects. Similarly, the resiliency of LAC’s financial systems has increased, reflecting significant reforms in financial legislation, regulation, and infrastructure that were introduced following the crises of the late‐1990s. These reforms have led, among other things, to a virtuous combination of financial deepening and a rising share of loans funded by local currency deposits (Figure 4).

LAC’s fiscal and external conditions have also registered improvements. To be sure, much of the good fiscal and external outcomes were driven by good luck. That is, LAC countries benefited from the benign external environment of the recent past, characterized by abundant liquidity, booming commodity prices, and vigorous global growth. Nevertheless, better policy frameworks are part of the story too. In the area of public finances, enhanced debt management systems and greater discipline in fiscal policy contributed to reductions in government debt burdens and improvements in the currency and term structure of such debts (Figures 5 and 6). The result was greater fiscal sustainability, even if, with the notable exception of Chile, LAC governments did not save sufficiently during the good times. Finally, in the external front, there was a substantial accumulation of international reserves across LAC countries, which was due not just to terms‐of‐trade windfall gains but also to efforts to self‐insure against capital flow reversals (Figure 7).

In sum, improved policy frameworks in LAC have contributed to reducing the weaknesses that used to be incubated in the monetary, financial, fiscal, and external fronts. These vulnerabilities tended to greatly magnify the adverse effects of external shocks. In the current crisis, such vulnerabilities are tamed and are, thus, not an independent source of shock amplification. As a result, many LAC countries are likely to avert a systemic financial crisis at home. However, this reduced vulnerability is insufficient to prevent bad consequences—the storm spreading from the center to the periphery is of such a formidable magnitude that its recessionary impact is already being felt. Moreover, if the global crisis becomes more acute or extends beyond 2009, fiscal and financial conditions can weaken to a point where they could well become an increasing part of the problem. Finally, while vulnerabilities have decreased for the LAC region as a whole, there is considerable heterogeneity across LAC countries. A few countries are still saddled with significant fiscal and public debt complications which limit budgetary maneuvering room. Many others, particularly among the smaller countries in Central America and the Caribbean, have heavily‐managed or pegged exchange rates and cannot therefore undertake counter‐cyclical monetary policy.

… and perhaps also, at least in some respects, compared to other emerging regions

Unlike many past experiences, some evidence suggests that LAC was this time in a relative good position vis‐à‐vis other emerging regions when the crisis hit. For instance, countries in LAC had on average lower inflation rates and were equipped with more flexible exchange rate arrangements compared to Eastern Europe and East Asia (Figures 8 and 9). Furthermore, in contrast with large current account deficits in Eastern Europe and South Asia, LAC was running current account surpluses before the crisis. Also, while significantly below South Asia, LAC’s ratio of international reserves to short‐term external debt compared favorably to those of East Asia and was above that of Eastern Europe (Figure 10). Moreover,

16

financial systems in LAC, although smaller on average than in other emerging regions, had a larger share of loans backed by local deposits, a factor contributing to resiliency to reversals in capital inflows. East and South Asia had a lower loan to deposit ratio while in Eastern Europe a very high share of credit was funded by foreign inflows (Figure 11).

LAC will most likely endure a recession this year

The 4th quarter of 2008 marked a clear point of inflexion for LAC and the world economy. Prior to that, LAC (a net commodity‐exporting region) had been enjoying a short‐lived decoupling stage underpinned by an accelerated rise in commodity prices. During that stage, even as the subprime crisis and economic slowdown spread through the economies of the center, LAC currencies strengthened, foreign direct investment continued flowing in, and growth kept apace. The key policy concern for LAC then was inflation, which was being pushed by rising international prices of foods and fuels. That situation began turning around as commodity prices fell, and came to an abrupt end with the financial devastation unleashed by the failure of Lehman Brothers in September 2008. As a result, financial flows and economic activity throughout the world took a highly synchronized nose dive, and LAC fell into a sort of global economic whirlpool.

The overall deterioration of economic conditions that was registered in the 4th quarter of 2008 has been unprecedented. During that quarter, on average for LAC, the cost of international borrowing for firms doubled; corporate issues of debt and equity securities came to a virtual halt; the flow of credit by private banks stagnated; remittances began contracting sharply; exports and imports shrunk by about 30 and 25 percent, respectively, as trade surpluses vanished; and industrial production fell by about 12 percent (Figures 12 and 13).

As these developments were linked to tectonic shifts in the advanced economies, LAC countries that are closely linked to the U.S. economy (including Mexico and the small open economies of Central America and the Caribbean) have felt a more direct and stronger initial impact. For other countries in the region, the repercussions are being felt with a lag. But in all cases, growth prospects for 2009 have been downgraded dramatically as news on economic performance of advanced and developing countries were revealed on the 4th quarter. For instance, as of August 2008 the Consensus Forecasts put GDP growth for LAC in 2009 at around 3.7 percent; by contrast, the most recent Consensus Forecasts (March 2009) sees LAC growth this year in the negative territory, at around ‐0.7 percent.

While the range of growth forecasts is wide—reflecting the more general uncertainty about world growth—few doubt that 2009 would be, at a minimum, a year of economic stagnation for LAC as a whole and, perhaps more likely, a year of recession.

LAC is especially vulnerable to a recession‐induced reversal of social gains

Poverty and inequality figures, as well as other social indicators, improved markedly in LAC during the last decade. For instance, infant mortality declined to 21.4 deaths per 1000 live births in 2006 from 36.1 in 1995, which is closely related to a larger access of the population to improved water and sanitation. Similarly, during the strong growth period of 2002‐2008, almost 60 million people in LAC were lifted out of poverty (measured at PPP‐adjusted US$4 a day) and 41 million left the ranks of extreme poverty (measured at US$2 a day). However, LAC still lags considerably other emerging regions in social dimensions. For instance, infant mortality is higher, educational achievement lower, basic infrastructure much less developed, and income distribution much more unequal in LAC compared to East Asia and Eastern Europe (Figures 14‐16). Given LAC’s unique combination of vibrant electoral processes with

17

high income and wealth inequality, these deficits in social indicators suggest that a reversal in the recently achieved social gains might be a comparatively more complicated affair to handle for LAC. Hence, there is a premium on preventing an undue contraction in public spending in health, education, basic infrastructure, and other social programs.

The scope for counter‐cyclical policy responses varies considerably across LAC countries

Perhaps the greatest scope for counter‐cyclical policy is in the monetary arena. LAC countries with robust inflation targeting regimes (Brazil, Chile, Colombia, Mexico, and Peru) are clearly in a better position, with exchange rate flexibility and high international reserves affording them maneuvering margins. These countries have in fact entered into aggressive monetary policy easing, especially since January 2009 (Figure 17). Some of them have also used actively their public banks to offset the decline in private bank credit. Monetary easing has helped cushion the decline in economic activity through two main channels. First, by lowering policy interest rates, it dampens the fall in investment and consumption. Second, by allowing the currency to depreciate, it helps curb imports while redirecting demand towards locally produced goods and services. Unfortunately, as noted earlier, many countries in LAC—particularly the small open economies in Central America and the Caribbean—lack the ability to conduct counter‐cyclical monetary policy.

The main challenge for fiscal policy in LAC is to manage the inevitable fall in tax collection (related to the economic downturn and fall in commodity prices) so as to protect expenditures in education, social security, and infrastructure. These expenditures are necessary to prevent a rise in poverty and inequality and lay the foundations for future growth. In practice, however, the maneuvering room for counter‐cyclical fiscal policy varies considerably among LAC countries (Figure 18). It is greater in countries where: (i) savings were accumulated during good times (Chile is an indisputable leader in this regard); (ii) expenditures are not unduly rigid and can thus accommodate suitable changes in composition; (iii) the debt situation is such that there is scope for prudent borrowing; and (iv) local financial markets are relatively deep. However, given that the shortfall in fiscal revenues is likely to be substantial, a major achievement for LAC would already be to maintain fiscal spending at the initially planned level while protecting vital social and infrastructure programs.

Some countries in the region have already announced fiscal stimulus packages. Nevertheless, there is a great deal of heterogeneity across countries with respect the composition and size of these packages. For instance, some countries have focused predominantly on tax cuts (Brazil), while others have planned to raise infrastructure spending (Mexico, Chile, and Peru). Moreover, some countries are reinforcing their social protection networks (Argentina and Chile) whereas others are providing incentives to non‐traditional exports (Peru). The size of these packages also varies widely, ranging from 0.3 percent of GDP for Brazil to 2.2 percent for Chile. It is difficult however to ascertain the extent to which the announced fiscal stimulus adds to already existing plans or reallocates already budgeted spending. Furthermore, the effectiveness of some the fiscal measures announced (e.g., tax cuts) will depend on the private sector’s willingness to spend.

Closing thoughts

The world is gripped by the broadest, deepest, and most complex crisis since the Great Depression. As the current crisis was originated in the advanced world, its resolution mainly depends on the policies implemented there, particularly on the success of the fiscal stimulus and financial rescue packages. There is still no consensus on the effectiveness of such policies or on when things will bottom out. What is clear is that the crisis is far from over, although the rate of decline seems to be slowing down in some

18

respects. In any case, the global nature of the crisis mutes two channels that have helped emerging markets rebound quickly from past crises—namely, the ability to export to the center and attract foreign direct investment from it. This time around, the real devaluations in LAC will not have those salutary effects on exports as long as the economies of the center and key emerging countries, particularly China, remain in crisis. In all, whether the large economies of the world rebound, remains stagnant, or further deteriorate will greatly determine the periphery’s prospects in the medium term.

While all emerging regions are being hit hard, the impact of the crisis has been very heterogeneous. The crisis is creating havoc in the financial systems of emerging countries where pre‐existing macro‐financial weaknesses were substantial—the most notable case is that of several countries in Eastern Europe. Emerging countries, where such weaknesses were low or non‐existent, are better able to avert a systemic financial crisis at home. LAC, fortunately, appears to be, by and large, in this latter category, thanks to significant institutional and policy improvements in macroeconomic and financial arenas achieved in recent years. These are now affording a number of LAC countries some room for counter‐cyclical policy responses.

However no emerging country, regardless of how well‐prepared or managed, is escaping the recessionary effects of the global crisis. The propagation of these effects is also marked by heterogeneity. For instance, countries that are tightly linked to world trade and highly integrated to the global production chain have experienced more severely the first‐round effects of the crisis on manufacturing production and employment. In contrast, the impact is lagged in countries where growth was supported mainly by domestic demand. Unemployment effects have also tended to be initially smaller in countries with a higher share of labor in the non‐traded sector. Finally, while LAC seems well‐positioned for a fast post‐crisis growth rebound, the recessionary effects of the current crisis threaten to reverse important social gains achieved in recent years. Such a reversal can be highly problematic for democratic‐but‐unequal LAC. The availability of financial resources as well as technical and policy advice from multilateral institutions can be highly relevant in this regard. It can help LAC countries in maintaining their spending plans, or adequately recomposing them, to protect vital infrastructure and social protection programs in the face of falling revenues.

19

Inflation in LACannual variation

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

1981 1994 1996 2007

280

Figure 1 Figure 2

Figure 3 Figure 4

Figure 5 Figure 6

Current account balance in LACas % of GDP

‐6%

‐5%

‐4%

‐3%

‐2%

‐1%

0%

1%

2%

3%

4%

1981 1994 1996 2006

LAC countries with exchange rate flexibility as % of the sample

0%

10%

20%

30%

40%

50%

60%

1981 1994 1996 2007

Loan to Deposits Ratio in LACin %

90%

95%

100%

105%

110%

115%

120%

125%

1981 1994 1996 2007

Total Pubic Debt in LACas % of GDP

25%

27%

29%

31%

33%

35%

37%

39%

1996 2007

Share of Domestic Debt in LAC as % of total public debt

35%

40%

45%

50%

55%

60%

65%

1996 2007

LAC figures usually calculated using data of LAC‐7 countries. Source: WDI and IFS

20

Figure 7 Figure 8

Figure 9 Figure 10

Figure 11 Figure 12

International Reserves in LACas % of GDP

7%

8%

9%

10%

11%

12%

13%

14%

15%

1981 1994 1996 2007

Inflation in selected regionsannual variation

4%

5%

6%

7%

8%

9%

10%

11%

12%

LAC ECA East Asia and Pacific South Asia

Countries with exchange rate flexibility in selected regions as % of the sample

0%

10%

20%

30%

40%

50%

60%


International Reserves in selected regionsas % of short‐term external debt

0%

100%

200%

300%

400%

500%

600%

700%

800%

900%

1000%


Loan to deposit ratio in selected regionsin %

60%

70%

80%

90%

100%

110%

120%

130%


Regional aggregates are calculated as simple averages for 2007. LAC figures usually calculated using data of LAC‐7 countries. Source: WDI and IFS





Regional aggregates are calculated as weighted averages. Source: World Bank DECPG

Industrial ProductionAnnual variation

‐25%

‐20%

‐15%

‐10%

‐5%

0%

5%

10%

15%

20%

Jan‐06

Mar‐06

May‐06

Jul‐0

6

Sep‐06

Nov

‐06

Jan‐07

Mar‐07

May‐07

Jul‐0

7

Sep‐07

Nov

‐07

Jan‐08

Mar‐08

May‐08

Jul‐0

8

Sep‐08

Nov

‐08

Jan‐09

LAC

ECA

East Asia

South Asia

21

Figure 13 Figure 14

Figure 15 Figure 16

Source: Haver Analytics and National Authorities Regional aggregates are calculated as simple averages for 2005. LAC figures usually calculated using data of LAC‐7 countries. Source: WDI.

Regional aggregates are calculated as simple averages for 2005. LAC figures usually calculated using data of LAC‐7 countries. Source: WDI.

Regional aggregates are calculated as simple averages for 2005. LAC figures usually calculated using data of LAC‐7 countries. Source: WDI.

Figure 17 Figure 18

Source: Bloomberg – National Authorities

This index is the weighted average of the relative score of the six different categories. The index as well as each category take values between 0 and 1. Higher values indicate higher constraints on the scope for counter‐cyclical fiscal police. Source: LCRCE Office calculations based on National Authorities data.

Monetary Policy Ratesin %

BRAZIL

COLOMBIA

4

6

8

10

12

14

16

18

Jan‐06

Apr‐06

Jul‐0

6

Oct‐06

Jan‐07

Apr‐07

Jul‐0

7

Oct‐07

Jan‐08

Apr‐08

Jul‐0

8

Oct‐08

Jan‐09

Apr‐09

Monetary Policy Ratesin %

CHILE

MEXICO

PERU

US

0

1

2

3

4

5

6

7

8

9

Jan‐06

Apr‐06

Jul‐0

6

Oct‐06

Jan‐07

Apr‐07

Jul‐0

7

Oct‐07

Jan‐08

Apr‐08

Jul‐0

8

Oct‐08

Jan‐09

Apr‐09

Index of constraints to implement counter‐cyclical fiscal police

0.0

0.2

0.4

0.6

0.8

1.0

Chile Brazi l Colombia Peru Mexico Argentina Ecuador Venezuela

Debt burden Primary deficits Commodity dependence

Expenditure rigidity Financing constraints Financing costs

Imports and Exports of Goods Growth in LACAnnual variation as of Feb‐09

‐80%

‐70%

‐60%

‐50%

‐40%

‐30%

‐20%

‐10%

0%

10%

ARG BRA CHI COL CRI ECU MEX PER GTM SLV PAN

Income inequality in selected regionsGini index

30

35

40

45

50

55

LAC ECA EAP‐7

Persistence to last grade of primary in selected regions total as % of cohort

82

84

86

88

90

92

94

96

98

100

LAC ECA EAP‐7

Mortality rate in selected regionsinfant per 1,000 live births

5

7

9

11

13

15

17

19

21

23

25

LAC ECA EAP‐7

22

23

2. REGULATORY REFORM: INTEGRATING PARADIGMS

Augusto de la Torre and Alain Ize

April 2009*

Abstract

The Subprime crisis resulted from the interplay of information asymmetry and control problems with failures to internalize systemic risk and recognize the implications of Knightian uncertainty. A successful reform of prudential regulation will thus need to integrate more harmoniously the three paradigms of agency, externalities, and mood swings. This is a tall order because each paradigm has different and often inconsistent regulatory implications. Moreover, efforts to address problems under one paradigm can exacerbate problems under the others. To avoid regulatory arbitrage and ensure that externalities are uniformly internalized, all prudentially regulated intermediaries should be subjected to the same capital adequacy requirements, and unregulated intermediaries should be financed only by regulated intermediaries. Reflecting the importance of uncertainty and mood swings, the new regulatory architecture will also need to rely less on market discipline and more on “holistic” supervision, and incorporate countercyclical norms that can be adjusted in light of changing circumstances.

Introduction

As in the case of the other two large financial crises in modern U.S. history, the Great Depression and the Savings & Loan (S&L) crisis, the Subprime crisis was triggered by the inability of financial intermediaries to withstand large macroeconomic price volatility.17 In the Great Depression, banks started failing when the stock market crash induced losses on their equity investments or the loans they had given to investors towards the purchase of stocks. In the S&L crisis, the main trigger was the rise in deposit rates that accompanied the increase in inflation of the late 1970s and the subsequent, sharp tightening of monetary policy. For the Subprime crisis, the trigger was the decline in housing prices. In all three cases, the crisis resulted from a rapidly rising wedge between the underlying value of financial intermediaries’ assets and liabilities, which prevented them from honoring the implicit insurance commitments they had made to their clients. High leverage and liquidity on demand, which limited the size of the buffers available against shocks, made these wedges lethal.

While the proximate triggers of these crises are fairly clear, the most interesting question is why financial intermediaries continue to contract such huge implicit insurance commitments while failing recurrently at honoring them, in the U.S. or elsewhere. Going back to the fundamentals of financial decision making, three possible explanations spring to mind: (i) managers of financial institutions understood the risks they were taking but made the bet because they thought they could capture the upside windfalls and leave the downside risks to others (the agency paradigm); (ii) managers understood the risks they were taking, yet went ahead because they did not internalize the social risks

* World Bank Policy Research Working Paper 4842 17 Throughout this paper we use the term “Subprime crisis” to denote the current, broader crisis of structured securitization and its propagation across financial markets and borders.

24

and costs of their actions (the externalities paradigm); and (iii) managers did not fully understand the risks they were running into; instead, they reacted emotionally to a constantly evolving, uncertain world of rapid financial innovation, with an excess of optimism on the way up and, once unexpected icebergs were spotted on the path, a gripping fear of the unknown on the way down (the mood swings paradigm).

These three paradigms reflect human condition in a nutshell. In the agency paradigm, the better informed are constantly tempted to take advantage of the less informed and, ultimately, the state. By contrast, in the externalities paradigm, financial intermediaries are free agents whose decisions do not necessarily coincide with the public good, or in the case of group coordination failures, with their own good. In the mood swings paradigm, like all market participants, managers of financial institutions have bounded capacity to deal with the genuine uncertainty lying ahead, which is naturally associated with bouts of risk euphoria (“this time around, things are really under control…”) followed by episodes of sudden alarm and deep risk retrenchment.

The next question that naturally comes to mind is why such similarly triggered crises have continued to recur notwithstanding the development over the last eighty years or so of a formidable set of prudential regulations precisely designed to prevent systemic failures. Not only has regulation failed abysmally but attempts to seek a safer regulatory path ahead seem in some cases to have made matters subsequently worse. For example, a key piece of regulatory legislation coming out of the Great Depression was the Glass‐Steagall Act that sought to shield commercial banks from stock market price fluctuations by barring them from investment banking. In turn, the S&L crisis launched the regulatory push towards securitization as a way to pass on to markets much of the risk associated with housing and other longer term finance. Yet, investment banks and securitization are precisely two ingredients at the epicenter of the Subprime crisis.

This paper argues that the failure of regulation largely resulted from a piecemeal approach to reform that looked at one paradigm at a time. In trying to address the central problem under one paradigm, they made the problems under the others worse. Thus, the creation of the Federal Reserve System in 1914 and introduction of deposit insurance after the Great Depression, which set the stage for the public lender‐of‐last‐resort function and were meant to alleviate the instability resulting from recurring runs on the banking system (a problem of externalities), exacerbated the agency‐moral hazard problem. In turn, the strengthening of prudential norms after the S&L crisis, meant to address the acute moral hazard manifestations observed during that crisis, indirectly exacerbated the externalities problem—it drove much of the intermediation outside the prudentially more tightly regulated sphere of commercial banking; once there, participants had less incentives (regulatory‐induced or otherwise) to internalize the externality and hold systemic buffers (liquidity or capital). This last problem of course came back to haunt us in the Subprime crisis.

Moreover, while following this game of tag and run between moral hazard and externalities, regulation missed all along another central suspect: asset bubbles growing and bursting under the impact of rapidly shifting animal spirits. In the Great Depression, the bubble and crash were driven by stock prices; in the Subprime crisis, they were driven by housing prices and the weaknesses of subprime mortgage lending suddenly emerging from the fog. To reconcile theory and facts, the third, missing (or much less developed) paradigm—which puts Knightian uncertainty and the associated mood swings (more than incentive misalignments) at center stage—needs to be recognized and dealt with.

Looking ahead, regulatory reform is largely complicated by the fact that the internal logic of each of the three paradigms leads to different and often inconsistent regulatory implications. In the pure agency paradigm, the only task of the regulator is to mitigate principal‐agent problems by fostering market discipline—mainly through the disclosure of ample, reliable information and by ensuring that financial

25

intermediaries’ “skin in the game” is sufficient to maintain their incentives aligned in the right direction. A properly set regulatory framework should thus eliminate the risk of systemic crises.

By contrast, in the pure externalities paradigm, as markets of their own cannot close the wedge between private and social costs and benefits, the relevant regulation cannot be “market friendly” and the supervisor’s role becomes more central. Moreover, because of the high cost associated with crisis‐proofing, the system’s exposure to some tail risk (akin to “one hundred year floods”) is likely to remain. The ex‐ante crowd coordination and control role of the supervisor needs therefore to double up, if a crisis materializes, with an ex‐post fireman role.

Finally, in the pure mood swings paradigm there are no incentive distortions but market participants cannot fully visualize the dynamic and systemic risk implications of market completion and innovation. Hence, markets are unlikely to provide efficient pricing signals. Unless effective safeguards can be put into place, this severely undermines the Basel II‐type, risk‐based regulatory architecture where every risk can presumably be assessed and translated into an efficient prudential norm. By the same token, the mood swings paradigm boosts the role (and responsibility) of the supervisor, who has to become a scout and a moderator, constantly looking for possible systemic trouble ahead and slowing down the system when uncertainty becomes too large.

To be successful, any reform of prudential regulation will need to integrate the key insights and sidestep the main pitfalls of all three paradigms in a way that limits inconsistencies and maintains a proper balance between financial stability and financial development. Overcoming these tensions will require a dialogue between researchers and policy makers whose perception of the world may be colored by different paradigms. One of the aims of this paper is to contribute to this dialogue.

The paper also proposes a set of basic objectives that any regulatory reform should seek to fulfill in a multi‐paradigm world. Reflecting the main current pitfall of un‐internalized externalities, the reform will need to improve the alignment of incentives by internalizing (at least partially) systemic liquidity risk, thereby lessening the likelihood of crises. However, it should do so in a way that ensures regulatory neutrality and leaves room for prudentially unregulated intermediaries to enter and innovation to thrive. At the same time, reflecting the pitfalls of uncertainty and mood swings, the reform will also need to pay more attention to the risks of financial innovation and rebalance the monitoring roles of markets and supervisors, with the latter acquiring more responsibilities but also more powers. Since in a world of externalities and uncertainty‐driven mood swings even the best regulation and supervision are unlikely to fully eliminate the risk of systemic crises, improving the systemic features of the safety net will continue to be an essential objective.

Consistent with these objectives, we propose: (i) making prudential norms also a function of the maturity structure of the intermediary’s liabilities; (ii) giving prudentially unregulated intermediaries the choice between becoming regulated (with the same capital adequacy requirements as commercial banks) or remaining unregulated subject to the condition of not funding themselves in the capital markets (in other words, prudentially unregulated intermediaries could only borrow from regulated intermediaries);18 (iii) giving the regulator more powers to authorize innovations and norm instruments; (iv) enabling the supervisor (through appropriate statutory powers, accountability, and tools) to play a more “holistic” role by focusing more on the system (its risks, evolution, links, etc.), and to set and calibrate (within bounds) countercyclical prudential requirements depending on changing circumstances, much as the interest rate is calibrated by monetary authorities; and (v) revisiting the

18 The obvious complement to this approach would be to ensure that all the direct and indirect credit risk exposures (on‐ and off‐balance sheet) of the regulated intermediaries are backed by capital (“skin‐in‐the‐game”), at a level which ensures regulatory neutrality.

26

deposit insurance to incorporate systemic risk, rethinking the LOLR as a risk absorber of last resort, and examining the feasibility of pairing them with a systemic insurance subscribed by all financial intermediaries.19

The rest of the paper is organized as follows. Section 2 goes back to the foundations and pitfalls of intermediary‐based finance and briefly retraces the steps and objectives of modern regulation. Sections 3 to 5 present alternative interpretations of the Subprime crisis from the perspective of each of the three paradigms. Section 6 sums up the main failures of regulation and emphasizes the deep contrasts that exist between the three paradigms when one tries to address these failures. Section 7 concludes by laying down a minimum set of basic objectives that would need to be met in order to ensure a more harmonious integration of the three paradigms.

The Foundations of the Current Prudential Framework

Finance seeks to bridge three basic gaps (Chart 1). First, there is an information and control gap (a principal‐agent problem) that reflects fund suppliers’ exposure to the idiosyncratic risks and costs involved in properly screening and monitoring fund users, and enforcing contracts with them. Second, there is a price volatility‐uncertainty gap that reflects fund suppliers’ aversion to becoming exposed to aggregate risks (market‐specific or systemic) over which they have no control. Third, there is a liquidity‐maturity gap that reflects fund suppliers’ “opportunistic” desire to maintain access to their funds and a quick exit option at all times. This third motive responds both to idiosyncratic risks (a quick exit disciplines fund users and mitigates agency problems) and aggregate risks (liquid portfolios and flights to cash mitigate exposure to uncertainty and mood shifts).

Reflecting transaction costs and borrower size, the bridging of these gaps takes on different forms along a continuum that goes from direct market contracting to intermediated contracting (Table 1). At the one extreme, markets bridge the principal‐agent gap through hard public information (arms‐length lending), the liquidity gap through the ability to trade financial contracts easily in deep markets, and the volatility gap through derivative contracts. Asset managers (mutual funds, pension funds, brokers, etc.) cover the middle ground. They help fund suppliers fill the agency gap through expert screening and continuous monitoring (including through direct board room participation), the liquidity gap through pooling, and the volatility gap through diversification. At the other extreme are financial intermediaries that engage in leverage. Commercial banks—the prototypical financial intermediaries—bridge the agency gap through soft private information (relationship lending), debt contracts (a disciplining device), and capital (skin‐in‐the‐game). They absorb the volatility gap and liquidity gap by funding themselves through debt redeemable at par and on demand, respectively, and by absorbing the ensuing risks through capital and liquidity buffers.20 Remarkably, debt and capital (hence leverage) play a key role in intermediaries’ ability to deal with each of the three gaps.

19 Needless to say, to avoid exacerbating cross‐border arbitrage, any such reform would require broad international agreement on the essence of the reforms and their modalities of implementation across borders. 20 In addition, intermediaries, unlike markets, can offer “incomplete” contracts that provide more ex‐post flexibility in adjusting to unforeseen circumstances that can lead to failures in honoring the contracts. See Boot et al. (1993) and Rajan (1998).

27

7

Chart 1. The gaps finance seeks to bridge and the pitfalls it enChart 1. The gaps finance seeks to bridge and the pitfalls it encounterscounters

Risk Gap Response Market Failure

Idiosyncratic

Aggregate

Pick and monitorborrowers

Contract agent

Stay liquidGrab opportunities

Adjust portfolioto risk appetite

Contract insurance

Agencyproblems

Externalities

Mood swings

InformationControl

LiquidityMaturity

VolatilityUncertainty

Table 1. Filling the finance gaps

Gap Channel of finance

Information/Control Liquidity/Maturity Volatility/Uncertainty

Markets Hard information and governance standards

Deep, liquid secondary markets

Derivative markets

Asset Managers Expert screening, direct board participation and monitoring the monitors

Pooling Diversification

Intermediaries Relationship lending, debt and capital (skin in game)

Pooling, demandable debt and capital/liquidity

(buffers)

Diversification, debt and capital (buffer)

By interposing their balance sheet between borrowers (through assets whose underlying value fluctuates with economic conditions) and investors (through liabilities whose value is fixed by contract), financial intermediaries become exposed to systemic risk. They may fail to address this risk in a socially optimal way, reflecting market failures that map all three gaps and paradigms portrayed in this paper. While we will describe these failures more fully in each of the three subsequent sections, a brief preview here will help establish the historic setting and rationale for the current regulatory framework.

Principal‐agent problems give rise to a variety of malfeasance manifestations, most importantly moral hazard.21 Should all depositors be well informed, banks could eliminate moral hazard to the satisfaction

21 The list of malfeasance manifestations with which bankers and other financial intermediaries have been associated over the ages also includes adverse selection, predatory lending, outright fraud and pyramid schemes (Ponzi finance). In this paper, we will broadly lump together all forms of malfeasance within the agency paradigm

28

of depositors by holding capital.22 But the mix of small uninformed depositors and larger, better informed investors can lead to inefficient equilibria in which banks and wholesale investors benefit at the expense of the retail depositors (or their deposit insurance).23 Governance issues compound the problem by superposing additional layers of moral hazard. In particular, bank managers may take decisions that benefit them in the upside but leave the downside mostly to the shareholders or investors.

The opportunistic behavior of fund suppliers or intermediaries faces an externalities problem. Financial intermediaries are exposed to runs by their depositors or lenders, triggered by self‐fulfilling panics or suspicions of intermediary insolvency. Even if they could limit this risk by holding sufficient capital and liquidity, their incentive to do so is limited by the fact that they do not internalize the social costs of a run, i.e., by the existence of externalities.24

The attitude of financial intermediaries (as well as that of other agents) towards price volatility also gives rise to a market failure in that their decisions in the face of uncertainty are influenced by mood swings. They incur bouts of excessive optimism (exuberance) during the upwards phase of financial expansions and excessive pessimism (extreme uncertainty aversion) during contractions. In either case, this compounds price volatility and can lead to sharp deviations from underlying fundamentals (bubbles).

Regulation has been designed to help intermediaries overcome the two first pitfalls, albeit not the third. The current regime rests on three key pillars: (i) prudential norms that seek to align incentives ex‐ante; (ii) an ex‐post safety net (deposit insurance and lender‐of‐last‐resort) aimed at enticing small depositors to join the banking system and forestalling contagious runs on otherwise solvent institutions; and (iii) a “line‐in‐the‐sand” separating the world of the prudentially regulated (mainly commercial banking) from that of the unregulated.

In turn, the line‐in‐the‐sand rests on at least three key arguments. First, regulation is costly and can produce unintended distortions. It can limit innovation and competition, and it needs to be accompanied by good, hence inherently costly, supervision. Second, extending bad oversight (oversight on the cheap) beyond commercial banking can exacerbate moral hazard—it can give poorly regulated intermediaries an undeserved “quality” label (hence an edge in the market place) and an easy scapegoat (blame the regulator if there is a problem). Third, investors outside the realm of the small depositor are but focus primarily on moral hazard because it is the only one that raises “prudential” issues, i.e., issues of risk management. 22 Moral hazard is a reflection of limited liability (limited capital). There is an important literature that questions the need for (and optimality of) capital requirements imposed from the outside. See in particular Kim and Santomero (1988), Berger, Herring, and Szego (1995), Diamond and Rajan (2000), and Allen and Gale (2005). 23 The literature has mostly stressed the “bright side” of wholesale finance, where small depositors free ride on the monitoring and disciplining services of larger investors (see for example Calomiris and Khan, 1991). However, Huang and Ratnovski (2008) recently showed that there is also a “dark side” to wholesale finance. In the presence of a noisy public signal on the state of the bank, wholesale investors may relax their monitoring and rely instead on an early exit as soon as there is any adverse change in the public signal, whether warranted or not. The fact that the smaller investors will stay put (which in their model reflects the presence of deposit insurance) facilitates the exit of the large investors. In this context, it is indeed surprising that the inherent tension within the deposit insurance as currently conceived—meant to cover only small depositors in non systemic events but de facto exposed to systemic losses resulting from early runs by the large depositors—has not received more attention. 24 There is a vast and rapidly expanding literature on the underpinnings of the demand for liquidity and the drivers of liquidity crises. In all cases there is a basic externality at the core of the respective models: liquidity has public good features which liquidity providers cannot fully appropriate. See: Diamond and Dybvig (1983), Holmstrom and Tirole (1998), Diamond and Rajan (2000), and Kahn and Santos (2008).

29

well informed and fully responsible for their investments. As a result, they should monitor adequately the unregulated financial intermediaries, making sure their capital is sufficient to eliminate moral hazard.

Consistent with this line‐in‐the‐sand rationale, only deposit‐taking intermediaries are prudentially fully regulated and supervised under the current regulatory architecture. In exchange, and reflecting their systemic importance, they benefit from a safety net. Other financial intermediaries (and all other capital markets players) neither enjoy the safety net nor are burdened by full‐blown prudential norms. Instead, they are mostly (if not only) subject to market discipline, enhanced by well known securities markets regulations focused on transparency, governance, investor protection, market integrity, etc.

Interestingly, the early history of regulatory intervention, which was marked by the introduction of the safety net, was more closely linked to externalities than to agency problems. However, subsequent regulatory developments came to be dominated by concerns about principal‐agent frictions, particularly moral hazard, which the safety net itself exacerbated. But at this point the logic of the line‐in the‐sand completely missed the obvious facts that, even if free markets take care of principal‐agent problems, they will (nearly by definition) neither internalize externalities nor temper mood swings and price risk appropriately where genuine uncertainty exists. Thus, the regulatory architecture that is in place today became seriously unbalanced.25

In fact, the line‐in‐the‐sand became porous and was widely breached during the build‐up to the Subprime crisis, as highly‐leveraged intermediation developed outside the confines of traditional banking—in what has now become known as the world of “shadow‐banking”—and the safety net had to be eventually sharply expanded, from the regulated to the unregulated.26 The explosive growth of “shadow banking”—driven by the originate‐to‐distribute model, which relied on the securitization of credit risk, off‐balance sheet transactions and vehicles, and fast expansion highly‐leveraged intermediation by investment banks, insurance companies, and hedge funds—has been so well documented elsewhere that it is not necessary to reiterate the details here.27 It is only worth stressing that, by radically expanding the interface between markets and intermediaries, the process brought a variety of new problems and issues. However, the same underlying pitfalls of agency problems, liquidity runs, and mood‐driven cycles reappeared with a vengeance.

In what follows, we interpret the story behind this shift to “shadow banking”—its roots, dynamics, and implications—from the vantage point of each of the three paradigms. As many of the observed features of the Subprime crisis can be consistent with more than one of the three paradigms, attribution is inherently problematic and conclusive proofs are virtually impossible. Hence, the strategy is to work out the internal logic of each paradigm taken by itself, so as to illustrate its potential explanatory power as well as highlight its internal limitations. We will also refer to structural factors such as financial innovation, competition, and regulatory arbitrage when useful to illustrate the inner workings of a particular paradigm, albeit such factors affect all paradigms. On the other hand, although we certainly

25 In modern terms, the prudential framework can be seen as a “line of defense” or “buffer” that partially shields public funds from bank losses by reinforcing market discipline and putting a positive price on the safety net. While focusing on capital, the existing prudential framework clearly goes beyond capital—it includes liquidity requirements, loan‐loss provisioning, fit and proper rules, loan concentration limits, prompt corrective actions, bank failure resolution procedures, etc. 26 Key players in the Subprime meltdown included commercial banks (the prototypical financial intermediaries) and other intermediaries that blossomed outside the banking system and became hyper‐leveraged (mainly investment banks but also insurance companies, hedge funds, as well as commercial banks themselves trespassing into securities markets through off‐balance sheet special investment vehicles—SIVs). 27 See for example Adrian and Shin (2008), Brunnermeier (2008), Gorton (2008), and Greenlaw et al. (2008).

30

recognize the importance of macroeconomic impulses such as the savings glut (and related macroeconomic imbalances) and the “Greenspan factor” (the long period of low interest rates), we restrict our attention to prudential failures because they are the ones that matter for regulatory reform.

The Agency Paradigm

The moral hazard‐agency story of the Subprime crisis is arguably the most popular.28 It posits that incentive distortions arising from unchecked principal‐agent problems (the heads‐I‐win‐tails‐you‐lose syndrome) are the source of trouble, inducing market participants to either pass on risks deceptively to the less informed or take on too much risk themselves with the expectation of capturing the upside or exiting on time and leaving the downside with someone else. The perversion of incentives can happen at one or several points of the credit chain between the borrower and ultimate investor, passing through the various intermediate links.

However, for moral hazard to start driving the show, it must be the case that the expected upside benefits come to dominate the expected downside costs (i.e., losing one’s capital or reputation). This can occur under two plausible scenarios: (i) an innovation (perhaps facilitated by deregulation) opens a world of new opportunities (the upside widens), or (ii) a macro systemic shock suddenly wipes out a large part of the intermediaries’ capital (the downside shrinks).29 Indeed, one can argue that in the case of the Subprime crisis it was the discovery of new instruments and intermediation schemes (securitization and shadow‐banking) which set the process in motion.30 The expansion of upside opportunities led to a moral hazard‐induced under‐pricing of risk, encouraging participants to make the bet and take the plunge.31 This process, which Basel I regulation encouraged, can be explained in part by regulatory arbitrage.32 However, poor regulation (that did not sufficiently align the incentives of principals and agents, whether the risk was acquired off or on balance sheet) can no doubt also be blamed.

Indeed, the build‐up phase of the crisis provided plenty of opportunities for all sorts of principal‐agent problems to expand and deepen. The multiplication of actors (borrowers, loan originators, servicers, securitization arrangers, rating agencies, asset managers, final investors) involved in the originate‐to‐distribute model not only reflected the increased sophistication and complexity of intermediation but also boosted the scope for accompanying frictions, including moral hazard, but also predatory lending, 28 See for example Caprio et al. (2008) and Calomiris (2008). 29 The sudden opening of profitable new business opportunities that set the cycle’s upswing into motion is what Fisher (1933) called a “displacement”. 30 By contrast, the S&L crisis can be viewed as driven by deregulation and the rise in interest rates that effectively de‐capitalized the system (a reduction of downside risks), unleashing the subsequent rounds of “betting for survival”. The process was exacerbated by the lack of fair value accounting (which aggravated information asymmetry problems while allowing insolvent institutions to continue operating normally) and generous regulatory forbearance. 31 There is a body of literature emphasizing moral hazard‐caused deviations of asset prices from their fundamental values. See for example Allen and Gale (1998). While these deviations may be interpreted as “bubbles”, the underlying models are typically static. 32 Basel I prudential standards encouraged securitization through differential risk weights (a mortgage held on a bank’s balance sheet is charged with a 50 percent risk weight, against only 20 percent if securitized). At the same time, although Basel I did incorporate some off‐balance sheet commitments, conversion factors limited their impact on capital. Banks could also circumvent regulation through innovations such as tranching and indirect credit enhancements, the use of the trading book rather than the banking book, and other balance sheet adjustments. See Tarullo (2008).

31

mortgage fraud, adverse selection, and other principal‐agent problems.33 The widespread preference of unregulated intermediaries to lever up on the basis of mainly short‐term funds can also be interpreted as driven by moral hazard. Managers (at least some of them and particularly, but not only, asset managers) also seemed to have danced eagerly to the moral hazard tune. While enjoying the high returns of the good times, they let their shareholders and investors deal with the losses in the bad times under the convenient excuse that everybody shared the same miseries.34

A good case can also be made that the state promoted moral hazard on the way up. Some argue, for example, that the widespread subsidies and guarantees provided to the house financing sector in an effort to boost access (exacerbated by Fannie Mae’s and Freddie Mac’s “quasi‐mandated” foray into the sub‐prime sector) can be blamed for launching the ball and boosting its moral hazard momentum once in play.35 The failure to control the build‐up phase can then be attributed to the regulator’s inability to win the cat‐and‐mouse game of regulatory arbitrage. Banks managed to stay on top by swiftly moving to the shadow‐banking world, with regulators hardly able to keep up.36 The extreme fragmentation and overlapping mandates of agencies that comprise the U.S. supervisory system was of course the final blow. Had the regulators been aware and statutorily able to do something, the necessary coordination was just too much to handle.

The agency paradigm is self‐contained in that it carries the seeds of its own demise. Once participants have taken the plunge, they have little or nothing more to lose by taking on additional risk. A dynamic could be thus unleashed that pushed bets higher and higher as less risky investment opportunities became gradually exhausted. Indeed, there is good evidence that risk taking by mortgage originators mushroomed over the cycle as less and less creditworthy borrowers were gradually let in.37 Such dynamics should be naturally unstable and eventually collapse on their own weight.38

Once the crisis hit, the liberal unfolding of the safety net under the gun of systemic contagion (lender‐of‐last‐resort by the Fed and bail outs by the Treasury) clearly validated any moral hazard incentives that might have led to the crisis. In particular, it facilitated the early exit of at least some of the well‐informed large investors, rewarding those who had lent imprudently (and allegedly knowingly). Another moral

33 Ashcraft and Schuerman (2008) analyze the “seven deadly frictions of asymmetric information” that unfolded with a vengeance in the originate‐to‐distribute world. 34 The managers masquerading their excessive tail‐risk taking as clever investment moves are dubbed by Rajan (2008a) as “fake‐alphas”. The perfect excuse for the bad times is defined by Calomiris (2008) as “plausible deniability”. Reflecting their greater concern for the short‐term bottom line than for the potential longer term risks (perhaps reflecting mostly backward looking compensation schemes), operational managers seem to have paid insufficient attention to the concerns of risk managers. On issues of managerial compensation and the scope for managerial “abuse”, see also Dewatripont and Tirole (1994), Brunnermeier (2008), and Gorton and Winston (2008). 35 Fannie Mae and Freddie Mac—the giant mortgage government‐sponsored enterprises—could meet their mandated social housing goals by buying eligible subprime mortgages. For a summary of public policy actions to promote housing finance see Calomiris (2008). 36 For good narratives along these lines, see Caprio et al. (2008), and Calomiris (2008). 37 On the propensity for increased risk taking, see Dell’ Ariccia et al. (2008), and Keys et al. (2008). Leamer (2008) goes further to argue that there was a gradual shift from hedge finance to speculative finance and then to outright Ponzi finance during the recent housing cycle. 38 In the end, the trigger for the crisis under the pure agency paradigm should still be a stochastic event (moral hazard would cease to operate if there was no longer a possible upside, as unlikely as it might be). That event, however, can be so small that it ceases to be relevant.

32

hazard booster in the ex‐post unfolding of the safety net was that, for the most part, large institutions were not closed and, perhaps more importantly, managers were allowed to stay in charge.39

In sum, the moral hazard tune does ring true in many respects. However, important questions remain. First, for shadow banking to be explainable by moral hazard, it must have allowed commercial banks to pile on more risk. However, whether, on balance, commercial banks ended up shedding or piling risk through securitization is not entirely clear, albeit some evidence seems to militate in favor of the latter.40 As intended by the early promoters of securitization, the sale of mortgage‐backed securities to investment banks should in and of itself, have reduced (not increased) commercial banks’ riskiness. In reality, however, much of the risk was never really divested away. Instead, commercial banks repurchased good chunks of the instruments they sold, for reputational as well as business continuity reasons, and remained committed to support investment banks through their back‐stop liquidity facilities (they were lenders of first resort to capital markets players). Moreover, they generally retained the more risky assets (or the more risky tranches) while shedding away the less risky ones.41 At the same time, they moved down the credit market to take on new and arguably higher risks associated with consumer, mortgage, and SME lending. They also accumulated more risk by engaging in widespread rating arbitrage (shopping for the most favorable ratings).42

Moreover, even if one believes that banks did accumulate more risk, it does not necessarily follow that this was induced by moral hazard. Indeed, commercial banks could have genuinely bought the risk under the presumption that it was safe for them to store it (they perceived the regulations to be too tight and their capital more than enough to cover the associated risks). Under this interpretation, to which we will come back under the externalities paradigm, commercial banks ventured into new markets and new instruments simply because they had a comparative advantage in doing so.

Perhaps more importantly, the main piece of the puzzle that does not quite fit this paradigm is the blatant asymmetry between the smart ones who are alleged to have consciously caused havoc and all the rest of the financial market participants who were not paying attention. In particular, why did the markets (informed investors and shareholders) fail to discipline financial intermediaries? In the end, many investors surely got it wrong and lost tons of money; a multitude of bank shareholders got wiped out; and many managers likely have had second thoughts about having played so eagerly the alpha card. In this context, supervisors must surely also be thinking that it is unfair to treat them as if they were the only ones asleep at the wheel.

The moral hazard story inherently requires a strong agency problem, caused either by high enforcement costs or deep crevices of information asymmetry. Arguably, principals (shareholders and large investors) lacked the incentives or regulatory tools that might have helped them align the actions of their agents (managers). However, it is difficult to believe that principals would not have taken early disciplinary action, if only by voting with their feet, had they really understood the risks agents were taking. Thus, setting aside the problem faced by the regulators as regards the growth of the unregulated sector, enforcement costs are not really consistent with the lengthy gestation of the build‐up to the crisis nor with the short‐term nature of the financing that supported that build‐up. A better case can perhaps be made for the intensification of information asymmetry resulting from the opacity, complexity, and 39 Curiously, while deposit insurance fully protected the small depositor, much less was done to protect the small borrower (that has been an important asymmetry as regards consumer protection). 40 Rajan (2005) presents evidence that suggests some increase in overall banking risk, as indicators of banks’ distance to default have not risen in many developed countries and bank earnings variability has not fallen in the United States. Instead, the risk premium implicit in bank stocks appears to have risen. 41 See for example Ambrose et al. (2005). 42 See Brunnermeier (2008).

33

interconnectedness of the new age housing finance market.43 Arguably, this could have provided a cover under which the ones at the top of the pack could have hidden their operations. Yet, it still remains hard to fathom that this “scam” would take place for such a long period, during which the asymmetry between those who were “in” and those who were “out” would linger unabated, and that this would happen in a market place where tips, news, and information are produced by the ton every minute.

The Externalities Paradigm

Externalities, the mirror image of individual opportunism, clearly play a major role in the collapsing phase of any crisis. Seeking to save oneself by running for the exits puts the others at increased risk of a major meltdown with extreme social costs, thereby exacerbating the violence of the downturn. But externalities also play a key role during the build up stage, making the system inherently more fragile. The failure to internalize the costs of a systemic crisis is at the core of the insufficient demand for prudential buffers, including in particular liquidity, which has features of a public good. Externalities can also induce bubble‐type deviations of asset prices from their fundamentals.44 They can also result in under‐production of information and monitoring (free‐riding) and over‐extension of credit during upswings, over‐contraction during downswings (in both cases, the marginal lender can “sour the market”, increasing the vulnerability of other lenders to a default). Last but not least, coordination failures (a form of un‐internalized externalities) can also play an important role in lengthening and aggravating the upwards phase of the cycle. Market participants may know it is in their best interest to prevent an asset bubble yet fail to do so because doing the right thing would only be optimal if everybody else in the group did it too. Supervisors, both across agencies and across countries, are similarly vulnerable to such coordination failures. For example, tightening regulation in isolation has a high cost, as business will quickly flow to the less regulated sectors or countries.

The lack of sufficient buffers was indeed at the core of the severity of the collapse. As in the case of traditional banking, shadow banking was financed mostly through short‐term obligations (and largely perceived to be redeemable at par), much of it through overnight repos. The potential for a bank‐type run was therefore there from the outset. But two additional factors made for a much more explosive situation. First, the financing came mainly from ready‐to‐run wholesale investors, thereby introducing a new, more unstable layer to the intermediation process. Second, the capital and liquidity buffers held by most shadow‐banking intermediaries to protect their short‐term liabilities from price fluctuations in the final asset (housing) were much smaller than in traditional commercial banking. This reflected the high leverage of self‐standing investment banks and (to a less extent) hedge funds, as well as the lack of capital put in by the final borrowers who benefited from high loan‐to‐value ratios and second mortgages. Thus, as documented elsewhere in detail, once a tail‐risk event materialized and pressures

43 See for example Gorton (2008). 44 Because individual agents do not internalize the general equilibrium impact on asset prices of fire sales under financial distress, they can bid up the price of these assets in excess of their socially optimal value. Lorenzoni (2007) develops a model along these lines and shows that competitive financial contracts can result in excessive borrowing ex‐ante and excessive volatility ex‐post. As in Holmstrom and Tirole (1998), agents cannot insure themselves against aggregate liquidity shocks due to a limited ability to commit to future repayments (this in turn reflects agency frictions). Korinek (2008) develops a paper along the same lines but applied to capital flows rather than domestic intermediation (in his model, agents borrow too much because they do not internalize the potential impact of an exchange rate move on a systemically‐induced need for sudden repayment).

34

to sell started to build up, the devastating downward spiral quickly dried up liquidity and brought markets to a standstill.45

In the shadow banking world, the externality pitfall of traditional banking operated with a vengeance, as everyone counted on everyone else’s for support but no one adequately internalized the systemic risks of such cross‐support. Investment banks counted on commercial banks (both for liquidity and for asset repurchases);46 commercial banks counted on market liquidity (why hold liquid backing against assets which you can sell at any time in the market place?); and leveraged intermediaries counted on credit default swaps and other forms of insurance issued by other leveraged institutions. In the process, a great fallacy of composition developed—leading market players (and supervisors) wrongly to believe that risk protections at the individual level would add up to systemic risk protection. Yet, markets for individual risk protection instruments could only continue functioning if some intermediary was willing to continue “making the market”.47

The extreme systemic fragility of such interconnectedness has by now become obvious.48 By unloading (selling) risk—for example through credit default swaps—to other financial institutions such as insurance companies, intermediaries further intensified the negative systemic externalities.49 Such transactions might have reduced the exposure of institutions individually but increased the exposure of the system as a whole. Yet, this move was openly encouraged by regulators (insured assets had a low or zero risk weight), who viewed it as a way to reinforce market discipline (again, an example where moral hazard and externality containment directly collided). The possible systemic costs of trading credit derivatives over the counter (without a central clearing counterparty or protocols for multilateral netting), rather than on an exchange, were not internalized either.

While the fragility brought about by externalities has received much attention in the crisis literature, an equally important consequence of un‐internalized externalities that has received much less attention is their implication for regulatory arbitrage. As in the case of moral hazard, the growth of shadow banking can also be explained as externality‐induced incentives to circumvent regulation. The key difference is one of intent. From an externality viewpoint, intermediaries were “doing nothing wrong” by finding new ways to take on more risk. Instead of seeking to take one‐sided bets with someone else’s money, as in the agency paradigm, the intermediaries engaged in regulatory arbitrage under the externalities parading were just searching for ways to match more closely their risk taking with their risk appetite, and they were doing so in a way which, from their own (limited) perspective, was sufficiently safe. From their individual viewpoint, regulations were “unnecessarily binding”.

In this sense, the intent of the Glass‐Steagall Act—to shift risk away from regulated intermediaries to capital markets and unregulated intermediaries—was fundamentally misguided. While it could have solved the agency problem (by shifting risks to the land of the well informed) if it had been done cleanly enough (i.e., without dragging the banking system into the mud and the safety net over the line‐in‐the‐sand), it exacerbated the externalities problem. Well‐informed investors can monitor the intermediaries to make sure they do not “cheat them” (play the moral hazard card). However, they have no incentives

45 See Greenlaw et al. (2008), Adrian and Shin (2007 and 2008), and Brunnermeier (2008). 46 Yet, there were no capital charges for such “reputational” credit lines (see Brunnermeier, 2008). 47 The linkages between securities market liquidity and funding liquidity, and the resulting increased scope for liquidity spirals are analyzed by Brunnermeier and Pedersen (2008). 48 The fact that most intermediaries traveled along the same path on both the way up and the way down, driven by similar incentives and risk management models, further boosted the systemic impact of these externalities. See Brunnermeier (2008). 49 Allen and Gale (2005) discuss the possible implications for systemic risk of such transfers.

35

to “internalize” the liquidity and other externalities.50 Instead, their incentive is to play it safe by investing very short and running at the first signal of trouble and to increase leverage by as much as is privately (not socially) optimal.

To be sure, reflecting regulatory shortcomings in the internalization of systemic liquidity risk (see below), incentives were not much better aligned for the regulated intermediaries. Nonetheless, capital in the regulated sector substantially exceeded that in the unregulated sector, reflecting systemic concerns of regulators for the commercial banking sector.51 Thus, the side‐by‐side existence of a regulated sector—where systemic concerns were partially factored in—and an unregulated sector—where externalities were not at all internalized—created a wedge in returns between the two worlds, giving rise to a fundamentally unstable construct. Investors left in droves the regulated intermediaries to join the world of the less regulated, highly leveraged and short funded intermediaries, rapidly raising their relative size and boosting systemic risk in the process. Moreover, because it involved sophisticated and unsophisticated investors, the exodus spread moral hazard throughout the presumably moral‐hazard free unregulated (or less regulated) world.

The resulting competitive pressures on commercial banks ultimately motivated the repeal of the Glass‐Steagall Act.52 However, by challenging commercial banks to compete head‐on with the blown‐up investment banks and on their turf, the repeal induced the former to find creative ways to shed their regulatory burden outside their balance sheet. Thus, oddly enough, the Glass‐Steagall Act resulted in a one‐two punch on the soundness of financial intermediaries. Its introduction boosted systemic risk outside commercial banking. Once this was done, its repeal boosted systemic risk within it.

As in the agency paradigm, supervisors come out severely bruised. They did not realize that their own well‐meaning regulation was setting into motion a deadly process of regulatory arbitrage that shifted intermediation to a field where inducements to internalize externalities were weaker or nonexistent, thereby contributing to asset over‐pricing and spreading liquidity risk all over the financial system. And even when supervisors caught up, they were unable to do much because in the cat‐and‐mouse game of regulatory arbitrage the mouse had trespassed over the line‐in‐the‐sand to a territory where prudential regulation was not unreasonably reluctant to enter. Investment banks, hedge funds, and the like were thus simply left out of reach.53 Moreover, even within the regulated world, the Basel‐inspired wave of prudential regulation focused little on liquidity. And when the norms addressed liquidity issues, they did so from a purely idiosyncratic perspective.54 To his defense, however, the externality‐conscious

50 A similar point was made by Bernanke (2006). 51 For example, investment banks’ leverage of around 25—compared to commercial banks’ leverage of only about 10—gave the former an obvious advantage. Although the SEC, as lead regulator, applied to investment banks the same Basel capital rules as for commercial banks, the differences in leverages resulted from the much lower capital requirements on “trading books” than on “banking books” and the fact that limits on gross leverage ratios only applied to commercial banks. 52 Pushed by the forces of competition and deregulation, commercial and investment banks seemed to have met somewhere in the “regulatory middle”. As the repeal of the Glass‐Steagall Act allowed commercial banks to encroach more directly on investment banks’ traditional fee‐based business, the former took on more fees in order to offset losses in intermediation margins. Also, and partly as a result of the deregulation of commissions for stock trading in the 1970s (that allowed low‐cost brokers to encroach on investment banks’ brokerage activities), self‐standing investment banks gradually shed their fee‐based business in favor of a highly‐leveraged margin‐based business. See Eichengreen (2008). 53 The move towards consolidated supervision of financial conglomerates was as far as prudential regulators were willing to extend their reach to protect the core banking system from capital market risks. 54 For example, liquidity norms generally advocate minimum ratio of liquid assets to liabilities to limit maturity mismatches. But this is simply not good enough from a systemic viewpoint where even short‐maturity assets can

36

supervisor may argue that systemic events such as the Subprime crisis are akin to “one‐hundred year floods”. They are too rare and unpredictable to be usefully internalized in prudential regulations. The social cost of doing so (note here the italics) would simply exceed the social benefits. Hence, a better option is to have a prompt correction regime and an efficient public rescue system.

The missing piece in this paradigm, which is otherwise convincing enough, relates to its dynamics. To be sure, the lack of sufficient internalization of systemic risks can lead as easily as moral hazard‐based incentives to a more fragile and vulnerable system. Yet, unlike in the agency case, the externalities paradigm in and of itself lacks inherent dynamics that gradually increase the precariousness of the equilibrium over time and eventually bring the system so close to the edge that the tiniest exogenous shock would throw it over.55 In the pure externalities paradigm, intermediaries continue to “manage” their risk, adjusting it to what is privately optimal and then just staying there. The large shock that eventually sent the financial system over the edge must have therefore come out of “left field”—an exogenous act of god, whose probability was independent of the degree of vulnerability of the system. However, as far as one can see, there was no such shock in the case of the Subprime crisis.

One could argue that, instead of an exogenous shock, the engine driving the financial system to its eventual collapse was a real sector‐driven business cycle. However, prudential norms are supposedly designed to allow financial systems to navigate unscathed through the ups and downs of the regular business cycle. Hence, this could only be a satisfactory explanation if the magnitude of the downturn was unprecedented and truly unexpected. Again, however, this does not seem likely. The financial crisis was unleashed in full force much before there was a marked real sector decline, with causality going mostly in the opposite direction.

Alternatively, one could tease out some endogenous dynamics within the externalities paradigm by associating the externalities driving the system to a prisoner’s dilemma. What market participants do individually (i.e., join the feast in the boom and the stampede in the bust) is clearly harmful to themselves and the group, but each participant would stop only if everyone else in the group did the same. That this type of coordination failure can generate some cyclical fluctuation stands to reason.56 That it can lead to a catastrophic and expected systemic collapse is more difficult to accept. In the absence of a non‐externalities related factor—either moral hazard (perhaps boosted by managers’ short incentive horizon) or a truly unexpected unfolding of events (a much bigger or much sooner meltdown than anyone could reasonably have expected)—one would think that at some point the downside risk to each individual participant of remaining in the game should dominate the upside risk. At that point, self‐preservation should de facto force coordination, keeping the group some distance away from the edge of the cliff.

become illiquid. Norms have failed to focus on systemic rollover risk, which is at the core of intermediaries’ vulnerability to runs. 55 Some recent analysis of the unfolding of the Subprime crisis stresses the extreme market fragility resulting from an unexpected market realignment in a context where all the large traders have similar underlying risk models and objectives (Khandani and Lo, 2008). However, it is not obvious that traders would have continued to operate so close to the edge if they had understood the true fragility of the environment in which they were operating and the huge potential costs of a meltdown. 56 For example, Abreu and Brunnermeier (2003) develop a model in which asset bubbles persist despite the presence of rational arbitrageurs because the latter cannot temporarily coordinate their selling strategies due to a dispersion of opinions.

37

The Mood Swings Paradigm

The starting point of the mood swings paradigm is the endogeneity of financial innovation within a broad process of financial development. The shift from traditional banking to shadow banking can be interpreted as the natural evolution of a rapidly deepening financial system in which markets and intermediaries increasingly complemented each other.57 Banks commoditized credit risk through the originate‐to‐distribute model and retained some credit risk to overcome agency problems.58 At the same time, they used their ability to provide first resort liquidity to help markets overcome the remaining liquidity gap associated with the yet nascent and still overly heterogeneous instruments. The pressures of competition, boosted by the steady entry and rapid growth of unregulated (or less regulated) brokers and intermediaries (particularly investment banks), were clearly at the heart of such a remarkable process of financial deepening and market completion.

However, the creation of new instruments and forms of intermediation went faster than the ability of market participants and supervisors to fully comprehend their implications and handle the risks and uncertainty associated with such a rapidly changing world. The opacity, complexity, and hidden interconnectedness of the Subprime world can thus be seen in the mood swings paradigm as bad side effects of an innovative process, but side effects that were either not intended or, if intended, not necessarily maliciously pursued.59 The inability to think through the potential systemic implications and fragilities of the new universe was the fundamental and critical failure.

This problem was compounded by a failure to fully comprehend the links between financial sector dynamics and the underlying asset price dynamics, and to adequately understand the feedback loop between rising asset prices and expanding credit. The possibility of a large and nation‐wide synchronized decline in housing prices (and the devastating implications this would have for the risk correlation assumptions underlying the presumed safety of credit default protections) was unthinkable because it had never happened since the Great Depression.60 Moreover, when delinquency rates on mortgages started to rise during the mini‐recession of 2002, the losses on mortgages were minimal because the housing market continued to boom.61 From this perspective, falling housing prices and their implications for the housing finance market appear not as “tail risk” but as a “black swan” event, a new reality that could not be anticipated from historical series.62

Faced with the world of the new and unknown, market participants involved in the Subprime process no longer had a steady frame of reference. On the way up, they found themselves in a truly new and wonderful territory which fueled a mood of optimism and exuberance. This was reinforced by the decline in observed macro‐financial volatility, predictable pricing and deep market liquidity, which further fed risk appetites and gave rise to pro‐cyclical leveraging.63 The low volatility environment not

57 Through securitization, markets benefit from the screening done by intermediaries and the latter benefit from the more efficient parceling and tailoring of risk carried out through the markets. See Gorton and Winston (2002), and Song and Thakor (2008). 58 This was certainly not a minor achievement—it involved standardizing the credit risk screening (through scoring and rating), breaking it up (through stripping and tranching) and dispersing it (by selling it to a wider base of investors and spreading it around through a new breed of credit risk derivatives). 59 Information got lost through the “chain of complexity” and banks became exposed in the process to heavy “pipeline risk”. See Brunnermeier (2008) and Gorton (2008). 60 See Gorton (2008) and Coval, Jurek, and Stafford (2008). 61 See Calomiris (2008). 62 See Taleb (2007). 63 Unlike commercial banks that targeted a constant leverage throughout the cycle, investment banks’ leverage was heavily pro‐cyclical. See Adrian and Shin (2007 and 2008).

38

only had the immediate mechanical effect of reducing values at risk but also, the more it persisted, the more it fed the feeling that “this time around, things are different and the good times are here to stay”. New forms of macro‐financial management and oversight, including the ever more sophisticated risk modeling, widespread divestment of risk through risk derivatives, and more effective and successful monetary management, were all major contributors to this optimistic picture.64 Feelings such as “everything is being taken care of”, “good men are now in charge”, and “systemic volatility is a memory of the past which has now been vanquished even by the Mexicos and Brazils of this world” became so prevalent that few really questioned them.

On the way down, the brutal downward swing in the prevalent market mood also fed the collapse. A significant dissonance would be enough to initiate the mood swing. In the Subprime crisis, the swing was arguably triggered when the CBX credit swap index on sub‐prime based instruments started going south, colliding with the still rosy assessments of the rating agencies.65 As long as there was widespread market agreement on a price vector, ensuring that instruments could continue to be unloaded on short notice, markets could go on functioning unperturbed (whether prices actually matched fundamentals was not that important as long as they were uncontested). However, by questioning the uniformity of market assessments, the drop in the CBX index suddenly raised the specter of “hidden icebergs lying ahead”. From euphoria, the mood shifted into acute Knightian uncertainty, where risk aversion swelled, driven by the fear of the unknown.66 The frenzied recoiling of investors was compounded by general market opacity—including the knowledge that intermediaries were deeply interconnected coupled with utter ignorance on the nature and specific details of this interconnectedness. Opacity thus intensified the massive sell out of securities and simultaneous flight to cash, with the resulting market collapse and evaporation of price signals further accentuating the downward spiral.67

In this paradigm, well‐meaning public policy also played a central role, both on the way up and on the way down. On the way up, a key and justifiable role for policy is to promote market completion within an evolutionary financial development process.68 Indeed, the set of policies designed to promote housing finance by jump‐starting the markets for new instruments such as securitization through guarantees and subsidies can be viewed as sowing the earliest seeds of the crisis. The Subprime crisis grew, in effect, in the “shadow” of the guaranteed world of Fannie Mae and Freddie Mac. While such policies can help overcome natural impediments to market development—particularly where collective action is difficult and network and scale effects are significant—they can also help promote the illusion that risk has been reduced to a point where it ceases to be a predominant concern. Public intervention also played (and continues to do so) a critical role on the way down. In a world of uncertainty and acute swings in risk aversion, only the State has the shoulders needed to function as the risk‐absorber‐of‐last‐resort during episodes of acute, systemic failure.69 In this view, the ex‐post unfolding of unprecedented

64 As Greenspan (1998) famously declared, the “management of systemic risk is properly the job of central banks” and “banks should not be required to hold capital against the possibility of an overall financial breakdown”. 65 See Gorton (2008). 66 Uncertainty aversion came on top of (and interacted with) increased volatility. See Brunnermeier (2008). 67 Panics end when information recomposes and becomes available. Intermediary‐based finance is in this sense much more vulnerable than market‐based finance, since prices are less likely to vanish in markets that do not rely on market‐making institutions. 68 A theoretical justification for government intervention in a context of incomplete markets can be found in Geneakoplos and Polemarchakis (1986). Gale (2004) shows that in the presence of incomplete markets there exists an implicit pecuniary externality that generally requires the imposition of capital requirements. 69 The seminal contribution as regards the role of the State as the residual absorber of risk is that of Arrow and Lind (1970). See also Caballero (2009) for a recent reinterpretation of the insurance role of the State in systemic crisis conditions. An intriguing argument can however also be made that instead of spreading risk over taxpayers

39

Fed’s lender‐of‐last‐resort activity and the U.S. Treasury’s bail out operations can be interpreted as a way to drain away from the system sufficient systemic risk so as to allow markets to spring back up to life and intermediaries to continue operating.

All in all, the mood swings paradigm presents a more rounded overall story than the other two paradigms, and a story with far‐reaching implications at that. Unlike the agency paradigm, it does not require a gigantic and unyielding asymmetry of information between market participants that are in‐the‐know and those that are out. Rather, it is a democratic paradigm where everybody was fooled. And unlike the externalities paradigm, it does not require a vengeful god to intervene exogenously with tail‐risk events to unleash the dynamics of a downward spiral. Instead, it has its own fully endogenous dynamics, with favorable returns and optimism feeding each other on the way up, adverse returns and pessimism on the way down. The dynamics are akin to Schumpeter’s creative destruction, where cycles are a natural part of the evolutionary process. However, unlike the traditional Schumpeterian process, where some do well while others perish at every point in the cycle, the dynamics in the mood swings paradigm are more like “Schumpeter on steroids”, as financial innovation cycles can have a devastating systemic impact because everyone follows the same path, up the bubble and down the abyss.

The mood swings paradigm, however, is not free of puzzles and difficulties. In particular, uncertainty‐driven mood swings are easy to invoke but harder to model.70 To be sure, one would expect rationality (even if bounded) and path dependence to constrain feasible outcomes. However, unlike incentive distortions under the moral hazard and externalities paradigms, which are firmly grounded in traditional economic theory, modeling mood shifts may require some departure from orthodox theory.71 In any event, it is also rather surprising that market participants were seemingly oblivious to the risks underlying the process of financial innovation. Did such obliviousness simply reflect a difficulty to look outside the box and connect the dots? Did such difficulty reflect the fact that markets do not reward

(current and future), risk might be more efficiently spread over existing debt holders by using debt equity swaps as an alternative to unconditional bail outs (see Veronesi and Zingales, 2008). 70 The importance of mood swings for financial bubbles and panics has been widely recognized. It finds its roots in Keynes’ animal spirits and Hyman Minsky’s writings on financial crises (see Minsky, 1975). More recently, it was popularized by Kindleberger (1996) and Shiller (2006). While many attempts have been made to model mood driven‐cycles within the traditional world of rational expectations with full information (see the seminal contribution of Azariadis, 1981), the conditions for such rational bubbles to exist have been shown to be rather limited (Santos and Woodford, 1997). However, moods play a much more important role once one assumes problems with the information (imprecision or uncertainty) or the way one deals with it, which, in turn, may (or may not) require abandoning the assumption of full rationality. Epstein and Wang (1994), and more recently Fostel and Geneakoplos (2008), showed that multiple priors can lead to models where beliefs influence asset prices in a fully rational world. In addition, Geweke (2001) and Weitzman (2007) showed that, when there is too much uncertainty, fully rational human behavior may not conform to the precepts of traditional economic theory as defined by the standard expected‐utility framework. Abandoning the assumption of full rationality opens up the scope for innate biases in the way economic agents process information and make decisions (see the recent surveys of behavioral finance in Barberis and Thaler, 2003, and Della Vigna, 2007). Attempts to explore the implications of such limitations for finance and credit cycles are making some headway. For example, Shleifer and Vishny (1997) showed that inefficient asset pricing driven by noise traders can persist despite the presence of rational arbitrageurs. Lo (2004) proposed an evolutionary approach to economic interactions. De Grauwe (2008) showed that it is possible to generate endogenous cycles when agents use simple heuristic rules to interpret the dynamics of a model they do not fully comprehend. 71 For example, they may be associated with biased perceptions under bounded rationality, or shifts in risk appetite under rational expectations, a non trivial distinction since one would expect risk pricing to be biased in the first case but not in the second. Another key modeling difficulty is the extent to which uncertainty goes beyond the underlying environment to include group behavior.

40

systemic risk gazing (a theme to which we will come back in the next section)? Or was something more sinister at play, either moral hazard or non internalized externalities? In particular, absent externalities, one wonders whether uncertainty alone could pack so much punch, particularly on the way down.

In sum, the overall picture one gets from systematically reviewing the three paradigms is that they all provide broadly plausible stories. Hence, they must all contain important grains of truth. Moreover, the paradigms seem to interact and feedback on each other in complex ways, one triggering the other or becoming more predominant at different stages of the cycle. Hence, a fully rounded story—one that does not leave key questions unanswered and fully accounts for the complexity of real life—requires combining the paradigms. However, multi‐dimensionality makes the challenges of policy reform that much more difficult. To these issues we now turn.

Paradigms and Regulation

In this section, we will briefly summarize what we perceive to have been the main failures of regulation and illustrate in broad terms how policy prescriptions to fix them will often not be independent of the paradigm of choice.

The great failures of prudential regulation evidenced by the Subprime crisis can be classified into: i) failures of scope; ii) failures of focus; and iii) failures of dynamics. Take the failures of scope first. The “line‐in‐the‐sand” philosophy simply did not work. The prevailing thinking was that opening a wide room for unregulated intermediaries to thrive was of little consequence to systemic stability. Knowledgeable investors would maintain them in line. Moreover, they were too small to be systemically important. Both assumptions turned out to be deadly wrong. The failure to internalize externalities in the unregulated world created a bias in favor of unregulated intermediaries that drew in unsophisticated investors in droves and made them grow explosively. In turn, this competitive bias induced banks to elude regulation by pushing risk outside their balance sheet and turn a somewhat blind eye to the risks taken by their borrowers. Thus, not only was risk not adequately internalized ex‐ante but also prudentially unregulated (or less regulated) intermediaries quickly grew to the point where they became systemically relevant players and, hence, had to be admitted ex‐post to the safety net, no questions asked.

Consider next the failures of focus. First, the prevailing regulatory framework established a neatly dividing line between the ex‐ante prudential norms and the ex‐post safety net. The ex‐ante regulatory framework focused on maintaining the soundness of assets, the ex‐post safety net on maintaining the liquidity of liabilities. The obvious loose end was the lack of ex‐ante internalization of systemic liquidity risk. Second, prudential regulation focused on the soundness of each institution under the assumption that the sum of sound institutions was equivalent to a sound system. However, as noted earlier, the Subprime crisis showed that this approach constituted a major fallacy of composition. It turned instead the approach on its head: the system is what matters most to the soundness of each institution.72 Third, traditional regulation focused on statistically observable risks and made much out of the sophisticated and complex risk modeling techniques that fed on these statistics. Yet, the Subprime crisis

72 Basel‐style regulation rewarded those institutions that covered their risks with products and services offered by other institutions. Yet, the Subprime crisis showed those atomized protections to be not only irrelevant (they provided a false sense of security, unraveling when most needed) but possibly counterproductive as well (they exacerbated contagion and the risk of overall systemic failure).

41

demonstrated that what you do not see is what will kill you (tail risks, black swans, and endogenous risk).73

Finally, consider the failures of dynamics. Basel‐style regulation was essentially static. Norms were time invariant (cycle independent) and the mandated capital buffers were assumed to be sufficient to carry the system through the business cycle.74 The Subprime crisis proved that approach wrong: static norms turned out to be pro‐cyclical, too loose on the way up, too tight on the way down. Last but not least, Basel‐style regulation failed to adequately incorporate the dynamic links between monetary and prudential policies. The central bank’s job adhered to ensuring macro stability and providing lender‐of‐last‐resort services, the supervisor’s to ensure financial prudency, and the two did not need to interact much. Yet, the insufficient attention of monetary authorities to the implications of their actions on financial developments, coupled with the insufficient attention of the supervisors to macro dynamics, deeply contributed to the crisis.75

A major problem when seeking to address these regulatory failures is that the best fix will most often depend on the paradigm. How one sees reform is thus essentially a function of the lens one uses. Table 2 synthesizes this discussion. The first questions in the table (under “foundations”) refer to the objectives of regulation. Although both the aims (reducing principal‐agent frictions or internalizing social costs) and the means (see below) differ, the need to align incentives through ex‐ante prudential norms is clear and uncontroversial under either the agency paradigm or the externalities paradigm. Instead, in the mood swings paradigm, the aim is to maintain innovation under control and to temper mood swings. While there is no obvious inconsistency between the two, aligning incentives and tempering moods are nonetheless clearly of a different nature.

In either case, the key question as regards the respective roles of markets and supervisors in achieving the mentioned objectives of regulation is whether risk can be priced (which in turn largely depends on whether systemic crises can be avoided). The answer is “yes” in the agency paradigm. Anyone who has enough “skin” invested in his own game will have incentives to maintain risk taking within socially acceptable bounds. Similarly, anyone with enough skin invested in somebody else’s game (and this can also be mandated by regulation) will have an incentive to look for the earliest signs of malfeasance. Markets can thus deliver efficient signals and function as early smoke detectors. Once principal‐agent problems are kept under control, systemic crises should not occur and historical statistics can become the bread‐and‐butter of day‐to‐day micro‐prudential risk management (i.e., help price risk across borrowers, institutions, and instruments). Accordingly, the main role of the agency supervisor is to put in place the necessary apparatus for markets to conduct their monitoring role effectively. Once this is done, his only residual role is one of compliance checking and crime policing (misrepresentation, fraud, looting, etc.).

73 While the regulatory framework has attempted to reduce the gap between risk and regulation (by upgrading from Basel I to Basel II), the Subprime crisis has brought into evidence severe issues of opacity, excessive complexity, and a misleading sense of control. See Tarullo (2008). 74 Spanish regulators were the only ones in the developed world that explicitly dealt with cyclical dynamics by introducing the so‐called “statistical provisions”—i.e., provisions that are built out of income during the upswing of the credit cycle and can be converted into specific provisions in the downward part of the cycle. This commendable approach was never embraced as part of the Basel creed, however. 75 Borio (2003), Goodhart et al. (2004), Rajan (2005), and White (2006) were among the few providing early forewarnings of the dangers of this approach.

42

Table 2. A Synthetic Overview of Regulatory Issues and General Policy Responses

Paradigm

Dimensions Issue Agency Externalities Mood Swings

What is the main problem?

Betting with someone’s else

money

Opportunistic behavior that

conflicts with the social good

Mood swings in an uncertain, evolving world

What should ex‐ante prudential norms do?

Align incentives through skin in

the game

Align incentives through

internalizing externalities

Temper moods and domesticate

creativity

Can risk be priced? Yes Probably not fully (one hundred year floods)

Probably not (unless Moses‐like supervisor)

How effective is market discipline?

Potentially very effective

Ineffective (inability to estimate or withstand

systemic risk)

Ineffective (inability to

comprehend or withstand

systemic risk)

Foundations

What is the role of the supervisor?

Enhancer of market discipline‐

crime police

Crowd manager‐fireman

Scout‐moderator‐fireman

Scope Should the line in the sand be redefined?

No Yes Not necessarily

Does fair value accounting help?

Yes, it is fundamental

No, it exacerbates externalities

No, it exacerbates mood swings

Are systemic liquidity norms needed?

No Perhaps Yes Probably Yes Focus

How important to look at the system?

Not important Very important Fundamental

Should prudential and monetary authorities coordinate?

Yes, but not tightly

Tightly Very tightly

Dynamics Are dynamic, macro‐prudential norms needed?

No Yes, rule‐based Yes, judgment‐

based

By contrast, the scope for market help is marginal at best in the externalities paradigm, where the key dimension of risk is dynamic rather than cross sectional. It is likely to be socially too expensive to put in place fully crisis‐proof prudential buffers. If so, risks of one hundred year floods (truly extraordinary events) will persist and markets can only help internalize externalities (i.e., provide systemic insurance) if they are able to calibrate the risks and costs of such events, and to withstand their strains. Neither is likely, however. For one thing, tail risks are unlikely to be estimated with precision, even when a

43

sufficiently long statistical history is available. For another, given the contrast between the huge scale of a systemic crisis and its low probability, this is an aggravated case of catastrophe insurance. In view of the difficulties that the latter has faced, it is dubious that full‐blown, market‐based systemic insurance will see the light of day any time soon.76

The scope for market assistance is limited even further in the mood swings paradigm. As in the externalities paradigm, risk is systemic and dynamic. However, rather than tail risks that can be ultimately modeled, exceptional bumps ahead are more in the nature of “black swans” (observations that cannot be inferred from previous data series) or “endogenous risk” (risk endogenously created by market participants).77 Hence, risk pricing becomes inherently difficult, not only because statistical history provides few clues as to what might be popping up ahead, but also because markets that are shaped by alternative bouts of euphoria and despair are unlikely to provide efficient, fundamentals‐based pricing signals. Thus, absent an effective oversight to prevent such financial system drifts (which, as argued below, will need to rely on greatly expanded supervisory skills and powers), Basel II’s aspiration to make regulation rest on internal risk management models, bolstered by risk‐rating agencies and market valuations, crumbles. This aspiration presupposes that risk dominates uncertainty and markets are efficient, two premises that an unbridled mood swings paradigm debunks.78

The only scope for markets to play a role in the mood swings paradigm would be taking bets on whether the system as a whole is headed in the right direction or likely to crash. While dedicated and well trained observers may well be able to detect an incoming iceberg through the fog, grasping how the system is wired and understanding the possible cracks is likely to require hefty investments and sophisticated skills. Hence, “systemic risk gazing” is unlikely to be a profitable market activity and should be viewed instead as a public good. Upgrading the role of the supervisor to provide such “holistic supervision” should therefore become a key component of reform. However, as discussed below, this will require, in addition to sound judgment and vision, sufficient independence and accountability—a tall order indeed.

Consider next some of the key implications for the nature of prudential regulation. As regards the scope of regulation (the “line in the sand”), the discrepancies between the three sides are obvious. A supervisor grounded in the agency paradigm would insist that allowing unregulated intermediaries to operate freely is the proper thing to do. Informed investors will naturally migrate to the unregulated world where innovation can thrive, risks and returns will likely be higher, and—as long as information is timely and reliable—users of funds will be appropriately disciplined. However, for the reasons already noted above, his externalities colleague would be dead set against the idea of allowing prudentially unregulated intermediaries to operate side by side with the regulated sector. The mood swings supervisor would be of a more mixed mind. Unregulated intermediaries could make his life more difficult as uncontrolled innovation, pushed along by the forces of competition and regulatory arbitrage, could set eventually the system on the wrong track. However, provided all innovation is regulated, he might find this to be manageable.

As regards the focus of regulation, the discrepancies across paradigms as regards the scope for market discipline have profound implications for the way risk is both reported and managed. Consider accounting issues first. In the agency paradigm, fair value accounting is clearly the superior alternative. Ensuring that changes in market values are immediately reflected in balance sheets is essential to contain the risk of a moral hazard‐driven bubble where undercapitalized intermediaries are allowed to

76 However, as proposed by Kashyap, Rajan, and Stein (2008), it might be feasible to set up private partial insurance schemes in the form of additional capital becoming available under stressful systemic events. 77 See Danielsson and Shin (2002). 78 De Grauwe (2008) makes a similar point.

44

continue operating normally. However, fair value accounting can be problematic under the other two paradigms. By enhancing the impact of one intermediary’s actions on the balance sheets of other intermediaries, it exacerbates externalities. At the same time, and perhaps more importantly, it magnifies the impact of liquidity or mood swing‐induced deviations in asset prices from their longer run fundamentals.

Consider risk management issues next. Are systemic liquidity norms needed? Clearly “no” under moral hazard (this is not a relevant problem), “perhaps” under externalities (as long as the ex‐ante social benefits exceed the ex‐ante social costs), and “probably yes” under mood swings. In the latter case, because crises are endogenous events rather than acts of god, they are likely to be more recurrent. Hence, unless the supervisor is convinced that he will be able to always navigate the ship around the icebergs, taking the proper systemic precautions is a good idea (multiple layers of steel against water inroads will better protect the keel).

How important is it to look at the system as a whole? In the agency case, this is not the proper way to look at the problem. Systemic events arise from individual malfeasance and this is where the emphasis should stay. Instead, in the externalities paradigm, a systemic perspective is naturally called for. Indeed, this is exactly what one does when one “internalizes the externalities”. In the mood swings paradigm, the focus on the whole is perhaps even more fundamental. Crises are manifestations of collective excesses and it is impossible to understand the dynamics of the whole by summing up the idiosyncratic risks and dynamic paths of individual institutions.

In this context, the answer to the question “how tightly should the prudential and monetary authorities coordinate?” is rather self‐evident. In the agency case, not much coordination is needed. Instead, the Greenspan doctrine seems to apply: let the prudential authority make sure that incentives are properly aligned and the monetary authorities make sure that the ship is sailing at the proper speed (i.e., take care of the cycle). In the externalities paradigm, the two authorities should instead closely consult each other to make sure that intermediaries are not unduly vulnerable to tail‐risk events and that the supervisor is sufficiently aware of where the cycle might go. In the mood swings paradigm, there should be very tight coordination between the two authorities and possibly even no major differentiation between them. By contributing to mood swings, monetary policy becomes an integral part of the prudential story. And the prudential risks ahead become a key dimension of monetary policy decision making. Hence, prudential and monetary adjustments are joined at the hip.

Along similar lines, are macro‐prudential, dynamically adjusted norms needed? In the stationary moral hazard world, the answer is clearly negative. Instead, in the externalities paradigm, the exposure to exogenous shocks and fluctuations provides a good basis for cycle‐adjusted norms because it allows prudential buffers to be real buffers, i.e., to be built up during the good times and used up during the bad times. In addition, these norms can help coordinate the actions of individual agents and thus overcome the prisoner’s dilemmas‐type situations. Given that the externalities are known (or knowable); this militates in favor of rules over discretion. The mood swings paradigm also makes a strong case for anti‐cyclical prudential norms but for a different reason. Rather than systematically limiting the ship’s speed under clear weather, the main motive in this case is to lift up the yellow flag when, under foggy weather, “icebergs may possibly be lying ahead”. Hence, mood swings provide a rationale for a judgment‐based anti‐cyclical framework, much as the one in effect for monetary policy—a framework where an independent body would have the discretion to calibrate the anti‐cyclical prudential instrument in light of evolving circumstances.

45

Consider finally the need for (and purpose of) a safety net (Table 3). To a large extent, this question relates to the scope for learning. In a system where learning is possible, it may be preferable to let agents face the hardships of financial crises and learn from experience.79 In the agency paradigm the system is not dominated by uncertainty and mood swings and, hence, should be broadly stationary (even if subjected to innovation). Therefore, agents should eventually learn. This might take a few crises and significant bruises (which in turn require that the ex‐post safety net not systematically validate the ex‐ante expectation of bailouts) but wisdom should eventually arise from the pain.80 Correspondingly, it would be better if the lender‐of‐last‐resort (LOLR) function did not exist. Bank runs are healthy manifestations of market discipline. Stopping runs unnecessarily protects banks that should fail and aggravates the misalignment of incentives for all other banks. Similarly, deposit insurance can only be justified by consumer protection but, given its adverse moral hazard implications, a pure agency supervisor would probably conclude that, on balance, the world would be a better place without it.81

By contrast, in the externalities paradigm, the nature of the problem makes learning irrelevant. As long as externalities are not internalized, participants only see their side of the story, no matter what. Moreover, there is no possible learning from exogenous and random acts of god or from self‐fulfilling runs in a multiple equilibrium world. Thus, to the extent that it is too expensive for society to prevent runs through large ex‐ante buffer requirements, an efficient LOLR becomes a socially superior solution and the cornerstone of the regulatory edifice. Also, as his forebears after the Great Depression, an externalities supervisor would conclude that deposit insurance is needed to induce the small uninformed depositors to join the banking system while preventing them from crying wolf and causing systemic havoc without justification. Again, however, having fire safety only a 911 call away hardly promotes incentives for keeping a fire extinguisher at home, another good example of regulatory collision between the paradigms.

79 The scope for learning is crucial for determining the need for any regulation, not just the safety net. Indeed, a good case can be made that even without a regulatory reform crises should convince principals (shareholders and investors) that they need to improve their control on agents (managers). 80 The remaining question, of course, is whether such a system would be “fair” to the smaller and less educated consumers who might be scared away and remain forever on the fringes but in the end this is likely to be an issue of consumer protection more than systemic stability. 81 Indeed, from a pure moral hazard perspective, the expansion of the safety net (particularly the creation of deposit insurance) can be seen as a mistaken knee‐jerk reaction that has come back to haunt the current regulatory architecture and the goal should be to get rid of it. See for example Herring and Santomero (2000), Gale (2004), and Calomiris (2008).

46

Table 3. The Need for A Safety Net

Paradigm

Issue Agency Externalities Mood Swings

Can players learn on their own?

Probably Yes No Apparently not

Is an ex‐post LOLR needed?

No, it is Counterproductive

Yes, to provide systemic liquidity

Yes, to absorb systemic risk

Is a deposit insurance needed?

Probably not Yes, to limit risks of

“wrong” runs Yes, to limit impact of mood swings

Interestingly, as regards the scope for learning, the mood swings paradigm lies somewhere in the middle. The constantly evolving environment makes some learning possible but tricky. One would think that agents should learn to be more cautious and eventually come to realize that, even if the scope for the truly new is constrained by path dependence, nasty surprises can emerge and that “not all that glitters is gold”. History has amply demonstrated that this is not the case, however. Moreover, learning in this paradigm is somewhat of an oxymoron. Believing that one has finally “learned the lesson” can boost over‐confidence in one’s ability to navigate through the obstacles, thereby setting in motion a mood swings‐induced bubble. The uncertainty conscious supervisor would thus agree with his externalities colleague as to the core importance of the LOLR. However, as already noted, he would expect the LOLR mainly to absorb systemic risk rather than provide liquidity. Similarly, he would agree that a deposit insurance is needed to “calm down” the frayed nerves of investors when moods start to turn ugly.

Towards a New Regulatory Framework

The discussion in the previous sections suggests that the design of a proper regulatory architecture faces two major challenges.82 The first is to build a regulatory framework that takes into account all three paradigms and avoids solving problems in one paradigm at the cost of making matters sharply worse in another. The second challenge is to find an adequate balance between financial stability and financial development. Extreme solutions—a crisis‐proof system that hardly intermediates or a thriving system that frequently collapses of its own weight—are of course to be avoided.

A fully specified reform proposal that meets these challenges lies much beyond the scope of this paper (even more so since the devil is in the implementation details). There is however a minimum set of basic objectives that, in our view, any new prudential architecture should seek to fulfill, either because they 82 A number of important and detailed proposals to fix the regulatory framework have already seen the light of day. See for example Financial Stability Forum (2008), Basel Committee on Banking Supervision (2008 a, and b, and 2009), Institute for International Finance (2008), and Goldstein (2008). The November 2008 Declaration of the G‐20 Summit on Financial Markets and the World Economy identifies the “root causes of the crisis”, sets out “common principles for reform of financial markets” and sketches an “action plan” to implement such principles. Rather than questioning the basic architecture and foundations of the current framework, these proposals have so far and for the most part sought to maintain (and build upon) this framework. While this approach is clearly understandable from a practitioner’s perspective, its longer term success will very much depend on the extent to which the key issues and interactions underpinning all three paradigms discussed in this paper are satisfactorily addressed.

47

cut across paradigms or they are absolutely central to one of the paradigms. Given the popularity of the agency paradigm, the reform agenda will likely be strong in addressing principal‐agent issues (including through governance improvements, changes in management compensation schemes, and increased skin‐in‐the‐game requirements). Hence, we focus in this section mainly on the objectives of the regulatory reform needed to address central issues under the externalities and mood swings paradigms.

The first objective, which is particularly relevant to the externalities paradigm but applies to all, is full regulatory neutrality. In a world where regulation is not applied uniformly, financial flows will sooner or later find the line of least resistance, giving unregulated financial institutions a competitive advantage and making them grow to the point where they become systemic behemoths. There are two possible solutions to this quandary. One is to make all financial intermediaries fit within the universal banking mode. This solution, however, would limit entry unduly and promote the preponderance of very large, too‐big‐to‐fail, financial conglomerates with limited creativity and large non‐competitive rents.

The alternative—which we find to be superior—is to maintain a distinction between commercial banks and other non‐deposit taking financial intermediaries, but make the latter choose between being prudentially regulated or being unregulated. All regulated intermediaries would need to satisfy the same prudential requirements (capital adequacy in particular) as commercial banks and in exchange benefit from LOLR services.83 However, reflecting their reduced responsibilities towards retail investors and the payment system, regulated non‐bank intermediaries would be subject to a lower entry capital (i.e., the minimum capital needed to open) and less cumbersome fit‐and‐proper tests than those applicable to commercial banks (otherwise all non‐bank intermediaries would become universal banks). The unregulated intermediaries, by contrast, would not need to satisfy capital adequacy requirements nor be subjected to an entry capital threshold. In exchange, however, they would be restricted to funding themselves only from regulated intermediaries, banks or non‐banks (i.e., they could not borrow directly from—or acquire contingent liabilities with—the market).84

This proposal has many benefits. As in the case of universal banking, it would comply with regulatory neutrality. Because unregulated intermediaries could only fund themselves from regulated intermediaries, a dollar lent to a final borrower through an unregulated intermediary would end up paying the same capital charge as a dollar lent through a regulated intermediary. Hence, systemic risk would be evenly internalized across all possible paths of financial intermediation, whether they involve regulated intermediaries or not.85

83 Following the same logic of regulatory neutrality, all asset‐backed securities issued with some form of recourse (including reputational) to the regulated intermediary, or purchased by a regulated intermediary, should carry an equity tranche retained by the issuer at least equivalent to the uniform capital adequacy requirement imposed on the intermediation system. 84 Thus, hedge funds that wish to remain unregulated would be allowed to borrow only from banks or other regulated intermediaries. In addition, they (as well as all other prudentially unregulated financial institutions) would not be permitted to engage as counterparties in credit derivatives transactions and other forms of default hedging and insurance (these give rise to contingent liabilities whose payment at the time they fall due may exert systemic stress by requiring asset fire sales). At the same time, a clear dividing line would also need to be established between financial and non‐financial corporations, with the latter not being allowed to engage in finance operations beyond basic trade credit 85 Some regulatory bias between intermediated debt and direct debt issues would persist, since systemic risk would be internalized only in the former case. However, because it would not involve leveraged intermediation or expose financial intermediaries, this residual bias should be much less problematic and more manageable. Notice also that our proposal is only meant to address the systemic risks associated with debt‐funded intermediation, but not those attached to unleveraged asset managers such as mutual funds, whose contribution to downward

48

At the same time, in contrast with universal banking, the proposed scheme would favor innovation and competition. Because they would not need to meet any entry capital requirements, unregulated intermediaries could start from scratch. This would facilitate the entry of the smaller players, possibly into “niche” or “boutique” intermediation. The most innovative and successful would eventually grow to become regulated and gain direct access to the capital markets. In turn, the most successful of the regulated non‐bank intermediaries could grow further to become universal banks, thereby authorized to tap deposits and take on full payment system responsibilities.86 The cost of oversight would remain low, however, as the activities of the unregulated would be monitored on a contractual basis by the regulated intermediaries that lend to them.87 This would effectively “delegate” supervision to the regulated intermediaries, creating a two‐tiered “nursery” system in which the start‐ups could prosper and grow under the watchful eye of the better‐established (and more experienced) institutions.

Most importantly, this proposal does not rely on artificial boundaries set up by the regulator between “systemically important” and “systemically unimportant” financial intermediaries, based on size, activity, or some risk‐based measure of systemic impact (such as the recently proposed CoVar).88 Such distinctions are bound to create unending distortions or be very difficult, if not impossible, to implement operationally. If the distinction is based on a simple objective criterion, such as size, unregulated intermediaries could multiply and engage in “systemic herding”. They would individually benefit from the lighter regulation by staying just below the size threshold but become just as systemically important as a whole as in the case where unregulated intermediaries of any size were allowed to operate. On the other hand, risk‐based distinctions, even if based on meaningful and uncontested models (by no means an obvious proposition), are bound to create grey zones with an uneven playing field as regards both the intensity of regulation and access to the safety net. In such a context, reclassifying institutions in and out of the systemically important list is likely to be an operational and political conundrum. Instead, by treating all intermediaries equally subject only to a simple choice by the intermediary itself, our proposal is much simpler and operationally quite easy to implement.

The second objective, particularly relevant to the externalities paradigm but also consistent with all three paradigms, is to keep the system reasonably close to a stable path (hence enhancing the scope for prices to reflect fundamentals) through a better alignment of incentives. In this regard, a key missing piece in the current framework is the internalization of systemic liquidity risk. Proposals have been made to penalize maturity mismatches between assets and liabilities. However, since short assets are likely to become as illiquid as long assets under systemic events, it seems preferable to focus on the maturity of the funding structure, irrespective of that of assets.89 By inducing final investors to hold at least part of the liquidity risk instead of pushing it back on the system, this should reduce the system’s exposure to liquidity events. In any event, a liquidity‐related norm would need to be properly calibrated

liquidity spirals is tempered (albeit not eliminated, particularly under conditions of structural or temporary asset market illiquidity) by the marking‐to‐market of their liabilities. 86 In this scheme, development banks could play a particularly important and relatively novel role. They could nurture innovation and promote competition and access by financing unregulated intermediaries and helping them grow. Their lower aversion to risk (supported by the State’s higher risk sharing capacity) would give them a natural edge over private regulated intermediaries. 87 Kambhu, Schuermann, and Stiroh (2007) discuss the benefits (and limitations) of such indirect monitoring of hedge funds by regulated entities and conclude that it is a preferable alternative to direct regulation. 88 See Brunnermeier et al. (2009). 89 Penalizing maturity mismatches could encourage intermediaries to lend short. This would push liquidity risk on to borrowers but would not eliminate it from the system as it would increase the risk of defaults under systemic stress. Moreover, when several banks lend to the same borrower, it could encourage run‐like loan recalls by banks that could further exacerbate systemic stress.

49

to reflect social costs and benefits, could take many alternative forms (a special capital charge, a risk‐adjusted insurance premium, or both), and would need to reconcile the inherent pro‐cyclicality of nearly any norm based on contemporaneous risk with the need for counter‐cyclical adjustments.90 None of the above is trivial.91

The third (and closely related) objective is to continue improving the safety net, reflecting its centrality to the externalities and mood swings paradigms. (Even with vigilant supervision and sufficient internalization of externalities, the high social costs of crisis‐proof systems and the uncertain turns taken by continually evolving financial systems render the full elimination of crisis a socially undesirable endeavor.) The objective of improving the safety net calls for: (i) reviewing the pricing of deposit insurance schemes to better reflect their de facto systemic exposure; (ii) examining whether access to the LOLR should be paired with a systemic insurance that all prudentially regulated intermediaries (whether deposit‐taking or not) should subscribe to; and (iii) rethinking the LOLR from a mood swings perspective, i.e., as a risk absorber of last resort. As noted, under our proposal for the scope of prudential regulation, all regulated intermediaries would have equal access to the LOLR. In contrast, unregulated intermediaries would be allowed to fail under an efficient bankruptcy code (this would allow the less successful intermediaries to exit promptly, thereby maintaining the vitality of the system).

The fourth objective relates to the importance of keeping a tighter rein on the possible downstream risks of financial innovation, particularly (but not only) from a mood swings perspective. This would require giving the regulator more powers to regulate, standardize, and authorize all forms of innovation (whether in instruments, institutions, or markets) and to subject them to much more rigorous pre‐approval and road‐testing, much as in the case of new drugs for the FDA.92

The fifth objective is realigning the respective monitoring roles of markets and supervisors to address the underlying weaknesses of market discipline under both the externalities and mood swings paradigms. Markets can no doubt continue to play an important ex‐ante role in helping align incentives with respect to principal‐agent frictions. However, it would be foolish to expect market discipline to prevent externality‐ or mood swings‐induced systemic crises. Moreover, imposing market discipline ex‐post, once the system is deeply out of equilibrium and a crisis is unfolding, is fraught with danger.93

By contrast, in the multi‐paradigm world, the supervisor would be naturally expected to have such a tough and complex responsibility that reasonable doubts exist as to whether its implementation lies in the feasible range. Unlike in the pure agency paradigm, he can no longer relax and concentrate on relatively simpler policing tasks once he has put in place the necessary arrangements to promote market discipline (hence, self‐regulation). Instead, the “holistic” supervisor of the mood swings paradigm provides a valuable scouting, moderating, and coordination service to society that markets cannot provide. To this end, he should be able to connect the dots, understand the forest beyond the trees, and

90 The direction towards which incentives need to be aligned (and moods tempered) shifts abruptly depending on the phase of the cycle: the upward phase calls for taking less risk and accumulating capital, the downward phase for taking more risk and using up capital. 91 Additional ways to better internalize systemic liquidity risk might also include limits on gross leverage, an in‐depth review of the differentiated capital requirements on trading books versus banking books, and some form of liquidity buffer (i.e., a prudential norm encouraging the holding of systemically safe assets). On the latter, see Morris and Shin (2008). 92 A very similar recommendation can be found in Buiter (2008). By the same token, the tight linkages between financial innovation and deregulation also call for special attention to the potentially destabilizing market implications of regulatory reform (unduly exuberance or moral hazard‐induced dynamics). 93 The failure of Lehman Brothers provides a vivid recent illustration of the risks attached to 11th hour attempts to limit moral hazard by restricting access to the safety net.

50

look ahead for possible systemic trouble. He would need the means and the clout to help coordinate expectations around systemically sustainable paths. This in turn calls for a deeper informational role—i.e., to provide systemically oriented information and benchmarks to help intermediaries think systemically and fashion their risk assessments accordingly. However, deeds will need to be added to words, which will require boosting the supervisor’s capacity (and skills) to exert judgment‐based discretionary interventions to slow down credit cycles, or restrict specific forms of intermediation that may become riskier as they develop. Given evolutionary uncertainty, macro‐prudential regulation cannot be entirely rule‐based. Instead, counter‐cyclical prudential norms may have to be at least in part judgment‐based, calibrated discretionally in view of changing circumstances, much as the interest rate is calibrated by monetary authorities.94 Of course, what shape and form such an instrument could take is hardly a trivial issue.

The stronger powers of the “holistic” supervisor would also be accompanied by a tougher responsibility and, with it, a risk of calamitous failure. If things go well, financial market participants will reap the benefits and the supervisor would be an unsung hero. If things go wrong, moral hazard will have a field day: “it was the regulator’s fault, hence the state’s responsibility to pay for damages.” Moreover, initial success in stirring the system may breed complacency and irrational exuberance leading to a crash down the line. Avoiding these pitfalls will require combining hard‐wired rules (that maintain the system within reasonable bounds) with an institutional reform that is commensurate with the supervisor’s new terms of reference (including his enhanced powers and responsibilities), and sufficiently strong to overcome the multiple difficulties associated with the use of discretion. Finding the right implementation modalities and regulatory mix between rules and discretion is likely to be one of the toughest yet most central challenges of prudential regulatory reform in the years ahead.95

94 Indeed, reflecting more tenuous and complex links between the instrument and the final objective, a pure rule‐based macro‐prudential policy could be even more elusive than a pure Taylor rule‐based monetary policy. Instead, having to explain and justify decisions could help promote progress on macro‐systemic prudential analysis, much as has been the case with inflation targeting for monetary policy. 95 In this context, to avoid regulatory capture, a particularly hard look will need to be given to the political economy of regulation (see Demirguc‐Kunt and Serven, 2009). This problem can become trickier when the supervisor needs to round off his views partly based on those who are closer to the market, including financial intermediaries. At the same time, however, players should realize that systemic adjustments should affect all players equally (provided regulation is truly neutral) and are for the common good, which should ease the way for fruitful coordination, much as in the case of monetary policy.

51

References

Abreu, Dilip and Markus Brunnermeier, 2003, “Bubbles and Crashes”, Econometrica, 71‐1, pp 173‐204.

Adrian, Tobias and Hyun Song Shin, 2007, “Liquidity and Leverage”, Mimeo, Federal Reserve Bank of New York and Princeton University.

Adrian, Tobias, and Hyun Song Shin, 2008, “Financial Intermediaries, Financial Stability, and Monetary Policy”, Federal Reserve Bank of New York Staff Report 346.

Allen, Franklin and Douglas Gale, 1998, “Bubbles and Crises”, Wharton School Working Paper 98‐01.

Allen, Franklin, and Douglas Gale, 2003, “Capital Adequacy Regulation: In Search of a Rationale”, in Economics for and Imperfect World by Joseph E. Stiglitz, Richard Arnott, Bruce Greenwald, and Ravi Kanvur. MIT Press.

Allen, Franklin, and Douglas Gale, 2005, “Systemic Risk and Regulation”, Wharton Financial Institutions Center Working Paper No. 95‐24.

Ambrose, Brett, Michael Lacour‐Little, and Anthony Sanders, 2005, “Does Regulatory Arbitrage of Asymmetric Information Drive Securitization?”, Journal of Financial Services Research.

Arrow, Kenneth, and Robert Lind, 1970, “Uncertainty and the Evaluation of Public Investment Decisions”, American Economic Review 60, pp 364‐378.

Ashcraft, Adam, and Til Schuermann, 2008, “The Seven Deadly Frictions of Subprime Mortgage Credit Securitization”, Mimeo, Federal Reserve Bank of New York.

Azariadis, Costas, 1981, “Self‐Fulfilling Prophecies”, Journal of Economic Theory, 25, 380‐96.

Barberis, Nicholas, and Richard Thaler, 2003, “A Survey of Behavioral Finance”, in Handbook of the Economics of Finance, George Constantinides, Milton Harris and Rene Stulz, editors, Elsevier.

Basel Committee on Banking Supervision, 2008a, “Principles for Sound Liquidity Risk Management and Supervision”, Bank for International Settlements, June.

Basel Committee on Banking Supervision, 2008b, “Proposed Revisions to the Basel II Market Risk Framework”, Bank for International Settlements, July.

Basel Committee on Banking Supervision, 2009, “Proposed Enhancements to the Basel II Framework”, Bank for International Settlements, January.

Berger, Allen, Richard Herring, and Giorgio Szego, 1995, “The Role of Capital in Financial Institutions”, Wharton School Working Paper 95‐01.

Boot, Arnoud, Stuart Greenbaum, and Anjan Thakor, 1993, “Reputation and Discretion in Financial Contracting”, American Economic Review, 83, pp 1165‐1183.

Borio, Claudio, 2003, “Towards a Macroprudential Framework for Financial Supervision and Regulation?”, BIS Working Paper 128.

Brunnermeier, Markus, 2008, “Deciphering the 2007‐2008 Liquidity and Credit Crunch”, Journal of Economic Perspectives, (forthcoming).

Brunnermeier, Markus, Andrew Crockett, Charles Goodhart, Avinash Persaud, and Hyun Shin, 2009, The Fundamental Principals of Financial Regulation: 11th Geneva Report on the World Economy

52

Brunnermeier, Markus, and Lasse Pedersen, 2008, “Market Liquidity and Funding Liquidity”, NBER Working Paper 12939.

Buiter, Willem, 2008, “Lessons from the Global Credit Crisis for Social Democrats”, Mimeo, European Institute (December).

Caballero, Ricardo, 2009, “A Global Perspective on the Great Financial Insurance Run: Causes, Consequences, and Solutions”, Vox, January 23.

Calomiris, Charles, 2008, “The Subprime Turmoil: What’s Old, What’s New, and What’s Next”, Mimeo, Columbia University.

Calomiris, Charles, and Charles Khan, 1991, “The Role of Demandable Debt in Structuring Optimal Banking Arrangements”, American Economic Review 81, pp 497‐513.

Caprio, Gerard, Asli Demirguc‐Kunt, and Edward Kane, 2008, The 2007 Meltdown in Structured Securitization”, World Bank Working Paper 4756.

Coval, Joshua, Jakub Jurek, and Erik Stafford, 2008, “The Economics of Structured Finance”, Harvard Business School Working Paper 09‐060.

Danielsson, Jon, and Hyun Song Shin, 2002, “Endogenous Risk”, Mimeo, London School of Economics.

De Grauwe, Paul, 2008, “Lessons from the Banking Crisis” A Return to Narrow Banking”, Mimeo, University of Leuven.

De Grauwe, Paul, 2008, “Animal Spirits and Monetary Policy”, Mimeo, University of Leuven

Dell’ Ariccia, Giovanni, Deniz Igan, and Luc Laeven, 2008, “The US subprime mortgage crisis: A Credit Boom Gone Bad?” Vox, February 4.

Della Vigna, Stefano, 2007, “Psychology and Economics: Evidence from the Field”, NBER Working Paper 13420

Demirguc‐Kunt, Asli, and Luis Serven, 2009, “Are All the Sacred Cows Dead?” World Bank Working Paper 4807.

Dewatripont, Mathias, and Jean Tirole, 1994, The Prudential Regulation of Banks, MIT Press.

Diamond, Douglas, and Phillip Dybvig, 1983, “Bank Runs, Deposit Insurance, and Liquidity”, Journal of Political Economy 109, pp 287‐327.

Diamond, Douglas, and Raghuram Rajan, 2000, “A Theory of Bank Capital”, Journal of Finance 55, pp 2431‐2465.

Eichengreen, Barry, 2008, “Origins and Responses to the Crisis”, Mimeo, University of California, Berkeley.

Epstein, Larry, and Tan Wang, 1994, “Intertemporal Asset Pricing under Knightian Uncertainty”, Econometrica, 62:2, pp 283‐322.

Financial Stability Forum, 2008, “Report of the Financial Stability Forum on Enhancing Market and Institutional Resilience”, Basel (April).

Fisher, Irvin, 1933, “The Debt Deflation Theory of Great Depressions”, Econometrica Vol 1, pp 337‐57.

Fostel, Ana and John Geneakoplos, 2008, “Leverage Cycles and the Anxious Economy”, American Economic Review 98:4, pp 1211‐1244.

53

Gale, Douglas, 2004, “Notes on Optimal Capital Regulation”, New York University, mimeo.

Geneakoplos, John, and Heracles Polemarchakis, 1986, “Existence, Regularity and Constrained Sub‐optimality of Competitive Allocations when the Asset Market is Incomplete”, in Essays in Honour of K. J. Arrow, vol 3, Heller, W., Starret, D. and Starr, R. (Cambridge).

Geweke, John, 2001, “A Note on Some Limitations of CRRA Utility”, Economic Letters, 71 (3): 341‐345

Goldstein, Morris, 2008, “Addressing the Financial Crisis”, Peterson Institute for International Economics, Washington D.C.

Goodhart, Charles, Boris Hofman, and Miguel Segoviano, 2004, “Bank Regulation and Macroeconomic Fluctuations”, Oxford Review of Economic Policy 20(4), pp 591‐615.

Gorton, Gary, 2008, “The Subprime Panic”, NBER Working Paper 14398.

Gorton, Gary and Andrew Winston, 2002, “Financial Intermediation”, Wharton School Working Paper 02‐28.

Greenlaw, David, Jan Hatzius, Anil Kashyap, and Hyun Song Shin, 2008, “Leveraged Losses: Lessons from the Mortgage Market Meltdown”, US Monetary Policy Forum Report No. 2 , Rosenberg Institute, Brandeis International Business School and Initiative on Global Markets, University of Chicago Graduate School of Business.

Greenspan, Alan, 1998, “The Role of Capital in Optimal Banking Supervision and Regulation”, Federal Reserve Bank of New York Economic Policy Review 4‐3, pp 163‐68.

Herring, Richard, and Anthony Santomero, 2000, “What is Optimal Financial Regulation?”, Center for Financial Institutions Working Paper 00‐34, Wharton School.

Holmstrom, Bengt, and Jean Tirole, 1998, “Private and Public Supply of Liquidity”, NBER Working Paper 5817.

Huang, Rocco, and Lev Ratnovski, 2008, “The Dark Side of Bank Wholesale Funding”, Risk Analysis and Management Conference, World Bank and IMF.

Institute for International Finance, 2008, “Final Report of the IIF Committee on Market Best Practices: Principles of Conduct and Best Practice Recommendations” July.

Kashiap, Anil, Raghuram Rajan, and Jeremy Stein, 2008, “Rethinking Capital Regulation”, Mimeo, University of Chicago and Harvard University.

Khan, Charles, and Joao Santos, 2008, “Liquidity, Payment and Endogenous Financial Fragility”, Conference on Liquidity: Concepts and Risks.

Keys, Benjamin, Mukherjee Tanmoy, Amit Seru, and Vikrant Vig, 2008, “Did Securitization Lead to Lax Screening: Evidence from Subprime Loans 2001‐2006”, EFA 2008 Athens Meeting Paper.

Khandani, Amir, and Andrew Lo, 2008, “What Happened to the Quants in August 2007: Evidence from Factors and Transactions Data”, NBER Working Paper 14465.

Kim, Deasik, and Anthony Santomero, 1988, “Risk in Banking and Capital Regulation”, Journal of Finance 43, pp 1219‐33.

Kindleberger, Charles, and Robert Aliber, 1996, Manias, Panics, and Crashes: A History of Financial Crises, Wiley Investment Classics.

Korinek, Anton, 2008, “Regulating Capital Flows to Emerging Markets”, University of Maryland, Mimeo.

54

Leamer, Edgard, 2008, “Housing is the Business Cycle”, NBER Working Paper 13428.

Lo, Andrew, 2004, “The Adaptive Markets Hypothesis”, The Journal of Portfolio Management, 30th Anniversary Issue.

Lorenzoni, Guido, 2007, “Inefficient Credit Booms”, NBER Working Paper 13639.

Minsky, Hyman, 1975, John Maynard Keynes, Columbia University Press

Morris, Stephen, and Hyun Song Shin, 2008, “Financial Regulation in a System Context”, Brooking Papers on Economic Activity, Fall 2008 Conference Draft

Rajan, Raghuram, 1998, “The Past and Future of Commercial Banking Viewed Through an Incomplete Contract Lens”, Journal of Money, Credit and Banking 30, pp 524‐550.

Rajan, Raghuram, 2005, “Has Financial Development Made the World Riskier?” NBER Working Paper 11728 (November).

Rajan, Raghuram, 2008a, “Bankers’ Pay is Deeply Flawed”, Financial Times, January 8.

Rajan, Raghuram, 2008b, “A View of the Liquidity Crisis”, mimeo, University of Chicago.

Santos, Manuel, and Michael Woodford, 1997, “Rational Asset Pricing Bubbles”, Econometrica, 65:19‐57.

Schumpeter, Joseph, 1934, The Theory of Economic Development, Harvard University Press.

Shiller, Robert, 2006, Irrational Exuberance, Princeton University Press.

Shleifer, Andre, and Robert Vishny, 1997, “The Limits of Arbitrage”, Journal of Finance, 52:35‐55

Song, Fengshua, and Anjan Thakor, 2008, “Financial System Architecture and the Co‐evolution of Banks and Capital Markets”, Pennsylvania State University and Washington University (Mimeo).

Taleb, Nassim Nicholas, 2007, The Black Swan: The Impact of the Highly Improbable, Random House.

Tarullo, Daniel, 2008, Banking on Basel, Peterson Institute for International Economics.

Veronesi, Pietro, and Luigi Zingales, 2008, “Paulson’s Gift”, Mimeo, University of Chicago.

Weitzman, Martin, 2007, “Subjective Expectations and Asset‐Return Puzzles”, American Economic Review, 97(4): 1102‐1130.

White, William, 2006, “Is Price Stability Enough?” BIS Working Paper 205 (April).

55

3. HOW HAS POVERTY EVOLVED IN LATIN AMERICA AND HOW IS IT LIKELY TO BE AFFECTED BY THE ECONOMIC CRISIS?

Joao Pedro Azevedo, Ezequiel Molina, John Newman, Eliana Rubiano and Jaime Saavedra

June 2009

Abstract

After establishing the recent history of what has happened to regional poverty in LAC, the note presents simulations of the potential poverty impact of the current crisis. A range of simulations are presented, drawing upon alternative specifications of the relation between per capita GDP growth and poverty and a range of estimates of how GDP per capita in different countries is expected to evolve in 2009.

For almost all of the 1980s and 1990s, the number of poor and extreme poor in Latin America and the Caribbean rose. Despite the growth episodes observed in the nineties, poverty rates stagnated. The number of poor climbed from 160.5 million in 1981 to 240.6 million by 2002, and of extreme poor from 90 to 114 million. But since 2002 the number of poor has decreased at unprecedented speed – so much so that in 2008 the number of poor is estimated to have fallen to 181.3 million and the number of extreme poor to 73 million. That is, almost 60 million people moved out of poverty while 41 million left the ranks of the extreme poor. Unfortunately, the recent worldwide recession has put an end to that progress and the number of poor are now projected to increase.

Based on GDP growth forecasts for May 2009, the aggregate poverty rate for LAC is estimated to rise 1.1 points. This would mean that there would be 8.3 million more poor people in 2009 than in 2008. A more pessimistic forecast will move the increase in the poverty rate to 2 points and increase in the number of poor to 13 million. The aggregate extreme poverty rate is estimated to rise 0.5 points. This would mean a further 3.6 million would fall into extreme poverty.

Introduction

This note examines the recent evolution of poverty in Latin America and estimates what is likely to happen to poverty as a result of the current economic crisis. It presents new estimates for the average poverty and extreme poverty rates and for the number of extreme poor and poor for LAC, based on PPP $4 and $2 international poverty lines. While the World Bank uses international poverty lines of PPP $2 a day for poverty and PPP $1.25 a day for extreme poverty when reporting world figures, applying these lines yields a level of poverty in PPP terms that is too far below the national figures to be of interest in Latin American countries. An analysis of the national poverty and extreme poverty lines used in Latin America suggest that international poverty lines of PPP$ 4 a day for poverty and PPP$ 2 a day for extreme poverty are more appropriate (See Annex 1).

After establishing the recent history of what has happened to regional poverty in LAC, the note presents simulations of the potential poverty impact of the current crisis. A range of simulations

56

are presented, drawing upon alternative specifications of the relation between per capita GDP growth and poverty and a range of estimates of how GDP per capita in different countries is expected to evolve in 2009.

Evolution of Poverty in LAC over the recent past

Evolution of poverty rates and link to movements in per capita GDP Figure 1 illustrates how extreme poverty and per capita GDP have evolved between 1981 and 2008 in Latin America, while Figure 2 shows a similar evolution of poverty and per capita GDP96. The patterns are quite striking in both cases, as there are clearly four distinct periods. In three of the periods, the evolution of poverty rates move is an almost exact mirror image of the evolutions in per capita GDP. It is only during the 1990s (a “lost decade” in LAC in terms of poverty reduction), where the link between movements in per capita GDP and movements in poverty is broken97. Over that period, per capita GDP continued to grow as it had in the 1980s, but poverty rates did not decline.

96 The data for 1981 to 2005 are taken from the WB’s regional aggregation module of POVCALNET, after setting the PPP international poverty lines at $2 and $4 a day. This module weights the individual country data by their respective populations and interpolates the data (as needed) so as to produce observations for all countries for every 3 years between 1981 and 2005. The data for 2006 were calculated by the authors using 2006 and 2007 individual country data from POVCALNET from Argentina, Brazil, Chile, Paraguay, Uruguay, Bolivia, Colombia, Peru, Ecuador, Venezuela, Costa Rica, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, the Dominican Republic and Jamaica. The sample is representative of 95.2% of the population in Latin America, but representative of only 40% of the population in the Caribbean. This procedure differs from the data used in the regional aggregation module (i.e. the data from 1981‐2005) in that it does not include data from Guyana, Haiti, St Lucia, Suriname and Trinidad and Tobago. These countries together make up a small fraction of the total population in LAC. Data for 2007 and 2008 are projections. 97 Even with the break in the 1990s, the simple correlation between the LAC regional aggregates of per capita GDP and both extreme and moderate poverty rates is around ‐0.88.

57

The change in poverty rates and per capita GDP over the four distinct periods can be seen clearly in Table 1.

Table 1. Changes in Poverty and Per Capita GDP in LAC over different episodes Episode GDP pc change 2 USD a day Poverty Change 4 USD a day Poverty change 1981‐1984 ‐2.88 1.16 1.35 1984‐1990 0.12 ‐1.03 ‐1.31 1990‐2002 1.03 ‐0.03 0.06 2002‐2006 2.88 ‐1.74 ‐2.55 2002‐2008* 3.01 ‐1.40 ‐2.16 Annex 2 presents similar graphs of the trends in poverty and growth in GDP per capita for other regions of the world. It is apparent that in no other region did one observe such a protracted period of significant growth in GDP per capita without a resulting decrease in poverty as was observed in Latin America during the 1990s.

The aggregate figures are heavily affected by what happened in Brazil, Mexico and Colombia, which make up 57.3 percent of the population of LAC. Figures 3‐4 present information on the behavior of per capita GDP growth and changes in poverty for the entire distribution of countries in Latin America. The box plots show the minimum and maximum values, and the values for the 25th, 50th and 75th percentiles. The median value (50th percentile) is shown as a bar in the middle of the box.

58

Fig. 3. Distribution of Changes in Per Capita GDP in LAC

Fig. 4. Distribution of Changes in Poverty in LAC (At $4 a day PPP)

During the period 1981‐1984, virtually all countries experienced declines in per capita GDP and increases in poverty. Between 1984 and 1990, the growth in per capita GDP picked up to the point where the median value was slightly above zero. Poverty stopped increasing in most countries, but the median value of changes in the poverty rate was close to zero. The decline in poverty in the aggregate figures over this time period (Fig. 1 and 2) indicates that it was the larger countries that experienced declines in poverty. Over the 1990s there was a better performance in per capita GDP growth, but a very wide dispersion in the poverty results. It was not until 2002‐2008 when both the performance in per capita GDP growth and in poverty became strong across the entire distribution.

59

Evolution in the number of poor

Table 2 illustrates changes in the number of poor 1981 and 2008. While there are six instances when extreme poverty and poverty rates declined (and 2 when they increased), there are fewer periods when the number of poor declined. During the eighties, a period of weak growth and high volatility, poverty rates fell slightly, but the number of poor increased. During the nineties, a period of stronger growth, the same pattern emerges, and, actually, the number of poor is substantially higher by the end of the decade. The big break in Latin America came in the period between 2002 and 2006. Extreme poverty fell 6 points and 36 million people moved out of extreme poverty. In turn, the poverty rate fell more than 10 points and 56 million people moved out of poverty. Projections for 2007 and 2008, show an additional decrease, of at least two additional points of further extreme poverty and moderate poverty rates reduction98. Hence, during the strong growth period 2002‐2008, almost 60 million people were lifted out of poverty, and 41 million left the ranks of extreme poverty. It should be noted that towards 2007 and 2008 there seems to be already a deceleration on the rate of poverty reduction.

Table 2. Poverty and Extreme Poverty ‐ Rates and Number of Poor (1981‐2008)

Year Extreme Poverty Rate (PPP $2 a day)

Number of Extreme Poor (millions)

Poverty Rate (PPP $4 a day)

Number of Poor (millions)

1981 24.6 90.0 48.5 160.5 1984 28.1 109.5 52.6 205.0 1987 24.9 103.0 47.5 196.8 1990 21.9 95.8 44.7 196.0 1993 20.7 95.5 44.4 205.1 1996 22.0 106.8 46.1 223.6 1999 21.8 110.7 45.5 230.8 2002 21.5 114.0 45.4 240.6 2005 17.1 94.2 38.8 213.6 2006 14.6 78.0 35.2 188.0 2007* 13.6 75.1 33.3 183.6 2008* 13.1 73.3 32.5 181.3 * Projected

The Rapidly Developing Worldwide Economic Crisis

The increase in food and energy prices in 2007 and 2008 raised concerns that the continuation of the good times in Latin America might be threatened. The arrival of the worst worldwide economic crisis since the Great Depression to Latin America has made it clear that the threat is now a reality. The period of rapid per capita GDP growth that Latin America experienced between 2002 and the middle of 2008 has come to an end. The concern today is how long and how deep will the recession be and how severely will poverty be affected. The downturn during the last quarter of 2008 was particularly dramatic and each month seems to bring worse news than the month before. While the industrialized countries were the first to experience rapid downturns in projected growth, the projections of GDP growth for most developing countries

98 2007 and 2008 are still projections as not all large countries have data for 2007, and only a few have infromation for 2008.

60

(including those of Latin America) are now following the industrial countries downwards. Figure 5 indicates how the predictions for 2009 GDP growth in the US, Canada, the UK, the Euro Zone and Japan have declined every month since January 2008. There is no indication that the bottom has been reached. Figure 6 presents forecasts for Latin American countries which are covered by Consensus Forecasts. A similar downward trend in the forecasts is evident.

Figure 5: Trends in Consensus Forecasts for 2009 GDP growth in US$,

Canada, Euro Zone, UK and Japan

-6.0

-4.0

-2.0

0.0

2.0

4.0

Jan

14, 2

008

Feb

11, 2

008

Mar

10,

200

8

Apr 1

4, 2

008

May

12,

200

8

Jun

09, 2

008

Jul 1

4, 2

008

Aug

11, 2

008

Sep

08, 2

008

Oct

13,

200

8

Nov

10,

200

8

Dec

08,

200

8

Jan

12, 2

009

Feb

09, 2

009

Mar

09,

200

9

GD

P (%

cha

nge )

Canada Euro zone Japan UK USA

Source: Consensus Forecast

Figure 6: Trends in LAC Consensus Forecasts for 2009 GDP growth

Source: Consensus Forecast

61

As of April 2009, Brazil, Mexico, Chile, Ecuador, Argentina and Venezuela are all forecasted to have negative GDP growth. Moreover, given the rates of population growth that prevail in LAC many more countries (15) are forecast to experience a negative rate of growth in per capita GDP. As with the industrialized countries, every month the forecasted growth rates in GDP have been revised downwards and there is no clear indication that the bottom has been reached.

The dramatic reversal in growth rates is apparent in Figure 7 which plots the number of countries that have experienced negative growth in any given year between 1980 and 200799. The figures for 2009 are the projected number of countries that, as of March 2009, are expected to experience negative per capita GDP growth. It is apparent that after falling to unprecedented low levels between 2002 and 2007, the number of countries that are now projected to have negative growth in per capita GDP has shot up sharply in both LAC and the World.

Figure 7: Number of countries with negative growth in per capita GDP

0

10

20

30

40

50

60

70

80

90

1980 1983 1986 1989 1992 1995 1998 2001 2004 2007

Year

Num

ber o

f cou

ntrie

s

World LAC

Note: Year 2009 projected Source: World Bank World Development Indicators

Figure 8 provides additional detail on the size of the decline in per capita GDP for all Latin American countries over all past periods when per capita GDP growth was negative. The magnitude of the decline is represented by both the size of the bar and the darkness of the color. 100 Larger declines in per capita GDP are represented by larger bars and darker colors. For example, the shock in Argentina between 2001 and 2002 is represented by a large dark bar.

Included in this figure are the 2009 projected growth rates in per capita GDP for those countries which are projected to have declines. One can observe that, as of March 2009, these projected declines have not yet reached some of the past levels. However, the contrast between the period 2003‐2008 when then were virtually no countries with negative per capita GDP growth and 2009 when virtually all countries are expected to suffer negative growth in per capita GDP is dramatic.

99 The figure includes those countries that were just starting in a given year, as well as those countries that were repeaters – ones that might have been in their second or more consecutive period of negative per capita GDP growth. 100 Using both the size of the bar and color to represent the magnitude is done to create a stronger visual impact and to make the periods of greatest decline stand out from the other periods.

2009 Projected

62

63

Potential Impact of the Current Worldwide Recession on Poverty in LAC

This note analyzes the potential impact on poverty in two ways:

a) By describing what have been the changes in poverty during previous downturns; and b) By estimating elasticities of poverty reduction with respect to changes in per capita GDP

and using those elasticities, together with projected growth rates to estimate likely changes in poverty rates and the number of poor.

All of this analysis is subject to the very strong caveat that we may be facing a situation that represents something very different from what has been faced in the past. While there have certainly been periods of negative per capita GDP growth in many LAC countries at the same time (mainly during the first half of the 1980s), this has not ever happened after a period of universal good growth and fast poverty reduction. Whereas previous periods of negative growth in per capita GDP over the last 25 years have usually been triggered by changes in developing countries, the current period has been triggered by events in industrialized countries. While many countries had entered into previous periods of negative growth in per capita GDP with public sector deficits, macroeconomic balances in most –not all‐ LAC countries around the end of 2008 had been strong, which gives some countries room to implement countercyclical policies. Finally, some of the factors that appear to have contributed to the recent significant gains in poverty between 2002 and 2008 (expansion of conditional cash transfers, greater effectiveness of some public transfer programs, and rising remittances) were either not present or at much lower levels during periods of previous downturns.

Changes in Poverty During Previous Downturns

Unfortunately, it is not possible to analyze what happened to poverty during all periods of negative growth in per capita GDP because poverty data are not collected annually in many countries in the region. Moreover, there is far less poverty data available during the 1980s, when there were many periods of crises.

The analysis presented in this section makes use of all available data on poverty, taken from the SEDLAC data bank of comparable LAC household surveys.101 The SEDLAC database contains data on poverty reported by national statistical agencies and calculated on the basis of national poverty lines. These data are used in this exercise instead of the POVCALNET World Bank poverty data (where poverty is measured using international poverty lines) because the SEDLAC data have more observations on poverty over consecutive years. This ensures that periods of changes in poverty can be matched to periods of negative growth in per capita GDP.

Figure 9 presents information on how poverty has changed for all the periods of negative per capita GDP growth ‐ for which poverty data are available (34 in total). Note that

101 See www.depeco.econo.unlp.edu.ar/cedlas/sedlac/ SEDLAC is a joint initiative between the Centro de Estudios Distributivos , Laborales y Sociales ‐ Universidad de la Plata and the LAC region of the World Bank.

64

65

there were no periods of negative per capita GDP growth in the countries for which poverty data are available after 2003. Figure 9 captures multiple dimensions of the poverty response. The figure shows the beginning and end year for all periods of negative per capita GDP growth. The size of the bar shows the magnitude of the decline in per capita GDP and the color shows the change in poverty that occurred during the period of negative growth. Finally, the number under the bar indicates the number of years between the time when poverty stopped falling to when the poverty level recovered to the level of poverty that prevailed before the onset of the decline in per capita GDP. If poverty did not rise over the period of decline in per capita GDP, the recovery period is denoted as zero years.

Table 3 summarizes some of the information from the observations in Figure 9.102 In this table, we divided the size of the GDP declines into ones that could be characterized as large, medium or small103. For each category of decline in per capita GDP, the table presents information on the average cumulative loss in per capita GDP and the average cumulative change in poverty over the period when per capita GDP fell. The table also presents the average number of years it took to recover to the level of poverty that prevailed before the crisis.

Table 3. Effects on Poverty of GDP Shocks

(Using data from 19 LAC countries over period 1981‐2006) Cumulative Loss in Per Capita GDP Large

(> 3 % loss) Medium

(Between 1.5 and 3 percent loss)

Small (Less than 1.5 percent

loss) Average cumulative loss in per capita GDP

‐7.2 ‐1.8 ‐0.6

Average cumulative change in poverty rate

4.4 0.64104 ‐0.08

Average years it takes to recover poverty loss

3 3 1

Countries with projected losses in per capita GDP for 2009 (Consensus forecasts as of March, 2009)

Ecuador (‐3.34%)Mexico (‐3.83%)

Guatemala (‐1.62%) Argentina (‐1.63%) Paraguay (1.82%)

Colombia (‐.37 %) Panama (‐0.8 %)

El Salvador (‐0.93 %) Chile (‐0.97 %)

Nicaragua (‐1.09%) Honduras (‐1.14%) Brazil (‐1.25%)

Dom. Rep. (‐1.42)

102 The data on poverty change and GDP per capita that underpin figure 9 are reported in Annex 4. 103 This classification is based on the size of the cumulative loss in GDP and is not based on the length of the period of negative per capita GDP growth. The exact divisions are somewhat arbitrary, but were made in such a way as to correspond to some of the projected declines in per capita GDP that are forecast for 2009. 104 This average does not include Jamaica, which appears to be an outlier. The Jamaican experience has been a puzzle and is discussed in The Road to Sustained Growth in Jamaica, World Bank (2004) and Osei, P. (2002) , “A Critical Assessment of Jamaica’s National Poverty Eradication Programme”, Journal of International Development, Vol. 14, pp. 773‐88.

66

The average years to recover from a poverty loss are measured from the time when poverty stops increasing to the time when it recovers to the level that prevailed before the onset of the crisis. As a crisis itself lasts typically for one or two years, a country may actually suffer through a period of 4‐5 years before it is able to get back to its position before the fall.

The last row presents the March 2009 projections for per capita GDP for 2009. The GDP growth forecasts are taken from the Consensus Forecasts and these were converted to GDP per capita projections using population data from the World Development Indicators. As is evident from the table, the projected losses of GDP per capita in 2009 for Mexico and Ecuador could already be considered large by historical standards. In these cases, it may take three years to recover the losses in poverty.

Estimating the Impact of the Economic Crisis on Poverty

The previous section simply presented the information on past relations between declines in per capita GDP and past changes in poverty. This information can help frame the expectations for the likely change in poverty. An alternative approach to considering what might be the effect of a given change in per capita GDP is to estimate the elasticity of changes in poverty to changes in per capita GDP and then use the estimated elasticities to predict the change in poverty for a specific projected growth rate for per capita GDP. Poverty elasticities were estimated using a specification similar to Ravallion (1995)105 and Adams (2004)106.

It should be noted that these elasticities make use of the periods of positive increase in per capita GDP growth and declines of poverty. We did estimate the elasticities with a spline to allow the coefficient on GDP growth to take on a different value depending on whether growth was positive or negative. However, the difference was not statistically significant. These results are presented in Annex 3. We also estimated elasticities using the POVCALNET data instead of the SEDLAC data and the differences were not very great. Table 4 reports the different elasticities and the significance level of the estimated elasticities using the SEDLAC data.

Table 4

Elasticities of Changes in Poverty Measure with changes in per capita GDP

Poverty Line Elasticity P value National Extreme* ‐2.63 0.00 National Moderate* ‐1.62 0.00

Note: (*) SEDLAC data; std errors clustered at country level; population weighted point estimates; controls: Log(Gini) and time trend.

The estimated elasticities presented in Table 4 can be used to simulate the impact of the economic crisis on poverty in LAC. This is done by taking forecasted rates of growth of GDP, converting them into rates of growth of GDP per capita and using the elasticities to generate an estimated effect on poverty. The forecast economic growth rates are taken from the March 16th LAC Consensus Forecasts, which compile

105 Ravallion (1995) used the private consumption component from national accounts. Since our main objective here was to use these elasticities to simulate the impact of changes on GDP per capita, we chose to follow the approach of Adams (2004) and others, who have used per capita GDP as one the regressors. 106 The specification is of first difference on the LOG, in order to take into account for country specific fixed effects, and also includes the Gini coefficient and a time trend.

67

data from several sources and report the mean, minimum and maximum expected growth rate for the period.107 The baseline values for poverty rates and numbers are calculated by taking the 2006 and 2007 poverty rates in SEDLAC and projecting them forward to 2008, using the estimated elasticities and the preliminary estimates of 2008 growth rates.

Table 5 reports the estimated impact of the economic crisis on moderate poverty and extreme poverty, using poverty defined by national poverty lines and reported in SEDLAC.108 Using the mean consensus forecast, aggregate poverty rates for LAC are estimated to rise 1.14 points. That would result in 8.3 million more poor people than in 2008. About half of those people that will fall into poverty are in Mexico, about a fifth in Brasil, and the rest are distributed in Argentina, Colombia, Ecuador, Guatemala and Venezuela. The aggregate extreme poverty rate for LAC is estimated to rise 0.53 points to, which would generate an increase of approximately 3.6 million in the number of extreme poor.

As today’s pessimistic forecast seems to be turning rapidly into tomorrow’s mean forecast, it is worth noting the estimated poverty rates and numbers associated with the pessimistic forecast. In this case, aggregate poverty rates are estimated to rise by 2.05 points generating 13.5 million additional poor people in 2009. Aggregate extreme poverty rates, under the pessimistic forecast, are estimated to increase almost one point, generating an increase of 6 million in the number of extreme poor. In the region, the final performance both of growth and poverty will depend on the magnitude of the downturn in industrialized countries and on the speed and effectiveness of anticyclical packages that most countries are already implementing. Figures 10 illustrate that the number of poor projected for LAC in 2009 is beginning to rise to the levels that prevailed back in 2006.

If one compares the projected number of poor people in 2009 to a prediction based on what was expected to be 2009 GDP growth back in January 2008, the increase in the number of people would be 13 million (Table 6). As there was a more optimistic outlook for growth in 2009 when forecasts were made back in January 2008 (see figure 6), the estimated poverty impact is greater for this counterfactual. In other words, now we expect, by the end of 2009 , 13 million more poor people than what would have been observed had past growth been maintained. In the case of extreme poverty, comparing the change to what had been expected back in January 2008 yields an increase of 6.1 million more extreme poor than what had been expected.

107 The GDP growth rates used in the analysis are provided in Annex 5. The estimated poverty impacts for the individual countries are also presented in Annex 5. If a particular country only reported data for the mean, the same number was used on the minimum and maximum scenario. Moreover, a few countries (St Lucia, Haiti and Jamaica) were not covered by the LAC Consensus Forecast, in those cases the World Bank GEP 2009 forecast was used, also on the three scenarios. 108 Similar tables of results are presented in Annex 4, using data from POVCALNET and elasticities estimated from POVCALNET data.

68

Table 5: Poverty Impact of Slowdown in 2009 (changes with respect to observed levels in 2008)

Consensus Forecast for Latin America and the

Caribbean* Mean Pessimist Optimistic Moderate Poverty

Absolute change in incidence (pp) 1.14 2.05 0.58 Percentage change in incidence (%) 12.64% 22.79% 6.39% Absolute change in number of poor (,000) 8,325 13,491 5,144 Percentage change number of poor (%) 5.37% 8.70% 3.32%

Extreme Poverty Absolute change in incidence (pp) 0.53 0.94 0.28 Percentage change in incidence (%) 5.87% 10.40% 3.06% Absolute change in number of poor (,000) 3,603 5,909 2,169 Percentage change number of poor (%) 7.16% 11.75% 4.31%

Note: (*) Consensus Forecast as of May/2009 for Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela.; For the countries in which theres was no Consensus Forecast esimates, the World Bank Forecast as of March 2009 was used; For the countries which Consensus Forecast did not report a minimum or a maximum value, the average reported value was used for both the optimistic and the pessimistic scenario. Countries: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, El Salvador, Honduras, Mexico, Paraguay, Peru, Uruguay, and Venezuela (90% of the region population covered by Povcalnet). Elasticities estimated using Sedlac data (17/nov/2008).

69

Table 6: Poverty Impact of Slowdown in 2009 (as compared to expected poverty levels had past growth rates continued)

Consensus Forecast for Latin America and the

Caribbean* Mean Pessimist Optimistic Moderate Poverty

Absolute change in incidence (pp) 2.30 3.22 1.74 Percentage change in incidence (%) 8.66% 12.09% 6.54% Absolute change in number of poor (,000) 13,015 18,181 9,834 Percentage change number of poor (%) 8.66% 12.09% 6.54%

Extreme Poverty Absolute change in incidence (pp) 1.07 1.48 0.82 Percentage change in incidence (%) 12.70% 17.52% 9.70% Absolute change in number of poor (,000) 6,075 8,381 4,641 Percentage change number of poor (%) 12.70% 17.52% 9.70%

Note: (*) Consensus Forecast as of May/2009 for Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela.; For the countries in which theres was no Consensus Forecast esimates, the World Bank Forecast as of March 2009 was used; For the countries which Consensus Forecast did not report a minimum or a maximum value, the average reported value was used for both the optimistic and the pessimistic scenario. Countries: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, El Salvador, Honduras, Mexico, Paraguay, Peru, Uruguay, and Venezuela (90% of the region population covered by Povcalnet). Elasticities estimated using Sedlac data (17/nov/2008).

Figure 10. Trends in the Number and Projected Number of Poor in Latin America

The aggregate data include actual and forecasted values for Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Paraguay, Peru, Uruguay, and Venezuela.

70

Conclusions

For almost all of the 1980s and 1990s, the number of poor and extreme poor in Latin America and the Caribbean rose. Despite the growth episodes observed in the nineties poverty rates stagnated. The number of poor climbed from 160.5 million in 1981 to 240.6 million by 2002, and of extreme poor from 90 to 114 million. Since 2002 the number of poor has decreased at unprecedented speed – so much so that in 2008 the number of poor is estimated to have fallen to 181.3 million and the number of extreme poor to 73 million. Hence, during the strong growth period 2002‐2008, almost 60 million people were lifted out of poverty, and 41 million left the ranks of extreme poverty. It should be noted that towards 2007 and 2008 there seems to be already a deceleration on the rate of poverty reduction.

Unfortunately, the recent worldwide recession has put an end to that progress and the number of poor are now projected to increase. However, compared to past periods of negative growth, in most cases the current projected declines in GDP have not yet approached the largenegative shocks that Latin America experienced throughout the eighties and late nineties. The large shocks of the past averaged a loss in per capita GDP of 7.2 percent and generated increases in poverty rates that were, on average, 4 percentage points. And, historically, it has proved difficult for countries to recover quickly and get back to the poverty level that prevailed before the shock. On average it has taken 3 years to get back to the poverty level prior to the shock, for all but the smallest negative shocks.As of March 2009, only in Mexico and Ecuador are the projected declines in GDP for 2009 expected to be relatively large. Given the consequente larger increases in poverty rates it might take about three years to recovre from the povrety losses. More countries could be in similar situations as the worldwide recession deepens.

If the projected growth rates are not yet at the level that has prevailed in individual countries in the past, what is noteworthy in this crisis is how it has hit all countries and how rapidly the projected growth rates are trending downwards. Whereas in 2007 and 2008, no country in Latin America was experiencing negative growth in per capita GDP, today 15 countries are projected to have negative per capita GDP growth in 2009. The downward trend in projections for Latin America are following the pattern observed in industrialized countries. Using elasticity estimates and the mean LAC Consensus Forecast for GDP growth , aggregate poverty rates are estimated to rise 1.14 points. That would result in 8.3 million more poor people than in 2008 in Latin America and the Caribbean. About half of those people that will fall into poverty are in Mexico, about a fifth in Brasil, and the rest are distributed in Argentina, Colombia, Ecuador, Guatemala and Venezuela. Aggregate extreme poverty rates are estimated to rise 0.53 points, increase that would generate a rise of approximately 3.6 million in the number of extreme poor. Using a pessimistic forecast – which, if the recent trend continues will turn rapidly into tomorrow’s mean forecast‐ aggregate poverty rates are estimated to rise by two points generating 13 million additional poor people in 2009. Aggregate extreme poverty rates, under the pessimistic forecast, are estimated to increase 0.94 points, generating an increase of almost 6 million in the number of extreme poor.

Annex 1. Construction of regional poverty estimates for Latin America

The World Bank has created estimates of regional poverty estimates for Latin America and other regions of the world using PPP $1.25 and $2 a day international poverty lines, corresponding to extreme and moderate poverty. Based on an analysis of the PPP equivalents of national poverty lines, this note argues that using PPP $2 a day for extreme poverty and PPP $4 a day for moderate poverty would be more appropriate for Latin America.

71

The following table presents information on the PPP $ a day equivalent of national extreme poverty lines across countries and over time. These values were obtained by converting the reported local currency poverty lines into PPP equivalents using the PPP Exchange Rates for household consumption from the 2005 International Comparison Program, adjusted for inflation using the national CPI.

Source: SEDLAC data base, PPP conversion factors from World Bank POVCALNET It is worth noting that the values of the national extreme poverty lines in PPP terms are quite consistent over time for each country. The average coefficient of variation is 0.07 and there is not a great range in the coefficient of variation across countries. The range extends from 0.01 (Chile) to 0.15 (Honduras). There is more of a variation looking across countries, with the lowest value at 1.04 for Nicaragua and the highest at 2.73 for Mexico. The coefficient of variation across countries is 0.35.

The following table presents equivalent values of the PPP $ a day equivalent of national poverty lines, calculated in a similar fashion.

72

Source: SEDLAC data base, PPP conversion factors from World Bank POVCALNET The values are still relatively consistent across time within a country. The average coefficient of variation is 0.06 and the range extends from 0.002 (Panama) to 0.15 (Honduras). However, in terms of the variation across countries, there is considerably greater variation in the national poverty lines in PPP terms across countries in moderate poverty than in extreme poverty. This is because all countries use some variation of a food or caloric based extreme poverty line. After accounting for differences in prices with the PPP adjustment, the remaining differences are due to differences in the combination of food that would yield the minimum requirements. The moderate poverty lines are typically defined by multiplying the extreme poverty line by the inverse Engel coefficient and this varies somewhat more across countries.

The observation that there is a considerable range in the national moderate poverty lines in PPP terms implies that comparisons of estimated national poverty rates are potentially misleading. All comparisons of poverty should be made using the same international PPP line.

We choose to consider extreme poverty as PPP $2 a day, because that is the round figure that minimizes the distance from the observed national extreme poverty lines. Similarly, we choose to consider moderate poverty as PPP $4 a day for the dame reason. However, this does not mean that all countries should necessarily use the $2 a day or $4 a day PPP lines. Nicaragua may want to compare their poverty situation to that of other countries using a PPP $1.25 a day line that more closely approximates their

73

own particular national poverty line. Mexico may choose a different one. The important point is to use the same PPP $ a day international poverty line consistently across all comparisons.

The following three graphs illustrate the relation between GDP per capita in PPP terms and the PPP $ a day equivalent to the national poverty line, national extreme poverty line and the ratio of the moderate to extreme poverty line. The graphs show that there is more variation across countries in the PPP $ a day equivalent of the national poverty lines than there is of the extreme poverty lines.

74

75

Annex 2. Trends in Regional Poverty and Growth in GDP per Capita

Figure A.2: Evolution of poverty and GDP

East Asia ‐ 2USD a day

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Hea

dcou

nt P

over

ty R

ate

0

1000

2000

3000

4000

5000

cons

tant

200

5 in

tern

atio

nal $

2 USD a day GDP per capita, PPP (constant 2005 international $)

East Asia ‐ 4USD a day

60.0

70.0

80.0

90.0

100.0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Hea

dcou

nt P

over

ty R

ate

0

1000

2000

3000

4000

5000

cons

tant

200

5 in

tern

atio

nal $


ECA‐ 2USD a day

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

18.0

20.0

1989 1991 1993 1995 1997 1999 2001 2003 2005

Hea

dcou

nt P

over

ty R

ate

5000

6000

7000

8000

9000

10000co

nsta

nt 2

005

inte

rnat

iona

l $


ECA‐ 4USD a day

25.0

30.0

35.0

40.0

45.0

50.0

1989 1991 1993 1995 1997 1999 2001 2003 2005

Hea

dcou

nt P

over

ty R

ate

5000

6000

7000

8000

9000

10000

cons

tant

200

5 in

tern

atio

nal $


Middle East and North Africa‐ 2USD a day

15.0

17.0

19.0

21.0

23.0

25.0

27.0

29.0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Hea

dcou

nt P

over

ty R

ate

4000

5000

6000

cons

tant

200

5 in

tern

atio

nal $


Middle East and North Africa‐ 4USD a day

50.0

52.0

54.0

56.0

58.0

60.0

62.0

64.0

66.0

68.0

70.0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Hea

dcou

nt P

over

ty R

ate

4000

5000

6000

cons

tant

200

5 in

tern

atio

nal $


Sub‐Saharan Africa‐ 2USD a day

70.0

71.0

72.0

73.0

74.0

75.0

76.0

77.0

78.0

79.0

80.0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Hea

dcou

nt P

over

ty R

ate

1000

1300

1600

1900

cons

tant

200

5 in

tern

atio

nal $


Sub‐Saharan Africa‐ 4USD a day

90.0

91.0

92.0

93.0

94.0

95.0

96.0

97.0

98.0

99.0

100.0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004

Hea

dcou

nt P

over

ty R

ate

1000

1300

1600

1900

cons

tant

200

5 in

tern

atio

nal $


Source: World Bank, World Development Indicators and POVCALNET

76

Annex 3. Alternative Specifications of Poverty Elasticities

Table A3.1 presents the coefficients of the Poverty‐GDP pc elasticity using a linear spline transformation. This exercise allows the estimation the relationship between y and x as a piecewise linear function, which is a function composed of linear segments. In this particular case, the first linear segment represents the Poverty‐per capita GDP elasticity for periods of economics crises (negative per capita GDP change), while the second linear segment represents the Poverty‐per capita GDP elasticity for periods of economic growth (positive per capita GDP change). Two models were estimated, with two different specifications. The first model looked at the extreme poverty elasticity, and the second looked at the moderate poverty elasticity. The two alternative specifications considered the full dataset and all observations but Argentina, given the very particular magnitudes of the changes in this country. All models were estimated using ordinary least squares (OLS) algorithm, with standard errors clustered at the country level.

These models allow us to test the equality of the elasticities during periods of economics crisis and growth. As it can be seen in the last two lines of Table A5.1 none of the specifications allowed the rejection of the hypothesis that the coefficients are identical.

Table A3.1: Poverty Elasticities using Splines (OLS)

Extreme Poverty Moderate Poverty

Full Without Argentina Full Without Argentina d_lngdppc: (.,0) ‐2.481 ‐1.494 ‐1.332 ‐0.732 (1.221) (0.822) (0.684) (0.392) d_lngdppc: (0,.) ‐3.767** ‐3.965* ‐2.423* ‐2.547* (1.178) (1.315) (0.929) (1.037) d_lngini3 ‐0.015 ‐0.011 ‐0.010 ‐0.008 (0.008) (0.008) (0.006) (0.006) Year ‐0.001 ‐0.002* ‐0.000 ‐0.001 (0.001) (0.001) (0.001) (0.001) Constant 2.534 4.426* 0.698 1.806 (1.862) (1.646) (1.148) (1.053) Adj.R‐squared 0.335 0.301 0.348 0.310 Obs. (unweighted) 107 102 96 91 Test: d_lngdppc_a1 ‐ d_lngdppc_a2 = 0 F‐stat 0.323 1.346 0.529 1.632 P‐val 0.580 0.271 0.480 0.226 Note: clustered standard errors; population weighted. Inference: * p<0.05, ** p<0.01, *** p<0.001

77

Table A3.2 presents the same models presented in Table A3.1, estimated as median regression. The advantage of this model over the OLS is the fact that it is much more robust to the presence of outliers. As it can be seen the moderate poverty elasticities for periods of economics crisis [d_lngdppc: (.,0)] are not very different, however the same is not true in the case of extreme poverty.

Table A3.2: Poverty Elasticities using Splines (median regression) Extreme Poverty Moderate Poverty

Full Without Argentina Full Without Argentina d_lngdppc: (.,0) ‐3.721*** ‐1.794*** ‐1.671*** ‐0.932*** (0.000) (0.011) (0.026) (0.000) d_lngdppc: (0,.) ‐2.559*** ‐3.578*** ‐1.916*** ‐2.020*** (0.000) (0.008) (0.021) (0.000) d_lngini3 ‐0.019*** ‐0.013*** ‐0.008*** ‐0.006*** (0.000) (0.000) (0.000) (0.000) Year ‐0.002*** ‐0.002*** 0.002*** 0.002*** (0.000) (0.000) (0.000) (0.000) Constant 3.346*** 4.306*** ‐3.775*** ‐3.152*** (0.000) (0.054) (0.140) (0.000) Obs. (unweighted) 107 102 96 91 Note: clustered standard errors; population weighted. Inference: * p<0.05, ** p<0.01, *** p<0.001 Annex 4. Estimates of Poverty Impact Using POVCALNET data and Elasticities calculated using POVCALNET data

This annex presents tables similar to tables 5 and 6 in the text, but using POVCALNET data and elasticities calculated from POVCALNET data instead of the SEDLAC data and elasticities calculated from the SEDLAC data (Table A6.1). For this case moderate poverty corresponds to PPP $4 a day and extreme poverty corresponds to PPP $2 a day. The estimated impacts are lower than in the case of the estimates based on the SEDLAC data. This is due to lower estimated elasticities. It is probably the case that calculating poverty changes over a longer period results in less sensitivity to changes in per capita GDP.

Table A4.1 Elasticities of Changes in Poverty Measure with changes in per capita GDP

Poverty Line Region GDP per capita P value $ 2 a day PPP LAC ‐1.37 0.044 World ‐1.37 0.000 $ 4 a day PPP LAC ‐0.96 0.001 World ‐0.14 0.708

Note: (*) SEDLAC data; std errors clustered at country level; population weighted point estimates; controls: Log(Gini) and time trend.

78

Table A4.2. Poverty Impact of Slowdown in 2009 (4 US dollars a day PPP) Consensus Forecast for Latin America and the Caribbean*

Mean Pessimistic Optimistic (4 US dollars a day PPP) Absolute change in incidence (pp) 0.31 0.58 ‐0.07 Percentage change in incidence (%) 2.40% 4.42% ‐0.55% Absolute change in number of poor (,000) 3,990.02 5,489.36 1,801.88 Percentage change number of poor (%) 2.20% 3.03% 0.99% (2 US dollars a day PPP) Absolute change in incidence (pp) 0.13 0.26 ‐0.08 Percentage change in incidence (%) 0.99% 1.99% ‐0.59% Absolute change in number of poor (,000) 1,631.77 2,370.05 453.30 Percentage change number of poor (%) 2.23% 3.23% 0.62%

79

Annex 5. GDP Growth Forecasts used in estimating the Poverty Impact of the Worldwide Recession

Table A5.1 GDP Growth

LAC Consensus Forecasts***

2007 * 2008 *

WB GDP Growth Forecast 2009** (from WB GEP 2008)

WB GDP Growth Forecasts 2009 ** (from WB GEP 2009)

IMF Forecasts 2009 Mean Pessimistic Opptimistic

Argentina 8.70% 6.60% 4.70% 1.50% 0.00% ‐0.69% ‐3.80% 2.10% Bolivia 4.60% 4.10% 4.20% 3.58% 4.00% 1.80% 1.80% 1.80% Brazil 5.40% 5.20% 4.50% 2.83% 1.80% ‐0.05% 0.00% 1.80% Chile 5.10% 4.20% 5.00% 3.45% 2.20% 0.01% ‐1.45% 1.60% Colombia 8.20% 3.70% 4.80% 2.64% 2.00% 0.85% ‐1.00% 2.70% Costa Rica 6.77% 4.00% 4.90% 3.92% 1.00% 0.00% 0.00% 0.00% Dominican Republic 8.50% 5.20% 4.80% 2.57% 1.80% 0.00% 0.00% 0.00% Ecuador 1.90% 2.50% 2.70% 0.80% 1.00% ‐2.30% ‐2.30% ‐2.30% El Salvador 4.20% 2.00% 4.00% 2.65% 2.50% 0.40% 0.40% 0.40% Guatemala 5.70% 2.80% 5.00% 3.13% 3.00% 0.80% 0.80% 0.80% Guyana 5.50% 4.80% 3.50% 4.00% 4.50% 4.00% 4.00% 4.00% Haiti 3.50% 3.00% 4.00% 3.80% 2.50% 3.80% 3.80% 3.80% Honduras 6.30% 3.10% 4.75% 3.96% 2.00% 0.60% 0.60% 0.60% Jamaica 1.20% 0.90% 3.10% 0.80% ‐1.00% 0.80% 0.80% 0.80% Mexico 3.20% 2.00% 3.60% 1.12% ‐0.30% ‐2.82% ‐5.00% ‐1.50% Nicaragua 3.50% 2.20% 4.50% 1.51% 1.50% 0.20% 0.20% 0.20% Panama 11.50% 7.80% 7.10% 3.27% 7.80% 0.80% 0.80% 0.80% Paraguay 6.80% 4.20% 3.80% 3.01% 2.00% ‐0.10% ‐0.10% ‐0.10% Peru 9.00% 8.50% 6.10% 5.19% 6.00% 4.01% 1.80% 6.20% Trinidad and Tobago 5.50% 6.20% 7.10% 6.60% 4.50% 6.60% 6.60% 6.60% Uruguay 7.40% 4.67% 3.80% 2.81% 2.50% 1.40% 1.40% 1.40% Venezuela, RB 8.40% 5.30% 4.24% 0.98% ‐2.00% ‐0.24% ‐4.10% 3.80% LAC 5.79% 4.48% 4.38% 2.18% 0.98% ‐0.63% ‐2.10% 1.19%

Notes: (*) Source World Bank, World Development Indicators; (**) Source: World Bank Global Economic Forecasts, 2008 and 2009. The GEP 2009 forecasts were made in November 2008; (***) Consensus Forecast as of March 16th; For the countries in which there was no Consensus Forecast estimate, the World Bank Forecast as of March 2009 was used; For the countries which Consensus Forecast did not report a minimum or a maximum value, the average reported value was used for both the optimistic and the pessimistic scenario.

80

Table A5.2 Per Capita GDP Growth LAC Consensus Forecasts***

2007* 2008*

WB GDP Growth Forecast 2009** (from WB GEP 2008)

WB GDP Growth Forecasts 2009 ** (from WB GEP 2009)

IMF Forecasts 2009 Mean Pessimistic Opptimistic

Argentina 7.64% 5.60% 3.72% 0.55% ‐0.93% ‐1.62% ‐4.70% 1.15% Bolivia 2.77% 2.31% 2.41% 1.80% 2.21% 0.05% 0.05% 0.05% Brazil 4.16% 3.95% 3.26% 1.61% 0.59% ‐1.24% ‐1.19% 0.59% Chile 4.07% 3.18% 3.98% 2.44% 1.20% ‐0.96% ‐2.41% 0.61% Colombia 6.21% 2.44% 3.53% 1.40% 0.76% ‐0.37% ‐2.20% 1.46% Costa Rica 4.79% 2.52% 3.41% 2.44% ‐0.44% ‐1.42% ‐1.42% ‐1.42% Dominican Republic 6.97% 3.72% 3.32% 1.13% 0.36% ‐1.41% ‐1.41% ‐1.41% Ecuador 0.85% 1.44% 1.64% ‐0.24% ‐0.04% ‐3.31% ‐3.31% ‐3.31% El Salvador 2.82% 0.65% 2.62% 1.29% 1.14% ‐0.93% ‐0.93% ‐0.93% Guatemala 3.15% 0.34% 2.49% 0.66% 0.53% ‐1.61% ‐1.61% ‐1.61% Guyana 5.47% 4.87% 3.57% 4.07% 4.57% 4.07% 4.07% 4.07% Haiti 1.42% 1.23% 2.21% 2.02% 0.73% 2.02% 2.02% 2.02% Honduras 4.46% 1.31% 2.94% 2.16% 0.23% ‐1.14% ‐1.14% ‐1.14% Jamaica 1.67% 0.53% 2.72% 0.43% ‐1.36% 0.43% 0.43% 0.43% Mexico 2.25% 0.97% 2.56% 0.10% ‐1.30% ‐3.79% ‐5.96% ‐2.49% Nicaragua 2.86% 0.88% 3.15% 0.21% 0.19% ‐1.09% ‐1.09% ‐1.09% Panama 9.44% 6.09% 5.40% 1.63% 6.09% ‐0.80% ‐0.80% ‐0.80% Paraguay 4.60% 2.42% 2.03% 1.25% 0.26% ‐1.81% ‐1.81% ‐1.81% Peru 7.78% 7.30% 4.92% 4.02% 4.82% 2.85% 0.67% 5.02% Trinidad and Tobago 5.63% 5.83% 6.73% 6.23% 4.14% 6.23% 6.23% 6.23% Uruguay 7.29% 4.54% 3.67% 2.69% 2.37% 1.27% 1.27% 1.27% Venezuela, RB 6.64% 3.59% 2.55% ‐0.66% ‐3.59% ‐1.86% ‐5.66% 2.11% LAC 4.58% 3.30% 3.20% 1.03% ‐0.16% ‐1.75% ‐3.21% 0.05% Notes: (*) Source World Bank, World Development Indicators; (**) Source: World Bank Global Economic Forecasts, 2008 and 2009. The GEP 2009 forecasts were made in November 2008; (***) Consensus Forecast as of March 16th; For the countries in which there was no Consensus Forecast estimate, the World Bank Forecast as of March 2009 was used; For the countries which Consensus Forecast did not report a minimum or a maximum value, the average reported value was used for both the optimistic and the pessimistic scenario.

81

4. LABOR MARKETS AND THE CRISIS IN LATIN AMERICA AND THE CARIBBEAN (A PRELIMINARY REVIEW FOR SELECTED COUNTRIES)109

Samuel Freije‐Rodríguez and Edmundo Murrugarra

June 2009*

Abstract

Countries in Latin America and the Caribbean are experiencing the impact of the international financial crisis on labor markets across different dimensions, such as employment, wages and the quality of labor market arrangements. This note reviews a selected group of countries to assess the speed and severity of labor market impacts. It identifies patterns in the changing labor market conditions, such as specific sectors or types of workers being affected. It also describes countries’ preparedness and capacity to respond to the crisis and the specific policy responses being implemented. The review finds a large variation in impacts and responses in the context of increases in unemployment rates that range from 0.4 to 2.1 percentage points. The impacts of the crisis are evolving rapidly but seem to have a more noticeable negative effect among salaried workers in Brazil and Chile whereas in Colombia non‐salaried workers have been affected the most. Mexico shows both types of workers as being seriously hit by the recession.

LAC Responses on Labor Markets: Information and Capacity

Countries in Latin America and the Caribbean are experiencing the impact of the international financial crisis on labor markets across different dimensions, such as employment, wages and the quality of labor market arrangements. The policy response to these impacts is reflecting both Governments’ ability to monitor the actual impacts of the crisis, as well as their capacity to consistently (and effectively) respond with programs and interventions.

This note reviews a selected group of countries to assess the speed and severity of labor markets impacts. It identifies patterns in the changing labor market conditions, such as specific sectors or types of workers being affected. It also describes the policies announced or adopted to respond to the crisis. At this stage, the note provides only a systematic description of labor market facts and policies. This groundwork stock‐taking is necessary for future analytical work. In this regard, the evidence presented below highlights the importance of further studies to understand two main questions: i) why different countries in the region show different labor market adjustments to the crisis and ii) what explains the adoption and effectiveness of the wide range of policy responses adopted by governments.

*LCR CRISIS BRIEFS SERIES. 109 This is the first of a series of notes to monitor the status of the labor market during this crisis period. Forthcoming issues will extend the study to other countries of the area, and complement the analysis of the countries included with further data. It will also report advances in labor market policies adopted by governments as the crisis evolves. This note was produced by a joint team from the Social Protection and Poverty and Gender Units including Georgina Pizzolitto, Diana Hincapie, Pablo Acosta and, Rodolfo Beazley, under the guidance of Helena Ribe and Jaime Saavedra.

82

Understanding the size and speed of the impacts of the crisis on labor markets depends on the availability of high frequency data. Countries included in this review fall into two categories: those with higher frequency data (i.e., monthly or quarterly labor statistics) and those with low frequency data (i.e., annual or bi‐annual statistics). For the latter group, however, leading indicators such as Social Security records provide some indicative information on the nature of the labor market response to the crisis.

Current Labor Market Impacts of the Global Crisis

Regarding the impact of the crisis, all countries with timely data show an increase in unemployment rates. Chile registers an increase in the unemployment rate of 2.1 percentage points for the latest month in record with respect to the same month one year ago. Colombia and Mexico register near one percentage point changes. Brazil shows an even smaller increase (0.4 percentage points) but a very rapid rise in recent months. These rises in unemployment are accompanied by important falls in net job creation since mid 2008. Countries differ on the main force that explains this fall in net employment creation. On one hand, Brazil, Chile and Mexico, register a deep fall in salaried net job creation since mid 2008. In Colombia, on the other hand, it is net job creation of non‐salaried jobs (i.e. employers, self‐employed and unpaid workers) what is driving the fall in total employment. All countries show a net destruction of non‐salaried jobs towards the end of 2008 or the beginning of 2009, although the Mexican is the largest fall in both relative and absolute terms. In general, non‐salaried workers represent a large portion of total informal employment. Hence, it can be argued that the crisis is having a more noticeable negative effect on the formal sector in Brazil and Chile whereas in Colombia the informal sector has been the most affected. Mexico shows both sectors been seriously hit by the recession.

A preliminary distributive analysis of the Mexican case shows that the poorest households and the households in the North of the country (where tradable activities concentrate) are the most affected by the crisis. The evolution of the impacts over time is also important. In Brazil, the early labor market impacts of the crisis were clearly observed in industrial states like Sao Paulo, but are now larger in the urban areas of poorer Northeast states like Recife and Salvador.

The impact of the crisis on earnings and wages is mixed. While Brazil and Chile show growing real earnings, Colombia registers falling real wages both in retail commerce and blue‐collar manufacturing (two of the most common occupations). Mexico shows a dual pattern whereby blue‐collar manufacturing workers have steady real wages whereas workers in retail commerce have falling real earnings. A pattern emerges whereby the more open economies like Chile, Colombia and Mexico have all endured an important shock due to the international crisis, but Chile has been able to prevent a fall in real wages, whereas Colombia and Mexico have endured declines in real wages, for at least some sectors. On the other hand, in a less open economy like Brazil the impact in the labor markets appears limited to losses of employment in the export oriented activities with no significant impact on earnings.

For other countries (e.g. El Salvador, the Dominican Republic), timely data is lacking but leading indicators from Social Security data suggest major job losses in recent months. However, given the limited size of formal employment, the full impact of the crisis cannot be fully ascertained.

Current Policy Responses

In relation to the policies actions adopted in response to the crisis, countries like Brazil, Chile or Mexico are taking advantage of labor market policy instruments already in place, such as unemployment insurance, or wage subsidies to keep or hire workers,. In these countries, the duration of unemployment insurance has been extended either in specific sectors or across sectors aiming at protecting formal workers from longer unemployment spells. Wage subsidies are being utilized for targeted and

83

vulnerable groups. In Chile, the wage subsidy is targeted at the youth with earnings below a certain threshold, and with formal employment. In Colombia, the (implicit wage) subsidy is broader since the Government established general payroll tax holidays for new small and medium size firms. Still, the actual coverage of these policy changes (unemployment insurance, wage subsidies) is limited to the formal sector, and it is unclear how much the take up for these measures will be. There is also renewed attention amongst policy makers to finding ways to protect or create jobs, and several are considering temporary employment programs (TEP), although countries are still dealing with implementation issues (Mexico). Minimum wages, a traditional labor market policy variable, have not been a key instrument in managing the current global crisis. Minimum wage increases have responded instead to the planned periodic adjustment, although in some cases (like in El Salvador), the adjustment took place earlier than planned to protect workers’ purchasing power due to high inflation during 2008. Finally, firms are receiving special subsidies and loans or benefiting from corporate tax reductions in Chile and Mexico in order to protect employment. While some of these interventions are aimed at providing a finance cushion to small and medium enterprises, their coverage, take up and impact cannot be fully assessed still.

Table 1: LAC Labor Market Interventions (Changes to policies since Oct. 2008)

MinimumWage Training

Unemp. Insurance

Wage subsidies*

Public Works

Argentina x

Bolivia x x x Brazil x x

Chile x(1) x x

Colombia x

Guatemala x

El Salvador x

Honduras x x Jamaica x x x

Mexico x x x

* This includes wage subsidies and other labor cost reductions. (1) Tax credit and leave for training activities. Source: Crisis Policy Response ‐ LAC Social Protection Unit.

Latin America in International Perspective

When compared to other countries in the world with available data, Latin American countries are not among the worst hit by the crisis. Table 2 includes year‐to‐year changes in open unemployment rates as an indicator of the impact of the crisis on the labor market for a selection of countries. Spain, The United States and small open economies/cities in Asia (Singapore, Hong‐Kong and Taiwan‐China) show an increase of more than 50 percent in their open unemployment rates in the past year. On the other hand, most Latin American countries register changes in their unemployment rates of less than 30 percent (Chile and Mexico) and even at or below the 10 percent mark (Brazil, Colombia, Peru and Uruguay).

Absolute changes are also a good indicator of the severity of the impact. In this regard, Chile is the only country with a rise in the open unemployment rate of more than 2 percentage points, so far. This mark is similar to the increases observed in Hong‐Kong and Taiwan, but still well below the record marks of

84

Spain (7.9 percentage points), the U.S. (3.9 percentage points), Turkey (3.9 percentage points), Canada (2.6 percentage points) and Sweeden (2.3 percentage points).

Table 2: Change in Open Unemployment Rates in selected countries

rate period rate periodin percentage

pointsin percentage

changeSpain 9.6 Q1 2008 17.5 Q1 2009 7.9 82%USA 5.2 May‐08 9.1 May‐09 3.9 75%Singapore 2.6 Q1 2008 4.4 Q1 2009 1.8 69%Hong‐Kong 3.3 Mar‐08 5.3 Mar‐09 2.0 61%Taiwan (China) 3.8 Apr‐08 5.8 Apr‐09 2.0 51%Canada 6.2 May‐08 8.8 May‐09 2.6 42%Sweden 6.0 Apr‐08 8.3 Apr‐09 2.3 38%UK 5.2 Q1 2008 7.1 Q1 2009 1.9 37%Turkey 11.6 Feb‐08 15.5 Feb‐09 3.9 34%Australia 4.3 Apr‐08 5.6 Apr‐09 1.3 30%Mexico 4.0 Q1 2008 5.1 Q1 2009 1.1 28%Chile 7.7 Mar‐08 9.8 Mar‐09 2.1 28%Thailand 1.7 Q1 2008 2.1 Q1 2009 0.4 27%Norway 2.5 Q1 2008 3.1 Q1 2009 0.6 24%Hungary 8.0 Q1 2008 9.7 Q1 2009 1.7 21%Portugal 7.6 Q1 2008 8.9 Q1 2009 1.3 17%Peru 7.1 Mar‐08 7.8 Mar‐09 0.7 10%Uruguay 7.6 Apr‐08 8.3 Apr‐09 0.7 10%Morocco 9.5 Q4 2007 10.4 Q4 2008 0.9 9%Colombia 12.1 Feb‐08 12.9 Feb‐09 0.8 6%

Germany (1) 7.8 May‐08 8.2 May‐09 0.4 5%Brazil 8.6 Mar‐08 9.0 Mar‐09 0.4 5%Philippines 7.4 Q1 2008 7.7 Q1 2009 0.3 4%Poland 8.1 Q1 2008 8.3 Q1 2009 0.2 2%South Africa 23.5 Q1 2008 23.3 Q1 2009 ‐0.2 ‐1%Egypt 9.1 Q4 2007 8.8 Q4 2008 ‐0.3 ‐3%

Source: International Labor Office, LABORSTA‐Internet

initial final annual change

85

ANNEX: Summary of recent country experiences

BRAZIL

Labor market facts

The global economic slowdown is having a sharp impact on Brazilian labor markets but recovery may occur soon. After years of sustained employment creation and falling unemployment, the unemployment rate in Metropolitan Areas experienced its largest increase ever from 6.8 in December 2008 to 9.0 in March 2009 (Figure 2). Most of the net job losses so far seem to be among non‐salaried workers, but there is an important decline in net job creation among salaried workers since mid 2008 (see Figure 3) which is consistent with previous findings that “…the countercyclical rise in unemployment and informality is driven primarily by a reduction in hiring in the formal sector, rather than increased labor shedding…” (see LCR Crisis Brief by Bill Maloney).110 Wages in the informal sector, which are less likely to be affected by minimum wage regulations, seem to be adjusting downwards during recent months, partly due to the high food inflation during 2008.

• Unemployment rates had a large increase in January 2009, in a period of historically low levels. While January is a month with seasonal increases in unemployment, 2009 showed the largest jump (21 percent) in the number of unemployed, compared to the increases in January 2007 (10.7 percent) or January 2006 (10.6 percent).111 Still, this sudden increase in the number of unemployed in urban Brazil is taking place when the unemployment rate is at historical low levels: the average unemployment rate between October and December 2008 was 7.5 percent, the lowest since 2002. This recent increase in unemployment has been particularly acute in Sao Paulo where the unemployment rate reached 9.4 percent (compared to December’s 7.1%). By February the unemployment rate stayed around 8.4 percent and in March the number of unemployment insurance requests actually declined, suggesting a gradual improvement in labor market outcomes.

• The sharp impact on labor markets is primarily driven by the employment reduction in the informal labor market. Labor market informality in Brazil has been traditionally countercyclical, and was declining during the recent fast growth period. Since the onset of the crisis, the number of informal workers112 in Metropolitan Areas was reduced by 3.2 percent (90 thousand workers) between January 2008 and January 2009, compared to an increase in the number of formal workers113 by 4.5 percent (about 400 thousand workers) in the same period.

• After an early ‐‐ and substantial – net employment reduction in the formal sector, labor demand shows signs of stability by March 2009. According to the roster of formal workers in the private sector (Caged),114 close to 800 thousand formal jobs were lost between November 2008 and January 2009 ‐‐ with a peak of 654 thousand in December 2008. Most of the decline in formal jobs took place because of the sudden reduction in hiring rather than firing, especially in States like Sao Paulo

110 The number of underemployed workers ‐‐ as measured by those working an insufficient number of hours – increased by 11 percent between January 2008 and January 2009, or an additional 60,000 underemployed workers, suggesting other margins of adjustment to the crisis. This rising underemployment is better observed in Sao Paulo, Rio de Janeiro and Belo Horizonte. Still, underemployment as measured here, represent only 700,000 workers compared to more than 21 million workers in the PME population. 111 Data from Pesquisa Mensual de Emprego (PME). 112 Informal workers are defined as those without Employment Card (sem Carteira de Trabalho assinada). Other analyses could include those self‐employed (conta propia). 113 Formal workers are defined as those with Employment Card (com Carteira de Trabalho assinada). 114 Caged stands for Cadastro Geral de Empregados e Desempregados and excludes informal sector workers.

86

and Minas Gerais.115 In 2008, the average monthly number of workers hired and fired in Sao Paulo was around 425 and 358 thousand respectively, but in December the number of hired workers dropped to only 271 thousand, causing a major drop in formal jobs. By March 2009, the formal job losses seem to have halted overall, and in fact may have reverted in some states, but not including the industrial sector in Sao Paulo. In April 2009, unemployment figures in Sao Paulo finally dropped, representing 38 thousand less unemployed in urban Sao Paulo.

• In addition, informal workers are facing declining real wages in recent months. Average real wages in urban areas have increased during 2008, and real wages in January 2009 were almost 6 percent higher than those in 2008. These higher real wages hide important differences across types of employees, since those informal workers (sem Carteira) saw their wages decline during the 2008 year and in January 2009 by 8 and 3 percent, respectively.

115 This pattern is also identified as a stylized fact by Bosch and Maloney (2009).

Brazil Labor Market Indicators (2006‐2009)

Figure 1

55

56

56

57

57

58

58

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

Perc

enta

ge o

f the

wor

king

age

pop

ulat

ion

Months

Activity rate

Activity rate 12 per. Mov. Avg. (Activity rate)

Figure 2

10.40

10.10

8.60

9.00

6

8

10

12

200

6

Jul

200

7

Jul

200

8

Jul

200

9

Per

cen

tage

of t

he

acitv

epo

pula

tion

months

Unemployment rate (Recife, Salvador, Belo Horizonte, Rio de Janeiro, Sao Paulo, Porto Alegre)

Figure 3

-200

-100

0

100

200

300

400

500

600

700

800

900

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

Thou

sand

of w

orke

rse

Months

Net job creation/destruction

salaried workers non-salaried workers

Figure 4

600

700

800

900

1,000

1,100

1,200

1,300

1,400

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

Rea

l inco

me

for p

rivat

e se

ctor

wor

kers

(R

$ of

Mar

ch 2

009)

Months

Real income of private sector workers

Private sector workers with social security Private sector workers without social security

87

Recent Labor Market Policies

The rapid increase in unemployment at the beginning of the 2009 and the lower real due to the 2008 inflation made the Government take two actions related to labor markets:

• First, to address the sharp increase in unemployment the Government extended the unemployment compensation period (Seguro Desemprego, SD) for two additional months for the “most affected sectors and states.” Originally the SD covered from three to five months and it was extended to three to seven installments. Every month, around 600 thousand people receive the benefit, and in 2008, the total cost was R$13.8 billion. The additional payment is expected to cost R$2.2 billion and only those laid off after December 2008 are eligible. Funding for these benefits come from the FAT (Fundo de Amparo ao Trabalhador) which might use up to 10 percent of its technical reserves.

• Second, the Government increased in February the minimum salary from R$415 to R$465 (about 6.4 percent) following its periodic adjustment pattern. The minimum salary not only applied to working individual but it also defines the benefit amount for several social security and social assistance transfers. This increase is expected to represent R$24.3 billion in additional salaries. The R$50 increase will impact the Federal Government’s in R$8.5 billion (about 0.3 percent of GDP), mostly affecting the social security accounts with additional R$7.8 billion. Other benefits also linked to the minimum wage, like salary bonus (abono salarial) and unemployment compensation will also be readjusted. Overall, the labor markets policy response in Brazil is addressing the vulnerable population affected by the decline in external demand and facing unemployment. Other interventions like housing construction and increase social assistance transfers (Bolsa Familia) are part of the measures of the Government.

CHILE

Labor market facts

Unemployment rates have been increasing, quarter‐to‐quarter, since the first quarter of 2008, changing the downward trend that these rates showed during the previous two years (see Figure 6). During the quarter ending in April 2009, unemployment rate was at 9.8 percent, higher than in the same period of 2008 (7.6 percent) and of 2007 (6.8 percent). On the other hand, the activity rate, has stayed at the same level than in April last year (i.e., around 56.2 percent of the working age population) (Figure 5)

Employment creation came to a halt in February 2009. After three years of annual growth of employment between 1 and 4 percent every month, the quarter ending in February 2009 showed an employment growth rate of 0.1 percent. Since March 2009 there has been net job destruction of around 30 thousand jobs per month. Salaried jobs have declined steeply going from an annual creation of around 200 thousand up until November 2008 to only 41 thousand in February 2009 and net job losses of nearly 40 thousand in the quarter ending in April 2009.. The number of non‐salaried workers has had an annual decline of around 40 thousand during the end of 2008, but this figure does not differ much from what had been seen in previous years and have actually had a small rise in recent months (see Figure 7 ). Hence, most of the deceleration in employment creation is due to an abrupt reduction in salaried employment creation.

88

Interestingly, real wages have, during the last three months, reverted the downward trend observed for most of year 2008 (see Figure 8) although some occupations (mostly in the service sector) have lower real wages than in January 2007.

Chile Labor Market Indicators (2006‐2009)

Figure 5

51.0%

53.0%

55.0%

57.0%

2004 Ju

l

2005 Ju

l

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

perc

enta

ge o

f th

e w

orki

ng

popu

latio

n

month

activity rate

12 per. Mov. Avg. (activity rate)

Figure 6

8.6

6.8

7.6

9.8

4.0

6.0

8.0

10.0

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

perc

enta

ge o

f th

e la

bor f

orce

month

Unemployment Rate

Figure 7

-200.00

-100.00

0.00

100.00

200.00

300.00

400.00

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

thos

uan

ds o

f wor

kers

Net job creation/destruction (year-to-year)

non-salaried workers salaried workers

Figure 8

96.00

98.00

100.00

102.00

104.00

106.00

108.00

2006

Ap

ril

July

Oct

obr

e

2007

Ap

ril

July

Oct

obr

e

2008

Ap

ril

July

Oct

obr

e

2009

Abr

il

ind

ex

( 20

06

=1

00

)

month

real earnings per hour


• Chile launched one of the most ambitious fiscal stimulus plans to buffer the impacts of the global economic crisis. In addition to a public investment plan for US$ 1.485 billion (1 percent of GDP) and tax reductions or holidays accounting for up to US$ 1.455 billion (another 1 percent of GDP), president Bachelet announced in early January 2009 several measures related to labor markets:

• First, a new Youth Employment Subsidy Law (30 Jan 09) provides a 30 percent subsidy of the annual income for those individuals aged between 18 and 24, with finished secondary education and working in a formal position. It only covers workers with monthly (annual) incomes below

89

Ch$360,000 (aprox. US$ 600). This Law also includes extensions to maternity benefits and additional leave associated to training activities.

• Second, in January 2009 a Law was approved providing for temporary income tax reductions for individuals and tax credits for firms that carry out training activities with their workers. Also, the law includes an extraordinary benefit of Ch $ 40,000 (aprox. US$ 67). For families and individuals that are beneficiaries of certain social programs (Subsidio Familiar. Asignación Familiar, Chile Solidario, Asignación Maternal). More recently, the Government has passed legislation to facilitate the access to credit for medium and small enterprises.

• In early 2008, and not related to the current international crisis, the Chilean Government had adopted important wage measures. In May it approved a bonus of 20.000 pesos (approximately 14% of the minimum wage) for workers with low incomes. Then, in July, it increased the minimum wage from 144000 to 159000 pesos (that is from US$ 291 to US$ 321, approximately).

COLOMBIA

Labor market facts

Recent unemployment rates in Colombia show a deterioration with respect to similar periods in years before (see Figure 10). Unemployment rate in March 2009 was 12.0 percent, 0.8 percentage points above the year before and 0.1 percentage points above the mark for 2007. The January rate, usually the highest in the year, reached a level not seen since 2005. Participation rates have been on the rise. Active labor force represents 60.7 percent of the percentage of working age population for the month of March 2009. This is one of the highest rates recorded in the country in the last three years.

Net job creation was in decline most of year 2008 because of a net destruction of salaried workers and a by declining net job creation of non‐salaried workers. However, since the last quarter of 2008, the growth of non‐salaried workers has declined while salaried jobs have been on the rise (see Figure 11).

On the other hand, real wages show an important deceleration since the middle of 2007 for retail commerce and since early 2007 for blue‐collar workers in manufacturing (see Figure 12). This fall in real wages is associated to consumer inflation being above 7% since June 2008, while nominal wages have grown at a rate below 5% in both retail and manufacturing.

Recent Labor Market Policies In a recent intervention during a meeting of the Inter‐American Development Bank, President Alvaro Uribe announced that Colombia will approach employment protection with a focus on investment, firm activity, and good quality jobs. In particular, President Uribe mentioned three main policy interventions to protect employment during the crisis:

• Temporary reductions in para‐fiscal contributions for small and medium size enterprises, during their first three years of operation, in an effort to revitalize the creation of small and medium firms.

90

Colombia Labor Market Indicators (2006‐2009)

Figure 9

54

55

56

57

58

59

60

61

62

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

Perc

enta

ge o

f wor

king

age

pop

ulat

ion

Months

Activity Rate

Activity Rate 12 per. Mov. Avg. (Activity Rate)

Figure 10

11.3

11.911.2

12.

8

9

10

11

12

13

14

15

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

Perc

enta

ge o

f act

ive

lab

or p

opu

latio

n

Unemployment Rate

Figure 11

-1500

-1000

-500

0

500

1000

1500

2006

2007

2008

2009

Net Job creation/destruction


Figure 12

-6.0%

-4.0%

-2.0%

0.0%

2.0%

4.0%

6.0%

8.0%

2006 Ju

l.

2007 Ju

l.

2008 Ju

l.

2009

Ann

ual p

erce

ntag

e ch

ange

Months

Change in real wages

change in real wages for blue collar workers in manufacturing (YoY)

change in real wages for workers in retail commerce (YoY)

• To enhance the employability of vulnerable individuals, an additional 250,000 vacancies in SENA (a

technical training institution) will be made available for individuals between 18 and 30 years. The estimated cost of this training effort is C$250 billion (0.7 percent of GDP). An important stated element in the Colombian strategy is the need to maintain workers within the formal sector (affiliated to Social Security).

• In addition to the large public investment program included in the budget, the Government, through the Export Bank of Colombia (Bancoldex), will launch C$5 billion facility for funding small and medium enterprises.

91

EL SALVADOR

Labor market facts

Unemployment and Informal employment rates in El Salvador are produced annually and the most recent figures corresponds to year 2007, making impossible a short term inspection of the labor market. Unemployment has been quite stable around the 7% for the last 5 years, with a peak for youth unemployment in 2005 (see Figure 13). Informal employment, on the other hand, has declined since 2004, although it is still very high (see Figure 14).

El Salvador Labor Market Indicators (2006‐2009)

Figure 13

0

2

4

6

8

10

12

14

16

2001 2002 2003 2004 2005 2006 2007

Per

cent

age

Unemployment rates

15-22 years old 23-30 years old31-64 years old 15-64 years old

Figure 14

0

20

40

60

80

100

2001 2002 2003 2004 2005 2006 2007

Percen

tage

Workers without formal labor contracts

15‐30 years old 31‐64 years old

Figure 15

550000

580000

610000

640000

670000

700000

59

60

61

62

2001 2002 2003 2004 2005 2006 2007

SS

Con

tribu

tors

% e

mpl

oyed

for 1

0+ y

ears

old

Employment rates and social security contributors

Employment Rate Contributors to Social Security

Figure 16

-2%

0%

2%

4%

6%

8%

10%

Jan

01M

ay

Sep

Jan

02M

ay

Sep

Jan

03M

ay

Sep

Jan

04M

a y

Sep

Jan

05M

ay

Sep

Jan

06M

ay

Sep

Jan

07M

ay

Sep

Jan

08M

a ySe

p

Annu

al %

cha

nge

Annual change in Social Security contributors

There is, however, monthly data on number of contributors to the Social Security System. Figure 15 shows that social security contributions are associated to the growth of economic activity and that they have declined noticeably since January 2007. Recent numbers are the lowest in more than five years, indicating a serious blow to the formal employment in the economy.

92


The attention in El Salvador has been focused on the electoral process for the last months, and it is now when the future and current administration are discussing broad policy actions in a joint Crisis Team. The main messages in the public policy dialogue include the revision of the 2009 budget which should be revised to reflect the slower growth and more limited fiscal resources.

On labor markets, the main discussion is about urban employment programs especially focused on the youth. This discussion is expected to be evolving rapidly as the administration changes in June 2009. Since employment and other interventions will require additional fiscal resources, the Government has been studying the rationalization of some subsidies with a regressive pattern (water, electricity).

THE DOMINICAN REPUBLIC

Labor market facts

Unemployment rates in the Dominican Republic have been declining since year 2005. However, there seems to be a reversion of this trend during the second semester of 2008. In fact, the unemployment rate (broad definition) grew to 14.2 percent in October 2008, a 0.2 percentage point increase with respect to April 2008. The conventional unemployment rate leveled off at 4.2 percent, after five consecutive semesters in decline. Meanwhile, the inactivity rate increased for a second consecutive semester to 44.6 percent: the highest since April 2004.

The former rates can be interpreted as early indicators of worsening labor market conditions (see Figure 17). Another indicator is the deceleration in the number of workers affiliated to the Dominican Republic pension system. During most of years 2007 and 2008, the annual increase of affiliated workers was above 8%. In the last quarter of 2008, the rates of growth declined noticeably (see Figure 19)

Annual data for real wages have recorded a slight increase in 2008, although still remain well below the levels before the financial crisis of 2004 (see Figure 18). Monthly data from the Pension System shows that the increase in real wages among affiliated workers occurred during the second part of year 2008 (see Figure 20).

93

Dominican Republic Labor Market Indicators (2006‐2009)

Figure 17

25.0

30.0

35.0

40.0

45.0

50.0

0.0

5.0

10.0

15.0

20.0

25.0

Apr-

00Ju

l-00

Oct

-00

Jan-

01Ap

r-01

Jul-0

1O

ct-0

1Ja

n-02

Apr-

02Ju

l-02

Oct

-02

Jan-

03Ap

r-03

Jul-0

3O

ct-0

3Ja

n-04

Apr-

04Ju

l-04

Oct

-04

Jan-

05Ap

r-05

Jul-0

5O

ct-0

5Ja

n-06

Apr-

06Ju

l-06

Oct

-06

Jan-

07Ap

r-07

Jul-0

7O

ct-0

7Ja

n-08

Apr-

08Ju

l-08

Oct

-08

Inac

tives

-Pe

rcen

tage

of w

orki

ng a

ge p

opul

atio

n

Une

mpl

oym

ent -

Perc

enta

ge o

f the

Lab

or fo

rce

Unemployment and Inactive population

unemployment (conventional definition) unemployment (broad definition) inactive

Figure 18

0

5

10

15

20

25

30

35

40

2000

2001

2002

2003

2004

2005

2006

2007

2008

Rea

l DR

$ p

er h

our

(bas

e 20

00)

Year

Earnings in the formal and informal sector

informal employment earnings formal employment wages Figure 19

-25.0%

0.0%

25.0%

50.0%

75.0%

0

250000

500000

750000

1000000

2006

2007

2008

2009

Per

cent

cha

nge

Nom

ber o

f wor

kers

months

change in social security af f iliates (YoY) social security af f iliates

Figure 20

-15%

-10%

-5%

0%

5%

10%

15%

2006

July

2007

July

2008

July

2009

Cha

nge

in R

eal

Wag

es (p

erce

ntag

e)

Months

Change in real wages

Real wages for workers with social security


The Government of the Dominican Republic initially responded to the crisis by creating a social dialogue mechanism called “National Unity Summit to confront the International Crisis”. The Summit deals with a wide range of issues, among them employment and social policy. As a result of this dialogue, many specific proposals have been suggested in relation to (i) employment generation, (ii) support to micro and medium enterprises, (iii) temporary employment programs, (iv) boosting housing construction, and (v) supporting business incubation centers

More recently, the government has announced a one year extension of health coverage for unemployed workers who earned less than DR$10.000 (around US$285) and their families. Furthermore, a US$ 400 million (1% of GDP) public works plan will be adopted in 2009 to promote both employment and economic growth.

On a related matter, the government has responded to the crisis with measures that reduce costs of transportation, trying to maintain exports and tourism, two major drivers of employment. The impact of these measures has not been assessed yet.

94

MEXICO

Labor market facts

Recessions have been accompanied by a rapid deterioration of labor markets in Mexico. The Tequila crisis of 1995 and the deceleration of economic growth of 2001‐2002 both brought about increases in unemployment and in informal employment rates. The current developments show that Mexico is heading to a severe fall in output growth (GDP fell ‐8.2 percentage points y‐o‐y in the first quarter of 2009) which will be accompanied with rising unemployment and informality.

Monthly data on unemployment records 5.3 percent for the month of February 2009. This is the highest mark since the third quarter of 1996, when the Mexican economy was getting out of the Tequila crisis. The rate for April was 5.26, still higher than the rate for any month during the last five years (see Figure 22).

Mexico Labor Market Indicators (2006‐2009)

Figure 21

57.77

54

56

58

60

62

2006

2007

2008

2009

perc

enta

ge o

f wor

kin

g ag

e po

pula

tion

month

activity rate

activity rate 12 per. Mov. Avg. (activity rate)

Figure 22

5.25

0.00

1.00

2.00

3.00

4.00

5.00

6.00

2006

2007

2008

2009

perc

enta

ge o

f act

ive

labo

r for

ce

month

unemployment rate

unemployment rate

12 per. Mov. Avg. (unemployment rate)

Figure 23

-1600

-1200

-800

-400

0

400

800

1200

1600

2005

2006

2007

2008

2009

thou

san

ds o

f wor

kers

quarter

Net Job Creation/Destruction (year-to-year)


Figure 24

80

85

90

95

100

105

110

2006 Ju

l

2007 Ju

l

2008 Ju

l

2009

inde

x (2

003=

100)

Months

Earnings

earnings in retail commerce (seasonally adjusted) wages for blue collar workers in manufacturing (seasonally adjusted)

As another indicator of the dearth of jobs, the activity rate has also been declining and reached 57.3 percent of the working age population in March 2007. This is the lowest activity rate in three years. By the fourth quarter of 2008, Mexico’s total employment fell by nearly 750,000 workers with respect to the fourth quarter of 2007. Interestingly, salaried employment grew by 530,000 workers for the same

95

period, whereas self‐employment, employers and non‐salaried workers declined by 1.28 million. Figure 23 shows that year 2008 recorded the highest destruction of self‐employment of the last five years.

The impact of the crisis is felt not only in loss of employment, but in falling wages as well. On the one hand, wages are declining in sectors such as commerce. On the other hand, wages have remained stable in the sectors that showed a sharp reduction in employment, such as manufacturing (see Figure 24)


The Mexican economy has had both active and passive labor market programs in place for several years. However, the country is the member of the OECD with the lowest expenditure in labor policies as a percentage of GDP.

As a response to the crisis, President Felipe Calderón announced in early January 2009 a National Agreement for the Economy and Employment (Acuerdo Nacional en Favor de la Economía Familiar y el Empleo). The agreement includes 25 activities grouped into five pillars. The activities directly related to labor markets are the following:

• an expansion of 2.2 billion pesos (a 0.02% of GDP) of the Temporary Employment Program (Programa de Empleo Temporal, PET);

• 2 billion pesos (a 0.02% of GDP) for employment subsidies for exporting firms; • extended ability to withdraw funds from retirement accounts for unemployed individuals; • extension of coverage of health insurance up to six months after dismissal for unemployed workers

and their families; and • 1.25 billion pesos (a 0.01% of GDP) for enhancing the employment intermediation services of the

Labor Secretary. In addition, the “Acuerdo” announced a national infrastructure program for 2009 of 570 billion pesos (a 5.87% of GDP) in combined investment between the public and private sectors.

Some Distribution aspects

A preliminary analysis of the Mexican Labor Force Survey (ENOE) allows for a first look at the distributive impact of the recent labor market performance. Sorting employment changes by position of employment and type of economic activity reveals if the patterns in net job creation/destruction are the same across different industries. Manufacturing is the only economic activity that has recorded a net destruction of salaried jobs. It also registers net destruction of all types of non‐salaried jobs. Actually, manufacturing represents more than 60 percent of all the job losses between the fourth quarter of 2007 and 2008 (seeTable 3). All the other activities, with small exceptions, have a different pattern: net job creation of salaried positions and net job destruction of non‐salaried jobs. Very few activities, however, have net employment creation. Even non‐tradable activities as construction register net job destruction.

The severity of the impact on the manufacturing sector highlights the international channel of diffusion of the crisis in Mexico. Another piece of the evidence in this regard is the distribution of household income by geographic region. The regions in the north of Mexico (where tradable activities concentrate) have experienced the most severe fall in nominal household incomes. Income losses are registered for households in all quintiles of the distribution in Northern Mexico, whereas in the rest of the country household income losses are mostly concentrated in the bottom quintiles. (see Table 4)

96

Table 3: Distribution of Net job creation by employment position: Fourth quarter 2008 (year‐to‐year)

Salaried Employer Self-employed Non-paid TotalNon specified (2,834) (1,859) 77 (14,713) (19,329)Agriculture 142,157 (62,270) (59,752) (122,918) (102,783)Power generation and mining 21,009 591 (852) 224 20,972Manufacturing (246,251) (78,733) (107,208) (28,247) (460,439)Construction 129,022 (92,764) (113,050) (100) (76,892)Trade and Commerce 108,647 (114,572) (185,711) (83,832) (275,468)Restaurants and Hotels 42,061 11,544 (5,445) (12,454) 35,706Transport and communications 108,765 (3,055) (19,909) 3,533 89,334Financial and private services 50,425 (29,623) (42,864) 10,531 (11,531)Social services 121,898 (19,534) (11,119) (6,854) 84,391Other services 45,553 (46,332) (41,115) (4,719) (46,613)Government and diplomatic services 11,103 0 0 1,562 12,665Total 531,555 (436,607) (586,948) (257,987) (749,987) Table 4: Changes in household income per capita by region and initial family labor income quintile:

Fourth quarter 2008 (year‐to‐year) (in nominal Mexican pesos) Region

1 2 3 4 5 TotalNorthwest -13.7% -4.3% -2.8% -4.5% -6.8% -5.6%North -34.9% -7.4% -3.2% -1.7% -6.3% -5.4%Northeast -39.4% -9.6% -6.0% -4.7% -5.6% -6.6%Midwest -38.8% -6.4% -3.3% -2.6% 1.4% -1.5%Mideast -20.9% -2.9% -0.8% 1.9% 0.4% 0.0%South -16.2% -3.7% -1.2% 0.1% -0.1% -0.4%Eastern -32.0% 0.3% 1.9% 0.9% 2.5% 1.5%Yucatan -5.6% 2.3% 4.1% 2.0% -4.5% -1.6%

Quintile

97

5. HOW MUCH ROOM DOES LATIN AMERICA AND THE CARIBBEAN HAVE FOR IMPLEMENTING

COUNTER‐CYCLICAL FISCAL POLICIES?*

Cesar Calderón and Pablo Fajnzylber

April 2009**

Abstract

Latin America’s government debt has exhibited a clear downward trend since 2003. While this has been partly due to rapidly increasing commodity prices, more sustainable fiscal policies have also been a contributing factor. In effect, in a significant break with the past, cyclically adjusted government balances have risen (fallen) in response to increases (reductions) in debt levels. However, Latin governments have continued to under‐save in good times and therefore fiscal policy has remained pro‐cyclical, thus weakening the ability to protect the poor and maintain infrastructure investments during bad times. Financing and institutional constraints to more counter‐cyclical fiscal policies still remain in most countries. They are lowest in Chile, followed by Brazil and Colombia, and highest in Ecuador and Venezuela. Looking forward, long‐term sustainability considerations cannot be ignored as decisions are made regarding the size, composition and targeting of fiscal stimulus packages.

Financing and institutional barriers to counter‐cyclical fiscal policies remain

Counter‐cyclical fiscal policies have recently been the focus of increasing attention by Latin American policy‐makers. As suggested by the stimulus packages recently announced by various G20 countries (Figure 1), such policies are being considered a potentially important tool for mitigating the negative impacts of the current global economic slowdown. In the case of LAC, both the size and the composition of the fiscal stimulus packages that have been announced vary considerably across countries. While some countries have focused predominantly on tax cuts (Brazil), others have planned to raise infrastructure spending (Mexico, Chile and Peru). Moreover, some countries are reinforcing their social protection networks (Argentina and Chile) whereas others are focusing on providing incentives to non‐traditional exports (Peru). As for the size of the packages that have been announced, it ranges from 0.6 percent of GDP in Brazil to 2.2 percent for Chile. Overall, the stimulus measures that have been announced by Argentina, Brazil, Chile, Mexico and Peru are equivalent to 1 percent of their combined GDP. Still, it is difficult to ascertain the extent to which the announced fiscal stimulus measures add to already existing plans or reallocate expenditures that had already been budgeted.

One important concern, in this context, is that Governments may be placing excessively high expectations on their ability to implement discretionary counter‐cyclical fiscal policies in an effective manner. After all, as documented in an extensive empirical literature, few developing countries have been able to implement such fiscal programs in the past.116 The great majority of emerging market

*The views in this note are entirely those of the authors and do not necessarily represent the views of the World Bank, its executive directors and the countries they represent. **LCR Crisis Briefs Series. 116 See Lane (2003), Kaminsky et al. (2005), Talvi and Végh (2005) and, Ilzetzki and Végh (2008).

98

economies have systematically cut taxes and raised expenditures during booms, while being forced to adopt contractionary policies during busts, when domestic and external credit constraints become binding.117 True, access to domestic and international financial markets has increased considerably for many Latin American governments during the present decade. This, in principle, could have reduced the financing constraints which in the past limited their ability to implement counter‐cyclical fiscal policies. However, as is well known, this situation has changed drastically with the drying of private credit markets after the onset of the current global financial crisis. Many LAC countries are now facing considerable challenges just to roll‐over their current stock of private and public debt. As a result, in order to finance possible fiscal expansions, or at least to avoid a fiscal contraction, most governments would probably have to rely solely on their own resources, complemented by multilateral financing. The risk is that many of the countries in the region could not have the necessary resources to finance substantial stimulus packages without compromising their hard‐gained macroeconomic stability.

A related concern is that some of the political incentives and weaknesses in budgetary institutions which in the past have contributed to the pro‐cyclicality of the region’s fiscal policies may be hard to alter in the short run. This could limit the scope for shifting, at least in a sustainable manner, to a more counter‐cyclical approach. In particular, governments have been traditionally unable to deal effectively with political pressures during expansions, which have limited their ability to generate significant surpluses during good times.118 To the extent that these political and institutional constraints remain unchanged, it could be difficult to scale back expansionary programs once the region’s economies start their cyclical recovery. This could in turn negatively affect their future creditworthiness and further limit the potential for shifting, in the medium term and on a sustainable basis, from pro‐cyclical to counter‐cyclical fiscal policies.

Although LAC has traditionally had a poor fiscal policy track record …

Latin America’s pro‐cyclical approach to fiscal policy has negatively affected its long term growth through at least two channels. First, pro‐cyclicality has helped amplify economic fluctuations.119 Second, governments have tended to penalize public investment in the fiscal adjustment programs implemented during downturns.120 Besides hurting growth, this anti‐investment bias has arguably also had unintended negative consequences on long term government solvency.121 This effect has been reinforced by the fact that most Latin American countries have failed to systematically adjust their fiscal policies to the requirements of long term debt sustainability, at least until the 1990s.122 In particular, during this period, countries experiencing increasing ratios of debt to GDP were not able to systematically tighten their discretionary revenue and expenditure policies. Finally, the pro‐cyclicality of the region’s fiscal policies has made it difficult to expand social safety nets. In this sense, the behavior of fiscal policies has been

117 See Gavin, Hausmann, Perotti and Talvi (1996) and Caballero and Krishnamurthy (2004). 118 See Tornell and Lane (1999), Braun (2001), Talvi and Végh (2005), Perry (2007), Alesina, Campante and Tabellini (2008), Ilzetzky (2008). 119 See Fatás and Milhov (2007) and Perry (2007). On the pro‐cyclicality of the region’s fiscal policies, see Gavin and Perotti (1997), Suescún (2005), Perry (2007), Perry , Servén, Suescún and Irwin (2007) and the references therein. 120 The “perversity” of fiscal adjustments biased against public investments is related to the fact that the latter have the potential to increase future government revenues, for instance through tolls, tariffs and growth‐related increases in general tax collection. See Easterly and Servén (2003) and Calderón and Servén (2004). 121 See Servén (2007). 122 See Suescún (2005) and Perry , Servén, Suescún and Irwin (2007).

99

especially harmful for the poor, given their fewer assets, limited access to credit and lower ability to smooth consumption during downturns.123

…this record has improved during the present decade

Have these stylized facts been altered during recent years? Increasing concerns with debt sustainability, for example, have been apparent in many LAC countries during the last decade. In fact, the region has reduced its net dependency on external capital inflows. As shown in Figure 2, Latin America’s general government debt as a share of GDP has exhibited a clear downward trend after 2003. This has been the result of sound macroeconomic and financial policy frameworks as well as better debt management practices involving improvements in currency and term composition. Moreover, countries have adopted more flexible and credible monetary policy frameworks; they have increased substantially their level of reserves, shifted to current account surpluses (or lower deficits) and deepened their local currency debt markets.

Figure 2 also presents the region’s actual and cyclically‐adjusted structural primary balances, both measured as shares of GDP.124 During most of the past two decades differences between these two variables have been very small, which is consistent with previous evidence on the relative weakness of Latin American automatic stabilizers – which can be approximated as the difference between actual fiscal outcomes and their structural counterparts.125 The main exceptions are the years 2000 and 2003, in which automatic stabilizers appear to have contributed to reducing the volatility of the region’s fiscal policy: operating in an expansionary fashion when cyclically‐adjusted balances were being increased (in 2000) and in a contractionary way when structural balances were being reduced (in 2003).

In addition, Figure 2 shows that after a decline during the second half of the 1990s, structural primary balances rose by about 3 percentage points between 1998 and 2008. This fiscal improvement was not driven, however, by an increased ability of governments to resist political pressures to increase primary expenditures. Instead, it can attributed largely to improvements in debt management and the fact that rising fiscal revenues, associated to a large extent to rapidly increasing commodity prices, were able to outgrow primary expenditures (Figure 3). Still, higher structural balances allowed governments to increase their emphasis on social spending, including on education, health and targeted transfers.

And fiscal policies have become more sustainable, even while remaining pro‐cyclical

In order to assess the extent to which the above policy changes altered the cyclical properties of the region’s fiscal policies, we use data for 17 LAC countries for the period 1990‐2008. We summarize the behavior of fiscal policy by estimating a fiscal policy rule in which primary balances – or alternatively government revenues or expenditures – depend on a measure of the state of the cycle, measured through the output gap, and the lagged value of the economy’s total debt as a share of GDP.126 We allow 123 See Perry (2007). 124 The construction of the structural balance follows the OECD methodology as outlined by Fatas and Mihov (2009). 125 See Suescun (2007) for evidence on the weakness of the region’s automatic tax stabilizers, in comparison with industrial countries, which the author attributes to the relatively smaller size of LAC governments and the smaller share of income taxes found in this region. 126 We closely follow the methodology proposed by Fatas and Mihov (2009) and correct for the possibility of reverse causality from fiscal policy to the level of economic activity using instrumental variables. We instrument for the lagged dependent variable and the output gap with actual and lagged values of the foreign output gap, lagged domestic output gap, actual and lagged values of international oil prices.

100

for the cyclical behavior and the responsiveness of fiscal policies to debt levels to vary from the 1990‐2002 to the 2003‐2008 period. As dependent variables we first use cyclically adjusted fiscal outcomes, so as to capture the behavior of the discretionary component of fiscal policy which responds endogenously to the economic cycle. Alternatively, we use the component of fiscal policy that is linked to automatic stabilizers. Moreover, to capture the joint effect of endogenous discretionary policies and automatic stabilizers, we repeat our analysis with actual values of fiscal outcomes as dependent variables.

Our main findings are, first, that after 2002 fiscal policies have become sensitive to long term sustainability considerations, even when controlling for the effect of increasing commodity prices. Thus, observed fiscal balances and their discretionary component have tended to significantly rise (fall) in response to increases (reductions) in the level of Government debt (Table 1).127 Our second main finding is that during the present decade Latin America’s fiscal policies have continued to behave in a pro‐cyclical way, being expansionary in countries experiencing booms and contractionary in those going through downturns.128 Policy reactions to the state of the business cycle, however, are statistically significant only for observed government expenditures – not for primary balances or for government revenues.

Third, looking in more detail at the region’s endogenous discretionary policies, we find that the cyclically‐adjusted component of government revenues tends to behave in a counter‐cyclical manner, while that of government expenditures is significantly pro‐cyclical. Not surprisingly, the behavior of the structural primary balance is a‐cyclical. Fourth, with regard to the automatic stabilizer component of fiscal policies, we find that it is significantly pro‐cyclical for government revenues and the primary balance but counter‐cyclical in the case of government expenditures. Fifth, the behavior of government expenditures is dominated by their pro‐cyclical discretionary component, which more than compensates for their counter‐cyclical automatic stabilizer element. And sixth, government revenues appear to be dominated by their pro‐cyclical automatic component, which more than compensates for counter‐cyclical discretionary policy changes.

There are, however, some notable differences between the behavior of fiscal policies across the group of 7 largest LAC countries and the rest of the region. 129 As illustrated in Figure 4, the discretionary component of government expenditures is relatively more pro‐cyclical in LAC7, while that of government revenues and primary balances is relatively more counter‐cyclical in this group of countries. In contrast, LAC 7 countries exhibit weaker counter‐cyclical automatic stabilizers in the area of government expenditures, but stronger and more pro‐cyclical automatic revenue and primary balance stabilizers. Overall, considering both discretionary and automatic changes in primary balances (jointly), LAC7 countries exhibit a less pro‐cyclical behavior.

127 The results in Table 1 do not change significantly when we also control for changes in export‐weighted commodity prices. 128 In our framework, booms (downturns) are defined as periods in which the growth of observed output is above (below) that of its cyclically‐adjusted structural component. 129 LAC 7 countries include Argentina, Brazil, Chile, Colombia, Mexico, Peru and Venezuela. Rest of LAC includes Bolivia, Costa Rica, Ecuador, Guatemala, Honduras, El Salvador, Nicaragua Panama, Paraguay, and Uruguay.

101

Fiscal space for financing stimulus packages varies considerably across LAC9 countries

We assess the extent to which LAC countries are in a position to implement fiscal stimulus packages without jeopardizing their fiscal sustainability and macroeconomic stability.130 To that end, we construct a composite index of “lack of space for fiscal stimulus” which depends on the following six factors: levels of public debt, primary deficits, commodity dependence, expenditure rigidity, access to finance and borrowing costs.131 We combine these six dimensions into an aggregate index.132 Higher scores indicate higher constraints for the financing of fiscal stimulus packages. As shown in Figure 5, Ecuador and Venezuela display the largest constraints for financing fiscal stimulus packages, while Chile displays by far the lowest constraints. Among the rest of the countries in our sample, Brazil and Colombia appear to be slightly better positioned than Peru, Mexico and Argentina. In most countries, the most important barrier to countries’ ability to implement stimulus packages is related to their limited access to domestic and external financing. The main exceptions are Chile, for which commodity dependence is the main factor, Brazil, for which the existing debt burden is the dominant factor, and Mexico, where high primary deficits are the main constraint, at least relative to other countries in the region.

Some desirable traits of counter‐cyclical stimulus packages

As shown above, the room for implementing counter‐cyclical fiscal policies varies considerably across the region. However, some key characteristics of fiscal stimulus measures are likely to be considered desirable by all countries.133 First, given the hard‐gained achievements of the region in the area of debt‐management, most countries are likely to place a large value on sustainability considerations. In practice, given the various constraints described above, this calls for exercising caution in terms of the size of potential stimulus packages. In terms of their composition, it calls for an emphasis on measures that could easily be scaled down once countries start recovering, or which could generate future increases in fiscal revenues – e.g. as in the case of growth‐enhancing investments in infrastructure. Second, when designing their fiscal responses to the crisis, most countries are likely to give a large weight to targeting issues, so as to try and protect the region’s achievements in the social front and help the most vulnerable cope with the downturn. In this respect, social safety nets based on means‐tested transfers and workfare programs may be preferable to general increases in public sector wages. Finally, all countries are likely to seriously consider the possible trade‐off between the timeliness of fiscal interventions and their potential effectiveness and efficiency. This may imply, for example, giving priority to avoiding the suspension of ongoing or pre‐appraised projects instead of starting new and un‐tested public investment projects.

130 We focus on a group of nine countries comprising Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Mexico, Peru and Venezuela. 131 Appendix 1 describes the motivations for focusing on these factors, as well as the specific indicators used to measure them, and country rankings for each of them. 132 We normalize the six components of the index to scores in the unit interval [0,1]. We do not cover Bolivia because of lack of data on borrowing costs. 133 See Spilimbergo, Symansky, Blanchard and Cotarelli (2008) and Kraay and Serven (2008).

102

References

Alesina, A., F. Campante, and G. Tabellini (2008) “Why is fiscal policy often pro‐cyclical?” Journal of the European Economic Association 6(5): 1006‐1036

Braun, M. (2001) “Why is fiscal policy pro‐cyclical in developing countries?” CIPPEC, manuscript

Caballero, R., and A. Krishnamurthy (2004) “Fiscal policy and financial depth.” NBER Working Paper 10532, May

Calderón, C. and L. Servén, eds. (2004) “Trends in Infrastructure in Latin America, 1980‐2001,” Policy Research Working Paper No. WPS 3401, World Bank, Washington, DC.

Debrun, X., J. Pisany‐Ferri, and A. Sapir (2008) “Government size and output volatility: Should we forsake automatic stabilization?”

Easterly, W. and L. Servén, eds. (2003) The Limits of Stabilization: Infrastructure, Public Deficits, and Growth in Latin America, Palo Alto: Stanford University Press and World Bank.

Fatas, A., and I. Mihov (2001) “Government size and automatic stabilizers.” Journal of International Economics

Fatas, A., and I. Mihov (2003) “The case for restricting fiscal policy discretion.” Quarterly Journal of Economics 118(4): 1419‐1447

Fatás, Antonio and Ilian Mihov (2007). “Fiscal Discipline, Volatility and Growth”, in Guillermo Perry, Luis Servén and Rodrigo Suescún (eds.), Fiscal Policy Stabilization and Growth: Prudence or Abstinence? Washington D.C.: The World Bank.

Fatas, A., and I. Mihov (2009) “The Euro and fiscal policy.” NBER Working Paper 14722, February

Galí, J. (2004)”Government size and macroeconomic stability.” European Economic Review 38:

Gavin, M., R. Hausmann, R. Perotti, and E. Talvi (1996) “Managing fiscal policy in Latin America and the Caribbean: Volatility, pro‐cyclicality, and limited creditworthiness.” Inter‐American Development Bank, Office and the Chief Economist, Working Paper 326

Gavin, Michael and Roberto Perotti (1997). "Fiscal Policy in Latin America". NBER Macroeconomics Annual. Cambridge and London: MIT Press. pp. 11‐61.

Ilzetzki, E. (2008) “Rent‐seeking distortions and fiscal pro‐cyclicality.” University of Maryland, manuscript, December

Ilzetzki, E., and C.A. Végh (2008) Pro‐cyclical fiscal policy in developing countries: Truth or fiction? NBER Working Paper 14191, July

Kaminsky, G.L., C.M. Reinhart, and C.A. Végh (2005) “When it rains, it pours: Pro‐cyclical capital flows and macroeconomic policies.” In: Gertler, M., and K. Rogoff, eds., NBER Macroeconomics Annual 2004. Cambridge, MA: The MIT Press

Kraay, A and L. Servén (2008) “Fiscal Policy Responses to the Current Financial Crisis: Issues for Developing Countries,” Manuscript, World Bank, Washington, DC.

Lane, P.R., 2003. “Business cycles and macroeconomic policy in emerging market economies.” International Finance 6: 89‐108

Perotti, R., 2007. “Fiscal policy in developing countries: A framework and some questions.” The World Bank Policy Research Working Paper 4365, September

103

Perry, G. (2007): “Fiscal Rules and Pro‐Cyclicality”, in Guillermo Perry, Luis Servén and Rodrigo Suescún (eds.), Fiscal Policy Stabilization and Growth: Prudence or Abstinence? Washington D.C.: The World Bank.

Perry, G., L. Servén, R. Suescún and T. Irwin (2007): “Overview: Fiscal Policy, Economic Fluctuations and Growth”, in Guillermo Perry, Luis Servén and Rodrigo Suescún (eds.), Fiscal Policy Stabilization and Growth: Prudence or Abstinence? Washington D.C.: The World Bank.

Servén, L. (2007): “Fiscal Discipline, Public Investment and Growth”, in Guillermo Perry, Luis Servén and Rodrigo Suescún (eds.), Fiscal Policy Stabilization and Growth: Prudence or Abstinence? Washington D.C.: The World Bank.

Spilimbergo, A., S. Kaminsky, O. Blanchard, and C. Cotarelli (2008): “Fiscal Policy for the Crisis”, IMF Staff Position Note, SPN/08/01, International Monetary Fund, Washington DC.

Suescún, R. (2005): “Pro‐Cyclical Fiscal Policy in Latin America: Where do we Stand?” World Bank LAC Regional Study, World Bank, Washington, DC.

Suescún, R. (2007): “The Size and Effectiveness of Automatic Fiscal Stabilizers in Latin America”, in Guillermo Perry, Luis Servén and Rodrigo Suescún (eds.), Fiscal Policy Stabilization and Growth: Prudence or Abstinence? Washington D.C.: The World Bank.

Talvi, E. and C.A. Végh (2005) “Tax base variability and pro‐cyclical fiscal policy in developing countries.” Journal of Development Economics 78, 156‐190

Tornell, A., and P.R. Lane (1999) “The voracity effect.” American Economic Review 89: 22‐46

104

Annex 1: Index Components

Public Debt. Lower levels of public debt may signal successful efforts of fiscal consolidation as well as better debt management practices. As an indicator of debt levels we use the general government debt as a percentage of GDP over 2005‐2008. As shown in Figure 6, Chile is by far the best positioned country in this respect, while Brazil exhibits the largest level of public debt in our sample, and is followed by Argentina, Bolivia and Mexico.

Primary deficits. During periods of liquidity constraints —either in domestic or international capital markets— governments may need to rely on liquidity buffers to finance stimulus programs. We focus on the average primary balances between 2005 and 2008. Figure 7 shows that between 2005 and 2008 Chile had the largest primary surplus whereas Venezuela had the lowest – despite the sharp increase in this country’s fiscal revenues thanks to rising oil prices.

Commodity dependence. Our indicator of choice is the response of central government revenues (as a percentage of GDP) to a 10% increase in the Reuters/Jefferies CRB index of commodity prices. Figure 8 shows that the LAC countries that are exporters of oil and natural gas (Bolivia, Ecuador, Mexico and Venezuela) appear to be the most vulnerable, in terms of reductions in fiscal revenues, to the ongoing declines in commodity prices. At least in the LAC9 group, the countries with the lowest sensitivity of fiscal revenues to fluctuations in commodity prices are Brazil and Colombia.

Expenditure rigidity. Countries with a higher share of earmarked spending arguably have a smaller room to undertake discretionary fiscal policies.134 We measure this through the share of mandatory spending in total spending, where mandatory spending is the sum of public wages, interest payments, social security payments and transfers to regions (or provinces). Figure 9 shows that mandatory spending was lowest in Colombia and largest in Brazil, Mexico and Venezuela. In Argentina and Venezuela, about three quarters of mandatory spending come from social security and transfers. Ecuador and Bolivia display the largest contribution of public wages while Colombia shows the largest contribution of interest payments.

Access to finance. Countries with deeper local currency debt markets and those with less restrictive access to world capital markets are expected to face lower financing constraints. Our composite index uses the following variables to measure this factor: (a) capital raisings by the private sector in the domestic market (as % of GDP) as a measure of depth of local currency debt markets, (b) gross capital inflows (i.e. FDI, portfolio equity, portfolio debt, and other investment) as % of GDP as a measure of access to funds abroad, and (c) an indicator variable that accounts for the fact that some countries have swap lines with foreign central banks and some LAC governments pre‐qualify for the flexible credit line (FCL) with the IMF. Figures 10 and 11 show the size of capital raisings by LAC9 countries and the gross capital inflows to those countries during 2005‐2008. Overall, Chile and Brazil appear to be the least constrained in terms of access to domestic and external financing.

Borrowing costs. The financial burden of financing stimulus packages would be lower for countries with lower sovereign spreads. We focus on the average EMBI sovereign spread of LAC countries between 2007 and 2008. Figure 12 shows that Argentina, Venezuela and Ecuador display the largest sovereign spreads while investment grade countries such as Chile, Brazil and Mexico have smaller spreads.135

134 Fatas and Mihov (2003). 135 We do not include measures of local borrowing costs as the papers issued by different countries vary considerably in terms of their maturity. However, thanks to extensive arbitrage between local and external bond markets, cross‐country differences in local borrowing costs are likely to be similar to those in external borrowing costs.

105

Annex 2: Figures and Tables

Figure 1. Estimated Cost of Fiscal Discretionary Measures in 2009

(as % of GDP)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

ItalyBrazilIndia

France

Indonesia UK

Argentina

Mexico

Germany

Canada US

AustraliaSpain

JapanChina

S. Arabia

Korea

Russia

Figure 2. Gross general government debt and primary balance in LAC region, 1990‐2008 (% of GDP )

0

10

20

30

40

50

60

70

80

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Government Debt Actual balance (rhs) Structural balance (rhs) Figure 3: LAC Revenues and Primary Spending

(Percent of GDP)

20.0

22.0

24.0

26.0

28.0

30.0

32.0

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Revenues Ciclycally Adjusted Revenues

Primary Spending Ciclycally Adjusted Primary Spending

Figure 4: Sensitivity of Fiscal Policies to the Economic Cycle in Latin America and the Caribbean

0.260.19 0.15

0.03

0.24

-0.13-0.05 -0.08

-0.27 -0.22 -0.21

-0.15

‐0.40

‐0.30

‐0.20

‐0.10

0.00

0.10

0.20

0.30

Expenditures :LAC7

Expenditures :Rest of LAC

Revenues : LAC7

Revenues : Rest of LAC

PrimaryBalance: LAC7

PrimaryBalance: Rest of

LACEndogenous discretionary policy Automatic Stabilizers

106

Figure 8. Reduction of Central Government Revenues as a result of a 10% decline in commodity prices (% of GDP )

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Brazil Colombia Argentina Peru Chile Mexico Bolivia Ecuador Venezuela

Figure 9. Central Government: Mandatory Spending (% of total spending, average 2005‐2008 )

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

Colombia Bolivia Chile Ecuador Argentina Peru Mexico Brazil Venezuela

Public wages Interest payments Social security and transfers

Figure 5. Aggregate index of lack of space for fiscal stimulus

0.0

0.2

0.4

0.6

0.8

1.0

Chile Brazil Colombia Peru Mexico Argentina Ecuador Venezuela

Debt burden Primary deficits Commodity dependenceExpenditure rigidity Financing constraints Financing costs

Figure 6. Public Debt as % of GDP (2005‐2008 Average)

0

10

20

30

40

50

60

70

Argentina Bolivia Brazil Chile Colombia Ecuador Mexico Peru Venezuela

Figure 7: Primary Balance as % of GDP (2005‐2008 Average)

0

1

2

3

4

5

6

7

8

9


Figure 10. Capital raisings ‐ Private issues (% of GDP, 2005‐2008 Average )

0

1

2

3

4

5

6


Figure 11. Gross Capital Flows (% of GDP, 2005‐2008 Average )

‐8

‐6

‐4

‐2

0

2

4

6

8

10

12


Figure 12. Borrowing Costs: EMBI Sovereign spreads (in basis points, 2007‐2008 Average )

0

500

1000

1500

2000

2500

3000


107

Table 1Fiscal Policy Reaction Function: Model with 2003 break in output gap and government debtDependent variable: Fiscal indicator as % of GDP (FI)Method: Instrumental Variables

Actual Cyclically‐adjusted Automatic stabilizersPrimary Government Primary Primary Government Primary Primary Government PrimaryBalance Revenues Expenditure Balance Revenues Expenditure Balance Revenues Expenditure

I. All LAC CountriesOutput gap, 1990‐2002 ‐0.109486 ‐0.066482 0.209820** 0.001678 0.081457 0.249023** ‐0.181486*** ‐0.246536*** ‐0.060828***

[0.100086] [0.103084] [0.106947] [0.104136] [0.101633] [0.104293] [0.016243] [0.016112] [0.009430]Output gap, 2003‐2008 ‐0.10535 ‐0.060685 0.211770* 0.004452 0.088687 0.251989** ‐0.180441*** ‐0.246137*** ‐0.061561***

[0.101527] [0.104511] [0.108066] [0.105610] [0.103208] [0.105446] [0.016472] [0.016332] [0.009560]Govt Debt, 1990‐2002 0.006994 ‐0.005468 ‐0.009512* 0.007275 ‐0.005738 ‐0.008646 0.00038 0.000447 ‐0.0004

[0.004982] [0.005668] [0.005663] [0.004824] [0.005672] [0.005607] [0.001364] [0.001351] [0.000771]Govt Debt, 2003‐2008 0.020370*** 0.002222 ‐0.012584 0.024904*** 0.006603 ‐0.012623 ‐0.002169 ‐0.000937 0.000735

[0.007432] [0.008076] [0.008248] [0.007043] [0.007985] [0.008156] [0.002142] [0.002115] [0.001222]Lagged fiscal indicator 0.444984*** 0.682914*** 0.845723*** 0.437704*** 0.565476*** 0.844531*** 0.161839*** 0.113259*** 0.244445***

[0.106828] [0.092082] [0.082748] [0.107885] [0.084132] [0.079674] [0.048194] [0.043339] [0.052917]No. Countries 17 17 17 17 17 17 17 17 17No. Observations 272 272 272 272 272 272 306 306 306

II. LAC 7 Countries (a)Output gap, 1990‐2002 ‐0.037676 ‐0.041078 0.196103 0.2348 0.142881 0.254821 ‐0.208988*** ‐0.267301*** ‐0.053806***

[0.164666] [0.121295] [0.164085] [0.192193] [0.126647] [0.158613] [0.025564] [0.024641] [0.016643]Output gap, 2003‐2008 ‐0.023594 ‐0.029414 0.200864 0.250227 0.156735 0.261228 ‐0.207845*** ‐0.267365*** ‐0.055178***

[0.168506] [0.123204] [0.166281] [0.196760] [0.128824] [0.160861] [0.026051] [0.025102] [0.016962]Govt Debt, 1990‐2002 ‐0.000763 ‐0.00989 ‐0.001902 0.001669 ‐0.014657 0.002041 ‐0.004023 ‐0.005303 ‐0.001944

[0.014259] [0.013865] [0.018588] [0.014380] [0.014638] [0.018524] [0.003888] [0.003745] [0.002521]Govt Debt, 2003‐2008 0.008195 ‐0.002954 ‐0.005339 0.012009 0.000133 ‐0.00554 ‐0.005435 ‐0.003938 0.001002

[0.014810] [0.012530] [0.016004] [0.014901] [0.013349] [0.016018] [0.004300] [0.004132] [0.002783]Lagged fiscal indicator 0.134514 0.630896*** 0.749189*** 0.006668 0.502029*** 0.788488*** 0.138092* 0.084851 0.282845***


III. Rest of LAC (b)Output gap, 1990‐2002 ‐0.144559 ‐0.069438 0.145708 ‐0.133101 0.027987 0.188266 ‐0.145566*** ‐0.224928*** ‐0.079714***

[0.129731] [0.161922] [0.135007] [0.128318] [0.155761] [0.132233] [0.021507] [0.022241] [0.009787]Output gap, 2003‐2008 ‐0.144969 ‐0.068222 0.14432 ‐0.135258 0.029068 0.187705 ‐0.144717*** ‐0.224477*** ‐0.080097***

[0.131105] [0.163918] [0.136214] [0.129618] [0.158015] [0.133485] [0.021755] [0.022488] [0.009890]Govt Debt, 1990‐2002 0.008552* ‐0.005486 ‐0.013607** 0.009440* ‐0.004229 ‐0.012773** 0.001114 0.00126 ‐0.000046

[0.005088] [0.006830] [0.005454] [0.004989] [0.006899] [0.005393] [0.001291] [0.001329] [0.000548]Govt Debt, 2003‐2008 0.026800*** 0.005217 ‐0.013895 0.031569*** 0.011017 ‐0.013397 ‐0.001317 ‐0.00063 0.000448

[0.008710] [0.011281] [0.009458] [0.008181] [0.011159] [0.009359] [0.002264] [0.002324] [0.000985]Lagged fiscal indicator 0.547577*** 0.731633*** 0.894010*** 0.608808*** 0.665810*** 0.880947*** 0.207247*** 0.168877*** 0.126472*


Standard errors in brackets *** p<0.01, ** p<0.05, * p<0.1(a) LAC 7 countries include Argentina, Brazil, Chile, Colombia, Mexico, Peru and Venezuela. (b) Rest of LAC includes Bolivia, Costa Rica, Ecuador, Guatemala, Honduras, El Salvador, Nicaragua Panama, Paraguay, and Uruguay.

108

6. CRISIS IN LAC: INFRASTRUCTURE INVESTMENT AND THE POTENTIAL FOR

EMPLOYMENT GENERATION

Laura Tuck, Jordan Schwartz and Luis Andres

May 2009*

Abstract

Infrastructure investment is a central part of the stimulus plans of the LAC region as it confronts the growing financial crisis. This paper estimates the potential effects on direct, indirect, and induced employment for different types of infrastructure projects with LAC‐specific variables. The analysis finds that the direct and indirect short‐term employment generation potential of infrastructure capital investment projects may be considerable—averaging around 40,000 annual jobs per US$1billion in LAC, depending upon such variables as the mix of subsectors in the investment program; the technologies deployed; local wages for skilled and unskilled labor; and the degrees of leakages to imported inputs. While these numbers do not account for substitution effect, they are built around an assumed “basket” of investments that crosses infrastructure sectors most of which are not employment‐maximizing. Albeit limited in scope, rural road maintenance projects may employ 200,000 to 500,000 annualized direct jobs for every US$1billion spent. The paper also describes the potential risks to effective infrastructure investment in an environment of crisis including sorting and planning contradictions, delayed implementation and impact, affordability, and corruption.

Introduction: Latin America and the Caribbean’s Stimulus Packages

As of February 2009, the largest economies in Latin America and the Caribbean (LAC) have announced stimulus packages that commit governments to increase spending on public works.136 The programs range in size from 0.4 percent to 1.6 percent of each country’s Gross Domestic Product (GDP). Extrapolating these commitments to the region as a whole suggests that governments in the region plan to invest approximately an additional US$25 billion in 2009 in public works—which is about 20 percent beyond the originally planned budget allocations.137 This represents an additional 0.5 to 1.0 percent of

*LCR Crisis Briefs Series. 136 Public Works, in this case, mainly refers to infrastructure investments in transport, energy, and water and sanitation but may also include public housing and public edifices such as schools and hospitals. The division of expenditure expected among categories and sub‐sectors is still unclear in many of the pronouncements. 137 While the authors have used UNECLAC, IMF and country‐level public expenditure data to attempt to verify the additionality of the stimulus announcements over originally budgeted expenditures, it remains to be seen whether fiscal, political and disbursement constraints will allow for these resources to be mobilized in the months to come. Also, the stated stimulus plans vary in degree of specificity and clarity as to timing, resourcing and additionality. The extrapolation to the region is based on the average levels of stated public works stimulus for 9 countries in LAC including the region’s 5 largest economies covering 80% of LAC’s GDP.

109

GDP in commitments in public works, raising public capital spending levels to somewhere between 3.0 to 4.0 percent of GDP for the region as a whole.138 (See Annex 1 for details.)

The stimulus packages are comprised mainly of public works. Although some of the programs may include investments in public housing and edifices, to date the majority of the projects and programs announced focus on the core infrastructure sectors of transport, water and sanitation, and energy. While discussions continue about the effects of these investments on short‐term aggregate demand, the employment generation potential of these investments remains a central feature139. This is particularly important as LAC’s unemployment levels rise in the face of the growing crisis. This note provides a preliminary estimate of the employment generation140 potential for different types of infrastructure investments as per the LAC stimulus packages.

The Impact of Infrastructure Investment on Short‐Term Employment Generation

The most comprehensive way to calculate the labor impacts of a single infrastructure investment or program is to consider three levels of employment generation stemming from the investment141:

• Primary Impact: Those directly employed on site to undertake the task at hand; • Secondary Impact: Those indirectly employed in the manufacture of materials and

equipment that are supplied to the initial investment; and • Tertiary Impact: The induced employment generated by the direct and indirect jobs

created. This includes all of the jobs supported by consumer expenditures resulting from wages in the two previous levels.

Using an Input‐Output Model that considers all levels of inputs to construction, the US Federal Highway Administration has estimated employment generated or supported from investments in highways.142 Keeping in mind the shortcomings of these calculations—and adjusting for them with available data from LAC on wages, leakages by sub‐sector143 and skilled and unskilled labor divisions—the approach to calculating direct and indirect jobs provides a basis for estimating the employment generation potential of investments in all areas of infrastructure in countries outside of the US. A review of project documents, IEG reports, and sectoral ESW provides a sufficient starting point for this analysis with information about construction costs and direct employment levels for a variety of infrastructure projects across Latin America. By assigning wage assumptions to workers according to skill sets, 138 In recent years, the LAC Region has seen additional investments in infrastructure from private sector sources totaling between 1 and 2 percent of GDP per year. This range includes telecommunications investments which are mostly private throughout the region but does not include private housing stock. Calderon and Serven (2004a,b). 139 For a summary of other related issues such as risks to effective infrastructure investment, impact of public expenditure stimulus on short‐ and long‐term growth, the balance between new investments vs. maintenances see Schwartz, Andres, Dragoiu (2009). 140 This note uses the term “employment generation” to refer to annualized short‐term jobs mobilized directly or indirectly as the result of an investment. It does not consider substitution effects or imply change in the long‐term labor stock. 141 See, for example, Heintz and Pollin (2009) and Romer and Bernstein (2009),. 142 JOBMOD2.1: A Comprehensive Model for Estimating Employment Generation from Federal‐Aid Highway Projects (2006). Boston University Center for Transportation Studies under subcontract to Battelle Memorial Institute for U.S. Department of Transportation Federal Highway Administration Office of Transportation Policy Studies See also Weels (2008) for more recent results of this model. 143 In calculating secondary labor generation, a portion of machinery and equipment inputs are assumed to be imported depending upon the technology deployed in the sub‐sector providing a very basic discount for leakage.

110

estimating domestic and foreign content for both materials and equipment, a levelized set of results in terms of direct and indirect annualized employment can be calculated for a given sum of money expended—in this case US$1billion. These estimates do not account for substitution effect and are thus most applicable to economies with slack labor conditions and high unemployment among day laborers and construction workers.

The most important result of this summary analysis is that the range of direct employment impacts is tremendous: from 750 direct short‐term positions per US$1billion spent on coal‐fired generation projects to 100,000 short‐term positions for water supply and sanitation network expansion. In addition, the results are highly sensitive to assumptions about wages and the division between skilled and unskilled workers. The direct employment generation potential of a public works project is thus highly sensitive to: the sectoral allocation of the proposed program; the technology to be employed in each project, and the local labor market traits of the country in question. Indirect job estimates are also highly sensitive to the division between locally produced versus imported inputs.

With those sensitivities in mind, a “prototypical mix” of infrastructure investment144 implemented in LAC would generate about 40,000 direct and indirect short‐term positions per US$1billion spent. Even with an assumed multiplier of 2.0 for further induced employment and no crowding out or substitution effect, this would mean employment generation of about 2 million jobs for the incremental US$25 billion so far proposed as stimulus in 2009 in the LAC region. This would represent about 7 percent of LAC’s estimated unemployed in 2009. The estimates correspond to capital investment projects in various countries across the region. [See Annex 2 for details.]

There is a sub‐set of infrastructure interventions, which, however limited in scope, may provide the opportunity for even greater direct employment benefits: rural road maintenance. Such programs typically145 invest up to 90 percent of the total project costs in labor activities. Regional data suggest that between 200 and 500 annualized positions are generated for every million dollar spent on these initiatives by employing unskilled workers in rural areas paid at the minimum wage. The jobs generated from labor intensive maintenance projects would, in turn, generate very few indirect jobs because of the lack of material and equipment inputs. Nevertheless, labor intensive projects coupled with well‐targeted social programs may be considered a highly progressive intervention for reducing the impact of the crisis on poor communities. Again, the primary employment generation numbers assume no substitution effect.

In advising governments on the design of stimulus plans, it becomes clear that the employment story is complex and the investment decision should be made in the context of the explicit objectives of the government in the medium to long term. Beyond the varying labor results, fast and significant expansion of infrastructure investments presents important practical challenges to the efficiency and the efficacy of the stimulus program. A shortlist of these challenges might include: sorting and planning contradictions, delayed disbursement and impact, the affordability of these packages, and corruption risks.

144 Based on country experience with infrastructure investment we assumed a composition of these stimulus as: 50% in Transport (25% in highways, 20% in urban roads, and 5% in rural roads), 30% in Electricity (25% in generation of electricity and 5% in rural electrification), and 20% in Water and Sanitation (15% in coverage expansion and 5% in treatment plants). We simulated different composition and the estimations were significantly robust. 145 Peru ‐ Second Rural Roads Project (P044601); data from Mexico Subsecretario de Transporte, MTC, and Guatemala ‐ Second Rural and Main Roads Project (P055085).

111

Sorting and Planning Contradictions: Infrastructure investments often contain complementarities (e.g., modes of transport assets along a supply chain) or substitution effects (e.g., rail versus road for transport or gas versus electricity supply for heating). They might also contain contradictory character traits intended to meet different objectives. For example, a road investment component might meet employment goals, but could contribute to automobilization and higher carbon emissions in the long‐term. A renewable energy component might meet environmental objectives but may not demonstrate significant employment benefits given the high import components. Governments will need to call upon the capacity for ex ante project evaluation; cross‐sectoral convening ability; and the authority to prioritize, scale and permit projects according to impact analysis.146 This will enable the development of investment packages that converge short‐term goals of stimulus with the long‐term goals of sustainable growth.

Delayed Implementation and Impact: The lifecycle of project preparation for medium or large‐scale projects is generally 1 to 3 years, although projects that are simply awaiting financing may be “shovel ready.” However, it is possible that projects that have been sitting in pipeline will require new demand studies, updated cost projections or even recalibrated willingness and ability to pay analyses given the shifting resources of consumers and the changing prices of inputs in the crisis environment. Moreover, several countries in LAC habitually disburse less than their annually expected disbursements—typically around 75 percent of plan. Given the importance of timeliness in generating stimulus effect, delays and slow disbursements would have a significant and perhaps irreversible effect on the impact of the project.

Affordability of the Stimuli Packages: The potential scope, size and timing of LAC’s proposed stimuli packages will be determined by fiscal space—the room in a government’s budget that allows it to provide resources for additional projects without jeopardizing the sustainability of its financial position or the stability of the economy. In other words, fiscal space must exist or be created if extra resources are to be made available for worthwhile government spending147. LAC’s proposed stimuli packages present enormous demand on the limited fiscal space in the region and the room for aggressive responses is heterogeneous across the region.148

Corruption Risks: Emergency environments often create the impetus for shortcuts, particularly as they relate to time‐consuming safeguard practices. In a crisis situation, governments may feel justified in seeking to bypass lengthy procurement policies such as international competitive bidding, pre‐qualification, and re‐bidding in the case of insufficient competition or non‐responsive bids149. The temptation to trade time for competition raises the risk of corruption, collusion, and public skepticism. Rushed procurement processes run the risk of being self‐defeating and costly elements of stimulus.

146 The Global Experts Team for Public Sector Management, in conjunction with LCSPS, is undertaking a study of best practices in the management of stimulus programs that includes a review of US and other OECD institutional arrangements. 147 Heller (2005). 148 Calderon and Fajnzylber (2009) 149 Kenny (2007).

112

Conclusions

Infrastructure investment is already a central part of the stimulus plans of LAC as the region confronts the growing financial crisis. The employment generation potential of the infrastructure investment component of stimulus may be considerable—averaging around 40,000 jobs per US$1billion in LCR for a basket of investment. This excludes the tertiary effects of induced employment from direct and indirect employee consumption.150 Albeit limited in scope, rural road maintenance projects initiated through micro‐enterprises may produce 200,000 to 500,000 direct jobs per US$1billion of disbursements. Levels of employment generation per package of investments is highly sensitive to local wages, the division among skilled and unskilled workers, the sector under consideration (i.e., the component pieces in the “basket”), the technology being deployed in each project investment, the degree of importation of inputs, and— in areas without slack labor conditions—substitution effect.

To understand the impact of investments in times of crisis, policy makers will benefit from sector‐level analysis, comparative technology analysis, and data on the sourcing of inputs. In addition, in order to assure the effectiveness of infrastructure investments in a crisis environment, governments may wish to consider the strengthening of planning processes which weigh the trade‐offs associated with multiple investments; procurement processes which are robust in the face of time pressures; and disbursement plans which keep up with the levels of expected investment activity. Finally, short‐term plans for infrastructure investment are most effective when viewed in the context of the long‐term objectives of growth and poverty alleviation which remain infrastructure’s fundamental contribution to economic activity.

150 Although US estimates from highway construction more than double the employment generation estimates when induced jobs are added, it should be remembered that other forms of transfers from government—tax credits, CCTs, food stamps—would also generate induced employment

113

Bibliography

Calderon, C. and P. Fajnzylber (2009) “How much room does Latin America and the Caribbean have for implementing counter‐cyclical fiscal policies?” LCR Crisis Briefs Series, The World Bank.

Calderon, C. and L. Servén (2004a) “The Effects of Infrastructure Development on Growth and income.” Policy Research Working Paper Series 3400, The World Bank.

Calderon, C. and L. Servén (2004b) "Trends in infrastructure in Latin America, 1980‐2001," Policy Research Working Paper Series 3401, The World Bank.

Coenen, G. and R. Straub (2005), “Does Government Spending Crowd in Private Consumption? Theory and Empirical Evidence for the Euro Area,” International Finance, 8(3): 435‐470.

Consensus Economics (2009) “Latin America Consensus Forecast,” February 16, 2009.

Heintz, J. and R. Pollin (2009) “How Infrastructure Investment Supports the U.S. Economy: Employment, Productivity and Growth,” Political Economy Research Institute (PERI), University of Massachussetts Amberst, January 2009.

Heller, P. (2005) “Understanding Fiscal Space,“ IMF Policy Discussion Paper PDP/05/4, Fiscal Affairs Department. March 2005.

Kenny, C. (2007) “Infrastructure governance and corruption: where next?,” Policy Research Working Paper, WPS4331. The World Bank.

Romer, C. and J. Bernstein (2009), “The Job Impact of the American Recovery and Reinvestment Plan.” January 8, 2009.

RDEL (2009) “Job and Economic Development Impact Models” National Renewable Energy Laboratory.

Schwartz, J., L. Andres, and G. Dragoiu (2009) “Crisis in LAC: Infrastructure Investment, Employment and the Expectations of Stimulus,” The World Bank.

UNECLAC (2009) “The reactions of Latin American and Caribbean governments to the international crisis: an overview of policy measures up to 30 January 2009.”

UNEP (2008) “Green Jobs: Towards decent work in a sustainable, low‐carbon world,” United Nations Environment Programme (UNEP).

Wells, J. (2008) “Transportation Spending, An Inefficient Way to Create Short‐Term Jobs,” The official blog of the U.S. Secretary of Transportation, http://fastlane.dot.gov/2008/09/chief‐economist.html (Accessed March 12, 2009)

Acknowledgements:

The authors would like to acknowledge the research assistance of Georgeta Dragoiu, María Claudia Pachón and Darwin Marcelo Gordillo and to thank the following colleagues for their inputs and suggestions: Daniel Benítez, Philippe Benoit, Cesar Calderón, Rodrigo Chaves, Cecilia Corvalán, Augusto de la Torre, Marianne Fay, Jose Luis Guasch, Ada Karina Izaguirre, Emmanuel A. James, William Maloney, Nick Manning, Marisela Montoliu Muñóz, Nicolas Peltier‐Thiberge, Jaime Saavedra, Tomas Serebrisky, María Angélica Sotomayor, Aiga Stokenberga, Theo David Thomas, Maria Vagliasindi, and Ariel Yepes. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors, and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. The authors are also solely responsible for any incomplete or inaccurate data.

114

Annex 1: Stimulus Pans for LAC

The Table below summarizes the region’s major commitments to economic stimulus that have been announced in recent weeks and the estimation of region‐wide investment levels. The five countries included in the table represent over 75 percent of the region’s population and GDP.

Table A1: Stimulus Plans for LAC, 2009 Ratio Stimulus vsTotal Investment

$B % GDP $B % GDPArgentina 4.4 1.6% 17.1 6.1% 25.7%Brazil 6.7 0.5% 23.3 1.7% 28.8%Chile 0.7 0.4% 4.7 2.7% 15.0%Mexico 6.9 0.8% 43.6 4.8% 15.8%Peru 1.6 1.3% 5.8 4.6% 27.6%LAC (*) 25 0.5% to 1.0% 125 3% to 4% 20%(*) For the LAC estimates, we extrapolated the figures to those countries without information.

2009 Stimulus PkgInvestment in Public Works

Total Public Works(2009)

Source: UNECLAC (2009), IMF, National Legislation Data, Consensus Economics (2009), and Authors’ calculations. Note: Many countries have proposed multiyear packages. These estimates capture the additional investments in public work to be implemented in 2009 in addition to the budget proposed previously for the year 2009. (*) For LAC’s estimates, we extrapolated the figures to those countries without information.

Annex 2: Annual Direct Employment

The table below provides the results for an estimate of Annual Direct Employment per US$1Billion spent.

Table A2: Employment Levels for Representative Infrastructure Capital Investment Projects in LAC, by Country and Sub-sector

Qualifiedworkers

Non-qualifiedworkers

Domesticinputs (mainly

material)

Foreigninputs(mainly

equipment)

Others Total

Annual DirectEmployment

(per US$1B/yr) [*]

TransportColombia - Access to neigborhoods (streets) 15% 6% 49% 16% 14% 100% 22,500Colombia - Feeder routes for Transmilenio 27% 23% 6% 99% 35,833Brazil - Roads 3% 9% 22% 63% 3% 100% 16,577Argentina- Rosario - highways 1.3% 0.3% 60% 38% 0% 100% 1,650

Water and SanitationHonduras - Improvement on water captation 28% 12% 40% 20% 100% 43,333Honduras - Rehabilatation of water networks 30% 20% 40% 10% 100% 58,333Honduras - Expansion of water networks 20% 30% 40% 10% 100% 66,667Honduras - New treatment plant 10% 10% 80% 0% 100% 25,000Colombia - Expansion of WSS networks 8% 56% 32% 4% 100% 100,000Brazil - Rain Drainage networks 8% 16% 48% 28% 0% 100% 34,001Brazil - Sewerage 4% 11% 68% 17% 0% 100% 21,746

EnergyUS - Solar PV 3%-5% 100% 2,700US - Wind Power 4%-6% 100% 3,400US - Biomass 1%-2% 100% 700US - Coal-fired 1%-2% 100% 750US - Natural gas-fired 2%-4% 100% 1,700Brazil - Hydropower 5%-10% 100% 4,500Peru - Rural Electrification 14% 7% 26% 53% 0% 100% 23,000

[*] These estimates were based on an hourly wage of $3 for non qualified workers and $6/hr for qualified one for 2,000 working hours a year.

43%

95%-97%94%-96%98%-99%98%-99%

90%-95%96%-98%

Source: World Bank project documents: Honduras ‐ Water and Sanitation Program (P103881), Colombia ‐ Bogota Urban Services Project (P074726), Brazil ‐ Bahia Poor Urban Areas Integrated Dev (P081436), and Argentina ‐ Santa Fe Road Infrastructure (P099051). Energy estimates are from the UNEP (2008), Peru ‐ Rural Electrification (P090116), and Brazil ‐ Cana Brava hydropower plant. Authors’ calculations.

115

7. HOW WILL LABOR MARKETS ADJUST TO THE CRISIS? A DYNAMIC VIEW

William Maloney

March 2009*

Abstract

Tracking flows of workers among different sectors of employment during economic downturns can shed light on the mechanism of labor market adjustment and inform the design of safety net programs. Though patterns may differ across recessions, we find that the generally countercyclical rise in unemployment and informality is driven primarily by a reduction in hiring in the formal sector, rather than increased labor shedding. Further, changes in the rate of separations from informality are the largest determinant of changes in unemployment. Both suggest that safety nets should focus less on formal job loss per se and more generally on movements in family incomes, perhaps revealed through self targeting mechanisms.

Past crises suggest that as GDP falls unemployment and informality will rise. For Brazil and Mexico the elasticity of the unemployment rate with respect to output averages roughly ‐4.5; for unemployment and the elasticity of the share of the labor force in informal employment averages about .2.

Understanding the flows of workers among sectors that generate these movements in aggregates can help understanding the mechanisms of adjustment of labor markets during crisis, and inform the design of safety net programs. At any moment in time, changes in any labor market indicator, such as the unemployment rate, or the share of formal employment, is driven by changes in flows into and out of those employment states from and to other states. A change in the unemployment rate, for example, could be supported with a variety of combinations of flows. In the US literature, for example, Shimer (2007) and Hall (2005) among others have argued that most new unemployment is caused by employers ceasing hiring, rather than firing workers. The reason for this is an active subject of debate.

Following workers as they move among sector across several periods of economic downturn in Brazil and Mexico suggests several stylized facts about how Latin America’s labor markets adjust to macro economic shocks.151

• Consistent with the US literature, the share of formal employment is procyclical with an elasticity of approximately .2‐.3. That is, while not always the case, formal employment generally falls during recessions. This occurs primarily because of a reduction in hiring and hence greater difficulty of finding formal jobs from inactivity, unemployment and informal jobs rather than because of increased separation from formal jobs.

*LCR Crisis Briefs Series. 151 See Bosch and Maloney (2008) for details.

116

• Transitions between informality and formal employment, in fact, slow down during downturns. Conversely, in recoveries, flows increase in both directions suggesting increased matching across both the informal and formal sectors of the economy. The symmetry of flows, as opposed one where workers transit unidirectionally from unemployment to informality to formality to retirement suggests that job matches in the informal sector are not overall, considered inferior. The queuing view of informality as disguised informality is true for some, but not the majority of those holding informal jobs.

• The unemployment rate is countercyclical, rising as output falls, with an elasticity with respect to output of roughly ‐4.5. It is driven primarily by increased job separations of informal workers. Shedding of workers from the formal sector, while important, has not been the dominant driver.

• Informality is also countercyclical, not primarily because of increased shedding from the formal sector, but because unemployed workers cannot find jobs in the formal sector.152 Fundamentally, informal job finding rates show much less cyclical volatility than formal ones and hence the sector winds up hiring a disproportionate number of job seekers in downturns.

• Together, these stylized facts offer an updated mechanism of the informal sector as a safety net, although without the connotation of a general inferiority of informal employment. The sector absorbs the majority of the newly unemployed, and contributes most to changes in unemployment. However, it is not primarily a direct safety net for those losing formal sector jobs.

Why does this matter?

Downturns differ from one another. At the most basic level, the elasticities of unemployment and formality vary across episodes within these two countries and we could expect the underlying dynamics to change as well. Hence, extrapolating the lessons from history to the present global crisis is not without risks. However, several implications emerge from the stylized facts presented above:

• Job loss in the formal sector is not sufficient as a targeting criterion. New entrants to the labor force, or family members seeking to augment falling real incomes will find themselves unable to access formal employment and will recur to the informal sector. Because no one in the family has lost a job, their declining prospects will not show up on official job registers.

• To the degree that job loss is a criterion, the fact that most is from the informal sector suggests that tracking formal employment rolls, or targeting formal workers will miss a critical part of the story.

• The increased absorption of labor in the informal sector implies a fall in average earnings there. For instance, small business owners will see more competition from new entrants in the midst of a decline in overall spending. Again, average family incomes, rather than job loss per se, needs to be a central focus of targeting efforts.

• The increased rate of job separations from the informal sector suggests that, while the informal sector is absorbing more workers, it is doing so in a dynamically frenetic way, shedding labor at a very high rate as well. This may arise because many of the new entrants to the sectors soon find their micro enterprise to be unviable. Informality is thus something of an unstable safety net.

152 More generally, the cyclical patterns can be more complex depending on the nature of the economic shock. From 1988‐91 in Mexico, the informal sector expanded during the boom in non‐tradables‐construction, services, transport, a pattern also found in Brazil and Colombia at various times (Fiess et al 2008). However, the present shock is originating from the exterior largely through demand for exports which tend to be more formal sectors. The more straightforward interpretation as a negative shock to the formal sector is appropriate.

117

• That said, informality cannot be considered a criterion in itself for social protection expenditures. In good times, opinion surveys in both countries confirm the view gleaned from the transition patterns above that a substantial majority consider self‐employment an attractive sector to enter. In times of crisis, the relative share of involuntary entrants rises, but a sizable share remains voluntarily informal (see annex II). The informal salaried in both good and bad times are substantially less voluntary than the self employed (see Perry et. al 2008).

Policy

• Policies seeking to preserve jobs by raising the costs of firing are unlikely to have first order effects since formal sector separations are not the primary drivers of unemployment and informality growth. In fact, they may further depress hiring in the formal sector by increasing the long run labor cost.

• Targeting should be based more on family income than on registered job loss. Unfortunately, because of the CCT’s reliance on means testing for targeting, sudden falls in family income with moderate duration may be missed by infrequent periodicity of means testing. That said, the fact that in Mexico, children do drop out of school when a parent loses a job suggests a need for the kinds of incentives that CCTs offer.

• A self targeting system such as envisaged in workfare programs such as Trabajar or the PET can obviate the need for means testing. For instance, setting the program wage sufficiently low means that only those truly in need will apply for the program.

• The choice of types of works programs, whether simple with low materials content or more sophisticated infrastructure projects with a lower budget share transferred to workers depends substantially on the local context and bjectives of the government.153

Annex 1: Details on adjustment mechanisms

Fujita and Ramey (2007) offer a means of decomposing movements of labor market aggregates, such as the unemployment rate, or the share of formal employment into the principle contributions of flows to and from different sectors. Table 1 reports this breakdown for Brazil and Mexico, across a long period, and during times of crisis. In general, between 85 and 80% of movements in the unemployment rate are driven by inflows from informality (I). Outflows reductions in outflows to formal jobs (F) contribute roughly 20%; to informal salaried work, relatively little, and to self employment (S), essentially zero.

153 See Ravallion (1999) and Maloney (2000)

118

Table 1: Relative Contribution of Flows to Aggregate Movements in Unemployment and Formal Employment (In percent)

Unemployment Rate I‐Inflows Outflows‐F Outflows‐I Outflows‐S Error Whole Sample Mexico 0.82 0.20 0.02 ‐0.03 ‐0.01 Brazil 0.69 0.22 0.11 ‐0.03 0.01 Recession Mexico 0.76 0.24 0.04 ‐0.01 ‐0.02 Brazil 0.65 0.26 0.09 0.00 0.00 Formal Employment I‐flows Outflows‐I Outflows‐S Outflows‐U Error Whole Sample Mexico 0.69 0.08 0.01 0.19 0.04 Brazil 1.22 ‐0.21 ‐0.16 0.18 ‐0.02 Recession Mexico 0.84 ‐0.05 ‐0.01 0.17 0.05 Brazil 1.31 ‐0.15 ‐0.22 0.08 ‐0.02

Notes: The table presents the contribution of the cyclical component of each flow to cyclical volatility of the unemployment rate and the share of formal employment for Mexico and Brazil following Fujita and Ramey, 2007. We define recession as output below trend. O=Out of the Labor Force, U=Unemployment, E=Employment, S=Informal Self‐Employed, I=Informal Salaried, and F=Formal Sector, all as proportions of working age population. Data for Mexico (left panels) is drawn from the quarterly National Urban Labor Survey (ENEU) from 1987:Q1 to 2004:Q4. Data for Brazil (right panels) is drawn from the Monthly Employment Survey (PME), quarterly averaged from 1983:Q1 to 2001:Q2.

Underlying these results are differing responses of flows among particular sectors to downturns. Figure 1 presents the raw and detrended job finding probabilities across time in Brazil and Mexico. The former are somewhat clearer, however Brazil experienced a steady rise in informality across the sample period and this evolution muddies somewhat the cyclical patterns. What is clear is that flows from unemployment into formal employment show the greatest volatility across the cycle, decreasing more than any other sector of employment during downturns.

Figure 2 shows the analogous series for job separations. In this case, transitions from formality to unemployment vary far less than either flows from informal salaried work or informal self employment.

In the simulations underlying table 1, the contribution of each possible flow to the evolution of unemployment and formal employment is undertaken by modeling how all flows interact to generate these aggregates, and then sequentially holding one flow or another fixed and measuring the resulting impact on the evolution of the aggregate.

119

Figure 1: Job Finding Rates from Unemployment (Levels and Cycle): Mexico and Brazil

Mexico Brazil

.1.2

.3.4

.5.6

1987q1 1991q3 1996q1 2000q3 2005q1

U-S U-IU-F

0.1

.2.3

.4

1983q1 1988q1 1993q1 1998q1 2003q1

U-S U-IU-F

-.4-.2

0.2

1987q1 1991q3 1996q1 2000q3 2005q1

U-S (HP) U-I (HP)U-F (HP)

-.4-.2

0.2

.4

1983q1 1988q1 1993q1 1998q1 2003q1

U-S (HP) U-I (HP)U-F (HP)

120

Figure 2: Job separation Rates towards Unemployment (Levels and Cycle): Mexico and Brazil

Mexico Brazil 0

.05

.1.1

5

1987q1 1991q3 1996q1 2000q3 2005q1

S-U I-UF-U

0.0

2.0

4.0

6.0

8.1

1983q1 1988q1 1993q1 1998q1 2003q1

S-U I-UF-U

-.4-.2

0.2

.4

1987q1 1991q3 1996q1 2000q3 2005q1

S-U (HP) I-U (HP)F-U (HP)

-.4-.2

0.2

.4

1983q1 1988q1 1993q1 1998q1 2003q1

S-U (HP) I-U (HP)F-U (HP)

Notes: Figure 1 shows the transition rates from unemployment (U=Unemployment) into the three employment sectors (S=Informal Self‐employed, I=Informal Salaried, and F=Formal Sector). Figure 2 shows the transitions rates into unemployment (U) from all three sectors of employment. Transition rates are inferred from the continuous time transition matrix for each period obtained following the procedure by Geweke et al. (1986) outlined in Bosch and Maloney (2009) Section III. Computations are based on 10.000 Monte Carlo replications. The series have been smoothed using a 4 quarter moving average to remove high frequency fluctuations. The bottom panels shows the series logged and de‐trended using an HP filter with lambda 1600. Data for Mexico (left panels) is drawn from the quarterly National Urban Labor Survey (ENEU) from 1987:Q1 to 2004:Q4. Data for Brazil (right panels) is drawn from the Monthly Employment Survey (PME), quarterly averaged from 1983:Q1 to 2001:Q2. Shaded areas indicate recessions.

121

Annex II: Cyclical changes in the level of disguised unemployment in the informal sector

The Mexican employment survey suggests that there is a component of informality that correspond to disguised unemployment and which varies as expected over the business cycle.. Figure 3 plots the proportion of workers who respond positively to the question “Have you been looking for a job over the last two months?” and who had not changed employment status from the quarter before as a possible proxy for the degree of dissatisfaction with the current job coupled with the availably of alternative jobs. Search intensity is generally higher in the informal salaried sector, perhaps, reflecting the relative youth of that group although the magnitudes (and hence differences) are not large: in the upturns of mid‐1990s and 2000, search rates were equal across sectors at roughly 1‐2%. These very low levels suggest that we may not be fully capturing the degree of involuntariness in the sector. As additional information, the National Microenterprise survey suggests that in 1992, roughly 65% of those entering informal self employment from formal salaried work replied doing so voluntarily. The rate of search is equivalent to the formal sector at this time, and somewhat below the informal salaried sector suggesting a higher degree of involuntary entry there.

The share searching is strongly countercyclical implying that as the labor market becomes slack and the access to the formal sector from all sectors decreases, dissatisfaction increases. In the informal sectors, the percentage searching for better jobs peaks at just under 7% during the 1995 crisis, a gap of slightly over 4% points over the formal sector . This suggests that in fact the sector contained more workers who were forced into bad matches. This makes sense if during the crisis only the informal sector was hiring and absorbing more unemployed as a share of the workforce than during booms.

Figure 3: Searching While Employed: Mexico

0.0

2.0

4.0

6.0

8

1987q1 1991q3 1996q1 2000q3 2005q1

Searching(F) Searching(S)Searching(I)

Note: Quarterly data from the National Urban Labor Survey (ENEU) 1987:Q1 to 2004:Q4. Searching (j) refers to the proportion of employed workers in sector j who claim to be looking for a new job and have not changed employment status in the previous quarter.

122

References:

Bosch, M. and W. F. Maloney (2009) Cyclical Movements in Unemployment and Informality in Developing Countries, mimeo, Office of the Chief Economist for Latin America, the World Bank

Fiess, N. M. Fugazza, W. F. Maloney (2008) Informality and Macro economic Fluctuations” IBRD working paper.

Geweke J., R. Marshall and Gary A. Zarkin (1986) “Exact Inference for Continuous Time Markov Chain Models,” Review of Economic Studies 53(4). Pp653‐69

Hall, R. (2005); “Employment Efficiency and Sticky Wages: Evidence Flows in the Labor Market,” Review of Economics and Statistics 87(3)397‐407.

Maloney, W. (1990) “Evaluating Emergency Programs: Intertemporal and Financing Considerations” mimeo, Latin American and Caribbean Region.

Shimer, R. (2007), “Reassessing the Ins and Outs of Unemployment” Mimeo, University of Chicago.

Perry, G., W. F. Maloney, O.S. Arias, P Fajnzylber, A.D. Mason, J Saavedra‐Chanduvi (2007) Informality, Exit and Exclusion, World Bank Latin American and Caribbean Studies.

Ravallion, M (1999) “Appraising Workfare,” World Bank Research Observer 14(1): 31‐48.

123

8. WHAT IS THE LIKELY IMPACT OF THE 2009 CRISIS ON REMITTANCES AND POVERTY IN LATIN AMERICA AND THE CARIBBEAN?

Gabriel Demombynes, Hector Valdés Conroy, Ezequiel Molina and Amparo Ballivián

April 2009*

Importance of Remittances for LAC’s Fight against Poverty and Inequality

Remittances have long been known to be an important factor for development and poverty reduction in Latin America and the Caribbean (LAC).154 Just how important depends on what variable is chosen for the analysis, since there is wide variation in remittances’ importance by country, depending on the viewpoint:

• In total dollar terms, the largest remittance receivers in 2008 were: Mexico ($26bn), Colombia ($4.5 bn), Brazil ($4.5 bn), Guatemala ($4.4 bn), and El Salvador ($3.8 bn).

• As share of GDP the largest receivers in 2007 were: Honduras (24.5%), Guyana (23.5%), Haiti (20%), and Jamaica (19.4%).

• In per capita terms, the largest receivers in 2008 were: Jamaica ($827), El Salvador ($555), Dominica ($413), Honduras ($395), and Guyana ($377).

• By fraction of households receiving remittances, the most important were: Haiti (49%), Jamaica (40%), Guyana (33%), El Salvador (27%), Nicaragua (22%), and Honduras (20%).

Remittances are an important driver of income growth, but only for those fortunate enough to receive them. The distribution of remittances across income quintiles using post‐transfer income can be very different than the distribution using pre‐transfer income. Previous analysis for 11 LAC countries for which micro‐data is available, has shown that 30 percent of households in these countries are in the first quintile of the pre‐remittances income distribution, but only 10 percent of households are in the lowest quintile of the post‐remittance income distribution. The different conclusions arrived at using post‐ versus pre‐transfer income simply reflect the fact that transfers (including remittances) are an important driver of income growth among the households fortunate enough to have access to these transfers.155 In other words, if remittances were to exogenously disappear, or substantially decrease as a result of exogenous shocks, the incomes of the poorest in LAC would have a negative impact.

The distribution of remittances across pre‐remittance income groups also shows significant variation among the 11 countries in LAC for which this data is available.156 Roughly speaking, three groups can be distinguished:

• In Mexico, 61 percent of the households receiving remittances fall in the bottom 20 percent of non‐remittances income, whereas only 4 percent of them are in the top 20 percent. Similarly, in Paraguay 42 percent of recipients are in the bottom 20 percent of the distribution, and only 8

*LCR Crisis Briefs Series. 154 For a complete analysis of the pros and cons of the subject see “Remittances and Development: Lessons from Latin America,” P. Fajnzylber and J.H. López, editors. World Bank, 2008.

155 See P. Fajnzylber and J.H. López, op.cit., chapter 2, and “Mexico: Income Generation and Social Protection for the Poor,” World Bank 2005. 156 Data in this paragraph is taken from Fajnzylber and López, 2008, p.33.

124

percent are in the top 20 percent. Other countries where at least 30 percent of the recipients of remittances are in the lowest 20 percent are Ecuador, El Salvador, and Guatemala.

• In contrast, in Peru and Nicaragua the distribution of remittances across households is completely different. For example, in Peru fewer than 6 percent of the households that receive remittances belong to the bottom 20 percent, while 40 percent belong to the top 20 percent. In the case of Nicaragua, only 12 percent of the recipients are in the bottom 20 percent, while 33 percent belong to the top 20 percent.

• In between these extremes, there are four countries—Bolivia, Honduras, the Dominican Republic and Haiti—where remittances recipients are found at similar rates among households in the bottom and top quintiles.

Evolution of Remittances flows to LAC and Impact of the 2009 Crisis

After strong growth in the 1990’s, since 2004, overall remittances to the LAC region have stagnated (see figures 1 and 2). Following this stagnation period, the overall flow of remittances started to decline since the onset of the global financial crisis. Monthly remittance data from central banks for individual countries for which recent monthly data is available157 show declines in remittances for most countries. Rates of growth in these countries started to become negative in the last quarter of 2008. The unweighted average change in remittances’ flows to these countries has declined by 7.3 percent in recent months from levels one year earlier. Taking into account the weights of different countries in the overall remittance stream, the total level of remittances in November and December 2008 for these six countries was 8.3 percent below the level for the same two months in 2007. In Central America, the weighted average decline in the first quarter of 2009 was 4.4% lower than the same period of 2008.

Figure 1. Total Flow of Remittances to LAC (percentage of GDP)

Figure 2. Actual and Expected Growth of Remittances to LAC

(year‐to‐year % change)

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

-10%

-5%

0%

5%

10%

15%

20%

25%

30%

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

e

2009

f

2010

f

2011

f

Base Low

Source: DECPG estimates. GDP figures are from World Development Indicators 2007. Does not

Source: DECPG estimates. Note: Does not include Chile, Suriname, and Uruguay.

In Mexico rates of growth started to be negative for several consecutive months as early as January 2008 in nominal terms, but remittances have increased in real terms. A sharp depreciation of the nominal exchange rate, coupled with lower inflation, has more than offset the decline in remittances in nominal dollars. IMF projections are that, in real terms, remittances to Mexico increased by 9.8 percent (annualized) in the last quarter of 2008. Yet, for the region as a whole, the gains from depreciation

157 Mexico, Dominican Republic, El Salvador, Guatemala, Honduras and Jamaica.

125

would be eroded by a continued decline in foreign exchange flows, and in dollarized economies or countries with more inflexible exchange rates the depreciation cushion is either absent or small.

World Bank projections based on GDP growth projections indicate that nominal remittances to LAC will fall by 4.4 percent in the base case and 7.7 percent in the low case scenario.158 Projections vary significantly by country. These projections are based on GDP forecast of both sending and receiving countries. But, if remittance senders in the U.S. are disproportionately affected by the recession, remittances may decline more than GDP‐based forecasts imply. To examine this, we look at the three main factors that determine the volume of remittances: (a) the number of immigrants, (b) their employment status and (c) their earnings.

The number of potential remittance senders from the United States is very unlikely to decline substantially. 159 There are 45 million Latinos in the country, of which about 13 million send remittances. Of these, 8 million send remittances on a regular basis, and an additional 5 million are occasional senders. The net positive flow of immigrants may drop from its recent levels of 1 million per year (from all countries of the world), but it is extremely unlikely that the net flow would turn negative. Research has shown that LAC migrant’s main motivation is to send money back home and, since they lack access to social security or other sources of income, they typically go to extraordinary lengths to remain employed or find new employment. Only illegal migration could change quickly in response to economic cycles. However, illegal immigrants are quite resilient to worsening economic times because there are able to adjust more quickly than the native‐born competitors for the same jobs, as they are able to change residence swiftly for work‐related reasons. In addition, earning expectations in the home countries are not better and enhanced border controls dissuade migrants from temporarily returning home to explore other opportunities.

Figure 3: Average Weekly Wages of Hispanics in the Bottom Decile, by Quarter, Not Seasonally Adjusted,

Current Dollars

Source: Bureau of Labor Statistics, Current Employment Statistics survey

On the other hand, the U.S. labor market conditions of potential senders are deteriorating, and this is likely to decrease the average amount sent. While immigrants comprise a varied group, remittance senders to LAC are typically recently arrived, young, married men with little education, low earnings,

158 For a description of the forecast methodology see the annex to Rahta, Mohapatra and Xu (2008). 159 About three quarters of the remittances flow to the LAC region originate in the United States. For this reason, the impact of the financial crisis on remittances to Latin America depends mainly on the behavior and circumstances of migration to the U.S. and remittance senders already in the U.S.

126

and little familiarity with formal banking systems. As they accumulate more time in the United States, immigrants become less likely to send remittances and the average amount sent drops. Therefore, we looked at the unemployment rate for Mexican immigrants who arrived in the last decade and found that it rose to 12.8 percent in January 2009 from 8.4 percent a year earlier (overall U.S. unemployment rose from 5.7 to 8.6 percent during the same period.).

Despite these large changes in employment, for those who have jobs, wages have remained roughly the same or even increased. The median real weekly wages of Hispanics were at their highest level of all time in February 2009, and average real weekly wages of construction workers, were also at a historically high level. Furthermore, average wages of Hispanics in the bottom decile and quartile have been approximately flat since late 2008. Combining both effects, we find that the total earnings of likely remittance senders from the US have declined by about 4.4 percent. Although data on remittances relative to the income of individual migrants is not readily available, coupling median weekly earnings for recent Hispanic immigrants with average amounts of remittances suggests that, very roughly, remittances may constitute between 15 and 25 percent of a typical low‐income migrant’s earnings. Unless this proportion has changed substantially as a result of the crisis, GDP‐based projections of remittances to the LAC region seem to be robust.

Several factors explain why the impact of the crisis on remittances is not larger: (i) although migration is likely to decrease, remittances depend on the stock of migrants, not on their flows, and the stock of migrants is not expected to change significantly, (ii) the latest unemployment and salary figures for Hispanic citizens in the United States do not support a predicted large drop on remittances, (iii) historically, remittance flows worldwide have shown less volatility than other financial flows, including foreign direct investment and official development assistance, and (iv) not all immigrants send money home, so immigrant unemployment among those who do not send remittances, or send them very occasionally, will not have a significant effect on overall remittance volumes. Overall, the evidence of a connection between U.S. macroeconomic conditions and remittance flows is mixed, but even if there is some positive correlation, it is countered by deteriorating macroeconomic conditions in the receiving country.160 Further, the depth and global nature of the current crisis sets it apart from recent economic downturns, so historical patterns may not hold.

Implications for Poverty and Inequality in LAC

The relative importance of remittances in the receiving country’s economy and the position of remittance receivers in the income distribution —and thus the effect of remittances on poverty— also vary greatly by country. Estimates of the effect of the drop in remittances on poverty imply that a drop in remittances of 4.4 percent would increase poverty across the region ranging from a 0.3 percent point increase in the headcount in Chile (around 51,000 additional poor) to a 1.6 percent point increase in Haiti (around 152,000 additional poor).161 In Central America, Guatemala (125,000 additional poor) and Honduras (68,000 additional poor) would be most affected.

The impact of projected remittances changes on inequality is also likely to show wide variations, but data shortcomings make it difficult to draw conclusions. Unfortunately, not all household surveys have

160 Fajnzylber and López find that remittances are pro‐cyclical with respect to US economic growth and countercyclical with respect to recipient countries’ growth, with the former effect larger than the later. 161 We use elasticities of poverty to remittances calculated by Fajnzylber and López. The average elasticity for the region is ‐0.4. We use a $4/day poverty line, which is the average of the national poverty lines in the LAC region.

127

data on remittances and, for those who do, remittances questions have been added only recently. There is, however, some recent data for Guatemala and for Guyana, among the few countries where the last two household surveys included remittances data. In Guatemala, remittances as a share of expenditure have grown between 2000 and 2006 for all income groups, but they have almost tripled for the lowest quintile. Hence, any fall in remittances is likely to hurt a significant share of the poorest quintile, although not more than other income quintiles. Like the case of Guatemala, in Guyana remittances as a share of consumption for the lowest quintile have increased from 23 to 36 percent between 1992 and 2006, suggesting that the poorest quintile would be particularly hard hit from any decrease in remittances.

Conclusions

Overall poverty and inequality impacts from the recent crisis through the remittances channel are likely to be limited, but may be of some significance in a few countries. Although there is large uncertainty about the depth and duration of the crisis, the effect it will have on remittance flows, and its impact on poverty via the remittances channel, available evidence points to a relatively small decline of remittance flows to LAC, resulting in relatively small increases in poverty rates. Yet, these declines would imply falling into poverty for about 3.4 million additional people. More data collection is needed to be able to ascertain the distributional and gender impacts of the current crisis on remittances.

One of the few short run policy options available to LAC governments to reduce the impact of the crisis on poverty is related to the cost of sending remittances. For remittances originating in the U.S. these vary from US$ 9.17 for a US$ 200 transfer to El Salvador to US$ 19.04, on average, for the same amount transferred to Dominican Republic. Now more than ever it becomes imperative to reduce these costs. Possible ways to reduce these costs include: (i) issuance of consular cards that allow migrants to open bank accounts, (ii) postal agreements, (iii) promote the use of mobile telephones for money transfers among private providers of telecommunication services, (iv) improve information on transfer costs both among senders and recipients and (v) promote more competition in the market for these financial services, including lowering regulatory cost of opening bank branches.

For the most vulnerable recipient families, short run compensation of losses through existing or new social protection programs is also a possibility. In the long run, governments can continue to enhance the development impact of remittances. The impact of remittance transfers on the growth rate of recipient economies depends on whether this transfer is invested or consumed. In the LAC region, remittances are used mostly for consumption purposes. Governments can continue to provide incentives to increase the proportion of remittances devoted to investment, by raising education levels of remittance receivers (which increases the returns to investment); improving the functioning of financial systems so that, even if remittance recipients themselves are not inclined to invest, the proportion of remittances that is not consumed is channeled to investment through financial intermediation; and reducing macroeconomic distortions and institutional weaknesses that reduce incentives to invest.

128

129

9. WILL FDI BE RESILIENT IN THIS CRISIS?

Cesar Calderon and Tatiana Didier

February 2009* Although FDI flows have tended to remain resilient during previous crises, they may not behave in a similar fashion during the current crisis. Why? In past crises, the stability of FDI flows was significantly associated with an increase in mergers and acquisitions (M&A), reflecting "fire‐sale FDI". In the present crisis, by contrast, M&A activity decreased significantly in the last quarter of 2008, and this trend may continue as long as the global crisis constrain the purchasing ability of foreign (acquiring) firms. These developments further illustrate that the nature of the current crisis differs considerably from previous ones, suggesting that certain key lessons from past crisis lessons might not apply in the current context.

Introduction

The outlook for private capital flows to emerging market economies, and especially to Latin America, has deteriorated substantially in recent months. Net flows to the region are expected to decline to US$ 43 billion in 2009, down from US$ 89 billion in 2008. This implies a decline of approximately 75% from the record net flows in 2007 —US$ 183.6 billion (IIF, 2009). While all components of net capital flows are projected to decline, the largest drop is expected on net private debt flows —an estimated decline of US$ 40 billion in 2009.162

Recent studies have shown that the various types of financial flows have different dynamics along business cycles and in particular during crises.163 For example, using a panel dataset of 66 countries during 1970‐2003, Levchenko and Mauro (2007) find that net equity flows, FDI flows, are not only less volatile than net debt flows (i.e., portfolio debt flows, bank lending, and trade credit) but also more resilient during episodes of sudden stops. In fact, the authors find that net debt flows explain almost entirely the reversal in the financial account that characterizes a sudden stop.

The resilience of FDI flows during crises episodes have been attributed to an increase in fire‐sale foreign direct investment. The tightening of liquidity constraints for domestically owned firms during these episodes has been associated with an increase in foreign acquisitions, usually at significantly low prices. This foreign ownership may facilitate the access of financially‐constrained domestic firms to world capital markets by bringing transparency, better management, and improved technology.

In this note, we argue that, although FDI flows remained resilient during previous crises, they may not behave in a similar fashion during the current financial crisis. Why? Recent crises, including the Asian crisis for example, were circumscribed to the emerging market world. Firms

*LCR Crisis Briefs Series. 162 The decline in net private debt flows is even larger, more than US$ 100 billion, if we compare the net private debt flows in 2009 relative to its record level in 2007. 163 See, for example, Sarno and Taylor (1999) and Levchenko and Mauro (2007).

130

buying out Asian firms were not affected by the liquidity crisis and had ample access to financial resources. In contrast, the current crisis has imposed severe liquidity constraints not only on the owners of (financially‐constrained) firms in emerging markets but also on the potential foreign buyers of these firms.

We analyze the SDC mergers and acquisition database, which contains high frequency data on all cross‐border mergers and acquisitions since 1985. The dataset includes all corporate transactions (public and private ones) involving at least 5% of the ownership of a company, with a total of 607 thousands of transactions until November of 2008. Merger and acquisition deals usually take a long time to be completed. Thus, in order to capture the right incentives behind these transactions, we consider the announcement date as the date of a merger and acquisition. Lastly, all announced deals that have not been withdrawn are considered.

Evidence from previous crises

Although East Asian countries experienced a sharp reversal in net portfolio equity and debt flows during the 1997‐98 financial crises, they witnessed a significant increase in FDI inflows. Krugman (2000) and Aguiar and Gopinath (2005) provide evidence that this increase in FDI inflows was driven by an increase in foreign acquisitions.164 This finding is confirmed by the data shown in Figure 1, which plots the number of cross‐border merger and acquisition (M&A) deals during this period for East Asian countries.

The 1997‐98 crises also affected other emerging market countries, namely countries in LAC and Eastern Europe. Cross‐border merger and acquisition activity for these two regions are also plotted in Figure 1. The graph shows a clear upward trend in the number of cross‐border M&A deals over this period. For example, LAC‐7 countries had on average around 40 deals per quarter before the crises, whereas during the second half of 1998, this number increased around 50% to 60 deals per quarter. A similar pattern is observed in Eastern European countries after the Russian crisis in the third quarter of 1998. These countries went from an average of 26 deals per quarter before the crisis, to 52 deals per quarter six month after the onset of the crisis.

In line with the existing evidence, Figure 1 suggests that during a crisis affecting only emerging economies, foreign investors from developed countries still have access to financial resources. They are thus able to take advantage of cheaper investment opportunities in crisis‐affected countries. Consequently, there is a significant increase in foreign ownership of domestic firms in financially‐constrained countries.

The current financial crisis

In contrast with the Asian/Russian crises, the current financial meltdown was originated in the United States and has then spread to both developed and emerging countries. Consequently, severe liquidity constraints are affecting not only the owners of (financially‐constrained) firms in emerging markets but also the potential foreign buyers of these firms. In other words, cross‐

164 The evidence is explained not only by the elimination of policies unfavorable to foreign ownership in East Asia but also by the perception of multinational firms that they can buy Asian companies at significantly lower prices.

131

border mergers and acquisitions in emerging economies should not increase as it has during previous crises.

The data shown in Figure 2 confirms this hypothesis. If LAC‐7 countries are considered, the number of announced deals has fallen almost 60%, from around 105 deals per quarter in the second quarter of 2007 to only 44 in the last quarter of 2008. There is a similar decrease in the value of the cross‐border mergers and acquisitions. Foreign acquisitions amounted to about US$ 11.6 billions in the second quarter of 2007 and have declined to only US$4.6 billions in the last quarter of 2008. It should be noted that the decrease in M&A activity is not an issue only for Latin American countries. Similar patterns can be observed in other regions. For example, the number of cross‐border M&A deals has decreased 25% in East Asia and almost 70% in Eastern Europe between the second quarter of 2007 and the last quarter of 2008.

Furthermore, the main argument discussed in this note also implies a similar dynamics for the evolution of mergers and acquisitions in the United States. As the crisis has spread to the rest of the world, we should also observe a decrease in cross‐border M&A in the United States, which is confirmed in Figure 2. The number of cross‐border M&A deals starts to fall in the first quarter of 2008 and has followed a downward trend since then, decreasing almost 70% in the last quarter of 2008 vis‐à‐vis the second quarter of 2007.

Conclusion

In sum, the evidence presented in this note suggests that FDI flows might not remain as resilient as they have been during previous financial crisis. A larger decrease in foreign direct investments should be expected. These developments further illustrate that the nature of the current crisis differs considerably from previous ones, suggesting that certain key lessons from past crisis lessons might not apply in the current context.

132

Figure 1Cross-Border M&A Activity: Number of Deals

0

10

20

30

40

50

60

70

80

1996

.4

1997

.1

1997

.2

1997

.3

1997

.4

1998

.1

1998

.2

1998

.3

1998

.4

1999

.1

1999

.2

1999

.3

LAC-7 East Asia Eastern Europe

Figure 2Cross-Border M&A Activity: Number of Deals

0

20

40

60

80

100

120

2006

.4

2007

.1

2007

.2

2007

.3

2007

.4

2008

.1

2008

.2

2008

.3

2008

.4

0

50

100

150

200

250

LAC-7 East Asia Eastern Europe United States (RHS)

133

References

Aguiar, M., Gopinath, G., 2005. Fire‐sale foreign direct investment and liquidity crises. The Review of Economics and Statistics 87(3), 439‐452.

Institute of International Finance, 2009. Capital flows to emerging market economies. Washington, DC: Institute of International Finance, January.

Krugman, P., 2000. Fire‐sale FDI. In: Edwards, S. (Ed.), Capital Flows and the Emerging Economies: Theory, Evidence, and Controversies. Chicago, IL: The University of Chicago Press for the NBER, pp. 43‐60.

Levchenko, A., Mauro, P., 2007. Do some forms of financial flows help protect against “sudden stops”? The World Bank Economic Review 21(3), 389‐411.

Sarno, L., Taylor, M., 1999. Hot money, accounting labels, and the permanence of capital flows to developing countries: an empirical investigation. Journal of Development Economics 59, 337‐364.

134

135

10. PATTERNS OF FINANCING DURING PERIODS OF HIGH RISK AVERSION: HOW HAVE LATIN FIRMS FARED IN THIS CRISIS SO FAR?

Tatiana Didier

May 2009*

Abstract

This note examines the extent to which firms in Latin America have been able to raise capital through debt and equity securities as well as syndicated loans, both abroad and domestically, since the onset of the 2008 global financial crisis. The public and the private sectors alike lost access to foreign sources of financing during the height of the turbulence. Furthermore, two months after the Lehman Brothers’ collapse, only government‐owned firms and governments themselves were able to re‐enter international markets to some extent and raise capital. Thus, the evidence suggests an important role for government guarantees in attracting foreign investors in times of high risk aversion. In domestic and syndicated loan markets, there has been a marked decrease in the total amount raised, although they have remained a viable option for the private sector in Latin America. To the extent possible, non‐government borrowers have been able to raise capital in these markets and have generally met their rollover needs. In contrast, the role of sovereign guarantees in attracting local investors seems to have been more important in Eastern Europe and Southeast Asia, where government entities have accounted for respectively 80 and 44 percent of all new issues in local markets, compared to less than 15 percent in LAC. Introduction

Governments in advanced economies, especially in the U.S., have announced large fiscal stimulus and financial rescue packages to help them recover from the current economic downturn. However, these packages will need financing. How will this affect the access of emerging economies to foreign capital markets? These packages may generate a global crowding out by mobilizing savings towards richer countries and elevating the real interest rate which, in turn, may raise the cost of borrowing. In other words, firms and governments in developing countries might find it harder, and possibly more costly, to raise capital in international financial markets. In the same fashion, governments throughout emerging markets, including several Latin American countries, have also announced relatively large fiscal stimulus packages to help their economies recover from their current slump. Hence, the question on how to finance these packages also arises. Will governments rely on domestic capital markets, thus possibly affecting the access of the private sector to local sources of financing? This note aims to describe the current evolution of access to capital markets, both domestically and abroad, with a focus on whether emerging markets are being crowded out of international capital markets, and if so, the role of domestic capital markets.

Access to international capital markets by the private sector in emerging market economies has deteriorated substantially since the start of the current financial crisis. This is evidenced by high frequency data on domestic and cross‐border issuance of securities (equity and debt) and syndicated loans from 1990 to February 2009, which this notes analyzes. The dataset (SDC New Issues database)

*LCR Crisis Briefs Series.

136

includes all transactions by the private sector in local and international capital and syndicated loan markets, and by governments in foreign capital markets. Government issues in local markets are not covered in this dataset and hence not addressed in this note. The local banking sector is only covered through the extent of local banks’ participation in the syndicated loan market. This type of evidence provides a clear, although incomplete, assessment of access (or lack thereof) to finance in local and international capital markets by the private and public sectors, as well as their rollover needs.

Volume of New Security Issues

During 2006 and 2007, the seven largest countries of Latin America (LAC‐7 countries) raised a total of $167 and $184.8 billion, respectively, with an average $13.9 and $15.4 billion per month, either through domestic or foreign capital markets in equity, bonds, and syndicated loans. These numbers are even more striking if the first three quarters of 2008 are considered, in which an average of $22.4 billion per month was raised, amounting to a total of $201.7 billion. However, between October 2008 and February 2009, firms and governments in Latin America were only able to obtain around $11 billion per month, 58% less than the same period in 2007‐08. Figure 1 shows the substantial decrease in the issuance of new securities or syndicated loans. New corporate and sovereign issues abroad since September 2008 have fallen 10% compared to the same period a year earlier, whereas in domestic markets and in the syndicated loan market, the amount raised by corporations in 2009 represents a striking decline, of 71% and 89%, respectively. Despite this marked decrease in the amount of new syndicated loans, they still accounted for most of the new issues during the height of the crisis, from September to November 2008. New loans accounted for, on average, 78.2% of all new issues during this period. During the same period a year earlier, they accounted for only 30% of capital raisings, where both new security issues at home and abroad were significantly larger.

Rollover Needs of the Private Sector

Figure 2 allows a comparison of new capital raised with expiring liabilities (aka rollover needs) of the private sector. It suggest that LAC‐7 countries seem to have faced a credit crunch in October 2008, i.e., at the height of the crisis.165 Mexico and Chile were the most affected with a gap between rollover needs and new issues of around $300 million. The other LAC‐7 countries also faced a large gap. It should be noted though that $18 billion of the $21.3 expiring liabilities in October 2008 can be accounted for by a maturing syndicated loan of the Brazilian mining giant Companhia Vale do Rio Doce, which had raised $50 billion in January of 2008.

The aggregate numbers shown in Figure 2 also suggest that the private sector was generally able to meet its rollover needs from November 2008 onwards, even though new issues have fallen significantly in all markets throughout Latin America. For instance, the private sector in LAC‐7 countries was able to raise $13.3 billion in excess of its expiring liabilities between November 2008 and February 2009. Furthermore, most of the new capital raised in this period by the private sector, over 97% of the total amount, was raised locally or in the syndicated loan market. In other words, although foreign capital markets have remained mostly closed since September 2008, firms have been able to rely on local markets or on the syndicated loan market for both rollover and new issues.

165 It is possible that firms raise capital in anticipation to its rollover needs in the near future. Hence, the comparison on a monthly basis of these two figures might not be an accurate measure of the gap in firms’ rollover needs. A detailed firm‐level analysis of the timing of new issues is needed but it is beyond the scope of this short technical note.

137

The Role of Governments and Government‐Owned Firms

The data presented in Figures 1 and 2 and described above suggests a decline in access to new capital in all markets since September 2008 for the private sector. Furthermore, Figure 3 shows that governments or government‐owned firms have been responsible for a significant share of the total capital raised since then, and especially since December 2008. Hence, the date suggests an important role for government guarantees in attracting investors and syndicated lenders to Latin firms in times of high risk aversion.

For instance, over 50% of the total amount raised can be accounted by governments or government‐owned firms alone in December, almost 66% in January, and more than 30% in February. However, between January and September 2008, the total capital raised by these two agents was on average only 11.5% of the total amount being raised every month. As depicted in Figure 3, the private sector might have been hit harder than the public sector during this crisis. In Figure 4, domestic and foreign markets are analyzed separately to evaluate the role of sovereign guarantees in different markets.

a. Foreign Capital Markets

Foreign capital markets were closed for two months after the Lehman/AIG episode as shown in the top panel of Figure 4. However, since December 2008, the public sector (including government‐owned firms) has been able raise new capital abroad, being the first sector to tap into new capital in international capital markets. Governments from Brazil, Colombia, Mexico, and Peru issued new debt abroad. Furthermore, between September 2008 and February 2009, only the public sector has had access to foreign capital, with the only exceptions being InPar S.A. from Brazil and Mexoro Minerals Ltd. from Mexico. In other words, governments or government‐owned firms have been responsible for close to 100% of all foreign new issues since September 2008. While this prevented the total of new issues in foreign markets from declining too much, it also suggests a potentially important role for government guarantees in attracting foreign investors in times where access to markets for the private sector is hindered by high risk aversion.

b. Syndicated Loan Markets

Contrary to the patterns observed in foreign securities markets, the middle panel in Figure 4 suggests that syndicated loans have been accessible to the private sector, including both government‐owned and non‐government‐owned firms. However, lack of data prevents an analysis of the role of sovereign guarantees in loans given by the domestic banking sector.

c. Domestic Capital Markets

The fall in new capital raising issues in local markets becomes evident in the bottom panel of Figure 4.166 On average, the private sector was able to raise $5.5 and $5.6 billion per month from January 2007 to September 2008, whereas in the period October 2008 to February 2009 the monthly average decreased to $2.5 billion. Nevertheless, as opposed to the patterns observed in the foreign markets, there is little evidence that government guarantees are important for attracting capital from local investors during periods of high risk aversion. The amount raised by government‐owned firms represents less than 15% of the total amount raised through new equity or debt issues in domestic markets. If the same period in the previous year is considered, the percentage is similar, at around 10%.

166 Government issues in local markets are not included in these numbers. The local banking system is also not analized.

138

The total amount raised in domestic markets has not been concentrated in a particular country. New issues throughout the region are still taking place. For instance, on average Brazil and Mexico accounted for a larger share of the amount raised locally, around 40% and 20% in the period from October 2008 to February 2009, followed by Colombia and Chile, which represented 11% and 10%, respectively. However, if the same period a year earlier is considered, Brazil and Mexico accounted for 71% and 25%, respectively, whereas Chile and Colombia represented together less than 5%. In February 2009, new issues by Colombian and Argentinean firms represented 33% and 35% of the total amount raised, respectively, suggesting that sizeable new issues across the region are taking place.

Lastly, despite this fall in the amount of capital raised in local markets, the new capital raised has not been concentrated in few issues by large firms. The number of firms that have been issuing new capital has been comparable to historical averages. For example, in December 2008, 4 firms from Argentina, 9 firms from Brazil, and 7 from Mexico raised new capital in local markets. An exception though is Colombia where a greater number of firms have gained access new capital in local markets recently – on average only 1 firm raised new capital locally in the 2000s, whereas in the first two months of 2009, almost 5 firms each month had access to new capital locally.

Comparison with other Emerging Markets

Firms in other regions of the world are facing similar problems to the ones faced by Latin American companies. Figure 5 shows capital raising activity in Southeast Asia and Eastern Europe in both domestic and foreign markets. Similarly to the patterns observed for LAC‐7 countries, access to foreign capital markets has been almost non‐existent for the 2‐month period after the spread of the turmoil in U.S. financial markets to the rest of the world. Furthermore, only governments or government‐owned firms have been able to raise any new capital in foreign markets since then. Once more, close to 100% of all foreign issues can be traced back to the public sector.

Similarly to what has been observed in Latin America, new issues in domestic markets have decreased in both Eastern Europe and Southeast Asia, as shown in Figure 5. However, while domestic markets remained a viable source of financing for the private sector in Latin America, government guarantees might have played a more important role in domestic markets in these two regions over this period. For instance, government‐owned firms have been responsible for 44% of all new issues in domestic capital markets in Southeast Asia since October 2008, whereas previously in 2008, the share of new issues by government‐owned firms has been on average less than 25% every month. The patterns in Eastern Europe are even more striking: capital raised by government‐owned firms represented 80% of all new issues in local markets after October 2008 versus 20% in the first nine months of 2008.

Conclusion

New security issues in international capital markets throughout emerging markets have dropped sharply since the onset of the 2008 global financial crisis. Both the public and the private sector lost access to foreign sources of financing during the height of the turbulence. However, since December 2008, only two‐months after the Lehman event, only governments or government‐owned firms from for Latin America have been able to raise capital abroad. Similar patterns are observed in Eastern Europe and Southeast Asia. Therefore, the evidence suggests an important role for government guarantees in attracting foreign investors in times of high risk aversion. In domestic markets, the amount of new issues has also decreased significantly during the current financial crisis. However, while domestic and syndicated loan markets remained a viable source of financing for the private sector in Latin America,

139

government guarantees seem to have played an important role in Eastern Europe and Southeast Asia. In sum, the private sector in Latin America, and in emerging markets more broadly, has been living in a world of scarce foreign capital and have been relying, to the extent possible, on local markets, on domestic banking sources of finance, and on their own cash, to cover its financing needs. Without access to foreign markets and with lesser ability to raise capital domestically, the private sector has been facing stronger credit constraints than the public sector (including government‐owned firms) during these last turbulent months.

140

Annex 1. Figures

* Does not include government bonds in domestic markets.

Figure 1. Capital Raising Activity by LAC-7 Countries

Capital Raised

0

10,000

20,000

30,000

40,000

50,000

60,000

May

-07

Jun-

07Ju

l-07

Aug

-07

Sep-

07O

ct-0

7N

ov-0

7D

ec-0

7Ja

n-08

Feb-

08M

ar-0

8

Apr

-08

May

-08

Jun-

08Ju

l-08

Aug

-08

Sep-

08O

ct-0

8N

ov-0

8D

ec-0

8Ja

n-09

Feb-

09

US$

Mill

ioon

Domestically* Abroad Synd. Loans

Figure 2. LAC-7: Capital Raising Activity and Rollover Needs of the Private Sector


All Markets*

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

Jan-

07

Mar

-07

May

-07

Jul-0

7

Sep-

07

Nov

-07

Jan-

08

Mar

-08

May

-08

Jul-0

8

Sep-

08

Nov

-08

Jan-

09

Mar

-09

May

-09

Jul-0

9

Sep-

09

Nov

-09

Total Amount Raised Rollover Needs

141


Figure 3. The Role of Governments

Percentage of Capital Raised by Governments or Government-Owned Firms

0%

10%

20%

30%

40%

50%

60%

70%

Nov

-07

Dec

-07

Jan-

08

Feb-

08

Mar

-08

Apr

-08

May

-08

Jun-

08

Jul-0

8

Aug

-08

Sep-

08

Oct

-08

Nov

-08

Dec

-08

Jan-

09

Feb-

09

142


Figure 4. The Role of Governments in Domestic and Foreign Markets

Total Amount Raised Abroad

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

May

-07

Jun-

07Ju

l-07

Aug

-07

Sep-

07O

ct-0

7N

ov-0

7D

ec-0

7Ja

n-08

Feb-

08M

ar-0

8A

pr-0

8M

ay-0

8Ju

n-08

Jul-0

8A

ug-0

8Se

p-08

Oct

-08

Nov

-08

Dec

-08

Jan-

09Fe

b-09

US$

Mill

ioo n

Total Government or Government-Owned Firms

Total Amount Raised Domestically*

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

May

-07

Jun-

07

Jul-0

7

Aug

-07

Sep-

07

Oct

-07

Nov

-07

Dec

-07

Jan-

08

Feb-

08

Mar

-08

Apr

-08

May

-08

Jun-

08

Jul-0

8

Aug

-08

Sep-

08

Oct

-08

Nov

-08

Dec

-08

Jan-

09

Feb-

09

US$

Mill

ioon

Total Government-Owned Firms

Total Amount Raised through Syndicated Loans

02,0004,0006,0008,000

10,00012,00014,00016,00018,00020,000

May

-07

Jun-

07Ju

l-07

Aug

-07

Sep-

07O

ct-0

7N

ov-0

7D

ec-0

7Ja

n-08

Feb-

08M

ar-0

8A

pr-0

8M

ay-0

8Ju

n-08

Jul-0

8A

ug-0

8Se

p-08

Oct

-08

Nov

-08

Dec

-08

Jan-

09Fe

b-09

US$

Mill

ioon

Total Government-Owned Firms

143

* D

oes n

ot in

clud

e go

vern

men

t bon

ds in

dom

estic

mar

kets

.

Figu

re 5

. Cap

ital R

aisi

ng A

ctiv

ity in

the

Wor

ldE

aste

rn E

urop

eSo

uthe

ast A

sia

Tot

al A

mou

nt R

aise

d A

broa

d

02,

000

4,00

06,

000

8,00

010

,000

12,0

0014

,000

16,0

0018

,000

20,0

00

May-07Jun-07Jul-07

Aug-07Sep-07Oct-07

Nov-07Dec-07Jan-08Feb-08Mar-08Apr-08

May-08Jun-08Jul-08

Aug-08Sep-08Oct-08

Nov-08Dec-08Jan-09Feb-09

US$ Millioon

Tota

lG

over

nmen

t or G

over

nmen

t-Ow

ned

Firm

s

Tot

al A

mou

nt R

aise

d D

omes

tical

ly*

050

01,

000

1,50

02,

000

2,50

03,

000

3,50

04,

000

4,50

05,

000

May-07

Jun-07Jul-07

Aug-07Sep-07

Oct-07Nov-07

Dec-07Jan-08

Feb-08Mar-08

Apr-08May-08

Jun-08Jul-08

Aug-08Sep-08

Oct-08Nov-08

Dec-08

Jan-09

Feb-09

US$ Millioon

Tota

lG

over

nmen

t-Ow

ned

Firm

s

Tot

al A

mou

nt R

aise

d A

broa

d

0

2,00

0

4,00

0

6,00

0

8,00

0

10,0

00

12,0

00

14,0

00

May-07Jun-07Jul-07

Aug-07Sep-07Oct-07


May-08Jun-08Jul-08

Aug-08Sep-08Oct-08


US$ Millioon

Tota

lG

over

nmen

t or G

over

nmen

t-Ow

ned

Firm

s

Tot

al A

mou

nt R

aise

d D

omes

tical

ly*

0

5,00

0

10,0

00

15,0

00

20,0

00

25,0

00

30,0

00

35,0

00

May-07Jun-07Jul-07

Aug-07Sep-07Oct-07


May-08Jun-08Jul-08

Aug-08Sep-08Oct-08


US$ Millioon

Tota

lG

over

nmen

t-Ow

ned

Firm

s