SSB - Monetarism and Structural Adjustment - The Case of ...Strtrl Adjtnt n Mzb pprtr 9/11 h frth hptr ntn rtrpt n pt-nd-pndn n pl n Mzb h frt prd ftr 1975 hrtrzd b ntrll plnnd n-th

94/11 Rapporter Reports

Klaus Mohn

Monetarism and Structural Adjustment- The Case of Mozambique

)4/11 Rapporter Reports

Klaus Moh►

Monetarism and Structural Adjustment- The Case of Mozambique

Statistisk sentralbyrb • Statistics NorwayOslo-Kongsvinger 1994

Mindre enn 0,5av den brukte enhetenMindre enn 0,05av den brukte enhetenForelopige tallBrudd i den loddrette serienBrudd i den vannrette serien

Less than 0,5 of unitemployedLess than 0,05 of unitemployedProvisional or preliminary figureBreak in the homogeneity of a vertical seriesBreak in the homogeneity of a horizontal series

0

0,0

Standardtegn i tabellerTall kan ikke forekommeOppgave manglerOppgave mangler forelopigTall kan ikke offentliggjoresNull

Symbols in tables Category not applicableData not availableData not yet availableNot for publicationNil

Symbol

ISBN 82-537-4005-0ISSN 0332-8422

Emnegruppe59 Andre samfunnsokonomiske emner

EmneordMakrookonomiMozambikVekstmodellØkonometri

Design: Enzo Finger DesignTrykk: Falch Hurtigtrykk

Abstract

Klaus Mohn

Monetarism and Structural Adjustment The Case of Mozambique

Reports 94/11 • Statistics Norway 1994

This report is the result of a project undertaken by the Research Department of Statistics Norway to recapitulate theimpact of economic policy on macroeconomic development in Mozambique. A brief review of the historical andeconomic origins of the structural adjustment paradigm is offered in the first chapter, whereas the second chaptercontains a survey of the background and theoretical content of monetarism in development economics. The post-independence economic policy of Mozambique is reviewed in the third chapter.

A panel data set containing monetary macroeconomic indicators for ten countries in sub-Saharan Africa over theperiod 1980 to 1991 is presented and commented in the fourth chapter. Finally, a simple single-equation econome-tric model of economic growth is constructed and simulated to test and illustrate some hypotheses concerning mone-tarism and structural adjustment. The results seem to suggest a significant, but very modest spill-over from monetaryto real economic variables. These results are probably due to the regulated macroeconomic environment which hasbeen quite typical for developing countries in sub-Saharan Africa.

Keywords: Econometrics, Growth models, Macroeconomics, Mozambique.

3

Rapporter 94/11 Structural Adjustment in Mozambique

Contents1. Introduction 71.1 Overview 71.2 What triggered the structural adjustment approach? 81.3 The never-ending debate 9

2. Monetarism in developing countries 112.1 The fundamental ideas 112.2 A monetarist macroeconomic framework 122.3 The rationale for financial sector reforms 132.4 Why monetarism might not work 15

3. Structural adjustment in Mozambique 173.1 Post-independence economic policy 173.2 The Economic Recovery Program 173.3 Looking ahead 19

4. Ten countries in sub - Saharan Africa 214.1 Introduction 214.2 Regional differences in population and production 214.3 Monetary growth and inflation 234.4 Interest rates 244.5 Terms of trade and the current account 25

5. Testing the impact of policy change 275.1 A heuristic model of growth 275.2 Econometric specification 295.3 Estimation 305.4 Discussion of results 31Country-specific fixed effects 31Broad money 32Terms of trade 32The exchange rate 32Net official transfers 33Interest rates 34The structural model 34

6. Simulations and policy experiments 376.1 Post-sample simulation 376.2 The reference scenario 376.3 Discrete shocks in policy variables 386.4 Continuous policy experiments 396.5 Accommodating monetary control 40

7. Conclusions 43

8. Literature 45

Issued in the series Reports from Statistics Norway 47

5


1. Introduction*

1.1 OverviewThis report represents the final documentation of a projectundertaken by the Research Department of Statistics Nor-way to recapitulate the impact of the late macroeconomicpolicy in Mozambique, with sidelong glances and compari-sons to the development of other economies in the region.The time span of primary interest is the period after 1987, inorder to evaluate the effects of the structural adjustmentplan implemented in cooperation with the multinationaldonor community

The introduction offers a brief retrospect of the history ofmonetarism as a school of thought within developmenteconomics. Various attempts to evaluate the impact of struc-tural adjustment are then reviewed. The general conclusionfrom empirical research on the area is that the structural ad-justment programs seem to have had moderate beneficiaryeffects when all adjustors are included in the data set. How-ever, this conclusion soon trembles if the scope is restrictedto sub-Saharan Africa, where basic conditions for economicgrowth seem to be less developed than for e.g. the averageWorld Bank client. The history of Mozambique makesthese features especially important, but the level at whichthe Mozambican economy had settled by the middle of the1980's still seems to have offered an outstanding potentialfor short to middle-term gains from structural adjustment.

General ideas of monetarism in development economics arereviewed somewhat closer in chapter 3. The main concernfor extreme monetarists is to accommodate a stable econ-omy by controlling the rate of monetary growth (and noth-ing else). The underlying idea is that the real economy hassatisfactory self-controlling properties if only macroecon-omic stability and low inflation is secured. Some detrimen-tal effects of inflation in developing countries are thereforediscussed separately to illustrate how high inflation may re-duce the cost-competitiveness, discourage savings and in-vestments, and cause capital flight. These effects are partlydue to standard macroeconomic theory, and partly to pecu-liarities of the economies in developing countries.

Next, a simple formal macroeconomic accounting frame-work is set up to illuminate the different sources of mon-etary growth in a stylized developing economy. Financialgaps are identified in the private sector, the public sector,and by the current account. All these financial gaps can bedirectly linked with the money supply via simple nationalaccounting equations. Thus, changes in the money supplyare determined by the savings surplus in the private sector,the public budget deficit, and the current account.

Financial sector reforms have developed into an importantand integrated part of most structural adjustment programs.The sub-Saharan experience with such reforms is also re-viewed in chapter 2, in trying to draw some conclusions forfuture plans for the financial sector. The outcome of finan-cial sector reform programs so far seems to suggest that toomuch has been spent, and too little has been achieved. Theoverall experience is that a restructuring the financial sys-tem has few or no positive effects unless it is accompaniedby close attention the development of the real economy.For the future, financial reform programs are therefore rec-ommended to take a somewhat more modest approach, start-ing from basics, always waiting for backing from the realside of the economy.

The last part of chapter 2 presents some critical remarks tothe policy recommendations of typical monetarist structuraladjustment programmes. First, it is stressed that financialliberalization and an according hike in interest rates mayhave significant contractive effects via the supply-side ofthe economy, because the price of credit plays an importantrole for the variable costs of most producers in developingcountries. Second, the exchange rate is equally importantfor the domestic price-formation, as imported inputs are ofextreme importance for developing countries in general,and for manufacturing industries in particular. This impliesthat also devaluations and cuts in the inflow of capital mayhave suppressive effects on the national income level, atleast in the short run. As both these mechanisms workmainly through the supply-side of the economy, they leaveroom for stagflationary effects of monetarist policies in de-veloping countries.

* This project has been supported by a grant from the Norwegian Corporation for Development Cooperation (NORAD). I am indebted to Adne Cappelen,Stale Johnsen, and Olav Ljones for expert comments and suggestions.

7

Structural Adjustment in Mozambique Rapporter 94/11

The fourth chapter contains a retrospect on post-inde-pendence economic policy in Mozambique. The first periodafter 1975 was characterized by a centrally planned econ-omy with extreme lack of human and physical capital re-sources, both extremely important factors for economic de-velopment. This situation was worsened by the civil war,which throttled the economy for almost the entire last de-cade. The result is an economy where standard macroecon-omic analysis has almost no interest, at least not before1987. Economic development seems to have been triggeredby the inflow of capital accommodated by the Economic Re-covery Plan. However, the medium-term potential for econ-omic growth seems to have been exhausted by the begin-ning of the 1990's. The peace treaty of 1992 offers quitesome optimistic prospects, but the costs of war are formi-dable, and so is the growing debt-burden. This still makesMozambique heavily dependent on external funds, also forthe unforeseeable future.

Chapter 4 offers a comparison of social and economic indi-cators among ten countries in sub-Saharan Africa during theperiod 1980-1991.To track the development that has takenplace over the last decade, the sample period is divided intotwo sub-periods. Different growth rates and levels are thenpresented and compared for the two sub-periods, and for thedifferent countries. The main impression is that Botswanastands out as a wealthy nation compared to most other coun-tries of the region. On the other hand, Mozambique is thepoorest country of the sample, as measured by annual percapita gross national product (US$ 80). Regarding the econ-omic performance over the last decade, the general im-pression is mixed. For Mozambique there seem to havebeen a significant turnaround in domestic production imme-diately after the implementation of the Economic RecoveryPlan, but systematic success of economic policy is not veri-fied by the descriptive analysis of chapter 4.

Econometric methods are applied in chapter 5 to establishan empirically-based relationship between the growth in thegross domestic product and different policy variables as-sumed to be under some degree of public control. The vari-ables of our econometric model include broad money sup-ply, the terms of trade, the exchange rate, net official trans-fers, and interest rates. The estimation suggests that the rateof economic growth is invariant to devaluations and change-s in net official transfers. However, the other variables seemto influence the rate of economic growth during the 1980's,both via short-effects, and via long-run structural effects.

To illustrate the workings of the model, it is first simulatedon historical data for Mozambique over the period 1986-1991, and the results lend support to the idea that macro-economic analysis hardly was applicable for the Mozambi-can economy prior to 1987. However, the model offers afairly good description of the evolution of GDP under theEconomic Recovery Plan. Second, the model is simulatedfor the period 1993-2005 to illustrate the projected effectsof some policy experiments. A slow-down in the rate ofmonetary growth produces a reduction in the rate of econ-

omic growth, and is thus contractionary, both in the shortand long run. An improvement in the terms of trade seemsto have undisputed advantageous effects, whereas the effectof an increase in the interest rates is mixed. An once-and-for-all increase in the interest rate is contractive in the shortrun, but increases the long-run rate of growth via the structu-ral parameter. This may be interpreted by applying the ideasof the financial repression literature, where regulated inter-est rates are claimed to discourage domestic savings,thereby limiting the funds available for domestic invest-ments.

Chapter 6 outlines some experimental simulations and pro-jections for the economy of Mozambique. These experi-ments illustrate some important points concerning the ef-fects of macroeconomic policy. First, there is a dilemma in-volved in the choice of gradual versus "cold turkey"strategies. Gradualism often dominates the alternative ap-proach with respect to expected viability of the structural ad-justment plans, but may be less effective when it comes tomeasuring the exact changes in economic indicators overtime. Second, the presentation of chapter 6 underscores theimportance of patience when implemented macroeconomicstabilization plans for developing countries are to be evalu-ated. The reason is that the macroeconomic processes seemto be really sluggish in these countries.

Finally, chapter 7 offers some concluding remarks and callsfor further research on macroeconomic modelling for pol-icy-planning in developing countries.

1.2 What triggered the structural adjustmentapproach?

During the 1970's, important parts of the world economysaw dramatic cyclical shifts and turns. The oil price shocksof 1973 and 1979, increasing inflationary pressures in thedeveloped countries, and a rightward shift in politics in im-portant western countries, produced a new scepticism to-wards the ruling approach to macroeconomics. It grew in-creasingly clear that Keynesian macroeconomics could notprovide a fully satisfactory explanation of the inflationarypressures that seemed to have been caused mainly by the oilprice shocks. Therefore, a worldwide concern for supplyside macroeconomics emerged. In short, the economic pol-icy of the industrialized countries reflected a switch of inter-est from aggregate demand management and employmentconsiderations towards increasing monetary restraint and in-flation control.

The developments in the real world did not escape attentionin the more academic fields of macroeconomics and devel-opment economics. Thus, the popularity of the advocates ofstabilization, deregulation, and liberalization went up. Onthe other hand, the former neo-keynesian consensus lostground, both in the industrialized world and in the develo-ping countries. The content of the emerging stabilizationplans for developing countries must be understood on thisbackground of historical events and real world phenomena.Together with the second oil price shock, the drop in OECD

8


growth rates caused a dramatic negative external shock tothe economies of the third world. The implication was adeterioration in the terms of trade in most developing coun-tries, again causing increasing deficits, both internal and ex-ternal. The developing countries had to react to the externalshocks, and those who responded quickly could somehowchoose their own line of policy. But most developing coun-tries were not able to respond, and so the misery developeduntil the eruption of the debt crisis of 1982.

The result was that the multinational donor society feltforced to take action on behalf of the countries that hadfailed to adjust their economic policy to the shifts in themacroeconomic climate. Structural adjustment lending andgeneral conditionality principles were proposed as answersto the problems. Officially initiated by World Bank Presi-dent MacNamara in Manila in 1979, programme lending"has ever since been associated with a particular lendingwindow of the Bank: An instrument defined as quick-dis-bursing, exceptional balance of payment financing, basedon economy wide conditionality" (Demery [1993]). Havingcorrected for distortions by fictitious shadow-prices for a de-cade, it was now time to get actual prices right. The econo-mists behind the resulting adjustment plans were indeed in-fluenced by the recent shift of paradigm in (esp. US, UK,and German) macroeconomics. Therefore, the plans thatwere designed came to rely quite heavily on supply-sideeconomics, and so-called monetarism.

The structural adjustment programmes were intended tocause long-term changes, which in turn could increase theeconomic growth potential, and secure a sustainable repay-ment rate of the national debt. This conflicted with the for-mer stabilization policies, which were criticized for narrow-ing the scope to short-run demand management in responseto temporary and reversible shocks. A major differencefrom earlier approaches was also the emerging negligenceof special peculiarities of the economy of developing coun-tries. The new programmes of structural adjustment re-flected a view that the macroeconomic mechanisms of theindustrialized world are equally valid also in the developingpart of the world.

This was largely how monetarists overcame structuralistsin the approach to macroeconomics in developing countries.Thus, the worldwide shift of macroeconomic paradigm to-wards supply side economics came to mean a lot also formacroeconomic policy design in the third world.

1.3 The never-ending debateFor many developing countries, the entrance of the 1980'scame to mean not just any new decade, but a total shift ineconomic conditions and policy. Structural adjustment pro-grammes were introduced all over the world, but the resultswere at best mixed, at least in the short run. Together withthe debt crisis of 1982 and political instability in many de-veloping countries, the 1980's came to involve a signifi-cantly renewed economic and political climate for develo-ping countries.

The structural adjustment programmes focused on threebroad issues: Macroeconomic stability, lifting price con-trols, and public sector reform. Macroeconomic stabilitywas to be accomplished by tight money and credit policiesto bring down inflation, deregulation should handle the con-version to market prices, and privatization was the domi-nant instrument in the public sector reforms. However, thedesigners of the structural adjustment programmes were notallowed to work undisturbed, and were subject to quitemassive criticism from the start. The former structuralists re-grouped and formed the neo-structuralist macro critique.Again, attention was paid to the distinctive features of theeconomy in developing countries. But a main differencewas now the attempts to fit the new ideas into a frameworkthat also left space for supply side mechanisms. Anexample is Bruno (1979), who incorporates imperfect com-petition and unofficial credit markets in a macroeconomicmodel with flexible prices. Over the 1980's, the neo-structu-ralist macro critique produced a number of interesting ideas,some of which also surfaced in practical applications of de-velopment economics.

Another source of criticism was the social content of thestructural adjustment programmes. This was brought to thesurface of public debate by UNICEF in 1987, who arguedthat the lending programmes of the World Bank onlyexacerbated the adverse income distribution and social un-rest. Slowly, the response to this appraisal surfaced throughthe introduction of so-called social action programmes.After a time, pre- and post-programme income surveys alsocame into use, to map the distributional impact of the struc-tural adjustment programmes. This critique has indeed beenfruitful, and today the programmes of the World Bank arenot only viewing to efficiency matters, but they also showconcern for the income distribution. It has also been ac-knowledged that a smoother transition to a market economymay substantially influence the political viability of thestructural adjustment programmes (cf. the recent develop-ments in Russia).

The World Bank has reviewed the practice of adjustmentlending in a number of documents (e.g. World Bank [1988,1990, 1992]), and Summers and Pritchett (1993) sum-marizes some of the findings of these reports. There aremajor identification problems emerging when the structuraladjustment programmes are to be evaluated. First, it isproblematic to separate the effects from adjustment fromother aspects influencing the economy. Although a develo-ping country has undergone structural adjustment, the im-provement in performance may actually be contributed toother factors. More generally, a structural adjustment pro-gramme's ability to predict an upturn does not automat-ically imply that the upturn was caused by the programme.Second, a structural adjustment programme normally con-sists of two parts: A macroeconomic policy component anda capital inflow component. To distinguish between partialeffects of these two components is quite arduous, not to sayimpossible. Third, it is important to clarify to what extentthe countries involved really have submitted to the recom-

9


mendations made by the IMF and the World Bank. In Af-rica, both Zambia and Tanzania have emerged as typicalon/off adjusters, and therefore the economic performance ofthese countries can not be attributed solely to the adjust-ment programmes (or other factors, for that matter).

As Summers and Pritchett (1993) develops into a somewhatpartial treatment of World Bank policy, it does not deservetoo extensive treatment in this connection. Still, their con-clusions are worth noticing. Their scope is on adjustmentlending, and according to the World Bank reviews adjust-ment lending generally causes faster growth, higher exports,reduced deficits, and increased savings. Thus, adjustmentlending is claimed to work. Second, it is argued that thepositive effects of adjustment are of medium to long-termcharacter, requiring patience in evaluation of the reforms.Third, and interesting for sub-Saharan Africa, it is pointedout that adjustment lending generally seems to have workedmore effectively in middle-income countries than in low-in-come countries. Summers and Pritchett (1993) actually statethat "adjustment has helped, but has not solved the long-rungrowth problems of most low-income countries, especiallyin sub-Saharan Africa " (italics added). The general criti-cism against structural adjustment thus seems to bear ahigher degree of relevance for sub-Saharan Africa than forthe average World Bank client.

This impression is substantiated by a closer look at one ofthe latest policy research reports from the World Bank(1993). In this report the progress, payoffs, and challengesof adjustment in sub-Saharan Africa is subjected to an em-pirical analysis. The conclusions confirm the seeminglygeneral view taken by the multinational donor society: sub-Saharan Africa is to be considered as a sick child. Accord-ingly, it is concluded that the areas of general macroecon-omics and incentives need more of the same medicine,while the public and financial sectors desperately want achange in medicine. Thus, the fundamental view of how toaddress the macroeconomic problems of sub-Saharan Af-rica remains largely unchanged.

More interesting perhaps, are the comments to the WorldBank (1993) offered by economists at the Research Depart-ment of Statistics Norway (Cappelen et al. [1993]). Theyconclude that the methods of the World Bank do not sufficefor a good explanation of the links between inflow of aid,policy reforms, and macroeconomic performance Further,it is argued that policy reforms may have caused significantpositive changes even in the absence of capital inflow,again highlighting the identification problem involved inevaluating adjustment lending. Finally, and probably moreseriously, it is suggested that some of the World Bank's con-clusions seem to be biased in favour of structural adjust-ment lending. Although one may get the opposite im-pression by reading World Bank documents, there is still noconsensus over the design and impact of structural adjust-ment.

When trying to explain why the response of sub-Saharan Af-rica to structural adjustment is minimal compared to theother adjusters, the World Bank (1993a) draws attention tofeatures that are characteristic for sub-Saharan countries.Among these are institutional weakness, severe governmentfailures, and lack of human capital. If one were to range thecountries of sub-Saharan Africa according to these indica-tors, Mozambique would probably be struggling near thebottom of the list. In conclusion, the structural adjustmentprogrammes seem to have had a slightly positive effect onaverage World Bank clients. This implies that empirical ana-lyses on the matter may draw some supportive conclusionsif all adjusters are included in the data set. However, theseconclusions are easily modified if the scope is restricted tosub-Saharan African adjusters, where the point of departureis worse than in the rest of the third world. To what degreethe argument is further extendable to the situation in Mo-zambique remains to be seen, and hopefully, the next chap-ter will shed some light on this question.

10


2. Monetarism in developing countries

2.1 The fundamental ideasIn short, monetarism is the view held by several neoclassi-cal economists that the quantity of money has a predomi-nant influence on the price level, and that the objectives ofmacroeconomic policy are best achieved simply by target-ing the rate of growth of the money supply.

The following exposition is devoted to illustrating the cen-tral ideas of the monetarist approach to macroeconomic pol-icy, and to stabilization and structural adjustment in develo-ping countries. The outline is general, in the sense that itaims at providing a description of the ruling principles be-hind the construction of structural adjustment programmesfor representative clients of the multinational donor society.However, without considerable loss of generality, sometechnical modifications are made to accomplish the specialfeatures of under-developed countries like Mozambique.These modifications particularly concern the repressed capi-tal and financial markets, features that have implications forthe transmission of monetary impulses. But before we con-centrate on factors that influence on the money stock, thebasic ideas of monetarism require a general introduction.The fundamentals of monetarism may be illustrated by theidentity:

(1) PYMV

whereby the price level (P) multiplied by the gross nationalproduct (Y) must equal the money supply (M) times the vel-ocity ratio (V). Although describing an accounting identity,equation (1) does not offer a proper behavioral explanationof the relationship between the four variables.

The identity of equation (1) may be transformed to relatethe respective rates of change, which must imply:

(2) P+j, r--in+i)

Equation (2) states that the rate of inflation Q.)) plus the rateof GDP growth (y) must equal the monetary growth rate (tit)corrected for changes over time in the velocity of money(i, ). Monetarists now claim that the physical volume of out-put is determined largely by some underlying long-runequilibrium in the markets for commodities and labour, and

that GDP growth is unaffected by monetary factors. Fur-ther, it is claimed that changes over time in the velocity ofmoney are negligible compared to changes in the moneystock, implying for equation (2) that v = 0 . Incorporatingthese modifications, equation (2) reduces to:

(3) 13=1h-4

Thus, for a given level of national income, the rate of infla-tion is given by the growth in the money supply. Thisclaimed long-run relationship between money and pricessubstantiates the view that inflation results primarily fromover-expansion of the money stock, and that inflation canbe checked by proper adjustment of the monetary policy in-struments. Further, equation (3) illustrates how a constantgrowth rate of the money stock also will secure a constantrate of inflation. This describes the core of Milton Fried-man's celebrated declaration that inflation is always andeverywhere a monetary phenomenon.

Every structural adjustment programme contains an elementof macroeconomic stabilization. As we have seen, this ap-proach to macroeconomics is dominated by so-called mone-tarists, who target the expansion of money and credit to cre-ate a stable and low-inflation macroeconomy. The reductionof inflation is thus crucial to the claimed intentions of thestructural adjustment programmes. There are several reason-s why inflation is a threat to developing economies. Theseinclude both traditional effects, and mechanisms that aremore specific to the third world.

First, with sticky wages and fixed exchange rates, the costcompetitiveness is reduced as a direct result of a higher rateof inflation than the rest of the world. This loss of competi-tiveness may be counteracted by a continuous devaluationof the currency to keep the real exchange rate constant.This is exactly what is observed in Mozambique and theother adjusting countries south of Sahara

Second, the inflation rate influences the real interest rates,especially if nominal interest rates are subject to regula-tions. Regulation and segmentation are common features offinancial markets in developing countries, and Mozambiqueis no exception. Low real interest rates will in turn cause

11


low domestic savings and capital flight. Thus, bringingdown inflation may increase real interest rates and improvethe efficiency of the financial markets. However, the ob-served pattern of capital flight can not be explained entirelyby interest rate differentials, a fact that was accentuated forMozambique in the middle 1970's. This illuminates thethird detrimental effect of a high inflation rate: a special re-lationship between inflation and capital flight is claimed fordeveloping countries. The explanation is that domestic andforeign assets may have different risk components. With aconstant real exchange rate, foreign assets represent fairlysafe investments. On the other hand, as the real value of do-mestic assets depend on the inflation rate, a volatile rate ofinflation will make domestic assets a risky investment. Ifthe level of inflation also indicates the volatility of inflation,a higher level of inflation will cause a shift away from do-mestic assets, resulting in capital flight.

Third, the above link between inflation and capital flight im-plies that an increase in the rate of inflation will reduce theavailability of domestic funds for investment. Thus, thereseem to be a special indirect link between inflation and in-vestment in representative developing countries.

These different adverse effects of a high and rising inflationrate provide an explanation for the multinational donor so-ciety's concern with the rate of inflation. The next section il-lustrates the monetarist policy measures in a simple macro-economic framework.

2.2 A monetarist macroeconomic frameworkThe following exposition aims at providing a simple techni-cal framework for analyzing some of the monetary mechan-isms in developing countries. This will serve as a point ofdeparture for closer looks at the different factors that in-fluence the growth of the stock of money. This kind ofmodel framework is very useful to understand the ideasunderlying the structural adjustment approach to develop-ment economics. In our model the private banking system ismerged with the central bank, and the aggregated "moneysupplier" is referred to as the banking system. For Mozam-bique this is not an especially far-fetched assumption, asstate-owned banks control over 95% of total assets(IMF[ 1992b1). With this simplifying assumption, our modelframework contains five different holders of liabilities.These are: the government, private households, privatefirms, the banking system, and a foreign sector. However,the underdevelopment of the capital markets implies quite alot of restrictions on different agents' access to capitalassets, and this will be highlighted by the explicit and im-plicit assumptions made in the exposition below. The cardi-nal point of reference is the national income identity:

(4) C+S+TEC+1+G±(A—B)+F

where C is private consumption, S is domestic saving, taxrevenues are given by T, I represents investments, and G ispublic expenditures. Exports are given by A, imports by B,and F represents net transfers from abroad. Transformed

into a national flow-of-funds identity, equation (4) mustalso imply:

(5) (S-1)+(T—G)EA—B+F

expressing that any domestic imbalances must be offsetthrough the external accounts. The first term on the left sideof equation (5) may be interpreted as net private saving, thesecond term is the surplus of the public sector, whereas theright side represents the current account surplus in the bal-ance of payments. Allowing for international flow of publiccapital only, balance in the current account implies:

(6) A—B+F=e(AR* + A B*)

where e is the exchange rate, R * is the change in foreignreserves in the banking system, and B

* represents changes

in government borrowing abroad, materializing in terms offoreign sale of government bonds. The assumption that pri-vate firms and households are excluded from holdingforeign assets captures the effects of a highly regulatedforeign exchange market. However, this is not critical forthe central results.

As we can see from equation (6), foreign borrowing is thusone of the sources of finance for the government's budgetdeficit. Another possibility is domestic sale of bonds (B),and these are assumed to be bought exclusively by the bank-ing system, excluding both private firms and private house-holds from the government bond market. This yields for thefinancing of the budget deficit:

(7) G — T = e AB* + AB

With restrictions on private international capital flows, asaving surplus in the private sector must imply accumula-tion of the money stock (M) or a reduction in bank loans tothe private sector (L):

(8) S — I = AM-AL

Inserting equations (6), (7), and (8) into equation (5) nowgives for the change in the money stock:

(9) AM = eAR* + AB + AL

Equation (9) establishes a direct link between national in-come identity of equation (4) and the evolution of the mon-etary base, and may be used to highlight the monetary trans-mission mechanisms stressed by the monetarist approach tomacroeconomics. The first term on the right-hand side ofequation (9) represents the effect on the money stock origin-ating in a change in the foreign reserves in the banking sys-tem. The second term shows the impact on the money sup-ply from an increase in the banking system's claims on thepublic sector, and thereby links the budget deficit to themoney supply. The third effect incorporates the impact onthe money supply from a change in credit to the private sec-tor. This macroeconomic framework illustrates the main

12


ideas behind the financial gap analyses offered by the multi-national donor agencies. Financial gaps in the private, pub-lic, and external sectors of the economy cause changes inthe money stock, and these monetary impulses feed directlyonto the inflationary process. That may explain some of themultinational donors' concern for financial gaps.

The first term on the right-hand side of equation (9) linksthe money supply mechanism to the current account. Thisterm is in accordance with the monetarist approach to thebalance of payments, the extension of the traditional mone-tarist approach to international macroeconomics. The mon-etary approach to the balance of payments distinguishes it-self from other approaches by focusing directly on the ef-fects from the money market on the balance of payments,rather than working through the implied changes in thegoods and asset markets. This school of thought is also anadvocate of floating exchange rate regimes. The rationalefor this conclusion is that a floating exchange rate is con-sidered to adjust in order to equilibrate the internationaltrade flows. For equation (9) this implies that e will settle ata value to satisfy:

(10) eAR* = 0

If this is were complete story, any country would be able tocontrol its domestic money stock independently of interna-tional commodity and capital flows by letting its currencyfloat. Thus, the domestic stock of money may still betargeted by policy measures, and this is exactly what linksthe monetary balance of payments approach to traditionalmonetarism.

The second term on the right-hand side of equation (9) illus-trates the link between the public budget deficits and theprocess of money creation. This highlights a commonsource of critique against the governments of developingcountries, who are accused of financing their budget deficitsby an inflation tax, or so-called seigniorage. The reason isthat when the government issues new money, it really takesinterest-free loan in the private sector. In turn, the privatesector will have to bear the costs in terms of higher infla-tion. Similar results may be reached by an examination ofthe required reserve ratio. The banking system is required tohold a certain amount of reserves, and these are determinedby the required reserve ratio as a constant share of total de-posits. The money multiplier is negatively influenced bythis required reserve ratio, which implies that "freezing"money in the banking system suppresses the efficiency ofmonetary policy. This conclusion rests crucially on the as-sumption that required reserves in the banking system alsocorrespond to actual reserves held. However, in developingcountries it has been quite common to use the reserve re-quirement ratio as a policy instrument, to finance the publicbudget deficits (cf. Fry [1988]). This implicitly means thatthe anti-inflationary effects of keeping reserves are under-mined, and therefore it is claimed that for practical purposesthe required reserve ratio may be considered an instrumentof inflation tax in developing countries. This seemingly per-

verted content of the reserve requirement ratio is betterunderstood if we keep in mind the extremely tight connec-tion between the central bank and the government in develo-ping countries, especially in some sub-Saharan countries.

The last term on the right-hand side of equation (9) de-scribes the change in credit to the private sector. Thus, con-trolling the stock of money also implies monitoring the flowof credit to the private sector. Accordingly, the supply ofcredit is usually choked by restrictions like credit ceilingsand directed credit facilities. In Mozambique, credit is nor-mally out of the question for ordinary consumers, andlargely also for small-scale producers. The credit controltherefore seems to contain some adverse effects on the dis-tribution of income. The implications for future policy de-sign are commented in the next section.

2.3 The rationale for financial sector reformsFinancial aspects of economic development have played asignificant role in the theoretical literature of developmenteconomics for decades (see e.g. Gurley and Shaw [1955,1960], McKinnon [1973] and Shaw [1973]). However, dur-ing the 1980's these ideas seem to have established them-selves to an increasing extent also in the multilateral donorsociety's design of structural adjustment programs in Af-rica, and a financial sector reform has recently also been im-plemented in Mozambique (IMF[1992b1). The public sec-tor's reliance on the banking system has, coupled with highand unstable inflation rates, caused negative real interestrates in many African countries. Low real interest rates dis-courage savings, whereas the variability of real interestrates presents a threat to private investment. Lifting the in-terest rate regulations has therefore been an important partof financial sector reforms in these countries, but so far thepositive results have been lacking almost totally. One expla-nation is that the competition of the banking system in Afri-can countries is limited because of the small number of sup-pliers of banking services. This leaves little room for compe-tition and market-determined interest rates, in turn restrict-ing the room for deregulation.

Table 2.1 Some indicators of financial deepening 1990

Country Bank deposits to Public sector Total loansmoney supply (M2) loans to GDP to GDP

Botswana 0.53 0.01 0.10Kenya 0.40 0.06 0.24Malawi 0.74 0.03 0.10Mozambique 0.74 0.23 0.36Tanzania 0.64 0.25 0.28Zambia 0.82 0.11 0.31Zimbabwe 0.69 0.12 0.31

Japan 0.92 0.17 1.35Norway 0.83 0.11 0.79West-Germany 0.89 0.26 1.12

Source: IMF (1990), World Bank (1992b).

13


In Mozambique, the failure to bring down inflation has keptmost real interest rates negative, despite nominal interestrates of over 40% on bank loans. The experience has shownthat interest rate reforms are unlikely to succeed, unless theoverall macroeconomic environment is stable and sound.For Mozambique this seems to suggest that a well-function-ing financial sector is critically dependent on a reduced andstabilized inflation rate and an independent central bank, tosecure a sustainable long-term credit and monetary policy.

Another characteristic of many African countries, includingMozambique, has been the variety of preferential credits,available for a number of sectors and activities. Political in-fluence has made the directed credit programs favour publicprojects, thereby crowding out private sector investments.As the interest rates of the directed credits have adjustedonly with a lag (or not at all), the programs may also belooked upon as a way of disguising subsidies. An ingredientof most of the financial sector reform programs has there-fore been to cut down on these directed credit facilities, thecentral idea being that project selection should be market-based, not administered.

As the financial reform programs started, many Africanbanks were badly managed, and even insolvent, both ele-ments stressed in the financial sector study of Mozambique(IMF[1992b]). An important task has therefore been to re-capitalize the banking system in a broad sense. For the coun-tries involved, this has in part proved overwhelmingly ex-pensive, depending on the terms achieved by the banks re-garding their bad loans. Neither has privatization of publicbanks proved especially successful. The overall experienceseem to be that consolidation and recapitalization of thebanking system has few or no positive effects unless it is ac-companied by close attention to the general policy frame-work.

Another problem of almost all sub-Saharan countries con-cerns the level of skills of bank managers and their em-ployees. There is room for improvement on all levels in thisarea, but in fields like simple accounting and auditing theproblems seem to be especially serious. This is well-knownalso in Mozambique, as a considerable part of the blue-col-lar bank staff lack the most elementary skills of mathe-matics, and many managers know less than a minimumabout accounting and auditing. The designers of the finan-cial reform programs are to blame, as attempts to improvethe financial expertise over the 1980's have been almost ab-sent. Where the training aspect has been an integrated partof the reform, the results are at this stage weak. However,one should bear in mind that educational policies of thistype have a typical long-term profile, suggesting that it stillmay be too early to judge their full impact. A related aspectis the adaptation of an appropriate legal environment, con-sidered by experts as an important part of the financial sec-tor reforms. Nonetheless, reforms of the legal system do notseem to have received the resources they deserve in desig-ning and carrying through the structural adjustment pro-grams.

In connection with the financial sector reforms, some efforthas also been put into the development of domestic moneymarkets. Where the fiscal and budgetary situation is some-what stable, this seems to have been wise. One reason isthat the establishment of a money market enables the cen-tral bank to perform market operations in order to manipu-late the money supply. In addition, interbank markets let thebanks provide each other with liquidity, instead of beingforced to turn to the central bank for every temporary needof funds. More resources can then be devoted by the centralbank to controlling the overall liquidity level. Domesticmoney markets also increase the availability of financial in-struments, and this is likely to raise domestic savings.

Although the time span is short, it may well be claimed thatthe financial sector reforms of the 80's to a large extenthave failed. The policy designers reached ambitiously at aproblem that seemed important, overlooking the complexityand the interlinkages to the rest of the economy. The chal-lenge of the future will therefore be to start from scratchand go somewhat more carefully ahead. Efficient paymentsand credit allocation, safe and sound banking, and financialdiscipline must be enforced well before resources are of-fered to the more idealistic ideas of money markets and ex-tended financial deepening.

The experience with banking reforms suggests that recapita-lizing and restructuring should not be addressed without at-tending also the real side problems of the financial system,so that the return of financial distress can be minimized. Fu-ture plans for restructuring of the banking system should fur-ther emphasize the importance of down-sizing and manage-ment and performance criteria. This will help in identifyingand removing lemons from the banking system, securing animproved environment for suitable investors.

To improve the quality of services, one should also apply astrategy that will increase the competition of the bankingsystem. However, this does not mean that an overall lib-eralization of entry should be introduced. As a result of thefinancial sector reforms, some countries are already over-banked. The country-specific design must therefore dependon the existing aggregate financial structure, taking into con-sideration the fact that liberal bank entry may in turn causefinancial distress. The strategy of removing the directedcredit facilities should be prolonged to secure profitablebank operation and improved monetary control. Directcredit channels are vulnerable to loan defaults and politicalcaptivity, reducing their efficiency in the allocation ofcredit. The development of market-based credit allocationshould therefore be followed also in the future.

Agriculture and small-scale industry have proved to be im-portant engines of growth in Mozambique during the late1980's. The existing banking system is generally not at alldesigned to take care of the needs of these sectors. Conse-quently, the informal markets are left to provide credit to ag-riculture and small-scale industry. Dealing with small-scalecustomers is expensive for the formal bank sector, and these

14


activities are therefore left to the informal markets in theshort term. Instead of trying to regulate the informal mar-kets, in the longer run the governments should provide theformal banking system with conditions that make it possibleto compete with traditional informal credit suppliers, alsowhen it comes to small-scale industry and agriculture. Thiswill imply a more efficient intermediary role of the formalbanking system and at the same time improving the domes-tic payment systems. Still, to the extent that the existence ofinformal markets have adverse effects on the economy assuch, action should be taken to link the formal and the infor-mal credit markets. Then the policy instruments of the mon-etary authorities can be directed towards the informal sectorthrough the formal banking system.

The financial reform programs have failed for a number ofreasons, the most important being that the overall economicenvironment has not adjusted to support the financial devel-opment. This has caused a rapid return of general problemslike mismanagement, lack of financial discipline, financialdistress, and inefficient payment systems and credit alloca-tion. However, in refreshing the development strategies forMozambique, one should not jump to the conclusion thatfinancial reforms should be abandoned, but make sure thatthey are appropriately phased. When planning for the fu-ture, financial reform programs are recommended to take asomewhat more modest approach, starting from basics, al-ways waiting for backing from the real side of the economy.

2.4 Why monetarism might not workThe crux of any criticism against monetarism and structuraladjustment is captured by looking closer into different econ-omists' evaluation of short-term and long-term consider-ations. The monetarist approach is clearly concerned mainlywith long-term macroeconomic stability in well-functioningmarket economies, stressing the stability over time of beha-vioral economic relations (e.g money demand). The implica-tion is that also the macroeconomic framework is one ofcontinuous equilibrium. Policy recommendations are charac-terized by a general laissez-faire attitude towards the role ofthe state. One of the few policy measures that should be ap-plied, according to traditional monetarists, concerns thegrowth of the money supply, and this should be measured toaccommodate a situation of long-run stability with a lowand constant rate of inflation. Thus, short-run instability andmacroeconomic disequilibrium are features that are re-garded as less relevant to any description of the economy.

On the other hand, the neo-Keynesian is more of a disequili-brium approach, focusing less on typical long-run problems.Among the factors that are stressed are market-imperfec-tios, capacity utilization, and unemployment. The policy im-plications rest on an explicit care for short-run phenomena,and accordingly, neo-Keynesians recommend manipulationof aggregate demand, to prevent production and employ-ment from fluctuating too dramatically in the short run.

According to the above distinction, it may be claimed thatthe two approaches are not developed to handle the same

macroeconomic problems, and therefore, that they shouldnot be subject to direct comparisons. Further, it is temptingto draw the conclusion that monetarism and structural ad-justment do not give proper answers because they do notformulate the appropriate questions. This conclusion restson the assumption that the short-term problems are more im-portant for developing countries than their long-run prob-lems, and now we really approach the core of the ongoingdebate. The challenge is therefore to compose an economicand political strategy that represents an optimal mix ofshort-run and long-run instruments.

This problem may well be illustrated by the case of agricul-tural production in Mozambique. Devaluations, decreasinginflation and transition to market-based food production aremeasures supposed to increase exports and purchasingpower of the rural poor over the medium term. In the mean-time they are suffering because of rapidly rising domesticprices, credit squeeze, and the riskiness of selling to a mar-ket. Together with a seriously underdeveloped infrastruc-ture this might provide an explanation of the lack of supplyresponse in agricultural production.

In addition to the basic conflicting views regarding short-run and long-run perspectives, there are significant diver-gences regarding the impact of the different policymeasures. As noted above, monetarists tend to stress that de-veloping economies are to be treated as any other economy,and that there are stable behavioral foundations with globalvalidity. This view is opposed by the new structuralistmacro critique. This school of thought underscores the spe-cial features of the economy in developing countries, leav-ing room for quite extraordinary results regarding the tradi-tional instruments of economic policy.

As we saw in the previous section, a distinct feature ofevery financial sector reform programme for developingcountries is the lifting of interest rate ceilings to liberate thefinancial markets. These efforts to make the financial mar-kets more efficient may have real effects that are unique fordeveloping countries, and these effects are largely neglectedin the multinational donors' structural adjustment pro-grammes. Below, we will look more closely into some ofthe reasons why interest rate liberalization may have con-tractionary and inflationary effects.

First, there are special features of the supply side of develo-ping economies that may cause perversion of standard re-sults concerning tight monetary policies and increased inter-est rates. Credit financing of working capital needs is cru-cial supporting a production process with extreme lack ofliquidity in developing countries, especially in manufactur-ing industries. Tight monetary policies and increased real in-terest rates will increase the firms' costs of short termcredit. This will push the marginal costs upwards, and forgiven prices, production will decrease. On impact one mayexperience that tight monetary policies may cause inflation-ary pressures via the supply side of the economy. These areperverted effects of monetary policies in developing coun-

15


tries that are stressed in a number of neo-structuralist works(e.g. Bruno [1979], Taylor [1981]).

Second, to the extent that consumers are active in the creditmarkets, one may experience contractionary effects of tightmonetary policies also via the demand side of the economy.The reason is that a contraction of the monetary base will in-crease the costs of finance also for households, squeezingout consumption expenditures and aggregate demand. In ad-dition, households must be assumed to place all their saving-s, whatever these might amount to, in unofficial credit mar-kets, where the rate of return exceeds the deposit rate of-fered by the banking system. One of these is typically theblack market for foreign exchange. An increased official in-terest rate is then likely to cause a reduced black-market pre-mium, and those who have placed their savings in foreignexchange are therefore likely to suffer from a negative shiftin real wealth. On an aggregate level this third-world spe-cific effect may reinforce the contractionary effects of tightmonetary policies via the demand side of the economy (seeMontiel [1991]).

Third, a narrow connection between the commercial bank-ing system, the central bank, and the government opens forquite peculiar macroeconomic effects concerning interestrates, financial liberalization and inflation. An importantsource of government finance is credit extended by the cen-tral bank, especially in socialist countries. With Tanzania asa point of reference, Collier and Gunning (1991) illustratehow the process of money creation is influenced by the inti-macy between the government and the banking system in asocialist developing country. Their analysis may have relev-ance also for Mozambique. Collier and Gunning (1991)have the banking system's credit to the government as apoint of reference. The banking system's assets are totallydominated by credit to the government, or to public enter-prises. Financial sector reforms and interest rate liberaliza-tion in such an environment result in an escalation of inter-est payments from the government, again making the bud-get deficit increase. As far as money creation is influencedby the budget deficits, the rate of inflation may thus also bethreatened by financial sector reforms.

Fourth, devaluations have a special effect in developingcountries, reflecting the producers' critical dependence onimported inputs. We touched upon the same idea in chapter2, when commenting on the recent history of economic pol-icy in Mozambique. It is widely acknowledged that im-ported inputs play an important role in the production pro-cess of most developing countries. Table 3.1 reveals thatMozambique is no exception in this respect. Since the im-plementation of ERP, the access to foreign currency and im-ports have increased, and this development is in generaltracked surprisingly precisely by the growth in GDP. How-ever, the continuous devaluations of the currency have grad-ually squeezed the access to foreign funds, and the effecthas been to throttle the import of inputs, again causing GDPgrowth to stagnate. In the long run Mozambique will haveto improve the international competitiveness to be able to

pay for her imports. Nonetheless, the urban and rural poorsuffer increasingly because of the emphasis on long-run pol-icy. More generally, if the real wage rate is already at sub-sistence minimum, devaluations may cause hunger, unlessthe income reduction is accompanied by continuous inflowof funds. This illuminates the dilemma of choosing the rightmix of short and long-run economic and political strategies.

That contractions of the money supply may have significantreal effects is close to an established fact in standard macro-economic analysis. Many industrialized countries take theconsequence as they measure short-run strategies to counter-act cyclical changes in employment and aggregate demand.However, in mapping out the strategies for developing coun-tries, short run considerations seem somewhat neglected.Comparing the urgency of the social and economic prob-lems in industrialized and developing countries, this can beconsidered as a striking observation. This view also findssupport in the evaluation of the Norwegian developmentstrategy concerning Mozambique made by Brochmann andOfstad (1990). As argued earlier, Mozambique will be acountry in severe economic and social crisis in many yearsto come. Accordingly, economic and political measuresshould be adopted to secure a development strategy with ahuman face, whereas the adjustment to a long-run sustain-able situation should be undertaken with extreme care.These considerations are highly important, and should befollowed even if they imply deviations from the plans madeby the multinational donor society.

16


3. Structural adjustment inMozambique

3.1 Post-independence economic policyThe macroeconomic setting of the early post-independenceperiod was characterized by FRELIMO's attempts to intro-duce a centrally planned economic environment. As the dra-matic exodus of 1974 had drained the country of bothhuman and physical capital, the prospects of fast recoverywere poor. During the last half of the 1970's, enterprisesleft by the Portuguese were taken over by the state, initiallyof practical as well as political reasons. As the confidencein leadership and ideology increased, the approach to econ-omic policy was heavily influenced by FRELIMO's endeav-our to introduce a socialist system of rapid development andequality. One may suspect the transition of the late 1970'sto have been as spectacular as today's structural adjustmentplans, and according to Bowen (1992), the post-inde-pendence economic and political strategy was completely indisharmony with the existing socio-economic conditions.

FRELIMO gradually acknowledged that the economy wasin need of some sort of restructuring, and in 1980 the firstten-year plan of economic policy was set out. A rigid sys-tem of central planning was introduced, and to accomplish"victory over underdevelopment" the gross national productwas meant to rise by 17% a year. Emphasis was placed ongiant development projects like state farms, heavy industry,and energy production. Other dominant features of the ten-year plan were the organization of the peasantry into com-munal villages, and the collectivization of agriculture to in-crease production. By the fourth FRELIMO congress of1983 it had become clear that the ten-year plan was afailure. The huge public projects were growing out of con-trol, and were losing enormous amounts of money. Themanagerial capacity was lacking, both for the state farmsand for industrial projects.

The ten-year plan was revised, and emphasis was now puton existing capacity. All new projects were of the small-scale type, and a division of the state farms into more practi-cal units was initiated. At the same time, the government re-sponded to the decline in production by credit expansion.However, these measures were not sufficient to stop the de-cline in all sectors of production. The expansionary mon-etary policy opened perilous financial gaps over the period1981-1986: domestic credit increased by 150% while the

country's net financial assets practically dried out. In addi-tion came the South African destabilization campaigns andthe civil war, both aspects that seriously throttled the flowof most goods and services. Finally, as a result of extensivedomestic price-distortions, the private producers lacked thenecessary incentives for production to rise.

3.2 The Economic Recovery ProgramTo halt the economic contraction, and to gain access tomultilateral donor funds, Mozambique joined the IMF andthe World Bank in 1984. Thus, structural adjustment lend-ing came to reach Mozambique, and by 1987 the inflow ofaid and credit was conditioned on the implementation of aneconomic recovery program (ERP) designed by the Bretton-Woods institutions. The ambition of the ERP was threefold.First, the level of production should rise, to establish a mini-mum level of per capita consumption and income, with spe-cial emphasis on the rural poor. Second, the public and ex-ternal deficits were to be closed. Third, the macroeconomicconditions should be prepared to support a faster pace ofmedium and long-term economic growth. From now on Mo-zambique followed a standard structural adjustment plan, in-corporating characteristic features such as augmentation ofmarket forces, improvement of private capital access, down-scaling of the public sector, contraction of domestic credit,and dramatic devaluations. After the implementation of theeconomic recovery plan, IMF (1992, 19931) and the WorldBank (1992b) summarize the development in macroecon-omic indicators as shown in table 3.1.

The ERP seems to have caused a certain immediate upturnin production and domestic absorption, but this growth le-velled out already by the beginning of the 1990's. A varietyof explanations have been put forward to explain this stag-nation. These may be sorted into groups: external shocks,the costs of peace, and failure of economic policy. Externalshocks include factors such as political changes in Europe,changes in the inflow and composition of aid, the oil crisisof 1990, an erratic domestic insecurity situation, and va-rying climatic conditions. Export markets and sources of ex-ternal finance collapsed after the breakdown of the formerpolitical regimes in Eastern-Europe.

17

Table 3.1 Key macroeconomic indicators for Mozambique 1986-1993


1986 1987 1988 1989 1990 1991 1992 1993 1

Real growth ratesGDP 2.3 4.6 5.5 5.4 1.3 2.6 - 2.3 5.6GDP per capita -0.8 2.0 2.8 2.7 - 1.3 0.1 -3.9Gross investment 51.4 15.6 16.5 7.3 8.5 2.5 -4.2 21.0Ratios to GDPDebt service .. 39.8 37.3 35.3. 40.3 48.3 67.9Gross investment 9.7 23.8 33.4 35.5 38.1 40.8 43.5 50.2Domestic saving - 1.1 -10.5 - 16.4 - 16.6 - 11.7 -9.7 - 15.1 -22.6Budget deficitBefore grants .. .. 25.3 25.5 29.3 26.0 29.8 30.8After grants 9.5 20.3 11.3 8.91 2.4 5.8 6.2 - 3.4Exports 3.6 11.9 15.1 15.4 15.9 22.6 26.7 29.7Imports 14.4 46.1 65.0 67.5 65.7 73.0 85.3 102.4Current accountBefore grants -15.1 - 45.7 - 52.9 - 58.8 - 54.7 - 53.9 - 65.0 - 103.0After grants - 22.6 -28.9 - 12.6 - 17.3 -21.0 -41.7

Real exch. rate (1987 = 100) 438.5 100 63.3 63.4 63.1 52.6 42.1 41.0Annual rate of inflation 32.5 163 50.1 42.1 49.2 33.2 45.1 39.6

' Estimates.I) Given the current estimate of population growth at 2.6% per annum.Source: IMF (1992, 1993a), World Bank (1992b).

Mozambique used to have a good relationship with the for-mer eastern block in Europe, especially with the formerGDR, and this offered Mozambicans opportunities in theform of export markets, guest working, and access to capitaland certain technological expertise. It is a widely held viewthat these factors are important in explaining the stagnationin economic performance around 1990. At the turn of thedecade, some 14,000 guest workers returned from East Ger-many, as a direct result of the reconciliation between thetwo German states. This caused the elimination of an im-portant source of labour training and foreign exchange. Indi-rectly, and perhaps as important, this repatriation also seemsto have increased the already existing unrest in the Mozam-bican labour force, as the degree of labour militancy hasshown a marked increase over the last few years.

When trying to explain the setback of the early 1990's, theWorld Bank (1993b) emphasizes the role of the level andcomposition of the inflow of aid. From 1990 we have seen asignificant reduction in import support and programme aid,whereas food and emergency aid has risen moderately,mostly due to the drought of 1992. This implies a change inthe distribution of aid away from typical growth-promotingfunds towards short-term emergency aid. The World Bankalso emphasizes the fact that the total inflow of aid has de-clined since 1990, especially as a result of the politicalevents in Eastern Europe and the Soviet Union. Combinedwith the oil crisis of 1990, these external shocks may ex-plain a part of the stagnation that occurred in 1990.

Climatic conditions vary considerably from year to year, anelement that has a particularly influential bearing on agricul-tural production. With the exception of 1992, the climaticconditions were not particularly worse in the early 1990's

than over the 1980's. In spite of this, the annual growth rateof agricultural production fell from 4 per cent in 1989 to 1.1and 1.9 per cent in 1990 and 1991, respectively (cf. table3.1). Going further back than 1989, the stagnation in agricul-tural production becomes even more aggravating, as agricul-tural production increased by more than 7 per cent a year,both in 1987 and 1988. Thus, the implementation of theERP had at best some temporary effects on the agriculturalproduction, leaving long-term issues largely unsolved.

By the costs of peace, we refer to the parts of the peace pro-cess that may have negative influences on the potential foreconomic development and growth, especially in the shortterm. The multinational donors are also increasingly con-cerned with what they refer to as the increased public expen-ditures in connection with the demobilization. The newarmy is planned to total 30,000 soldiers. This implies thedemobilization of some 100,000 troops. At an expense ofUS$ 700 per soldier this will cost the Mozambican govern-ment around US$ 70 million. In addition, there are consider-able cost elements connected to the process of resettlement.The repatriation of refugees from neighbouring countriescauses the population to increase, again pressing per capitagrowth rates downward. In addition, a great many of the for-mer refugees are unemployed, homeless, and extremelypoor. In the short run, they will therefore increase public ex-penditures for social services and immediate economic as-sistance. All these relatively acute aspects of the peace pro-cess have undoubtedly increased the expenditure side of thepublic budgets in the short term, whereas the observableeconomic benefits of peace are typically longer-term ef-fects. As the peace was not settled until late 1992, the se-curity situation was still unstable both in 1990 and 1991. Aresult was that industrial production was harmed by the de-

18

Table 3.2 The evolution of;production by sector 1986-1992


1986 1987 1988 1989 1990 1991 1992*)

Agriculture - 0.6 7.0 72 4.0 1.1 1.9 - 8.0Industry - 18.6 - 4.3 8.9 6.8 - 8.3 - 0.5 0.0Construction - 4.4 44.8 16.0 3.0 1.5 3.0 2.0Transport - 11.8 1.7 - 9.8 10.2 19.8 13.2 4.0Commerce - 6.8 - 0.7 3.5 4.0 2.5 2.0 2.0GDP 2.3 4.6 5.5 5.4 1.3 2.6 - 2.3

*) Estimates.Annual growth rates in per cent. Source: IMF (1992), the World Bank(1992b).

struction of electricity supply in 1990. Thus, the stagnationof economic development and growth may also have beencaused by factors directly related to the insecurity of thecivil war, and to the establishment of peace.

There are also economists who will have it that parts of theERP itself may have caused the stagnation of the early1990's. Critics of the structural adjustment plans for develo-ping countries often point to industries' critical dependenceupon imported inputs. If foreign currency is a scarce re-source under these conditions, a bottleneck in the supply-side of the economy will evolve. If this kind of supply-con-straint is binding, and the availability of foreign currencyand imports is increased, one would expect an immediateupturn in production. This is almost exactly what can be ob-served in Mozambique directly after the implementation ofthe ERP. However, another element of the ERP was thegradual liberalization of the market for foreign currency.This implied that the metical was devalued continuouslyover the late 1980's, making the domestic price of importsrise. With the described link between imports and outputsupply, the bottle-neck may actually have narrowed againaround the turn of the decade. This special feature of pro-duction technology may thus provide a partial explanationof the stagnation following the implementation of the ERPwith a three-year lag.

Another aspect of the ERP itself concerns the monetary andcredit policy. Pivotal to almost every structuralist adjust-ment programme is the ambition to reduce inflation. Avail-able monetary instruments are applied to reduce the growthof the money and credit supply, as the links between moneysupply and inflation are essential according to the theoreti-cal origins of the structural adjustment programmes. To be-lieve in this prescription, one really has to "swallow acamel", and that is that the implemented monetary policyhas no or few negative real effects. To the extent that Mo-zambique is producing at a natural price-inelastic full-em-ployment level, tight money and credit policy may leaveGDP unchanged, and inflation may be brought down. How-ever, assuming that a contraction of the credit volume willleave production levels unchanged is quite absurd, espe-cially for a country like Mozambique.

Accompanied by increasing real interest rates, the tightmonetary policy of ERP is likely to have had significant ne-gative effect on the evolution of production levels, at leastin the short run. Not only has the credit-accessibility beenreduced, but the price of the available credit has also esca-lated. One may dispute the relevancy of this argument, asthe real interest rates are still very low, not to say negative.However, the rise in financial costs is nonetheless real, bothfor consumers and producers. As short-term credit makesup an important part of the variable costs of production intypical developing countries, the credit squeeze has prob-ably caused contractionary effects also via the supply-sideof the economy. Even more important, these supply side ef-fects may have counteracted the overall anti-inflationary ob-jective of the structural adjustment policy.

3.3 Looking aheadTo comment on the prospects for post-war development, itmay be useful to draw a distinction between the short termand the medium to long term. Main immediate challengesare embodied in keywords as demobilization, resettlement,and emergency aid. On the other hand, the design of thecontinued policy reform is of utmost importance for themiddle to longer-term growth and development possi-bilities. The level of welfare and development is quite extra-ordinary. Initially colonized by one of the backward nationsof Europe, the struggle for independence resulted in an ex-haustive exodus in 1974. Misguided central planning andcivil war further worsened the overall economic situationup to the middle of the 1980's. Although a certain turn-around can be traced for the first years following the im-plementation of the ERP, it is a highly relevant questionwhether the inflow of aid in connection with the ERP hasbeen as effective as might be expected. Thus, the long runissues of economic development and improved welfare stillremain completely unsolved.

Mozambique is today one of the most aid-dependent coun-tries of the world, and per capita income qualifies the popu-lation among the world's absolute poorest. Among theclients of the World Bank, Mozambique relates to the restof Southern Africa as this region relates to the averageWorld Bank client. Therefore, it is striking how little thisspecial situation of Mozambique seem to have influencedon the design of policy prescriptions. Rather, a strategy ofswift stabilization, liberalization, deregulation, and privatiz-ation seem to have been imposed on a country lacking mostbasic prerequisites for economic progress. This has left Mo-zambique as one of the most debt-ridden and aid-dependentcountries of the world, whereas the benefits seem at leastquestionable.

Food support and emergency aid will have to dominate theuse of capital inflow in Mozambique in many years tocome. The rural poor have no buffer whatsoever, by whichthey could meet the climatic variations, and thus they are ex-tremely vulnerable. In the short run, extensive attention willalso have to be given to the problems concerning demobili-zation and resettlement. This is a critical factor in creating

19


the premises for a supportive environment in the medium tolonger term, not only for economic development, but alsofor political stability. These efforts will take up a significantpart of public expenses over the next years, whereas theeconomic benefits in terms of increased production andeconomic growth must be awaited with patience.

Regarding the economic policy design, Mozambique is nowin many respects floating down a river of no return. Mostparts of the process that has been launched are irreversible.This implies that the policy measures must be concentratedon adjusting the speed and exact direction of the initiatedmovement, as the main lines are largely predetermined bythe ongoing economic reforms. In addition, the sequencingof the different parts of the programmes can still be control-led, and this instrument should be applied to foster elementsregarded as influential, and to bridle any parts that are to beconsidered as ineffective or out of time/place. The overallobjectives of the new National Reconstruction Plan (NRP)are the reduction of poverty, enhancement of growth and ad-ministrative training. Except for stressing the importance ofinstitution-building, management training and human capi-tal formation, nothing much has changed since the im-plementation of the ERP. One exception is the long-term ex-ternal conditions, which have improved significantly afterthe civil war was brought to an end. The new security situ-ation and the revised objectives of policy reform are promi-sing to the extent that they may allow a larger share of thetotal inflow of grants and loans to benefit small-scale indus-try and farming in rural areas. This would mean a signifi-cant improvement, as the inflow of aid over the last yearslargely has been concentrated to urban and/or typically se-cure areas.

Today's Mozambique is characterized by a highly shatteredeconomic environment. In addition, the country's situationwill be one of deep economic crisis for years to come. Bothhuman and physical capital is desperately needed to gener-ate any increase in welfare levels. The most serious mistakeanalysts may make in designing the economic policy forMozambique is to believe uncritically that this economywill react to macroeconomic policy instruments accordingto the same rules that seem to apply for the countries of theindustrialized part of the world.

20


4. Ten countries in sub - Saharan Africa

4.1 IntroductionIn order to construct indicators for social and economic per-formance and development, the World Bank's WorldTables 1992 have been consulted to establish a panel dataset for ten countries in sub-Saharan Africa. Many consider-ations influenced the selection of countries. First, to the ex-tent that the data availability allowed it, the main cooper-ators of the Norwegian Agency for Development Cooper-ation (NORAD) were given preference. Unfortunately,Namibia have to be excluded due to lack of data, leaving uswith five countries of the above described category. Sec-ond, five more countries were included to increase the totalnumber of observations for the planned econometricstudies. These countries are located in the same regions asthe main cooperators of NORAD, to some degree sharingsocial, political, and economic characteristics with the restof the panel. The countries included in the panel are (abbre-viations in brackets):

Botswana (BWA)Ethiopia (ETH)Malawi (MWI)Rwanda (RWA)Zambia (ZMB)

Burundi (BDI)Kenya (KEN)Mozambique (MOZ)Tanzania (TZA)Zimbabwe (ZWE)

The country-specific time series range over the period 1980to 1991 (12 years), giving a total of 120 observations. InWorld Tables 1992, the information on Mozambique is veryscarce, and therefore supplementary sources have been usedto construct time series for Mozambique that correspond tothose of the other countries. These sources include TheWorld Bank (1992b) and IMF (1992). For some of the othercountries there were also "holes" in the World Tables, andthese are repaired by using alternative data sources (mainlyIMF [1993]), and in a few cases by extrapolation. Still, wellabove 95% of the records of the panel are the untouchedfigures from the data base produced by the InternationalBank for Reconstruction and Development.

4.2 Regional differences in population andproduction

The countries in our data set are very different, and the aimof the following exposition will be to shed some light on the

most important variations. Table 4.1 reproduces some demo-graphic characteristics of the countries involved. The totalfertility rate is the total number of child births per womanthrough the fertile ages, and indicates the growth potentialof the population. For Mozambique, the fertility is close tothe average for the ten countries of the table. Botswana andZimbabwe are countries with comparatively low rates oftotal fertility, whereas Malawi and Rwanda stand out ascountries where the average number of children per womanis relatively high.

The infant mortality rates measure the number of childrenper thousand who die every year before reaching one yearof age. For this indicator, the simple average for the tencountries is 99, and thus Mozambique lies well above mostof the other countries with respect to infant mortality. Fur-ther, there are only two countries with a lower life expec-tancy than Mozambique. Although crude, these twomeasures indicate that social conditions are relatively poorin Mozambique compared to the rest of the sample. The rateof urbanization is not outstanding in any direction in Mo-zambique, but as the civil war has made people flee to thecities, the rural population must be expected to increase con-siderably more than the urban population in the years tocome. When measured by the size of the total population,

TotalFertilityRate

InfantMortalityRate(deaths perthousand)

LifeExpec-tancyat Birth(years)

UrbanPopu-lation,% oftotal

TotalPopu-lation

Botswana 4.5 36.4 67.6 29.0 1.29Burundi 6.8 106.0 46.7 5.7 5.60Ethiopia 7.5 129.8 48.3 13.3 52.89Kenya 6.5 65.7 59.0 24.4 25.02Malawi 7.6 149.2 46.2 12.1 8.80Mozambique 6.5 136.3 46.8 28.3 16.14Rwanda 8.3 119.8 47.9 8.0 7.40Tanzania 6.6 115.3 47.3 34.3 25.27Zambia 6.7 82.8 48.7 50.8 8.37Zimbabwe 4.8 47.6 60.4 28.3 10.08

21

Figure 4.1 Per capita GNP . in 1991 (US$)

Botswana

Burundi

Ethiopia

Kenya

Malawi

Mozambique

Rwanda

Tanzania

Zambia

Zimbabwe

0 500 1 000 1 500 2 000 2 500 3 000

Figure 4.2 Per capita net official transfers In 1991 (US$)

1


Mozambique is almost exactly average size in the con-structed panel. Observe also the dramatic variation betweencountries with respect to population size. Botswana is smallwith its 1.3 million inhabitants, whereas Ethiopia must feeda population 40 times larger than that of Botswana. Thisvariation should be kept in mind when figure 4.1 is exam-ined.

Figure 4.1 illustrates how per capita gross national productvaried over the ten countries in 1991. Gross national pro-duct includes gross domestic product at market prices plusfactor income from abroad. Botswana seems to be ex-tremely rich compared to the rest of the sample. This is alsotrue, and with vast mineral resources and large foreignreserves, Botswana is receiving reduced attention from themultilateral donor community.

On the other hand, Mozambique is on the bottom of thescale with an estimated US$ 70 per capita GNP. Accord-ingly, one should think that Mozambique would have acorrespondingly high priority with the donors.

0 50 100 150 200

250

But Mozambique is not the only country characterized bylow per capita income figures. Other extremely poor coun-tries include Burundi (US$ 210), Ethiopia (US$ 120), Ma-lawi (US$ 230), and Tanzania (US$ 100). Except from Bot-swana, the most well-off country seem to be Zimbabwe,with US$ 610 in per capita GAP.

Figure 4.2 illustrates the variation in per capita net officialtransfers, as reported by the World Bank in World Tables1991. Net official transfers comprise net transfer paymentbetween governments of the reporting country and the restof the world, and thus they should pick up quite a bit ofwhat is referred to as development aid.

Again, the most striking feature concerns Botswana. Itseems that Botswana has received transfers from govern-ments of the rest of the world amounting to more than US$230 per capita in 1991. The low population of Botswanatend to inflate the inflow of development aid, whenmeasured in per capita terms. The country also has severaladvantages that make her especially popular among the do-nors. During the 1980's, Botswana was located on the fron-tier to apartheid, and gained considerable support for hold-ing South Africa at a distance. In addition, Botswana has atradition of peace and democracy, both extremely importantprerequisites for development cooperation and for econ-omic growth. The mineral resources of Botswana have alsocontributed to the rapid formation of substantial liquid na-tional wealth. Among the other countries of interest we ob-serve that together with Burundi, Mozambique receivedrelatively much in the form of net official transfers com-pared to other countries in the region. Ethiopia, Kenya, andMalawi are localized near the bottom of the scale, partlydue to political controversies in recent years. However,there is reason to believe that this situation can be improvedif the political unrest is settled.

A dramatic inflow of development aid may also have cre-ated unhealthy structural mechanisms in the economy.These effects are described in a theoretical context by vanWijnbergen (1984), where it is argued that continuous in-flow of capital in the form of development aid may have ef-fects similar to those of petroleum income, causing the so-called Dutch disease. The idea of the Dutch disease is that aconstant (costless) inflow of foreign currency may have ad-verse effects, as they often cause resources to be directedfrom production of tradeables towards production of untra-deable goods. The resulting increased competition for do-mestic inputs make production costs increase in both sec-tors, and on impact the terms of trade deteriorate. Overtime, the downscaling of the competitive industry may alsohave additional negative effects on the competitiveness, ashuman capital formation usually depends on the generallevel of activity (learning-by-doing effects).

Figure 4.3 summarizes the economic growth performancefor the ten countries over the 1980's. And again the winneris Botswana. With an annual growth rate of 10% in bothhalves of last decade, Botswana has enjoyed an enviable

Botswana

Burundi

Ethiopia

Kenya

Malawi

Mozambique

Rwanda

Tanzania

Zambia

Zimbabwe

22

BotswanaBurundiEthiopia

KenyaMalawi

MozambiqueRwanda

TanzaniaZambia

Zimbabwe

0 10

20

30

40

50

❑ 1980-1985 M1986 - 1991


KenyaMalawi

MozambiqueRwanda

TanzaniaZambia

Zimbabwe

El 1 9 8 0 - 1 9 8 5 ■1986-1991

Structural Adjustment in MozambiqueRapporter 94/11

Figure 4.3 Average annual GOP growth (%)1980-1991 Figure 4.4 Average annual monetary growth (%)1980.11991

rate of economic development. Burundi has also had stableand satisfactory growth rates over all twelve years, but thelevel at which this country still struggle still makes it trans-fer-deserving (cf. fig 4.2). The most dramatic turnaroundseems to have taken place in Mozambique, where an aver-age growth rate of —5% during the period 1980-1985 hasbeen transformed to a positive average annual growth ofabout 4% over the last half of the 1980's. As we saw in thelast chapter, this development is about to stagnate, but theprospects for post-war Mozambique should still be promi-sing. Ethiopia, Kenya, Malawi, and Tanzania are other coun-tries where the economic development seems to have beenaccelerating over the last decade.

Rwanda reveals a somewhat bleak pattern in figure 4.3, asthe turnaround for this country has the wrong sign. From asteady annual growth of almost 3%, the GDP growth hasstagnated, and over the last half of the 1980's domestic pro-duction has fallen by more than 2% a year. For Zimbabwe,the situation is comparable to that of Rwanda, as the GDPgrowth rate has fallen from 4% to 1% over the period. Zam-bia has been a typical on/off adjustor, and the absence of asignificant turnaround in production must be analyzed withthis feature in mind.

4.3 Monetary growth and inflationAs pointed out in chapter 2, monetarists tend to stress thelinks between the money supply mechanisms and inflation.This section reports the actual figures on the subject for oursub-Saharan panel. Figure 4.4 illustrates the annual growthin money supply over the period. The applied definition ofthe money supply is broad, including all monetary andquasi-monetary liabilities of the country's financial institu-tions to residents other than the central government. Inspite of the efforts made by the International MonetaryFund and the World Bank, only two of the countries of oursample have actually reduced the growth in the money sup-ply over the 1980's. These countries are Burundi andRwanda. All the other countries report a higher growth in

the money supply over the last six years than over the firstsix-year period.

Ethiopia, Kenya, and Malawi seem to have had a verymoderate acceleration in the growth of the money supply.For all the other countries the growth in the money supplyhas increased significantly over the 1980's. Mozambiquestands out with the highest monetary growth rate for thefirst half of the decade, and the second highest rate ofgrowth during the last period. Only Zambia has experienceda more dramatic acceleration in the monetary growth ratethan Mozambique. Along with Tanzania, Zambia, and Zim-babwe, the monetary growth has also been extraordinary inBotswana. However, for Botswana an accompanyinggrowth in production seems to have suppressed the inflation-ary impulses of the money supply. Thus the growth in themoney supply poses a threat to inflation primarily if the le-vels of production have stagnated, as confirmed by figure4.5.

Figure 4.5 reports the average growth in the consumer priceindex, or in other words, the average annual rate of inflationfor our panel of sub-Saharan countries. In figure 4.5, six outof ten countries seem to have reduced their annual rates ofinflation over the 1980's. This is somewhat in contrast withfigure 4.4, where only two out of ten countries have re-duced their annual monetary growth over the same period.Tanzania is an example, revealing a dramatic increase in themonetary growth rate in figure 4.4, whereas the rate of infla-tion seems to be reduced over the decade.

Not surprisingly, however, most of the countries charac-terized by high rates of inflation are also countries withhigh rates of monetary growth and/or low growth in produc-tion. Thus, Mozambique is one of the countries with veryhigh inflation, especially over the last sub-period. However,the immediate impression is that the monetary growth andthe growth of production do not suffice to explain the sharpincrease in inflation rates in Mozambique towards 1990.Annual average rates of inflation have risen from 20% to

23

Botswana easamBurundi magi

EthiopiaKenya

MalawiMozambique

RwandaTanzania Buszasson

Zambia Zimbabwe minomi

0 20 40 60

80

01980-1985 111986- 1991

isevai-J1111117717=51

100

Figure 4.5 Average annual inflation (%) 1980-1991


KenyaMalawi

MozambiqueRwanda

TanzaniaZambia

Zimbabwe

EIBEINI1

0 20

40 60

80 100

U1980-1985 M1986 -1991


Figure 4.6 Average annual nominal lending rates 1980-1991


KenyaMalawi

MozambiqueRwanda

TanzaniaZambia

Zimbabwe

d.Figure 4.7 Average annual real lending rates 1980-1991

E1980 - 1985 M1986 -1991

24

63%, monetary growth by half this distance, whereas thegrowth in production has increased sharply. Thus, the rapidinflation in Mozambique over the last part of the 1980'sseems to need supplementary explanations in addition to thegrowth rates of money supply and production. Zambia is an-other country where inflation has increased dramaticallyover the 1980's. This corresponds well with the messagefrom figure 4.4, where the highest monetary growth rateover the last sub-period is reported for Zambia. Somewhatmore striking is the inflation performance in Zimbabwe. Inspite of an increasing monetary growth rate and stagnatingGDP growth, Zimbabwe reports a decreasing rate of infla-tion.

Figures 4.4 and 4.5 illustrate some important problemsabout macroeconomic analysis for developing countries.Combined with the general growth performance, the figuresshed some light on the macroeconomic relationship be-tween the growth in production, the growth in the moneysupply, and inflation. On the other hand, the tendencies areweak and unstable, implying that firm conclusions under nocircumstance can be drawn.

4.4 Interest ratesChapter 3 gave a brief comment on the rationale for finan-cial sector reforms in developing economies. These are de-signed by the multilateral donors in order to improve thequality and efficiency of the financial system. As we haveseen, the lifting of interest rate regulations is an importantelement of these financial sector reforms. Figure 4.6 reportsaverage nominal lending rates for our ten sub-Saharan coun-tries. For half of the countries in the sample, there has beena sharp increase in nominal interest rates over the 1980's.These are Kenya, Malawi, Mozambique, Tanzania, andZambia. Two countries (Burundi and Rwanda) reveal insig-nificant variation over the ten-year period, whereas threecountries (Botswana, Ethiopia, and Zimbabwe) have re-duced their nominal lending rates. However, as most econ-omic agents respond to real interest rates, figure 4.6 islimited to illustrating the actual action taken to deregulatenominal interest rates.

The impact of deregulation and the macroeconomic policieson real interest rates are examined in figure 4.7. The real in-terest rates are calculated by subtracting the rate of inflationfrom the nominal interest rates illustrated in figure 4.6, andthese measures provide a measure of the real cost of credit,as inflation over time will reduce the value of a givenamount of credit. According to figure 4.7, only two coun-tries have seen an increase in the cost of credit over thewhole period. These are Kenya and Tanzania. The real costof credit seems to have fallen for the rest of the countriesfrom the first to the second sub-period.

The point of establishing positive real interest rates is oftenunderlined in the financial sector reform programmes. Fig-ure 4.7 illustrates that seven countries seem to have hadpositive real interest rates over the first half of the decade,whereas this number is reduced to four over the second sub-

:Figure 48 Tents of trade adjustriteht 19804991.„

Average terms of trade index for 1980-1985and 1986-1991 (1987=100)

❑ 1980-1985 M1986 - 1991


KenyaMalawi

MozambiqueRwanda

TanzaniaZambia

Zimbabwe

0 20 40

60 80 100 120 140 160

Annual average 1980-1985 and 1986-1991


KenyaMalawi

MozambiqueRwanda

TanzaniaZambia

Zimbabwe

-50 -40

D1980-1985 M1986 - 1991

25

-30 -20 1 0


period. Again, Mozambique and Zambia stand out at thebottom of the scale, with negative and falling real interestrates over the whole period, in spite of the fact that nominalinterest rates have escalated sharply during the decade inboth these countries.

4.5 Terms of trade and the current accountA characteristic element of almost every structural adjust-ment program is the outward orientation of economic pol-icy. Rapid devaluations of the currency are undertaken toimprove the competitiveness, again increasing exports, andincreasing the ability to pay for imports. The terms of tradeare defined as export prices compared with import prices,and thus they measure the ratios at which the exports of onecountry will exchange for those of its trading partners. Forsmall open economies, the terms of trade are exogenouslyfixed by world market prices, at least for a constant patternof trade. But the terms of trade may be increased if the tradepattern is adjusted to facilitate an improved exploitation ofcomparative cost-advantages. When calculated as the ratioof a country's index of average export prices to the averageimport price index, the result is an index for the terms oftrade.

In figure 4.8 such an index forms the basis for the reportedterms of trade adjustment. An increase in the reported indexmeans an improvement in the terms of trade. Thus, for sixout of ten countries figure 4.8 implies a deterioration incompetitiveness.

This worsening in the terms of trade is most significant forBurundi, Malawi, and Rwanda, while Ethiopia, Kenya, andZimbabwe seem to have experienced more moderate nega-tive changes in competitiveness. Botswana is the onlycountry where the improvement in the terms of trade is sig-nificant. However, a moderate improvement in the terms oftrade is also observed for Mozambique, Tanzania, and Zam-bia. Figure 4.9 illustrate how the current account has de-veloped over the 1980's. The reported current account is thesum of net exports of goods and non-factor services, net fac-tor service income, and net private transfers.

Botswana again reveals the most striking pattern, with a sig-nificant improvement in the balance of the current accountover the decade. From an average annual deficit of about12% of GDP, the current account of Botswana has turnedinto an average annual surplus of 10% of GDP. This trend isshared with Zimbabwe, which also seems to have turned thecurrent account deficit into a modest surplus during the1980's. These are the only countries reporting an annualaverage current account surplus over the period 1986 -1991. However, there are some other countries that have im-proved their situation by reducing the surplus. Among theseare Ethiopia, Malawi, Rwanda, and Zambia. The rest of thesample report an increased current account deficit over the1980's. Together with Tanzania, Mozambique representsthe most serious deterioration. For Mozambique, the aver-age annual deficit has increased from 18.5% to 49% ofGDP. Whatever the explanations are for this development,

the economic policy of the 1980's can hardly be claimed tohave improved the medium-term balance of payments situ-ation of the Mozambican economy.

The presentation of the data set has been rough, leaving forthe reader to draw conclusions regarding the implications ofstructural adjustment. Generally, however, it seems that thestructural adjustment policies has succeeded only partiallyin achieving their primary goals of reduced monetarygrowth, improved terms of trade, reduced inflation, in-creased real interest rates and economic growth. Further,there is a certain tendency that countries with relativelyhigh monetary growth are better off than countries withcomparably low growth in the supply of broad money. Tobe able to test these conclusions in a more rigid framework,the next chapter applies econometric methods to the data setpresented above.

Figure 4.9 The current account as . percentage share of GDP1980-1991

I


5. Testing the impact of policy change

5.1 A heuristic model of growthThe analysis in this chapter applies a more rigid method toanalyze the data set presented in the previous chapter.More precisely, the aim is to establish relations between out-put growth and various explanatory variables. Hopefully,this will reveal a bit more of how the change in selectedmacroeconomic indicators has influenced the evolution ofthe gross domestic product over the twelve-year periodfrom 1980 to 1991. Econometrics provides us with a scien-tific method in which the tools of economic theory, mathe-matics, and statistical inference are applied to economicphenomena. This enables us to fit mathematical models to agiven set of real world observations of the macroeconomy.

With macroeconomic time series for ten countries over atwelve-year period, the total number of observations is 120.This should suffice to test some simple but stringent hypo-theses concerning macroeconomic policy in the countries ofour sample. The applied mathematical relations must begeneral, in order to capture effects which are common to theten countries. However, as this study must be regarded as in-troductory and cursory, the outline must necessarily berather sketchy. A reasonable point of departure seems to beto explain the evolution of gross domestic product. Thiscorresponds to explaining the growth rate of the economyfor the ten countries of our panel. Thus, the dependent vari-able of the below analysis is the rate of change in the grossdomestic product.

The next step is to select the independent, or explanatory,variables. In concordance with the previous chapter, wewould like to test whether the most typical features of thestructural adjustment programme have contributed to thegrowth in the gross domestic product. Ideally speaking, thiswould require an exhaustive set of empirically-based beha-vioural relations for the economy. As the scope of this workis limited, we will have to go about the problems in alterna-tive way.

A kind of heuristic approach has become quite typical whenthe impact of structural adjustment is to be estimated byeconometric methods. Rather than specifying a full-blownmacroeconomic model, economists try to identify factorswhich have affected the growth potential in the countries

which are studied. The World Bank reports reviewed bySummers and Pritchett (1993) draw in part on suchmethods, and recent work published in international jour-nals confirm a growing consensus on this kind of methodo-logy.

Faini et al. (1991) offer a statistical analysis based onpooled time series for 93 developing countries, and the aimis to identify the existence of a turnaround in macroecon-omic indicators between the two periods 1978-1981 and1982-1986. Stylized models are estimated where thechange i performance indicators are taken to depend on au-tonomous policy changes, changes in the external environ-ment, and on participation in IMF-World Bank adjustmentprograms. After controlling for initial conditions and exter-nal factors, Faini et al. (1991) find no evidence of a statisti-cally significant improvement in macroeconomic perfor-mance among loan recipient countries. However, the timespan of their analysis is short, suggesting that improvementsmay still occur.

Morley (1992) applies a similar set of methods to analyzethe effects of devaluation during stabilization programmesin developing countries. Based on pooled time series for 28developing countries during the period 1973-1990, Morley(1992) estimates a simple relationship between the capacityutilization of the economy and different fiscal and monetaryvariables, including the real exchange rate. This study sug-gests that devaluations of the currency have significant ne-gative effects on the volume of production. Further, it is ar-gued that the process by which this link takes place is slugg-ish, as the effect of devaluation takes two years to materi-alize.

King and Levine (1993) represents a recent work in whichthe links between the financial sector and the real sector areinvestigated by means of econometric methods. Using apanel data set for about 80 countries from 1960 through1989, King and Levine (1993) study whether higher levelsof financial development are associated with economic de-velopment. Regressing long-run per capital GDP growthagainst a variety of financial indicators, it is concluded thatthe levels of financial deepening are strongly and robustlycorrelated with the general potential for economic develop-

27


ment. Further, it is argued that the financial variables areable to predict the subsequent values of the growth indica-tors. From this, King and Levine (1993) conclude that thereis a clear causality between financial and real development,but this conclusion is strongly contested by other economist-s in the field (e.g. Lucas [1988]).

These three example should illustrate that the below modelrelates to a certain heuristic tradition within empirical devel-opment economics. The next step is to pick the macroecon-omic indicators which are suspected to have influenced therate of economic development in our ten sub-Saharan coun-tries. At the same time, these must be factors which may be(and have been) manipulated by economic policy, or morespecifically by the structural adjustment programmes. Typi-cal targets of the structural adjustment programmes includethe rate of monetary expansion, outward orientation, the ex-change rate, the inflow of capital, and the interest rates.

First, we would like to include some sort of monetary vari-able, as structural adjustment programmes typically includerather strict controls with the money supply. A broadlydefined money supply can be taken to reflect the amount ofdomestic credit to the economy. Financial markets charac-terized by credit-rationing and interest rate regulations pro-vide a direct link between the money supply and the volumeof domestic credit because of the direct monetary trans-mission mechanism through the banking system. In addi-tion, credit plays a special role in the production technologyin developing countries, being the only source of externalfinance. This means that monetary policy might have astronger impact on the level of economic activity in develo-ping countries than in industrialized countries. The mon-etary variable of our model is approached through a broadlydefined money supply ( Me ), including all monetary andquasi-monetary liabilities of a country's financial institution-s to residents other than the central government.

Second, our analysis should include a variable which re-flects the outward-orientation of almost every structural ad-justment programme. The terms at which a country partici-pate in the international markets for goods are important tothe growth potential for the economy as a whole. The termsof trade index reported in World Tables ( ) is therefore in-cluded as an explanatory variable in the below simplifiedmodel of output growth. For a small open economy, theprices of exports and imports are fixed by the world market,and for given exports and imports, the level of the terms oftrade index will therefore be out of reach. However, structu-ral adjustment implies an adjustment also in the countries'trade with the rest of the world, and exports and imports aretherefore not constant. Accordingly, appropriate policymeasures may be applied to restructure exports and imports,and this will push the terms of trade index upwards.

Third, we analyze the growth impact of changes in thenominal exchange rate. As illustrated in chapter 2 and 3,rapid devaluations of the domestic currency are almost pivo-tal to the structural adjustment programs for sub-Saharan

Africa. This is done to neutralize the effect from domesticinflation on the international competitiveness. However, thereal world effects of devaluations in developing countriesare strongly debated. As mentioned in chapter 2, devalu-ations may have contractionary effects in developing coun-tries. These effects are especially due to the crucial role ofimported inputs. To investigate whether devaluations seemto have influenced on the growth and level of GDP, thenominal exchange rate (Es ) is also included as an inde-pendent variable in the below model.

Fourth, our investigation should also allow for positivegrowth effects from capital inflow. It is often claimed thatdevelopment aid has significant short-term beneficial ef-fects because it increases the import capacity of the econ-omy. Imports are of extreme importance for the productiontechnology, especially for manufacturing industries. An-nual net official transfers ( Fe ) are included to capture theeffects on domestic production originating from the inflowof development aid.

Finally, the model should incorporate some of the ideasfrom the financial development literature Financial sectorreforms are present in most structural adjustment pro-grammes, and these include a set of policy measures. De-regulations of the credit markets are intended to smooth theflow of funds from savers toward real investment. At thesame time, credit is the only source of external finance forthe producers, and liberalization of interest rates there im-plies an increase in the costs of finance, and the marginalcosts of production. Nominal interest rates ( Re ) are intro-duced to test these hypotheses in an empirical framework.

In standard macroeconomic approaches, the effects of allthese monetary variables are closely interrelated, and iso-lated effects on the gross domestic product are therefore nor-mally difficult to identify. For example, the effects on econ-omic activity from interest rate changes are usually closelycorrelated with the effects of general monetary factors, suchas capital inflow, exchange rates and monetary growth.Thus it would be hard, not to say impossible, to isolate inter-est rate effects from other monetary effects in a model ofthis type for an industrialized country with competitive mar-kets for money and capital.

However, the access to domestic and international financialmarkets has been strictly regulated in all the countries ofour sample Access to official domestic credit have tradi-tionally been restricted by credit ceilings and interest rateregulations, and these are generally under public control.Markets for foreign exchange have also been administeredover the entire period, as the exchange rates have beenfixed, and determined by the central bank according to rul-ing principles of economic policy. In addition, official mar-kets for foreign exchange have been close to absent, and thedistribution of this rationed asset has been undertaken bythe public administration. The problems concerning endoge-neity and simultaneity may thus not be too severe when wedeal with data for developing countries, like the ones in our

28


sample. It follows that a simplified short run model ofgrowth for this kind of countries may include monetarygrowth, exchange rates and interest rates as explanatoryvariables.

The listed set of variables does not at all constitute an ex-haustive set of macroeconomic growth factors. Of course,this study neglects a number of interesting magnitudeswhich are suspected to influence the rate of economicgrowth. Recent works of economic growth has pointed outthe importance of human capital formation (e.g. Lucas[1988, 1993]), and it would therefore be of interest to in-clude a variable to account for the variation over countries,and over time, in the change and level of education. How-ever, the different reported school enrolment ratios showlittle variation over time, and therefore they are likely to becorrelated with the constant term of our estimated models.There is also a serious lack of data when this information isto be gathered for the ten countries of our panel.

Another variable that may have interest is the public budgetdeficit. In the short run, one would expect that an increasein the budget deficit would stimulate aggregate demand, butthe long run effect is often claimed to be adverse with re-spect to economic growth. Unfortunately, the sources of thedata set are not generous enough to let us include the budgetdeficit as an explanatory variable.

5.2 Econometric specificationThe next step is to specify the relationship which is to formthe basis for the econometric estimation. Our model includethe real gross domestic product as a dependent variable.The explanatory variables are the money supply ( Me ), aterms of trade index ( T, ), the nominal exchange rate(Es ), annual net official transfers ( Ft ), and the nominal in-terest rate ( Re ). As a point of reference, we assume that along-run economic relationship exists between the level ofthe gross national product and the independent variables in-troduced above. The idea is that any long-run real nationalincome level ( Y, ) is associated with a specific level of theabove listed independent variables through a long-termequilibrium structure.

This long-run relationship may be stated generally as:

(11) Yt =g(111,,T t ,E t ,F o R t )

where g(•) is the mathematical correspondence between thelong-run level of the gross domestic product and theexogenous policy variables. For competitive economies,monetarists would typically claim that the influence of mon-etary factors on this long-run equilibrium is zero( g'm = 0 ). An improvement in the terms of trade are likelyto have positive influence on evolution of GDP ( g'T > 0 ),while the effects of devaluations are strongly debated

(g = ? ). To the extent that the recorded net official trans-fers embodies the variation in the inflow of developmentaid, the effect on the national income is usually assumed tobe non-negative ( g'm 0 ). Deregulating the financial sec-tor means a liberalization of interest rates, and this is as-sumed to be growth promoting ( g'R > 0 ). Our results willhopefully enable us to draw conclusions regarding the ex-tent to which these general results seem valid for the develo-ping economies of our data set. For simplicity of expositionand to facilitate a convenient interpretation of the results,the explicit formulation of the above function is charac-terized as multiplicative:

(12) yt = A/WI 71;42 EcL3 J4 Rta5

where A is a constant term. Equation (12) now implies thatthe a • 's (j = 1, 2, 3, 4 ) can be interpreted as partial elas-ticities, representing the percentual change in national in-come accompanied by a one per cent increase in the actualindependent variable. Furthermore, this functional form isespecially convenient for simple econometric analysis (ordi-nary least squares), where linearity in the parameters is a re-quired property for estimation of the parameters. We nowmay transform equation (12) to read:

(13) y, = ao + a i mt + a2 t„ + a3 et + ft + a5 rt

where ao InA and small types generally represent the natu-ral logarithm of the corresponding capitals ( x, laXt ).Equation (13) represents the long-run relationship of ourmodel, and may describe theoretical phenomena pretty well.However, the real world is typically characterized byshocks and rigidities, to which the economy must respondcontinuously. The process by which our data set was gener-ated will typically reflect these dynamics, and equation (13)will have limited value when real-world phenomena are tobe described. What is required is therefore a dynamic modelwhich implicitly encompasses a long-run economic relation-ship like the one in equation (13).

This need is met by the error-correction model, where thechange in the left-hand variable depends on changes in theexplanatory variables in addition to a so-called error-correc-tion term (Engle and Granger [1987]). The error-correctionterm reflects the deviation from the lagged long-run equili-brium of equation (13). Accordingly, this variable's par-ameter, the error-correction coefficient, reflects the share ofthe deviation from the long-run equilibrium which is equili-brated in every period. Thus, the error-correction coefficientserves as a measure of the sluggishness in the dynamic pro-cess behind the data set. Letting , denote the percentualchange over time in the variable X, and introducing ut aswhite empirical noise, the empirical error-correction formu-lation corresponding to equation (13) is given by: 1

1 Formally, the relative time derivative is defined by:

dX, 1 dlnx t

dt . X

t dt

29


(14) j7„ = + „+ a21,+ a 3e. „ + a4f,+ a5r,+ + u,

where X is the error-correction coefficient, and ecmt is theso-called error-correction mechanism, defined by:

(15) ecm, = y, - [ao + a i m, + a2t, + a3e, + ag ft + a5r,

We are now ready to formulate the estimated econometricequation, from which both short-run and long-run effectsmay be derived. Letting subscript i denote the differentcountries of the sample and introducing a io country-specificfixed effects, the econometric equation which is estimatedis given by:

(16) Y it = (7,0 + a l Mit + a2 ti, + a3 eit + a4lit + a5

b0 Y b1 mit--1 b3 e

+ 64 f it-1 + b5 + u,„

The coefficients of the differentiated variables may now beinterpreted as short-run effects of shifts in the macroecon-omic policy, whereas the level-effects are more closely con-nected to the underlying long-run equilibrium. Indeed, froman estimated version of equation (16) estimates for the long-run parameters of equation (13) can be derived as:

j = 1,2,3,4,5

Table 5.1 Estimated growth impulsesDependent variable: GDP growth (pt ), estimationmethod: Ordinary Least Squares.

Variables Model 1 Model 2 Model 3 Model 4

Mit 0.07 - 0.04 0.05 0.05 -t it 0.01 0.01 0.01

eit -0.01 -0.01 -0.01

fit 0.00 0.01 0.00

f it -0.24 -0.21 -0.24 - 0.25**

Yi t-1 -0.14 -0.11 *** -0.13 - 0.11 ***

mi t-i 0.02 0.02 0.03***

0.03 -

ti t-1 0.04 0.04 - 0.04*lc*

0.04-

e i t-1 0.00 -0.00 -0.01

f it-1 -0.00 0.01 0.00

r it-1 0.00 0.07 0.06 0.05

CON ST 0.39 -0.14

BWA 0.07- 0.07 - 0.07-BDI 0.02 0.02 0.02 **ETH -0.00KEN 0.01 0.01 0.01MWI 0.00MOZ 1)

RWA -0.01

TZA 0.01

ZM B -0.01

ZWE 0.01

R 2=0.49 R 2=0.60 R2=0.59 R 2 =0.59F(12, 98) =7.1 F(21$9) = 6.44 F(14, 96) =9.9 F(9, 101) = 15.9

*** Significant at 5% level.** Significant at 10% level.

Significant at 10% level with a one-tailed test.I) Mozambique serves as the point of reference in this model.

AA "i0

(17) aio = To-

Aa -= gio

s.e. = 0.041 s.e. = 0.037 s.e. = 0.036 s.e. = 0.035

702(2) = 2.6 x2(2) = 7.1 x 2 (2) = 9.2 X2(2) = 7.2

V i

Consequently, the specification in equation (16) opens forboth short-run and long-run effects originating from shiftsin the explanatory variables. This concludes the general for-mulation of an econometric model, and we now turn to theestimated results.

5.3 Estimation

The data set is transformed to fit the formulation of equa-tion (16), and the relation is then estimated by ordinary leastsquares. Different versions of the model are tried out, and asmall subset of the estimated models is presented in table5.1.

The first four lines of the table contain the effects of dif-ferentiated exogenous variables, and may be interpreted asshort-run elasticities. Then there are five lines for laggedlevel-effects, followed by the common constant term.Country-specific fixed effects are captured by the last tenlines of table 5.1, and the abbreviations correspond to thosedefined in section 5.1. We will first comment on the techni-cal differences between the versions of the model, before re-

sults concerning economic effects and policy implicationsare discussed more thoroughly.

The first model excludes any variation between the coun-tries of the sample, except due to the included explanatoryvariables, as all country-specific constant terms are left out.The constant term of the model may be interpreted as the au-tonomous rate of growth, or the rate of evolution in thegross domestic product which cannot be attributed to any ofthe other explanatory variables. Thus, model 1 implicitly as-sumes that the degree to which variation in the explanatoryvariables explains changes in aggregate output is identicalbetween the ten countries.

There may, however, be constant country-specific effectsthat have affected the economic growth potential over the1980's. To take one example, the mineral resources of Bot-swana may have improved the general economic climaterelative to the other nine countries. Another example is thevariation over countries in the educational level of the popu-lation. This must be assumed to influence the variation in

30


the growth rates between the countries. A crude measure isintroduced in model 2 to control for such country-specificvariations in the autonomous rate of GDP growth.

Model 2 incorporates all the effects of equation (16), andmay thus be referred to as the full-blown version of oursimple model of economic growth. This model includes amaximum of variables in addition to the country-specificconstant terms. The constant term is allowed to vary overthe countries in the sample to isolate the growth effects ofthe four ordinary independent variables. To be able estimatethe model by econometric methods, one of the country-spe-cific constant terms had to be left out. Therefore, in the sec-ond model Mozambique serves as a reference country fromwhich all the other country-specific constant terms aremeasured as deviations. However, the results from this sec-ond step of the procedure indicates that quite a few of thecountry-specific fixed effects do not deserve their positionin the model, as they do not deviate significantly from zero.The next step therefore excluded the insignificant country-specific dummy variables as well as the common constantterm, leaving us with the results of the third column of table5.1.

The last step of the estimation procedure was to excludealso the ordinary explanatory variables which do not seemable to contribute significantly to the explanation of the rateof economic growth. The result is the "final" version of themodel, and this is reported in the far-right column of table5.1. In this preferred version of the model, the selection ofcountry-specific fixed effects reflects the statistical signific-ance of such effects, and so does the included explanatoryvariables.

The establishment of the estimated econometric equationmay thus be divided into four steps: the estimation of thegeneral version of the model, the estimation of the full-blown model, the selection of country-specific fixed effects,and finally, the selection of ordinary explanatory variables.We have started from a general specification and repeatedestimation has brought the model closer to a more specificand restricted form. This is in consonance with general prin-ciples of econometric modelling of time series.

In evaluating the different versions of the model, we haveutilized some statistical diagnostics, and these are reportedin the bottom line of table 5.1. The first indicator of the va-lidity of the model is the R 2 , measuring the proportion ofthe total variation in the dependent variable that is ex-plained by the estimated model. None of the estimatedmodels are able explain more than 60 per cent of the vari-ation in the data set, as measured by the reported R2 . The in-terpretation of the results should therefore be exercised withcautiousness. Second, the reported F-statistic indicates thejoint statistical significance of all the estimated parameters.The null-hypothesis that all estimates are jointly equal tozero is rejected for all the reported models. Third, the stand-ard error of the estimation is reported. The standard error ofthe regressions are quite high, but as we are working with

panel data for developing countries, one must expect quite abit of unexplained variation. Fourth, a X2 test statistic iscomputed to test whether the residuals are normally dis-tributed. Due to outliers, this hypothesis of non-normality isrejected only for model 1. All these diagnostics have beenused in the gradual approach of the final preferred model,and except from the Jarque and Bera (1980) test of nor-mality, they have all improved during the procedures of esti-mation. Statistical significance and a sign in accordancewith hypotheses in development economics should indicatethe relevance of the included variables, but the absolute sizeof the different coefficients must be interpreted with care.Further, econometric models that combine macroeconomictime series for different countries can not easily be taken torepresent appropriate behavioral economic relationships.However, the estimated effects in table 5.1 tell us some-thing about the partial correlations between economicgrowth and different other macroeconomic indicators.

5.4 Discussion of resultsAll results will be commented using model 4 as the centralreference. The other models of table 5.1 serve to illustratethe procedures of estimation, where the aim has been to findthe model that describes the data-generating process in thebest way. On this background, model 4 seems preferable sofar. Below, the different elements of the model are dis-cussed, starting with the country-specific constant terms.The estimated effects of changes in the broad money sup-ply, the terms of trade, devaluations of the domestic cur-rency, net official transfers, and nominal interest rates arethen clarified. Finally, estimates for the underlying structu-ral model are presented to shed some light on the long-rununderlying equilibrium, on which the growth model is con-ditioned.

Country-specific fixed effectsThe constant terms of the model are allowed to vary be-tween the countries of the sample. Mozambique serve as thepoint of reference in model 2, and the country-specific fixedeffects are measured as deviations from the constant termfor Mozambique (estimated to be zero). In model 4, fixedgrowth effects are appended for three countries. The esti-mated fixed country effects include both positive and nega-tive coefficients. However, this does not automaticallymean that the growth rates vary according to the sign ofthese fixed effects. Rather, the fixed effects are meant to iso-late constant country-specific growth factors apart from theincluded explanatory variables.

For all the countries of the sample there are many growth-explaining variables that are left out of our model. If thesum of these effects has been somewhat constant over theestimation period, they are captured by the constant terms.Thus, the fixed effects give a hint of how large proportionof the variation in growth rates that cannot be explained byvariation in the ordinary explanatory variables of the model.Thus they may be interpreted as some kind of "autono-mous" rates of growth. For Ethiopia, Malawi, Mozambique,Rwanda, Tanzania, Zambia, and Zimbabwe, the estimated

31


fixed-effects do not seem significantly different from zero,and therefore, (the average of) these countries serve as thepoint of reference for model 3 and model 4. For the threeother countries of the sample there seem to be excluded con-stant growth-factors that produce a significantly higher rateof growth than for the rest of sample over the period. Theidea may be illuminated by the case of Botswana, where theautonomous growth rate seem to be 5.5 percentage pointshigher than for the other countries, e.g. Mozambique. Thissuggests that Botswana may be an economy that is quite dif-ferent from the Mozambican economy, and that all the im-pulses to growth are not directly comparable for these twoeconomies. The variation exposed in the figures of chapter4 seem to support this impression. Following the same lineof reasoning, Ethiopia, Malawi, Rwanda, Tanzania, Zam-bia, and Zimbabwe seem to be economies where the mech-anisms of economic development to a larger extent are com-parable to those of Mozambique.

Broad moneyAccording to table 5.1, monetary growth seems to have hada significant, but very modest, positive effect upon thechange in the gross domestic product over the 1980's. Theestimated short-run elasticity of broad money is positiveand relatively significant, but perhaps not as stable as weideally would like to see it. In the preferred model, mon-etary growth has a positive estimate of about 0.05, and thisestimate is significantly positive at a 10% significancelevel. This implies that it is likely that an expansion in thevolume of domestic credit has moderate positive effects onthe real rate of growth in the short run. On the other hand,this result also implies that contractions of the monetarybase are likely to cause the gross domestic product to fall inthe short run. However, the coefficient of the lagged levelof broad money has even more interesting characteristics.The partial level-effect of domestic credit is highly signifi-cant in our final specification, and seems quite robust to thedifferent versions of the growth model. This suggests thatthere exists a significant positive correlation between nomi-nal money growth and the gross domestic product also inthe medium term. Thus, our data set lends support to the hy-pothesis that changes in the volume of domestic credit arepositively correlated with the change of the national incomelevel. This reflects the observation that countries in oursample which can show a relatively advantageous growthperformance usually also have monetary growth rateshigher than the sample average (cf. chapter 4).

As mentioned above, the estimates concerning broad moneysuggest that the set of monetary constraints which usuallyaccompany the structural adjustment programmes mighthave contractive effects on the economic growth potentialof the countries of our sample. This analysis suggests thatthe monitoring of monetary growth should be exercisedwith modesty if medium-term economic growth is the pri-mary target of the macroeconomic policy. A tempting con-clusion is further that the multinational donor communityconcern with monetary control has been exaggerated, but asthe estimated effects of our model are very small, it is diffi-

cult to draw clear-cut conclusions. We should also remem-ber from chapter 2 that there is a strong link between mon-etary growth and inflation, and that inflation may have ad-verse effects which require separate treatment.

Terms of tradeMacroeconomic balance, fiscal restraint, and outward-look-ing policies are keywords of nearly all programs of structu-ral adjustment. To test the importance of orientation to-wards international goods-markets, our model of growth in-clude an index for the terms of trade. This index is extracteddirectly from World Tables 1992, and is intended tomeasure the ratio at which the exports of one country ex-change for those of its trading partners. For small open econ-omies, both export and import prices are determined interna-tionally, and for given amounts of export and imports, theterms of trade are thus exogenous. However, the reportedindex is defined as the index of average export pricesrelative to the average import price index. Changes in quan-tities of exports and imports will thus influence on the termsof trade. An increase in the terms of trade may therefore beinterpreted as a movement towards a trade equilibriumwhere comparative cost advantages are optimally exploited.This kind of structural adjustment is claimed to be of vitalimportance for the growth-potential of developing coun-tries. Changes in the terms of trade index will therefore beapplied to reflect general variations in the international com-petitiveness of the countries involved.

From table 5.1 we see that the immediate net effect uponthe economic growth rate from a change in the terms oftrade index is negligible. Therefore, this variable is ex-cluded in the final specification of our model. However, theestimated level effect lies very stable at 0.04 in all the differ-ent specifications, and is highly significant in the finalpreferred model. More precisely, the estimate suggests thatcountries with improving terms of trade have performed sys-tematically better than countries with constant terms oftrade over the 1980's. Adjustment of the trade pattern there-fore seems to be of vital importance for the medium-termgrowth potential, whereas the short-term effects of terms oftrade adjustment seem to be of minor importance.

The exchange rateA devaluation means that import prices rise relative to ex-port prices. If domestic exporters are producing competi-tively and without rationing, this will normally bring aboutan increase in exports. However, a devaluation also impliesthat imports become relatively more expensive, and this ef-fect should not be neglected, especially not in analyses ofdeveloping countries. As imports are of extreme importancein the production process in most developing countries, de-valuations may have contractive effects via the supply sideof the economy, because they tend to increase the marginalcosts of production. This is stressed in a number of theoreti-cal works on development economics, especially within theneo-structural macro critique (e.g. Bruno [1979], Krugmanand Taylor [1978]), and is also pointed out by Cappelen

32


(1992) in a recent comment on the economic recovery planin Mozambique.

Data for exchange rates are approached through the conver-sion factors reported by the World Bank (1992b). This con-version factor represents the annual average of market ex-change rates for countries quoting rates in units of nationalcurrency per US$. Further, an index is constructed to beable to compare the evolution in the exchange rate betweenthe different countries of the sample. Accordingly, an in-crease in this index of exchange rates will reflect a devalu-ation of the currency, making the interpretation of the esti-mated coefficients quite straight-forward.

The estimated short-run effect of changes in the exchangerate does not support the hypothesis concerning contractiveeffects of devaluation, at least when measured by the totalnet effect on GDP growth. An increase in E t implies a de-valuation of the nominal currency, and this seems to pro-duce a modest, but statistically insignificant, negative effecton the rate of growth, as reported in models 1-3. This shortrun change in the exchange rate index is therefore excludedint the final preferred version of the model. The estimatedlevel effect from changes in the exchange rate have similarproperties, and is also excluded in model 4.

The preferred model in table 5.1 thus seems to suggest thatthe export-promoting effects of devaluations tend to balancethe contractive effects due to an import intensive productiontechnology. Seemingly, our estimated model of growth forten countries in Sub-Saharan Africa thus lends only limitedsupport to hypotheses concerning contractive effects of de-valuation. However, there are certain statistical problems atthis point of the analysis. Partial regressions suggest that themonetary growth rate is closely correlated with the rate ofdevaluation. The reason for this might be that monetarygrowth is used as an indicator of future inflation. As devalu-ations have been implemented to counteract the effects onthe international competitiveness from domestic inflation,this will cause a close relationship between monetarygrowth and the exchange rate. This is a typical problem ofeconometric analysis, and is referred to as multicollinearity.The consequence of this property of the data set a low de-gree of precision in the estimated coefficients. For our esti-mated model it also implies that it will be hard to identifythe isolated effects on the economic growth rate fromchanges in the money supply and adjustments of the ex-change rates, respectively.

Net official transfersInflow of capital in the form of development aid is fre-quently claimed to have substantial growth-promoting ef-fects on developing countries, but various contributionswithin the macroeconomic development literature now castdoubt on this earlier quite generally accepted conclusion(e.g. Cassen et al. [1986] and Pedersen [1991]).

As most of the countries in our sample are excluded from in-ternational credit on market-based terms, inflow of capital

largely take place in form of subsidized credits and develop-ment aid. Short-term demand-side effects are likely to occurafter changes in the net capital inflow of this type, but rent-seeking and other non-productive activities may squeezeout even the short-term effects of capital inflow. In the me-dium to long-term there are important effects from net trans-fers concerning macroeconomic structure and economic in-centives (e.g. van Wijnbergen [1984]), and these may neu-tralize the growth impact of net transfers. The empirical evi-dence is, accordingly, ambiguous in this area. To test for theinfluence of development aid, net official transfers has beenincluded in the estimated model.

Net official transfers comprise net transfer payments be-tween governments of the reporting country and the rest ofthe world. As chapter 4 revealed, this variable takes on highvalues for Botswana. This suggests that the recorded vari-able may include more than official development aid. Bot-swana has generated international funds from their miningactivities. If net official transfers include some of the inter-est payments from these funds, it would typically skew thedistribution in favour of Botswana, as none of the othercountries of the sample are likely to receive official interestpayments from abroad. This should be held in mind whenthe results are interpreted.

Our estimated model of the far-right column of table 5.1does not confirm either of the hypotheses concerning inflowof aid and economic growth. The estimated short-term ef-fect from a change in net official transfers is positive, butvery small, and not statistically significant, and is excludedin the final version of the model. The level-effect is moreclosely connected to the long-run equilibrium, but has simi-lar properties, and is therefore also excluded in thepreferred model 4. As measured by net official transfersover the 1980's, our model is thus not able to uncover in-fluences from the inflow of development aid on the rate ofeconomic growth.

At this point, the model does not seem appropriate for an ex-planation of the turnaround in Mozambique in 1987. As theinflow of donor funds increased dramatically due to the im-plementation of the ERP, growth rates turned significantlypositive in Mozambique. Over the last seven years, this indi-cates a close relationship between the import capacity andthe rate of economic growth. On the other hand, the poten-tial for a turnaround seem to have been exceptional in Mo-zambique, due to the extraordinary macroeconomic, and se-curity-related, conditions at the middle of the 1980's. Thus,the inflow of capital to Mozambique may have been farmore effective for generating economic growth over theshort to medium term than for the other countries. To the ex-tent that these effects are temporal, they also shed somelight on the stagnation that has taken place in Mozambiqueafter 1990. Expansionary effects of development aid arelikely to be relevant also for the other countries in the re-gion, but our estimated model is not able to pick up these ef-fects.

33


Interest ratesThe nominal interest rate is included to account for two dif-ferent aspects of the macroeconomy in developing coun-tries. First, an increase in interest rates is likely to produce acontraction of the volume of production, because the imme-diate response of investment demand will be negative. Thiseffect is identical to the one found in standard textbook ex-positions of IS/LM-analysis. Second, the interest rate is in-cluded to expose the effects from financial liberalizationupon the rates of economic growth. We therefore hypo-thesize that interest rates have a negative short-term effect,whereas the long-run structural effect from increasing inter-est rates is positive.

The estimated immediate effect from an increase in thenominal interest rates is negative, fairly stable and very sig-nificant. This effect suggests a negative short-term growthimpulse from an increase in the real interest rates. In magni-tude, the effect seems to be far more dramatic than of achange in the broad money supply, and the estimate is alsomore precise. Thus our data set lends support to the notionthat interest rates are negatively correlated with the level ofproduction in the short-run. The level effect is also rela-tively stable, but not as significant as the short run effect ofinterest rate changes. More interesting, perhaps, is the factthat the estimated long-run effect has the opposite sign ofthe short-run effect. The effect of an increase in interestrates is thus expected to have a positive long-run influenceon the rate of economic growth.

This long-run result contradicts the standard hypothesis ofmacroeconomic theory, but is well in line with the tradi-tional literature of financial development (cf. Fry [1988]).In industrialized countries, the interest rate tend to be nega-tively correlated with the level of national output, also inthe medium term. A reason is that competitive capital mar-kets redirect financial capital towards investment demand inthe case that interest rates should fall, and this contributes toan increase in aggregate demand. Things look quite differ-ent in representative developing countries.

First, the fixed interest rate ceilings place restrictions on thesupply of loans, and the volume of outstanding loans is lessthan it otherwise would be. Second, the regulations of nomi-nal interest rates usually imply low or negative real depositrates, and therefore nobody wants to place their savingswith the banking system unless for special reasons. Third,low interest rates tend to cause capital flight, because the do-mestic financial system can not compete with the interna-tional capital market. Fourth, rationing in official financialmarket almost inevitably give rise to grey or black marketsfor credit, rising the marginal costs of finance dramatically.Deregulations of the financial system is often claimed tomobilize savings, both domestic and foreign. Further, thebanking system is affected, as financial resources can besupplied to more profitable projects at a higher interest rate.Accordingly, an increase in interest rates may cause an in-crease also in the gross domestic product, at least in thelonger run.

The ideas that the real effects of interest rates changes havea credit-availability channel which is special for developingcountries have been subject to theoretical and empirical re-search for decades (Mohn [1992]). Influential early theoreti-cal contributions include McKinnon (1973) and Shaw(1973), and the literature on development economics hasbeen quite seriously concerned with the role of the financialsector in the macroeconomy ever since (cf. Fry [1988]).

In an empirical study of macroeconomic investment beha-viour in Greece, Voridis (1993) estimates robust and posi-tive coefficients for user costs in his investment relations.This is claimed to reflect the expansionary effect from in-creasing interest rates on capital formation, again increasingthe availability of loanable funds. In a more general model,using panel data for 80 countries, King and Levine (1993)also finds strong support for the hypothesis that financial de-velopment is positively correlated with the rate of economicgrowth. Hence, the above estimated growth model for Sub-Saharan countries can be taken to be in accordance withthese two recent empirical works on related topics.

The structural modelAs pointed out under the specification of the econometricmodel, specified growth model is conditioned on someunderlying long-term equilibrium relationship through theso-called error-correction mechanism. The change in thegross-national product is thus assumed to continuously re-duce the deviation from an underlying long-term structuralequilibrium. The parameters of this underlying equilibriumare derived by assuming that all changes approach zero, andthen solving for the dependent level variable. This solutionfor the long-run structural elasticities is illustrated formallyby equation (17) above.

The estimates of the underlying long-run macro productfunction are derived according to equation (17), and the re-sulting structural elasticities are presented in table 5.2. Aswe see, the long-run elasticities are directly related to the es-timates of the growth model via the estimated error-correc-tion coefficient. As the adjustment towards the long-runequilibrium in table 5.2 is continuously hampered by thesluggishness of the error-correction mechanism, the

Table 5.2 Derived estimates for the structural parameters

Variable

Estimated structural elasticity

Broad money

0.18Terms of trade

0.36Exchange rate

0.00Net official transfers

0.00Interest rate

0.45Error correction term

0.11

Dependent variable: GDP-level ( )

34


presented structural elasticities can not be taken too lite-rally. The bottom line of table 5.2 reports the estimatederror-correction coefficient. The estimate of 0.11 impliesthat only 11 per cent of the deviation from the deviation iscorrected in each period, suggesting a moderate pace of ad-justment.

All the estimated structural parameters are consistent withthe previously commented effects in the dynamic model.Therefore, the below presentation of implications will notinclude any further discussion of the structural parameters.For broad money the long-run structural elasticity is esti-mated at 0.18. This implies that a one per cent increase inthe supply of broad money will increase the gross nationalproduct by 0.18 per cent. For the terms of trade index thelong-run elasticity is estimated at 0.29. Thus, the influenceof improving terms of trade on the level of GDP seem tohave been significantly positive over the 1980's for thecountries of our sample. On impact, changes in the nominalexchange rate effect do not seem to have had significant ef-fects on the evolution of gross domestic product, but the iso-lated effects from devaluations are hard to identify, as theyare closely correlated with the growth in the money supply.

The inflow of net official transfers seems not to be corre-lated with the level of GDP in the medium to long-term, aresult which resembles the conclusions of Faini et al. (1991)in their econometric evaluation of growth-oriented adjust-ment programs. As pointed out above, net official transfers,however, may be a inappropriate proxy for the inflow of de-velopment aid, which is the main variable of interest. Thelast structural parameter is the long-run elasticity of thenominal interest rate, and this is positive and twice the mag-nitude of the elasticity of broad money. This is in accord-ance with the discussion above, and the conclusions con-cerning interest rate deregulation and inflation control arethus maintained.

Some of our results indicate that shifts in economic policymay have short-term contractive effects, while the structuralparameters are positive. Further, the sluggishness of the esti-mated error-correction model indicates that it may take con-siderable time for policy shifts to increase the level of na-tional income. Confronted with the real world dilemmas ofdevelopment economics, this is an important piece of infor-mation, because it implies a general call for patience in theevaluation of macroeconomic policy reforms.

35

figOre6,1 Post-saMple simulation ftitMozambiqUe 19844.931

Estimated and observed GDP (1987-prices). Mt.bill.


6. Simulations and policy experiments

6.1 Post-sample simulationAny estimated econometric model should be subject tosimple tests of validity, and some such tests may easily beconstructed using the data set and the fitted equation. An il-lustrative and quite transparent test is to simulate the modelusing historical data, to see how well it describes the devel-opment which formed the basis for estimation. The Mozam-bican economy forms the point of reference for all theexamples and simulations presented below.

To investigate the performance of our single-equationmodel, it is simulated using historical data for Mozambiqueover the period 1980 to 1991. Simulations on observationsprior to 1986 revealed that for the early years of the sample,our model does not seem appropriate for explaining thegrowth in domestic production in Mozambique. The predic-tions for the beginning of the eighties were far higher thanthe observed economic development. This suggests thatother factors influenced the rate of growth over this periodthan those embodied in our model. However, we must re-member that we are simulating an economy at civil war.The general social and economic instability associated withthe civil-war situation are likely to have caused rather er-ratic shifts in the level of economic activity, and these cannaturally not be captured by our model. The predictive

600

500

7

-

------

------------

400 -7

300 -_ _ _ Observed

200 - — Estimated

100 -

01986'

1987 1988 1989 1990 1991

power of our equation, however, seem to improve consider-ably as the Mozambican government joined the multina-tional donor community, and initiated the first structural ad-justment plan in 1987. For some of the last years of the eigh-ties, our model projects rates of growth which are quite nearthe realized growth of the Mozambican economy. Thus theestimated model seem to capture quite a lot of the variationin gross domestic product. Somehow, the insecurity situ-ation might have settled by this time, stabilizing the econ-omy at a level reflecting all adverse effects of the civil war.

Figure 6.1 illustrates some one of the problems which werediscussed at greater length in chapter 3. The first three yearsfollowing the implementation of the ERP were charac-terized by positive and steady growth at an annual 4-5 percent. Toward the beginning of the 1990's, the growth poten-tial seems to have stagnated, and figure 6.1 illustrates thatthis stagnation can not be attributed to variation in the mac-roeconomic indicators of our simple model. Speaking interms of our model, the recent delay of economic develop-ment in Mozambique remains unexplained. For discussionsof ad hoc explanations, cf. chapter 3.

One may well object that the estimated model provides arather dubious description of the macroeconomic growthrate during the last decade. However, one should take intoaccount the fact that our model consists of only one equa-tion. Further, the modest aim is to explain the economicmechanisms of a developing country at war, using onlythree macroeconomic indicators. In this perspective, themodel's predictive power may nonetheless be approved of.

6.2 The reference scenarioIn the following, we shall conduct some experimental pro-jections of the macroeconomic level of activity and its re-sponse to policy-shifts, using the Mozambican economy asan example. The initial year of these simulations is 1993,and the model is calibrated to fit this year's observationsperfectly. For the future years of our simulations, we needprecise assumptions regarding the evolution in theexogenous variables. Further, in order to throw the effectsof the exogenous policy shocks into relief, we also need toestablish some kind of reference scenario, against which theeffects of economic policy shifts can be compared. This

37

Figure 6.3 GDP growth under:the reference scenario

6

5

4

3

Reference scenario

2

1

0 '1993 1996 1999 2002 2005 2008 2011


Projected evolution in gross domestic product (1987-prices), Mt.bill.1 200 -

1 000

800

600

400

200

0 '1992

1996 2000

2004

2008

2012

base line projection over twenty years is built on assump-tions which largely extrapolate the late development in thedifferent exogenous variables.

The reference scenario for Mozambique is established asfollows. First, we assume a constant monetary growth rate25 per cent. This magnitude is also representative for themonetary growth over the last five years, and may belooked upon as an average over the growth in the broadmoney supply over recent years. Second, the observed dete-rioration of the terms of trade is extrapolated, implying a2.5 per cent annual decrease in the terms of trade index. Fi-nally, we take the interest rate to be constant at 37 per cent,which is reported by IMF (1993) to be the relevant interestrate for 1993. Figure 6.2 presents the annual levels of GDPfor the reference scenario, whereas the projected growthrates are illustrated in figure 6.3. Observe that the rate ofgrowth is falling over time also in the reference alternative,reflecting the assumed continuous deterioration of the termsof trade, and the fact that the error-correction mechanism re-duces the rate of growth as the GDP level increases. Over

Projected annual growth rates in per cent

the twenty year projected period, the reference scenario im-plies an average annual growth rate of 4.3 per cent. Thisallows the gross domestic product of Mozambique todouble about every fifteen years.

The next step is to introduce the policy experiments. Ingeneral, there are two main groups of alternatives when weare to implement exogenous changes in our model. First,one may shift the exogenous variable in a once-and-for-allmanner. This will affect the level of the variables involved,whereas the growth rate of the variable will be left un-changed. Second, one may pursue continuous macroecon-omic policy plans, implying a gradual change in the differ-ent exogenous variables. This will affect the growth rates ofthe exogenous variables, and the levels will be affected indi-rectly and continuously. Combinations of these two maingroups of alternatives are also perfectly possible, leaving uswith a large possible number of policy plans to be evalu-ated. It is beyond this report to be totally exhaustive on thispoint, and therefore we restrict ourselves to presenting theimplications of partial changes in the variables of our esti-mated model. However, the implications of theseexogenous influences will be illustrated both for the dis-crete and for the continuous case.

6.3 Discrete shocks in policy variablesThe preferred estimated model includes broad money, theterms of trade, and the interest rate as exogenous policyvariables. This opens for three partial one-shot experiments.First, we hold the supply of money constant between 1993and 1994, whereas the monetary growth rates are assumedto remain unchanged (at an annual 25%) from 1995 to2011. This alternative is referred to as "One-shot contrac-tion of broad money" in figure 6.4. The second experimentis to assume a discrete (10%) improvement in the terms oftrade index in 1994, whereas the modest deterioration of thereference alternative is assumed to continue from 1995.This alternative is referred to as "One-shot improvement inthe terms of trade" in figure 6.4. The third experiment in-volves the interest rate, and here we assume a discrete in-crease from 37 to 45 per cent in 1994, and that this level ismaintained throughout the period. This alternative isreferred to as "One-shot increase in interest rates" in figure6.4.

As one might have expected, the impact of a contraction ofthe supply of broad money is a reduction in the gross domes-tic product, relatively to the reference scenario. The short-term effect implies a 1.2 per cent reduction from the first tothe second year, whereas the long term effect of this experi-ment seems to be a gross domestic product about 3.5 percent below the projection of the reference scenario. Further,a discrete increase in the terms of trade produces an in-crease in the level of activity, relatively to the base-line pro-jection. As our model does not include short-term effects ofchanges in the terms of trade index, the projection does notreact until the second year after the change, when thelagged level has responded. Over the period, the effect of adiscrete 10 per cent increase in the terms of trade seems to

Reference scenario

38

Figure 6.5 Simulated continuous policy experiments

One-shot contractionof broad money

— — —One-shot improve-ment in the termsof trade

One-shot increase ininterest rates

— — .Reduced monetary Constant terms Increased interestgrowth of trade rates

Rapporter 94/11

Structural Adjustment in Mozambique

Figure 6.4 Simulated one -shot policy changes

Projected deviations in GDP level relative to the reference scenario (%)

4Projected deviations in GDP level relative to the reference scenario (%)

be a gross domestic product about 3 per cent higher than inthe reference scenario. This illustrates the long-run positiveimpact of restructuring the international trade pattern.

The third example illustrates the projected effects of a dis-crete increase in the interest rate level from 1994, from anannual 37 per cent in the reference scenario, to a constant45 per cent in the present projection. As commented above,this produces a significant contractive effect in the shortrun, as the GDP growth rate falls from 5.6 to 4.0 per centfrom one year to another. However, the long-run structuralimpact of increasing interest rates seems to be favourable,again increasing annual GDP growth above the reference al-ternative. After a five-year period, an increase in the interestrate level has advantageous effects relative to the projectedreference every year of the period. This illustrates the long-term gains that might occur if the capital markets are re-structured to facilitate an efficient flow of domestic savingstowards domestic investments. The multinational donors aredevoted to these problems through their financial sector re-form programmes, commented in chapter 2.

6.4 Continuous policy experimentsThe second group of policy alternatives incorporate changesin the growth rates of the different independent variables.We still keep the reference alternative described above un-changed, and the implications of the alternative policy plansare measured as deviations from the reference scenario. Themeasures will also still be restricted to three partial experi-ments, each incorporating an isolated change in the growthrate of the three independent variables. In the first experi-ment we reduce the growth in the money supply from an an-nual 25 per cent to an annual 20 per cent over the completeperiod. This alternative is referred to as "reduced monetarygrowth" in figure 6.5. Second, we keep the terms of tradeconstant, in stead of extrapolating the deterioration assumed

in the reference alternative. This experiment is referred toas "constant terms of trade" in figure 6.5. Finally, we let theinterest rate rise by 0.5 percentage points every year throug-hout the period. This case is referred to as "interest rate in-crease" in figure 6.5.

As expected, the reduction in the rate of growth in broadmoney makes GDP growth slow down. By the end of theperiod this experiment yields an annual gross domestic pro-duct about 9 per cent below the projection of the referencescenario. This deviation will also continue increasing, be-cause the structural effect of this experiment implies anever-lasting negative deviation in the rate of economicgrowth, again relative to the reference alternative. Figure6.5 also illustrates how an improvement in the terms oftrade may produce an increase in the rates of growth, im-plying a continuous increase also in the deviation presentedin the figure. As there are no short-term dynamic effectsconnected to the terms of trade, the estimated structural ef-fect of our model produces a undisputable positive effect onthe evolution in the gross domestic product.

However, this is not the case for the last policy experiment,involving a continuous increase in the interest rates. Aschapter 5 illustrated, there seem to be adverse short-run ef-fects connected with an increase in the interest rates. Thatinterest rates are negatively correlated with the level of ac-tivity in the short run is not surprising, as this is the standardhypothesis of Keynesian macroeconomic theory. On theother hand, our model incorporates a positive long-run struc-tural effect of increasing interest rates, and this is not in ac-cordance with traditional macroeconomics. Figure 6.5 sug-gests that the impact of an interest rate increase is negativein the short run, but this negative influence is continuouslyhindered by the positive structural effect. With our choiceof policy, the structural effect of interest rate increase seems

39

Figure 6, 7 Economic growth under the complex policyexperiment:

Reference scenario

Policy experiment


to outweigh the negative dynamic effect only after about tenyears.

These experimental projections underscores the dilemma ofchoosing the right macroeconomic policies; The positive ef-fects of gradualism may be modest, and will often occurwith many years delay. However, the short term adverse ef-fects of structural adjustment may be reduced to trifles ifgradualism is chosen. On the other hand, the one-shotchanges of the previous section may have serious short-term adverse effects. Still the structural effects breakthrough at an earlier stage with this kind of policy, making"cold turkey" strategies more efficient, when measured byproductive gains alone. Thus, to some extent one may un-derstand those who claim gradualism to be slow and ineffi-cient. But one should also remember that gradualism repre-sents a much more considerate approach to the problems ofstructural adjustment. The proper answers are therefore farfrom evident. Social objectives and preferences concerningthe distribution of income may thus still make the gradualapproach best suited in pursuing the economic and politicalplans of the government.

6.5 Accommodating monetary controlOur model includes a positive spill-over from nominal mag-nitudes to real economic development through the trans-mission mechanism for monetary impulses. This impliesthat a contraction of the broad money supply has disadvant-ageous effects on the real level of activity. On the otherhand, one should not jump to the conclusion that an expan-sion of the broad money supply is the appropriate measureto increase the rate of GDP growth. The reason is that ourmodel overlooks the adverse effects of high and increasingrates of inflation. As discussed in chapter 2, inflation mayhave severe consequences on domestic production, capitalmovements, and interest rates. Our conclusions may belooked upon as valid only for a given rate of inflation. Tar-geting the broad money supply to reduce inflationary press-ures may thus still have relevance. The final simulated ex-periment is therefore constructed to illustrate the real effectsof reduced monetary growth, and to illuminate whatmeasures that might be taken to counteract these direct ne-gative effects on domestic production. To illustrate theideas we construct a specially tailored experiment.

First, we let the annual rate of growth in the broad moneysupply fall from 25 to 20 per cent from 1993 to 1994. From1995 to 2004 the same annual growth rate is reduced by 0.5percentage points every year. From 2004 to 2011, the mon-etary growth rate is held constant at the resulting 15 percent per annum. In isolation, this measure produces a con-traction in the gross domestic product when measured rela-tively to the reference scenario. If the production level is tobe maintained, the control of the money supply will there-fore have to be followed by appropriate accommodatingmeasures. To illustrate this, we first let the terms of tradeindex improve at a rate of 2.5 per cent in 1994, and this rateof improvement is then assumed to fall gradually towardsan annual 1 per cent in 2011. Finally, the interest rate is in-

Projected evolution in gross domestic product

1 200 -

1 000

800

600

400 -

200

1993 1996 1999 2002 2005 2008 2011

creased gradually to 50 per cent over the first three years,and is then held constant throughout the period. The impli-cations for the level and growth of GDP are illustrated infigures 6.6 and 6.7, respectively.

From figure 6.6 we observe that the reference scenarioyields a higher GDP level than the policy experiment forthe first five years after the implementation of policymeasures. Also GDP growth rates are higher for the refer-ence alternative, but only for the first two years. These re-sults are due to the short-term adverse effects of the in-crease in interest rates, which are at work together with thereduced monetary growth. However, the growth rates of ourpolicy experiment picks up by 1996, and when measured interms of economic growth, the alternative dominates the ref-erence scenario from then and throughout the period. Forthe level of activity, this simulated policy experiment im-

Projected annual GDP growth (%)6 -

5

4

3

2

1

••■

Reference scenario

Policy experiment

01993 1996 1999 2002 2005 2008 2011

40

Structural Adjustment in MozambiqueRapporter 94/11

Figure 6.8 Benefits of the comptex polity;experiment

Projected deviation in GDP level relative to the reference scenario (%)

4

3

2

1

- 1

-2 1993 1996 1999 2002 2005 2008 2011

plies a higher gross domestic product than the reference al-ternative for every year after 1999.

Figure 6.8 illustrates the projected benefits of our complexpolicy experiment. Compared to real world experience, theapplied policy measures are not especially dramatic. None-theless, it takes 3 years to improve on the growth rates, and6 years to improve on the GDP levels of the reference alter-native. This illuminates the sluggishness of the estimatedmodel, a feature which is likely to characterize also the ac-tual functioning of the macroeconomy in developing coun-tries. If the macroeconomy respond largely to laggedchanges and levels of different macroeconomic indicators,instead of their current values, the economy will typicallybe sluggish in its response to external shocks and policychanges. The error-correction formulation of our single-equ-ation econometric model is one example of such mechan-isms. This should also remind us of the pittance which is re-quired in evaluating implemented macroeconomic stabiliza-tion plans for countries such as Mozambique.

The presented simulations should not under any circum-stance be taken as literal projections of the evolution of theMozambican economy. Rather, they should be consideredas illustrative examples of macroeconomic mechanismswhich may be of relevance for the design of economic pol-icy. It has not escaped our attention that there are numerousfactors which affect the growth of an economy, and that ourestimated equation only captures a minimal fraction ofthese. Nonetheless, it is hoped that the above expositionwill stimulate stringent and empirically oriented thoughtson how to design the macroeconomic reforms for develo-ping countries.

Policy experiment

41


7. Conclusions

The design of appropriate macroeconomic structural adjust-ment policies for developing countries remains a complexand highly debated issue. Many reasons may contribute tothe explanation of the background for this debate. First, themultinational donor community offers integrated structuraladjustment programmes which can be criticized for beingstereotype, and inconsiderate with respect to country-spe-cific variations in political, economic and social conditions.This is especially relevant for Mozambique, as this countryis extremely poor and debt-ridden, also by World Bankstandards. In addition, a relatively standard structural adjust-ment programme was implemented in the domestically un-stable situation which accompanied the civil war of the1980's. The macroeconomy of Mozambique has thus lite-rally been worlds apart from the model economies of themultinational donor community. One can not expect aneconomy like this to respond to external shocks in the sameway as other developing countries, not to mention the econ-omies of the industrialized world.

Second, the structural adjustment programmes often take along-term view of the macroeconomy, trying to improve thegeneral structure over a number of years. In addition, manyof the recommended changes in policy have short-term ad-verse effects, and these tend to harm a population alreadystruggling with extreme poverty. For countries like Mozam-bique this may be provoking, considering the urgency ofthis country's economic and social needs. As stressed by theANC in their thoughts on macroeconomic plans for post-apartheid South-Africa, it might be time to introduce a moresignificant social profile to the restructuring efforts in Af-rica (cf. Tjdnneland [1990]). Looking back on the 1980's,the "trickle-down" effects from a general increase in econ-omic growth has not been convincing. To ensure the via-bility of economic reforms the long-term restructuringshould therefore be accommodated by appropriate short-term measures, to secure a minimum consumption level forall the population, but in particular the poor. This also re-sembles the recent recognition from the designers of reformpolicies for Russia.

Third, the effects of many of the policy measures that havebeen applied in the adjustment programmes are hamperedby the sluggishness of the macroeconomic mechanisms indeveloping countries. The simple simulations of chapter 6have illustrated how sluggish the macroeconomy may re-

spond to exogenous shocks and policy changes if economicagents respond to lagged macroeconomic indicators insteadof their updated current values. In evaluating the results ofstructural adjustment this should be taken into consider-ation. As stressed by policy designers, and by the analysisabove, it may take many years to reach the final turnaroundin a destabilized and underdeveloped economy like e.g. Mo-zambique.

Thus, future policy plans should be increasingly tailored,taking proper account of the country-specific properties ofthe economy, in addition to general features and character-istics of the macroeconomy in developing countries. Fur-ther, the feasibility of structural adjustment should be in-creasingly secured through proper short-term measures, tosoften the short-term adverse effects of structural adjust-ment, and to guarantee a minimum consumption level fortransfer-deserving households. Finally, the implementationof restructuring should be accompanied by general informa-tion, reflecting the government's commitment to the macro-economic policy. This might improve the understanding ofthe needs for structural adjustment, and reduce the generalimpatience in a population expecting real improvements intheir standards of living.

In many cases, the success of structural adjustment hingescritically on the policy-designers' ability to smooth the tran-sition to a sustainable path of development. This requires anoverview of the consequences of alternative policy plans,and optimal decision-making will therefore depend criti-cally on advisory support on as many as possible of theproblems of concern. One useful device for this type ofeconomic and political planning is a mathematical imitationof economic mechanisms, or more precise, an empirically-based model of the macroeconomy. This will make it pos-sible to estimate the quantitative effects of different policymeasures and other external shocks to the economy, andcould thus be an important piece of support to the designersof economic policy. Although a macroeconomic model willnot provide the precise answers to all problems involved bystructural adjustment, this kind of integrated framework cancontribute to a better understanding of the underlying mac-roeconomic mechanisms of a developing economy, andthereby improve the efficiency of the economic and politi-cal strategies.

43


8. Literature

Barro, R. J. and Lee, J. (1993), "International compari-sons of educational attainment", Journal of Monetary Econ-omics, vol. 32, no. 3, pp. 363-394.

Bowen, M. (1992), "Beyond Reform: Adjustment and Pol-itical Power in Contemporary Mozambique", Journal ofModern African Studies, vol. 30, no. 2, pp. 225-279.

Bruno, M. (1979) , "Stabilization and Inflation in a semi-in-dustrialized economy", in: Dombusch, Rudiger and Fren-kel, Jacob A. (eds.), International economic policy, theoryand evidence, Johns Hopkins University Press, Baltimore,pp. 270-289.

Brochmann, G. and Ofstad, A. (1990), "Mozambique:Norwegian assistance in a context of crisis", Report R 1990:4, Christian Michelsens Institutt (CMI), Bergen.

Cappelen, A. (1992), "Okonomisk utvikling i Mosambiketter ERP", (Economic Development in Mozambique underthe ERP), unpublished mimeo in Norwegian, Statistics Nor-way.

Cappelen, A., Mohn, K., Rolland, M. and Aasness, J.(1993), "Comments to: Adjustment in Sub-Saharan Africa:Progress, Payoffs, and Challenges", unpublished mimeo inNorwegian, Statistics Norway.

Cassen R. et al. (1986), Does aid work?, Clarendon Press.

Collier, P. and Gunning, J. W. (1991), "Money Creationand Financial Liberalization in a Socialist Banking System,Tanzania 1983-88", World Development, vol.19, no. 5, pp.533-538.

Conway, P. (1994), "An Atheoretic Evaluation of Successin Structural Adjustment", Economic Development and Cul-tural Change, vol. 42, no. 2, pp. 267-292.

Cornia, G., Jolly, R. and Stewart, F. (1987), Adjustmentwith a Human Face, Oxford University Press, Oxford.

De Long, B. and Summers, L. H. (1993), "How stronglydo developing economies benefit from equipment invest-ments?", Journal of Monetary Economics, vol. 32, no. 3,pp. 395-416.

Easterly, W. and Rebelo, S. (1993), "Fiscal Policy andEconomic Growth: An Empirical Investigation", Journal ofMonetary Economics, vol. 32, no. 3, pp. 417-458.

Easterly, W., Kremer, M., Pritchett, L., and SummersL. H. (1993), "Good Policy or Good Luck? CountryGrowth performance and Temporary Shocks", Journal ofMonetary Economics, vol. 32, no. 3, pp. 459-484.

Edwards, S. (1994), "The Political Economy of Inflationand Stabilization in Developing Countries", Economic De-velopment and Cultural Change, vol. 42, no. 2, pp. 235-266.

Engle, R. F. and Granger, C. W. J. (1987), "Cointegrationand Error-Correction - Estimation and Testing", Econome-trica, vol. 55, pp. 251-276.

Faini, R., de Melo, J., Senhadji, A. and Stanton, J.(1991), "Growth-Oriented Adjustment Programs: A Statisti-cal Analysis", World Development, vol. 19, no.8, pp.957-967

Fischer, S. (1992), "Growth, Macroeconomics , and Devel-opment", in: Fischer, S. and Blanchard, 0. (eds.), Macro-economics Annual 1992, pp. 329-379, National Bureau ofEconomic Research.

Fischer, S. (1993), "The Role of Macroeconomic Factors inGrowth", Journal of Monetary Economics, vol. 32, no. 3,pp. 485-512.

Fry, M. (1988), Money, Interest, and Banking in EconomicDevelopment, Johns Hopkins University Press, Baltimoreand London.

Gurley, J. G. and Shaw, E. S. (1955), "Financial Aspectsof Economic Development", American Economic Review,no. 45, september, pp. 515-538.

Gurley, J. G. and Shaw, E. S. (1960), Money in a Theoryof Finance, The Brookings Institution, Washington D.C.

Hermele, K. (1990), "Mozambican Crossroads - Econ-omics and Politics in the Era of Structural Adjustment", Re-port R 1990: 3, Christian Michelsen Institute, Bergen.

45


International Monetary Fund (1991), International Finan-cial Statistics, Yearbook 1991.

International Monetary Fund (1992), "Mozambique -Policy Framework Paper 1992-1994".

International Monetary Fund (1993a), "Mozambique -Policy Framework Paper 1994-1996".

International Monetary Fund (1993b), International Fin-ancial Statistics, Yearbook 1993.

Jarque, C. M. and Bera, A. K. (1980), "Efficient Tests forNormality, Homoskedasticity and Serial Independence ofRegression Residuals", Economic Letters, vol. 6, pp. 255-259.

King, R. G. and Levine, R. (1993a), "Finance and Growth:Schumpeter might be right", Quarterly Journal of Econ-omics, vol. 108, no. 2, may, pp. 717-737.

King, R. G. and Levine, R. (1993b), "Finance, Entrepe-neurship, and Growth: Theory and Evidence, Journal ofMonetary Economics, vol. 32, no. 3, pp. 513-542.

Krugman, Paul og Taylor, Lance, (1978) , "Contraction-ary Effects of Devaluation", Journal of International Econ-omics, vol.8, pp. 445-456.

Kyle, S. (1991), "Economic Reform and Armed Conflict inMozambique", World Development, vol. 19, no. 6, pp. 637-649.

Lucas, R. E. (1988), "On the Mechanics of Economic De-velopment", Journal of Monetary Economics, vol. 22, pp. 3-42.

Lucas, R. E. (1993), "Making a Miracle", Econometrica,vol. 61, no. 2, pp. 251-272.

McKinnon, R. I. (1973), Money and Capital in economicdevelopment, The Brookings Institution, Washington D.C.

Mohn, K. (1992), "Finanssektor og Økonomisk utvikling"(Financial sector and economic development), Report 28/92(in Norwegian), Center for Research in Economics andBusiness Administration (SNF), Bergen.

Montiel, Peter J. (1991), "The Transmission Mechanismfor Monetary Policy in Developing Countries", IMF StaffPapers, vol.38, no.1, March, pp. 83-108.

Morley, S. A. (1992), "On the Effect of Devaluation duringStabilization Programs in LDC's", Review of Economicsand Statistics, vol. no. pp. 21-28.

Pedersen, K. R. (1991), "U-hjelp - Nytter det ?", (Develop-ment aid - does it work ?), Working paper (in Norwegian),Center for Research in Economics and Business Administra-tion (SNF), Bergen.

Romer, P. M. (1986) , "Increasing Returns and Long RunGrowth", Journal of Political Economy, vol. 94, no.5, pp.1002-1037.

Romer, P. M. (1993), "Idea Gaps and Object Gaps in Econ-omic Development", Journal of Monetary Economics, vol.32, no. 3, pp. 543-574.

Shaw, E. (1973), Financial Deepening and Economic De-velopment, Oxford University Press, Oxford.

Sims, C. A. (1980), "Macroeconomics and Reality", Econ-ometrica, vol. 48, no. 1, January, pp. 1-48.

Summers, L. H. and Pritchett, L. H. (1993), "The Structu-ral Adjustment Debate", American Economic Review,Papers and Proceedings, vol. 83, no. 2, may, pp. 383-389.

Taylor, L. (1980), "IS/LM in the Tropics, Diagrammaticsof the New Structuralist Macro Critique", in Weintraub, Sid-ney and Cline, William R. (eds.), Economic Stabilization inDeveloping Countries, Brookings Institution, WashingtonD.C.

Tjunneland, E. N. (1990), Southern Africa after Apartheid,Christian Michelsen Institute, Bergen.

van Wijnbergen, S. (1984),"The Dutch Disease - A Dis-ease After All ?", Economic Journal, vol. 94, no. 373,march, pp. 41-55.

Voridis, H. (1993), "Ceilings on Interest Rates and Invest-ment: The Example of Greece", Review of Economics andStatistics, vol. 75, no. 2, may, pp. 276-283.

World Bank (1989), "Adjustment Lending: An Evaluationof Ten Years of Experience", Policy and Research Series,no. 1, Washington D.C.

World Bank (1990), "Adjustment Lending Policies for Sus-tainable Growth", Policy and Research Series, no. 14,Washington D.C.

World Bank (1992a), "Adjustment Lending and Mobiliza-tion of Private and Public Resources", Policy and ResearchSeries, no. 22, Washington D.C.

World Bank (1992b), "Mozambique - Financial SectorStudy", report no. 10269-MOZ, Southern Africa Depart-ment, Washington D.C.

World Bank (1993), "Adjustment in Sub-Saharan Africa:Progress, Payoffs, and Challenges", Policy Research Re-port, no. 2, Washington D.C.

46

Utkommet i serien Rapporter fra Statistisk sentralbyrietter 1. januar 1993 (RAPP)

Issued in the series Reports from Statistics Norwaysince 1 January 1993 (REP)

ISSN 0332-8422

Nr. 92/26 Nils Øyvind KrysslOpsdata ogkrysslOpsanalyse 1970-1990.1993-230s. 140 krISBN 82-537-3783-1

- 92/29 Charlotte Koren og Tom Kornstad:Typehusholdsmodellen ODIN.1993-Ms. 75 kr ISBN 82-537-3797-1

- 93/1 Naturressurser og miljo 1992.1993-144s. 115 krISBN 82-537-3844-7

93/IA Natural Resources and the Environment1992. 1993-154s. 115 krISBN 82-537-3855-2

93/2

93/3

93/4

- 93/5

- 93/6

- 93/7

- 93/8

93/9 Erling Holmoy, Bodil M. Larsen ogHaakon Vennemo: Historiskebrukerpriser pa realkapital.1993-63s. 90 kr ISBN 82-537-3832-3

Nr. 93/10 Runa Nesbakken og Steinar Strom:Energiforbruk til oppvarmingsformal ihusholdningene. 1993-4Is. 75 krISBN 82-537-3836-6

93/11 Bodil M. Larsen: Vekst og produktivitet iNorge 1971-1990.1993-44s. 75 kr ISBN 82-537-3837-4

- 93/12 Resultatkontroll jordbruk 1992. Tiltak motavrenning av nxringssalter og jorderosjon1993-79s. 90 krISBN 82-537-3835-8

93/13 Odd Frank Vaage: Mediebruk 1992.1993-38s. 75 kr ISBN 82-537-3854-4

- 93/14 Kyrre Aamdal: Kommunal ressursbruk ogtjenesteyting Makromodellen MAKKO.1993-94s. 100 krISBN 82-537-3857-9

- 93/15 Olav Bjerkholt, Torgeir Johnsen og KnutThonstad: Muligheter for en bwrekraftigutvikling Analyser pa World Model. 1993-64s. 90 krISBN 82-537-3861-7

93/16 Tom Langer Andersen, Ole Tom Djupskasog Tor Arnt Johnsen: Kraftkontrakter tilalminnelig forsyning i 1992 Priser,kvantum og leveringsbetingelser. 1993-42s.75 krISBN 82-537-3864-1

- 93/17 Steinar Strom, Tom Wennemo og RolfAaberge: Inntektsulikhet i Norge 1973-1990. 1993-99s. 100 krISBN 82-537-3867-6

93/18 Kjersti Gro Lindquist: Empirical Modellingof Exports of Manufactures: Norway 1962-1987. 1993-124s. 100 krISBN 82-537-3869-2

- 93/19 Knut Roed: Den selvforsterkendearbeidsledigheten Om hystereseeffekter iarbeidsmarkedet. 1993-95s. 90 krISBN 82-537-3870-6

93/20 Dag Kolsrud: Stochastic Simulation ofKVARTS91. 1993-70s. 95 krISBN 82-537-3952-4

- 93/21 Sarita Bartlett: The Evolution ofNorwegian Energy Use from 1950 to1991. 1993-142s. 100 krISBN 82-537-3890-0

Nr. 93/22

Klaus Mohn: Industrisysselsetting ogproduksjonsteknologi i norske regioner.1993-59s. 90 krISBN 82-537-3910-9

- 93/23 Torbjorn Eika: Norsk okonomi1988-1991: - Hvorfor steg arbeids-ledigheten sa mye? 1993-38s. 75 krISBN 82-537-3912-5

Anne Brendemoen: Faktorettersporsel itransportproduserende sektorer.1993-49s. 75 kr ISBN-82-537-3814-5

Jon Holmoy: Pleie- og omsorgs- tjenesten ikommunene 1989.1993-136s. 100 krISBN 82-537-3811-0

Magnar Lillegard: Folke- og boligtelling1990 Dokumentasjon av de statistiskemetodene. 1993-48s.90 kr ISBN 82-537-3818-8

Audun Langorgen: En okonometriskanalyse av lonnsdannelsen i Norge.1993-48s. 100 kr ISBN 82-537-3819-6

Leif Andreassen, Truls Andreassen, DennisFredriksen, Gina Spurkland og YngveVogt: Framskriving av arbeidsstyrke ogutdanning MikrosimuleringsmodellenMOSART. 1993-100s. 100 krISBN 82-537-3821-8

Dennis Fredriksen og Gina Spurkland:Framskriving av alders- og uforetrygdved hjelp av mikrosimulerings-modellen MOSART. 1993-58s. 90 krISBN 82-537-3945-1

Odd Frank Vaage: Feriereiser 1991/92.1993-44s. 75 krISBN 82-537-3831-5

47

Nr. 93/24 Kristin Rypdal: Anthropogenic Emissionsof the Greenhouse Gases CO 2 . CH4 andN20 in NorwayA Documentation of Methods ofEstimation, Activity Data andEmission Factors. 1993-65s. 90 krISBN 82-537-3917-6

93/25 Skatter og overfainger til private Historiskoversikt over satser my. Arene 1975-1993.1993-75s. 90 krISBN 82-537-3922-2

93/26 Thor Olav Thoresen: Fordelings-virkninger av overfOringene tilbarnefamilier Beregninger vedskattemodellen LOTTE. 1993-42s.75 kr ISBN 82-537-3923-0

93/27 Odd Frank Vaage: Holdninger til norskutviklingshjelp 1993.1993-41s.75 kr ISBN 82-537-3931-1

93/28 Kjetil Bofasthet, flytting ogutdanningsniva i kommunene Atte arskullfulgt gjennom aldersfasen15-35 ar Del 1: Ostlandet. 1993-174s. 115kr ISBN 82-537-3935-4

93/29 Kjetil SOrlie: Bofasthet, flytting ogutdanningsniva i kommunene Atte arskullfulgt gjennom aldersfasen15-35 ar Del 2: Sorlandet ogVestlandet. 1993-179s. 115 krISBN 82-537-3936-2

93/30 Kjetil Sortie: Bofasthet, flytting ogutdanningsniva i kommunene Atte arskullfulgt gjennom aldersfasen15-35 ar Del 3: Trondelag og Nord-Norge.1993-165s. 115 krISBN 82-537-3937-0

Nr. 93/31

Erling Holm0y, TorbjØrn Hxgeland,Oystein Olsen og Birger StrOm: Effektivesatser for nxringsstotte. 1993-178s. 115 krISBN 82-537-3947-8

- 94/1 Torstein Bye, Adne Cappelen, TorbjØrnEika, Eystein Gjelsvik og Oystein Olsen:Noen konsekvenser avpetroleumsvirksomheten for norsk Økonomi.I994-54s. 95 krISBN 82-537-3956-7

- 94/2 Wenche Drzwi, Lisbeth Lerskau, OysteinOlsen og Nils Martin StOlen: Tilbud ogetterspOrsel etter ulike typer arbeidskraft.1994-56s. 95 krISBN 82-537-3950-8

- 94/3 Hilde-Marie Brans Zakariassen:Tilbud av arbeidskraft i Norge En empiriskanalyse pa kvartalsdata for perioden 1972 til1990. Under utgivelse

- 94/4 Resultatkontroll jordbruk 1993 Tiltak motavrenning av nxringssalter og jorderosjon.Under utgivelse

- 94/5 Haakon Vennemo: A Growth Model ofNorway with a Two-way Link to theEnvironment. 1994-57s. 95 krISBN 82-537-3985-0

- 94/6 Odd Frank Vaage: Feriereiser 1992/93.1994-49s. 80 kr ISBN 82-537-3983-3

- 94/7 Magnar Lillegard: Prisindekser forboligmarkedet. Under utgivelse-

- 94/8 Grete Dahl, Else Flittig og Jorunn Lajord:Inntekt, levelcar og sysselsetting forpensjonister og stOnadsmottakere ifolketrygden.Under utgivelse

48

CReturadresse:Statistisk sentralbyraPostboks 8131 Dep.N-0033 Oslo

Publikasjonen kan bestilles fra:

Statistisk sentralbyr5SaIg- og abonnementservicePostboks 8131 Dep.N-0033 Oslo

Telefon: 22 86 49 6422 86 48 87

Telefaks: 22 86 49 76

eller:Akademika - avdeling foroffentlige publikasjonerMollergt. 17Postboks 8134 Dep.N-0033 Oslo

Telefon: 22 11 67 70Telefaks: 22 42 05 51

ISBN 82-537-4005-0ISSN 0332-8422

Pris kr 80,00

2CST)

C

3(t.

0NCU39".CCD

Statistisk sentralbyràStatistics Norway

9 882 7 8

Documents

SSB - Monetarism and Structural Adjustment - The Case of ...Strtrl Adjtnt n Mzb pprtr 9/11 h frth hptr ntn rtrpt n pt-nd-pndn n pl n Mzb h frt prd ftr 1975 hrtrzd b ntrll plnnd n-th