Investment in emerging market economies

Empir Econ (2012) 43:97–119DOI 10.1007/s00181-011-0457-0

Investment in emerging market economies

Tuomas A. Peltonen · Ricardo M. Sousa ·Isabel S. Vansteenkiste

Received: 4 April 2010 / Accepted: 15 December 2010 / Published online: 18 March 2011© Springer-Verlag 2011

Abstract This article empirically models investment in emerging economies. Usingdynamic panel estimation methods and quarterly data for 31 emerging economies forthe period 1990:1–2008:3, we show that (i) the GDP and the cost of capital are the keyfundamental determinants of investment; (ii) financial factors (such as equity prices,credit and lending rate) play a relevant role on the dynamics of investment, in particular,for Asian and Latin American countries; (iii) investment growth exhibits substantialpersistence and (iv) crises episodes magnify the negative response of investment.

Keywords Investment · Credit · Asset prices · Emerging markets

JEL Classification E22 · E44 · D24

Ricardo Sousa would like to thank the International Policy Analysis Division of the ECB for its hospitality.

T. A. Peltonen · I. S. VansteenkisteEuropean Central Bank, Kaiserstraße 29, 60311 Frankfurt am Main, Germanye-mail: [email protected]

I. S. Vansteenkistee-mail: [email protected]

R. M. Sousa (B)London School of Economics, Financial Markets Group (FMG), Houghton Street,London WC2 2AE, UKe-mail: [email protected]

R. M. SousaDepartment of Economics and Economic Policies Research Unit (NIPE),University of Minho, Campus of Gualtar, 4710-057 Braga, Portugale-mail: [email protected]

123

98 T. A. Peltonen et al.

1 Introduction

There exists a broad consensus that a country’s long-run economic performance isstrongly determined by private investment. In addition, it is pivotal to a country’seconomic growth, employment situation and foreign exchange position. For emerg-ing market economies, investment is particularly relevant as it contributes to theircatching-up process with advanced countries.

While, broadly speaking, there seems to exist a consensus among theorists andpractitioners on modelling aggregate consumption, there is no equivalent agreementon modelling aggregate investment. In fact, two distinct types of models for aggre-gate investment compete in the literature: (i) the traditional neoclassical model, i.e.the Jorgenson (1963) approach and (ii) the alternative Q approach by Tobin (1969).According to the neoclassical model, output and the cost of capital are the major causaldeterminants of investment. In the alternative Q approach, investment is mainly deter-mined by the Tobin’s Q ratio, i.e. the ratio of the market valuation of a firm’s securitiesto the replacement cost of the physical assets they represent.

In the case of emerging market economies, the neoclassical flexible-acceleratormodel has been the most popular approach, whereby investment is typically consid-ered to be negatively related to the cost of capital and positively related with the rateof growth of real output per capita. Nevertheless, testing the neoclassical model hasbeen difficult because key assumptions (such as perfect capital markets and little orno government investment) are generally inapplicable and data for certain variables(capital stock, real wages and real financing rates for debt and equity) are normallyunavailable.

Accordingly, research has proceeded in several directions. While these efforts havenot yet produced a comprehensive model of investment behaviour in emerging mar-ket economies, they led to a growing literature on the effects of financial factors oninvestment.1

One of such factors is the interest rate, and the notion that business spending onfixed capital falls when interest rates rise is a theoretically unambiguous relationshipthat lies at the heart of the monetary transmission mechanism.

Another financial factor which has been seen as affecting private investment is theaccess to credit (Fazzari et al. 1988; Whited 1991). Firms, especially in emerging mar-ket economies, may face binding financial constraints in domestic capital markets dueto endogenous credit rationing or limited banks’ capacity to extend new loans (Stiglitzand Weiss 1981). The importance of these factors is confirmed in studies which per-sistently postulate that private investment expenditure is constrained by inefficiency

1 Apart from the ‘fundamental’ and ‘financial’ factors, other studies have identified several additionalexplanatory variables playing a role in private investment, namely (i) public investment (Blejer and Khan1984); (ii) the domestic inflation rate (Dornbusch and Reynoso 1989); (iii) large external debt burdens(Borensztein 1990); (iv) income per capita; (v) exchange rate volatility (Serven 2003); (vi) the quality ofpolitical institutions (Agnello and Sousa 2009); (vii) corporate tax policy (Auerbach 1983) and (viii) mea-sures of natural resource endowments (Papyrakis and Gerlagh 2004; Johnson and Soenen 2009). Howeverand not surprisingly, these are hard to quantify and quite time invariant, and are unlikely to capture the richdiversity in institutional arrangements that exists, particularly, in developing countries.

123

Investment in emerging market economies 99

of the banking sector and the presence of repression in financial markets (Shaw 1974),and the (lack of) availability of financial resources (Tybout 1983).

Finally, there exists an empirical literature on the role played by equity prices ininfluencing investment behaviour. In fact, in the presence of information asymmetriesin financial markets, a rise in share prices will improve the balance sheet position ofthe firm and, thereby, the ability of the firm to directly fund projects or to provide col-lateral for external finance (Fischer and Merton 1984) and/or to raise external funding(Bernanke and Gertler 1986). Moreover, if the role of management is to maximise thewealth of shareholders, then it should respond to changes in market valuation as theyreflect the long-run market value of the firm (Bosworth 1975).

Despite the unquestionable role of investment and the widespread recognition thatit is a key element into generating economic growth, there is surprisingly little researchfor emerging market economies, a fact that cannot be detached from the scarcity ofdata. Moreover, the distinction between fundamental and financial factors explaininginvestment remains crucial and has not been fully explored yet.2

The major goal of this article is, therefore, to uncover the determinants of investmentin emerging market economies. We do so by improving and extending the existingliterature in several directions. First, we look not only at the effects of the traditionalvariables that capture the fundamentals of investment (such as the GDP and the costof capital), but also consider its financial determinants (namely, the lending rate, theamount of credit to the private sector and the equity price). In doing so, we build an‘encompassing’ model that can ultimately test the validity of the neoclassical theoryof investment and the Q approach. Second, we use data at a high frequency, that is,quarterly data from 1990:1 to 2008:3, and are, therefore, able to obtain more preciseestimates of the impact of the various explanatory variables on investment. Third, weassess the robustness of the results to different estimation methods, sample selectionor the inclusion of additional control variables, such as the inflation rate, the publicinvestment, the external debt, the commodity prices and the corporate spread.

Using a panel of 31 emerging economies, we show that GDP and the cost of cap-ital are among the fundamental determinants of investment in accordance with theneoclassical model. The estimated effects are economically sizeable and statisticallysignificant: (i) a 10% rise in real GDP increases investment by 14% in the short-runand (ii) a 10% increase in the cost of capital reduces investment by 3%.

In addition, we find that financial factors play an important role: (i) a 10% rise inequity price increases investment 0.2% and (ii) a 100 basis point increase in real lend-ing rate reduces investment by 0.1%. However, we find that credit is not a statisticallysignificant factor for investment in the full sample.

The regional analysis suggests that while credit availability and equity markets seemto be the main financing sources of investment in emerging Asia, for Latin America,the banking sector and the capital markets are the drivers among financial factors. Asfor emerging Europe, the empirical evidence shows that financial determinants areless relevant.

2 In the case of private consumption, Peltonen et al. (2009) show that income and financial variables areinfluential in emerging markets.

123


Controlling for the degree of financial development, we find that the elasticity ofinvestment with respect to output and lending rate is larger for countries with lowstock market capitalization (in percentage of GDP), whereas the credit elasticity ofinvestment is stronger in countries with a high stock market capitalization.

Finally, the results suggest that investment growth exhibits a substantial persistenceand responds sluggishly to shocks. This may be an important reason for concern—particularly, in the case of an economic downturn—taking into account that theseeconomies have often witnessed episodes of economic, financial and currency crises.In fact, despite the small short-run elasticity of investment to equity or lending rate,the empirical findings show that both the short-run and the long-run elasticities ofinvestment with respect to GDP are quite large. As a result, the effects of a slowdownof the economic activity may be amplified by this intrinsic characteristic of investmentin emerging markets.

The rest of the article is organized as follows. Section 2 presents the estimationmethodology. Section 3 describes the data. Section 4 discusses the empirical results.Finally, Sect. 5 concludes with the main findings and policy implications.

2 Estimation methodology

The empirical model for the estimation of investment can be summarized as follows:

log Ii,t = β0 log Ii,t−1 + log X′j i,tβ1 j + log Y

′j i,tβ2 j + vi + εi,t (1)

with i = 1, . . . , N t = 1, . . . , Ti j = 1, . . . , K

where Ii,t stands for the investment of country i at time t , X j i,t is a vector of j ‘fun-damental’ determinants, Y j i,t is a vector of j ‘financial’ explanatory variables, the βsare parameters to estimate, vi are country-specific effects and εi,t is the error term. Inthe set of fundamental determinants, we include the GDP and the cost of capital. Inaddition, the equity prices, the amount of credit to the private sector, the lending rateand the corporate bond spread are considered among the financial factors impactingon private investment.

When model (1) is estimated using ordinary least squares (OLS), substantial com-plications arise. In fact, in both the fixed and random effects settings, the lagged depen-dent variable is correlated with the error term, even if we assume that the disturbancesare not themselves autocorrelated.

Arellano and Bond (1991) developed a generalized method of moments (GMM)estimator that solves the problem referred above, allowing one to eliminate coun-try-specific effects or any time invariant country-specific variable. In addition, it alsotackles the endogeneity issue that may be due to the correlation of the country-specificeffects and the independent variables. Consequently, first differencing (1) removes vi ,

and produces an equation estimable by instrumental variables:

� log Ii,t = β0� log Ii,t−1 + � log X′j i,tβ1 j + �log Y

′j i,tβ2 j + �εi,t (2)

with i = 1, . . . , N t = 1, . . . , Ti j = 1, . . . , K

123


where � is the first difference operator, while the variables and parameters are definedas in (1). Following Holtz-Eakin et al. (1988), one can instrument the differencedpre-determined and endogenous variables with their available lags in levels: levels ofthe dependent and endogenous variables, lagged two or more periods; levels of thepre-determined variables, lagged two or more periods. The exogenous variables canbe used as their own instruments.

A problem of this difference-GMM estimator is that lagged levels are weak instru-ments for first-differences if the series are very persistent and efficiency can be in-creased by adding the original equation in levels to the system (Blundell and Bond1998). If the first-differences of explanatory variables are uncorrelated with the indi-vidual effects, both lagged values of the first-differences of the explanatory variablesand of the dependent variable can be used as instruments in the equation in levels. Inthis case, the estimation combines the set of moment conditions available for the first-differenced equations with the additional moment conditions implied for the levelsequation. Blundell and Bond (1998) show that this system GMM estimator is prefer-able and, for this reason, the current article also uses that estimation methodology.

3 Data

The dataset consists of an unbalanced panel of 31 main emerging economies, 10 fromemerging Asia (China, Hong Kong, India, Indonesia, Korea, Malaysia, Philippines,Singapore, Taiwan and Thailand), 6 from Latin America (Argentina, Brazil, Chile,Colombia, Mexico and Peru), 12 from emerging Europe (Bulgaria, Croatia, CzechRepublic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Slovakiaand Slovenia) and 3 other countries (Israel, South Africa and Turkey).

In what concerns the time span of the dataset, we use quarterly data for 1990:1–2008:3 (where available). The main source for the National Accounts data is HaverAnalytics: Investment (Gross Fixed Capital Formation at constant prices), and GDP(at constant prices) that is used to proxy for economic activity and the business cycle.Moreover, the cost of capital is proxied using the ratio of investment deflator to GDPdeflator.

Regarding financial variables, stock price indices (composite indices) are obtainedfrom Haver Analytics and Global Financial Database (Argentina, Chile, Colom-bia, Croatia, Czech Republic, Hong Kong, Israel, Korea, Peru, Philippines, Russia,Singapore and South Africa). The availability of credit is proxied by claims on privatesector, which is provided by the IMF International Financial Statistics (IFS). The inter-est rate available to firms is proxied by lending rate or the interbank rate (Romania andTurkey) also from the IMF IFS. Finally, the data for the additional regressions in thesensitivity analysis are obtained as follows: inflation is measured by the annual changeof the consumer price index (all items) from Haver Analytics, external debt is mea-sured as international debt securities by the Bank for International Settlements (BIS),3

3 International debt securities cover bonds and notes (long-term issues) and money market instruments(short-term issues) placed in international markets by all issuers (i.e. government, corporations and financialinstitutions). They comprise all foreign currency issues by residents and non-residents in a given countryand all domestic currency issues launched in the domestic market by non-residents. In addition, domestic

123


public investment by government consumption expenditure (at constant prices) fromHaver Analytics and commodity prices by Reuters/Jefferies CRB all commodities fu-tures price index. To proxy for the interest rate at which firms are able to raise capitalin the international capital markets, we use the CEMBI spread from JP Morgan.

Over the estimation sample, emerging economies have faced several severe eco-nomic and financial crisis that may have influenced investment. In order to controlfor this, we construct a crisis dummy that takes the value of 1 if either the change(year-on-year) of real GDP or real equity price index is more than two times thecountry-specific standard deviation of the variable, and 0, otherwise. In addition, thequarters immediately before and after the peak of the crisis are also marked with 1,and all other (normal) periods are marked with 0.

For the econometric analysis, data are transformed in several ways. First, the finan-cial variables are deflated using the GDP deflator, with the exception of Singapore,where the CPI index (all items) is used. Second, data on real GDP, real investment andthe corresponding deflator for China are annual, and, therefore, we interpolate themusing a cubic spline conversion method. In addition, the following missing data pointsare also interpolated using cubic spline conversion method: credit (Hong Kong, 1990–1993; South Africa 1991:3–1991:4) and lending rate (Argentina, 2002:2). 4 Third, thefollowing variables are seasonally adjusted using the X11 ARIMA procedure:5 grossfixed capital formation at constant prices (India, Korea, Mexico and Romania), grossfixed capital formation at current prices (India and Korea), GDP at constant prices(Korea and Romania), GDP at nominal prices (Korea), and claims on private sector(all countries).

Table A.1 in Appendix A provides a detailed description of the variables and datasources used in the analysis, while Table A.2 provides the sample period and thenumber of observations per country.6

4 Empirical results

In this section, we present the estimation results from the dynamic panel as defined inEq. 2 of Sect. 2. The section is divided in two subsections with the main results pre-sented in Sect. 4.1 and the sensitivity analysis in Sect. 4.2. In the benchmark model,the set of explanatory variables includes: the lag of investment, gross domestic prod-uct, cost of capital, equity prices, domestic private credit and lending rate, a dummyvariable for economic/financial crises, and a constant.

Footnote 3 continuedcurrency issues launched in the domestic market by residents are also considered as international issues ifthey are specifically targeted at non-resident investors.4 The cubic spline interpolation method is well suited to the data characteristics and is commonly usedin the literature. Nevertheless, the estimation results are qualitatively similar to the ones based on a linearinterpolation.5 The other series are seasonally adjusted either by the national source or by Haver Analytics.6 The sample period refers to the period over which all explanatory variables for a given country areavailable.

123


The results for the full sample, as well as for three subsamples, namely, emergingAsia, emerging Europe and emerging Latin America countries are presented in Tables1 and 2. In the sensitivity analysis, we add the following variables to the benchmarkmodel: inflation rate, public investment, external debt, commodity price index andcorporate bond spread. In addition, we estimate the models for two equally sizedsubsamples according to their stock market capitalization per GDP to proxy for thelevel of financial development. The results for the sensitivity analysis are presented inTable 3. Moreover, we assess the robustness of the results by applying several econo-metric technics to Eq. 2, namely: (i) the pooled OLS (dynamic OLS) estimator; (ii)the pooled OLS estimator with time effects;7 (iii) the pooled OLS with both timeand country effects; (iv) the fixed effects estimator; (v) the random effects estimatorand (vi) the IV/GMM estimator. In the cases of the IV/GMM and the Blundell andBond (1998) estimators, the crisis dummy and a constant are considered to be strictlyexogenous, while the remaining ones are included in the set of endogenous variables.The empirical results for the other estimation methods are presented in Tables B.1,B.2, B.3, B.4, B.5, B.6, B.7, B.8 of Appendix B.

4.1 Main results

Tables 1 and 2 show that the lag of investment is statistically significant for all specifi-cations, reflecting a strong persistence of investment growth and its sluggish responseto shocks (the short-run elasticity being 0.55). Investment smoothing is in line withconvex adjustment models of investment and a recurring feature in empirical invest-ment models.

Besides lagged investment growth, we also show that GDP growth and the cost ofcapital are statistically significant. In the case of output growth, we find a positiverelation with the short-run elasticity being 1.36 and the long-run elasticity 3.04. Boththe short-run and the long-run elasticities of investment with respect to GDP are eco-nomically substantial: a 10% increase in real GDP would imply a 14% increase ininvestment in the short-run, and a 30% increase in the long-run. As a result, the effectsof a slowdown of the economic activity may be amplified by this intrinsic characteris-tic of investment in emerging markets. In contrast, we find that an increase in the costof capital reduces investment growth with an estimated short-run elasticity of −0.27and a long-run elasticity of −0.61. Overall, such findings are consistent with the neo-classical model of investment and is generally supported by the empirical literature(Blomstrom et al. 1996).

In addition to lagged investment and the cost of capital, which are variables thathave been applied earlier both to advanced and emerging market economies alike, wealso include a group of financial variables that are seen to be relevant for investmentequations for emerging market economies.

Private sector credit is a potentially important determinant of investment growth.In this context, King and Levine (1993) show that investment and the share of creditallocated to the private sector are positively correlated. In addition, credit availability

7 “Time effects” refer to time-specific dummy variables for each period.

123


Table 1 Dynamic panel regressions: short-run elasticities for full sample, Asia, Europe and Latin America

Full sample Asia Europe Latin America

L. Investment 0.5525*** 0.5428*** 0.4605*** 0.4177***[0.0320] [0.0457] [0.0777] [0.0610]

GDP 1.3641*** 1.0415*** 1.3759*** 1.9401***

[0.1718] [0.1864] [0.2648] [0.2158]

Cost of capital −0.2736*** −0.0605 −0.2679*** −0.1091**

[0.1027] [0.1070] [0.0918] [0.0484]

Equity 0.0211 0.0237* 0.0134 0.0315***

[0.0172] [0.0133] [0.0122] [0.0068]

Credit −0.0453 0.0821* 0.0400 −0.1116*

[0.0445] [0.0448] [0.0248] [0.0664]

Lending rate −0.0014*** 0.1481*** 0.0550 −0.0015***

[0.0003] [0.0517] [0.0339] [0.0003]

Crisis −0.0257* −0.0714*** 0.0320** −0.0738***

(Econ+Fin) [0.0136] [0.0197] [0.0146] [0.0189]

Constant −0.0353*** −0.0330*** −0.0410*** −0.0364***

[0.0064] [0.0080] [0.0103] [0.0050]

Observations 1,469 527 439 344

# Of countries 31 10 12 6

Sargan p value 0.00 0.00 0.14 0.00

Hansen p value 0.99 0.99 0.99 0.99

AR2 p value 0.02 0.29 0.05 0.13

Note: Heteroscedasticity and serial correlation robust standard errors in brackets. All series are in log dif-ferences∗ Statistically significant at 10% level; ∗∗ at 5% level; ∗∗∗ at 1% level

Table 2 Dynamic panel regressions: long-run elasticities for full sample, Asia, Europe and Latin America

Full sample Asia Europe Latin America

GDP 3.0483∗∗∗ 2.2780∗∗∗ 2.5503∗∗∗ 3.3318∗∗∗Cost of capital −0.6114∗∗∗ −0.1323 −0.4966∗∗∗ −0.1874∗∗Equity 0.0472 0.0518∗ 0.0248 0.0541∗∗∗Credit −0.1012 0.1796∗ 0.0741 −0.1917∗Lending rate −0.0031∗∗∗ 0.3239∗∗∗ 0.1019 −0.0026∗∗∗Note: ∗ Statistically significant at 10% level; ∗∗ at 5% level; ∗∗∗ at 1% level

was also shown to be an important determinant of private investment (Oshikoya 1994).Despite this, the results do not confirm these findings. In all cases, the credit variableis not statistically significant. This feature may reflect the fact that credit growth alsoincludes credit to the household sector (and not just to the corporate sector).

In contrast, the lending rate is found to be statistically significant in all estimatedmodels with an estimated short-run elasticity of −0.0014 and a corresponding long-run

123


Table 3 Dynamic panel regressions: results for models with additional regressors and for samples for highvs. low stock market capitalisation

Additional Additional High stock market Low stock marketregressors 1 regressors 2 capilisation capilisation

L. Investment 0.5193*** 0.6090*** 0.5376*** 0.4486***

[0.0279] [0.0361] [0.0428] [0.0552]

GDP 1.4479*** 0.9256*** 1.1627*** 1.8083***

[0.1673] [0.1854] [0.1907] [0.2729]

Cost of capital −0.2864*** 0.0069 −0.121 −0.2691**

[0.0793] [0.0965] [0.0766] [0.1120]

Equity 0.0114 0.0444*** 0.0210* 0.0123

[0.0121] [0.0136] [0.0110] [0.0171]

Credit −0.0208 −0.0434 0.0343 −0.0638

[0.0443] [0.0668] [0.0449] [0.0450]

Lending rate −0.0395 −0.1214** −0.0016*** −0.0527*

[0.0312] [0.0585] [0.0003] [0.0270]

Inflation 0.0597 0.2037

[0.0374] [0.1325]

Public investment 0.0647 −0.004

[0.1046] [0.0837]

External debt 0.0009 −0.0296*

[0.0054] [0.0177]

Commodity 0.0071 −0.0088

[0.0187] [0.0294]

Corporate spread 0.062

[0.0510]

Crisis −0.0218 −0.0794*** 0.0246

(Econ+Fin) [0.0162] [0.0191] [0.0203]

Constant −0.0445*** −0.0271*** −0.0314*** −0.0447***

[0.0078] [0.0095] [0.0063] [0.0104]

Observations 1,215 346 850 619

# Of countries 30 19 15 16

Sargan p value 0.00 0.00 0.00 0.00

Hansen p value 0.99 0.99 0.99 0.99

AR2 p value 0.03 0.52 0.10 0.13


elasticity of −0.0031. This implies that a 100 basis point increase in real lending ratedecreases private investment by 0.1% in the short-run, and by 0.3% in the long-run.Despite the estimated elasticities are economically rather small, they highlight theimportance of the link between monetary policy and real activity and the role of bank

123


lending channel in the monetary policy transmission mechanism, and are in accordancewith the works of Tybout (1983) and Bernanke and Gertler (1986).

Finally, we find that stock prices have a positive impact on investment with ashort-run elasticity of 0.021 and a long-run elasticity of 0.0472. However, the esti-mated coefficients are not statistically significant. These results suggest a relativelysmall economic impact of the equity market on investment, in line with the findingof Morck et al. (1990)). Similarly, they are consistent with the work of Blanchard etal. (1993), who argue that it may be optimal for the firm to respond to fluctuations instock prices by simply restructuring its financing patterns without altering investment.Moreover, they give rise to the idea that large and liquid stock markets can facilitatepooling of savings and channel capital to investment projects (Greenwood and Smith1997).

Besides the results for the full sample of emerging market economies, we alsoestimate the same models for subsamples where we consider countries from threeregions: (i) emerging Asia; (ii) Latin America and (iii) emerging Europe. In general,the empirical evidence confirms the importance of the lagged investment and outputgrowth in shaping aggregate investment growth. Whereas investment process seemsto be most persistent in emerging Asia (with a short-run elasticity of 0.5), it is the leastpersistent in Latin America (with a short-run elasticity of 0.4). Regarding sensitivityto output changes, we find that the estimated elasticities with respect to real GDP arethe highest in Latin America (a short-run elasticity of 1.94 and long-run elasticityof 3.33) and the lowest in emerging Asia (an estimated short-run elasticity of 1.04and a long-run elasticity of 2.28). Moreover, in emerging Europe and Latin America,the cost of capital has a statistically significant and economically important negativeimpact on investment (the short-run elasticities being −0.27 and −0.11 and −0.50and −0.19 for the long-run elasticities, respectively). By contrast, in Emerging Asia,the cost of capital is not statistically significant. All in all, one can conclude that vari-ables derived from the neoclassical model seem to be useful in modelling investmentprocess, especially, in emerging Europe.

Financial factors seem to play a more important role in the other regions anda higher fit can be achieved by encompassing the model that explains investmentbehaviour in a richer framework structure. In emerging Asia, credit growth andchanges in the stock market price are statistically significant and with a positivesign. Regarding credit, the estimated short-run elasticity is 0.08 and the long-runelasticity is 0.18, implying a rather significant influence of the credit channel in theinvestment dynamics of the region. The estimated elasticities for the equity price(0.024 in the short-run and 0.052 in the long-run) suggest a much smaller impacton investment. In Latin America, the equity market is a stronger determinant ofinvestment growth, with the estimated short-run elasticity being 0.03 (or 0.054 inthe long-run), i.e. significantly higher than in emerging Asia or emerging Europe.Moreover, the lending rate has a statistically significant negative impact on invest-ment growth in Latin America: the estimated short-run elasticity is −0.0015 (long-run−0.0026), which implies that the impact of lending rates on investment is small inmagnitude.

Finally, as regards to the investment sensitivity crisis dummy, it can be seen thatthe effect is negative (and statistically significant) in both emerging Asia and Latin

123


America, with a short-run elasticity of around −0.07 in both cases. However, in emerg-ing Europe, the estimated elasticity is positive, with the a short-run elasticity of 0.03.Such results suggest that contagion effects and volatility spillovers play an importantrole in the transmission of economic, financial and currency crises to private invest-ment (Caporale et al. 2005). They also reflect the strength of the wealth dynamicsassociated to such crisis episodes (Sousa 2010a) and the nexus between monetarystability and financial stability (Mallick and Mohsin 2010; Sousa 2010b).

4.2 Sensitivity analysis

The results for the sensitivity analysis are presented in Table 3 for models with addi-tional regressors (Columns 1 and 2), as well as for the subsamples based on the levelof financial development (Columns 3 and 4).

One factor which has been seen as relevant is the domestic inflation rate. Indeed,high rates of domestic inflation may adversely affect investment by (i) increasingthe riskiness of longer-term investment projects; (ii) reducing the average maturityof commercial lending and (iii) distorting the information content of relative prices.In addition, they can also be considered an indicator of macroeconomic instabilityand the country’s inability to conduct sound macroeconomic policies (Dornbusch andReynoso 1989).

Similarly, public investment or public infrastructure provision can influence privateinvestment in developing countries, although its effect might be ambiguous (Blejer andKhan 1984). On the one hand, public investment may be complementary to (and, thus,support) private investment. On the other hand, it may detract from private investmentactivity to the extent that it substitutes or generates important “crowding-out” effects.This may occur when the investment involves semigovernmental enterprise producinggoods that compete with the private sector, when heavy spending for public capitalprojects leads to high interest rates, severe credit rationing, heavier current or future taxburdens (Aschauer 1989), or when it impacts on long-run fiscal sustainability (Gabrieland Sangduan 2010).

The presence of large external debt burdens has also been suggested as a factorreducing investment activity (Borensztein 1990). First, it implies a high debt-servicepayment, which reduces the incentives for investment as much of the future returnsmust be repaid. Second, it lowers the amount of funding that a country can obtainthrough trade, because it make it harder or more costly to finance private investment.

Finally, measures of natural resource endowments, in particular, commodity pricescan play a critical role in emerging markets’ investment (Papyrakis and Gerlagh 2004;Johnson and Soenen 2009).

Column 1 presents the results for models where we add the abovementioned vari-ables, namely inflation, public spending, external debt and commodity prices, to theset of explanatory variables included in the benchmark model.8 Overall, we find thatthe inclusion of these additional variables does not qualitatively alter the benchmark

8 The sample used in the sensitivity analysis includes only 30 countries (therefore, excluding Taiwan) andhas 254 observation less as the additional variables are not available.

123


results. As before, the neoclassical model seems to be a good description of invest-ment in emerging markets, as the estimated coefficients for lagged investment, realGDP and cost of capital are highly statistically significant with expected signs. More-over, both the equity price and the lending rate remain important determinants ofinvestment, which emphasizes the need to include these financial variables in thespecification.

Another financial variable that we consider is the corporate spread (CEMBI). Theuse of the bond prices (as opposed to equity prices) was recently proposed by Gilchristand Zakrajsek (2007), who, using data for the US, suggest that they may be a more suit-able proxy for the Tobin’s Q. In fact, despite being less liquid, the bond market mightbe less susceptible to bubbles than the equity market. In addition, different degrees ofasymmetric information, market segmentation, and heterogeneity in adjustment costsand stochastic processes can explain such choice.

In Column 2, we report the results where the corporate spread is included as apotential determinant of investment. The empirical evidence shows that this variableis not statistically significant. This can be explained by the very limited availabilityof the time series for the CEMBI yield (which reduces our sample period to merely 5years and 19 countries and also coincides with a period of rapid increase in corporatespreads). Moreover, the bond market development is still at an initial stage in manyemerging market economies, and may hence have a negligible impact on aggregatebusiness investment.

Furthermore, we split the sample in two subsamples based on the level of financialdevelopment, which is proxied by the ratio of stock market capitalization to GDP.9,10

Columns 3 and 4 summarize the empirical findings. The following interesting observa-tions arise. First, the reaction of investment to equity prices is roughly the same acrosssubsamples. Second, the estimated elasticities of investment with respect to the lendingrate are higher in absolute terms in the case of countries with low stock market cap-italization, whereas the sensitivity of investment to credit is positive and statisticallysignificant in the high stock market capitalization sample. Third, the elasticity withrespect to output is larger for countries with a low stock market capitalization. Thisimplies that firms in less developed financial systems are more sensitive to changes intheir revenue streams and retained earnings when making their investment decisions.Fourth, the crisis dummy seems to have a statistically significant and negative impacton investment in the high stock market capitalization sample, thereby showing thatcountries where capital markets are more often used to finance investment operationsalso become more vulnerable and exposed to financial crises.

9 Stock market capitalization per GDP is measured at the end of 2008 and arranged to the ratio and split totwo equal sized subsamples. The countries with high stock market capitalization per GDP are, respectively,the following: Hong Kong, South Africa, Singapore, India, Taiwan, Malaysia, Chile, Israel, Korea, China,Croatia, Thailand, Brazil, Colombia and Philippines.10 One can argue that stock market capitalization per GDP is only a rough approximation of the level offinancial development. However, the aim of the sensitivity analysis is to show that results are relativelyrobust to the choice of the estimation sample or method. Another way to split the sample would be basedon the GDP per capita. Again, the main empirical findings are corroborated and the results are availableupon request.

123


As a final robustness check, we consider alternative estimation methods. Theseresults are presented in Tables B.1, B.2, B.3, B.4, B.5, B.6, B.7, B.8 of the Appendix.They broadly confirm the findings using the dynamic panel estimation based on theBlundell and Bond (1998) approach.

5 Conclusion

The sudden emergence of the 2007–2008 crisis, its severity and potentially long-lastingeffects, became crucial elements for understanding the impact of external influences,oil prices, stock markets or even duration dependence on the likelihood of an expansionand contraction ending (Castro 2010a).

In what concerns emerging market economies, their financial markets have beenhit hard by the global financial crisis. Financial wealth losses between the end of2007 and the end of 2008 are estimated to be around USD 2 trillion in Latin Americaand USD 9 trillion in developing Asia and the New Industrialised Countries. Suchfinancial losses clearly reflect the importance of wealth dynamics and have substantialimplications for growth.

In this article, we analyse to what extent fundamental and financial variables influ-ence private investment growth in emerging market economies. Using a panel of 31emerging economies for 1990:1–2008:3, we find evidence supporting the neoclassi-cal model of investment, as GDP and the cost of capital are important fundamentaldeterminants. The estimated effects of these variables are large in magnitude: a 10%increase in real GDP has a positive impact in private investment of 14%, while a 10%increase in the cost of capital reduces it by 3%.

Among financial variables, the lending rate helps explaining the investment dynam-ics: a 100 basis point increase in the lending rate leads to a fall in investment of 0.1%.In contrast, the availability of credit or equity prices do not seem to be a statisticallysignificant factor in the full sample, but financial and economic crises have a negativeimpact on private investment.

The sensitivity analysis confirms that the benchmark results are overall robust todifferent estimation methods, sample selection, or inclusion of additional control vari-ables. In particular, the impact of the output and the cost of capital on investmentremains fairly significant for regional subsamples. However, there are some differ-ences with regards to the response of investment to financial variables: (i) in emergingAsia, credit and equity prices are, for most estimation methods, statistically significantand have an associated coefficient that is positive; (ii) for Latin America, the equitymarket and the banking sector (via the lending rate) seem to be the driving sources ofinvestment financing and (iii) in emerging Europe, financial factors play a less relevantrole.

When one takes into account the degree of financial development, the empiricalevidence suggests that investment is more sensitive to changes in the output and thelending rate in countries with low stock market capitalization (in percentage of GDP),whereas credit conditions impinge more dramatically in countries with a high stockmarket capitalization.

123


As far as other potential determinants of private investment are concerned—namely,inflation, public investment, external debt and commodity prices—our results do notsupport their inclusion in the benchmark model.

The empirical findings also show that investment growth displays high persistence,a feature that deserves particular attention. In fact, in the event of an economic down-turn or an episode of financial or currency crisis, investment may fall dramaticallyin emerging markets as a reflex of its large long-run elasticity with respect to GDP.In addition, the prominent role played by fundamental factors suggests that publicpolicies that lower the cost of capital can help stimulating private investment. In thiscontext, monetary policy can be crucial, in particular, if it targets financial conditions(Castro 2010b; Sousa 2010a), as it can have both a direct effect on lending rates andcredit availability and an indirect effect on the present value of future tax deductionsvia interest payments.

6 Appendices

A Data and summary statistics

See Table A.1 and A.2

Table A.1 Data sources

Variable Source Definition Remark

Investment HA Gross fixed capital formation CP, SA

GDP HA Gross domestic product CP, SA

Cost of capital HA Ratio investment / GDP deflator CP, SA

Equity HA/GFDa Composite index Deflated

Credit IMF code SAP Claims on private sector Deflated

Lending rate IMF code IPb Lending rate Deflated

Inflation HA Annual change in CPI SA

Public spending HA Government consumption expenditure CP, SA

External debt BIS International debt securities

Commodity price HA Reuters/Jefferies CRB all commodities index Deflated

Corporate spread JPM CEMBI Deflated

Notes: In the source section, HA stands for Haver Analytics, GFD for Global Financial Database, IMF forInternational Monetary Fund IFS statistics, BIS for Bank for International Settlements, JPM for JP Morgan.In the remark section, CP means constant price, SA means seasonally adjusted, and Deflated means deflatedusing the GDP deflatora For Argentina, Chile, Colombia, Croatia, Czech Republic, Hong Kong, Israel, Korea, Peru, Philippines,Russia, Singapore, South Africa; b interbank rate for Romania and Turkey

123


Table A.2 Sample period andnumber of observations percountry

Country Obs Sample period

Argentina 55 1995:1–2008:3

Brazil 43 1998:1–2008:3

Bulgaria 28 2001:4–2008:3

Chile 67 1992:1–2008:3

China 43 1997:2–2007:4

Colombia 51 1996:1–2008:3

Croatia 27 2002:1–2008:3

Czech Republic 43 1998:1–2008:2

Estonia 44 1997:2–2008:1

Hong Kong 68 1991:4–2008:3

Hungary 47 1997:1–2008:3

India 42 1998:2–2008:3

Indonesia 30 2001:2–2008:3

Israel 47 1997:1–2008:3

Korea 74 1990:2–2008:3

Latvia 31 2001:1–2008:3

Lithuania 31 2001:1–2008:3

Malaysia 46 1997:2–2008:3

Mexico 57 1994:3–2008:3

Peru 71 1991:1–2008:3

Philippines 71 1990:2–2007:4

Poland 36 1998:1–2006:4

Romania 27 2002:1–2008:3

Russia 47 1997:1–2008:3

Singapore 71 1990:2–2007:4

Slovakia 42 1998:2–2008:3

Slovenia 36 1998:1–2006:4

South Africa 74 1990:2–2008:3

Taiwan 27 2002:1–2008:3

Thailand 55 1995:1–2008:3

Turkey 38 1999:1–2008:3

B Additional estimation methods

See Table B.1, B.2, B.3, B.4, B.5, B.6, B.7, B.8

123


Table B.1 Dynamic panel regressions: full sample

Pooled Pooled OLS Pooled OLS Fixed Random IV/GMMOLS with time

effectswith time andcountry effects

effects effects

L. Investment 0.5296*** 0.5409*** 0.4961*** 0.4887*** 0.5296*** 0.5988***

[0.0243] [0.0258] [0.0260] [0.0176] [0.0173] [0.0361]

GDP 1.3054*** 1.2470*** 1.4970*** 1.5320*** 1.3054*** 0.7774***

[0.0768] [0.0777] [0.0897] [0.0666] [0.0614] [0.1470]

Cost of capital −0.2032*** −0.2049*** −0.2131*** −0.2077*** −0.2032*** −0.1018

[0.0461] [0.0469] [0.0505] [0.0399] [0.0387] [0.1100]

Equity 0.0174*** 0.0218*** 0.0143* 0.0092* 0.0174*** 0.0537***

[0.0064] [0.0083] [0.0082] [0.0054] [0.0054] [0.0124]

Credit −0.0166 −0.0165 −0.0281 −0.0280* −0.0166 −0.0278

[0.0172] [0.0170] [0.0196] [0.0154] [0.0134] [0.0267]

Lending rate −0.0024** −0.0026*** −0.0022*** −0.0021*** −0.0024*** −0.0027***

[0.0011] [0.0008] [0.0008] [0.0005] [0.0005] [0.0005]

Crisis −0.0355*** −0.0348*** −0.0286** −0.0287*** −0.0355*** −0.0360*

(Econ+Fin) [0.0134] [0.0132] [0.0132] [0.0092] [0.0090] [0.0212]

Constant −0.0329*** 0.0109 0.0255 −0.0395*** −0.0329*** −0.0135**

[0.0033] [0.0313] [0.0333] [0.0031] [0.0029] [0.0053]

Observations 1,469 1,469 1,469 1,469 1,469 1,438

# Of countries 31 31 31 31 31 31

R2 0.78 0.79 0.80 0.77 0.78 0.76

Hansen J stat 0.00

Note: Heteroscedasticity and serial correlation robust standard errors in brackets. All series are in logdifferences∗ Statistically significant at 10% level; ∗∗ at 5% level; ∗∗∗ at 1% level

Table B.2 Dynamic panel regressions: Asia



effects effects

L. Investment 0.5559*** 0.6021*** 0.5773*** 0.5328*** 0.5559*** 0.6054***

[0.0394] [0.0404] [0.0416] [0.0278] [0.0273] [0.0599]

GDP 0.9727*** 0.7997*** 1.0017*** 1.1537*** 0.9727*** 0.6258***

[0.1049] [0.1116] [0.1300] [0.1079] [0.0960] [0.1728]

Cost of capital −0.0835 −0.0979 −0.1371 −0.1123 −0.0835 −0.0517

[0.0812] [0.0893] [0.0994] [0.0782] [0.0728] [0.0935]

Equity 0.0173* 0.0109 0.0088 0.0122 0.0173* 0.0479***

[0.0091] [0.0115] [0.0114] [0.0098] [0.0096] [0.0170]

Credit 0.0873** 0.0964** 0.1108*** 0.0984*** 0.0873*** 0.0436

[0.0367] [0.0381] [0.0407] [0.0353] [0.0333] [0.0318]

123


Table B.2 continued



effects effects

Lending rate 0.0710 0.1028 0.1261 0.0857 0.0710 0.0536

[0.0820] [0.0881] [0.0887] [0.0694] [0.0692] [0.1185]

Crisis −0.0803*** −0.0771*** −0.0657*** −0.0687*** −0.0803*** −0.0800***

(Econ+Fin) [0.0204] [0.0195] [0.0198] [0.0140] [0.0135] [0.0309]

Constant −0.0302*** 0.0436 0.0357 −0.0389*** −0.0302*** −0.0146**

[0.0054] [0.0397] [0.0380] [0.0055] [0.0050] [0.0071]

Observations 527 527 527 527 527 517

# Of countries 10 10 10 10 10 10

R2 0.79 0.83 0.83 0.78 0.79 0.78

Hansen J stat 0.00


Table B.3 Dynamic panel regressions: Europe



effects effects

L. Investment 0.5192*** 0.5311*** 0.4771*** 0.4731*** 0.5192*** 0.6454***

[0.0455] [0.0474] [0.0485] [0.0377] [0.0366] [0.0976]

GDP 1.1218*** 1.0753*** 1.3559*** 1.3600*** 1.1218*** 0.5176*

[0.1482] [0.1572] [0.1717] [0.1430] [0.1311] [0.2976]

Cost of capital −0.2039*** −0.1970** −0.2311*** −0.2279*** −0.2039*** −0.1453

[0.0753] [0.0854] [0.0861] [0.0714] [0.0699] [0.1542]

Equity 0.0138 0.0180 0.0080 0.0056 0.0138 0.0440***

[0.0098] [0.0124] [0.0126] [0.0095] [0.0093] [0.0169]

Credit 0.0348 0.0385* 0.0102 0.0100 0.0348 0.0306

[0.0232] [0.0230] [0.0290] [0.0280] [0.0228] [0.0264]

Lending rate −0.0134 −0.0382 −0.0245 −0.0093 −0.0134 0.0400

[0.0289] [0.0345] [0.0343] [0.0283] [0.0269] [0.0447]

Crisis 0.0283 0.0258 0.0195 0.0254 0.0283* 0.0419**

(Econ+Fin) [0.0185] [0.0192] [0.0180] [0.0175] [0.0172] [0.0210]

Constant −0.0317*** −0.1201 −0.0938 −0.0357*** −0.0317*** −0.0115

[0.0063] [0.0910] [0.0893] [0.0075] [0.0064] [0.0113]

Observations 439 439 439 439 439 427

# Of countries 12 12 12 12 12 12

R2 0.66 0.70 0.72 0.60 0.66 0.64

Hansen J stat 0.00

Note: Heteroscedasticity and serial correlation robust standard errors in brackets. All series are in logdifferences. ∗ Statistically significant at 10% level; ∗∗ at 5% level; ∗∗∗ at 1% level

123


Table B.4 Dynamic panel regressions: Latin America



effects effects

L. Investment 0.4162*** 0.4299*** 0.4184*** 0.4072*** 0.4162*** 0.5148***

[0.0410] [0.0432] [0.0441] [0.0326] [0.0322] [0.0626]

GDP 1.9118*** 1.8175*** 1.8591*** 1.9439*** 1.9118*** 1.4663***

[0.1455] [0.1497] [0.1549] [0.1235] [0.1209] [0.2807]

Cost of capital −0.1126 −0.133 −0.1408 −0.1220* −0.1126 0.2703

[0.0852] [0.0812] [0.0900] [0.0737] [0.0717] [0.2061]

Equity 0.0357*** 0.0609*** 0.0636*** 0.0360*** 0.0357*** 0.0840***

[0.0121] [0.0191] [0.0197] [0.0116] [0.0114] [0.0163]

Credit −0.1114*** −0.1592*** −0.1439*** −0.0984*** −0.1114*** −0.1399***

[0.0282] [0.0296] [0.0319] [0.0260] [0.0246] [0.0438]

Lending rate −0.0022** −0.0167* −0.0171* −0.0022*** −0.0022*** −0.0044***

[0.0011] [0.0095] [0.0098] [0.0006] [0.0006] [0.0008]

Crisis −0.0697*** −0.0545** −0.0553** −0.0692*** −0.0697*** −0.0933***

(Econ+Fin) [0.0244] [0.0224] [0.0232] [0.0192] [0.0188] [0.0334]

Constant −0.0363*** 0.0602** 0.0656** −0.0379*** −0.0363*** −0.0238**

[0.0063] [0.0305] [0.0317] [0.0055] [0.0054] [0.0102]

Observations 344 344 344 344 344 338

# Of countries 6 6 6 6 6 6

R2 0.85 0.90 0.90 0.85 0.85 0.82

Hansen J stat 0.00


Table B.5 Dynamic panel regressions: additional regressors 1



effects effects

L. Investment 0.5583*** 0.5687*** 0.5160*** 0.5063*** 0.5583*** 0.6422***

[0.0272] [0.0292] [0.0294] [0.0196] [0.0191] [0.0488]

GDP 1.2680*** 1.2222*** 1.5195*** 1.5260*** 1.2680*** 0.6538***

[0.0939] [0.0937] [0.1059] [0.0785] [0.0735] [0.1667]

Cost of capital −0.1520*** −0.1503*** −0.1798*** −0.1566*** −0.1520*** −0.0278

[0.0510] [0.0527] [0.0571] [0.0439] [0.0427] [0.1233]

Equity 0.0141** 0.0166** 0.0048 0.0027 0.0141** 0.0348***

[0.0067] [0.0084] [0.0084] [0.0061] [0.0060] [0.0101]

Credit −0.0108 −0.0071 −0.0421* −0.0370* −0.0108 −0.0023

[0.0183] [0.0182] [0.0220] [0.0191] [0.0150] [0.0220]

123


Table B.5 continued



effects effects

Lending rate −0.0666*** −0.0689*** −0.0715*** −0.0588*** −0.0666*** −0.1811*

[0.0205] [0.0215] [0.0193] [0.0175] [0.0175] [0.1071]

Inflation 0.0585 0.0752 0.2074*** 0.025 0.0585** 0.063

[0.0367] [0.0597] [0.0719] [0.0365] [0.0282] [0.0561]

Public investment 0.0265 0.0461 0.0625 0.0454 0.0265 0.1675

[0.0381] [0.0384] [0.0437] [0.0360] [0.0346] [0.1255]

External debt −0.0022 −0.0015 0.0056 0.007 −0.0022 −0.0139

[0.0053] [0.0055] [0.0052] [0.0053] [0.0050] [0.0132]

Commodity 0.0115 0.0233 0.2239*** 0.0147 0.0115 0.0314

[0.0142] [0.0566] [0.0723] [0.0146] [0.0145] [0.0198]

Crisis −0.0336** −0.0340** −0.0284* −0.0289*** −0.0336*** −0.0369*

(Econ+Fin) [0.0148] [0.0146] [0.0146] [0.0103] [0.0100] [0.0221]

Constant −0.0369*** −0.0549*** −0.0361** −0.0420*** −0.0369*** −0.0204***

[0.0042] [0.0140] [0.0167] [0.0043] [0.0036] [0.0059]

Observations 1,215 1,215 1,215 1,215 1,215 1,185

# Of countries 30 30 30 30 30 30

R2 0.78 0.79 0.81 0.77 0.78 0.76

Hansen J stat 0.00


Table B.6 Dynamic panel regressions: additional regressors 2



effects effects

L. Investment 0.6305*** 0.6587*** 0.5240*** 0.5232*** 0.6305*** 0.6779***

[0.0441] [0.0473] [0.0559] [0.0391] [0.0355] [0.0893]

GDP 0.8482*** 0.8059*** 1.1540*** 1.1434*** 0.8482*** 0.217

[0.1672] [0.1733] [0.2191] [0.1739] [0.1324] [0.2564]

Cost of capital 0.0058 0.0101 0.0649 0.1442* 0.0058 0.0795

[0.0677] [0.0714] [0.0788] [0.0767] [0.0641] [0.1215]

Equity 0.0343*** 0.0374*** 0.0146 0.0193* 0.0343*** 0.0509***

[0.0091] [0.0123] [0.0117] [0.0105] [0.0099] [0.0194]

Credit −0.0362 −0.0314 −0.0694 −0.0301 −0.0362 −0.0104

[0.0420] [0.0418] [0.0471] [0.0368] [0.0287] [0.0411]

Lending rate −0.1263** −0.1244** −0.1782*** −0.1334** −0.1263** −0.306

[0.0508] [0.0518] [0.0530] [0.0534] [0.0535] [0.2004]

123


Table B.6 continued



effects effects

Inflation 0.2026* 0.1966 0.2196 0.0586 0.2026*** 0.2355

[0.1117] [0.1641] [0.1939] [0.1099] [0.0662] [0.1805]

Public investment 0.0038 0.029 0.0238 −0.0142 0.0038 0.0981

[0.0475] [0.0511] [0.0520] [0.0563] [0.0513] [0.2509]

External debt −0.0209 −0.0250* −0.0242 −0.0181 −0.0209 −0.0017

[0.0145] [0.0144] [0.0219] [0.0217] [0.0149] [0.0225]

Commodity −0.0149 0.002 0.3945*** −0.0085 −0.0149 −0.0207

[0.0236] [0.0796] [0.1307] [0.0252] [0.0233] [0.0300]

Corporate spread 0.0650* 0.0475 −0.0271 0.0471 0.0650* 0.0986

[0.0390] [0.0418] [0.0470] [0.0390] [0.0380] [0.0748]

Constant −0.0240*** −0.0187 −0.016 −0.0225** −0.0240*** −0.0059

[0.0071] [0.0155] [0.0264] [0.0113] [0.0069] [0.0091]

Observations 346 346 346 346 346 326

# Of countries 19 19 19 19 19 19

R2 0.74 0.75 0.79 0.60 0.74 0.70

Hansen J stat 0.00


Table B.7 Dynamic panel regressions: high stock market capitalisation



effects effects

L. Investment 0.5334*** 0.5411*** 0.5111*** 0.5060*** 0.5334*** 0.6171***

[0.0307] [0.0330] [0.0334] [0.0218] [0.0215] [0.0507]

GDP 1.1646*** 1.0563*** 1.2759*** 1.3653*** 1.1646*** 0.6434***

[0.0845] [0.0883] [0.1022] [0.0838] [0.0754] [0.1761]

Cost of capital −0.1509** −0.1726*** −0.1717** −0.1438*** −0.1509*** −0.0386

[0.0612] [0.0636] [0.0713] [0.0552] [0.0530] [0.1372]

Equity 0.0147* 0.0221** 0.0182* 0.0087 0.0147* 0.0587***

[0.0079] [0.0106] [0.0111] [0.0077] [0.0076] [0.0117]

Credit 0.0455* 0.0609** 0.0686*** 0.0513** 0.0455** −0.0049

[0.0257] [0.0258] [0.0264] [0.0234] [0.0226] [0.0363]

Lending rate −0.0023** −0.0024*** −0.0021*** −0.0021*** −0.0023*** −0.0025***

[0.0010] [0.0007] [0.0008] [0.0006] [0.0005] [0.0007]

Crisis −0.0813*** −0.0770*** −0.0643*** −0.0680*** −0.0813*** −0.0766***

(Econ+Fin) [0.0187] [0.0194] [0.0201] [0.0124] [0.0121] [0.0245]

123


Table B.7 continued



effects effects

Constant −0.0319*** 0.0237 0.0084 −0.0400*** −0.0319*** −0.0115*

[0.0043] [0.0392] [0.0399] [0.0040] [0.0037] [0.0070]

Observations 850 850 850 850 850 835

# Of countries 15 15 15 15 15 15

R2 0.78 0.80 0.81 0.78 0.78 0.76

Hansen J stat 0.00

Note Heteroscedasticity and serial correlation robust standard errors in brackets. All series are in logdifferences∗ Statistically significant at 10% level; ∗∗ at 5% level; ∗∗∗ at 1% level

Table B.8 Dynamic panel regressions: low stock market capitalisation



effects effects

L. Investment 0.4904*** 0.5096*** 0.4464*** 0.4355*** 0.4904*** 0.5548***

[0.0363] [0.0395] [0.0412] [0.0298] [0.0292] [0.0732]

GDP 1.5874*** 1.4650*** 1.7373*** 1.8258*** 1.5874*** 1.1930***

[0.1486] [0.1570] [0.1674] [0.1090] [0.1040] [0.3001]

Cost of capital −0.2141*** −0.1783** −0.2143*** −0.2492*** −0.2141*** −0.1153

[0.0668] [0.0703] [0.0713] [0.0567] [0.0556] [0.1834]

Equity 0.0171* 0.0202* 0.013 0.0086 0.0171** 0.0419**

[0.0093] [0.0119] [0.0118] [0.0075] [0.0075] [0.0177]

Credit −0.0495** −0.0387* −0.0528** −0.0617*** −0.0495*** −0.046

[0.0218] [0.0220] [0.0269] [0.0205] [0.0170] [0.0293]

Lending rate −0.0741*** −0.0778*** −0.0689*** −0.0660*** −0.0741*** −0.1384

[0.0195] [0.0210] [0.0185] [0.0180] [0.0178] [0.1066]

Crisis 0.0219 0.02 0.0191 0.0204 0.0219* 0.0271

(Econ+Fin) [0.0152] [0.0146] [0.0143] [0.0134] [0.0131] [0.0268]

Constant −0.0390*** −0.0306*** −0.0346*** −0.0442*** −0.0390*** −0.0276***

[0.0051] [0.0055] [0.0114] [0.0049] [0.0044] [0.0105]

Observations 619 619 619 619 619 603

# Of countries 16 16 16 16 16 16

R2 0.79 0.81 0.83 0.78 0.79 0.78

Hansen J stat 0.00


123


References

Agnello L, Sousa RM (2009) The determinants of public deficit volatility. European Central Bank, ECBworking paper no. 1042. http://www.ecb.int/pub/pdf/scpwps/ecbwp1042.pdf

Arellano M, Bond S (1991) Some tests of specification for panel data: Monte Carlo evidence and an appli-cation to employment equations. Rev Econ Stud 58:277–297

Aschauer DA (1989) Does public capital crowd out private capital?. J Monet Econ 24:171–188Bernanke B, Gertler M (1986) Agency costs, collateral, and business fluctuations. National Bureau of

Economic Research, NBER working paper no. 2015. http://www.nber.org/papers/w2015.pdfBlanchard O, Rhee C, Summers L (1993) The stock market, profit, and investment. Q J Econ 108:115–136Blejer M, Khan M (1984) Government policy and private investment in developing countries. IMF Staff

Pap 31:379–403Blomstrom M, Lipsey R, Zejan M (1996) Is fixed investment the key to economic growth? National Bureau

of Economic Research, NBER working paper no. 4436. http://www.nber.org/papers/w4436.pdfBlundell R, Bond S (1998) Initial conditions and moment restrictions in dynamic panel data models.

J Econom 87:115–143Borensztein E (1990) Debt overhang, credit rationing, and investment. J Dev Econ 32:315–335Bosworth B (1975) The stock market and the economy. Brookings Pap Econ Act 2:257–290Caporale GM, Cipollini A, Spagnolo N (2005) Testing for contagion: a conditional correlation analysis.

J Emp Financ 12:476–489Castro V (2010a) The duration of economic expansions and recessions: More than duration dependence.

J Macroecon 32:347–365Castro V (2010b) Are central banks following a linear or nonlinear (augmented) Taylor rule? J Fin Stab

(forthcoming)Dornbusch R, Reynoso A (1989) Financial factors in economic development. National Bureau of Economic

Research, NBER working paper no. 2889. http://www.nber.org/papers/w2889.pdfFazzari SM, Hubbard RG, Petersen BC (1988) Financing constraints and corporate investment. Brookings

Pap Econ Act 19:141–206Fischer S, Merton R (1984) Macroeconomics and finance: the role of the stock market. Carnegie-Rochester

Conf Ser Public Policy 21:57–108Gabriel VJ, Sangduan P (2010) Assessing fiscal sustainability subject to policy changes: a Markov switching

cointegration approach. Emp Econ (forthcoming). http://www.econ.surrey.ac.uk/discussion_papers/2009%5CDP03-09.pdf

Gilchrist S, Zakrajsek E (2007) Investment and the cost of capital: new evidence from the corporate bondmarket. National Bureau of Economic Research, NBER working paper no. 13174. http://www.nber.org/papers/w13174.pdf

Greenwood J, Smith B (1997) Financial markets in development, and the development of financial markets.J Econ Dyn Control 21:145–181

Holtz-Eakin D, Newey W, Rosen HS (1988) Estimating vector autoregressions with panel data. Econome-trica 56:1371–1395

Johnson R, Soenen L (2009) Commodity prices and stock market behavior in South American countries inthe short run. Emerg Mark Fin Trade 45:69–82

Jorgenson DW (1963) Capital theory and investment behavior. Am Econ Rev 53:247–259King RG, Levine R (1993) Finance and growth: Schumpeter might be right. Q J Econ 108:717–737Mallick SK, Mohsin M (2010) On the real effects of inflation in open economies: theory and empirics. Emp

Econ. doi:10.1007/s00181-009-0328-0Morck R, Shleifer A, Vishny RW (1990) The stock market and investment: is the market a sideshow?.

Brookings Pap Econ Act 21:157–216Oshikoya T (1994) Macroeconomic determinants of domestic private investment in Africa: an empirical

analysis. Econ Dev Cult Change 42:573–596Papyrakis E, Gerlagh R (2004) The resource curse hypothesis and its transmission channels. J Comp Econ

32:181–193Peltonen T, Sousa RM, Vansteenkiste I (2009) Wealth effects in emerging market economies. European

Central Bank, ECB working paper no. 1000. http://www.ecb.int/pub/pdf/scpwps/ecbwp1000.pdfServen L (2003) Real exchange-rate uncertainty and private investment in LDCS. Rev Econ Stat 85:212–

218Shaw E (1974) Financial deepening in economic development. Oxford University Press, Oxford

123

http://www.ecb.int/pub/pdf/scpwps/ecbwp1042.pdf

http://www.nber.org/papers/w2015.pdf



http://www.econ.surrey.ac.uk/discussion_papers/2009%5CDP03-09.pdf

http://www.econ.surrey.ac.uk/discussion_papers/2009%5CDP03-09.pdf



http://dx.doi.org/10.1007/s00181-009-0328-0

http://www.ecb.int/pub/pdf/scpwps/ecbwp1000.pdf


Stiglitz J, Weiss A (1981) Credit rationing in markets with imperfect information. Am Econ Rev 71:393–410Sousa RM (2010a) Consumption, (dis)aggregate wealth, and asset returns. J Emp Financ 17(4):606–622Sousa RM (2010b) Housing wealth, financial wealth, money demand and policy rule: evidence from the

euro area. N Am J Econ Financ 21:88–105Tobin J (1969) A general equilibrium approach to monetary theory. J Money Credit Bank 1:15–29Tybout JR (1983) Credit rationing and investment behavior in a developing country. Rev Econ Stat 65:598–

607Whited TM (1991) Investment and financial asset accumulation. J Financ Intermed 1:307–334

123

Documents

Investment in emerging market economies