Institutions and the external capital structure of countries

Journal of International Money and Finance 28 (2009) 367–391

Contents lists available at ScienceDirect

Journal of International Money and Finance journal homepage: www.elsevier.com/locate/jimf

Institutions and the external capital structure of countriesq Andre´ Faria a, Paolo Mauro b, * a b

Barclays Global Investors, London, United Kingdom International Monetary Fund, United States

a b s t r a c t JEL classification: F21 F34 F36 Keywords: Foreign direct investment Portfolio equity External debt External liabilities

A widespread view holds that countries that finance themselves through foreign direct investment and portfolio equity, rather than bonds and loans, are less prone to crises. But what determines countries’ external capital structures? In a cross-section of advanced economies, emerging markets, and developing countries, we find that equity-like liabilities as a share of countries’ total external liabilities are positively and significantly associated with indicators of educational attainment, openness, natural resource abundance and, especially, institutional quality. These relationships are robust to attempts to control for possible endogeneity, suggesting that better institutional quality may help improve countries’ external capital structures. Ó 2008 Andre´ Faria. Published by Elsevier Ltd. All rights reserved.

1. Introduction A widespread view holds that the external capital structure of countries (that is, the relative shares of items such as foreign direct investment, portfolio equity, and external debt in a country’s external finance) is an important determinant of economic performance and propensity to crises. Indeed, this view has been reinforced by a number of recent emerging market crises, and some authors have argued that it would be desirable for emerging market countries to reduce their reliance on debt and increase the role of equity in their external capital structures (Rogoff, 1999). Equity finance makes it possible for domestic producers to share risk with foreign investors, thereby helping stabilize domestic consumption and improving domestic producers’ ability to undertake projects with high risk and high

q The paper was written while both authors were at the IMF. * Corresponding author. Barclays Global Investors, United States. E-mail addresses: [email protected], [email protected] (A. Faria), [email protected] (P. Mauro). 0261-5606/$ – see front matter Ó 2008 Andre´ Faria. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jimonfin.2008.08.014

368

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

expected return. Moreover, liquidity crises have often been triggered by sudden stops in debt flows, but are unlikely to be generated by sudden stops in equity flows. In addition, one form of equity-like finance, namely, foreign direct investment is often viewed as especially desirable because it is associated with technological transfer (Borensztein et al., 1998). However, for policies aimed at improving capital structures to be formulated, or even for the effects of capital structures to be accurately estimated, it is first necessary to understand the factors underlying countries’ existing capital structures: this is the objective of the present paper. Many previous studies have sought to identify the determinants of either total capital flows or FDI flows only. Previous work on the composition of countries’ capital flows has been more limited. And only recently have new data sets made it possible for researchers to begin analyzing the composition of the stocks of countries’ liabilities. We restrict our analysis to external liabilities, rather than the whole balance sheet. This makes for a simpler analysis and cleaner parallels with the corporate finance literature, which considers the composition of firms’ liabilities. In our study, we devote substantial attention to the role of institutional quality (e.g., the absence of corruption, red tape, or political violence) as a potential determinant of capital structures.1 This is consistent with increased emphasis on institutional variables in explaining economic development and, more specifically, capital flowsdsee, for example, Alfaro et al. (2008), and Lothian (2006). Indeed, an additional motivation for our analysis is the recently identified association between weak institutional quality and severe crises (Acemoglu et al., 2004). The mechanism underlying such association has not yet been uncovered. If institutional quality turns out to be associated with a more crisis-prone external capital structure, this might be a plausible channel through which weak institutions lead to higher frequency and severity of crises. A number ofdoccasionally conflictingdhypotheses have been put forward regarding the impact of institutions on the composition of capital structures. We conjecture, for example, that weak institutions may have an especially deleterious impact on FDI: investors considering foreign direct investment may be particularly concerned about the likely exposure to requests for bribes and the need to work through red tape. Similarly, weak institutions in a recipient country (including lack of transparency in the corporate sector and weak corporate governance)2 may deter international investors from acquiring portfolio equity stakes there. Wei (2001) suggests that weak institutions may reduce the relative importance of FDI in total liabilities: foreign banks are more likely than foreign direct investors to be bailed out in the context of a crisis, and are therefore more willing to invest in corrupt countries; thus, as countries with weak institutions are usually crisis-prone, they will tend to have a smaller share of FDI. Albuquerque’s model (2003) implies that better institutions lead to a greater share of FDI in total liabilities after controlling for a country’s international financial constraints. In contrast, a formal theory by Razin et al. (1998) suggests that informational asymmetries would be associated with a higher share of FDI, and a lower share of portfolio equity, in total external liabilities. The role of institutional quality has been somewhat under explored in previous empirical studies of the determinants of external capital structures. Indeed, pioneering work by Lane and Milesi-Ferretti (2001a,b) on the composition of the stocks of countries’ external liabilities analyzed a limited number of potential correlates (namely, openness, economic size, and per capita GDP).3 Other authors have analyzed the impact of institutional variables on total capital flows; the composition of capital flows, rather than stocks; and specific subcomponents of the stock of liabilities drawn from different data sources. In a cross-section of about 40 advanced and developing countries, Alfaro et al. (2008) find that institutional quality is a key determinant of total capital flows. In a panel of advanced and developing countries, Albuquerque (2003) finds the share of FDI in total flows to be negatively and significantly

1 A recent wave of studies has empirically analyzed the relationship between indicators of institutional quality and economic variables such as investment and growth (Knack and Keefer, 1995; Mauro, 1995); foreign direct investment (Be´nassy-Que´re´ et al., 2005; Wei, 2000a,b); development outcomes (Kaufmann et al., 1999); economic and political instability (Acemoglu et al., 2003); and the severity of crises (Johnson et al., 2000). 2 As reported by Acemoglu and Johnson (2003), corporate governance and political governance are highly correlated, so that it is difficult to disentangle their individual effects. 3 Lane (2004) empirically analyzes the determinants of long-term external debt levels, including the role of institutions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

369

associated with good credit risk ratings, but unrelated to indicators of institutional quality.4 In a crosssection of advanced and developing countries, Hausmann and Ferna´ndez-Arias (2000) consider the determinants of the share of FDI flows in total flows, using averages for 1996–98, and find no relationship with institutional quality.5 In a gravity model of bilateral FDI stocks (drawn from OECD data) and bank loan stocks (drawn from BIS data) applied to a common sample of about 10 source countries and 20 recipient countries, Wei (2001) finds that weaker institutions are associated with less FDI and more bank loans. In this paper, we take the same approach as Lane and Milesi-Ferretti (2001a,b) by focusing on crosscountry variation and, more important, working with stocks, rather than flows. Stocks are the object of capital structure theories in the corporate finance literature, and empirical studies of the determinants of corporate structures typically rely on cross-sectional data on liability stocks at the firm level. Moreover, as Lane and Milesi-Ferretti (2001b) put it, ‘‘[the] stock position is the relevant state variable in a macroeconomic model, and capital flows arise to close the gap between desired and actual stock positions.’’6 Consistent with our focus on the composition of liability stocks, and our interest in the fundamental, slow-moving, determinants of cross-country differences, our empirical analysis is based upon cross-country regressions of time-series means from 1996 to 2004.7 To measure institutional quality, we rely on subjective indicators.8 In an attempt to reduce potential endogeneity bias, we use instrumental variables, following a strategy used by previous studies.9 We find that the key determinants of countries’ external capital structures include institutional quality and, to a lesser extent, educational attainment, openness, and the availability of natural resources. Holding other factors constant, better institutions tilt countries’ capital structures significantly toward equity and away from debt. Our main results are confirmed by a host of robustness exercises. Here we list a few examplesdmore are included in an extensive robustness tests section. First, we show robustness to changes in the sample of countries, data sets, and estimation technique (including robust regressions; bounded dependent variable estimation; and Fama–Macbeth regressions). Second, we document robustness to the introduction of additional or alternative explanatory variables, such as domestic stock market capitalization, international financing constraints, and capital controls or equity market liberalization. The rest of the paper is organized as follows. Section 2 surveys some of the existing hypotheses regarding the potential determinants of countries’ external capital structures. Section 3 describes the data, presents the empirical strategy, and reports the main results and the robustness tests. Section 4 concludes. 2. Existing theories and hypotheses A generally accepted theory of the external capital structure of countries has not yet been developed, though several studies have sought to draw lessons for international finance from the corporate

4

Albuquerque’s (2003) empirical analysis abstracts from country fixed effects. Of these studies, only Alfaro et al. (2008) use instrumental variables to address possible endogeneity issues. 6 In a related vein, Wei (2001) sketches a model in which a multinational allocates FDI stocks among countries. 7 Our baseline cross-country regression of time-series means is equivalent to a between estimator regression. We obtain essentially identical results in regressions where the dependent variables are based on stock measures for any single year during the sample period; or when using Fama–MacBeth regressions (see below). 8 Some authors have questioned the validity of an approach based on subjective indicators of institutional quality combined with instrumental variables, and have argued in favor of objective measures of institutional quality (Glaeser et al., 2004; Przeworski, 2004). We agree that developing better objective measures of institutional quality compared with the existing ones is a high priority task for further research. Nevertheless, our view is that subjective indicators are preferable, because they capture de facto institutional quality: countries may have excellent legislation on their books, but what matters is whether such legislation is applied and enforced in practice. 9 The consultants who produce such subjective indicators of institutional quality might be influenced in their judgment by the structure of a country’s liabilities, or by other factors that are correlated with the structure of liabilities. The instruments we work with include settler mortality and population density in the 1500s from Acemoglu et al. (2003), ethnolinguistic fractionalization from Mauro (1995), and British legal origin from La Porta et al. (1998). 5

370

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

finance literature, which has extensively analyzed capital structures at the firm level.10 In this section, we provide a brief review of full-fledged theories and less formal hypotheses regarding external capital structures. A first theory, by Albuquerque (2003), focuses on the problems of expropriation and imperfect enforcement of international financial contracts.11 The theory assumes that FDI is less subject to expropriation than are other liabilities, though the validity of this assumption may depend on the specific economic sector in which FDI is undertaken. On the whole, Albuquerque (2003) suggests that much of FDI is of an intangible nature (technology, brand names) and thus difficult to expropriate. Under this view, the optimal contract between international investors and financially constrained countries, which are unable to pre-commit not to expropriate, will usually take the form of FDI. Therefore this theory predicts that countries with tighter financial constraints will finance themselves primarily through FDI. The theory may also be interpreted to predict thatdfor given financial constraintsdworse institutions (greater ease of expropriation of FDI) will lead to a lower share of FDI in total external liabilities.12 A second theory, by Razin et al. (1998) focuses on the role of informational asymmetries, and foresees a pecking-order in countries’ external capital structures, as in the corporate finance literature. Firms would finance themselves first through FDI (a parallel to retained earnings and, therefore, internal equity), then debt, and then portfolio equity (external equity). In fact, to circumvent informational barriers, foreign multinationals would favor placing their own managers in the recipient country and thus investing abroad through FDI. To the extent that weak institutional quality (such as a poorly regulated stock market) may be taken to proxy for more severe informational asymmetries, it would be associated with a larger share of FDI, and a lower share of portfolio equity, in total external liabilities.13 As mentioned in Section 1, early empirical tests of the relationship between indicators of institutional quality and variables related to countries’ capital structures have reached a variety of results. In a cross-section of countries (including advanced economies), Hausmann and Ferna´ndez-Arias (2000) document no relationship or a negative relationship between the ratio of FDI inflows to total private capital inflows and institutional quality. In contrast, Wei (2000a,b, 2001) and Wei and Wu (2002) find that weak institutions tilt capital inflows toward bank loans and away from FDI, consistent with their hypothesis that foreign direct investors are less likely to be bailed out than are foreign banks in the event of a crisis. Other studies have identified a number of additional factors that may affect countries’ capital structures, with special attention to FDI.14 Such factors include human capital, natural resources, economic size, and openness. Human capital may act as a stronger ‘‘pull’’ factor for FDI (Borensztein et al., 1998) than other forms of capital such as portfolio equity or debt. Natural resources may also attract FDI to a greater extent than they do other types of capital, as suggested by Hausmann and

10 While corporate finance reasoning cannot be trivially applied to the international setting, the literature on sovereign debt (Eaton and Gersovitz, 1981, 1984; Cole and English, 1991, 1992; Cole and Kehoe, 1995; Bulow and Rogoff, 1989) may be interpreted as being in a broadly similar vein. Hart and Moore’s (1994) analysis of default and renegotiation when one of the sides to a financial contract cannot commit to the contract has particular resonance in international finance. Attempts to extend corporate finance reasoning to the international finance setting are reviewed in Borensztein et al. (2004). A broader survey of theories of capital structures in the domestic corporate context is Myers (2001). Rajan and Zingales (1995) and Booth et al. (2001) have analyzed the effects of government policies, laws, and regulations on the domestic capital structures of the G-7 countries, and developing countries, respectively. Using firm level evidence for a cross-section of 39 developed and developing countries, Fan et al. (2006) find that firms operating in more corrupt countries tend to have capital structures with less equity. 11 Albuquerque’s (2003) main interest is in why FDI flows are less volatile than other capital flows, and he focuses on financial constraintsdempirically proxied by credit risk ratings. For an alternative theoretical analysis on related issues, see Schnitzer (2002). 12 This is our interpretation of Albuquerque’s (2003) model, though it is not emphasized by the author. It is based upon the author’s simulations in Table 2, p. 370, and interpreting the parameter q as the ease with which FDI may be expropriated. (Other types of capital can always be fully expropriated in the model.) 13 There is a conceptual difference between informational asymmetries and institutional weaknesses, and this interpretation may not have been intended by the authors. 14 Lim (2001) reviews the literature on the determinants of FDI.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

371

Ferna´ndez-Arias (2000) and Lane and Milesi-Ferretti (2001b). Indeed, in many cases natural resources might lie unexploited or even undiscovered without the crucial expertise provided by multinationals (Markusen, 1997). However, the tangible nature of FDI aimed at extracting natural resources might make it especially vulnerable to expropriation once it is in place. Larger economic size (proxied by measures such as total GDP) also attracts FDI, which provides an opportunity to better serve the local market (possibly circumventing trade barriers). Finally, openness may reduce the need for ‘‘tariffhopping’’ FDI, though the ease with which products can be exported increases a country’s appeal as a destination for FDI. With a variety of existing theoretical hypotheses, the relationship between countries’ external capital structures and variables such as institutions is ultimately an empirical question. Early empirical tests have not reached definitive conclusions, largely owing to data constraints. In the next section, we provide new empirical evidence on this question, drawing on data sets which have become available recently and which provide far greater country coverage and better cross-country consistency than was the case in the past. 3. Empirical analysis This section briefly describes the data, presents the empirical strategy, and reports the results. Appendix 1 describes the data sources and variable definitions in greater detail. 3.1. Data sources and variables used The data on our dependent variabledexternal liabilities and their subcomponentsdwere assembled by Lane and Milesi-Ferretti (2006), updating and extending their initial exercise (Lane and MilesiFerretti, 2001a,b) from 67 to 145 countries. In the Lane and Milesi-Ferretti (2006) classification, external liabilities comprise FDI, portfolio equity, debt (consisting of portfolio debt, loans, currency, and deposits), and financial derivatives. Lane and Milesi-Ferretti’s new database improves on alternative sources, notably the International Investment Position (IIP) reported in the International Monetary Fund’s International Financial Statistics, in terms of both country coverage and appropriate correction for valuation effects.15 In the robustness section, we show that our main results are similar using the IIP data set. More generally, evidence of robustness to changes in data source is provided by the observation that the working paper version (Faria and Mauro, 2004), where we obtained very similar results, was entirely based upon a previous vintage of the IIP. Potential explanatory variables include the size of the economy (total GDP in U.S. trillions of dollars at constant 2000 prices); the level of economic development (GDP per capita in U.S. thousands of dollars at constant 2000 prices); openness (sum of imports and exports over GDP); the relative importance of natural resources (share of exports of fuels, metals, and ores as a ratio of GDP); human capital (percentage of population over 25 that has attended some secondary schooling); financial development (private credit to GDP)16; a dummy variable for transition economies; and an index of institutional quality. This last variable is the simple average of six institutional indicators drawn from

15 The maximum number of countries covered by IIP for the period 1996–2004 is 106 for some categories of external liabilities. (A thorough description of the IIP data is provided in IMF, 2002a.) While the IIP reports FDI at market value for some countries and at book value for others, Lane and Milesi-Ferretti’s database provides portfolio equity at market value and FDI at book value consistently for all countries. Unfortunately, neither data set distinguishes between public and private liabilities, or between bank loans and non-bank finance. While a public/private decomposition would be interesting, in practice it may not be too informative, because many loans originally extended to private entities are assumed by the sovereign borrower when repayment difficulties emerge. 16 Our baseline measure of financial development, private credit to GDP, is recommended by Levine et al. (2000) as the preferred indicator of financial development, because it proxies for higher levels of financial services and therefore greater financial intermediary development, even though it does not directly measure the amelioration of information and transaction costs. Another advantage is the large country coverage in the data. In the robustness section, we look at alternative measures of financial development.

372

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Kaufmann et al. (2006)17: voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption.18 In the full country sample of Kaufmann et al. (2006), each index ranges between 2 and 2 for the vast majority of countries, with a mean of 0 and a standard deviation of 1.19 We focus our analysis on two groups of countries: the whole sample, including countries at all levels of economic development, and a sample of developing and emerging market countries only. The reason for looking at both samples separately is twofold. First, while advanced economies have substantial gross assets and liabilities, emerging markets and developing countries have traditionally had limited external gross assets.20 Second, while we do control for GDP per capita, we are not only interested in the heterogeneity between advanced countries, on the one hand, and developing and emerging market countries, on the other, but also in the heterogeneity within developing and emerging market countries; indeed, as mentioned in motivating our study, developing countries and emerging markets may deserve special attention because they are more crisis-prone than are advanced economies. Our baseline sample consists of the 94 countriesd22 advanced and 72 emerging/developing, listed in Appendix 1dfor which all of our key explanatory variables are available (at least for 1 year of the years between 1996 and 2004 for each country).21 We regress the time-series mean of the dependent variable for the available years on the time-series means of the explanatory variables. For the typical country in the sample (that is, the cross-country average of time-series means in the sample of 94 countries used in the baseline regressions), FDI is 27 percent of total liabilities, portfolio equity 6 percent, and debt 67 percent. Throughout the paper we report the results for ‘‘total equity’’ (defined as the sum of FDI and portfolio equity).22 Table 1 reports the descriptive statistics for the variables used in this study.

17 Instead of simple averaging of the six subcomponents, one could consider extracting a common component (for example, the first principal component obtained by applying principal components analysis to the six series). This yields essentially the same resultsdnot only in this paper but also in the broader literature on institutional quality. 18 In our view, the indices compiled by Kaufmann et al. (2006), now known as World Bank Governance Indicators (WGI), are ‘‘the state of the art’’ among indicators of institutional quality, in the sense that they are a summary measure of the largest set available of such indicators. These indices are based on several hundred individual variables measuring perceptions of governance, drawn from 31 separate data sources constructed by 25 different organizations, ranging from think-tanks to governments, multilateral organizations and commercial firms (for example, Freedom House, Heritage Foundation, World Economic Forum, U.S. State Department, European Bank for Reconstruction and Development, Economist Intelligence Unit, and Political Risk Services). They report values every other year beginning in 1996, and annually beginning in 2002. 19 The range for the institutional quality index is narrower because we exclude countries without adequate data coverage for other variables, and because of the averaging of the six governance indicators. 20 Blonigen and Wang (2005) argue against pooling advanced economies together with emerging markets and developing countries in empirical studies of FDI, on the grounds that advanced economies experience large two-way FDI flows, whereas emerging markets and developing countries have traditionally been almost exclusively recipients of FDI. 21 Our baseline sample size is smaller than that of Lane and Milesi-Ferretti (2006) primarily owing to constraints on the availability of data on educational attainment. We eliminate offshore financial centers from the sample (13 countries in total, of which six are non-high income countries). However, the main result of the paperdthe positive and significant correlation between institutional quality and the share of equity-like components in total liabilitiesdstill holds at the 1 percent significance level when offshore financial centers are included; the only change vis-a`-vis the baseline regressions refers to the increase in the coefficient and statistical importance of economic size in explaining cross-country shares of total equity in total liabilities for the sample of non-high income countries. Results not reported for the sake of brevity. 22 Whether FDI and portfolio equity should be treated separatelydas emphasized in some of the literaturedis an open question: FDI’s conceptually distinctive feature compared with portfolio equity is the existence of a long-term relationship between the direct investor and the enterprise, and a significant degree of influence on the management of the enterprise. At a practical level, the balance-of-payments statistics usually define FDI on the basis of whether the direct investor has at least 10 percent of the ordinary shares or voting power (for an incorporated enterprise) or the equivalent (for an unincorporated enterprise); but several countries have chosen to permit qualifications from that criterion when a direct investor owns less than 10 percent of an enterprise but has an effective voice in management, or when the investor owns more than 10 percent but does not have an effective voice in management. In this paper, we focus on the aggregate, FDI plus portfolio equity, ‘‘total equity.’’ In the robustness section, we explore the differential impact of the independent variables on the subcomponents of equity. In the working paper version, using a different data set and period coverage, we look into these disaggregated components in more detail.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

373

Table 1 Descriptive statistics: averages 1996–2004. Variable Whole sample Total equity Institutional quality index GDP (constant 2000 US$ trillions) GDP per capita (constant 2000 US$ thousands) Financial development Natural resources Openness Human capital Transition Non-high income countries Total equity Institutional quality index GDP (constant 2000 US$ trillions) GDP per capita (constant 2000 US$ thousands) Financial development Natural resources Openness Human capital Transition

Minimum 0.03 1.56 0.001 0.12

Maximum

Mean

Median

Standard deviation

Coefficient of variation

0.84 1.84 9.55 36.90

0.33 0.11 0.31 6.80

0.30 0.16 0.02 2.04

0.16 0.87 1.12 9.64

0.50 8.17 3.61 1.42

0.04 0.00 0.21 0.03 0

2.04 0.97 2.08 0.90 1

0.48 0.21 0.75 0.40 0.16

0.30 0.10 0.66 0.39 0

0.43 0.24 0.38 0.23 0.37

0.90 1.16 0.51 0.58 2.31

0.04 1.56 0.001 0.12

0.84 1.09 1.24 10.92

0.33 0.28 0.08 2.01

0.30 0.32 0.01 1.42

0.17 0.54 0.19 2.04

0.52 n.a. 2.24 1.01

0.04 0.00 0.24 0.03 0

1.80 0.97 2.08 0.77 1

0.32 0.23 0.76 0.33 0.19

0.25 0.11 0.65 0.25 0

0.31 0.25 0.39 0.21 0.40

0.95 1.08 0.52 0.63 2.05

Sources and notes: The whole sample consists of 94 observations and the non-high income countries sample consists of 72 observations; the classification of countries according to the income level follows the Global Development Network Growth Database. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI, and is expressed as a share of total international liabilities. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP, from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling, from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

The list of potential explanatory variables we consider in our baseline specifications is relatively parsimoniousdnot an unnatural choice in light of the limited number of countries for which data are available and the need to attain a sufficient number of ‘‘degrees of freedom’’ in the estimation. Several of these potential explanatory variables are correlated with each other (Table 2), highlighting the importance of using multivariate regressions. Nevertheless, as mentioned in Section 1 and explained in detail in an extensive ‘‘robustness tests’’ section presented below, our results are confirmed when we introduce additional explanatory variables. 3.2. Results Considering the univariate correlations between the shares of equity in total liabilities and factors potentially associated with liability composition, a number of significant correlations emerge (Table 3). For the whole sample, total equity as a share of total liabilities is significantly and positively correlated with institutional quality, financial development, openness, and human capital; across the sample of non-high income countries, total equity is significantly correlated with the variables mentioned above (though the correlations have a different order of magnitude), as well as with country size and level of economic development. Not surprisingly (given that the shares of the various components of liabilities need to sum to 1), the remaining component of total

374

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Table 2 Pairwise correlations (independent variables): averages 1996–2004. Institutional quality index

GDP

GDP per capita

Financial development

Natural resources

Openness

Human capital

Transition

Whole sample Institutional quality index GDP GDP per capita Financial development Natural resources Openness Human capital Transition

1 0.30*** 0.84*** 0.70*** 0.33*** 0.15 0.68*** 0.01

1 0.51*** 0.54*** 0.15 0.25** 0.33** 0.10

1 0.71*** 0.13 0.03 0.65*** 0.16

1 0.27*** 0.07 0.53*** 0.23**

1 0.12 0.03 0.01

1 0.29*** 0.40***

1 0.42***

1

Non-high income countries Institutional quality index GDP GDP per capita Financial development Natural resources Openness Human capital Transition

1 0.10 0.68*** 0.44*** 0.36*** 0.37*** 0.48*** 0.31***

1 0.30*** 0.35*** 0.10 0.25** 0.13 0.09

1 0.33*** 0.06 0.18 0.55*** 0.22*

1 0.20* 0.30*** 0.27** 0.12

1 0.16 0.04 0.03

1 0.47*** 0.42***

1 0.70***

1

Sources and notes: ***significant at 1%; **significant at 5%; *significant at 10%. The whole sample consists of 94 observations and the non-high income countries sample consist of 72 observations; the classification of countries according to the income level follows the Global Development Network Growth Database. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling, from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Variables are time-series means for the available years during the period 1996– 2004. Appendix 1 provides further details on sources and variable definitions. Table 3 Pairwise correlations of total equity with each independent variable: averages 1996–2004. Institutional quality index Whole sample 0.29*** Non-high income countries 0.56***

GDP

GDP per capita

Financial development

0.07

0.04

0.24**

0.24**

0.48***

0.43***

Natural resources

Openness

Human capital

Transition

0.08

0.32***

0.29***

0.04

0.03

0.37***

0.37***

0.04

Sources and notes: ***significant at 1%; **significant at 5%; *significant at 10%. The whole sample consists of 94 observations and the non-high income countries sample consists of 72 observations; the classification of countries according to the income level follows the Global Development Network Growth Database. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI, and is expressed as a share of total international liabilities. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling, from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

375

liabilities (unreported) bears relationships with all these variables with opposite signs to those of total equity. Turning to multivariate regressions, we begin by focusing on the determinants of the share of total equity in total liabilities in the whole sample (Table 4). Our main, and most robust, finding is that institutional quality is positively and significantly associated with total equity in essentially all specifications and samples, and controlling for a variety of other explanatory variables. The magnitude of the coefficient is economically significant: for example, a one-digit improvement in the institutional quality index is associated with an 18 percentage point increase in the ratio of total equity to total liabilities controlling for economic size and economic development in the whole sample (column 2).23 That magnitude is also reasonably robust to changes in the set of controls and sample considered. Other variables seem to play a role, too, with a statistically and economically significant impact in a number of specifications. Total equity is positively correlated with economic size (perhaps because market size tends to attract foreign investors), butdwhen institutional quality is included as a regressordnegatively with the level of economic development (in the whole sample but not in the sample that excludes advanced economies). The ratio of private credit to GDP, a proxy for domestic financial development, is positively associated with the share of total equity in total liabilities in most specification, and occasionally significant. Openness, natural resource abundance, and indicators of educational attainment are positively, significantly, and fairly robustly associated with the share of total equity in total liabilities. Transition countries have a significantly lower share of total equity controlling for the other baseline explanatory variables. The overall ability of these independent variables to fit the cross-sectional variation in the total equity share is considerable: the adjusted R2 coefficient is 0.37 in the whole sample, and 0.46 in the non-high income sample; the adjusted R2 coefficient in the univariate regressions using institutional quality alone is 0.07 in the whole sample, and 0.31 in the non-high income sample. 3.3. Robustness tests In this section, we outline a number of potential concerns regarding our main estimates, explain our approach in seeking to address them, and report the related findings, as follows. Some results are not shown to conserve space. 3.3.1. Changes in the sample The results are robust to other alternative samples, including the following: an enlarged sample that uses all available data for each specification (i.e., no longer restricted to those countries for which all explanatory variables are available)dsee Table 5; and a narrower sample consisting of the countries that have observations for all dependent and explanatory variables in all years between 1996 and 2004.24

23 In the institutional quality scale, one digit is approximately equal to one standard deviation within the full country sample of Kaufmann et al. (2006): taking the index at face value, this would be equivalent, for example, to improving the institutions of Jamaica to the level of those of Chile, or improving the institutions of Peru to the level of Slovenia. Of course, these comparisons between pairs of countries are only for illustration purposes. Our view is that the institutional quality index is useful in identifying broad cross-country correlations, but measurement error is often too large for comparisons between pairs of countries to be taken too seriously (see Kaufmann and Kraay, 2004). 24 Human capital and the institutional quality index only have information available for some of the years (for human capital we only have information, at best, for 1990, 1995, and 2000). Given the sluggishness of these variables, wherever data for a given year are not available, we use the data for the most recent available year. Narrowing the sample to those years for which data on institutional quality are available (1996, 1998, 2000, 2002-04) does not change the results. Narrowing the sample to the set of countries that have recent data for education (1995 and 2000) does not change the results either. We describe in detail the construction of these variables in Appendix 1.

376

Table 4 Total equity as a share of total international liabilities: averages 1996–2004.

Institutional quality index GDP GDP per capita Financial development Natural resources Openness Human capital Transition Constant Observations Adjusted R-squared

Non-high income countries

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

0.05*** (0.02)

0.18*** (0.03)

0.16*** (0.03)

0.17*** (0.03)

0.17*** (0.04)

0.18*** (0.02)

0.15*** (0.04)

0.12*** (0.04)

0.15*** (0.05)

0.18*** (0.05)

0.03*** (0.01) 0.01*** (0.003)

0.02*** (0.01) 0.01*** (0.003) 0.08 (0.05)

0.04***(0.01) 0.02*** (0.003) 0.07 (0.05)

0.04*** (0.01) 0.02*** (0.003) 0.01 (0.06)

0.14 (0.12) 0.01 (0.02)

0.09 (0.11) 0.01 (0.02) 0.11*** (0.04)

0.21** (0.09) 0.002 (0.01) 0.06 (0.06)

0.24** (0.10) 0.004 (0.02) 0.03 (0.06)

0.15** (0.07)

0.10 (0.08)

0.16** (0.07)

0.16** (0.07)

0.10** (0.05)

0.12** (0.05) 0.34*** (0.13) 0.19*** (0.07) 0.23*** (0.05) 94 0.37

0.11** (0.06)

0.16*** (0.06) 0.18 (0.16) 0.18** (0.08) 0.20*** (0.06) 72 0.46

0.32*** (0.02) 94 0.07

0.40*** (0.02) 94 0.22

0.37*** (0.03) 94 0.23

0.27*** (0.04) 94 0.29

0.38*** (0.02) 72 0.31

0.34*** (0.04) 72 0.33

0.30*** (0.05) 72 0.34

0.21*** (0.06) 72 0.41

Sources and notes: robust standard errors in parentheses. Ordinary least squares regressions. ***Significant at 1%; **significant at 5%; *significant at 10%. The classification of countries according to the income level follows the Global Development Network Growth Database. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI, and is expressed as a share of total international liabilities. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling, from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Dependent and independent variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Whole sample

Whole sample (1) Institutional quality index GDP GDP per capita Financial development Natural resources Openness Human capital Transition Constant Observations Adjusted R-squared

Non-high income countries (2)

0.05*** (0.02) 0.13*** (0.03)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

0.11*** (0.03)

0.16*** (0.03)

0.18*** (0.03)

0.13*** (0.03)

0.10*** (0.04)

0.07* (0.04)

0.15*** (0.03) 0.18*** (0.04)

0.14 (0.11) 0.01 (0.01)

0.09 (0.10) 0.01 (0.01) 0.10** (0.04)

0.28** (0.07) 0.01 (0.01) 0.01 (0.01)

0.03*** (0.01) 0.02* (0.01) 0.04*** (0.01) 0.05*** (0.02) 0.01*** (0.003) 0.01*** (0.003) 0.02*** (0.003) 0.02*** (0.003) 0.08* (0.05) 0.09 (0.05) 0.02 (0.003) 0.16** (0.06) 0.11** (0.04)

0.32*** (0.01) 129 0.06

0.38*** (0.02) 126 0.13

0.35*** (0.03) 124 0.13

0.24*** (0.04) 117 0.29

0.12 (0.07) 0.13*** (0.04) 0.32*** (0.12) 0.18*** (0.06) 0.23*** (0.04) 94 0.38

(10)

0.25*** (0.09) 0.004 (0.02) 0.04 (0.07)

0.17*** (0.06) 0.18*** (0.07) 0.16*** (0.05) 0.18*** (0.05) 0.16 (0.16) 0.17** (0.08) 0.38*** (0.02) 0.32*** (0.03) 0.29*** (0.04) 0.18*** (0.04) 0.19*** (0.05) 105 103 101 94 72 0.19 0.21 0.21 0.39 0.49

Sources and notes: robust standard errors in parentheses. Ordinary least squares regressions. ***Significant at 1%; **significant at 5%; *significant at 10%. The classification of countries according to the income level follows the Global Development Network Growth Database. The dependent variable is total equity as a share of total international liabilities. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Dependent and independent variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Table 5 Robustness tests: enlarging sample size.

377

378

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

3.3.2. Dropping potentially influential observations To show that our results are robust to changes in the sample of countries, we run the key regressions routinely dropping one country at a time, and find that no individual country has excessive influence on the results.25 3.3.3. Robust regressions From the partial correlation plots, we identify a potential outlier (Fiji), and four possible influential observations (Botswana, Chile, Tajikistan, and Trinidad and Tobago, for the whole sample; and Botswana, Chile, Namibia, Tajikistan, and South Korea, for the non-high income sample).26 Therefore we run robust regressions that take these cases into account by weighting the observations inversely to their residuals so that observations with smaller residuals have more weight after dropping influential observations (it is a form of weighted least squares regression). Results for the institutional quality index are unchanged for our preferred specifications (columns 4 and 5, for the whole sample, and columns 9 and 10 for the non-high income countries sample). One result that appears to be somewhat fragile is the (conditional) relationship between openness and total equity for the non-high income countries sample (Table 6).27 3.3.4. Fama–MacBeth procedure The results hold when we adopt the Fama and MacBeth (1973) approach, widely used in finance, as an alternative estimation procedure to explore both time-series and cross-section information. In the first step, this involves running cross-section regressions for each year. In the second step, the timeseries averages of the cross-sectional regression point estimates are used as point estimates for the coefficients of interest, and the standard deviations of the cross-sectional estimates are used to generate the standard errors for the estimates. The point estimates for all estimated coefficients are very similar to those reported in our baseline regressions, and all main results retain their high degree of statistical significance. 3.3.5. Bounded nature of the dependent variables To take into account that the shares of total equity by definition cannot lie outside the 0–1 range, we run the key regressions using a quasi-maximum likelihood procedure, as proposed by Papke and Wooldridge (1996), and obtain broadly similar results (not shown) to those reported above. 3.3.6. Shares of GDP Although we are mostly interested in the determinants of the composition of external liabilities, we also checked whether the same explanatory variables are associated with the size of total equity (and its subcomponents, FDI and portfolio equity) expressed as a share of GDP. The findings are similar: institutional quality, openness, and, for the non-high income countries sample, natural resources are

25 By dropping one country at a time for specifications (4) and (5) for the whole sample, and (9) and (10) for the non-high income countries sample (Table 4), the institutional quality index coefficient remains significant at the 1 percent level, the only exception being when we drop South Korea from the non-high income countries sample for specification (9), which makes the coefficient significant only at the 5 percent level and slightly smaller (0.11). By dropping Fiji, the results are strengthened; however, we keep Fiji in the sample as its inclusion goes against the main result of the paper. 26 An extreme version of this procedure would be to drop all observations we think are outliers or influential observations. First, we adopt a conservative approach and drop only the influential observations. For the regressions in the whole sample, the significance level for the institutional quality index remains at the 1 percent level, though the magnitude of the coefficient is lower (12–13 percentage points). For the regressions in the non-high income countries, for specification (10), the magnitude of the coefficient falls to 9 percentage points, and is significant only at the 10 percent level (for specification 9, the coefficient is only 4 percent and we cannot reject it is different from zero). By dropping Fiji (the outlier), institutional quality is again significant in all specifications, even when the influential observations are dropped. 27 For specifications (7) and (8), Fiji obtains zero weight and China is dropped from the regressions (technically, its Cook’s D value is larger than one, so an influential observation). The institutional quality index coefficient remains significant at the 5 percent level but its magnitude is smaller. For specification (5), the United States is dropped from the regressions because its Cook’s D value is larger than one.

Table 6 Robustness tests: accounting for outliers and influential observations.

Institutional quality index GDP GDP per capita Financial development Natural resources Openness Human capital Transition Constant Observations Adjusted R-squared

Non-high income countries

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

0.06*** (0.02)

0.18*** (0.03)

0.16*** (0.03)

0.18*** (0.04)

0.20*** (0.04)

0.18*** (0.03)

0.12*** (0.04)

0.09** (0.04)

0.16*** (0.05)

0.18*** (0.05)

0.03* (0.02) 0.01*** (0.003)

0.02 (0.02) 0.01*** (0.003) 0.08 (0.05)

0.03** (0.02) 0.02*** (0.003) 0.07 (0.05)

0.10*** (0.04) 0.02*** (0.01) 0.05 (0.06)

0.04 (0.14) 0.03** (0.01)

0.05 (0.14) 0.03*** (0.01) 0.10* (0.06)

0.22** (0.10) 0.002 (0.01) 0.09 (0.06)

0.23** (0.10) 0.002 (0.01) 0.01 (0.07)

0.18*** (0.07)

0.15** (0.07)

0.20*** (0.07)

0.18*** (0.07)

0.07* (0.04)

0.13*** (0.04) 0.33*** (0.12) 0.20*** (0.06) 0.22*** (0.05) 93 0.38

0.06 (0.04)

0.12** (0.05) 0.14 (0.14) 0.14** (0.06) 0.20*** (0.05) 72 0.50

0.31*** (0.02) 94 0.10

0.38*** (0.02) 94 0.24

0.35*** (0.03) 94 0.25

0.27*** (0.04) 94 0.30

0.37*** (0.02) 72 0.33

0.30*** (0.03) 71 0.39

0.25*** (0.04) 71 0.46

0.22*** (0.05) 72 0.45

Sources and notes: standard errors in parentheses. Robust regressions. ***Significant at 1%; **significant at 5%; *significant at 10%. The classification of countries according to the income level follows the Global Development Network Growth Database. The dependent variable is total equity as a share of total international liabilities. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006) also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling, from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Dependent and independent variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Whole sample

379

380

Table 7 Robustness tests: validity of the results for the IMF’s International Investment Position data set.

Institutional quality index GDP GDP per capita Financial development Natural resources Openness Human capital Transition Constant Observations Adjusted R-squared

Non-high income countries

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

0.04* (0.02)

0.17*** (0.04)

0.15*** (0.04)

0.16*** (0.04)

0.14*** (0.04)

0.19*** (0.03)

0.16*** (0.05)

0.13*** (0.05)

0.16*** (0.05)

0.20*** (0.05)

0.04*** (0.01) 0.03** (0.01) 0.01*** (0.003) 0.01*** (0.004) 0.10 (0.06)

0.32*** (0.02) 0.38*** (0.03) 70 70 0.03 0.19

0.35*** (0.03) 70 0.21

0.04*** (0.01) 0.04** (0.01) 0.02*** (0.003) 0.02*** (0.003) 0.09 (0.07) 0.01 (0.07)

0.24*** (0.04) 0.19*** (0.04) 0.01 (0.01) 0.01 (0.01) 0.11** (0.05)

0.27*** (0.04) 0.28*** (0.04) 0.01 (0.01) 0.01 (0.01) 0.07 (0.07) 0.05 (0.07)

0.18* (0.11)

0.10 (0.12)

0.30*** (0.07) 0.28*** (0.08)

0.10** (0.05)

0.11** (0.05) 0.53*** (0.12) 0.21*** (0.06) 0.16*** (0.06) 70 0.39

0.11* (0.06)

0.24*** (0.06) 70 0.26

0.37*** (0.02) 0.31*** (0.04) 48 48 0.29 0.46

0.27*** (0.05) 0.14** (0.06) 48 48 0.48 0.59

0.14*** (0.05) 0.43*** (0.16) 0.22*** (0.07) 0.11* (0.06) 48 0.67

Sources and notes: robust standard errors in parentheses. Ordinary least squares regressions. ***Significant at 1%; **significant at 5%; *significant at 10%. The classification of countries according to the income level follows the Global Development Network Growth Database. The dependent variable is total equity as a share of liabilities. Total equity consists of portfolio equity plus FDI. International liabilities and their components are from countries’ International Investment Position in the IMF’s International Financial Statistics (IFS). The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Dependent and independent variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Whole sample

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

381

estimated to play a role using that alternative dependent variable. The working paper version (Faria and Mauro, 2004) reports a host of tables analyzing this issue. 3.3.7. Validity of the results for the IMF’s International Investment Position data set We replicate the main regressions (as in Table 4) involving the share of total equity in total liabilities, with IIP data (Table 7).28 Reassuringly, despite using a different data set that covers different sets of countries, the main result of the paper remains unchanged: the institutional quality index coefficient is positively and significantly correlated with total equity at conventional levels (and the magnitude of the coefficients is close to the ones obtained with the Lane and Milesi-Ferretti data set). It is also worth recalling that the working paper version (Faria and Mauro, 2004), which had obtained very similar results, had been entirely based on a previous vintage of the IIP data set. Compared with that previous version of this study, one result that changes with the new Lane and Milesi-Ferretti (2006) data is that the ratio of portfolio equity to FDI no longer bears a clear-cut and robust relationship to institutional quality when other controls enter in the regression. 3.3.8. Possible special role of individual subcomponents of the institutional quality index Now beginning again from our baseline sample and data set, one might be interested in finding out whether a particular subcomponent of the institutional quality index is especially important in driving the results. We thus regress our dependent variables on each subcomponent of the institutional quality index, one at a time. In all regressions, we include our baseline controls (GDP, GDP per capita, financial development, natural resources, openness, human capital, and a transition dummy) (Table 8).29 (We do not report coefficients and p-values for the controls.) There is no clear pattern of a particular subcomponent being more important than others. (Indeed, theory suggests that various aspects of poor institutional quality would tend to go togetherdsee Mauro, 2004.) The coefficients of the different institutional subcomponents, whenever significant, have the same sign.30 3.3.9. Subcomponents of equity In an effort to disentangle the exact mechanisms whereby various factors may affect the composition of countries’ external liabilities, we run multivariate regressions of debt and each subcomponent of equity (FDI and portfolio equity) on the same set of potential determinants. This allows us to ask questions such as the following: when the institutional quality index improves by one digit, by how many percentage points of total liabilities do the shares of debt, FDI, and portfolio equity change, respectively? By identity, the sum of the changes in the shares of FDI, portfolio equity, and debt has to equal to 0. We find that institutional quality is positively and significantly associated with total equity, and, in particular, with FDI as a share of total liabilities. As countries’ institutional quality improves, the composition of external liabilities tilts in favor of FDI over debt. FDI may thus be especially vulnerable to institutional weaknesses such as red tape hurdles or expropriation through bribes. A one-digit improvement in the institutional quality index is associated with a 3 percentage point increase in the share of portfolio equity in total liabilities (though not statistically different from 0); a 14 percentage point increase in the FDI share; and a 17 percentage point decline in the share of debt (which includes portfolio debt, loans, currency, and deposits). Consistent with previous studies, a number of ‘‘pull’’ factors, such as openness, natural resources, and human capital are positively and significantly associated with the share of FDI in total liabilities and thus seem to play a role in attracting FDI. In some specifications, financial development, though not significantly associated with total equity as a share of total liabilities, is positively and significantly associated with portfolio equity as a share of total liabilities.31

28 We show regressions for the averages in the period 1996–2004 for the sake of comparability with the baseline regressions that use Lane and Milesi-Ferretti’s (2006) data set. Including 2005 in the IIP averages does not change the results. 29 As noted above, the six subcomponents that form the institutional quality index are: voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. 30 The few instances in which a subcomponent is not significant are due to the inclusion of Fiji in the sample. Dropping Fiji makes all subcomponents significant and increases or maintains the magnitude in almost all of them. 31 These results are strengthened when running robust regressions.

382

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Table 8 Robustness tests: role of individual subcomponents of institutional quality.

Institutional quality index Voice and accountability Government effectiveness Political stability Regulatory quality Rule of law Control of corruption

Whole sample

Non-high income countries

0.17*** 0.08*** 0.14*** 0.12*** 0.11*** 0.09** 0.13***

0.18*** 0.07** 0.13*** 0.11*** 0.08 0.08 0.17***

(0.04) (0.03) (0.04) (0.03) (0.04) (0.04) (0.03)

(0.05) (0.03) (0.05) (0.03) (0.05) (0.05) (0.04)

Sources and notes: robust standard errors in parentheses. Ordinary least squares regressions. ***Significant at 1%; **significant at 5%; *significant at 10%. The dependent variable is total equity as a share of liabilities. Each line of the table represents a regression. In each regression, the controls are GDP, GDP per capita, financial development, natural resources, openness, human capital, and a transition dummy. The whole sample includes 94 observations and the non-high income countries sample include 72 observations; the classification of countries according to the income level follows the Global Development Network Growth Database. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling, from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Dependent and independent variables are time-series means for the available years during the period 1996– 2004. Appendix 1 provides further details on sources and variable definitions.

3.3.10. Controlling for international financial constraints We now turn to changes in the list of explanatory variables. To relate our results to one of the propositions put forward by Albuquerque (2003), we add a control for international financial constraints. As a proxy, we use the number of years a country was in default between 1970 and 2001.32 We find that institutional quality is still positively and significantly associated with the share of equity in total liabilities and the ratio of FDI to total liabilities (Table 9). Financing constraints are positively and significantly associated with the share of equity in total liabilities, in particular FDI, as predicted by Albuquerque (2003). 3.3.11. Domestic stock market development instead of domestic credit The baseline regressions reported in the previous section have included the ratio of domestic credit to GDP as an explanatory variable. The results are robust to using instead alternative measures of domestic financial development, notably proxies for domestic stock market development, which may help attract not only portfolio equity, but also foreign direct investment, because multinational firms often tap domestic markets to finance local investments. Although domestic stock market capitalization is significant in a number of specifications, institutional quality remains highly significant in most specifications (Table 10). These results are robust to using the number of listed firms as an alternative proxy for domestic stock market development. 3.3.12. Controlling for international equity liberalization This is of course potentially most important for portfolio equity, and is also related to the previous point, as both portfolio equity flows and domestic stock market capitalization have become gradually more important in recent years. We run the regressions controlling for a dummy variable indicating whether a country had liberalized international access to its equity markets by 1995 (Bekaert et al.,

32 We thank a referee for suggesting this measure, whichdas suggested by Reinhart et al. (2003)dmay be viewed as more exogenous than credit ratings.

Table 9 Robustness tests: including international financing constraints.

Institutional quality index Financing Constraints GDP GDP per capita Financial development Natural resources Openness Human capital Transition Constant Adjusted R-squared

Non-high income countries (2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

0.06*** (0.02) 0.18*** (0.02)

0.16*** (0.03)

0.17*** (0.03)

0.17*** (0.03)

0.18*** (0.02)

0.16*** (0.04)

0.13*** (0.04)

0.15*** (0.05)

0.18*** (0.05)

0.006*** (0.002)

0.005*** (0.002)

0.006*** (0.002)

0.004* (0.002)

0.003* (0.002) 0.002 (0.002) 0.002 (0.002) 0.004* (0.002) 0.001 (0.003)

0.005*** (0.002)

0.03*** (0.01) 0.02*** (0.01) 0.04*** (0.01) 0.04*** (0.01) 0.01*** (0.003) 0.01*** (0.003) 0.01*** (0.003) 0.02*** (0.004) 0.07 (0.05) 0.06 (0.06) 0.01 (0.06) 0.13* (0.07) 0.12*** (0.05)

0.30*** (0.02) 0.38*** (0.03) 0.11 0.25

0.35*** (0.03) 0.25

0.23*** (0.04) 0.33

0.10 (0.07) 0.13*** (0.05) 0.29** (0.15) 0.17** (0.08) 0.21*** (0.05) 0.37

0.14 (0.12) 0.01 (0.02)

0.37*** (0.02) 0.31

0.09 (0.11) 0.01 (0.02) 0.11*** (0.04)

0.22** (0.10) 0.002 (0.01) 0.05 (0.05)

0.24** (0.10) 0.004 (0.02) 0.03 (0.07)

0.15** (0.07) 0.14** (0.07)

0.16** (0.07) 0.17*** (0.06) 0.17 (0.18) 0.17* (0.09) 0.19*** (0.06) 0.46

0.34*** (0.04) 0.30*** (0.04) 0.19*** (0.06) 0.32 0.34 0.42

Sources and notes: robust standard errors in parentheses. Ordinary least squares regressions. ***Significant at 1%; **significant at 5%; *significant at 10%. The whole sample includes 94 observations and the non-high income countries sample include 72 observations; the classification of countries according to the income level follows the Global Development Network Growth Database. The dependent variable is total equity as a share of liabilities. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. Financing constraints are measured by the number of years a country has been in default between 1970 and 2001. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Financial development is measured by private credit divided by GDP from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling, from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Dependent and independent variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Whole sample (1)

383

384

Whole sample

Institutional quality index GDP GDP per capita Market capitalization Natural resources Openness Human capital Transition Constant Adjusted R-squared

Non-high income countries

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

0.16*** (0.03)

0.14*** (0.03)

0.16*** (0.04)

0.17*** (0.04)

0.15*** (0.05)

0.14*** (0.05)

0.17*** (0.05)

0.19*** (0.05)

0.03*** (0.01) 0.01*** (0.003)

0.03** (0.01) 0.01*** (0.003) 0.08* (0.05)

0.05*** (0.01) 0.02*** (0.003)

0.05*** (0.01) 0.02*** (0.004) 0.03 (0.06)

0.13 (0.12) 0.01 (0.02)

0.12 (0.13) 0.005 (0.02) 0.07 (0.05)

0.24** (0.09) 0.01 (0.02)

0.26*** (0.09) 0.01 (0.02) 0.1 (0.1)

0.16* (0.09)

0.16* (0.09)

0.26*** (0.08)

0.28*** (0.09)

0.13*** (0.04) 0.32** (0.14) 0.18*** (0.06) 0.21*** (0.09) 0.33

0.13*** (0.05) 0.34** (0.16) 0.19*** (0.07) 0.21*** (0.06) 0.32

0.18*** (0.05) 0.16 (0.18) 0.15** (0.07) 0.16** (0.07) 0.43

0.20*** (0.06) 0.25 (0.21) 0.21** (0.09) 0.16** (0.07) 0.44

0.41*** (0.02) 0.16

0.39*** (0.03) 0.18

0.36*** (0.05) 0.23

0.34*** (0.05) 0.23

Sources and notes: robust standard errors in parentheses. Ordinary least squares regressions. ***Significant at 1%; **significant at 5%; *significant at 10%. The whole sample includes 78 observations and the non-high income countries sample include 56 observations; the classification of countries according to the income level follows the Global Development Network Growth Database. The dependent variable is total equity as a share of liabilities. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. GDP (U.S. trillions of dollars at constant 2000 prices) and GDP per capita (U.S. thousands of dollars at constant 2000 prices) are from the World Bank’s World Development Indicators (WDI). Market capitalization is the ratio of stock market capitalization divided by GDP, from the WDI. Natural resources are the percentage of ore, metals and fuels in total exports, built using data from the WDI. Openness is the sum of exports and imports divided by GDP, built using data from the WDI. Human capital is the share of population over 25 that attended at least some level of secondary schooling from the World Bank’s Education Indicators, EDSTATS (Education Attainment in the Adult Populationdfollowing Barro and Lee, 2001). Transition is a dummy variable that indicates whether a country belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries, from the Global Development Network Growth Database. Dependent and independent variables are time-series means for the available years during the period 1996–2004. Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Table 10 Robustness tests: alternative measure of financial development.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

385

2005). About one-third of the countries in our sample had not liberalized by that time, whereas several emerging markets had already liberalized by the early 1990s.33 The impact of institutional quality on external capital structures is essentially the same as in the main tables. We also run these same regressions using a financial liberalization reform index created by Detragiache et al. (unpublished): the institutional quality index remains significant at the 1 percent level and the financial development index is only significant (at the 10 percent level) when the only regressors are financial development, institutional quality, size, and level economic development.

3.3.13. Adding capital controls to the list of independent variables Our main results hold when we introduce standard measures of capital controls as an additional regressor.34 In all of our main regressions, the coefficient on institutional quality is essentially unchanged and always remains significant, whereas capital controls are never statistically significant. Capital controls are always positively (and significantly, in some specifications) associated with the share of total equity in total liabilities. The impact of capital controls on total equity comes from its positive (and sometimes significant) association with the share of FDI in total liabilities. Specifically, to proxy for capital controls, we use the sum of four dummy variables that take the value of one if the country has (a) multiple exchange rates; (b) current account restrictions; (c) capital account restrictions; and (d) export proceeds surrender requirements. (Thus, in each year, the summary measure takes integer values between 0 and 4.) For each country, we use the 1990–1995 average of this sum. The dummy variables are all drawn from the International Monetary Fund’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAR).35

3.3.14. Possible endogeneity of the institutional quality index We run regressions of liability components (as a share of total liabilities) on institutional quality using a variety of instruments such as settler mortality. The F-statistic in the first-stage regression confirms that the instruments are not ‘‘weak,’’ and that 2SLS estimation is thus warranted. The identifying assumption is that settler mortality (and/or the other instruments) affects institutional quality, and institutional quality in turn affects the composition of countries’ external liabilities, with no other links between liability structures and the instruments. In particular, for the identifying assumption to hold, there must be no direct channel from the instruments to liability structures. This leads us to use univariate regressions combined with a broad interpretation of ‘‘institutions’’ (Table 11). For example, column (2) reports the results of the share of total equity in total liabilities on an index of institutional quality, using settler mortality and population density in the 1500s as instruments (as in Acemoglu et al., 2001). This specification may be of interest to those who believe that settler mortality and population density in the 1500s affected institutions in the broad sense (the institutional quality index would then proxy for many aspects of institutions, perhaps even including educational attainment); and institutions in turn affected our dependent variable, with no direct channel from the instruments to liability structures. Column (4) reports the results using instead ethnolinguistic fractionalization (as in Mauro, 1995) and British legal origin (as in La Porta et al., 1998) as instruments. In all

33 The samples for which the data are available consist of 25–44 countries (depending on the specification). Institutional quality is significant at the 1 percent level in all specifications (except one in which it is significant at the 5 percent level); in such limited samples, the equity market liberalization dummy is never significant. Moreover, Bekaert et al. (2005) use an index of institutional quality as an instrument for equity market liberalization, suggesting that they view equity market liberalization as endogenous to institutional quality. Equity market liberalization appears with the expected (positive) sign and significant when the dependent variable is the share of portfolio equity in total liabilities: a country that liberalized its equity market before 1995 has a share of portfolio equity in total liabilities 2–3 percentage points larger than a country that did not liberalize. (For comparison, the sample mean share of portfolio equity in total liabilities is 7 percent.) 34 Ideally, in our set up, one would wish to control for restrictions on certain types of capital flows (such as FDI, or short-term flows). Unfortunately, reliable cross-country measures of capital controls by type of flow are not yet available. 35 In 1996, the format of the AREAR changed to more detailed dummies with no simple mapping to the previous system. In our view, and in light of the persistence of many aspects of capital controls, this is the best compromise measure in terms of precision of the capital controls measure, relevance for the questions we address, and availability for a broad cross-section of countries.

386

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Table 11 Robustness tests: two stage least squares regressions.

Institutional quality index Constant Observations R-squared in OLS First stage for institutional quality index Settler mortality Population density in 1500 Ethnolinguistic fractionalization British legal origin Constant R-squared in first stage F-statistic Hansen’s J statistic (p-value)

(1)

(2)

(3)

OLS

IV

OLS

(4) IV

0.12*** (0.02) 0.35*** (0.02) 51 0.35

0.13*** (0.03) 0.35*** (0.02) 51 n.a.

0.06*** (0.02) 0.30*** (0.02) 85 0.15

0.08*** (0.03) 0.29*** (0.02) 85 n.a.

0.26*** (0.08) 0.23*** (0.04)

1.16*** (0.37) 0.58 38.05 0.19

0.02*** (0.003) 0.45** (0.19) 0.72*** (0.16) 0.22 18.62 0.14

Sources and notes: robust standard errors in parentheses. ***Significant at 1%; **significant at 5%; *significant at 10%. The dependent variable (in the second stage) is total equity as a share of liabilities. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. Settler mortality is the logarithm of settler mortality for former colonies; and population density in the 1500s is the logarithm of population density in the 1500s for former colonies; both from Acemoglu et al. (2001). Ethnolinguistic fractionalization is the probability that two randomly selected persons from a given country will not belong to the same ethnolinguistic group from Mauro (1995). British legal origin is a dummy variable that attributes one to countries with English law or former British colonies or protectorates, from La Porta et al. (1998). Appendix 1 provides further details on sources and variable definitions.

cases, the coefficient on institutional quality rises compared with the ordinary least squares estimation. (To emphasize this point, we report the OLS results obtained with the same sample of countries as is available for instrumental variable estimation.) 4. Conclusion Previous studies have emphasized the importance of countries’ external capital structures for economic performance: reliance on equity-like instruments (FDI and portfolio equity) improves an economy’s ability to share risks with international investors; moreover, FDI is usually considered to be a vehicle for technological transfer. This study has shown that equity-like components in countries’ external capital structures are significantly associated with indicators of institutional quality, as well as educational attainment, and natural resources. This finding may help shed light on the mechanism underlying the observed correlation between weak institutional quality and severe crises (Acemoglu et al., 2004): weak institutions may tend to increase countries’ reliance on crisis-prone forms of financing, thereby increasing the frequency and severity of crises. Our interpretation of the results is that improving institutionsdobviously no easy task, and typically requiring a long timedmay help to promote a shift toward more desirable external liability structures. Moreover, measures aimed at improving countries’ external capital structures in a more direct manner should be evaluated carefully, because their effectiveness might be undermined by countries’ weak institutional quality. Acknowledgements We thank Geert Bekaert, Eduardo Borensztein, Simon Johnson, Gian Maria Milesi-Ferretti, Elias Papaioannou, Alan Taylor, Shang-Jin Wei, participants in seminars at the International Monetary Fund and the InterAmerican Development Bank, the CEPR conference on ‘‘Institutions, Policies, and Economic Growth’’, the Annual Congress of the European Economic Association, and the Annual Meeting of the Latin American and Caribbean Economic Association, and two anonymous referees for insightful suggestions; and Priyanka Malhotra and Martin Minnoni for able research assistance. The views expressed are those of the authors and do not necessarily represent those of the IMF or IMF policy.

Appendix 1. Sources and description of the variables

Whole sample (1) Institutional quality index GDP GDP per capita Financial development Natural resources Openness Human capital Transition Constant Observations Number of time periods Adjusted R-squared

Non-high income countries (2)

0.05*** (0.01) 0.16*** (0.01)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

0.14*** (0.01)

0.16*** (0.01)

0.14*** (0.01)

0.17*** (0.01)

0.14*** (0.01)

0.11*** (0.01)

0.14*** (0.01)

0.15*** (0.01)

0.03*** (0.001) 0.02*** (0.001) 0.03*** (0.001) 0.03*** (0.002) 0.01*** 0.01*** 0.01*** (0.0005) 0.02*** (0.0003) (0.0003) (0.0004) 0.07*** (0.01) 0.07*** (0.01) 0.03* (0.01)

0.16*** (0.01) 0.11*** (0.01) 0.21*** (0.01) 0.22*** (0.01) 0.01*** (0.002) 0.01*** (0.002) 0.003*** 0.002*** (0.002) (0.002) 0.11*** (0.01) 0.09*** (0.01) 0.03* (0.002)

0.17*** (0.01)

0.13*** (0.01)

0.19*** (0.01)

0.07** (0.02)

0.29*** (0.01) 545 9

0.08*** (0.02) 0.10*** (0.01) 0.12*** (0.02) 0.09*** (0.02) 0.21*** (0.01) 0.20*** (0.01) 545 545 9 9

0.41

0.49

0.32*** (0.02) 0.39*** (0.02) 737 737 9 9

0.36*** (0.02) 737 9

0.27*** (0.01) 737 9

0.08*** (0.02) 0.31*** (0.03) 0.13*** (0.01) 0.22*** (0.01) 0.37*** (0.02) 0.33*** (0.01) 737 545 545 9 9 9

0.09

0.26

0.34

0.40

0.24

0.30

0.37

0.18*** (0.01)

0.51

Sources and notes: robust standard errors in parentheses. ***Significant at 1%; **significant at 5%; *significant at 10%. The classification of countries according to the income level follows the Global Development Network Growth Database. The dependent variable (in the second stage) is total equity as a share of liabilities. International liabilities and their components are from Lane and Milesi-Ferretti (2006). Total equity consists of portfolio equity plus FDI. The institutional quality index is the simple average of six governance indicators from Kaufmann et al. (2006), also known as World Bank Governance Indicators (WGI): voice and accountability; political stability and absence of violence; government effectiveness; regulatory quality; rule of law; and control of corruption. Settler mortality is the logarithm of settler mortality for former colonies; and population density in the 1500s is the logarithm of population density in the 1500s for former colonies; both from Acemoglu et al. (2001). Ethnolinguistic fractionalization is the probability that two randomly selected persons from a given country will not belong to the same ethnolinguistic group from Mauro (1995). British legal origin is a dummy variable that attributes one to countries with English law or former British colonies or protectorates from La Porta et al. (1998). Appendix 1 provides further details on sources and variable definitions.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Table A1 Fama–MacBeth regressions.

387

388

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

Dependent variables The source for countries’ total external liabilities and their components in the baseline regressions (FDI, portfolio equity and debt) is the data set developed by Lane and Milesi-Ferretti (2006). All variables are in millions of U.S. dollars. Data are available at http://www.imf.org/external/pubs/ft/wp/ 2006/data/wp0669.zip. An alternative source for countries’ total external liabilities and their components (FDI, portfolio equity, portfolio debt, and other instruments) is the International Investment Position reported in the IMF’s International Financial Statistics. All variables are in millions of U.S. dollars. A thorough description of the methodology is available at http://www.imf.org/external/np/sta/iip/guide/index.htm. The dependent variables are expressed as ratios to total liabilities. The dependent variables used in the baseline regressions and in most specifications are, unless otherwise noted, a time-series mean of the variables of interest between 1996 and 2004, whenever available. Independent variables Institutional quality index Simple average of six institutional indicators (voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, control of corruption), drawn from Kaufmann et al. (2006), for all available years between 1996 and 2004 (available for 1996, 1998, 2000, and annually from 2002 on). The institutional quality index in a given year is formed only for countries that have information for all governance indicators in that year. Each institutional indicator is modeled by the authors as a standard normal distribution (zero mean, and standard deviation one), http://info.worldbank.org/governance/wgi2007/resources.htm. Gross domestic product Constant 2000 U.S. dollars for all available years between 1996 and 2004. Rescaled to trillions in the regressions to make results more legible. Source: World Development Indicators, World Bank, http:// devdata.worldbank.org/dataonline/. GDP per capita Constant U.S. dollars in 2000 for all available years between 1996 and 2004. Rescaled to thousands in the regressions to make results more legible. Source: World Development Indicators, World Bank. Financial development Private credit divided by total GDP for all available years between 1996 and 2004. Source: World Development Indicators, World Bank. Natural resources Percentage of ore, metals and fuels in total exports for all available years between 1996 and 2004. Source: World Development Indicators, World Bank. Openness Sum of imports and exports divided by total GDP for all available years between 1996 and 2004. Source: World Development Indicators, World Bank.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

389

Human capital Percentage of total population over 25 that attended at least some secondary schooling. Sources: Barro and Lee (2001) available from World Bank Education Indicators (EDSTATS), http://devdata.worldbank.org/edstats/td10.asp. Data refer to 1995 and 2000 for a vast majority of countries and to 1990 for a smaller set of countries (values for the U.S.S.R. attributed to Russia). Transition Countries that belonged to the former Soviet Union, former Yugoslavia, or ex-communist countries. Source: Global Development Network Growth Database, http://www.nyu.edu/fas/ institute/dri/dataset/Social%20Indicators%20Fixed%20Factors_7_2005.xls. Market capitalization as a share to GDP All available years between 1996 and 2004. Source: World Development Indicators, World Bank. Financing constraints Number of years in default between 1970 and 2001. A country is in default if any of the following sources considers it in default. Detragiache and Spilimbergo (2001), Manasse and Roubini (2005), and Reinhart et al. (2003). The independent variables used in the baseline regressions and in most specifications are, unless otherwise noted, a time-series mean of the variables of interest between 1996 and 2004, whenever available. Time-series means of variables are only formed using data for the years for which observations for the dependent variable are available. To enlarge the number of years available, for the years 1997, 1999, and 2001 we attribute to a given country’s institutional quality index the value for that same country in the previous year (1996, 1998, and 2000, respectively). In the baseline regressions, we restrict the sample to the set of countries for which we have information for all key variables in at least one year between 1996 and 2004. Instruments Logarithm of settler mortality: for former colonies. Source: Acemoglu et al. (2001). Logarithm of population density in the 1500s: for former colonies. Source: Acemoglu et al. (2001). Ethnolinguistic fractionalization: Probability that two randomly selected persons from a given country will not belong to the same ethnolinguistic group. Source: Mauro (1995). Countries The baseline sample used in the regressions consists of the following 94 countries: Algeria, Argentina, Australia*, Austria*, Bangladesh, Belgium*, Benin, Bolivia, Botswana, Brazil, Bulgaria T, Cameroon, Canada*, Chile, China, Colombia, CroatiaT, Czech RepublicT, Denmark*, Dominican Republic, Ecuador, Egypt, El Salvador, EstoniaT, Ethiopia, Fiji, Finland*, France*, Germany*, Ghana, Greece*, Guatemala, Haiti, Honduras, HungaryT, Iceland*, India, Indonesia, Iran, Ireland*, Italy*, Jamaica, Japan*, Jordan, KazakhstanT, Kenya, Kuwait*, LatviaT, LithuaniaT, Malawi, Malaysia, Mali, Mexico, MoldovaT, Mozambique, Namibia, Nepal, New Zealand*, Nicaragua, Niger, Norway*, Pakistan, Papua New Guinea, Paraguay, Peru, PolandT, Portugal*, Romania, Russia, Rwanda, Senegal, Slovak Republic T, SloveniaT*, South Africa, South Korea*, Spain*, Sri Lanka, Sudan, Swaziland, Sweden*, Syria, TajikistanT, Thailand, Togo, Trinidad and Tobago, Tunisia, Turkey, Uganda, United Kingdom*, United States*, Venezuela, Vietnam, Zambia, Zimbabwe.

390

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

The full sample (129 economies) includes the 94 countries listed above plus the following: AlbaniaT, Angola, ArmeniaT, AzerbaijanT, BelarusT, Bosnia and HerzegovinaT, Burkina Faso, Cambodia, Chad, Democratic Republic of Congo, Republic of Congo, Coˆte d’Ivoire, Equatorial Guinea, Gabon, GeorgiaT, Guinea, Kyrgyz RepublicT, Lao, Libya, MacedoniaT, Madagascar, Morocco, Myanmar, Nigeria, Oman, Qatar*, Saudi Arabia, Taiwan (Province of China), Tanzania, Turkmenistan, UkraineT, United Arab Emirates*, UzbekistanT, Yemen, YugoslaviaT. Countries marked with * are excluded from the non-high income sample. Countries marked with T are transition countries. The classification of countries according to the income level and transition from communism follows the Global Development Network Growth Database. We exclude from the sample all countries considered offshore financial centers (as listed in Table 1 of International Monetary Fund (2002b): Bahrain*, Costa Rica, Cyprus, Hong Kong (S.A.R. of China)*, Israel*, Malta*, Mauritius, Netherlands*, Panama, Philippines, Singapore*, Switzerland*, and Uruguay. In some specifications, we follow a more stringent interpretation of the table and also exclude Ireland, Japan, Malaysia, Portugal, Thailand, United Kingdom, and United Statesdthese countries either have banking facilities or parts of their territory in which offshore activities are allowed. References Abiad, A., Detragiache, E., Tressel, T., 2008. A New Database of Financial Reforms. Unpublished. IMF. Acemoglu, D., Johnson, S., Robinson, J., Tchaicharoen, Y., 2003. Institutional causes, macroeconomic symptoms: volatility, crises and growth. Journal of Monetary Economics (Carnegie-Rochester Conference Series) 50 (1), 125–131. Acemoglu, D., Johnson, S., 2003. Institutions, corporate governance, and crises. In: Cornelius, P., Kogut, B. (Eds.), Global Issues in Corporate Governance, Risk, and International Investment. Oxford University Press. Acemoglu, D., Johnson, S., Robinson, J., 2001. The colonial origins of comparative development: an empirical investigation. American Economic Review 91 (5), 1369–1401. Acemoglu, D., Johnson, S., Robinson, J., 2004. Institutions, volatility, and crises. In: Ito, T., Rose, A. (Eds.), Growth and Productivity in East Asia. NBER East Asia Seminar on Economics (EASE), vol. 13. University of Chicago Press. Albuquerque, R., 2003. The composition of international capital flows: risk sharing through foreign direct investment. Journal of International Economics 1 (2), 353–383. Alfaro, L., Kalemli-Ozcan, S., Lolosovych, V., 2008. Why doesn’t capital flow from rich to poor countries? An empirical investigation. Review of Economics and Statistics 90 (2), 347–368. Barro, R., Lee, J., 2001. International data on educational attainment: updates and implications. Oxford Economic Papers 53 (3), 541–563. Bekaert, G., Harvey, C., Lundblad, C., 2005. Does financial liberalization spur growth? Journal of Financial Economics 77 (1), 3–55. Be´nassy-Que´re´, A., Coupet, M., Mayer, T., 2005. Institutional Determinants of Foreign Investment. CEPII Working Paper 2005–05. Centre d’etudes prospectives et d’informations internationales, Paris. Blonigen, B., Wang, M., 2005. Inappropriate pooling of wealthy and poor countries in empirical FDI studies. In: Moran, T., Graham, E., Blomstro¨m, M. (Eds.), Does Foreign Direct Investment Promote Development? Institute for International Economics, Washington DC. Booth, L., Aivazian, V., Demirguc-Kunt, A., Maksimovic, V., 2001. Capital structures in developing countries. Journal of Finance 56 (1), 87–130. Borensztein, E., De Gregorio, J., Lee, J., 1998. How does foreign direct investment affect economic growth? Journal of International Economics 45 (1), 115–135. Borensztein, E., Chamon, M., Jeanne, O., Mauro, P., Zettelmeyer, J., 2004. Sovereign Debt Structure for Crisis Prevention. IMF Occasional Paper 237, Washington DC. Bulow, J., Rogoff, K., 1989. Sovereign debt: is to forgive to forget? American Economic Review 79 (1), 43–50. Cole, H., English, W., 1991. Expropriation and direct investment. Journal of International Economics 30 (3–4), 201–227. Cole, H., English, W., 1992. Direct investment: a doubtful alternative to international debt. Federal Reserve Bank of Minneapolis Quarterly Review 16 (1), 12–22. Cole, H., Kehoe, P., 1995. The Role of institutions in reputation models of sovereign debt. Journal of Monetary Economics 35 (1), 45–64. Detragiache, E., Spilimbergo, A., 2001. Crises and LiquiditydEvidence and Interpretation. IMF Working Paper 01/2. Eaton, J., Gersovitz, M., 1981. Debt with potential repudiation: theoretical and empirical analysis. Review of Economic Studies 48 (2), 289–309. Eaton, J., Gersovitz, M., 1984. A theory of expropriation and deviations from perfect capital mobility. Economic Journal 94 (373), 16–40. Fama, E., MacBeth, J., 1973. Risk return and equilibrium: empirical tests. Journal of Political Economy 81 (3), 607–636. Fan, J., Titman, S., Twite, G., 2006. An International Comparison of Capital Structure and Debt Maturity Choices. AFA Philadelphia Meetings. Available at SSRN: http://ssrn.com/abstract=423483. Faria, A., Mauro, P., 2004. Institutions and the External Capital Structure of Countries. IMF Working Paper 04/236. Glaeser, E., La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2004. Do institutions cause growth? Journal of Economic Growth 9 (3), 271–303.

A. Faria, P. Mauro / Journal of International Money and Finance 28 (2009) 367–391

391

Hart, O., Moore, J., 1994. A theory of debt based on the inalienability of human capital. Quarterly Journal of Economics 109 (4), 841–879. Hausmann, R., Ferna´ndez-Arias, E., 2000. Foreign Direct Investment: Good Cholesterol? Inter-American Development Bank Research Department Working Paper 416. International Monetary Fund, 2002a. International Investment Position: A Guide to Data Sources. International Monetary Fund, Washington DC. Available from: http://www.imf.org/external/np/sta/iip/guide/index.htm. International Monetary Fund, 2002b. Offshore financial centers. IMF Background Paper. Available from: http://www.imf.org/ external/np/mae/oshore/2000/eng/back.htm. Johnson, S., Boone, P., Breach, A., Friedman, E., 2000. Corporate governance in the Asian financial crisis. Journal of Financial Economics 58 (1–2), 141–186. Knack, S., Keefer, P., 1995. Institutions and economic performance: cross-country tests using alternative institutional measures. Economics and Politics 7 (3), 207–227. Kaufmann, D., Kraay, A., Mastruzzi, M., 2006. Governance Matters V: Governance Indicators for 1996–2005. World Bank Policy Research Working Paper 4012. Kaufmann, D, Kraay, A., Zoido-Lobato´n, P., 1999. Governance Matters. World Bank Policy Research Working Paper 2196. Kaufmann, D., Kraay, A., 2004. Governance Indicators, Aid Allocation, and the Millennium Challenge Account. Development and Comparative Systems No. 0405013 Economics Working Paper Archive at WUSTL. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. Journal of Political Economy 106, 1113–1155. Lane, P., 2004. Empirical perspectives on long-term external debt. Topics in Macroeconomics 4 (1). Available from: http://www. bepress.com/bejm/topics/vol4/iss1/art1. Lane, P., Milesi-Ferretti, G.M., 2001a. The external wealth of nations: measures of foreign assets and liabilities for industrial and developing countries. Journal of International Economics 55 (2), 263–294. Lane, P., Milesi-Ferretti, G.M., 2001b. External capital structure: theory and evidence. In: Siebert, H. (Ed.), The World’s New Financial Landscape: Challenges for Economic Policy. Springer-Verlag. Lane, P., Milesi-Ferretti, G.M., 2006. The External Wealth of Nations Mark II: Revised and Extended Estimates of Foreign Assets and Liabilities, 1970–2004. IMF Working Paper 06/69. Levine, R., Loayza, N., Beck, T., 2000. Financial intermediation and growth: causality and causes. Journal of Monetary Economics 46 (1), 3–77. Lim, E., 2001. Determinants of, and the Relation Between, Foreign Direct Investment and Growth: A Summary of the Recent Literature. IMF Working Paper 01/175. Lothian, J., 2006. Institutions, capital flows, and financial integration. Journal of International Money and Finance 25 (3), 358–369. Manasse, P., Roubini, N., 2005. ‘Rules of Thumb’ for Sovereign Debt Crises. IMF Working Paper 05/42. Markusen, J., 1997. Trade Versus Investment Liberalization. NBER Working Paper 6231. Mauro, P., 1995. Corruption and growth. Quarterly Journal of Economics 110 (3), 681–712. Mauro, P., 2004. The persistence of corruption and slow economic growth. IMF Staff Papers 51 (1), 1–18. Myers, S., 2001. Capital structure. Journal of Economic Perspectives 15 (2), 81–102. Papke, L., Wooldridge, J., 1996. Econometric methods for fractional response variables with an application to 401(k) plan participation rates. Journal of Applied Econometrics 11 (6), 619–632. Przeworski, A., 2004. Some Historical, Theoretical, and Methodological Issues in Identifying Effects of Political Institutions. Unpublished. New York University. Rajan, R., Zingales, L., 1995. What do we know about capital structure? Some evidence from international data. Journal of Finance 50 (5), 1421–1460. Razin, A., Sadka, E., Yuen, C., 1998. Pecking-order of capital inflows and international tax principles. Journal of International Economics 44 (1,), 45–68. Reinhart, C., Rogoff, K., Savastano, M., 2003. Debt intolerance. Brookings Papers on Economic Activity Spring (1), 1–74. Rogoff, K., 1999. Institutions for reducing global financial instability. Journal of Economic Perspectives 13 (4), 21–42. Schnitzer, M., 2002. Debt v. foreign direct investment: the impact of sovereign risk on the structure of international capital flows. Economica 69, 41–67. Wei, S.J., 2000a. How taxing is corruption on international investors? Review of Economics and Statistics 82 (1), 1–11. Wei, S.J., 2000b. Local corruption and global capital flows. Brookings Papers on Economic Activity 2, 303–354. Wei, S.J., 2001. Domestic crony capitalism and international fickle capital: is there a connection? International Finance 4 (1), 15–45. Wei, S.J., Wu, Y., 2002. Negative alchemy? Composition of capital flows, and currency crises. In: Edwards, S., Frankel, A. (Eds.), Preventing Currency Crises in Emerging Markets. The University of Chicago Press, pp. 461–506.