Trust and the accumulation of physical and human capital

European Journal of Political Economy 27 (2011) 507–519

Contents lists available at ScienceDirect

European Journal of Political Economy j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e j p e

Trust and the accumulation of physical and human capital Jacob Dearmon a,⁎, Robin Grier b a b

Department of Economics and Finance, Meinders School of Business, Oklahoma City University, Oklahoma City, Oklahoma 73106, United States Department of Economics, 729 Elm Avenue, University of Oklahoma, Norman, Oklahoma 73019, United States

a r t i c l e

i n f o

Article history: Received 8 July 2009 Received in revised form 19 February 2011 Accepted 2 March 2011 Available online 13 March 2011

JEL classification: O15 O43 O50

a b s t r a c t Recent empirical work has shown that trust plays an important role in economic development. In this paper, we delve deeper into the mechanism behind that relationship. Specifically, we investigate the effect of trust on human and physical capital while controlling for the fact that the two types of capital are simultaneously determined. In a sample of 50 countries from 1976 to 2005, we show that trust has a positive and significant effect on human capital and a non-linear effect on physical capital. Increasing trust in a low-trust country has a greater impact on the accumulation of physical capital than an identical increase in trust in a high-trust country. We go on to investigate the interaction between institutions and trust and find that institutional reform is less effective at promoting investment in countries with high levels of trust. © 2011 Elsevier B.V. All rights reserved.

Keywords: Trust Physical capital Human capital Development Investment Education

1. Introduction There is a growing consensus among economists that trust plays an important role in economic development. Both cross sectional and panel research have verified that countries with higher levels of trust also have higher per-capita incomes on average. This begs the question as to how trust actually promotes economic growth. Rather than focusing directly on the relationship of trust and economic growth, as much of the prior research has done, we dig deeper by investigating trust's role in the accumulation of human and physical capital. We estimate an econometric model of the determination of human and physical capital, allowing for simultaneous spillovers between the two. Our results indicate that the two types of capital are indeed simultaneously determined, and that trust is significantly related to the accumulation of both types of capital. We go on to show that while trust has a direct and positive impact on human capital, trust's effect on physical capital investment is non-linear. The relationship between trust and physical capital is characterized by diminishing marginal returns, which implies that increasing trust in a low-trust country will have a greater impact than an identical increase in trust for a high-trust country. We also find an interesting interaction effect between trust and institutions, whereby the effectiveness of institutional reform at promoting investment depends on a country's overall level of trust. Besides our key findings about trust, we also discover some other interesting results about the determinants of human and physical capital in our sample. We show that government spending and political instability negatively affect physical capital, while lagged real GDP growth and openness increase physical capital investment. Ethnic diversity is negatively associated with human capital, while

⁎ Corresponding author. Tel.: +1 405 208 5518. E-mail address: [email protected] (J. Dearmon). 0176-2680/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.ejpoleco.2011.03.001

508

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

dummy variables representing Protestantism and Catholicism are both positive and significant in the human capital equation. The joint endogeneity of human and physical capital means that the variables that are included in the physical (human) capital equation also indirectly affect human (physical) capital. We calculate the overall effect of these variables to determine their long-run relationship with capital. Our paper is organized as follows. Section 2 discusses the relationship between trust and human and physical capital. Section 3 outlines our empirical model, discusses the other independent variables in the system and details our identifying assumptions. Section 4 examines the results and analyzes the equilibrium quantitative effects implied by these outcomes. In Section 5, we analyze the robustness of the results by employing a jackknife. We conclude with some closing remarks in Section 6. 2. Trust and capital While much of the literature on trust has emphasized the relationship between it and economic growth, there has been some work detailing how trust may affect investment in human and physical capital. Bjørnskov (2009), for example, provides several ways in which education decisions are related to trust. First, as Coleman (1988) finds in an examination of the role of social capital on educational outcomes, higher levels of social capital both within the family and outside the family are associated with lower high school dropout rates.1 As Bjørnskov points out, human capital should be higher in social capital-rich communities due to the fact that non-kin individuals would be willing to help children other than their own, as they trust that other parents would eventually reciprocate the favor. Putnam (2000) finds that even in communities with high levels of material wealth, education of children may be poor if the adults’ social networks are deficient. There are also labor market reasons for believing that trust and education are related. Firms typically hire workers with high levels of human capital to perform complex tasks. Monitoring costs escalate as the complexity of these tasks increase. Under higher trust levels, these monitoring costs are reduced increasing the firm's demand for human capital rich workers. Further, under higher trust levels, workers with high levels of human capital may be better able to cooperate, work, and share information with other workers, thus raising the firm's return to hiring workers with higher levels of human capital.2 Empirically, previous research has confirmed a positive relationship between trust and educational outcomes. La Porta et al. (1997) find that increasing trust by one standard deviation produces a one half standard deviation increase in the percentage of graduates from high school. Likewise, Knack and Keefer (1997) report a positive relationship between secondary education and trust for a cross section of 28 countries. Bjørnskov (2009) finds additional support for this result in an extended cross sectional sample of 52 countries covering the period 1960–2000. There are also reasons to believe that the level of trust is positively related to the accumulation of physical capital. If both the quality and quantity of information are increased under trust, then firms would not only know about a larger variety of investment opportunities, but they could also more accurately assess their chance of success. Under higher levels of trust, the need for extensive contracts and the probability of outcomes ending in expensive litigation would be reduced as deals might be sealed with as little as a handshake. Higher trust levels would increase investment through trust-induced efficiency gains at the microeconomic level. Zak and Knack (2001) examine the aggregate impact of trust on physical capital using a cross section of 41 countries and find that trust increases investment.3 While some research has studied the link between trust and capital accumulation, all have done so in the context of a single equation model. Our model is unique in that we study trust while taking into account the possibility that the two types of capital may be simultaneously determined. In the next section, we discuss the motivation for modeling trust's effect on factor accumulation using a system of simultaneous equations and we explain the various measures used to identify those equations. 3. An empirical model of trust and capital 3.1. The relationship between human and physical capital There is a well-developed theoretical literature on the link between human and physical capital. Nelson and Phelps (1966) discuss how education fosters the ability to innovate and assimilate new technology. Likewise, Fishlow (1966) argues that physical capital accumulation in the US was driven by high levels of education in the 1900s. Romer (1993) argues that endogenous growth theory can help formalize the claim that idea gaps (a lack of human capital necessary to sustain economic growth) and object gaps (a lack of physical capital) are related to one another. Countries that lack one type of capital tend to lack the other one as well. If policy could be used to close the idea gap, it would generate external benefits by reducing the object gap.

1 Social capital refers to the interpersonal relationships that exist between people and other individuals in their communities. Bjørnskov (2006) finds that the main component of social capital for the areas of governance and life satisfaction is trust. Paldam and Svendsen (2000) argue that trust is the key component of social capital. 2 Yamamura (2009) examines the relationship between trust and human capital in the growth of the Japanese garment industry in Kojima, Japan over the period 1968 to 2005. 3 While it is possible that capital also influences trust, several authors (Uslaner (2008), Tabellini (2008), and Algan and Cahuc (2010)) note that trust exhibits highly stable behavior over extended periods of time. Trust's stability contrasts sharply with that of physical capital accumulation, which is highly variable over time, a result that suggests physical capital does not cause trust. As for human capital, Bjørnskov (2009) finds that the direction of causality runs from trust to education.

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

509

Caballe and Santos (1993) and Graça et al. (1995) show that increases in physical capital raise the return to education producing a positive spillover effect on the level of human capital. Upadhyay (1994) develops a model where technological innovation increases the demand for new types of human capital. Lucas (1993) and Greiner (1999) note that increases in physical capital must be matched by increases in human capital for growth in per-capita income to be sustainable. However, the literature is more limited with regards to the empirical link between human and physical capital. Benhabib and Spiegel (2005) verify a modified version of the Nelson and Phelps hypothesis that more educated countries grow faster. Grier (2002, 2005) models the relationship between human and physical capital in a system of simultaneous equations. Grier (2002) examines a panel of Latin American countries from 1965 to 1990, while Grier (2005) examines a panel of Sub-Saharan African countries from 1970 to 2000. In both cases, she finds that human and physical capital are simultaneously determined. 3.2. Data on trust, human capital, and physical capital We construct a simultaneous model of human and physical capital for a sample of 50 countries from 1976 to 2005. As is common in the literature (cf., Knack and Keefer (1997)), we measure trust with the question A165 from the World Values Survey (WVS) (European Values Study Group and World Values Association, 2006) dataset, which asks “Generally speaking would you say that most people can be trusted or that you can't be too careful in dealing with people?” Individual responses are either: (1) Most people can be trusted or (2) Can't be too careful in dealing with people. We re-coded all responses of (2) with a value of zero. We then average all the individual responses for each country and wave. Wave 1 was conducted over years 1981–1984, wave 2 over years 1989–1993, wave 3 over years 1994–1998, and wave 4 over years 1999–2004.4 Our human capital variable is taken from the Barro and Lee (BL) (2001) dataset. It represents the percentage of people over 15 who have reached or attained the secondary education level. This data is recorded at five-year intervals. To best accommodate the temporal structure of the trust variable, we select the first observation on educational attainment after the corresponding WVS wave.5 We measure investment in physical capital with data on investment share of real GDP taken from the Penn World Tables (PWT) (Heston et al., 2006). The annual frequency of this variable is transformed into the corresponding frequency of the World Values Survey (WVS) (European Values Study Group and World Values Survey Association, 2006) trust data by averaging over the time periods for which each survey wave took place. Therefore, data affiliated with survey wave 1 are averages from 1981 to 1984, data associated with survey wave 2 are averages from 1989 to 1993, data connected with survey wave 3 are averages from 1994 to 1998, and data associated with survey wave 4 are averages from 1999 to 2004. Estimating simultaneous systems can be difficult, especially in our case where the two left-hand side variables measure different types of capital. To properly identify the system, we need to find variables that are significantly related to one form of capital but not to the other. In the physical capital equation, we include lagged real GDP growth, lagged inflation, regime change, lagged government spending, trade openness, and trade liberalization. To identify the physical capital equation, we assume that the lag of inflation, regime change, lagged government spending, trade openness, and trade liberalization will have a direct effect on the accumulation of physical capital but not human capital. In the human capital equation, we include lagged real GDP growth, religion dummies, and a measure of ethno-linguistic fractionalization. To identify this equation, we assume that religion and ethno-linguistic fractionalization directly affect human capital and not physical capital. Below we discuss the variables in detail and provide theoretical and empirical support for our over-identifying assumptions.6 3.3. Macroeconomic indicators We include several macroeconomic variables in our model to capture the current investment environment for physical capital. First, we control for any potential effect that lagged inflation might have on investment in physical capital. The theoretical and empirical evidence (cf. Temple (2000)) paints a mixed picture of inflation's role in the development process. The traditional Phillips curve shows a negative relationship between inflation and unemployment, and if a decrease in unemployment is associated with an increase in economic growth (as suggested by Okun's law), it implies a positive relationship between inflation and output.7 To the extent that investment and output are highly correlated, this also suggests a positive effect of inflation on investment. In addition, the Tobin (1965) effect also implies a positive relationship between inflation and investment. Under higher levels of anticipated inflation, individuals will hold less in real money balances and invest more in a portfolio of assets. Alternatively, high inflation may create more uncertainty with regard to the profitability of investment, and as a result, investors may delay investment decisions. Servén (1997) finds that inflation uncertainty has a negative impact on investment. Pindyck and Solimano (1993) show a negative relationship between inflation and investment, but Crosby and Otto (2000) find that inflation has a positive impact on the capital stock in a sample of 34 countries. While inflation may cause investors to delay making new investments in physical capital, it is unlikely that inflation would cause the same kind of strategic behavior among young students starting secondary school. It is hard to see how inflation uncertainty would cause 12 or 13 year olds to delay starting middle school for a later date. For that reason, we 4

Therefore, each country has at most four, irregularly spaced observations in the WVS data. Appendix A lists which countries are in each wave. Specifically, we use educational attainment levels in 1985 for wave 1, levels in 1995 for wave 2, and levels in 1999 for wave 3. Wave 4 data comes from 2005 levels found in Barro and Lee (2010). 6 Note that previous research (Grier (2002, 2005)) has also employed similar overidentification assumptions in a model of physical and human capital. 7 With the exception of the Korean war and two oil shock periods occurring in the 70s and 80s, Yoon (2006) finds little evidence for this hypothesis using Korea as his sample. 5

510

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

assume that inflation has a direct effect on physical capital but only indirectly affects human capital. We calculate inflation for each wave and country based on the International Financial Statistics (IFS) (International Monetary Fund, 2007) CPI data from the IMF. Second, we control for the overall investment and educational climate by including lagged real GDP growth rates in our model. Using PWT data, the real GDP growth rate is calculated by taking the annual percentage change in the level of real GDP. These percentage changes are then averaged over each wave. Higher GDP growth reflects an optimistic economic outlook, which in turn, should spur new investment in physical capital, while simultaneously increasing the opportunity cost of acquiring human capital. Under periods of economic growth, attending school is more expensive due to lost wages that could have been obtained in the labor market. Third, while much of the literature on trade openness and development focuses on economic growth, there is also empirical support for the hypothesis that openness also positively affects investment levels.8 Seghezza and Baldwin (2008) present several reasons why increased openness to world trade should raise the accumulation of physical capital in a country. First, they argue that tradables tend to be more capital intensive than non-tradables. Combined with the fact that trade liberalization is likely to favor the tradables sector, countries that liberalize trade should have increases in the stock of their physical capital. Second, trade openness will likely raise demand for a country's tradable goods, which leads to increased demand for capital. Last, liberalization can raise the competitiveness of the domestic banking system by offering customers financial services from abroad. This would lower the cost of capital and make it relatively more attractive. Empirically, Wacziarg and Welch (2008) and Giavazzi and Tabellini (2005) find that trade liberalization is associated with higher overall levels of trade and investment. Levine and Renelt (1992) find in a sample of 119 countries from 1960 to 1989 that the empirical link between trade and investment is one of the few robust relationships that they can identify. They go on to argue that any positive relationship between trade and economic growth is probably a result of trade's direct effect on investment.9 The relationship between education and trade openness, however, is much less definitive. Sachs and Warner (1995) show that economically closed countries do not have lower average levels of education. Likewise, Harrison (1996) uses a couple of different measures of trade openness and finds no robust statistical relationship between them and education. For that reason, we assume that trade openness directly affects physical, but not human, capital. We use two different measures to test if trade openness is an important determinant of physical capital. The first variable is Wacziarg and Welch's (2008) updated version of the Sachs and Warner's (1995) openness variable. The original Sachs and Warner measure considers a country closed if at least one of the following criteria are met (Wacziarg and Welch 2008, p. 190): “1) average tariff rates of 40% or more, 2) nontariff barriers covering of 40% or more of trade, 3) black market exchange premium at least 20% lower than the official exchange rate, 4) state monopoly on major exports, 5) A socialist economic system.” Wacziarg and Welch update this variable by extending the time period to the 1990s.10 Using their updated measure, we construct a dummy variable called trade liberalization that takes the value of one if a country is, or becomes, open during a wave and zero otherwise. In addition to this binary indicator, we also include a variable that measures trade volume as well. Our openness variable comes from the PWT dataset (where it is called openk) and is calculated as imports plus exports divided by real GDP. Finally, we control for the effect of government spending. Grier and Tullock (1989) find a negative relationship between the growth of government's share of real GDP and GDP growth in a sample of OECD countries. Likewise, Barro (1991) identifies a negative relationship between economic growth and the overall size of government. Alesina et al. (2002) find that government spending depresses private investment. Giannaros et al. (1999), using a sample of 18 OECD countries, show that increases in the size of government are negatively related to investment.11 We measure government spending using PWT data on government's share of real GDP. Annual observations for each country are averaged over periods corresponding to survey waves. We also construct initial observations for the period 1976–1980 because we enter all of the macroeconomic variables into the system with a one-period lag. 3.4. Political instability We include a measure of political instability in our model to control for the negative effect that instability might have on new investment. If political instability creates an environment of uncertainty, investors will think twice before embarking on any long-term investment projects. In addition, in many developing regions of the world, the secondary markets for goods are scarce, making investments even harder to reverse (see, for example, Collier and Gunning (1999)). A lack of secondary markets would only exacerbate the effect of uncertainty on investment. Berthélémy and Söderling (2001) find that instability in the political environment has a significant and negative impact upon investment. Campos and Nugent (2002) show that regime instability has a negative contemporaneous impact on investment, and demonstrate that the causality runs from instability to investment and not the other way around. Svensson (1998), in a sample of around a 100 countries, finds that political instability depresses investment by lowering the enforcement of property rights. Le and

8 Bloom and Sachs (1998) find trade openness to be an important component of economic development in the Sub-Saharan Africa. Frankel and Romer (1999) and Alesina et al. (2005) both show a positive relationship between openness and economic growth. Rodriguez and Rodrik (2001), however, find no significant statistical relationship between trade and capital accumulation in a sample of 80 countries. 9 More recently, Razin et al. (2003) employ a two-country model with lumpy new investment costs to show that trade liberalization can cause a large increase in investment, but can also lead to investment volatility. In their model, a certain threshold must be met before a firm will invest. This threshold, which depends on both the price of the foreign good and the fixed cost of investment, changes when a country opens to trade, thus producing discrete jumps in the level of investment. 10 In addition, there are some other minor differences between their measure and the original variable, which are noted in their paper. 11 While most of the literature on government spending focuses on investment, it is theoretically possible that it could also directly affect human capital if overall spending is significantly related to spending on education. We re-estimated all of our models with government spending in both equations and found the variable to be consistently insignificant in the education equation. For that reason, we feel comfortable including it only in the physical capital equation.

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

511

Zak (2006) show that political instability is the largest reason behind capital flight. Altug et al. (2007) and Nishide and Nomi (2009) find that even the threat of a regime change can depress investment. Le (2004) shows that unconstitutional regime changes negatively affect private investment. Jong-A-Pin (2009), using exploratory factor analysis on 25 different political instability measures, finds four underlying factors: politically related violence, corruption, within political regime instability, and political regime instability. Of these four factors, he notes that only political regime instability is consistently and negatively related to economic growth. Since the empirical link between political instability and physical capital has been found to be robust, and the relationship between human capital and political instability is much more ambiguous both from a theoretical perspective and empirical perspective, we include regime change only as a regressor in the physical capital equation. Based on prior empirical evidence, we expect to find a negative coefficient on regime change. To measure political instability, we construct a variable, called regime change, which is based on the polity2 variable from the Polity IV (Marshall and Jaggers, 2004) database.12 Regime change takes the value 1 if there is a three-point movement in the polity2 variable at some point during a three year period and 0 otherwise. The regime change occurs in the first year that the three-point change takes place unless the regime change occurs between waves. If an inter-wave regime change occurs less than three years prior to the next wave, then the regime change takes the value of one for the next wave.13

3.5. Cultural variables Empirical work has shown that differences in religious beliefs or denominations can lead to differences in economic outcomes (c.f., Tomes (1984)). In particular, a body of research has demonstrated that Protestantism is correlated with better economic outcomes such as higher income levels or faster economic growth (c.f., Grier (1997)). Woodberry and Shah (2004) provide a detailed discussion of how Protestantism influences one of the primary engines of economic growth, namely the accumulation of human capital. From an empirical standpoint, Becker and Wößssmann (2008), using Prussian census data from 1816, find that higher shares of Protestants diminished the educational gap between genders. Becker and Wößssmann (2009) show that Protestantism's impact upon literacy can explain the entire variation in economic outcomes for a sample of 452 Prussian counties from the 1870s and 1880s. That is, Protestantism acts through human capital in influencing economic outcomes. Stark (2005), however, disputes the notion that there is a significant relationship between Protestantism and education, arguing instead that Christianity in general promoted education. Gallego and Woodberry (2008) show that increases in competition between Catholic and Protestant missionaries raised schooling in African provinces. Ewing (2000) finds that Catholics earn higher wages than non-Catholics and attributes this partially to gains in human capital from being Catholic. To our knowledge, there are no papers that test for an empirical relationship between religious denomination and investment in physical capital. For that reason, we construct two dummy variables for Protestantism and Catholicism with data from the Shared Global Indicators (SGI) (2005) dataset and include them only in the human capital equation. The variables are equal to one if the majority of the population is Protestant or Catholic, and zero otherwise.14 Ethno-linguistic diversity is also thought to be important to educational outcomes and development in general. Easterly and Levine (1997) find that ethnic diversity decreases human capital in Africa. Alesina et al. (1999) also identify lower public amounts of schooling in more ethnically diverse regions. Alesina and La Ferrara (2005) construct a model where the public provision of goods is lower under higher levels of ethnic fractionalization and find empirical support for this argument in a large sample of countries from 1960 to 2000. Poterba (1997), using a panel of US states covering the period from 1960 to 1990, finds that increasing the fraction of elderly residents decreases educational spending on a per-child basis. This effect becomes even larger if the children and elderly people are from different racial groups. Grob and Wolter (2007) discover similar results for a sample of Swiss Cantons for the years 1990 to 2002. As the fraction of elderly increases it has a negative effect on the provision for public schooling. Miguel (1999) examines Kenyan schools and finds a negative relationship between schooling and ethnic diversity. The effect of ethno-linguistic fractionalization is more clear-cut than its effect on the accumulation of physical capital. One reason for this ambiguity is that the literature has emphasized the relationship between ethnic diversity and income growth rather than investment and even those yield mixed results (for instance Lian and Oneal, 1997 and Nettle, 2000). Since we are already controlling for the uncertainty caused by economic and political instability in the physical capital equation, we believe that ethno-linguistic fractionalization should only directly impact education and not investment. For that reason, we include a widely used metric of ethnolinguistic fractionalization, elf, only in the human capital equation. Elf is defined as the probability that any two people from a country will not have the same ethno-linguistic background and is taken from the SGI dataset. Based on the results reported in the empirical literature, we expect that elf would have a negative impact on secondary education.

12 Polity2 is an indicator that takes into account the extent that a government exhibits autocratic and democratic qualities. It ranges from − 10 (most autocratic) to 10 (most democratic). 13 This particular variable measures major changes in the polity score occurring over short periods of time. Rapid transformations in governance would generate increased uncertainty in the economy and lead to lower investment levels, regardless of whether that movement is towards more democratic or autocratic regimes. We tested and found that splitting this variable into democratic and autocratic movements does not significantly change the results. 14 Berggren and Jordahl (2006), in an examination of the effect of economic freedom on the development of generalized trust, find that hierarchical religions such as Catholicism are negatively associated with trust.

512

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

Table 1 A model of trust and capital. Model 1 Investment Constant

0.125*** (0.039)

Investment Secondary education lag(real GDP growth) lag(inflation) Regime change lag(Govt. spending as % of GDP) Trade liberalization Openness

0.238** (0.101) 0.82*** (0.21) 0.003** (0.002) −0.029** (0.014) −0.338*** (0.072) −0.008 (0.015) 0.025** (0.011)

Protestant dummy Catholic dummy Ethno-linguistic Diversity Trust Trust

0.051 (0.051)

Model 2 Education 0.175*** (0.066) 0.662** (0.312)

−0.788* (0.461)

0.076*** (0.026) 0.063*** (0.024) −0.141*** (0.051) 0.272*** (0.078)

2

R2 N

0.515 115

0.499 115

Investment 0.072** (0.036)

0.242** (0.094) 0.813*** (0.207) 0.004** (0.002) −0.028** (0.013) −0.291*** (0.068) −0.007 (0.014) 0.027*** (0.01)

0.305** (0.142) −0.351** (0.159) 0.517 115

Education 0.156** (0.064) 0.814*** (0.296)

−0.887* (0.458)

0.078*** (0.026) 0.059** (0.023) −0.130*** (0.048) 0.245*** (0.077)

0.499 115

Notes: Standard errors are in parentheses. ***, **, * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. We also include wave dummies that are not reported.

4. Results 4.1. Model Table 1 presents the results of two different models. Model 1 uses trust as a linear regressor in both the education and investment equations whereas Model 2 also incorporates a trust squared term into the investment equation to capture any diminishing returns that might be occurring. We estimate both of these models with three stage least squares (3SLS), a procedure which addresses the cross correlation between the residuals in the two equations. The magnitude of the estimated cross-equation error correlation for each model is greater than one third, indicating that 3SLS is an appropriate modeling choice. We also test whether our over-identifying restrictions are valid with a Hansen–Sargan test using a J-statistic. The J-statistic is distributed as a χ2 with (L–K) degrees of freedom, where L is the number of instruments and K is the number of endogenous variables.15 We calculate and report this statistic for all of our estimations below. 4.2. Coefficient signs and significance Model 1 of Table 1 demonstrates that investment and education are significantly related to one another in our sample. Increases in investment are positively and significantly related at the 0.05 level to increases in secondary school attainment levels. Likewise, the coefficient on secondary education is also positive and significant at the 0.05 level in the investment equation. This result has important implications for policymakers because it indicates that increasing one form of capital will have spillover effects on the other type as well.16 We find that political instability, as measured by regime change, negatively and significantly impacts investment, while lagged average inflation has a significant and positive effect on investment. Countries with more government spending (as a % of GDP) also tend to have lower investment rates on average. While liberalization of trade is insignificant in both models, the openness measure is positive and statistically significant. Thus, the loosening of trade barriers does not significantly raise investment, but increasing the volume of trade does. The coefficient on lagged real GDP growth rate is statistically significant in the investment equation. 15

Note that the Hansen–Sargan test assumes a valid identification and only tests for whether the system is properly overidentified. Model 1's J-test statistic, which is distributed as a with 6 degrees of freedom, is 10.722. We cannot reject the null hypothesis that the over identifying restrictions are valid at the 0.05 level. 16

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

513

Fig. 1. The effect of trust on investment.

In the secondary education equation, we find that the religious dummy variables, Protestant and Catholic, are positive and significant for both models. The coefficient on elf is negative and significant at the 0.01 level in the secondary education equation, indicating that countries with higher levels of ethno-linguistic diversity also tend to have lower levels of secondary schooling on average. For the purposes of our paper, Model 1's most interesting result is that the level of trust has a positive and statistically significant relationship with education but not with investment. Higher trust countries tend to have higher levels of secondary education on average, but not significantly higher levels of investment. The lack of significance does not necessarily mean that the effect of trust is unimportant in the accumulation of physical capital. It is possible that there are diminishing returns to trust.17 For example, a low-trust developing country where property rights are not adequately enforced may benefit more from gains in trust than a high-trust developed country where property rights are more protected. To test for this possibility, we re-estimate Model 1 with the inclusion of a squared trust term in the investment equation. Model 2 reports these results, showing that the coefficients on trust and trust2 are both statistically significant in the investment equation. The coefficient on the level of trust is positive and significant, while the coefficient on the squared term is negative and significant (both at the 0.05 level).18 This indicates that as the level of trust increases its impact on physical capital accumulation becomes less pronounced. Furthermore, while these diminishing returns occur in the investment equation, they also spill over and affect secondary education.

4.3. The equilibrium quantitative effects of trust on capital We established above that trust is statistically significant in a model of investment and secondary education, but those results do not tell us anything about the economic importance of trust to capital. In this section, we calculate the quantitative effect of our independent variables on investment and secondary education to have a better idea of which variables play an important role in the evolution of capital. Since investment and education are jointly determined, there are indirect spillover effects that occur between them. Those spillover effects mean that variables that are excluded from the investment (education) equation will still indirectly affect investment (education). For example, while religion is only included in the education equation, it still has an indirect effect on investment through its direct effect on secondary education. In order to better understand the quantitative effect of trust on educational and investment outcomes, we conduct a bootstrap exercise consisting of 5000 regression runs for each of the 100 trust values that lie between 0.01 and 1, increasing the level of trust by 0.01 increments. For each regression run, we calculate Model 2's reduced form coefficients and combine them with trust and the mean value of all other variables in order to obtain the predicted values of investment and secondary education. The bootstrapped mean and bias-corrected confidence interval are constructed for each trust value from the results of 5000 bootstrapped regressions. Figs. 1 and 2 present the bootstrapped results. Fig. 1 shows trust's effect on investment, while Fig. 2 shows trust's effect on secondary education. These results reveal that trust has a positive, yet diminishing, effect on both investment and education. Put another way, our findings suggest that increasing trust will produce a larger effect in low-trust countries than in high-trust ones. For example, suppose a low-trust country increased its trust level from 5% to 21%, which corresponds to a movement of one 17 Roth (2009) also finds a nonlinear effect of trust on economic growth. At the individual level, Butler et al. (2009) find that the relationship between trust and income is humped-shaped, indicating that low-trust and high-trust individuals earn less income than those whose trust levels are closer to the population average. 18 Model 2's J-test statistic, which is distributed as a with 7 degrees of freedom, is 12.406. We fail to reject the null hypothesis that the over identifying restrictions are valid at the 0.05 level.

514

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

Fig. 2. The effect of trust on secondary education.

standard deviation (and is comparable to moving from Peru's trust level to France's trust level).19 According to our results, this increase in trust would raise investment by 80.7% of a standard deviation and education by 60.2% of a standard deviation. In contrast, the same one standard deviation increase in trust is less effective at increasing investment and education for a country operating at the mean (33%) level of trust. For instance, this increase, which is similar to moving from Austria's level of trust to New Zealand's level, only increases investment and education by 22.8% and 37.4% of a standard deviation, respectively. Further, the confidence intervals surrounding the mean are wider at the ends than in the middle denoting more uncertainty at low and high values of trust. A low trust country would be associated with a wide confidence interval on the left side of the graph. Moving one standard deviation (0.16) to the right would not only produce a better outcome, but also a more certain one as denoted by the narrowing confidence interval. Several other independent variables also yield large quantitative effects on investment and education. The religion dummy variables have the greatest quantitative impact on secondary education. Moving from a situation where the Protestant (Catholic) dummy is equal to zero to a scenario where it is equal to one is associated with an increase in secondary education of 68.2% (51.5%) of a standard deviation and an increase in investment by 34.1% (25.7%) of a standard deviation. Regime change also has an important economic effect on education and investment. The occurrence of a regime change during a survey wave decreases secondary education by 20.1% and investment by 51%.20

4.4. Institutional reform and trust It is possible that developing countries possess lower quality institutions on average than developed countries. If this is true, reducing institutional deficiencies may increase investment. As an example, reducing corruption could lead to higher levels of investment by decreasing the cost of doing business, while increasing the role of law and order may create a more favorable investment climate by enhancing the security of property rights. As we have found above, trust also plays an important part in shaping the investment climate by lowering transaction costs. Therefore, the effect of reforming institutions on the level of investment could potentially depend on the level of overall trust in a society.21 To the extent that trust is high, transaction costs might already be relatively low and institutional reform would be less effective in increasing investment.22 Ahlerup et al. (2009) analyze this complex relationship between institutions and trust and its effect on investment and economic growth. They modify the Knack and Keefer (1997) or Zak and Knack (2001)-style cross sectional regression model to include an institutional variable and an interaction term between that institutional variable and trust. This interaction term between trust and institutions is negative and statistically significant, meaning that increases in trust generate higher levels of investment when institutional quality is low. 19 Since trust is non-linear there is a difference between a standard deviation movement in trust and the marginal effect of changing trust at a single point. A standard deviation movement in trust requires us to calculate the change in the levels. 20 Elf decreases secondary educational attainment by 27% of a standard deviation and lowers investment by 13.5% of a standard deviation. Lagged real GDP growth increases investment by 26.1% of a standard deviation and reduces secondary education by almost 4.8% of a standard deviation. Lagged government's share of real GDP depresses investment in human and physical capital by 13.6% and 34.4% of a standard deviation respectively. Increasing trade volume or openness increases investment by 21.7% of a standard deviation and secondary education by 8.6% of a standard deviation. 21 While Hug and Spörri (2011) find that referendums can increase the link between trust and tax morale, the strength of this effect depends on institutions. Bjørnskov (2010) also examines the complex relationship between trust and institutions. 22 Bjørnskov et al. (2010) find that the effect of institutions on subjective well-being differs depending on the level of economic development.

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

515

Table 2 Trust and institutions. Model 3

Model 4

Investment Constant

0.009 (0.040)

0.18*** (0.06) 0.693*** (0.259)

Investment Secondary education

0.24** (0.096) 0.869*** (0.201) 0.004*** (0.001) −0.026** (0.013) −0.341*** (0.068) −0.019 (0.015) 0.017* (0.01)

lag(real GDP growth) lag(inflation) Regime change lag(Govt. spending as % of GDP) Trade liberalization Openness

−0.834* (0.450)

Protestant dummy

0.078*** (0.027) 0.056** (0.024) −0.142*** (0.047) 0.255*** (0.075)

Catholic dummy Ethno-linguistic Diversity Trust Institution Institution × trust R2 N

Education

0.272*** (0.09) 0.034*** (0.008) −0.065*** (0.017) 0.573 115

0.499 115

Investment 0.030 (0.051)

0.255*** (0.105) 0.677*** (0.209) 0.004*** (0.002) −0.028** (0.013) −0.339*** (0.07) −0.017 (0.015) 0.019* (0.01)

0.283** (0.118) 0.025*** (0.007) −0.055** (0.023) 0.549 115

Education 0.188*** (0.062) 0.608*** (0.285)

−0.772* (0.455)

0.079*** (0.026) 0.06** (0.024) −0.143*** (0.048) 0.27*** (0.076)

0.499 115

Notes: Standard errors are in parentheses. ***, **, * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. We also included wave dummies that are not reported.

In this section we examine the trust-investment relationship but in the framework of our simultaneous system. In particular, we add an institutional measure into Model 1's investment equation, both individually and as an interaction term with trust. We use two different variables as a measure of institutional quality, both of which are taken from the International Country Risk Guide (ICRG) (2006) Table 3 Jackknife coefficient distributions for Model 2. Mean

Std dev.

Min

Max

Investment equation Constant Secondary education lag(real GDP growth) lag(inflation) Regime change lag(govt. spending as % of GDP) Trade liberalization Openness Trust Trust2

0.072 0.241 0.811 0.004 −0.028 −0.291 −0.007 0.027 0.304 −0.351

0.010 0.028 0.067 0.000 0.003 0.018 0.003 0.001 0.025 0.026

0.053 0.096 0.412 0.003 −0.040 −0.344 −0.020 0.024 0.215 −0.432

0.116 0.302 0.951 0.005 −0.021 −0.239 0.003 0.032 0.371 −0.280

Education equation Constant lag(real GDP growth) Protestant dummy Catholic dummy Ethno-linguistic diversity Investment Trust

0.155 −0.885 0.078 0.059 −0.129 0.816 0.245

0.013 0.080 0.007 0.005 0.009 0.057 0.014

0.124 −1.043 0.057 0.040 −0.152 0.692 0.216

0.188 −0.467 0.096 0.071 −0.109 0.984 0.284

Each coefficient was estimated 50 times using N−1 countries in the sample.

516

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

dataset. The first variable is corruption, which takes a value of zero when corruption is highest and 6 when corruption is at its lowest. The second variable is law and order, which takes a value of six for the highest level of law and order and a value of zero for the lowest. Model 3 of Table 2 presents the results of including corruption, both individually and as an interaction variable with trust, in the investment equation. We find that reducing corruption is more effective at increasing the level of investment when trust is lower. As the level of trust rises, reducing corruption is less useful in growing investment. For example, decreasing the level of corruption at the lowest level of trust in our sample produces a 56.5% of a standard deviation increase in investment, and indirectly, an 18.9% of a standard deviation increase in education. If we examine this same movement in corruption occurring at the mean value of trust, we find that institutional reform produces a 22.6% of a standard deviation increase in investment and a 7.6% of a standard deviation increase in education. When we re-estimate the equation using law and order as a measure of institutional quality, we find similar results. Model 4 shows that improvements in law and order have less of an impact on investment levels when trust levels are higher. Increasing law and order at the lowest trust value produces a 39.7% of a standard deviation increase in investment and an 11.7% of a standard deviation increase in education. At the mean value of trust in our sample, this effect becomes much smaller as a one unit increase in law and order yields 11.3% of a standard deviation change in investment and 3.3% of a standard deviation change in education. 5. Robustness We test the robustness of our main results as found in Model 2 to changes in the underlying sample by performing a jackknife experiment. We run fifty 3SLS regressions with each regression having a different country removed. From these results, we construct a distribution of coefficient estimates. Table 3 lists summary statistics associated with the outcomes of these regressions. This jackknife exercise reveals several interesting results. First, all of the coefficient estimates for investment and secondary education are positive and tightly clustered about their means. The coefficient on investment in the education equation has a mean of 0.82. Regression runs generate a series of investment coefficient values that do not deviate very far from the mean with a maximum value of 0.984 and a minimum value of 0.692. This can also be seen by looking at the standard deviation associated with this coefficient, which is 0.057. Thus, the standard deviation of investment's coefficient is just under 7% of the mean. If we perform a similar analysis on the coefficient of secondary education in the investment equation, we find that the standard deviation of education's coefficient is 11.6% of the mean. Second, the experiment also indicates that our results on trust are robust.23 The coefficient on trust is never less than zero in any of the specifications, meaning that trust has a positive effect on education and investment in all 50 regressions. Likewise, the coefficient on the quadratic term is always negative. The trust coefficients are also tightly clustered about their means. The standard deviation on trust's estimated coefficient in the investment equation is 8.3% of its mean value, while the standard deviation of the estimated coefficient in the education equation is even lower at just 5.6% of the mean. The standard deviation of the quadratic term is 7.5% of its mean value.24 6. Conclusion In this paper, we explore the channels through which trust affects development by analyzing the relationship between it and human and physical capital. We find that human and physical capital are simultaneously determined, in that increases in one form of capital have a positive and significant effect on the other. We go on to show that trust is significantly related to human and physical capital. As the level of trust increases, its marginal impact on capital decreases, meaning that policies that increase trust would be most helpful in low trust countries. We also studied the interaction between institutions and trust and find that institutional reform is less effective at increasing the level of investment in high-trust countries. In addition to our results on trust, we find that dummy variables representing Protestantism and Catholicism are positively and significantly related to secondary education, while ethno-linguistic diversity is negatively related to education. Consistent with expectations, both political stability and a healthy economic climate are found to increase investment. Further work could investigate how policymakers could construct policies to best increase generalized trust. In particular, new research could build on recent contributions by Dohmen et al. (2008), Danielson and Holm (2007), and Chaudhuri and Gangadharan (2007) in identifying the determinants of trust formation. Knowing the key determinants of trust may allow policymakers to create more trust and promote economic development.25 Acknowledgements The authors would like to thank Kevin Grier, Ben Keen, Carlos Lamarche and two anonymous referees for valuable comments. 23

Each of the three coefficients on trust is significant at the 10% level or better in 100% of the jackknife regressions. The Protestant and Catholic variables have small standard deviations and their means do not include zero. The macroeconomic variables also have small standard deviations about their mean, reflecting the tight fit of the coefficient estimates. Thus, our results are robust to small changes in the underlying sample. 25 Other recent interesting work regarding trust and development includes Chan (2007) and Gustavsson and Jordahl (2008). 24

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

517

Appendix A1 Countries and waves. Country name

Algeria Argentina Australia Austria Bangladesh Belgium Brazil Canada Chile China Colombia Denmark Dominican Republic Egypt El Salvador Finland France Greece Hungary India Indonesia Iran Ireland Israel Italy Japan Jordan Korea, Republic of Mexico Netherlands New Zealand Norway Pakistan Peru Philippines Poland Portugal Singapore South Africa Spain Sweden Switzerland Tanzania Turkey Uganda United Kingdom United States Uruguay Venezuela Zimbabwe

Wave number 1

2

3

X X

X

X X

4 X X

X

X X X

X X

X X X X

X

X

X X X X

X X X

X

X X X X X

X X

X X

X X

X

X

X X

X X

X X

X X X

X

X

X X

X X X X X X X X X

X X

X X

X X

X X X X X X X X X X X X X X X

X X X X

X X X X

X

X

X X

X X X X

X X X X X X X X X X X X X X X X

Appendix A2 Summary statistics. Variable

Obs

Mean

Std. dev.

Min

Max

Investment Education Lag real GDP growth Regime Change Lag gov't. spending (% real GDP) Lag Inflation Trade Liberalization Openness Protestant Catholic

115 115 115 115 115 115 115 115 115 115

0.195 0.423 0.020 0.130 0.196 0.629 0.904 0.577 0.287 0.504

0.069 0.142 0.024 0.338 0.065 2.749 0.295 0.449 0.454 0.502

0.033 0.060 −0.047 0.000 0.073 0.006 0.000 0.103 0.000 0.000

0.400 0.799 0.121 1.000 0.516 22.824 1.000 3.960 1.000 1.000 (continued on next page)

518

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

Appendix A2 (continued) Variable

Obs

Mean

Std. dev.

Ethno-linguistic diversity Trust Trust2 Corruption Rule of law

115 115 115 115 115

0.292 0.327 0.134 3.998 4.600

0.237 0.164 0.119 1.419 1.310

Min 0.002 0.028 0.001 0.542 1.722

Max 0.930 0.665 0.443 6.000 6.000

References Ahlerup, P., Olsson, O., Yanagizawa, D., 2009. Social capital vs institutions in the growth process. European Journal of Political Economy 25, 1–14. Alesina, A., La Ferrara, E., 2005. Ethnic diversity and economic performance. Journal of Economic Literature 43, 762–800. Alesina, A., Baqir, R., Easterly, W., 1999. Public goods and ethnic divisions. Quarterly Journal of Economics 114, 1243–1284. Alesina, A., Ardagna, S., Perotti, R., Schiantarelli, F., 2002. Fiscal policy, profits, and investment. The American Economic Review 92, 571–589. Alesina, A., Spolaore, E., Wacziarg, R., 2005. Trade, growth, and the size of countries. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth Edition 1, vol. 1. North Holland, Amsterdam, pp. 1499–1542. Algan, Y., Cahuc, P., 2010. Inherited trust and growth. The American Economic Review 100, 2060–2092. Altug, S., Demers, F., Demers, M., 2007. Political risk and irreversible investment. CESifo Economic Studies 53, 430–465. Barro, R., 1991. Economic growth in a cross section of countries. Quarterly Journal of Economics 112, 1251–1288. Barro, R., Lee, J., 2001. International data on educational attainment: Updates and implications. Oxford Economic Papers 53, 541–563. Barro, R., Lee, J., 2010. A new data set of educational attainment in the world, 1950–2010. NBER Working paper no. 15902. NBER, Massachusetts. Becker, S., Wößssmann, L., 2008. Luther and the girls: religious denomination and the female education gap in nineteenth-century Prussia. Scandinavian Journal of Economics 110, 777–805. Becker, S., Wößssmann, L., 2009. Was Weber wrong? A human capital theory of the protestant economic history. Quarterly Journal of Economics 124, 531–596. Benhabib, J., Spiegel, M., 2005. Human capital and technology diffusion. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth Edition 1, vol. 1. North Holland, Amsterdam, pp. 935–966. Berggren, N., Jordahl, H., 2006. Free to trust: economic freedom and social capital. Kyklos 59, 141–169. Berthélémy, J., Söderling, L., 2001. The role of capital accumulation, adjustment, and structural change for economic take-off: empirical evidence from African growth episodes. World Development 29, 323–343. Bjørnskov, C., 2006. The multiple facets of social capital. European Journal of Political Economy 22, 22–40. Bjørnskov, C., 2009. Social trust and the growth of schooling. Economics of Education Review 28, 249–257. Bjørnskov, C., 2010. How does social trust lead to better governance? An attempt to separate electoral and bureaucratic mechanisms. Public Choice 144, 323–346. Bjørnskov, C., Dreher, A., Fischer, J., 2010. Formal institutions and subjective well-being: Revisiting the cross-country evidence. European Journal of Political Economy 26, 419–430. Bloom, D., Sachs, J., 1998. Geography, demography, and growth in Africa. Brookings Papers on Economic Activity 2, 207–289. Butler, J., Guiliano, P., Guiso, L., 2009. The right amount of trust. NBER Working paper no. 15344. NBER, Massachusetts. Caballe, J., Santos, M., 1993. On endogenous growth with physical and human capital. Journal of Political Economy 101, 1042–1067. Campos, N., Nugent, J., 2002. Who is afraid of political instability? Journal of Development Economics 67, 157–172. Chan, K., 2007. Trade, social values and the generalized trust. Southern Economic Journal 73, 733–753. Chaudhuri, A., Gangadharan, L., 2007. An experimental analysis of trust and trustworthiness. Southern Economic Journal 73, 959–985. Coleman, J., 1988. Social capital in the creation of human capital. The American Journal of Sociology 94, S95–S120. Collier, P., Gunning, J., 1999. Explaining African economic performance. Journal of Economic Literature XXXVII, 64–111. Crosby, M., Otto, G., 2000. Inflation and the capital stock. Journal of Money, Credit, and Banking 32, 236–253. Danielson, A., Holm, H., 2007. Do you trust your brethren? Eliciting trust attitudes and trust behavior in a Tanzanian congregation. Journal of Economic Behavior and Organization 62, 255–271. Dohmen, T., Armin, F., Huffman, D., Sunde, U., 2008. Representative trust and reciprocity: prevalence and determinants. Economic Inquiry 46, 84–90. Easterly, W., Levine, R., 1997. Africa's growth tragedy: policies and ethnic divisions. Quarterly Journal of Economics 112, 1203–1250. European Values Study Group and World Values Survey Association, 2006. European and world values surveys four-wave integrated data file, 1981–2004, v.20060423. http://www.worldvaluessurvey.org/. Ewing, B., 2000. The wage effects of being raised in the Catholic religion: does religion matter? American Journal of Economics and Sociology 59, 419–432. Fishlow, A., 1966. Levels of nineteenth-century American investment. The Journal of Economic History 26, 418–436. Frankel, J., Romer, D., 1999. Does trade cause growth? The American Economic Review 89, 379–399. Gallego, F., Woodberry, R., 2008. Christian missionaries and education in former colonies: how institutions mattered. Documentos de Trabajo, 339. Instituto de Economia. Pontificia Universidad Católica de Chile, Chile. Giannaros, D., Kolluri, B., Panik, M., 1999. An empirical analysis of the effects of government spending on capital investment: evidence from O.E.C.D. countries. International Economic Journal 13, 45–55. Giavazzi, F., Tabellini, G., 2005. Economic and political liberalizations. Journal of Monetary Economics 52, 1297–1330. Graça, J., Saqib, J., Philippopoulos, A., 1995. Interaction of human and physical capital in a model of endogenous growth. Economics of Planning 28, 93–118. Greiner, A., 1999. Endogenous growth through externalities of investment: a different approach. Atlantic Economic Journal 27, 86–90. Grier, R., 1997. The effect of religion on economic development. Kyklos 50, 47–62. Grier, R., 2002. On the interaction of human and physical capital in Latin America. Economic Development and Cultural Change 50, 891–913. Grier, R., 2005. The interaction of human and physical capital accumulation: evidence from Sub-Saharan Africa. Kyklos 58, 195–211. Grier, K., Tullock, G., 1989. An empirical analysis of cross-national economic growth, 1951–1980. Journal of Monetary Economics 24, 259–276. Grob, U., Wolter, S., 2007. Demographic change and public education spending: a conflict between young and old? Education Economics 15, 277–292. Gustavsson, M., Jordahl, H., 2008. Inequality and trust in Sweden: some inequalities are more harmful than others. Journal of Public Economics 92, 348–365. Harrison, A., 1996. Openness and growth: a time series-cross country analysis for developing countries. Journal of Development Economics 48, 419–447. Heston, A., Summers, R., Aten, B., 2006. Penn World Table Version 6.2. Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania. Hug, S., Spörri, F., 2011. Referendums, trust, and tax evasion. European Journal of Political Economy 27, 120–131. International Country Risk Guide, 2006. International Country Risk Guide. PRS Group http://www.prsgroup.com/. International Monetary Fund. International Financial Statistics [Online: Accessed Sept. 2007]. Washington D.C. Jong-A-Pin, R., 2009. On the measurement of political instability and its impact on economic Growth. European Journal Of Political Economy 25, 15–29. Knack, S., Keefer, P., 1997. Does social capital have an economic payoff? A cross-country investigation. Quarterly Journal of Economics 112, 1251–1288. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1997. Trust in large organizations. The American Economic Review 87, 333–338.

J. Dearmon, R. Grier / European Journal of Political Economy 27 (2011) 507–519

519

Le, Q., 2004. Political and economic determinants of private investment. Journal of International Development 16, 589–604. Le, Q., Zak, P., 2006. Political risk and capital flight. Journal of International Money and Finance 25, 308–329. Levine, R., Renelt, D., 1992. A sensitivity analysis of cross-country growth regressions. The American Economic Review 82, 942–963. Lian, B., Oneal, J., 1997. Cultural diversity and economic development: a cross-national study of 98 countries, 1960–1985. Economic Development and Cultural Change 46, 61–78. Lucas, R., 1993. Making a miracle. Econometrica 61, 251–272. Marshall, M., Jaggers, K., 2004. Polity IV project: political regime characteristics and transitions. Miguel, T., 1999. Ethnic diversity, mobility, and school funding: theory and evidence from Kenya. The Development Economics Discussion Paper Series, 14. London School of Economics, London. Nelson, R., Phelps, E., 1966. Investment in humans, technological diffusion and economic growth. The American Economic Review 56, 69–75. Nettle, D., 2000. Linguistic fragmentation and the wealth of nations: the fishman-pool hypothesis reexamined. Economic Development and Cultural Change 48, 335–348. Nishide, K., Nomi, E., 2009. Regime uncertainty and optimal investment timing. Journal of Economic Dynamics and Control 33, 1796–1807. Paldam, M., Svendsen, G., 2000. An essay on social capital: looking for the fire behind the smoke. European Journal of Political Economy 16, 339-366. Pindyck, R., Solimano, A., 1993. Economic instability and aggregate investment. In: Blanchard, O., Fischer, S. (Eds.), NBER Macroeconomics Annual, vol. 8. MIT Press, Cambridge, MA, pp. 259–302. Poterba, J., 1997. Demographic structure and the political economy of public education. Journal of Policy Analysis and Management 16, 48–66. Putnam, R., 2000. Bowling Alone: The Collapse and Revival of American Community. Simon and Schuster, New York. Razin, A., Sadka, E., Coury, T., 2003. Trade openness, investment instability, and terms of trade volatility. Journal of International Economics 61, 285–306. Rodriguez, F., Rodrik, D., 2001. Trade policy and economic growth: a skeptic's guide to the cross-national evidence. In: Bernanke, B., Rogoff, K. (Eds.), NBER Macroeconomics Annual, vol. 15. MIT Press, Cambridge, MA, pp. 261–338. Romer, P., 1993. Idea gaps and object gaps in economic development. Journal of Monetary Economics 32, 543–573. Roth, F., 2009. Does too much trust hamper economic growth? Kyklos 62, 103–128. Sachs, J., Warner, A., 1995. Economic reform and the process of global integration. Brookings Papers on Economic Activity 1, 1–118. Seghezza, E., Baldwin, R., 2008. Testing for trade-induced investment-led growth. Economia Internazionale / International Economics 61, 507–537. Servén, L., 1997. Irreversibility, uncertainty and private investment: analytical issues and some lessons for Africa. Journal of African Economies Supplement 6, 229–268. Shared Global Indicators Cross National Database accessed via the internet at http://www.ksghome.harvard.edu/126pnorris/Data/Data.htm. Stark, R., 2005. The Victory of Reason: How Christianity Led to Freedom, Capitalism, and Western Success. Random House, New York. Svensson, J., 1998. Investment, property rights and political instability: theory and evidence. European Economic Review 42, 1317–1341. Tabellini, G., 2008. The scope of cooperation: values and incentives. Quarterly Journal of Economics 123, 905–950. Temple, J., 2000. Inflation and growth: stories short and tall. Journal of Economic Surveys 14, 395–426. Tobin, J., 1965. Money and economic growth. Econometrica 33, 671–684. Tomes, N., 1984. The effects of religion and denomination earnings and the returns to human capital. The Journal of Human Resources 19, 472–488. Upadhyay, M., 1994. Accumulation of human capital in LDCs in the presence of unemployment. Economica 61, 355–378. Uslaner, E., 2008. The foundations of trust: macro and micro. Cambridge Journal of Economics 32, 289–294. Wacziarg, R., Welch, K., 2008. Trade liberalization and growth: new evidence. The World Bank Economic Review 22, 187–231. Woodberry, R., Shah, T., 2004. The pioneering protestants. Journal of Democracy 15, 47–61. Yamamura, E., 2009. Dynamics of social trust and human capital in the learning process: the case of the Japan garment cluster in the period 1968–2005. Journal of Economic Behavior and Organization 72, 377–389. Yoon, J., 2006. The co-movement of inflation and the real growth of output. Journal of Korean Economy 7, 213–229. Zak, P., Knack, S., 2001. Trust and growth. The Economic Journal 111, 295–321.