Institutional infrastructure and economic growth in member countries of the Organization of Islamic Cooperation (OIC)

Economic Modelling 51 (2015) 214–226

Contents lists available at ScienceDirect

Economic Modelling journal homepage: www.elsevier.com/locate/ecmod

Institutional infrastructure and economic growth in member countries of the Organization of Islamic Cooperation (OIC) Ly Slesman a, Ahmad Zubaidi Baharumshah a,b,⁎, Wahabuddin Ra'ees c a b c

Department of Economics, Faculty of Economics and Management, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia Financial Economics Research Centre, Faculty of Economics and Management, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia Department of Political Science, Kulliyyah of Islamic Revealed Knowledge & Human Sciences, International Islamic University Malaysia, 53100 Gombak, Selangor, Malaysia

a r t i c l e

i n f o

Article history: Accepted 9 August 2015 Available online xxxx Keywords: Conflict-preventing institutions Dynamic panel threshold model Economic growth Economic institutions Organization of Islamic Cooperation Political institutions

a b s t r a c t This paper examines the relationship between the quality of different dimensions of institutional infrastructure and economic growth in a panel of 39 member countries of the Organization of Islamic Cooperation (OIC). The empirical results confirm that better-quality political and economic institutions can have positive effects on economic growth. All in all, the evidence from nonlinear model reveals that the quality of political institutions that ensure stable government, less expropriation, and low external conflict are the core dimensions of an institutional matrix because they influence the growth effects of economic institutions, confirming the “hierarchy of institutions hypothesis.” The study also finds that when political and economic institutions are accounted for, institutions that prevent internal conflict and tensions arising from ethnic and religious conflicts do not have significant (positive) impacts on growth. Thus, institutional reforms to upgrade the quality of both political and economic institutions are crucial for development in OIC countries. © 2015 Elsevier B.V. All rights reserved.

1. Introduction A general consensus holds that weak institutional infrastructure is the fundamental constraint on countries' ability to accumulate productive factors (e.g., physical and human capital) and to innovate and adopt new technology (North, 1981, 1990). Weak institutions inadequately support private economic activities because they lead to expropriation activities as a result of low constraints on executive power, judicial manipulation, entry barriers to new entrepreneurs and technologies, corruption, and inefficient bureaucracy (Asoni, 2008). The bulk of the literature shows that having well-functioning of broad institutions is fundamental to achieving economic growth (see, inter alia, Acemoglu et al., 2001; Banerjee and Iyer, 2005; Carlsson and Lundstrom, 2002; Dawson, 1998, 2003; De Haan et al., 2006; Doucouliagos and Ulubasoglu, 2006; Fedderke, 2001; Gwartney et al., 2006; Hall and Ahmad, 2012; Hall and Jones, 1999; Heckelman and Stroup, 2000; Knack and Keefer, 1995; Rodrik et al., 2004).1

⁎ Corresponding author at: Department of Economics, Faculty of Economics and Management, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia. Tel.: +60 3 89467247; fax: +60 3 89486188. E-mail address: [email protected] (A.Z. Baharumshah). 1 Huang's (2010) studies show that political institutions can affect the level of financial development and hence economic growth through the investment channels, suggesting that the extent of benefits from financial development depends on governance (see also Anwar and Cooray, 2012).

http://dx.doi.org/10.1016/j.econmod.2015.08.008 0264-9993/© 2015 Elsevier B.V. All rights reserved.

Despite important advancement on the topic in the literature, the question remains as to the relative importance of different dimensions of broad institutions in the growth process, because institutions are multidimensional and thus may have differential effects on economic growth. Although several studies have emerged that focus on specific effects of different dimensions of the institutional matrix on growth—for example, democracy (Narayan et al., 2011a), market-supporting institutions (Bhattacharyya, 2009; Rodrik, 2005), and other institutional risks (e.g., Nawaz, 2015)—only a few look at the relative importance of political and economic institutions (Acemoglu and Johnson, 2005; Aidt et al., 2008; Flachaire et al., 2014; Siddiqui and Ahmed, 2013). Moreover, the focus has been on global samples with different country characteristics. In addition, institutional indices are highly correlated (Langbein and Knack, 2010), which may have prevented the analysis on the relative influence of different institutional dimensions on growth in a single empirical framework.2 Recent studies (Narayan et al., 2014, 2015) have successfully dealt with this issue using principal component analysis to extract a single institutional indicator from three or four correlated institutional indices (as measured in the International Country Risk Guide [ICRG]) in order to study its effects on stock market

2 It is well known that institutions are a broad concept and that existing indicators are highly correlated, thus they do not provide a clear distinction between different sets of institutions. Our study deals with this issue and tries to provide a clear distinction between different sets of institutions and their impact on economic progress.

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

215

returns. In this paper, we follow and extend these studies by extracting different unique dimensions of institutions from all twelve ICRG institutional indices in our attempt to shed additional light on the importance of the relative effects of different dimensions of the broad institutional matrix on economic growth in both the developed and developing economies. The relative importance of different sets of institutions can be drawn from a recent theoretical view suggesting that political institutions are the most important dimension in the aggregate institutional matrix, the “hierarchy of institutions hypothesis” (see Acemoglu and Robinson, 2000, 2008; Acemoglu et al., 2005). This hypothesis argues that “political institutions ‘set the stage’ in which economic institutions can be devised” (Flachaire et al., 2014, p. 213). It conjectures that the emergence and persistence of equilibrium economic institutions—for example, institutions that protect private property rights and enforce contracts—depend on political institutions. Political institutions determine the allocation of political power and set the constraints on its usage among competing individuals or groups. Because economic institutional arrangements differ in their distribution of resources, these individuals or groups, using their relative political power allocated by political institutions, seek to shape equilibrium economic institutional arrangements (i.e., the rules of the game) that align with their preferred distribution of resources.3 Thus, political institutions determine the distribution of political power,4 which in turn shapes equilibrium economic institutions. Examples of political institutions include forms of government (e.g., democracy vs. dictatorship) and the extent of constraints it places on political power holders (e.g., politician and political elites); see Acemoglu et al., 2005. In this sense, economic institutional arrangements that promote growth may not be chosen when political institutions are weak (i.e., a high concentration of political powers is in the hands of a single or a few individuals, and these power holders are subject to weak constraints).5 This hierarchy of institutions hypothesis (HIH) has received little attention in the empirical growth literature. We attempt to provide empirical evidence for this hypothesis, using both linear and nonlinear empirical frameworks, with respect to the relative role that political, economic, and conflict-preventing institutions play in the growth process in member countries of the Organization of Islamic Cooperation (OIC). We focus on these countries because they share the same religion and have similar cultures and, especially in the Middle East and Africa, have wide variations in institutional quality and growth experience.6 Poverty levels in these countries are substantially lower than those countries with similar levels of income. One reason why the situation prevails is that the dual Islamic practice of zakat (Islamic obligatory charity) and sadaqa (voluntary charity) encourages the rich member of the society to donate a percentage of their income and wealth to the poor. These are large sums of money and explain why poverty rate in the Arab world are low relative to income levels. However, these countries frequently experience internal, ethnic, and

religious as well as external conflict, which impedes economic progress.7 Such environments are not conducive for productive investment to flourish. Furthermore, reports published by the World Bank and the International Monetary Fund (IMF) indicate that youth unemployment (15–25 years old) in these countries is among the highest in the world. Thus, this set of countries provides a unique sample for assessing the relative influence of different institutional dimensions on economic growth. In this study, we focus on a panel of 39 OIC countries over the period 1983–2009, to exclude the period of uncertainty associated with the post-Arab Spring regime,8 which started in late 2010. This was a historic movement in the politics of the Middle East and North Africa (MENA), but its long-term impact remains unpredictable. Many of these countries are still undergoing complex political, social, and economic transitions. In addition, because there seem to be no systematic studies focusing on OIC countries in the context of developing countries in general and in comparison to non-OIC countries, for our robustness check, we also compare the evidence on this issue in OIC countries to that of global and other developing countries.9 Our study makes at least three important contributions to the literature. First, this study provides new empirical evidence on the relative importance of political, economic, and conflict-preventing institutions for economic growth in OIC countries, employing (newly constructed) uncorrelated institutional indicators. In doing so, we extract three different unique dimensions of institutions from existing highly correlated indicators, using the principal component analysis (PCA) method. To provide a comparative analysis, we construct three additional panels—a sample consisting of 112 global countries, a sample of 88 developing countries, and a sample of 50 non-OIC developing countries. We look at these subpanels separately to see whether any clear pattern emerges from the empirical analysis. Second, based on the newly constructed institutional indicators, we provide empirical scrutiny for the HIH for OIC countries based on both linear and nonlinear dynamic panel growth frameworks. This empirical strategy allows us to investigate the threshold effects in the link between institutions and growth. A striking feature of our results is that economic institutions have an enhancing effect on economic growth only after political institutions cross a certain threshold level, below which they have no effect on growth. Specifically, our results suggest that the positive effect of economic institutions on growth is observed when political institutions exceed the 1.99–3.45 range (on scale of 0–10). Our study provides empirical evidence as to whether political institutions are a “deep determinant” in the growth process in the sense that they influence economic institutions (and policies), as suggested by the HIH. Recently, Flachaire et al. (2014), using a finite mixture regression model on a global sample of 79 countries, show that two growth regimes emerge from political institutions, which in turn condition the differential growth effects of economic institutions. Our study differs from this important study in its focus (39 OIC in the context of

3 Since there is conflict of interests (over the choice of economic institutions), the prevailing choice of economic institutional arrangements would emerge in accordance with the preference of those having more/predominant political power. 4 There are two types of political power de jure (institutional) and de facto (economic affluence) political power. 5 The hierarchy of institutions hypothesis argues that earlier models rely on entrenched vested interests to erect barriers to technological innovation and development using economic affluence (i.e. de facto political power) (Krusell and Rio-Rull, 1996; Parente and Prescott, 1999) are inadequate, because only those having de jure (institutional) political power will be able to erect those barriers (Acemoglu and Robinson, 2000). 6 North institutional framework includes both formal and informal institutions (North, 1981, 1990). Informal institutions include, for example, culture, religions, trust, or social capital. By focusing on OIC with relatively homogenous informal institutions, we are most likely isolating independent role play by formal institutions (e.g. political, economic and conflict-preventing institutions) in their growth process.

7 A glance at the data on International Country Risk Guide (ICRG) and its components reveals that the 39-OIC developing countries score relatively lower on overall institution risks (ICRG average score of 5.5 on 0–10 scale, with higher score indicating lower institutional risk or higher institution quality) compared to 50-non-OIC developing countries (average score of 6). Compared with non-OIC developing countries, OIC tends to have higher corruption (4 vs. 5 for non-OIC), higher degree of incompetent bureaucracy (4 vs. 4.8), experience higher conflicts (ethnic conflicts, 5.8 vs. 6.5, internal conflict, 6.2 vs. 6.8), lower observance of rule of law and order (5.3 vs. 5.8). 8 Arab Spring denotes a revolution wave of demonstrations and protests (both nonviolent and violent), riots, and civil wars in the Arab world that first started in Tunisia at the end of 2010. Events in Tunisia were soon spread to Egypt, Yemen, Syria, Libya and other countries in the region. Sakbani (2011) provides a historical perspective on the Arab Spring. 9 Most of the studies on the issue have pooled all the countries in one panel. Apart from the global sample (112 countries) we divide the developing country subsamples into developing countries, non-OIC developing countries and OIC countries to highlight the differences, if any, on the role of economic, political and conflict-preventing institutions on economic progress. We thank an anonymous referee for this suggestion.

216

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

89 developing countries and a 112-country global sample) and empirical approach (a dynamic panel linear model and a dynamic panel nonlinear threshold regression framework). In addition, as mentioned above, we also tackle the problem of high correlation among institutional indicators. Furthermore, we also provide evidence on the role that conflict-preventing institutions, along with political and economic institutions, play in the growth process in these focused sets of countries. Third, we account for weak and excessively numerous instruments that may have plagued past studies and were based on a dynamic panel system generalized method of moments (S-GMM).10 To this end, we follow the strategy suggested by Roodman (2009b) to deal with the problem of weak and excessively numerous instruments. Our empirical analysis, which focuses on the OIC member countries, has three major findings. First, political and economic institutions are important for economic growth in the long run. Second, in line with the HIH, we find evidence from nonlinear empirical framework that economic institutions affect growth only in the presence of political institutions at a sufficient level of quality. This new finding is the major contribution of our paper: the role of political institutions in relation to that of economic institutions with respect to growth in particular. Many previous studies have failed to capture this effect, as their findings tend to suggest that different dimensions of broad institutions (e.g., political and economic institutions) affect growth in the “same direct manner but only differ in the sizes of the effects,”11 and many such findings were based on highly correlated institutional data. Our results on a panel of OIC countries are consistent with a few recent studies (Aidt et al., 2008; Flachaire et al., 2014) that focus on a global sample of cross-sectional country data. For the specific case of OIC countries, ignoring the role of political institutions can lead to spurious results by suggesting that economic institutions have no statistically significant effect on economic growth. Third, when these characteristics of political institutions (e.g., stable government with strong legislative and popular support; friendly diplomatic relations with the international community) and economic institutions (e.g., rule of law that protects property rights, enforces contracts, and the presence and perfection of markets) are controlled in the growth regressions; institutions that prevent internal conflict (including ethnic and religious) have no direct significant effect on long-run economic growth.12 Importantly, these results are found to be robust when controlled for weak and excessively numerous instruments. In a comparative empirical analysis of 112 countries, 88 developing countries, and 50 non-OIC developing countries over the period 1983–2009, we found that the HIH is present only in developing countries, regardless of whether OIC countries are included in the empirical analysis. In the global sample, even after the correlated nature of institutional data is corrected, we do not find any significant and reliable evidence of the impact of any of these institutions on growth when developed/industrialized countries are pooled with developing countries in one panel. This may be due to the experience of rich countries, in which most of them achieve the maximum scores for institutional quality. Furthermore, unlike the OIC sample, in developing countries in general, we found that only political institutions influence growth. However, when all the OIC countries are excluded from the panel 10 In different GMM (D-GMM), when variable is persistent (e.g. institutions), its lag level perform poorly as instruments for its differenced series. Many past studies on institutions that use D-GMM (e.g. Siddiqui and Ahmed, 2013) may have suffered from this problem. SGMM overcomes the weak instruments problem through the use of additional moment conditions, but suffers from the problem of too many instruments as the number of instrument exponentially increases with time dimensions. Past studies that used S-GMM (e.g., Nawaz, 2015) may have suffered from too many instrument problems. This causes Hansen J and difference-in-Hansen identification tests to perform very poorly (Bowsher, 2002). 11 Some of these studies includes (Bhattacharyya, 2009; Nawaz, 2015; Siddiqui and Ahmed, 2013; among others). 12 This finding may relay well with the abovementioned report by World Bank and IMF about the highest youth unemployment in the world among many Arab countries undergoing the so-called Arab Spring. We conjecture that better political and economic institutions ensuring economic prosperity (e.g., jobs and income) may reduce conflicts.

(i.e., non-OIC developing country sample), conflict-preventing institutions as well as political institutions turn out to be significant. These comparative results suggest one important thing in common: political institutions are the fundamental factor that influences long-run growth in developing countries. Overall, we show, at least in the specific case of OIC countries, that better-quality political and economic institutions may just be the key to conflict reduction and higher growth, even in the absence of a conscious effort to improve conflict-preventing institutions. However, in non-OIC developing countries, what matters most are political and conflict-preventing institutions. It appears that high-quality political institutions are the most important ingredient in ensuring wellfunctioning economic institutions and reducing conflict in OIC countries, specifically, and developing countries, in general. The rest of the paper is organized as follows. Section 2 contains the empirical model and an explanation of our methodology, and Section 3 describes the data. Section 4 presents our empirical findings, and Section 5 concludes the paper. 2. Model specification and methodology The goal of this paper is to examine the effects of institutional infrastructure on output growth in OIC countries. To this end, we apply a standard growth model used in the growth literature (Barro, 1991; Islam, 1995; Levine and Renelt, 1992; Mankiw et al., 1992). Consider the following model: yit −yi; t−1 ¼ β1 yi; t−1 þ β2 INSit þ β03 X it þ ηi þ μ it

ð1Þ

Eq. (1) can also be written as follows: yit ¼ ð1 þ β1 Þyi; t−1 þ β2 INSit þ β03 X it þ ηi þ μ it

ð2Þ

or 0 ~ yit ¼ βy i; t−1 þ β 2 INSit þ β 3 X it þ ηi þ μ it

ð3Þ

where subscript i and t are the country and time index, respectively, y is the log transformation of real gross domestic product (GDP) per capita, INS measures the quality of the institutional infrastructure, X is a vector of other control variables hypothesized to affect output growth, ηi is a time-invariant unobserved country-specific effect term, and μit is the usual error term. The main control variables comprise the log of initial income, population growth, the investment ratio, trade, and a log of life expectancy. Eq. (3) forms the basis for our estimation. As in Islam ~ in Eq. (3) is expected to be positive if it indi(1995), the coefficient β

cates conditional convergent and negative if divergent.13 To estimate Eq. (3), we apply the general method of moments (GMM) proposed by Arellano and Bond (1991), Arellano and Bover (1995), Blundell and Bond (1998), and Holtz-Eakin et al. (1988). For application of this method in the growth regression context, see, for example, Caselli et al. (1996) and, more recently, Siddiqui and Ahmed (2013). The objective is to express Eq. (3) as a dynamic panel regression and then take the first difference to eliminate country-specific effects. Then, the regressors in the first-difference equation are instrumented using the right-hand-side series lagged two periods or more, under the assumption that time-varying disturbances in the original equations are not serially correlated. This strategy is known as a difference GMM (D-GMM) estimation. Besides controlling for country-specific effects, D-GMM is able to deal with simultaneity biases, which are a wellknown problem in growth regressions. For instance, variables in growth

13 Alternatively, when Eq. (1) is to be adopted, as many researchers in the literature did, then the sign on β1 would be negative to indicate convergent (or positive to indicate divergent).

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

regressions such as the investment rate (and, in our case, the quality of institutions) may create some endogeneity problems.14 Following Arellano and Bond (1991), Eq. (3) can be transformed into a first-difference equation to eliminate country-specific effects as follows:

 ~ y yit −yi; t−1 ¼ β i; t−1 −yi; t−2 þ β2 INSit −INS i; t−1
 þ β3 X it −X i; t−1 þ μ it −μ i; t−1

ð4Þ

Under the assumption that the error terms are not serially correlated and the lag of explanatory variables is weakly exogenous, the possibility of simultaneity bias and the correlation between (yi, t − 1 − yi, t − 2) and (μit − μi, t − 1) can be addressed following Arellano and Bond (1991), by using lagged levels of the right-hand-side variables as instruments in the following moment conditions: h  i E yi; t−s  μ it −μ i; t−1 ¼ 0 for s≥2; t ¼ 3; …; T

ð5Þ

h  i E INSi; t−s  μ it −μ i; t−1 ¼ 0 for s≥2; t ¼ 3; …; T

ð6Þ

h  i E X i; t−s  μ it −μ i; t−1 ¼ 0 for s≥2; t ¼ 3; …; T

ð7Þ

If variables are persistent, however, their past values convey little information about their future changes, making their lagged value a weak instrument for their differenced series (Acemoglu and Robinson, 2008). This may be the case for institution variables, which may lead to a biased estimation of parameters in small samples and larger variance asymptotically. Arellano and Bover (1995) suggested a combination of the differenced Eq. (4) and level Eq. (3). Blundell and Bond (1998) showed that this estimator is able to increase the efficiency via its reduction in biases, and imprecision characterized the D-GMM estimator, especially the abovementioned weak instrument problem. Arellano and Bover (1995) and Blundell and Bond (1998) proposed a system general method of moments (S-GMM) estimator as follows. In addition to the moment conditions of Eqs. (5)–(7), the authors proposed that the S-GMM uses the following moment conditions: 

 E yi; t−s −yi; t−s−1  ηi þ μ it ¼ 0 for s ¼ 1

ð8Þ



 E INSi; t−s −INSi; t−s−1  ηi þ μ it ¼ 0 for s ¼ 1

ð9Þ



 E X i; t−s −X i; t−s−1  ηi þ μ it ¼ 0 for s ¼ 1

ð10Þ

The consistency of these S-GMM estimators rests on the assumption that the error terms are not correlated, the instruments used are valid, and the changes with additional instruments are uncorrelated with country fixed effects,  
 E Δ Z i; t−s  ηi ¼ 0 where Z i; t−s ¼ yi; t−s ; INSi; t−s ; X i; t−s : Three test statistics are used to test for these assumptions. The first test examines the hypothesis that the second-order error term Δ μ it in Eq. (4) is probably correlated with the first order, but not the second order. Then, the validity of all instruments is tested using the Hansen J-test of overidentification restrictions (Hansen, 1982), which are robust to heteroscedasticity and autocorrelation. Finally, we apply the 14 The literature has been careful in noting and accounting for the fact that institutions and economic growth jointly cause each other. However, some criticisms remain, especially with regard to instruments used in the cross-sectional data context (e.g., Glaeser et al., 2004). In a panel-data context, though the GMM estimation method can deal with the endogeneity of institutions, a recent study also points to the fragility of the link between institutions and growth when weakness and a proliferation of instruments are corrected for (see Vieira et al., 2012). In this study, we are careful in dealing with weakness and a proliferation of instruments in our panel study using S-GMM equipped with Roodman’s (2009b) correction strategy.

217

difference-in-Hansen C test to determine the validity of additional instruments, in which the E[ΔZi, t − s ⋅ ηi] = 0. If the nulls of these identification tests are not rejected, the models are adequately specified. Using the moment conditions in Eqs. (5)–(10), we estimate using the two-step system GMM, which is more asymptotically efficient than the one-step estimator. One-step estimators use weighting matrices that are independent of the estimated parameters. Two-step GMM estimators, however, use optimal matrices in which the moment conditions are weighted by a consistent estimate of their covariance. However, the application of the two-step estimator in small samples, as in our study, has several problems (Arellano and Bond, 1991; Roodman, 2009b). These problems are triggered by the proliferation of instrumental variables, which quadratically increase as the time dimension T increases. This can cause the number of instruments to be very large relative to the number of countries N— that is, when the instrument ratio i = N/j b 1, j being the number of instruments, leading the specification tests mentioned above to be invalid (Roodman, 2009a,b). For example, Hansen J and difference-in-Hansen tests may give the implausible probability value of 1.000. Bowsher (2002) demonstrates that as T becomes large, the overidentification tests behave very poorly.15 On top of this, the problem is aggravated with i b 1—as Roodman (2009b) shows, the results obtained are susceptible to Type I error. This means that the GMM estimators tend to make the underlying insignificant relationship appear statistically significant.16 Roodman (2009b) suggests that we reduce the dimensionality of the instrumental variable matrix to overcome these problems.17 We follow this strategy to control for weak and excessively numerous instruments in our empirical assessment. This is to ensure that the gain from S-GMM over D-GMM with respect to the weak instrument problem is not complicated by a proliferation of instruments. Furthermore, doing so would also subject our results to nonsignificant findings, as recent evidence suggests that controlling for the weakness and proliferation of instruments using the Roodman strategy makes the impact of institutions on growth fragile; see Vieira et al. (2012).

3. Data and empirical strategy We estimate Eq. (3) using S-GMM estimators on a panel of 39 OIC countries over the period 1983–2009, which were selected based on the availability of reliable data.18 The period is divided into nine nonoverlapping 3-year periods (1983–1985, 1986–1988, . . . 2007–2009). Data for some countries are missing for some years, so we are dealing with an unbalanced panel. The data used in the analysis were collected from the Penn World Table (PWT, version 7.1) and the World Bank's World Development Indicators (WDI). The dependent variable is the log of per capita real (weighted chained series) GDP (LRGDPC), as reported in PWT. Accordingly, the variables used in our model can be classified into stock and flow variables (Islam, 1995). Stock variables are measured at the beginning of each non-overlapping 3-year period, while flow variables are measured as the average over each 3-year period. Stock variables consist of logged initial income (measured in 1983, 1986, . . . 2007 if the dependent variable is measured in 1986, 1988, . . . 2009). Flow variables consist of the real investment share of real GDP (INV) 15 The author shows that the test is undersized and never rejects the null of joint validity at 0.05 or 0.10, rather than rejecting it 5% or 10% of the time as a well-sized test would. 16 In a Monte Carlo simulation on 8 × 100 panels, Windmeijer (2005) shows that reducing the instrument count from 28 to 13 reduces the average bias in the two-step estimate of the parameter of interest by 40%. 17 The instrument matrix is collapsed so that each instrumenting variable generates only one instrument column for each lag rather than one column for each time period and lag available to that period (see Roodman, 2009a,b). 18 Appendix A lists all these countries. As already mentioned, the Arab Spring reflected the historic movement in the politics of the Arab countries in the MENA region but its long-term impact remains unpredictable; and that those countries (e.g., Syria, Yemen, and Egypt) are still undergoing complex political, social, and economic transitions.

218

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

(PWT), population growth (WDI), log of life expectancy (proxy for human and health capital), and the degree of trade openness measured as exports plus imports over GDP (OPEN). Finally, our focal (flow) variables are twelve components that measure the quality of institutional infrastructure in the International Country Risk Guide (ICRG) of the Political Risk Services Group.19 ICRG indices are available over relatively long periods of time and were rescaled to range from 0 to 10, where a higher score indicates better quality. These twelve component indices are law and order, bureaucratic quality, democratic accountability, ethnic tension, protection from religious tension, nonmilitarized politics, no corruption, external conflicts, internal conflicts, investment profiles, socioeconomic conditions, and government stability. In general, ICRG indices reflect the quality of most dimensions for institutions and fundamentally define the incentive structures of the economic agents to engage in productive activities, whether to invest, learn and imitate, innovate, or upgrade their technological capabilities.20 Specifically, in addition to longer time and country coverage, these indices have the advantage over alternative existing measures of providing a wide range of deep institutional characteristics reflected in these Muslim countries. One set of such indices measures the ethnic, religious, and other internal and external conflicts that are prevalent in the countries under investigation. Appendix B provides a detailed definition and the source of each variable. ICRG indicators have been widely used in the past to illustrate the in-depth quality of North's (1981, 1990) underlying institutions that support private contracts, facilitating economic transactions and the checks and balances that constrain expropriation by politically powerful elites and interest groups, which fundamentally define the investment climate in countries. However, it is well known that measuring the quality of institutions can be difficult. Existing indicators are imperfect and highly intercorrelated and usually lack clear conceptual dimensions that purport to reflect the underlying institutions.21 Using component measures of ICRG, several attempts have been made to conceptually distinguish various fundamental aspects of institutions that support a market economy (Rodrik, 2000, 2005), which were shown empirically to determine growth (Bhattacharyya, 2009). Such attempts were drawn from North's (1981, 1990) well-accepted theoretical framework on private property rights and contracting institutions (Acemoglu and Johnson, 2005). And many use ICRG along with the World Governance Indicators (WGI)22 of the World Bank, among others, to measure them. One important study (Acemoglu and Johnson, 2005) shows that North's contracting and property right institutions can cause long-run growth. Some recent researchers make use of statistical methods (e.g., factor analysis) to sort out the commonality and uniqueness of these various institutional indices into uncorrelated component factor indices (Knoll and Zloczysti, 2012; Langbein and Knack, 2010; Siddiqui and Ahmed, 2013). To extract distinct uncorrelated measures of the quality of different dimensions in an institutional matrix and have relatively clear conceptual distinctions to study their relative importance in influencing economic performance in Muslim countries in a common empirical framework, we follow their lead and address these issues by employing principle component analysis (PCA) on ICRG components and sort out the unique factors and employ them as our institutional variables. PCA explores and converts a set of highly intercorrelated variables into 19 Since ICRG is available from only 1984, the first 3-year average (1983–1985) is the average of 1984–1985. 20 It is desirable also to use the measure of economic freedom (EF) to reflect the underlying institutions, but EF is not available annually and is measured at 5-year intervals. Given the already small number of countries in our sample, we excluded this option. 21 Many of the critiques are directed at the six-dimension World Governance Indicators (WGI) of the World Bank (Knoll and Zloczysti, 2012; Langbein and Knack, 2010). WGI covers a relatively shorter period of time in 1996, 1998, 2000, and from 2002 to the present. 22 We exclude these WGI measures due to the unavailability of data prior to 1996.

Table 1 Principle component factor analysis on 12 ICRG subcomponents. Factors

Eigenvalue

Difference

Proportion

Cumulative

Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Observation

4.598 1.762 1.205 1.097 0.645 346

2.836 0.557 0.108 0.452 0.042

0.383 0.147 0.100 0.091 0.054

0.383 0.530 0.631 0.722 0.776

Source: Authors' calculation.

Table 2 Factor analysis. Factor loading after rotation Mean Std. Dev.

Variable

6.396 1.853 Government stability 4.454 1.586 Socioeconomic conditions 5.534 1.691 Investment profile 6.648 1.929 Internal conflict 7.443 1.743 External conflict 4.145 1.466 Corruption 4.733 2.479 Military in politics 5.882 2.393 Religious tensions 5.318 1.979 Rule of law and order 5.784 2.136 Ethnic tension 4.588 2.083 Democratic accountability 4.151 2.193 Bureaucratic quality

Factor 1: INS1

Factor 2: INS2

Factor 3: INS3

Factor 4: INS4

0.8575 0.2853

0.0133 0.7729

0.1529 0.0288 −0.0446 −0.1071

0.8217 0.5872 0.5547 −0.2890 0.2404 0.0181 0.5358 0.4384 0.0741

0.2724 0.2484 −0.0958 0.6843 0.6442 0.0007 0.5672 0.3009 0.1312

−0.0361 0.6091 0.4469 0.3234 0.2418 0.8262 0.3296 0.5581 0.0780

0.1504

0.7635

−0.0687 0.3016

0.1075 0.1343 0.4069 0.1002 0.2018 0.1308 −0.0922 −0.2627 0.9047

Source: Factors are extracted using principle component analysis method, and rotation is performed using varimax with Kaiser normalization. The boldfaced numbers indicate the highest weights and correlation of the (i.e., most relevant) variables for each retained factor.

fewer uncorrelated component factors. Using PCA on all twelve ICRG components, we showed that they can be sorted into four factor unique types of institutional quality. We depart from the existing literature by using different institutional indices as our focal variables in the growth equations. The four factor components of institutional quality as measured by ICRG are extracted as follows. The exploratory factor analysis was applied on our 12-component ICRG index to summarize correlated structures into fewer factors associated with a unique variance; see Siddiqui and Ahmed (2013) for details. These factors were extracted using PCA. PCA forms each factor as a linear combination of all observed variables (i.e., all 12 ICRG components) with successive levels of their respective unique variances. These factors are uncorrelated to one another and are equal to the number of observed variables. The first factor component has the largest proportion of total variance, while the subsequent factors have progressively lower proportions of the remaining variances. The Kaiser criteria (Kaiser, 1974) were applied here. Accordingly, we retained factor components with an eigenvalue greater than one, and we dropped those with an eigenvalue of less than one. The results in Table 1 show that we retain the four factor components, with the first factor (INS1) explaining 38% of 72% of the total retained variance. Factor 2 (INS2), Factor 3 (INS3), and Factor 4 (INS4) account for 14.7%, 10%, and 9% of the remaining variance, respectively.23 Finally, a varimax orthogonal rotation procedure is deployed in the current work to distribute factor loading evenly among the factors.24 Table 2 shows that INS1 is uniquely explained by government stability, investment profile, and external conflicts, while INS2 is heavily loaded

23 Each INS1, INS2, INS3, and INS4 ranges between −3 and +3 but was rescaled to 0–10 using the formula (Vi − Vmin)/(Vmax − Vmin) × 10. 24 The varimax method simplifies the interpretation of factor by maximizing the variances of the variable loadings on each factor (see Siddiqui and Ahmed, 2013).

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

219

Table 3 Summary statistics.

Mean Minimum Maximum Standard deviation Observation Number of countries (N) Number of time periods (T) LRGDPCit Initial income (log) [LRGDPCi, t − 1] Population growth (POP) Investment ratio (INV) Life expectancy (log) [LIFE (log)] Trade openness (OPEN) INS1/Political institutions INS2/Economic institutions INS3/Conflict-preventing institutions INS4/Democratic accountability

LRGDPCit

LRGDPCi, t − 1

POP

INV

LIFE (log)

OPEN

INS1

INS2

INS3

INS4

8.0706 5.8792 11.979 1.3491 344 39 8.82 1.0000 0.9956 0.2247 0.3730 0.8286 0.5338 0.5810 0.3957 −0.0428 −0.0837

8.0304 5.8792 11.582 1.3380 344 39 8.82

2.6647 −1.1566 16.350 1.7529 350 39 8.97

22.471 −11.57 76.422 11.454 344 39 8.82

4.0968 3.6216 4.370 0.1903 351 39 9

71.110 15.461 209.792 35.234 344 39 8.82

5.556 0 10 2.096 346 39 8.87

4.556 0 10 2.024 346 39 8.87

4.940 0 10 1.824 346 39 8.87

4.679 0 10 2.027 346 39 8.87

1.0000 0.2197 0.3639 0.8195 0.538 0.5847 0.3699 −0.0448 −0.0958

1.0000 0.0951 0.0374 0.2157 0.1657 0.1300 −0.0754 −0.1449

1.0000 0.4848 0.4745 0.3215 0.2091 0.0698 0.1480

1.0000 0.4529 0.5101 0.4205 −0.0073 0.0534

1.0000 0.4319 0.2681 0.1707 0.0304

1.0000 0.0052 0.0233 0.0137

1.0000 −0.0116 −0.0118

1.0000 −0.0215

1.0000

on socioeconomic condition, corruption, nonmilitarized politics, rule of law and order, and quality of bureaucracy. Meanwhile, INS3 is loaded predominantly on internal conflicts, religious tension, and ethnic tension while INS4 takes on democratic accountability. To summarize, INS1 reflects closely the political institutions governing the daily affairs of the state and ensuring the absence of cross-border conflicts. These institutions ensure stable and responsible government and freedom from expropriation and external pressures. INS2 reflects the economic institutions ensuring socioeconomic justice, the protection of property rights, and the enforcement of contracts. The index INS3 reflects conflict-preventing institutions, which minimize social conflicts, and the index INS4 is an indicator of the quality of democratic institutions, that is, democratic accountability. The correlation matrix in Table 3 clearly displays a strong positive correlation between INS1 and INS2, and GDP is high. For the other sub-indices, the rank is negative and low, suggesting that the former is important for economic prosperity. Finally, we note that the correlation between all four sub-indices is lower, and, in some cases, it is negative. The summary statistics for these and other variables are reported in Table 3. 4. Empirical results and discussion Table 4 reports the main effects of the four institutional components on economic growth in the OIC countries. The diagnostic checks on S-GMM reveal the following. The model passed the AR (2) tests, as indicated by p-value showing that the serial correlation in the error terms is not second order. Overall, the validity of the instruments and that of the additional instruments used as a necessity for S-GMM are confirmed, as indicated by the p-values of the Hansen J and difference-in-Hansen C tests. Thus, judging test statistics from these tests, one can conclude that the estimated model is adequately specified. All core variables enter the dynamic growth regression with the correct sign, and they are statistically significant at the conventional levels, except population growth, which appears not to be robust to the exclusion of outliers. It should be noted that the robust significant positive coefficient on logged initial income suggests that conditional convergence is operating among these Muslim countries. The implied speed of convergent (λ) of 5.9%25 is in the range of about 5–7.9% recorded by the dynamic panel growth studies of Islam (1995) and Caselli et al. (1996) but appears to be much faster than the 2–3% recorded in Barro's (1991) cross-sectional growth studies. For other core variables, their magnitude effects are in line with the existing 25 ~ ¼1þβ ¼ Based on Eqs. (2) and (3), the speed of convergent (λ) is given by solving β 1 expð−λtÞ where t is the time gap between current and lagged income, which is 3 in this paper.

literature (Azman-Saini et al., 2010; Caselli et al., 1996; Siddiqui and Ahmed, 2013). Our main interest is the focal variables—the four major components of institutions identified using PCA. Results suggest that only the estimated coefficients on INS1 and INS2 are significant at the 1% and 5% significance levels, respectively. The positive sign on both of these indicators reflects the important role played by political and economic institutions in improving living standards in the countries examined. Clearly, this finding supports the conventional wisdom that good (-quality) institutions are crucial for economic prosperity. It appears to replicate the results found in some earlier studies (e.g., Acemoglu et al., 2001; Rodrik et al., 2004; Siddiqui and Ahmed, 2013). Political institutions (i.e., INS1) that govern the daily affairs of the state and ensure the absence of cross-border conflicts (i.e., INS1) exert a strong significant positive impact on economic growth. Our model indicates that a one-standard-deviation increase in INS1 is associated with an increase in annual growth rate of about 4.9% (obtained by multiplying the slope coefficient, 0.0698, by the standard deviation, 2.096 divided by 3, the time gap in our panel estimate). Indeed, this is a large magnitude to be ignored for developing economies. We now turn to economic institution (i.e., INS2), a variable measuring socioeconomic justice and the protection of property rights. The empirical result reveals that a one-standard-deviation increase in the conduciveness of this market-supported economic institutional environment would lead to an increase in annual growth rate of 3.8 percentage points. The latter finding accords well with Hall and Jones (1999), who also documented high-magnitude effects (5.14%) of institutional infrastructure on growth in a cross-section of countries. The estimated coefficients of conflict-preventing institutions (labeled INS3) and democratic accountability (labeled INS4) are both insignificant (even at the 10 % significance level). With respect to INS4, the result is in line with Hisamoglu (2014), who also finds an insignificant effect on growth for Turkey. This may reflect the fragility and inconclusive nature of the link between democracy and local social factors and growth, an issue to which we will return later.26

26 However, high-quality institutions for conflict management can serve to mitigate the negative effects of adverse external shocks and social conflict on long-run economic growth (see Rodrik, 1999; Schneider and Wagner, 2001). It should be mentioned that both studies use distinct measures of conflict-management institutions that this paper employed. Rodrik (1999), for example, used earlier aggregate ICRG comprising rule of law, bureaucratic quality, corruption, and government expropriation, which are instead closed to our measures of economic institutions (i.e., INS2), and alternatively by democracy, i.e., the Freedom House’s civil liberties and political rights. While Schneider and Wagner (2001) used variants of trust and civic corporation measures, which according to North (1990)'s framework reflect informal institutions.

220

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

Table 4 Institutions and economic growth in the OIC.

Constant Initial income (log) Population growth Investment ratio Life expectancy (log) Trade openness INS1/Political institutions INS2/Economic institutions INS3/Conflict-preventing institutions INS4/Democratic accountability Time dummies AR(2) test (p-value) Hansen J-test (p-value) Diff-in-Hansen test (p-value) Instruments Country/Observation

Coeff.

S.e.

p-value

−0.5438 0.8371 0.0139 0.0040 0.3893 0.0014 0.0698 0.0560 −0.0186 −0.0107

0.6542 0.0428 0.0071 0.0012 0.2053 0.0005 0.0215 0.0251 0.0131 0.0145 Yes 0.483 0.895 0.812 36 39/340

0.4110 0.0000 0.0560 0.0020 0.0660 0.0110 0.0030 0.0320 0.1660 0.4640

Note: S.e. indicates heteroskedasticity-robust standard errors. AR(2) test is on the null of no second-order residual serial correlation. Hansen J-test reports p-value for null hypothesis of instrument validity. Difference-in-Hansen test reports p-values for the null of validity of the additional moment restrictions necessary for system GMM. Following Roodman (2009b), the instrument columns are collapsed.

Table 5 summarizes the results when economic institutions (INS2) and political institutions (INS1) are excluded from the baseline model to assess whether the insignificant results of INS3 and INS4 are due to the fact that countries with better-quality political and economic institutions also tend to have more social harmony, making conflictpreventing institutions (INS3) irrelevant (insignificant). As shown in Table 5, such a contention is indeed the case here. Models 1–4 (Table 5) explore the exclusion of INS2. The purpose of doing this is to determine whether INS3 still has a role to play in the growth process. Notice that all the models presented in Table 5 passed the diagnostic checks and appears to be adequately specified. Surprisingly, the results show that, without controlling for economic institutions, conflictpreventing institutions (INS3) turn negative and have a statistically significant effect on long-run growth.27 Thus, this result suggests that though political institutions significantly and positively affect growth, without controlling for economic institutions, better-quality conflictpreventing institutions instead harm economic growth, at least for the countries examined (Models 2 and 3). This is contrary to the popular belief that improvement in conflict-preventing institutions is essential for volatile and conflict-prone Muslim countries, especially those located in Africa and the Middle East. In fact, this finding is in accordance with an alternative perspective that suggests that improvement in political institutions is fundamental to the development process and that without improvement in economic institutions, further efforts to ensure social harmony and alleviate the ethnic and religious conflicts through conflict-preventing institutions may be irrelevant. To further the analysis, we consider the interaction term between political institutions and economic institutions (i.e., INS1 × INS2) as an additional explanatory variable in our model apart from the standard variables used in the growth equation.28 The INS1 and INS2 are included independently in the regression as well for two reasons. First, the significance of the interaction term may be the results of the omission of these variables by themselves. In this way, it allows one to test jointly whether these variables affect growth by themselves or through the interaction term. Second, to ensure that the interaction term did not proxy

27 The interaction term between INS1 and INS3 was included in Models 2 and 3. It carries a negative sign but was insignificant at the conventional levels. In the discussion below, we explore this further using a nonlinear model (i.e., DPTR). 28 The reason for doing this is that political institutions can have also indirect influence on growth through economic institutions (Eicher and Leukert, 2009, p. 205).

for INS1 or INS2, these variables were included in the regression independently. The coefficient of the interaction term between INS1 and INS2 is statistically insignificant (0.0077, s.e. = 0.0070) but the individual coefficients carry a positive sign and are significant at the conventional levels.29 In other words, the effect of economic institutions and political institutions has not weakened when the interaction term is included in the model. In contrary, we observed that when the interaction term between economic institutions and conflict-preventing institutions (i.e., INS2 × INS3) is added to the regression, the coefficient is positive and significant at one percent level (0.0292, s.e. = 0.0098) while both the positive coefficient on INS2 and the negative coefficient on INS3 are insignificant at the conventional levels. In other words, the direct effect of economic institution and conflict-preventing institutions on growth disappears (or weakens) when the interaction term is included in the model. In societies with a high degree of social conflict, some groups may find it difficult to cooperate, and thus those with dominant power devise economic institutions that redistribute resources to themselves at the expense of others (Acemoglu et al., 2005; Rodrik, 1999), making conflict-preventing institutions ineffective or even harmful to growth, as their function may be reduced to rent-seeking activities (see also Schneider and Wagner, 2001). This may be due to the fact that ethnic and religious and other internal conflicts arise because of weak economic institutions, not because of low-quality conflictpreventing institutions. Thus, efforts to alleviate these conflicts through improvement in conflict-preventing institutions, instead of economic institutions, may harm economic growth, at least in the countries examined. Therefore, improvement in the quality of both political and economic institutions is the key to wealth creation development, to ensuring social harmony, and to preventing conflict. Next, Models 5–8 perform further robustness checks on our findings regarding the impact of economic institutions on economic progress by excluding the political institutions (INS1). Clearly, the results of these models reveal that, without controlling for INS1 (political institutions), improvement in the quality of other institutions (i.e., INS2, INS3, and INS4) has no significant effect on economic growth. This highlights an intuitive finding that political institutions are crucial for long-run growth; INS1 by itself along the core growth variables is significant at the 5% significance level or better (Model 1). However, this is not the case for economic institutions (Models 5–8). Indeed, INS2, when included in all the regressions without INS1, carries the expected sign but remains statistically insignificant. The latter findings illustrate the importance of a third factor (INS1) to explain the relationship between economic institutions and economic growth and their role in shaping differences in growth across countries. The evidence suggests that countries have different economic institutions because of the structure of political power and because political institutions are different. Thus, our results tend to support the hypothesis that even the best economic institutions will not work well in the absence of supportive political institutions. Both political and economic institutions may have been predominant in affecting growth in the absence of a conscious effort to improve conflict-preventing institutions (INS3), as shown in the overall results earlier in Table 4. Thus, the primary aim of institutional reforms in these OIC countries should be improving the quality of political institutions (e.g., stable government with strong legislative and popular support; friendly diplomatic relations with the international community) and then further efforts can be geared toward improving the quality of economic institutions. This may help reduce the negative effect of domestic conflicts and tensions on economic progress.

29 Due to space consideration, we choose not to report the results here. All these results are available upon request. For completeness, we explore this further below using dynamic panel threshold regression.

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

221

Table 5 Political institutions vs. economic institutions and economic growth in the OIC. Excluding INS2 (economic institutions)

Constant Initial income (log) Population growth Investment ratio Life expectancy (log) Trade openness INS1/Political institutions INS2/Economic institutions

Model 2

Model 3

Model 4

Model 5

Model 6

Model 7

Model 8

−1.4611 (1.0799) 0.8779⁎⁎⁎ (0.0432) 0.0098⁎ (0.0058) 0.0045⁎⁎ (0.0021) 0.5281⁎ (0.2939) 0.0013⁎ (0.0006) 0.0506⁎⁎ (0.0235) –

−0.8212 (0.6534) 0.8897⁎⁎⁎ (0.0333) 0.0090 (0.0057) 0.0048⁎⁎⁎ (0.0017) 0.3479 (0.2069) 0.0013⁎⁎ (0.0006) 0.0602⁎⁎⁎ (0.0219) –

−0.8423 (0.5455) 0.8985⁎⁎⁎ (0.0353) 0.0111 (0.0069) 0.0041⁎⁎⁎ (0.0013) 0.3377⁎ (0.1778) 0.0016⁎⁎⁎ (0.0005) 0.0527⁎⁎ (0.0221) –

−0.7849 (1.1993) 0.8877⁎⁎⁎ (0.0469) 0.0082 (0.0068) 0.0040⁎⁎ (0.0017) 0.3464 (0.3244) 0.0016⁎⁎⁎ (0.0005) 0.0453⁎⁎ (0.0215) –

−1.3246 (1.1871) 0.9184⁎⁎⁎ (0.0339) 0.0134⁎⁎ (0.0061) 0.0052⁎⁎⁎ (0.0013) 0.4164 (0.3280) 0.0012⁎⁎ (0.0006) –

−0.6205 (0.7178) 0.9344⁎⁎⁎ (0.0342) 0.0124⁎⁎ (0.0057) 0.0045⁎⁎⁎ (0.0011) 0.2199 (0.2122) 0.0012⁎ (0.0006) –

−0.9818 (0.7122) 0.9385⁎⁎⁎ (0.0412) 0.0108⁎ (0.0061) 0.0041⁎⁎⁎ (0.0011) 0.3059 (0.2062) 0.0013⁎⁎ (0.0006) –

−1.3346 (0.9302) 0.9193⁎⁎⁎ (0.0417) 0.0094 (0.0069) 0.0041⁎⁎⁎ (0.0012) 0.4273 (0.2537) 0.0014⁎⁎⁎ (0.0005) –

0.0045 (0.0141)

−0.0316⁎⁎ (0.0141) −0.0049 (0.0160) Yes 0.430 0.763 0.518 33 39/340

−0.0149 (0.0175) Yes 0.536 0.571 0.604 30 39/340

0.0115 (0.0233) −0.0135 (0.0144) 0.0050 (0.0137) Yes 0.698 0.522 0.532 33 39/340

0.0092 (0.0266)

−0.0303⁎ (0.0153)

0.0096 (0.0182) 0.0011 (0.0140)

INS3/Conflict-preventing institutions INS4/Democratic accountability Time dummies AR(2) test (p-value) Hansen J-test (p-value) Difference-in-Hansen test (p-value) Instruments Country/Observation

Excluding INS1 (political institutions)

Model 1

Yes 0.521 0.769 0.498 27 39/340

Yes 0.430 0.894 0.574 30 39/340

Yes 0.705 0.657 0.818 27 39/340

Yes 0.726 0.603 0.630 30 39/340

−0.00005 (0.0149) Yes 0.707 0.414 0.591 30 39/340

Note: S.e. indicates heteroskedasticity-robust standard errors. AR(2) test is on the null of no second-order residual serial correlation. Hansen J-test reports p-value for null hypothesis of instrument validity. Difference-in-Hansen test reports p-values for the null of validity of the additional moment restrictions necessary for system GMM. Following Roodman (2009b) the instrument columns are collapsed. ⁎⁎⁎ Statistical significance at the 1% level. ⁎⁎ Statistical significance at the 5% level. ⁎ Statistical significance at the 10% level.

4.1. Robust checks Are these findings robust to the exclusion of extreme observations (or outliers) that may drive the finding? Outliers may blur our findings and, hence, result in an erroneous conclusion. We address this question by excluding the outliers based on the DFITS test advocated by Belsley et al. (1980) to flag countries that have high combinations of residual and leveraged statistics. The outcome of the test suggests that Qatar is

Table 6 Robust analysis: outlier removal.

Constant Initial income (log) Population growth Investment ratio Life expectancy (log) Trade openness INS1/Political institutions INS2/Economic institutions INS3/Conflict-preventing institutions INS4/Democratic accountability Time dummies AR(2) test (p-value) Hansen J-test (p-value) Diff-in-Hansen test (p-value) Instruments Country/Observation

Coeff.

S.e.

p-value

−0.8358 0.8138 0.0082 0.0038 0.5124 0.0014 0.0727 0.0543 −0.0201 −0.0076

0.5656 0.0426 0.0082 0.0009 0.1890 0.0005 0.0206 0.0198 0.0141 0.0132 Yes 0.483 0.895 0.812 36 38/340

0.148 0.0000 0.3240 0.0000 0.0100 0.0110 0.0010 0.0090 0.1630 0.5670

Note: S.e. indicates heteroskedasticity-robust standard errors. AR(2) test is on the null of no second-order residual serial correlation. Hansen J-test reports p-value for null hypothesis of instrument validity. Difference-in-Hansen test reports p-values for the null of validity of the additional moment restrictions necessary for system GMM. Following Roodman (2009b), the instrument columns are collapsed. An outlier (Qatar) was identified and removed using the so-called DFITS test.

an extreme outlier.30 Table 6 reports the main results of the regression without Qatar. The core variables remain highly significant except that the wrong-signed population is no longer significant. More importantly, the statistical significance of INS1 and INS2 remains intact, with the INS1 coefficient increasing its magnitude impacts on growth.31 The result with respect to INS3 and INS4 appears to hold up. Thus, the problem of extreme observations or outliers is not a major concern in this paper.32 Recently, Hisamoglu (2014), using autoregressive distributed lag (ARDL) models, shows that his conflict-governing ability indicator (measured as the first principle component of internal conflicts and ethnic tension) has a robust, positive, and significant effect on growth in Turkey. We construct the same indicator and find that in the OIC countries in general this indicator is negative and insignificant, confirming our main finding using broader indicators that also include religious conflict. Furthermore, as noted in the methodology section, following Roodman's (2009b) strategy to control for the proliferation of instruments is crucial to reduce small-sample bias and increase the power of the overidentification test. Vieira et al. (2012), however, have shown that doing so can weaken the effect of institutions on growth. Thus, our results on the link between political and economic institutions and growth appear to be highly robust to the problem of weak and excessively numerous instruments. In general, our results on OIC countries confirm the prevailing belief that institutions are the root cause of underdevelopment. Specifically, 30 The result and graphical illustration of this test are available upon request from the authors. 31 One possibility is that the direct effect of institutions (i.e., INS1 and INS2) on growth identified here may work through better management of their resources, particularly oil, as a large fraction of their recorded GDP comes from the extraction of existing resources. This may be an important research issue for further study. 32 We also reran all the models presented in Table 5, and the results are highly robust. To conserve space, the results are not reported here but are available upon request from the authors.

222

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

reform efforts aimed at the removal of institutional bottlenecks/risks by improving the quality of political and economic institutions are the key to long-term prosperity. We find that improvements in political institutions that ensure a stable political environment, low levels of expropriation, and low external conflicts are the most important ingredients for long-run growth. Moreover, efforts to improve the quality of economic institutions that ensure low corruption, competent bureaucracy, socioeconomic justice, and the efficient and impartial functioning of the legal environment not only are important for long-term growth in the OIC but are crucial for addressing negative growth effects that result from various internal conflicts, such as ethnic and religious conflicts. 4.2. Nonlinear effects of political institutions on growth in OIC countries In this section, we conduct a separate analysis for OIC countries using dynamic panel threshold regression (DPTR) methods. Results reported in Table 5 based on system GMM seem to indicate that political institutions (INS1) have direct influence on the economic growth.33 Recently, several studies have shown that political institutions have contingency effects on other dimensions of the institutional matrix (Aidt et al., 2008; Flachaire et al., 2014),34 supporting HIH in a global sample of countries. The theoretical model also argues that political institutions can have a nonlinear effect on growth. For instance, the “institutionally induced nonconvergence trap” argues that weak institutions (with entrenched vested interests, possessing/capturing de jure political power) can trap economies in a low-growth equilibrium, preventing them from transitioning into a new and better growth equilibrium as they move closer to the technological frontier, which requires better use of technological innovation (Acemoglu et al., 2006). Based on the prediction of the theoretical model and the strong empirical support for hierarchy of institutions hypothesis, we further systematically address the nonlinear threshold effects of political institutions on the growth effects of economic and conflict-preventing institutions for OIC countries using a recently developed dynamic panel threshold regression (DPTR) methods proposed by Kremer et al. (2013).35 Table 7 summarizes the main results from the application of the DPTR. There are threshold effects for political institutions on the growth effects of economic and conflict-preventing institutions. First, economic institutions positively affect growth only when the quality of political institutions surpasses the threshold score of 2.0676 (on the scale of 0–10) over the medium to long run. Though it supports the main finding (linear effect), the presence of threshold effect emphasizes that the benefit from improvement in the quality of economic institutions can accrue only when supporting political institutions are of sufficient quality. Furthermore, the threshold restriction of political institutions on the growth effects of economic institutions appears to be less restrictive (i.e., lower threshold score). Using a different measure of political institutions (voice and accountability of the World Bank's World Governance Index), Aidt et al. (2008) found much higher 33 Earlier we observed that interaction effects between INS1 and INS2 and between INS1 and INS3 are statistically insignificant. This may be due to the nonlinear threshold effect of INS1 on the growth effects of INS2 and INS3. We thank an anonymous referee for a suggestion that lead us to this extension. 34 A part from Flachaire et al. (2014), Aidt et al. (2008) showed theoretically and empirically using cross-section instrumental threshold regression, that there are two distinct political institution regimes, low-political-accountability and high-politicalaccountability regimes, defining the effect of corruption on growth. Both studies, however, focus on global samples. We address this issue in OIC countries, using a recent dynamic panel threshold regression framework. 35 For details on the DPTR procedure, see Kremer et al. (2013). Briefly, Eq. (3) above becomes:

Table 7 Threshold effects of political institutions on economic growth in the OIC. Threshold variable = Political institutions (INS1)

^ Threshold estimates γ 95% confidence interval Institutions ^ (INS1: Political institutions ≤ γ) β 1

^ (INS1: Political institutions N γ) β 2 Covariates Initial income (log) Population growth Investment ratio Life expectancy (log) Trade openness ^δ1 (regime intercept) Time dummies Country/Observation

Institutions = INS2, economic institutions

Institutions = INS3, conflict-preventing institutions

2.0676 [1.9854, 3.4515]

1.6667 [1.4355, 3.6923]

−0.0078 (0.0107)

−0.0368 (0.0169)⁎⁎

0.0342 (0.0051)⁎⁎⁎

0.0086 (0.0058)

0.7501 (0.1258)⁎⁎⁎ 0.0122 (0.0063)⁎ 0.0020 (0.0012)⁎ 0.5761 (0.2316)⁎⁎⁎ −0.0006 (0.0004) 0.0651 (0.0608)

0.9415 (0.1281)⁎⁎⁎ 0.0086 (0.0067) 0.0014 (0.0007)⁎⁎ 0.6515 (0.2425)⁎⁎⁎ −0.0004 (0.0005) 0.0698 (0.0742)

Yes 39/341

Yes 39/341

Notes: All available lags of LRGDPC were used as instruments. As in Hansen (1999), each regime contains at least 5% of the observations. ⁎⁎⁎ Statistical significance at the 1% level. ⁎⁎ Statistical significance at the 5% level. ⁎ Statistical significance at the 10% level.

threshold scores of political institutions (0.5–0.75 on a scale of 0–1) in their study of corruption-growth linkages.36 Despite this, our results reach a similar conclusion that to set the stage for the positive growth benefit from improvement in economic institutions, minimum quality of political institutions (relatively stable government, less expropriation, and low external conflicts) governing the daily affairs of these OIC countries has to be achieved. This also suggests that the effects of economic institutions on economic growth are regime specific defined by political institutions. This is in line with the hierarchy of institutions hypothesis (Acemoglu et al., 2005; Aidt et al., 2008; Flachaire et al., 2014). Second, the threshold effect of political institutions is also recorded for the growth effect of conflict-preventing institutions. It shows that conflict-preventing institutions have a significant and negative influence on growth only when countries' score for the quality of political institutions falls below the threshold score of 1.6667 (on the scale of 0–10). This may suggest that any period in OIC countries characterized by an environment of unstable governance, and high expropriation and external conflicts (weak political institutions) would spawn conflicts, making conflict-preventing institutions ineffective. However, that the threshold restriction is quite low indicates that such negative effects of conflict-preventing institutions only occur in OIC countries with worse political institutional environment (e.g., Sudan, Sierra Leon, Nigeria). This justifies the negative effect of conflict-preventing institutions recorded in Table 5 (Models 2–3) for the OIC countries. The reason is that their political institutions have not reached the tipping point not to hinder economic growth. 4.3. Comparative evidence from the global sample and developing countries subsamples

0 ~ yit ¼ βy i; t−1 þ β 1 INSit I ðINS1it ≤γ Þ þ δ1 I ðINS1it ≤γ Þ þ β 2 INSit I ðINS1it Nγ Þ þ θ X it þ ηi þ uit

Finally, we assess the comparative evidence on the relative importance of political, economic, and conflict-preventing institutions for global and developing countries samples. We examine the issue for 112 countries in the global sample, a subsample of 88 developing countries, and a subsample of 50 non-OIC developing countries over the period 1983–2009. Applying PCA on the data for the respective

where INS1it is the threshold variable defining the effects of other institutions, INS2 and INS3, on growth; γ is the threshold value; δ1 is the regime intercept; and other variables are as defined above. For a recent application of DPTR on the effect of institutions on the capital flow-growth nexus, see Chen and Quang (2014) and Slesman et al. (2015).

36 Some plausible reasons: It may be due to data frequency (our threshold value is based on panel data), measures of political institutions, or influence of rich countries with high political institutional competencies in their sample.

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

223

Table 8 Institutions and growth: global and developing countries evidence.

Constant Initial income (log) Population growth Investment ratio Life expectancy (log) Trade openness INS1/Political institutions INS2/Economic institutions INS3/Conflict-preventing institutions Time dummies AR(2) test (p-value) Hansen J-test (p-value) Diff-in-Hansen test (p-value) Instruments Country/Observation

Global sample

Developing countries

Non-OIC developing countries

−1.2206 (.3899)⁎⁎⁎ 0.9649 (0.0088)⁎⁎⁎ 0.0236 (0.0023)⁎⁎⁎ 0.0017 (0.0004)⁎⁎⁎ 0.3490 (0.1076)⁎⁎⁎ 0.00005 (0.00014) −0.0067 (0.0043) 0.0071 (0.0095) 0.0291 (0.0060)⁎⁎⁎ Yes 0.356 0.043 0.790 65 112/994

−1.275 (0.2915)⁎⁎⁎ 0.9394 (0.0082)⁎⁎⁎ 0.0167 (0.0030)⁎⁎⁎ 0.0019 (0.0003)⁎⁎⁎ 0.4061 (0.0837)⁎⁎⁎ 0.0003 (0.0001)⁎⁎⁎ 0.0112 (0.0047)⁎⁎ −0.0036 (0.0066) 0.0031 (0.0051) Yes 0.378 0.174 0.808 73 88/779

−0.7371 (0.2446)⁎⁎⁎ 0.9943 (0.0132)⁎⁎⁎ 0.0619 (0.0074)⁎⁎⁎ 0.0019 (0.0007)⁎⁎ 0.1690 (0.0739)⁎⁎ −0.0004 (0.0002)⁎⁎ 0.0589 (0.0121)⁎⁎⁎ 0.0104 (0.0113) 0.0385 (0.0109)⁎⁎⁎ Yes 0.240 0.186 0.201 41 50/447

Note: S.e. indicates heteroskedasticity-robust standard errors. AR(2) test is on the null of no second-order residual serial correlation. Hansen J-test reports p-value for null hypothesis of instrument validity. Difference-in-Hansen test reports p-values for the null of validity of the additional moment restrictions necessary for system GMM. Following Roodman (2009b), the instrument columns are collapsed. ⁎⁎⁎ Statistical significance at the 1% level. ⁎⁎ Statistical significance at the 5% level.

samples, the 12 correlated ICRG institutional indicators fit uniquely into three uncorrelated components, namely, INS1 (political institutions), INS2 (economic institutions), and INS3 (conflict-preventing institutions) for the three samples.37 Table 8 reports the results of the investigation. For the global sample, we notice that the model does not pass the Hansen J overidentification test.38 This suggests that the significant result for conflict-preventing institutions is not reliable. Furthermore, it may also suggest that when the correlated nature of institutional data is accounted for, the model may not support the pooling of developed and developing countries. This may also be due to the contrasting nature of the institutional competencies of rich and poor countries, as most rich countries receive the maximum score for institutional quality. For developing countries and non-OIC developing countries samples, all diagnostic tests (i.e., AR(2), Hansen J, and difference-in-Hansen tests) suggest that the models are adequately specified. First, the result shows in general that only political institutions have a significant and positive influence on economic growth in developing countries.39 Second, the results show that when OIC countries are excluded from the developing countries sample (i.e., non-OIC developing countries sample), the influence of political institutions remains intact, while conflict-preventing institutions turn significant and positively influence growth. In a nutshell, while political and economic institutions are crucial for growth and conflict reduction in OIC countries, political and conflict-preventing institutions matter the most for long-run economic growth in non-OIC developing countries. This is an important distinction because it highlights the importance of different sets of uncorrelated dimensions of an institutional matrix for economic growth in developing countries: OIC countries (political and economic institutions), non-OIC developing countries (political and conflict-preventing institutions), and developing countries (political institutions). It also serves as a reminder that pooling all the countries in one panel may not reveal the role of different 37 For all three samples, ICRG components are loaded into INS1, INS2, and INS3 in almost identical manner as the results reported in Tables 1 and 2 for the OIC sample. One exception is democratic accountability, which was shown to fit into INS4 for the OIC sample, is now loaded heavily on INS2. Due to space limitations, all the table results of PCA for these three samples are not reported here but are available upon request from the authors. 38 The model still fails the test even when outliers were removed, and also even when, in addition to Roodman strategy, the lag ranges are further reduced to limit the number of instruments. 39 When INS1 is excluded, economic and conflict-preventing institutions remain insignificant.

dimensions of institutional infrastructure on economic progress in the countries being examined.40 In sum, the main finding of our paper robustly points out that highquality political institutions are the key to growth in developing countries including OIC countries. For OIC, political institutions set the conducive environment in which economic institution function to promote growth, and better qualities of both political and economic institutions are crucial in minimizing conflicts. 5. Conclusion This paper is motivated by the fact that different kinds of underlying institutional infrastructure can lead to different economic outcomes. Despite voluminous empirical evidence suggesting that cluster institutions are the fundamental cause of economic prosperity, no study on the relative roles that political, economic, and conflict-preventing institutions play in growth has been conducted regarding OIC countries plagued with frequent conflicts arising from ethnic and religious conflicts and other institutional uncertainties that gave rise to the recent Arab Spring. We fill this gap in this line of the literature by focusing on a comparison of the OIC countries with global countries, developing countries, and non-OIC developing countries. By extracting the unique institutional dimensions from broadly correlated institutional indicators for these countries, through the use of PCA, we address the problems of high correlation among these indicators and, to some degree, of the lack of clear conceptual dimensions reflected in the underlying institutions. Further, we also directly address the issue of the weakness and excess number of instruments that plagued the S-GMM estimator in our empirical assessment. Thus, we deal with recent concerns that existing evidence on the role of institutions in long-run growth may suffer from weak and excessively numerous instruments (Roodman, 2009b; Vieira et al., 2012) and show that our results on the relative importance of different (uncorrelated) sets of institutional matrices for growth are robust. Our results show political institutions are the most important dimension of the institutional matrix for economic growth in OIC countries, in particular, and developing countries, in general. More interestingly is the finding that in OIC countries, economic institutions positively and significantly affect growth only after political institutions are accounted 40 Recent study by Narayan et al. (2011b) on stock market convergent also made a similar point when they argued that one is unlikely to find convergent in stock market in a highly heterogeneous panel of global countries.

224

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

for and attain sufficient quality. Third, contrary to a popular belief in the importance of conflict-preventing institutions in these OIC countries, we found that these institutions are negatively affecting growth, especially at lower-quality political institutions. However, when both political and economic institutions are accounted for in the analysis, the negative influence of conflict-preventing institutions fades away. Furthermore, we extend the existing literature in showing that different uncorrelated sets of institutions matter differently for different groups of developing countries. In general, political institutions are the core determinant of growth in developing countries. In addition to this fundamental factor, economic institutions are important in OIC countries while conflict-preventing institutions are crucial for non-OIC developing countries. The implication is that the main institutional binding constraint to development in developing countries in general is political institutions. Thus, institutional reforms that aim to minimize this binding constraint can provide a conducive environment for economic growth and the functioning of economic and conflict-preventing institutions, especially in OIC countries and non-OIC developing countries. For instance, strong institutional constraints to checks and balance (the de jure and de facto) power structure of the elites and powerful groups, and having an efficient and independent judiciary, are some of the most important elements for ensuring better-quality political institutions, which can spill over into better functioning of other dimensions of the institutional matrix, such as institutions that protect private property and ownership rights and contract enforcement (economic institutions) and institutions that ensure social harmony (conflict-preventing institutions). Acknowledgements We are grateful to the editor and the two anonymous referees for helpful comments and suggestions. Financial support from Universiti Putra Malaysia [Grant no: GP-IPB/2014/9440900] is acknowledged. Appendix A. List of member countries of the Organization of Islamic Cooperation (OIC) included in the study

Albania Algeria Bahrain Bangladesh Burkina Faso Cameroon Côte d'Ivoire Cyprus Egypt Gabon Gambia, The Guinea Guinea-Bissau Indonesia Iran Jordan Kuwait Libya Malaysia Mali Morocco Mozambique Niger Oman Nigeria Pakistan Qatar Saudi Arabia Senegal Sierra Leone Sudan Suriname

ALB DZA BHR BGD BFA CMR CIV CYP EGY GAB GMB GIN GNB IDN IRN JOR KWT LBY MYS MLI MAR MOZ NER OMN NGA PAK QAT SAU SEN SLE SDN SUR

Syria Togo Tunisia Turkey United Arab Emirates Uganda Yemen

SYR TGO TUN TUR ARE UGA YEM

Appendix B. Variable, definitions, and sources

Variable Real GDP per capita (Y)

Definition

Source

PPP converted GDP per capita (chain series), at Penn World 2005 constant prices. Table Mark 7.1 (PWT) Population growth Annual population growth rate. World (POP) Development Indicators (WDI) Investment share Investment share of PPP converted GDP at PWT of real GDP (INV) 2005 constant prices Life Expectancy Life expectancy at birth indicates the number WDI (LIFE) (log) of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life. Trade openness Export plus import divided by GDP. WDI Government An assessment both of the government's ability Internationstability to carry out its declared program(s), and its al Country ability to stay in office. The risk rating assigned Risk Guide is the sum of three subcomponents, each with (ICRG), a maximum score of four points and a Political minimum score of 0 points. The Risk Service subcomponents are government unity, Group (PRS, legislative strength and popular support. It is Inc.) rescaled to 0–10 scale, higher score means better quality. Socioeconomic An assessment of the socioeconomic pressures ICRG, PRS conditions at work in society that could constrain government action or fuel social dissatisfaction. The risk rating assigned is the sum of three subcomponents, each with a maximum score of four points and a minimum score of 0 points. The subcomponents are unemployment, consumer confidence, poverty. It is rescaled to 0–10 scale, higher score means better quality. Investment profile An assessment of factors affecting the risk to ICRG, PRS investment that are not covered by other political, economic and financial risk components. The risk rating assigned is the sum of three subcomponents, each with a maximum score of four points and a minimum score of 0 points. The subcomponents are contract viability/expropriation, profits repatriation, payment delays. It is rescaled to 0–10 scale, higher score means better quality. Internal conflict An assessment of political violence in the ICRG, PRS country and its actual or potential impact on governance. The highest rating is given to those countries where there is no armed or civil opposition to the government and the government does not indulge in arbitrary violence, direct or indirect, against its own people. The lowest rating is given to a country embroiled in an on-going civil war. The risk rating assigned is the sum of three subcomponents, each with a maximum score of four points and a minimum score of 0 points. The subcomponents are civil war/coup threat, terrorism/political violence, civil disorder. It is rescaled to 0–10 scale, higher score means better quality. External conflict An assessment of both the risk to the ICRG, PRS incumbent government from foreign action, ranging from non-violent external pressure (diplomatic pressures, withholding of aid, trade restrictions, territorial disputes, sanctions, etc.) to violent external pressure (cross-border conflicts to all-out war). External conflicts can adversely affect foreign business in many ways, ranging from restrictions on operations, to trade and investment sanctions, to distortions in the allocation of economic

L. Slesman et al. / Economic Modelling 51 (2015) 214–226 Appendix B (continued) (continued)

Appendix B (continued) (continued) Variable

Corruption

Military in politics

Definition

225

Source

resources, to violent change in the structure of society. The risk rating assigned is the sum of three subcomponents, each with a maximum score of four points and a minimum score of 0 points. The subcomponents are war, cross-border conflict, and foreign pressures. It is rescaled to 0–10 scale, higher score means better quality. An assessment of corruption within the ICRG, PRS political system. Such corruption is a threat to foreign investment for several reasons: it distorts the economic and financial environment; it reduces the efficiency of government and business by enabling people to assume positions of power through patronage rather than ability; and, last but not least, introduces an inherent instability into the political process. The most common form of corruption met directly by business is financial corruption in the form of demands for special payments and bribes connected with import and export licenses, exchange controls, tax assessments, police protection, or loans. Such corruption can make it difficult to conduct business effectively, and in some cases may force the withdrawal or withholding of an investment. Although this measure takes such corruption into account, it is more concerned with actual or potential corruption in the form of excessive patronage, nepotism, job reservations, “favor-for-favors,” secret party funding, and suspiciously close ties between politics and business. These insidious sorts of corruption are potentially of much greater risk to foreign business in that they can lead to popular discontent, unrealistic and inefficient controls on the state economy, and encourage the development of the black market. The greatest risk in such corruption is that at some time it will become so overweening, or some major scandal will be suddenly revealed, as to provoke a popular backlash, resulting in a fall or overthrow of the government, a major reorganizing or restructuring of the country's political institutions, or, at worst, a breakdown in law and order, rendering the country ungovernable. This index is rescaled to 0–10 scale, higher score means lower corruption. The military is not elected by anyone. ICRG, PRS Therefore, its involvement in politics, even at a peripheral level, is a diminution of democratic accountability. However, it also has other significant implications. The military might, for example, become involved in government because of an actual or created internal or external threat. Such a situation would imply the distortion of government policy in order to meet this threat, for example by increasing the defense budget at the expense of other budget allocations. In some countries, the threat of military takeover can force an elected government to change policy or cause its replacement by another government more amenable to the military's wishes. A military takeover or threat of a takeover may also represent a high risk if it is an indication that the government is unable to function effectively and that the country therefore has an uneasy environment for foreign businesses. A full-scale military regime poses the greatest risk. In the short term, a military regime may provide a new stability and thus reduce business risks. However, in the longer term, the risk will almost certainly rise, partly because the system of governance will

Variable

Religious tensions

Rule of law and order

Ethnic tension

Democratic accountability

Bureaucratic quality

Definition be become corrupt and partly because the continuation of such a government is likely to create an armed opposition. In some cases, military participation in government may be a symptom rather than a cause of underlying difficulties. Overall, lower risk ratings indicate a greater degree of military participation in politics and a higher level of political risk. It is rescaled to 0–10 scale, higher score means lower risk. Religious tensions may stem from the domination of society and/or governance by a single religious group that seeks to replace civil law by religious law and to exclude other religions from the political and/or social process; the desire of a single religious group to dominate governance; the suppression of religious freedom; the desire of a religious group to express its own identity, separate from the country as a whole. The risk involved in these situations range from inexperienced people imposing inappropriate policies through civil dissent to civil war. It is rescaled to 0–10 scale, higher score means lower risk. Law and order are assessed separately, with each sub-component comprising zero to three points. The law sub-component is an assessment of the strength and impartiality of the legal system, while the order sub-component is an assessment of popular observance of the law. Thus, a country can enjoy a high rating—3—in terms of its judicial system, but a low rating—1—if it suffers from a very high crime rate of if the law is routinely ignored without effective sanction (for example, widespread illegal strikes). It is rescaled to 0–10 scale, higher quality means better quality. An assessment of the degree of tension within a country attributable to racial, nationality, or language divisions. Lower ratings are given to countries where racial and nationality tensions are high because opposing groups are intolerant and unwilling to compromise. Higher ratings are given to countries where tensions are minimal, even though such differences may still exist. It is rescaled to 0–10 scale, higher score means lower degree of ethnic tension. Measure of how responsive government is to its people, on the basis that the less responsive it is, the more likely it is that the government will fall, peacefully in a democratic society, but possibly violently in a non-democratic one. It is rescaled to 0–10 scale, higher score means lower risk. The institutional strength and quality of the bureaucracy is another shock absorber that tends to minimize revisions of policy when governments change. Therefore, high points are given to countries where the bureaucracy has the strength and expertise to govern without drastic changes in policy or interruptions in government services. In these low-risk countries, the bureaucracy tends to be somewhat autonomous from political pressure and to have an established mechanism for recruitment and training. Countries that lack the cushioning effect of a strong bureaucracy receive low points because a change in government tends to be traumatic in terms of policy formulation and day-to-day administrative functions. It is rescaled to 0-10 scale, higher score means better quality.

Source

ICRG, PRS

ICRG, PRS

ICRG, PRS

ICRG, PRS

ICRG, PRS

Sources: These definitions are extracted from ICRG Methodology of the PRS, Inc., WDI, and Penn World Table Mark 7.1 (PWT).

226

L. Slesman et al. / Economic Modelling 51 (2015) 214–226

References Acemoglu, D., Johnson, S., 2005. Unbundling institutions. J. Polit. Econ. 113, 949–995. Acemoglu, D., Robinson, J., 2000. Political losers as a barrier to economic development. Am. Econ. Rev. 90, 126–130. Acemoglu, D., Robinson, J., 2008. Persistence of power, elites, and institutions. Am. Econ. Rev. 98, 267–293. Acemoglu, D., Johnson, S., Robinson, J., 2001. The colonial origins of comparative development: an empirical investigation. Am. Econ. Rev. 91, 1369–1401. Acemoglu, D., Johnson, S., Robinson, J., 2005. Institution as the fundamental cause of longrun growth. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth. Elsevier, Amsterdam, Netherland, pp. 385–472. Acemoglu, D., Aghion, P., Zilibotti, F., 2006. Distance to frontiers, selection, and economic growth. J. Eur. Econ. Assoc. 4, 37–74. Aidt, T., Dutta, J., Sena, V., 2008. Governance regimes, corruption and growth: theory and evidence. J. Comp. Econ. 36, 195–220. Anwar, S., Cooray, A., 2012. Financial development, political rights, civil liberties and economic growth: evidence from South Asia. Econ. Model. 29 (3), 974–981. Arellano, M., Bond, S., 1991. Some tests of specification for panel data: Monte Carlo evidence with an application for employment equations. Rev. Econ. Stud. 58, 277–297. Arellano, M., Bover, O., 1995. Another look at the instrumental-variable estimation of error-components models. J. Econ. 68, 29–52. Asoni, A., 2008. Protection of property rights and growth as political equilibrium. J. Econ. Surv. 22, 953–987. Azman-Saini, W.N.W., Baharumshah, A.Z., Law, S.H., 2010. Foreign direct investment, economic freedom and economic growth: international evidence. Econ. Model. 27, 1079–1089. Banerjee, A., Iyer, L., 2005. History, institutions, and economic performance: the legacy of colonial land tenure systems in India. Am. Econ. Rev. 95, 1190–1213. Barro, R.J., 1991. Economic growth in a cross section of countries. Q. J. Econ. 106, 407–443. Belsley, D.A., Kuh, E., Welsch, R.E., 1980. Regression Diagnostics. Wiley, New York. Bhattacharyya, S., 2009. Unbundled institutions, human capital and growth. J. Comp. Econ. 37, 106–120. Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel data models. J. Econ. 87, 115–143. Bowsher, C., 2002. On testing overidentifying restrictions in dynamic panel data models. Econ. Lett. 77, 211–220. Carlsson, F., Lundstrom, S., 2002. Economic freedom and growth: Decomposing the effects. Public Choice 112, 335–344. Caselli, F., Esquivel, G., Lefort, F., 1996. Reopening the convergence debate: a new look at cross-country growth empirics. J. Econ. Growth 1, 363–389. Chen, J., Quang, T., 2014. The impact of international financial integration on economic growth: new evidence on threshold effects. Econ. Model. 42, 475–489. Dawson, J.W., 1998. Institutions, investment and growth: new cross-country and panel data evidence. Econ. Inq. 36, 603–619. Dawson, J.W., 2003. Causality in the freedom-growth relationship. Eur. J. Polit. Econ. 19, 479–495. De Haan, J., Lundstrom, S., Sturm, J.-E., 2006. Market-oriented institutions and policy and economic growth: a critical survey. J. Econ. Surv. 20, 158–191. Doucouliagos, C., Ulubasoglu, M.A., 2006. Economic freedom and economic growth: does specification make a difference? Eur. J. Polit. Econ. 22, 60–81. Eicher, T.S., Leukert, A., 2009. Institutions and economic performance: endogeneity and parameter heterogeneity. J. Money Credit Bank. 41, 197–219. Fedderke, J., 2001. Growth and institutions. J. Int. Dev. 13, 645–670. Flachaire, E., García-Peñalosa, C., Konte, M., 2014. Political versus economic institutions in the growth process. J. Comp. Econ. 42, 212–229. Glaeser, E.L., La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2004. Do institutions cause growth? J. Econ. Growth 9, 273–303. Gwartney, J.D., Holcombe, R.G., Lawson, R.A., 2006. Institutions and the impact of investment on growth. Kyklos 59, 255–273. Hall, S.G., Ahmad, M., 2012. Institutions-growth spatial dependence: an empirical test. Procedia Soc. Behav. Sci. 65, 925–930. Hall, R., Jones, C., 1999. Why do some countries produce so much more output per worker than others? Q. J. Econ. 114, 83–116. Hansen, L.P., 1982. Large sample properties of generalized method of moments estimators. Econometrica 50, 1029–1054. Hansen, B., 1999. Threshold effects in non-dynamic panels: estimation, testing, and inference. J. Econ. 93, 345–368. Heckelman, J., Stroup, M., 2000. Which economic freedoms contribute to growth? Kyklos 53, 527–544.

Hisamoglu, E., 2014. EU membership, institutions and growth: the case of Turkey. Econ. Model. 38, 211–219. Holtz-Eakin, D., Newey, W., Rosen, H., 1988. Estimating vector autoregressions with panel data. Econometrica 56, 1371–1395. Huang, Y., 2010. Political institutions and financial development: an empirical study. World Dev. 38, 1667–1677. Islam, N., 1995. Growth empirics: a panel data approach. Q. J. Econ. 110, 1127–1170. Kaiser, H.F., 1974. An index of factor simplicity. Psychometrika 39, 31–36. Knack, S., Keefer, P., 1995. Institutions and economic performance: cross-country tests using alternative institutional measures. Econ. Polit. 7, 207–227. Knoll, M., Zloczysti, P., 2012. The good governance indicators of the millennium challenge account: how many dimensions are really being measured? World Dev. 40, 900–915. Kremer, S., Bick, A., Nautz, D., 2013. Inflation and growth: new evidence from a dynamic panel threshold analysis. Empir. Econ. 44, 861–878. Krusell, P., Rio-Rull, J.-V., 1996. Vested interests in a positive theory of stagnation and growth. Rev. Econ. Stud. 63, 301–329. Langbein, L., Knack, S., 2010. The worldwide governance indicators: six, one, or none? J. Dev. Stud. 46, 350–370. Levine, R., Renelt, D., 1992. A sensitive analysis of cross-country growth regressions. Am. Econ. Rev. 82, 942–963. Mankiw, G., Romer, D., Weil, N., 1992. A contribution to the empirics of economic growth. Q. J. Econ. 107, 407–437. Narayan, P.K., Narayan, S., Smyth, R., 2011a. Does democracy facilitate economic growth or does growth facilitate democracy? An empirical study of sub-Saharan Africa. Econ. Model. 28, 900–910. Narayan, P.K., Mishra, S., Narayan, S., 2011b. Do market capitalization and stocks traded converge? New global evidence. J. Bank. Financ. 35 (10), 2771–2781. Narayan, P.K., Narayan, S., Thuraisamy, K.S., 2014. Can institutions and macroeconomic factors predict stock returns in emerging markets? Emerg. Mark. Rev. 19, 77–95. Narayan, P.K., Sharma, S.S., Thuraisamy, K.S., 2015. Can governance quality predict stock market returns? New global evidence. Pac. Basin Financ. J. http://dx.doi.org/10. 1016/j.pacfin.2015.02.007. Nawaz, S., 2015. Growth effects of institutions: a disaggregated analysis. Econ. Model. 45, 118–126. North, D.C., 1981. Structure and Change in Economic History, Norton, New York. North, D.C., 1990. Institutions, Institutional Change, and Economic Performance. Cambridge University Press, Cambridge, UK. Parente, S.L., Prescott, E.C., 1999. Monopoly rights: a barrier to riches. Am. Econ. Rev. 89, 1216–1233. Rodrik, D., 1999. Where did all the growth go? External shocks, social conflicts, and growth collapse. J. Econ. Growth 4, 385–412. Rodrik, D., 2000. Institutions for high-quality growth: what they are and how to acquire them. Stud. Comp. Int. Dev. 35, 3–31. Rodrik, D., 2005. Growth strategies. In: Aghion, P., Durlauf, S. (Eds.), Handbook of Economic Growth. Elsevier, Amsterdam, Netherlands, pp. 967–1014. Rodrik, D., Subramanian, A., Trebbi, F., 2004. Institutional rule: the primacy of institutions over geography and integration in economic development. J. Econ. Growth 9, 131–165. Roodman, D., 2009a. How to do xtabond2: an introduction to “difference” and “system” GMM in Stata. Stata J. 9, 86–136. Roodman, D., 2009b. A note on the theme of too many instruments. Oxf. Bull. Econ. Stat. 71, 135–158. Sakbani, M., 2011. The revolutions of the Arab Spring: are democracy, development and modernity at the gates? Contemp. Arab Aff. 4, 127–147. Schneider, F., Wagner, A.F., 2001. Institutions of conflict management and economic growth in the European Union. Kyklos 54, 509–532. Siddiqui, D.A., Ahmed, Q.M., 2013. The effect of institutions on economic growth: a global analysis based on GMM dynamic panel estimation. Struct. Chang. Econ. Dyn. 24, 18–33. Slesman, L., Baharumshah, A.Z., Wohar, M.E., 2015. Capital inflows and economic growth: does the role of institutions matter? Int. J. Financ. Econ. 20, 253–275. Vieira, F., MacDonald, R., Damasceno, A., 2012. The role of institutions in cross-section income and panel data growth models: a deeper investigation on the weakness and proliferation of instruments. J. Comp. Econ. 40, 127–140. Windmeijer, F., 2005. A finite sample correction for the variance of linear efficient twostep GMM estimators. J. Econ. 126, 25–51.