The Euro-adoption effect and the bank, market, and growth nexus: New evidence from EU panels

The Journal of Economic Asymmetries 12 (2015) 41–51

Contents lists available at ScienceDirect

The Journal of Economic Asymmetries www.elsevier.com/locate/jeca

The Euro-adoption effect and the bank, market, and growth nexus: New evidence from EU panels Andreas G. Georgantopoulos ∗ , Anastasios D. Tsamis, Maria-Eleni K. Agoraki Panteion University, Department of Public Administration 136, Sygrou Ave. 176 71, Athens, Greece

a r t i c l e

i n f o

Article history: Received 13 October 2014 Received in revised form 30 December 2014 Accepted 13 January 2015 Available online xxxx Keywords: Panel data European Union Economic growth Banks Stock markets GMM estimator

a b s t r a c t This study investigates the financial system–growth relationships for a panel that includes the twenty-eight member states of the European Union (EU) for the period 1999–2012. Considering that the Euro currency is currently adopted by only seventeen member states, the originality of this paper lies in that it assesses, for the first time in related literature, the financial system–growth nexus by dividing the full EU28 sample in two new panels, the Eurozone panel and the non-Euro countries panel. The basic argument behind this sample split is that the Euro-adoption prerequisites closer and more centralized political, economic, fiscal and financial cooperation between the Eurozone members and therefore different relationships may exist, from a finance–growth perspective, due to greater sample uniformity. To assess these relationships, the differenced and the system GeneralizedMethods-of-Moments (GMM) estimators are employed. The empirical findings derived from the full EU countries panel fail to support the bank-led growth hypothesis. However, stock markets are reported as weak but significant growth factors. From the Eurozone panel, strongly significant results are reported, documenting the contribution of the financial sector on growth. Finally, from the non-Euro countries panel, findings imply the significant negative impact of the banking sector on growth. Overall, these results could suggest that testing these relationships under the Euro-adoption criterion can indeed lead to different results. Policy makers should further improve the banking/financial regulatory framework, the credit allocation process and the banking competition in order for the financial system to promote EU’s economic development in more efficient ways. © 2015 Published by Elsevier B.V.

1. Introduction During the last two decades, forces such as globalization, technological change, deregulation and integration have fundamentally transformed the European and global financial industry. The second Banking Directive, the creation of the European Monetary Union in 1999, and the approval and implementation of the Financial Services Action Plan (FSAP) by the European Commission gave new impulse to the creation of a single European financial services market. Moreover, the recent financial crisis is milestone in the financial transformation that has provoked immense debate on the circumstances under which crises arise. Thus, it becomes particularly significant to understand the finance–growth nexus, extracting conclusions mainly for the European economy in the new global framework.

*

Corresponding author. E-mail addresses: [email protected] (A.G. Georgantopoulos), [email protected] (A.D. Tsamis), [email protected] (M.-E.K. Agoraki).

http://dx.doi.org/10.1016/j.jeca.2015.01.002 1703-4949/© 2015 Published by Elsevier B.V.

42

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

The literature on the relationships between the components of the financial system, namely, the banking sector and the financial sector, provides exhaustive evidence that support the significance of the financial system–growth nexus. These studies date back to the early works of Bagehot (1873) on the relationships between the financial system and industrialization for the case of England, and Schumpeter (1911, 1934), who suggests that the financial intermediary sector plays a significant role to productivity growth and technological advance by the allocation of savings to firms. Since then, the literature evolved with a rapid pace, especially during the last twenty years, and can be divided into two theoretical groups. The first group deals with the impact of stock market development on economic growth by employing proxies such as stocks traded, market capitalization and turnover ratio. The second group of literature concentrates on the relationship between banking sector developments, as measured by bank’s credit and liquid liabilities, and economic growth. Regardless of the proxies and countries employed, both strands of the literature are in consensus that the financial system promotes economic growth, although presenting different characteristics and significance levels across regions/countries. Moreover, this literature further expanded by employing measures of trade, namely, trade openness (the sum of imports and exports) and exports documenting not only the significant impact of trade on economic growth (see Shahbaz, 2012; Georgantopoulos & Tsamis, 2011; Yanikkaya, 2003; Baldwin & Forslid, 1998) but also the linkages between financial development and trade (Bordo & Rousseau, 2012; Taylor & Wilson, 2011; Baltagi, Demetriades, & Law, 2009; Hur, Raj, & Riyanto, 2006). This study focuses on the impact of financial and banking sector developments on economic growth for a full panel of the 28 European Union member countries (EU28). The contribution of this study to the related literature is threefold; first, this is the only study that focuses exclusively on the EU28 (Croatia became the 28th EU member state on 1 July 2013). Other studies deal with the financial system and growth nexus (see Section 2) either by employing data relating to a single European country or by using too many and mixed regional European panel data making it difficult to analyze in depth one particular region that presents similar financial attributes. Other studies include EU countries in panels mixed-up with other non-EU countries. Therefore, this study argues that EU member states, present different financial characteristics than non-EU countries, mainly due to the close economic and political cooperation that EU member states have developed. The EU operates through a system of supranational independent institutions and intergovernmental negotiated decisions by the member states. Some of the most prominent EU institutions are the European Commission, the Council of the European Union, the European Council, the Court of Justice of the European Union, the European Central Bank, the Court of Auditors, and the European Parliament. Although the EU is not a single country yet, a consolidated economic and political environment has been created for its member states and therefore scientific conclusions could be much more accurate dealing with these countries apart from other non-EU countries when employing panel data analysis. The second contribution of this study is that for the first time in this literature, the full EU sample is divided into a panel of Eurozone countries and non-Euro countries. The European Economic and Monetary Union (EMU) is a term for the group of policies aimed at integrating the economies of all EU member states. Both the seventeen Eurozone states and the eleven non-Euro states are EMU members. However, a Member State needs to comply with a number of economic and fiscal criteria before being able to adopt the Euro as a national currency and become a full member of the Eurozone. Therefore, the Eurozone countries enjoy an advanced stage of not only economic and trade, but also monetary integration binding the Eurozone countries tighter to each other than the rest of the non-Euro countries that have not adopted the Euro currency. The Euro currency (€) is managed and administered by the European Central Bank (ECB) and the Euro-system, which is composed of the central banks of the Eurozone countries. As an independent central bank, the ECB has sole authority to set monetary policy. The Euro-system participates in the printing, minting and distribution of notes and coins in all member states, and the operation of the Eurozone payment systems. The 1992 Maastricht Treaty obliges most EU member states to adopt the euro upon meeting certain monetary and budgetary convergence criteria, although not all states have done so yet. The United Kingdom and Denmark negotiated exemptions, while Sweden, that joined the EU in 1995, after the Maastricht Treaty was signed, turned down the Euro in a 2003 referendum, and has circumvented the obligation to adopt the Euro by not meeting the monetary and budgetary requirements. However, all nations that have joined the EU since 1993 have pledged to adopt the euro in due course. To summarize, the Eurozone members present different characteristics from the rest non-Euro countries in three ways; (i) common monetary policy, since the ECB is the sole responsible entity, (ii) closer political cooperation through the Euro Group, and (iii) limited but significant fiscal integration, through the peer review of each other’s national budgets, which can also be seen as a highly political act. Therefore, these characteristics of the Euro area can justify the separate treatment of the Euro-members apart from the rest of the non-Euro member states. Third, this study investigates the short-run relationships between financial sector, banking sector and economic growth. There are very few studies (Narayan & Narayan, 2013; Loayza & Ranciere, 2006; Kaminsky & Reinhart, 1999) that deal with these short-run relationships. However, four factors could be numbered that motivate this study to focus on a short-run timeframe: (i) the rapidly changing economic environment; since the beginning of the EU in 1957, then called European Economic Community (EEC), when the six founding member states, namely, Belgium, France, Italy, Luxembourg, the Netherlands, and West Germany signed the Treaty of Rome, numerous territorial, economic, political and social transformations have taken place, leading to a significantly expanded, and more stabilized and integrated European Union. Therefore, the risk of reaching to ambiguous and misleading results, could be higher when dealing with long period panel data, especially in the case of the EU, (ii) the concern that averaging data leads to information loss and rigid model that omits the

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

43

cross-country heterogeneity parameter (Loayza & Ranciere, 2006), (iii) the possibility that negative effects can occur on economic growth if the banking sector is over-liberalized in the short-run, urge this study to focus on investigating the short-run relationships between the selected variables, and (iv) the data limitation problem that researchers face when choosing to work on balanced cross-country panels. The rest of the paper is organized as follows. Section 2 provides a brief review on the recent literature. Section 3 introduces the data, the selected determinants of growth and the methodology. Section 4 presents the descriptive statistics and empirical results, while Section 5 summarizes and provides the concluding remarks with some policy implications. 2. Literature review Since the seminal contributions of Bagehot (1873) and Schumpeter (1911, 1934), as described briefly in the previous section, a large volume of studies deal with this topic. This section limits the discussion only to a selected body of empirical literature and emphasizes on studies that are more closely related to this paper in terms of data analysis and methodological application. The literature provides conflicting evidence regarding the impact of the financial system and its components, namely, stock markets and banks, on economic growth. A number of studies suggest that when banks and stock markets operate under a well-functioning economic environment, they can both stimulate growth and efficiently allocate resources through the reduction of information loss and transaction costs. King and Levine (1993a) examined the impact of various financial system indicators (liquid liabilities to GDP, claims on the non-financial private sector to GDP, and private credit). In this study, cross-sectional data from 80 developed and developing countries are employed, concluding that all four indicators have statistically significant and positive impact on growth (for related studies, see also, Bencivenga, Smith, & Starr, 1995; Bencivenga and Smith, 1991). However, these studies also note that the financial system can hurt economic growth since the continuously increasing resource allocation, which is associated to higher return on deposits, may lead eventually to lower interest rates on deposits. Therefore, in case sufficiently large externalities occur that are associated with savings and investment, the financial system will burden economic growth. Moreover, the literature on the finance–growth nexus is divided based on whether the financial sector and banking sector are substitutes, compliments or whether banks overcome markets and vice versa, as more significant stimulus of growth. Levine, Loayza, & Beck (2000) investigate the impact of financial intermediary variables, namely, liquid liabilities and private credit on growth for a cross-country panel of 74 countries and for a cross-section panel of 71 countries employing the Arellano and Bond (1991) panel-GMM estimator. They concluded that financial intermediary variables have significant and positive impact on growth. Cole, Moshirian, & Wu (2008) examine the impact of banking sector stock returns on growth for a panel of 38 countries employing the Arellano and Bond GMM estimator concluding that banking sector stock returns have more significant impact on a panel of developing markets than a panel of developed markets. Boyd and Prescott (1986) conclude that banks play a key role in reducing the loss of information leading to more efficient allocation of resources, while other studies propose that the financial sector underperforms in terms of resource allocation compared to the banking sector (Bhide, 1993; Stiglitz, 1985). Another group of studies stress that stock markets overcome banks in promoting economic growth. These theories emphasize that the financial sector is not only free from the monopolistic and conservative environment that the banking sector impose to the economy but also encourages innovation and competition stimulating growth in more effective ways. Shen and Lee (2006) suggest that only stock market development significantly impacts growth, by employing simple Ordinary Least Squares (OLS) and two-stage OLS procedures for a panel of 48 countries (see also, Allen & Gale, 2000). A different stream of literature proposes that both banks and stock markets boost economic growth, since these two components of the financial system improve in different ways the diffusion of information and allocation of resources. Levine and Zervos (1998) examine the relationship between banking sector development, stock market development and growth for a cross-section panel of 45 countries. They conclude that both components of the financial system are significant predictors of growth. Bittencourt (2012) finds that financial development significantly impacts growth by employing time series and panel data for four Latin American countries, while Campos, Karanasos, & Tan (2012) in their research for the case of Argentina, they also conclude that financial development impacts growth in a statistically significant and positive way (see also, Zhang, Wang, & Wang, 2012; Hasan, Wachtel, & Zhou, 2009; Beck & Levine, 2004). Finally, some recent studies, for example Anwar and Sun (2011), dispute on the positive relationship between the components of the financial system and growth. Their study employs the GMM estimator methodology on annual data for the period (1970–2007) suggesting that the level of financial development has contributed to the growth of the domestic capital stock in Malaysia but its impact on economic growth is statistically insignificant. An increase in the stock of foreign investment in Malaysia has contributed to an increase in the stock of domestic capital and economic growth but the stock of foreign investment is affected significantly only by the level of trade openness and its real exchange rate. To summarize, the majority of the theory suggests that the financial system and its components (banks and stock markets) significantly affect economic growth in positive and in some cases negative ways, depending on characteristics, such as, the levels of economic development, regulation effectiveness, competitiveness and/or market volatility and maturity.

44

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

3. Data analysis and methodology 3.1. Data analysis This study employs balanced panel data for the 28 member states of the European Union1 (EU28) for the period 1996–2012. This dataset captures almost the entire course of the reform process in the countries examined.2 The reforms were adopted gradually and supported the further improvement of the institutional framework as well as the more competent functioning of banks and financial markets in general, also in light of the participation in the Eurozone. Although all 28 countries are linked at economic, financial and trade levels, probably the most important dissimilarity lies on the fact that only seventeen member states have adopted the Euro as their national currency. As analyzed in the introductory section, this differentiation leads to two separate levels of intensity of economic cooperation. Therefore, to assess these differences at monetary level and present more accurate results on the finance–growth nexus, this study divides the full EU28 panel data set, based on whether an EU country is a member of the Euro area, creating the Eurozone panel3 and the non-Euro countries panel.4 The selected sample covers all Eurozone members as in 2011 (see also Leroy, 2014, adopting a similar practice). The hypothesis behind this panel division is that due to the various dissimilarities within the European Union, investigating only the full regional EU data set — let alone the regional European panel — may lead to ambiguous results and conclusions. To accept this hypothesis, significantly different empirical results and conclusions on the finance–growth relationships are expected, when running the GMM models for the full EU28 panel data set and for the above sub-sets. To investigate these relationships, the following variables are employed in the estimation procedure: Economic growth (GGDP), gross fixed capital formation as a percentage of GDP (GFCF), inflation (INFL), trade openness as a percentage of GDP (OPEN), market capitalization of listed companies as a percentage of GDP (MCAP), bank’s domestic credit as a percentage of GDP (CREDIT), and stocks traded as a percentage of GDP (STOCKS). All data are derived from the World Development Indicators (WDI), a reliable World Bank database. Market capitalization and stocks traded are used as proxies for the financial sector development. The literature on this topic reveals four functions of the financial sector, which are considered to promote economic growth; (i) financial intermediaries enable savings to be placed into long-term assets, which are more productive than short-term assets (Bencivenga & Smith, 1991), (ii) financial markets enable investors to diversify risk and promote liquidity. In the absence of financial markets, investors constrained by liquidity shocks are forced to liquidate long-term investments, hurting economic growth (Obstfeld, 1994), (iii) the financial sector provides better corporate control (Nieuwerburgh, Buelens, & Cuyvers, 2006), and (iv) stock markets are able to deliver evaluation and monitoring services more efficiently than individual investors, ensuring an efficient allocation of funds over the life of investment projects, due to acquiring information in advance (King & Levine, 1993b). On the other hand, a sharp rise of the stock markets may occur as a result of speculative pressure and generalized euphoria caused by the financial liberalization (Keynes, 1936; Singh, 1997). This bull market situation may increase interest rates significantly due to higher market risk, freezing investments and affecting negatively economic growth as a consequence (Federer, 1993). Devereux and Smith (1994) argue that risk diversification in trading activities may cause savings’ rates to decline, hurting economic growth, while same results may occur due to the use of market manipulation strategies through in-excess stock trading (De Long, Shleifer, Summers, & Waldmann, 1989). Finally, Fry (1997) concludes that stock markets appear to be weak intermediaries between households and businesses. Moreover, to assess the bank-led growth hypothesis, this study considers bank’s domestic credit as a proxy for the banking sector development, which is consistent with the literature. Several studies use this variable, as indicator of the banking sector development. One reason is due to the availability of data for a large number of countries. Banks compared to stock markets, intermediate differently, since equity funding is associated with business performance. However, negative effects on growth may arise not only from financial intermediation but from bank intermediation too. For example, banks tend to allocate credit, considering primarily transaction costs and the risk of default and ignoring the expected return on investment, while shrinking the average investment efficiency. Interest rates decline, since investments with lower returns can now become profitable in favoring current consumption against future consumption leading to a severe decline in savings’ rates with significantly negative social implications. Further to the selected proxies of the financial system, three macroeconomic determinants of growth are also included as variables, namely, capital formation, trade openness and inflation. The literature provides exhaustive evidence on the relationship between these variables and economic growth. However, with the exception of capital formation, where there is a general consensus in the relevant theory regarding its positive impact on growth, the literature provides opposing views for the last two selected variables. For example, Grossman and Helpman (1992) conclude that trade openness has a positive

1 The EU28 member states are: Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, and the United Kingdom. 2 This may be quite important considering the fact that in many transition countries the reform process has been completed during the 2000s, while in others this process is still underway. 3 The Eurozone countries are: Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Malta, the Netherlands, Portugal, Slovakia, Slovenia, and Spain. 4 The non-Euro countries are: Bulgaria, Croatia, Czech Republic, Denmark, Hungary, Latvia, Lithuania, Poland, Romania, Sweden, and the United Kingdom.

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

45

effect on new technology through the imports of goods and services. When new technology is applied in the domestic production process, productivity raises and becomes more efficient having a direct positive impact on economic growth. In a similar way, Levine and Renelt (1992) argue that through trade openness countries are exposed to a greater variety of profitable investment opportunities promoting indirectly economic growth. Finally, inflation is regarded in numerous studies as a significant growth factor. Given the assumption that money and capital are substitutes, a rise in inflation will increase the cost of holding money, leading to a shift from money to capital that causes a decline in interest rates and thus increases investments and growth (Mundell, 1965; Tobin, 1965). Other studies, however, argue that inflation increases the cost of investment, which in turn disrupts rational decision making and hurts productivity growth (Stockman, 1981; De Gregorio, 1993; Fischer, 1993). Furthermore, an increase in inflation may lead investors either to stall their investments (Pindyck & Solimano, 1993) or even encourage capital flight due to uncertainty (Fischer, 1993). 3.2. Methodology Based on the theoretical framework as briefly presented in Section 3.1, this study employs the Arellano and Bond Generalized-Method-of-Moments (GMM) estimator.5 Therefore, the cross-country growth regression is of the following form:

y i ,t = a0 y i ,t −1 + β X i,t + γ M i,t + ηt + εi ,t ,

i = 1, . . . , N , t = 1, . . . , T ,

(1)

where y refers to real GDP growth rate; α0 , β and γ are parameters to be estimated; X represents the set of macroeconomic explanatory variables, namely, gross fixed capital formation as a percentage of GDP, trade openness as a percentage of GDP and inflation; M is a vector of financial and banking sector development, employing stocks traded as a percentage of GDP, market capitalization as a percentage of GDP and bank’s domestic credit as a percentage of GDP as proxies; η stands for the unobserved country specific effect; ε is the error term; i and t represent country and time period, respectively.6 Moreover, Arellano and Bond (AB, 1991) propose to difference Eq. (1) in order to solve the underlying problem of correlation between the dependent variable and the error term and eliminate country specific effects. Although differencing eliminates the country specific effects and E (εi ,t − εi ,t −1 ) = 0, however, ( y i ,t −1 − y i ,t −2 ) is not independent of (εi ,t − εi ,t −1 ). The Arellano and Bond method solves this problem by employing two or more lags of the first difference of the growth rate as instruments. In regards with ( X i ,t − X i ,t −1 ) and (M i ,t − M i ,t −1 ) this study assumes that financial system variables and control variables are predetermined considering that E ( X i ,t , εi ,s ) = 0 and E ( M i ,t , εi ,s ) = 0 for s < t but zero for s ≥ t. Therefore, for these predetermined variables, one or more lagged levels are orthogonal to the differenced error term and thus form valid instruments for the respective first differenced explanatory variables. However, this study does not limit its methodological applications only to Arellano and Bond (1991) GMM estimator. The reasons are threefold: First, the country specific relationship between the financial system development and economic growth also presents significant research interest. However, the cross-country effect is eliminated in the AB-GMM estimator. Second, differencing may exacerbate the bias due to measurement errors in variables by decreasing the so-called “signalto-noise ratio” (see Griliches & Hausman, 1986). Finally, a third reason is to cross-check the robustness of the differenced GMM estimator. To address these issues, this study estimates again this model by employing the system dynamic panel data estimator presented by Arellano and Bover (1995) and Blundell and Bond (1998). The system-GMM estimator combines in a system the regression in differences with the regression in levels. The instruments for the differenced estimator rest the same, as above. However, the instruments for the regression in levels are the lagged differences of the corresponding variables. These are regarded as appropriate instruments under the assumption that although there may be correlation between the levels of the explanatory variables and the country-specific effect in Eq. (1), there is no correlation between the differences of these variables and the country-specific effect (Arellano & Bover, 1995). Another decision to be made is the choice between the one-step and the two-step estimators for both the differenced and the system GMM estimators. Arellano and Bond (1991) and Blundell and Bond (1998) argue that the asymptotic inferences based on the one-step estimator are more reliable in the sense that they have correct empirical size distributions. Considering the above argument, this study employs the one-step estimator, which is in line with recent studies that utilize this methodology (Narayan & Narayan, 2013; Cole et al., 2008). Finally, it should not be overlooked that the consistency of the GMM estimator lies on; (i) the validity of the assumption that the error terms are not serially correlated, and (ii) the validity of the instruments. To address these issues two commonly used diagnostic tests are employed, namely, the Sargan test and the Arellano and Bond (1991) test for autocor-

5 The endogeneity of explanatory variables can make the coefficient estimates obtained through the traditional OLS estimation biased and inconsistent (Greene, 2000). To take into account the possibility of endogeneity, following Arellano and Bond (1991) and Blundell and Bond (1998), we apply the system-GMM estimators. Using the previous observations of variables, i.e. the time-series dimension of panel data, as instruments, the GMM framework takes into account the fact that explanatory variables may be endogenous or at least weakly exogenous, and provides consistent coefficient estimates. 6 To guarantee robustness we control for country heterogeneity and temporal variation in the above specifications through the appropriate use of dummy variables (see Baltagi, 2001). These dummy variables have been found jointly statistically significant in virtually all equations, but they are not reported to save space.

46

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

Table 1 Descriptive statistics. GFCF

INFL

OPEN

MCAP

CREDIT

STOCKS

Panel A: Full sample of EU28 countries Mean 2.457 Median 2.847 St. Dev. 3.696 Skewness −1.137 Kurtosis 7.204 CV 1.504 Obs. 392

GGDP

21.584 21.073 3.995 0.653 3.862 0.185 392

3.459 2.576 4.274 6.399 57.916 1.235 392

110.572 97.796 51.927 1.574 6.559 0.470 392

53.117 39.106 46.324 1.730 7.433 0.872 392

361.414 108.052 1779.292 7.318 57.413 4.923 392

39.007 14.474 53.563 2.001 7.948 1.373 392

Panel B: Eurozone countries Mean 2.169 Median 2.326 St. Dev. 3.384 Skewness −0.752 Kurtosis 5.628 CV 1.560 Obs. 238

21.553 21.027 3.863 0.622 4.536 0.179 238

2.710 2.476 1.824 1.540 9.429 0.673 238

117.835 103.852 61.586 1.248 4.720 0.523 238

59.801 51.018 47.041 1.932 8.908 0.787 238

543.552 127.706 2266.332 5.606 34.201 4.169 238

42.046 21.739 50.889 1.597 5.304 1.210 238

Panel C: Non-Euro countries Mean 2.901 Median 3.625 St. Dev. 4.103 Skewness −1.577 Kurtosis 8.520 CV 1.414 Obs. 154

21.633 21.128 4.203 0.681 3.852 0.194 154

4.617 2.941 6.270 4.620 28.372 1.358 154

99.348 92.803 28.403 0.513 2.731 0.286 154

42.790 25.915 43.353 1.453 4.102 1.013 154

79.926 62.437 55.331 1.130 3.353 0.692 154

34.310 7.027 57.304 2.505 10.837 1.670 154

Note. GGDP: economic growth; GFCF: gross fixed capital formation as a percentage of GDP; INFL: inflation; OPEN: trade openness as a percentage of GDP; MCAP: market capitalization of listed companies as a percentage of GDP; CREDIT: bank’s domestic credit as a percentage of GDP; STOCKS: stocks traded as a percentage of GDP.

relation.7 The Sargan test of over-identifying restrictions, tests the overall validity of the instruments by assessing whether the instruments are uncorrelated with the error terms in the estimated equation.8 The null hypothesis in this case is that the instruments, as a group, are exogenous. Thus, failure to reject the null hypothesis could provide evidence that valid instruments are used. Second, we test whether there is a second order serial correlation with the first differenced errors. We test whether the differenced error term is second-order serially correlated. Under the null hypothesis of no second-order serial correlation, this test has a standard-normal distribution. The GMM estimator is consistent if there is no second order serial correlation in the error term of the first-differenced equation. The null hypothesis in this case is that the errors are serially uncorrelated. Thus, failure to reject the null hypothesis could supply evidence that valid orthogonality conditions and instruments are used. The second test examines the null hypothesis that the error terms are not serially correlated. This study reports only the test statistics and p-values for AR(2) because it is able to detect possible autocorrelation in levels. Therefore, failure to reject the null hypothesis for both tests implies robustness of the model. 4. Empirical results 4.1. Descriptive statistics Table 1 presents some selected descriptive statistics, namely, mean, standard deviation (St. Dev.), skewness, kurtosis and coefficient of variation (CV) as well as for all variables employed in this study, namely, economic growth (GGDP), gross fixed capital formation as a percentage of GDP (GFCF), inflation (INFL), market capitalization as a percentage of GDP (MCAP), bank’s domestic credit as a percentage of GDP (CREDIT), and stocks traded as a percentage of GDP (STOCKS). These summary statistics are reported not only for the full sample of the 28 countries of the European Union (EU28) but also for the two created sub-sets, under the “Euro-adoption criterion”, namely, the Eurozone panel (Panel B) and the non-Euro countries panel (Panel C). The summary statistics reveal a wide variation across the variables and the panels created. In terms of economic growth, assessing the currency-based splitting of data (Panels B and C), the highest average growth rate is reported for the non-Euro countries (2.9%), while GGDP for the Eurozone countries present higher variability. Moving on to the selected macroeconomic determinants of growth, GFCF and INFL average around 21.6% and 4.6% respectively, presenting higher figures for

7 A statistic commonly used is the R 2 . However, for dynamic models with endogenous variables, these indicators may not be sufficiently informative (Baum, Schaffer, & Stillman, 2003). 8 If not, the results are consistent with the presence of measurement errors, since the instruments used would be weak.

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

47

Table 2 D-GMM and S-GMM results for the full sample of EU28 countries.

GFCF INFL OPEN MCAP

D-GMM1

D-GMM2

D-GMM3

S-GMM1

S-GMM2

S-GMM3

0.657∗∗ (0.019) −0.163 (0.573) 0.103∗∗∗ (0.000) 0.035 (0.416)

1.816∗∗∗ (0.000) −0.072 (0.719) 0.281∗∗∗ (0.000)

1.953∗∗∗ (0.000) −0.022 (0.914) 0.285∗∗∗ (0.000)

0.484∗∗ (0.011) −0.086∗ (0.074) 0.298∗∗∗ (0.006) 0.086 (0.412)

1.655∗∗∗ (0.000) −0.162 (0.113) 0.319 (0.000)

1.611∗∗∗ (0.000) −0.166 (0.544) 0.379∗∗∗ (0.000)

−0.000 (0.689)

CREDIT STOCKS Sargan test AR(2)

14.596 (1.000) −0.195 (0.432)

12.897 (1.000) −0.284 (0.374)

0.045∗∗ (0.027) 15.614 (1.000) −0.241 (0.458)

−0.001 (0.569)

13.968 (1.000) −0.336 (0.493)

20.216 (1.000) −0.415 (0.519)

0.125∗∗ (0.013) 11.879 (1.000) −0.449 (0.537)

Note: GGDP: economic growth; GFCF: gross fixed capital formation as a percentage of GDP; INFL: inflation; OPEN: trade openness as a percentage of GDP; MCAP: market capitalization of listed companies as a percentage of GDP; CREDIT: bank’s domestic credit as a percentage of GDP; STOCKS: stocks traded as a percentage of GDP. The coefficients of the time and country dummies have been omitted from the regression output but are available upon request. AR(2): Arellano–Bond test that average autocovariance in residuals of order 2 is 0 (H0: no autocorrelation); Sargan: the test for over-identifying restrictions in GMM dynamic model estimation. The p-values are presented in parentheses (∗∗∗ , ∗∗ and ∗ indicate 1 per cent, 5 per cent and 10 per cent significance levels, respectively).

coefficient of variation (CV) than results from the Eurozone panel. On the other hand, OPEN presents an impressively higher mean for the case of Eurozone panel, reaching 117.8%, presenting a noteworthy higher CV too. Moreover, to briefly assess these statistics for the key explanatory variables, significant disparities exist across Panels B and C. The proxies of financial sector development, namely MCAP and STOCKS for Eurozone countries present higher average results than non-Euro countries (around 60% and 42% respectively) and lower variation according to the CV calculations. On the other hand, CREDIT, which is employed as a proxy of the banking sector development, reports more than six times higher mean and CV results in the case of the Eurozone panel (around 544%) than in the case of non-Euro countries panel. Finally, an overview on skewness and kurtosis statistics indicates that all data do not follow a normal distribution. 4.2. Results for the full panel of EU28 member states Table 2 reports the findings from the full panel set of the EU28 member states. The models seem to fit the panel data reasonably well, having fairly stable coefficients, while the Sargan test shows no evidence of over-identifying restrictions.9 As analyzed in the previous section, this study employs both the differenced (D-GMM) and system (S-GMM) GMM estimators. Therefore, columns 2–4 report the results from the application of the differenced-GMM estimator, while columns 5–7 tabulate the results from the system-GMM estimator. For each GMM estimator, three models are reported, in order to assess the impact of the key explanatory variables (MCAP, CREDIT and STOCKS) on economic growth separately. Moreover, the three selected macroeconomic variables (GFCF, OPEN and INFL) are included in all six models. Column 2 of Table 2 presents the results for the first differenced-GMM estimator model (D-GMM1), which employs market capitalization (MCAP) as a proxy of the financial sector development. These estimations report insignificant impact of MCAP on economic growth (GGDP). Moving on to the results for the macroeconomic variables, Gross fixed capital formation (GFCF) appears to have a positive and strongly significant effect on growth, at the highest significance level (1%), which is in line with the majority of studies on this topic. The last variable in Model 1 that reports significant and positive results is trade openness (OPEN) also at the 1% significance level. Moving on to Model 2 (D-GMM2), similar results are presented regarding GFCF, since it appears to promote growth at the 1% significance level, while trade openness also promotes economic growth significantly at the 1% significance level. However, Model 2 fails to trace significance for the case of bank’s domestic credit (CREDIT). Therefore, we are obliged to reject the bank-led growth hypothesis based on these findings. Model 3 (D-GMM3) results also imply that GFCF and OPEN are strongly significant growth factors, since both variables positively affect growth at the 1% significance levels. Finally, Model 3 of D-GMM reports statistically significant and positive evidence regarding the impact of STOCKS on growth at 5% significance level. The results for the system-GMM estimator (S-GMM) are very similar to those from the D-GMM estimator. GFCF and OPEN appear to significantly promote growth across all three models of the S-GMM estimator. Significant and positive results are

9 In order to ensure robustness we compared the various consistent GMM estimators (system GMM) to simpler static estimators, which are likely to be biased in opposite directions as regards the lagged dependent variables in panels with a small time dimension. In the first estimates, standard errors are biased, as discussed previously, although the bias is expected to be small. Indeed, the difference in the coefficients between the two estimation methods is found to be significant.

48

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

Table 3 D-GMM and S-GMM results for the Eurozone countries.

GFCF INFL OPEN MCAP

D-GMM1

D-GMM2

D-GMM3

S-GMM1

S-GMM2

S-GMM3

0.835∗∗∗ (0.000) 0.114 (0.568) 0.216∗∗∗ (0.001) 0.056∗∗∗ (0.009)

0.842∗∗∗ (0.000) 0.070 (0.701) 0.289∗∗∗ (0.000)

0.558∗∗∗ (0.002) 0.254 (0.201) 0.231∗∗∗ (0.000)

0.547∗∗∗ (0.000) 0.277 (0.241) 0.268∗∗∗ (0.000) 0.047∗∗∗ (0.000)

0.881∗∗∗ (0.000) 0.033 (0.761) 0.303∗∗∗ (0.000)

0.552∗∗∗ (0.000) −0.016 (0.982) 0.229∗∗∗ (0.000)

−0.000 (0.463)

CREDIT STOCKS Sargan test AR(2)

12.354 (0.513) −0.4935 (0.613)

15.714 (0.314) −0.534 (0.441)

0.083∗∗∗ (0.009) 13.242 (0.551) −0.472 (0.637)

−0.000 (0.132)

11.959 (0.448) −0.513 (0.427)

14.714 (0.257) −0.534 (0.441)

0.083∗∗∗ (0.000) 12.814 (0.383) −0.472 (0.539)

Note. GGDP: economic growth; GFCF: gross fixed capital formation as a percentage of GDP; INFL: inflation; OPEN: trade openness as a percentage of GDP; MCAP: market capitalization of listed companies as a percentage of GDP; CREDIT: bank’s domestic credit as a percentage of GDP; STOCKS: stocks traded as a percentage of GDP. The coefficients of the time and country dummies have been omitted from the regression output but are available upon request. AR(2): Arellano–Bond test that average autocovariance in residuals of order 2 is 0 (H0: no autocorrelation); Sargan: the test for over-identifying restrictions in GMM dynamic model estimation. The p-values are presented in parentheses (∗∗∗ , ∗∗ and ∗ indicate 1 per cent, 5 per cent and 10 per cent significance levels, respectively).

also documented in the case of STOCKS, at the 5% significance level, while S-GMM estimator failed to report significant relationships between MCAP, CREDIT and GGDP. The latter findings are also in accordance to D-GMM results. Overall, the results from the application of both the differenced-GMM and the system-GMM estimators for the full panel of the EU28 member states are similar across all six models. Regarding the financial sector developments, namely, market capitalization and stocks traded, only STOCKS is reported as a significant growth factor, documenting however the positive impact of stock markets on economic growth. Moreover, these models failed to report significant relationships between the banking sector variable (bank’s domestic credit) and growth. Therefore, we are obliged to accept the null hypothesis of no significant impact of CREDIT on GGDP. Moreover, an overview on the results for the macro-determinants of growth concludes that all variables appear to be significant growth factors, with the exception of INFL, where S-GMM2 is the only model that reports a weakly significant and positive impact on GGDP at 10% significance level. Therefore, due to the limited significant results in this case, we assume that no significant impact exists between INFL and growth in the case of EU28 panel set. Finally, the last two rows of Table 2 report the results from the examination of two commonly used diagnostic tests; (i) the Sargan test, under the null that the instruments employed are exogenous, and (ii) the AR(2) test, under the null of no autocorrelation in relation with the differenced residuals. Both tests are found to be insignificant across all models of the D-GMM and S-GMM estimators, since the reported p-values in parentheses are greater than 0.10. Therefore, these results support the robustness of all six models, implying that the selected instruments are exogenous and that all models are free from autocorrelation.10 4.3. Results from the Eurozone and the non-Euro countries panels Table 3 presents the results from the application of the D-GMM and S-GMM estimators for the Eurozone countries sub-set. Findings for the macroeconomic variables support the significant and positive contribution of GFCF and OPEN on growth, while INFL is found to have insignificant effect on GGDP. All six models applied, regardless of the type of the GMM estimator lead to these conclusions. On the other hand, results for the financial sector development variables indicate that stock markets are key growth factors, since both proxies (MCAP and STOCKS) appear to be strongly significant at 1% significance level and across all six models employed. However, CREDIT, is insignificant to growth, so, the contribution of the banking sector on GGDP cannot be supported. Comparing the results derived from the Eurozone panel to the empirical findings from the full EU28 sample (see Table 2), we find that similar results are reported, since GFCF, OPEN and STOCKS appear significant growth factors in both cases. The only noteworthy difference refers to MCAP, which is strongly significant in the case of Eurozone countries panel. Therefore, the new information gained from the application of GMM estimators on the Euro area data set, is limited to the significant estimations for both proxies of financial sector development, implying the strong relationships between the Eurozone’s

10 The Sargan and the AR(2) tests are presented in the last two rows of all subsequent tables. For all panels employed in this study, and for all types of GMM estimators, insignificant p-values are reported, implying that the selected instruments are exogenous and that all D-GMM and S-GMM models are free from autocorrelation in relation with the differenced residuals.

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

49

Table 4 D-GMM and S-GMM results for the non-Euro countries.

GFCF INFL OPEN MCAP

D-GMM1

D-GMM2

D-GMM3

S-GMM1

S-GMM2

S-GMM3

1.423∗∗∗ (0.000) −0.486∗∗ (0.014) 0.458∗∗∗ (0.000) −0.002 (0.941)

1.621∗∗∗ (0.000) −0.379∗∗ (0.049) 0.371∗∗∗ (0.001)

1.043∗∗∗ (0.000) −0.449∗∗∗ (0.008) 0.485∗∗∗ (0.000)

1.164∗∗∗ (0.000) −0.375∗∗ (0.041) 0.487∗∗∗ (0.000) −0.009 (0.739)

2.111∗∗∗ (0.008) −0.104∗ (0.086) 0.239∗∗ (0.041)

1.687∗ (0.081) −0.258∗ (0.056) 0.327∗∗ (0.017)

CREDIT

−0.281∗∗∗ (0.002)

STOCKS Sargan test AR(2)

14.697 (0.569) −0.648 (0.553)

11.634 (0.314) −0.719 (0.621)

0.115 (0.163) 17.639 (0.598) −0.534 (0.447)

15.697 (0.617) −1.314 (0.694)

−0.292∗∗ (0.013)

11.319 (0.319) −0.693 (0.634)

0.036 (0.583) 9.639 (0.341) −0.431 (0.536)

Note. GGDP: economic growth; GFCF: gross fixed capital formation as a percentage of GDP; INFL: inflation; OPEN: trade openness as a percentage of GDP; MCAP: market capitalization of listed companies as a percentage of GDP; CREDIT: bank’s domestic credit as a percentage of GDP; STOCKS: stocks traded as a percentage of GDP. The coefficients of the time and country dummies have been omitted from the regression output but are available upon request. AR(2): Arellano–Bond test that average autocovariance in residuals of order 2 is 0 (H0: no autocorrelation); Sargan: the test for over-identifying restrictions in GMM dynamic model estimation. The p-values are presented in parentheses (∗∗∗ , ∗∗ and ∗ indicate 1 per cent, 5 per cent and 10 per cent significance levels, respectively).

stock markets and growth. Thereby, accepting the hypothesis that different results occur when separating EU member states according to whether they have adopted the Euro or not, remains partially inconclusive from these results. The estimations from the non-Euro countries panel are reported in Table 4. To begin with the selected macroeconomic variables, similar results are tabulated for the case of GFCF and OPEN, since both determinants of growth have statistically significant and positive impact on growth across all six models, although the significance levels vary across various types of GMM models. Interestingly, this is the first time in this study that statistical significance is reported for the case of INFL, which is found to negatively affect growth across all six types of GMM models. Moreover, regarding the selected proxies of stock markets, both of them present insignificant results, therefore we are obliged to reject the stock market-led growth hypothesis for the case of non-Euro countries. Finally, findings derived from both the D-GMM and S-GMM models reveal that the banking sector has a statistically significant and negative effect on growth at 1% and 5% levels respectively. Overall, the results from the non-Euro panel present different results than the rest panel sets, basically in three cases; First, INFL has a significant and negative effect on GGDP, which is in line with the majority of studies that deal with the inflation–growth relationships. Second, this is the first time that CREDIT appears to have a significant effect on growth, and third, these are the sole results that indicate the absence of significant impact of the financial sector development on growth, since both MCAP and STOCKS have insignificant effects on the non-Euro countries’ economic growth. Therefore, these findings support the hypothesis that different finance–growth relationships can be traced when dealing with data from EU member states that have not adopted the Euro. So, assessing these relationships separately from the Eurozone countries can provide valuable and presumably more accurate information on the finance–growth nexus. Therefore, this study implies that significant differences exist between Eurozone member states and non-Euro countries in terms of economic, political, institutional and structural development. Characteristics such as the ownership structure, the market concentration, the size and importance of different delivery channels and the presence of foreign banks vary substantially across the EU member states. This could have important consequences on financial stability, as it affects profit and risk trade-offs, cost efficiency and ultimately has a bearing on the shock-absorptive capacity of the financial system. Eurozone economies have concurrently been confronted in recent decades with common transnational developments that have transformed their national financial markets and allowed them to converge. These countries have implemented numerous common regulatory changes, motivated by the need to achieve the level of harmonization required for the establishment of a single, competitive market for financial services. During the last few years, on the other hand, the majority of non-Euro countries, with the collaboration and assistance of international financial institutions, have taken concrete and far-reaching measures to reform their financial institutions and markets. The main differences between the applied financial policies in these two panels and the diverse financial reformations employed, can be attributed to the different institutional and financial management that is demanded by the adoption (or not) of the common European currency. The financial and sovereign debt crisis of 2008, although it slowed down the development of most EU member states can be regarded as an excellent example of different fiscal and monetary management between Eurozone countries and non-Euro countries that can be attributed to the fact that common currency EU countries were not in a position to adjust separately the Euro currency exchange according to the needs of their national economies. This fact forced Eurozone countries to speed-up the pace of political and economic integration in order to minimize the consequences of the severe Crisis. Considering that the “common currency EU model” appeared to be rigid under the pressure of the Crisis and that Eurozone countries experienced sharper recession relative to the non-Euro countries this fact alone can be seen as evidence of the

50

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

legal and politico-institutional disparities that these two panels present. This feature that Eurozone countries present, along with the paradox of common currency adoption but maintaining autonomous to an extent political and fiscal policy, are probably the most significant differences between the two panels under study, as well as the main causes of the different empirical results as tabulated and presented previously in this section. 5. Summary, concluding remarks, and policy implications The main goal of this study is to assess the relationships between the components of the financial system, namely, the banking sector and financial sector developments, for the case of the expanded European Union, for the period 1996–2012. For this purpose, balanced panel data are employed. Moreover, this paper, for the first time in this literature, divides the full EU28 panel, under the “Euro-adoption criterion” (i.e. whether an EU member country has adopted the Euro (€) as a national currency or not) creating two sub-sets, the Eurozone panel and the non-Euro countries’ panel. The motivation behind this sample split is supported by the hypothesis, that single currency EU member states are more economically, politically and fiscally dependent on each other than the rest of the non-Euro countries. Therefore, conclusions from the division of the initial EU28 panel should be more accurate than those, when dealing with all the EU countries. In order to accept this hypothesis, empirical results should be different than those from the full EU28 panel. Considering the above, this study differs from similar studies that employ mixed regional European panels (Narayan, Mishra, & Narayan, 2011; Narayan & Narayan, 2013), not only by focusing on the expanded EU28, but also by dividing this already mixed panel in sub-sets under the Euro-adoption criterion. The EU is a special case of multi-national union that although promoting a close economic cooperation between the European member countries, academics, researchers and policy makers should consider that it is not a single country yet. Considering all these variations, even narrowing a European sample to an EU full sample could possibly lead to biased results. The empirical analysis employs the differenced and system-GMM estimators. To assess the impact of the components of the financial system on economic growth, two proxies are used for financial sector development, namely, market capitalization and stocks traded, while bank’s domestic credit is used as a proxy for the banking sector development. Results for the finance–growth nexus suggest that the financial sector weakly promotes growth in the case of full EU28 sample, only in terms of STOCKS, since MCAP variable has insignificant effect on growth. However, this impact appears stronger when assessing the validity of the stock market-led growth hypothesis in the case of Eurozone panel, since both proxies of financial sector development are reported as significant growth factors. On the other hand, empirical results from the non-Euro countries panel report insignificant results for the finance–growth nexus. The banking–growth relationships are insignificant in both cases of the full EU28 panel and the Eurozone panel, while results from the non-Euro panel support the significant and negative impact of the banking sector on growth. Furthermore, this study employs some selected macroeconomic determinants of growth, namely, gross fixed capital formation, inflation and trade openness. Results across all panels and types of GMM models applied, suggest that gross fixed capital formation and trade openness significantly promote economic growth. These findings are in line with the majority of the related literature, in the sense that capital stock and openness are seen as stimuli for economic growth. However, inflation is found to have insignificant effect on growth across all panels, with the exception of non-Euro countries sub-set, where significant and negative relationships are reported. In conclusion, a number of implications arise from the empirical findings; First, robust results from all panels imply that capital stock and trade openness are key contributors of EU’s economic growth. Thus, these findings are added to the existing literature on growth, that encourage public investments that promote citizen’s quality of life and government policies aimed at opening-up the national economy. Second, a prerequisite seems to be a review on the stock markets regulatory framework in the case of non-Euro countries, since the impact of stock markets on growth is totally absent from this panel. Third, the banking-led growth hypothesis is not only rejected from all panels, but also in the case of non-Euro countries panel, results suggest that banks hurt growth in the short-run. Therefore, this paper suggests that the EU needs to focus on further improving banking regulations and the credit allocation process, while restoring banking competition seems to be a necessity, in order for banks to become significant growth contributors. Finally, fourth, exhaustive empirical results from all panels support the validity of the hypothesis of this study that different results may occur, when dealing with the EU countries separately from the rest non-EU countries. Moreover, dividing the full EU28 sample under the “euro-adoption criterion” also leads to interestingly different results, and hence, to different policy implications. Therefore, this study provides stimulus for a more disaggregate analysis of the impact of banking and finance developments on growth. In particular, it would be interesting to analyze which of the elements (related to the developments) are the most relevant in promoting growth. Finally, another possible area for future research could be to provide a more detailed analysis of the different country-specific characteristics, dividing the full EU regional sample into sub-regional panels based on geographical, economic, cultural and social similarities. References Allen, F., & Gale, D. (2000). Comparing financial systems. Cambridge: MIT Press. Anwar, S., & Sun, S. (2011). Financial development, foreign investment and economic growth in Malaysia. Journal of Asian Economics, 22, 335–342. Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies, 58, 277–297.

A.G. Georgantopoulos et al. / The Journal of Economic Asymmetries 12 (2015) 41–51

51

Arellano, M., & Bover, O. (1995). Another look at the instrumental-variable estimation of error-components models. Journal of Econometrics, 68, 29–52. Bagehot, W. (1873). Lombard Street (1962 ed.). Homewood, IL: Richard D. Irwin. Baldwin, R. E., & Forslid, R. (1998). Trade and growth any unfinished business? European Economic Review, 42(3), 695–703. Baltagi, B. H. (2001). Econometric analysis of panel data (2nd ed.). Chichester: John Wiley & Sons. Baltagi, B. H., Demetriades, P. O., & Law, S. H. (2009). Financial development and openness: evidence from panel data. Journal of Development Economics, 89, 285–296. Baum, C. F., Schaffer, M. E., & Stillman, S. (2003). Instrumental variables and GMM: estimation and testing. Stata Journal, 3(1), 1–31. Beck, T., & Levine, R. (2004). Stock markets, banks, and growth: panel evidence. Journal of Banking and Finance, 28, 423–442. Bencivenga, V. R., & Smith, B. D. (1991). Financial intermediation and endogenous growth. Review of Economic Studies, 58, 195–209. Bencivenga, V. R., Smith, B. D., & Starr, R. M. (1995). Transaction costs, technological choice, and endogenous growth. Journal of Economic Theory, 67, 53–117. Bhide, A. (1993). The hidden costs of stock market liquidity. Journal of Financial Economics, 34, 1–51. Bittencourt, M. (2012). Financial development and economic growth in Latin America: is Schumpeter right? Journal of Policy Modeling, 34, 341–355. Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87, 115–143. Bordo, M. D., & Rousseau, P. L. (2012). Historical evidence on the finance–trade–growth nexus. Journal of Banking and Finance, 36, 1236–1243. Boyd, J. H., & Prescott, E. C. (1986). Financial intermediary-coalitions. Journal of Economics Theory, 38, 211–232. Campos, N. F., Karanasos, M. G., & Tan, B. (2012). Two to tangle: financial development, political instability and economic growth in Argentina. Journal of Banking and Finance, 36, 290–304. Cole, R. A., Moshirian, F., & Wu, Q. (2008). Bank stock returns and economic growth. Journal of Banking and Finance, 32, 995–1007. De Gregorio, J. (1993). Inflation, taxation and long-run growth. Journal of Monetary Economics, 31, 271–298. De Long, J., Shleifer, A., Summers, L., & Waldmann, R. (1989). The size and incidence of the losses from noise trading. Journal of Finance, 44, 681–696. Devereux, M., & Smith, G. (1994). International risk sharing and economic growth. International Economic Review, 35, 535–550. Federer, P. (1993). The impact of uncertainty on aggregate investment spending. Journal of Money, Credit, and Banking, 25, 30–45. Fischer, S. (1993). The role of macroeconomic factors in growth. Journal of Monetary Economics, 32, 485–512. Fry, M. J. (1997). In favour of financial liberalization. The Economic Journal, 107, 754–770. Georgantopoulos, A., & Tsamis, A. (2011). The triangular causal links between economic development, FDI and exports: evidence from Turkey. Middle Eastern Finance and Economics, 15, 66–75. Greene, W. (2000). Econometric analysis. Prentice Hall. Griliches, Z., & Hausman, J. A. (1986). Errors in variables in panel data. Journal of Econometrics, 31, 93–118. Grossman, G. M., & Helpman, E. (1992). Innovation and growth: technological competition in the global economy. Boston: MIT Press. Hasan, I., Wachtel, P., & Zhou, M. (2009). Institutional development, financial deepening, and economic growth: evidence from China. Journal of Banking and Finance, 33, 157–170. Hur, J., Raj, M., & Riyanto, E. Y. (2006). Finance and trade: a cross-country empirical analysis on the impact of financial development and asset tangibility on international trade. World Development, 34(10), 1728–1741. Kaminsky, G. L., & Reinhart, C. M. (1999). The twin crises: the causes of banking and balance-of-payments problems. American Economic Review, 89, 473–500. Keynes, J. M. (1936). The general theory of employment, interest, and money. London: Macmillan. King, R. G., & Levine, R. (1993a). Finance and growth: Schumpeter might be right. Quarterly Journal of Economics, 108, 717–737. King, R. G., & Levine, R. (1993b). Finance, entrepreneurship, and growth. Journal of Monetary Economics, 32, 513–542. Leroy, A. (2014). Competition and the bank lending channel in Eurozone. International Financial Markets, Institutions and Money, 31, 296–314. Levine, R., Loayza, N., & Beck, T. (2000). Financial intermediation and growth: causality and causes. Journal of Monetary Economics, 46, 31–77. Levine, R., & Renelt, D. (1992). A sensitivity analysis of cross-country growth regressions. American Economic Review, 82, 942–963. Levine, R., & Zervos, S. (1998). Stock markets, banks and economic growth. American Economic Review, 88, 537–558. Loayza, N. V., & Ranciere, R. (2006). Financial development, financial fragility and growth. Journal of Money, Credit, and Banking, 38, 1051–1076. Mundell, R. (1965). Growth, stability and inflationary finance. Journal of Political Economy, 73, 97–109. Narayan, K. P., Mishra, S., & Narayan, S. (2011). Do market capitalisation and stocks traded converge? New global evidence. Journal of Banking and Finance, 35, 2771–2781. Narayan, K. P., & Narayan, S. (2013). The short-run relationship between the financial system and economic growth: new evidence from regional panels. International Review of Financial Analysis, 29, 70–78. Nieuwerburgh, S. V., Buelens, F., & Cuyvers, L. (2006). Stock market development and economic growth in Belgium. Explorations in Economic History, 43, 13–38. Obstfeld, M. (1994). Risk taking, global diversification, and growth. American Economic Review, 1310–1329. Pindyck, R., & Solimano, A. (1993). Economic instability and aggregate investment. NBER Macroeconomic Annuals, 8, 259–303. Schumpeter, J. (1911). Theorie der Wirtschaftlichen Entwicklung Leipzig. Germany: Duncker and Humblot. Schumpeter, J. A. (1934). The theory of economic development. Cambridge: Harvard University Press. Shahbaz, M. (2012). Does trade openness affect long run growth? Cointegration, causality and forecast error variance decomposition tests for Pakistan. Economic Modeling, 29(6), 2325–2339. Shen, C. H., & Lee, C. C. (2006). Same financial development yet different economic growth — why? Journal of Money, Credit, and Banking, 38, 1907–1944. Singh, A. (1997). Financial liberalisation, stockmarkets and economic development. The Economic Journal, 107, 771–782. Stiglitz, J. E. (1985). Credit markets and the control of capital. Journal of Money, Credit, and Banking, 17, 133–152. Stockman, A. C. (1981). Anticipated inflation and the capital stock in a cash-in-advance economy. Journal of Monetary Economics, 8, 387–393. Taylor, M. A., & Wilson, F. L. J. (2011). International trade and finance: complementaries in the United Kingdom 1870–1913 and the United States 1920–1930. Journal of International Money and Finance, 30(1), 268–288. Tobin, J. (1965). Money and economic growth. Econometrica, 33, 671–684. Yanikkaya, H. (2003). Trade openness and economic growth: a cross-country empirical investigation. Journal of Development Economics, 72(1), 57–89. Zhang, J., Wang, L., & Wang, S. (2012). Financial development and economic growth: recent evidence from China. Journal of Comparative Economics, 40, 393–412.