The Nexus between Financial Development and Economic Growth: Evidence from European Countries

The subprime crisis brought new challenges for the European countries that’s why this study examines the relationship between banking sector development, stock market development and economic growth, using annual data, for the period 1990-2015, in twelve European Countries (Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal and Spain). The principal component analysis was used to construct two new measures (banking sector development and stock market development). The Panel Vector Auto-Regressive model developed by Love and Zicchino (2006) and Granger causalities test was used. Results show that the model is endogenous, stable and that the shocks caused by the introduction of the euro and the subprime crisis are significant. By using dummies tools to control the crisis effects, the banking sector development and the stock market development show a bidirectional relationship. The results suggest that governments should implement stability policies of the banking sector development to attract foreign direct investment that impulses economic growth. Future research that evolves the nexus between financial development and economic growth should take into consideration the impact of the economic crisis in the countries.


Introduction
Economic growth is a recurrent concern in all economies. Decision-makers propose and implement several measures seeking more economic and commercial advantages (e.g. the implementation of the euro). In the last few years after the subprime crisis even more attention is given to the theme.
Thus, the objective of this paper is to analyse twelve European countries (Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal and Spain) by using a Panel Vector Auto-Regressive (PVAR) model. To verify the following hypotheses: i) The model is endogenous; ii) the shocks caused by the introduction of the euro and the subprime financial crisis are significant; and iii)the banking sector development in the presence of stock market development are significant.
So, to realize this, the study used GDP to capture the economic growth, and other important variables, such as inflation, trade openness to measure economic openness, foreign direct investment to measure financial openness, and used Principal Component Analysis (PCA) to create two single variables to assess stock market development and banking sector development.
Two dummies variables were created to capture the structural break in the model. The first of its kind for the euro implementation period, being statistically significant in the year 2001. The second shift type for the subprime crisis (from 2008 to 2010). The two dummies were significant in the model.
The results still indicate that in future research involving the banking markets and the stock markets, it's important to consider the shocks highlighted here. Also important is to define that the stock market is an important sector for developed economies, so governments must take these results into consideration and apply policies that increase and stabilize the structure of the stock market. so that investors can trust and be motivated to invest in stocks.
The structure of this study is divided as follows: Section 2 is dedicated to the literature review; Section 3 shows the data, methodology and the construction of the composite measures of BSD and SMD; Section 4 presents the empirical results and discussion; and Finally, Section 5 includes the concluding remarks.

Literature Review
Since it's such a wide and interesting subject to study, many researchers focus their attention on economic growth and its connection with all the other economic variables, such as banking sector development, stock market development, inflation, trade, foreign direct investment. It is known that since Schumpeter (1911), an analysis of the interaction between economic growth and alternatives has been used with safety. Goldsmith (1969), McKinnon (1973) and Shaw (1973) focus on the link between financial sector development and economic growth. While Lucas (1988) shows the financial sector only for economic growth. However other studies like Liu and Hsu (2006) Montes and Tiberto (2012), stress that in the long-run, stock market development is key in fostering economic growth.
The use of the inflation variable in studies on economic growth are common. Authors such as Boujelbene and Boujelbene (2010); Barro (2013); and Jalil, Tariq, and Bibi (2014) defend that controlled and stable inflation promotes business and investment decisions. So, it's obvious that inflation and stock market development are related.
According to the vast literature, it is difficult to identify whether it is economic growth that drives the other variables (e.g. inflation, trade liberalization, foreign direct investment, banking sector development, stock market development) or, if it's the variables that drive economic growth. With the support of empirical / mathematical analysis, it's possible to identify these relationships. Thus, it's also possible to categorize them in terms of the causal relationship between the variables, in four hypotheses, namely: i) Neutrality hypothesis -when there is no causality between variables, means that the variables are independent of each other; ii) Supply-leading hypothesis -when exists unidirectional causality between variables, means causality running from variables to economic growth; iii) Demand-following hypothesis -when exists unidirectional causality between variables, means causality running from economic growth to one or more variables; and iv) Feedback hypothesis -when there is a bidirectional causality between variables, means that the causality runs in both directions.
In Table 1 Table 1: Resume of the studies on causality with economic growth Table 1, shows that the relationship between economic growth and other economic variables. Several authors show a great concern with the topic, since it has been extensively studied in recent years. Among the countries and group of countries studied, researchers found several directions in causality relationship between the variables.

Data and Methodology
To test the relationship between economic growth, banking sector development and stock market development, other control variables were added, namely: inflation, economic and financial openness, so to detect the causality between variables, the estimation is realized by using a Panel Vector Auto-Regressive model (see Abrigo and Love 2015). Next, the data used in this paper are presented and this section is   The robustness of the composite index Banking Sector Development (BSD) was verified by the application of Bartlett's test for sphericity (Bartlett, 1950) and Kaiser-Meyer-Olkin of sampling adequacy (Kaiser, 1970), and table 3 presents the results of the test. Kaiser-Meyer-Olkin sampling adequacy measure indicates values above 0.500, so it's possible to apply the PCA. In the case of the Bartlett's test for sphericity, the null hypothesis was rejected with a pvalue less than 5% (0.000). Also shows a good result of Chi-square and finally reveals that the variables are significantly correlated. Note, the variable SMT was not accepted in the PCA of the Stock Market Development.

Methodology
In this study is applied a technique that combines the regular VAR approach that treats as endogenous all the variables in the system, with the unobserved individual heterogeneity from a panel-data approach (Grossmann et al., 2014). The application of a Panel Data Vector Auto-Regressive model was developed by Love and Zicchino (2006), and used the same methodology. The mentioned model, a first-order PVAR, uses an equation stated as follows in equation 1: Where, ‫ݐݖ‬ is vector variables, in this study they are: dlGDP, INF, dlTRd, dlFDI dlBSD and dlSMD. All variables are in natural logarithm followed by their first differences except INF (inflation). Γ corresponds to the constant vector, Γ ଵ ‫ݖ‬ ௧ିଵ to the matrix polynomial, ݂ the fixed effects in the model, ݀ ,௧ the effects of time, and the term of random errors is ݁ ௧ .
A technique applied by Love and Zicchino (2006) called "Helmert Procedure" (Arellano and Bover, 1995), to solve the problem of fixed effects correlated with the regression related to delays of the dependent variables, usually average differentiation procedure is used to eliminate fixed effects, is also used in the model to avoid the occurrence of biased coefficients.  It is important to point out the existence of a structural brake on the panel that does not allow capturing the unit root, i.e. the test doesn't confirm the real stationary effect of the variables, so it was not performed.

Empirical Results and Discussion
In the previous section 3, a preliminary analysis was performed to verify if the PVAR model was the most appropriate. Thus, it was confirmed that the PVAR test was the most appropriate to analyse this nexus between the variables. Note that the PVAR model was estimated using one lag and that all variables are in natural logarithms in their first differences (except inflation which is an index).

The
Generalized-Method-of-Moments (GMM) was used in the estimation PVAR. In table 6, it is possible to see the test results, after the insertion of the dummies variables to control the shocks and make the model stable and consistent.
Just a note that initial a PVAR test was conducted without the presence of the shocks as presented in table 5, but the model was not stable, and not valid to realize the test, so it's necessary and important the use of shocks in the model.  dlFDI to dlGDP and dlFDI to dlBSD; (iv) dlTRD to dlGDP and dlTRD to dlBSD; (v) dlSMD to dlGDP and dlSMD to dl statistical significance at 1% level, unidirectional causality but at 5% level of statistical significance it's possible to find: (i) dlFDI to dlTRD; (ii) dlBSD to dlGDP.
So, the variable with less causality relationship is dlGDP, dlBSD and dl if the model was possible to conduct our two main variables will not have many relations with the economic growth.
The Eigenvalue test is shown in figure 1 and the results indicate the instability in the model. Eigen value stability condition without shocks presence of endogeneity the instability on it suggests that exist some shocks that need to be controlled in a way that is possible to get a The dummies variables, impulse and shift, were used to absorb outliers and structural impacts and were applied to year 2001 and 2010, respectively. It is important to highlight the use of dummies tool in this paper, to capture the effects of two main situations: (i) integration of the countries in the Monetary Union, because for all of them it was necessary to have monetary stability (relevant for integration); issues created by the physical change of the currency and etc.; (ii) economic distortion caused by the subprime crises which leads to foreign assistance in some countries. In table 6, show the results of PVAR with control of the shocks.
So, the variable with less causality and dlSMD, so ossible to conduct our two main variables will not have many relations with the economic growth.
is shown in figure 1, and the results indicate the instability in  So, the variable with less causality relationship is dlGDP, dlTRD and dlBSD, if we look at two of these measures (dlTRD and dlBSD), it's possible to identify that the European single market can be one of the reasons to reduce these relationships. On the other side, the variable with more causality connexion is dlSMD; this can be explained by the fact that the stock market sector has an important economic role in developed economies. In resume, it is possible to mention that inflation, foreign direct investment, trade and stock market development collaborate to economic growth.
A stability test was also conducted to check the estimations validation (Lütkepohl, 2005). The results satisfy stability condition, because all eigenvalues are inside the circle unit. The eigenvalue test shows the real, imaginary and modulus values, details under figure 2.

Eigen value stability condition with shocks
A good estimation of the PVAR model passes through three moments that were executed in this article: the accomplishment of the diagnostic tests before estimating the model, the execution model and finally tests of robustness With the results of the impulse-response figure 3, it's possible by a graphical analysis how the variables react to an exogenous shock, and eturn to is equilibrium. The first interpretation we can redraw from the graph is that variables recover from the shock in a years period. Second is possible to identify dlBSD as the variable more stability in response to the shocks in all other variables and that dlSMD is a variable that have an intense response on the beginning of the shock, but after a four years period stabilizes. Third is possible to verify that dlTRD and dlFDI after the shock occur, when stabilized, both stay on a lower level than before. The first interpretation we can redraw from the graph is that almost all variables recover from the shock in a four period. Second is possible to identify dlBSD as the variable more stability in response to the shocks in all other variables and that dlSMD is a variable that the beginning of the shock, but after a four years period Third is possible to verify that dlTRD and dlFDI after the shock occur, when stabilized, both stay on a lower level      According to the above results dlTRD have a feedback hypothesis with dlGDP and with dlBSD, so it's important that the European countries try to attract foreign direct investment from outside Europe so that they can maintain a wealth and sustainable growth and be less vulnerable to crises, as is it possible to.
The way how the economic agents and the governments interact in develop countries with a strong dependence from the banking system, like the case under study were the variable dlBSD represent feedback hypothesis between dlTRD and dlSMD should aim as a principal objective the social well-being and create measures between public and private agents to reach this goal.
Even solid markets and stable financial systems can be vulnerable to the negative effects generated by a global crisis similar to the one experienced in 2008. This vulnerability can be reduced or at least minimized if the policies encourage the agents to diversify the investments and the areas of expertise e.g. investments in tourism, so these new areas of expertise should be boost in a way to recover economies from crises periods.
Sustainable growth now a days is a global concern, so the developed countries should encourage the financial system to make contributions to this major concern by helping with social projects, sustainable development, income inequality and with the economy in general. Corroborate with the suggestion of Charfeddine and Kahia (2019) that the banking operations should support green energy projects and low carbon emissions projects.
In the next section, we show the conclusion and contribution of this study to the economy. The study also suggests a topic for future studies.

Conclusion
In this study we estimated an Auto-Regressive Vector Panel model for a panel of twelve European countries and a time span from 1990 to 2015. The relationship between economic growth and banking sector development and stock market development was analysed.
By using a Principal Component Analysis mode, it was possible to create two new measures, one for banking sector development and the other for stock market development, which includes four financial dimension measures: depth, efficiency, stability and other; this way it was possible to realize the study with composite to identify specific financial markets (bank and stocks). After estimation the three hypotheses of this study were verified and confirmed. The model is endogenous; the shocks (introduction of the euro and the subprime financial crisis) are statistically significant and should be considered in future studies; and finally, banking sector development and stock market development have significant statistical relationships with The estimation results respond to the previous hypotheses according the existence of feedback hypothesis between economic growth and economic openness, (i.e. governments should implement a wellsupported economic openness, promoting economic trade -import export of goods and services).
However, it also shows neutrality hypothesis between economic growth to stock market development and financial openness. It denotes also that inflation (INF), foreign direct investment (FDI) and stock market development (BSD) grangercause economic growth. So, to promote economic growth, the policies should induce a healthy and sustained external relation to help economic openness, also some good financial resources to support a good stock market development because there is a relationship with economic growth.
As a main contribution, it's possible to refer to the set of countries chosen, because they are a homogenous group with the same currency and similar economic shocks like the subprime crises, which had a huge impact over the economy, and to refer to the new aspect of including the economic and financial openness in the study to interrelate with the financial development. For future studies, it is important to get a set of countries with different economic spheres to confirm the impact of financial development and economic openness over the economy. Another interesting suggestion would be the individual analysis of the countries to be able to compare them with the results obtained in the studies that analyse the panel countries.