Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1
Financial Development Convergence
Berrak Bahadir and Neven Valev
September, 2013
Abstract
We show that credit levels relative to GDP and other measures for financial development tend to converge across countries over time. The results are obtained using a broad sample of countries over many years and controlling for the quality of country-level institutions, the efficiency of financial institutions, and a range of macroeconomic variables. While we find evidence for convergence in the broad sample, we show that it levels off when countries reach a medium level of financial development. At high levels of financial development, convergence slows down even more and becomes negligible.
JEL Codes: G01, E22
Key Words: Financial Development, Convergence, Institutions
Contacts: Bahadir, University of Georgia ([email protected]); Valev, Georgia State University ([email protected]).
2
Financial Development Convergence
I. Introduction The last several decades have witnessed massive financial liberalization in emerging
markets. For the financial markets, the objective has been to foster financial development by
inviting international know-how and to reduce the “knowledge gap” (Goldberg, 2007) in finance
between developed and developing countries. The objective for the real economy has been to
provide funding for capital investment and to stimulate economic growth, thus helping emerging
markets reach the standard of living achieved in developed economies (Aghion et.al. 2005).
While the benefits of financial development are well established, to date there is little
systematic evidence indicating whether the financial development in developing countries is
catching up with the levels achieved by advanced countries. Bridging the financial development
gap is likely to help reduce the income gap between countries. If there is divergence, then the
gap between developed and emerging countries will keep growing, with important ramifications
not only for the financial sector but also for the real economy. In this paper, we investigate
whether such convergence has occurred using a broad sample of countries with a primary focus
on credit markets. Our findings suggest that:
• there is an underlying process of financial development convergence but it ceases as
countries reach higher levels of financial development;
• the convergence cannot be attributed to improvements in the country-level institutions or
in the efficiency of financial institutions;
• it is most likely driven by expanding the financial sector to reach new customers in
underdeveloped countries; as the gains from such investment wear off with diminishing
returns, convergence slows down.
3
There is an important difference between studying the determinants of financial
development and studying financial development convergence. Financial markets have evolved
substantially in advanced economies during the last several decades reflecting significant
improvements in financial sector know-how. The question for emerging markets then is not only
whether its citizens can access more financial services over time but whether they take full
advantage of the available international finance know-how. To answer that question, we must
look at the differences between countries over time, and not only at the growth of financial
services within one country over time. Similar to the economic growth literature, where the goal
is to know if incomes increase and, separately, if they increase relative to advanced countries, in
the financial development literature the objective is to know: 1) whether the access to financial
services increases over time and 2) whether the increase in a particular country outpaces or lags
behind the development in the leader countries. The literature has addressed the former question
on the determinants of financial development but has paid less attention to the latter question on
financial development convergence.
We can highlight the process of convergence with a couple of examples. In 2011 credit to
the private sector was 135 percent of GDP in Germany and two times smaller (66 percent) in
neighboring Poland. However, in 2000 the difference was even greater: credit in Poland (34
percent) was four times smaller than in Germany (146 percent). Going further east, in 2000
credit in Belarus was almost eight times smaller than in Germany (19 percent); by 2011 it was
only four times smaller (34 percent). We investigate whether a similar process can be observed
across other countries as well.
Our analysis extends the literature on the determinants of financial development, e.g. La
Porta et al. (1997, 1998); Djankov et al. (2007); and Beck et al. (2003). Similar to that literature,
4
we are interested in the factors that explain the differences in financial development across
countries. However, we analyze whether we observe financial development convergence or
divergence across countries. Our focus is on the changes in the country differences over time.
We also contribute to several recent papers on financial system convergence. Antzoulatos et al.
(2011) find evidence for non-convergence in a sample of thirty eight industrial and developing
countries during 1990 – 2005 whereas Veysov and Stolbov (2011) report convergence. Bruno et
al. (2012) find mixed evidence for convergences looking at household financial assets in the
OECD countries. Our contribution to these papers is threefold. First, we approach the question
with the entire financial structure database (Beck and Demirgüç-Kunt, 2009) using all countries
and years for which data are available. Second, we investigate financial development
convergence at different levels of economic and financial development. Third, we explore
whether the process of convergence can be attributed to institutional changes on the country level
or to changes in the efficiency of operation of financial institutions. These estimations are
important as institutions are the primary factor for financial development (La Porta et al. 1997)
while the efficiency of financial institutions reflects know-how and innovation.
The rest of the paper is structured as follows. In section II we discuss whether or not we
should expect convergence. Then, we present the data in Section III and the empirical results in
section IV. The paper concludes with final remarks in section V.
II. Should we expect financial development convergence?
Looking at the global economy, private credit as percent of GDP had doubled in the three
decades before the financial crisis of 2008. It increased from 87.7 percent of GDP in 1975, to
113.8 percent in 1985, 145.9 percent in 1995, and reached 162 percent in 2005. The growth has
5
been faster in some countries and slower (or nonexistent) in others but, nonetheless, the general
trend has been upward and quite steep.
Available time series do not tell us how much of that growth can be attributed to larger
loans to existing borrowers or to a larger number of borrowers. A recent project by the World
Bank (Demirgüç-Kunt and Klapper, 2012) has started collecting such data but it would be some
time before time series are accumulated. Nonetheless, the initial survey from 2011 shows that
over 30 percent of households in the U.S., UK, Sweden, Netherlands, Belgium, Denmark, and
other developed countries have mortgages while the percent of households with mortgages is less
than 10 percent in most of the rest of the world. In fact, less than 5 percent of households in
many countries in Africa, Latin America, and the former communist bloc have outstanding
mortgages. Given these numbers and that much of the growth in credit has come from growth in
household credit (Beck et al., 2012; Büyükkarabacak and Valev, 2010; Greenwood and
Scharfstein, 2012), we can conjecture that an increase in the number of borrowers can explain
much of the credit growth.
The increase in the number of borrowers can be attributed to several potential factors.
First, institutions that are important to the financial system could have improved over time
allowing banks to go down-market. Many emerging markets have made efforts to adopt
institutions that promote financial development. More efficient bankruptcy procedures, better
financial transparency rules, and stronger regulation and supervision of the financial system
could help expand the supply of external financing to households and firms. As we pointed out
earlier, the literature, e.g. La Porta et al. (1997, 1998), identifies institutions as the primary
determinant of financial development.
6
Second, financial liberalization has spread financial sector know-how to emerging
markets. The branches of international banks bring expertise in risk management, credit
evaluation, and other areas that facilitate the development of the financial sector. In some
emerging markets, such as the Ukraine, more than half of the entire financial system is foreign
owned. In some developing countries, mobile technology has played a vital role in expanding
financial services to rural areas. All of those could increase the efficiency of operation of the
financial system, bringing down the costs and allowing financial institutions to reach new
customers.
Third, financial liberalization and the general move towards market-based economies
around the world have opened the playing field for capital investment in financial services.
Domestic and international banks have invested in new branches in urban and rural areas
reducing the geographic distance to their customers. According to Demirgüç-Kunt and Leora
Klapper (2012), geographic distance to financial institutions is a major deterrent to the use of
financial services. Respectively, reducing the distances can help bring more households and
firms into the formal financial system even if institutions and technological know-how remain
the same. A few numbers from the IMF’s Financial Access Survey could illustrate that point.
The number of bank branches per 100,000 adults in middle income countries increased from 8.1
in 2004 to 12.2 in 2010. At the same time, the number of bank branches in the high income
countries decreased from 31.6 per 100,000 adults to 23.1.1
1 The data set begins in 2004 and cannot be used in our analysis as the series is too short. Moreover, there are limited data for low income countries.
The number of ATM’s per 1000
adults increased by about 50 percent in the high income countries (from 68.6 to 90.1). In middle
income countries, their number tripled from 10.1 to 29.0.
7
Fourth, the general growth of income across countries has pushed more households and
firms above a threshold level of income and savings that allow them to start using financial
services. Similar to the levels of credit, global GDP per capita in constant 2000 dollars has nearly
doubled from 3,594 dollars in 1975 to 5,700 dollars in 2005. Higher and more stable incomes
along with real estate and other assets that could serve as collateral allow new customers to
increase their demand for financial services and the financial system to provide them.
If the factors discussed above have played a more important role in less developed
countries compared to developed countries, then we should observe convergence in terms of
credit availability. If they have benefited primarily the already developed countries, then we
should observe divergence. As a parallel, the economic growth literature (Pritchettt, 1997;
Durlauf, 1996) finds that incomes per capita do not exhibit convergence on a worldwide scale. In
fact, looking at the most and the least advanced economies, income levels tend to diverge over
time. There are two reasons for that. First, new technologies are not readily adopted in countries
with poor education, institutions, and infrastructure. Thus, while new ideas raise the standard of
living in rich countries, they have had less impact on incomes in poor countries. Second, as
Romer (1986) shows, investment in new capital could have decreasing returns on a firm level but
it could exhibit increasing returns on a societal level if firms benefit from other firms’ investment
in new knowledge. Thus, rich countries get richer over time, benefiting from a virtuous cycle,
while poor countries are stuck in a vicious cycles or low investment and low productivity.
The same forces could be at play in the financial system. New financial sector
technologies may not be suitable in institutionally weak countries and financial innovations may
not take hold. Also, countries with weak institutions may find it difficult to invest the necessary
resources to improve their institutions. Thus, a financially underdeveloped economy may remain
8
such while the financial sectors in developed countries continue to innovate and expand.
However, it could also be the case that convergence occurs despite institutional differences. The
financial system could be adding new firms and households as customers at a more rapid rate in
the less developed countries simply because the starting base is smaller. In that environment,
capital investment in new bank branches and introducing new assets have a bigger impact in
terms of new customer generation compared to advanced economies where a higher percent of
the private sector is already “banked”. Then, we could observe convergence while holding
constant the quality of institutions and financial innovation.
The discussion suggests that, along with other controls, our models exploring financial
development convergence should incorporate: 1) measures for the quality of country-level
institutions; 2) measures of the efficiency of financial institutions, and 3) income levels. We want
to know if there is evidence for convergence even after controlling for institutions, bank
operating efficiency, and rising incomes. If such evidence is present, we could attribute the
process of convergence to capital investment in financial services.
Whether convergence is a byproduct of the improvements in the county-level institutions
or it results from additional investments in the financial sector has important implications. At
some point, the financial institutions have succeeded in reaching most potential customers and
there are fewer opportunities to reach new ones by simply opening another bank branch in a part
of town without any. In essence, the diminishing returns to capital investment slow down the
growth. From that point on, further convergence would be possible only if country-level
institutions improve and financial institutions innovate.
9
III. Data and methodology
We explore the convergence of several measures for financial development from the
Financial Structure Database (Beck and Demirgüç-Kunt, 2009). Our primary focus is on credit to
the private sector (CREDIT) as a percent of GDP since that variable has been used most
extensively in the literature to study financial development (Beck et al., 2007; Djankov et al.,
2007). It measures the level of firm and household credit by deposit money banks and other
financial institutions as a percent of GDP. We also report estimations with BANK CREDIT
which is the amount of credit extended by deposit money banks to the private sector as percent
of GDP; and LIQUID LIABILITIES, defined as the ratio of liquid liabilities to GDP. Summary
statistics and correlations are presented in the Appendix. Since we are interested in the rate of
growth of these variables, we also report summary statistics for the average annual rates of
change during the sample period.
We transformed the annual data into five-year periods: 1965-1969, 1970-1974, .., 2005-
2009. The objective was to preserve the time variation on the country level while at the same
time smooth out the annual variations. It is a method that has been used extensively in the
literature (Beck et al., 2000). For completeness, in unreported regressions we also use the annual
data panel and a cross section database with the country averages for the entire period, and
obtained qualitatively similar results. The estimation results are available on request.
We start our estimations with a simple empirical formulation for convergence:
( 1 )
10
where is the growth of financial development in country i during a five-year period t. is
the level of financial development in the first year of that same time period.2
Equation (1) estimates unconditional convergence as it does not control for any country
characteristics. Building on that, we add a vector of control variables :
We are interested in
the coefficient estimate for : implies convergence of the level of financial development
in the sense that financial development grows more rapidly in countries/periods with a lower
initial level of financial development; implies that differences in financial development
across countries persist over time; and provides evidence for divergence, i.e. finance
expands more rapidly in countries that are already more financially developed.
( 2 )
The base control set includes the level of real GDP PER CAPITA, INFLATION,
GOVERNMENT spending, and TRADE openness. Favorable macroeconomic conditions affect
credit growth by increasing the borrowing capacity of households and decreasing the default
rates likelihood. Thus, we expect an increase in the real growth of GDP to increase the credit
growth (Levine, 2005). High inflation is associated with high nominal interest rates and may also
be viewed as a proxy for poor macroeconomic management. Also, because the real profits of
banks may become harder to predict during periods of high inflation, banks may find it more
difficult to assess credit quality at these times. Huybens and Smith (1999) theoretically and Boyd
et al. (2001) empirically study the effects of inflation on financial development and conclude that
economies with higher inflation rates are likely to have smaller banks and equity markets. An
increase in government spending is likely to crowd out private borrowing and is negatively
associated with private credit growth. Finally, following Huang and Temple (2005) we include
2 We also experimented using the average level of financial development from the preceding five year period and obtained similar results.
11
trade openness, as the ratio of exports plus imports to GDP. We expect to see a positive
correlation between trade openness and private credit growth as international trade requires
financing and the exposure to international competitions stimulates the demand for capital
investment and innovation.
All variables are entered in logs and are timed at the beginning of the five year period to
reduce endogeneity issues. Alternatively, we used the average values from the preceding five-
year period as well as the average values during the contemporaneous period. The results are
similar and available upon request. All the models are estimated with generalized least squares
allowing for autocorrelation within countries as well as heterogeneous standard errors across
countries. For robustness we also estimate the models with fixed effects and robust standard
errors.
Next, we add institution variables to equation (2):
( 3 )
We start with the rule of law measure from the International Country Risk Guide (ICRG)
as well as a composite ICRG variable that averages three of the ICRG indexes: rule of law,
corruption, and bureaucratic efficiency. The advantage of the ICRG variables is that they are
collected on an annual basis and have some time variation. The disadvantage is that the series
start in 1984 reducing our sample size substantially. Then we add countries’ legal origin as it has
been identified as an important predictor of financial development (La Porta et al., 1997).
Finally, we add the religious composition of the population: CATHOLIC and MUSLIM, leaving
PROTESTANT as the omitted category, as the literature (Stulz and Williamson, 2003) has
identified the religious composition of the population as an important and exogenous predictor of
financial development. Specifically, countries with predominantly protestant population exhibit
12
higher levels of financial development. We are interested in whether there is evidence for
convergence after we control for institutions.
Then, we add several measures for the efficiency of financial institutions to the baseline
equation (2):
( 4 )
Specifically, we introduce the NET INTEREST MARGIN which is defined as the accounting
value of bank's net interest revenue as a share of its interest-bearing (total earning) assets; COST
is the total costs as a share of total income of all commercial banks; and RETRUN ON ASSETS
is the average return on assets (Net Income/Total Assets). Similar to equation (3), we are
interested in whether the evidence for convergence, i.e. the coefficient estimate for , is affected
by the inclusion of these variables.3
Next, we estimate equation (2) for different groups of countries based on their levels of
economic and financial development. And, finally, we follow Aghion et al. (2005) and estimate
an alternative version of equation (2) where we explain the growth of financial development
relative to the U.S. where the U.S. is taken as the frontier of development:
(3)
Here, we ask whether financial development grows faster in countries that are further away from
the U.S. in terms of financial development.
3 Unfortunately, we could not incorporate the share of foreign owned banks or other market structure variables in the analysis as their time and country coverage is limited.
13
IV. Results
Table 1 shows the results of estimating equation (1) using the various measures for
financial development. Recall that in those estimations, we do not include any control variables
but we allow for country-specific effects, heterogeneity across countries, and serial correlation
within countries over time. Looking across the three columns of results we see evidence for
convergence for all financial variables with a level of statistical significance falling below 1%
percent. Table 2 adds a range of macroeconomic control variables to the model. We still observe
statistically significant evidence for financial development convergence across the different
measures. Of the control variables, initial income per capita is positive and statistically
significant in all of the models whereas government spending has a negative and statistically
significant effect in two of the three regressions. Trade openness has a positive and statically
significant on the broader financial development measures – financial system credit and liquid
liabilities. Higher inflation reduces the growth of bank credit and liquid liabilities.
Next, in Table 3 we estimate equation (3) and add various measures for institutions to the
model specification. We focus on financial system credit as that variable is the most widely used
financial development measure in the literature. First, we introduce the ICRG variables, then we
add legal origin, and finally the variables for religious composition. Notice that the initial level
of financial development remains negative and statistically significant across all estimations. The
second column of results suggests that stronger rule of law is associated with more rapid growth
in financial system credit. The third column provides evidence that legal origin also matters. The
British, French, and German legal origins are associated with more rapid credit growth compared
to the omitted Scandinavian legal origin.4
4 There are no countries with Socialist legal origin in the sample.
14
Note that the rank order of the three legal origin variables in terms of coefficient size
suggests that the German legal system is associated with the most rapid credit growth, followed
by the French legal system, and then the British legal system. This runs counter to the results
reported in the earlier literature where the British legal system is associated with the highest level
of financial development. However, here we study the change of financial development. Thus,
the reported coefficient estimates on the legal origin variables provide additional evidence for
convergence: credit grows more rapidly in countries whose legal origin is, in principle,
associated with a lower level of financial development. Interestingly, these results also suggest
that institutions, while a major factor for financial development as identified in the literature do
not restrict completely the convergence of credit levels across countries. In other words, credit
levels converge over time despite differences in institutions across countries. Unlike legal origin,
there is no evidence that religious composition affects the growth of credit. Still, the evidence for
convergence remains statistically significant after controlling for religious composition.
The sizes of the estimated coefficients on the initial credit level, i.e. the estimate of in
Table 3 range between -0.1 and -0.06. If we take the -0.1 estimate, an increase in initial credit of
10 percentage points, e.g. an increase in the credit to GDP ratio from 50 percent to 60 percent, is
associated with a decline in the growth of credit by about 1, i.e. by one percentage point. That is
an economically sizable effect given that the average annual growth rate of credit in the sample
is 4.93 percent. For example, controlling for other factors, as credit in Turkey increased from 14
to 38 percent between 1990-2010, the rate of growth of credit slowed down by 2.4 percentage
points.
In Table 4 we add the four variables that measure the efficiency of financial institutions.
The return on assets is the only variable with statistical significance and has a positive effect on
15
credit growth. More importantly, looking across all the estimations, the evidence for credit
convergence remains similar after the inclusion of the efficiency variables. It seems that
convergence is not a byproduct of an increased efficiency of operation of the financial
institutions.
Next, Table 5 shows estimates from estimating split samples according to the level of
economic development of a country. We use the World Bank categorization to identify four
income groups: low, lower middle, upper middle, and high income. The coefficient estimate on
initial credit is negative and statistically significant in all four samples but its size differs. The
most economically important evidence for convergence is observed in the low income countries
and the slowest convergence is observed in the high income group of countries. The evidence for
the middle income countries is mixed, with a stronger effects for upper middle income countries,
but overall they fall in-between the low and the high income groups in terms of the size of the
coefficient on initial credit.
Table 6 repeats that exercise with groups of countries split according to their average
level of financial development during the sample period. Following Rioja and Valev (2004), we
run estimations for three groups: low, middle, and high level of financial development. The
results suggest a similar pattern to what we observe in Table 5. Convergence is most rapid in
countries with low financial development, followed by the middle group. Convergence is
markedly slower among countries with a high level of financial development.
Next, we report two sets of robustness tests. In Table 7 we estimate the benchmark results
with the ICRG variables, fixed country-specific effects and robust standard errors. As before,
initial credit is negative and statistically significant and the control variables have similar effects.
Then, in Table 8 we follow Aghion et al. (2005) and estimate the benchmark models but taking
16
the U.S. as a frontier county. Here we explore the convergence to the level of financial system
credit in the U.S. The results confirm our earlier findings of a negative and statistical significant
effect of initial credit levels on the subsequent growth in credit.
V. Conclusion
We present evidence that the financial system credit to GDP ratio of different countries,
and other measures of financial development, converge over time. Countries with a lower initial
level of financial development experience more rapid growth of financial development. We
observe convergence for different measures of financial development, with various control sets,
and estimation techniques. Most importantly, convergence is present even after we control for
the quality of institutions. Institutions have been identified as the most important fundamental
determinant of financial development. We find, however, that the levels of financial
development across countries become more similar over time despite the persistent differences in
institutions.
Our analysis has centered on credit variables, leaving aside the question whether stock
markets also exhibit convergence. There are reasons to expect different dynamics. The threshold
level of institutions and economic development that allow stock markets to develop may be
higher for stock markets than for bank credit. Also, while lending to firms and households is
primarily a local activity, stock market activity is concentrated in several large financial centers.
Therefore, our conjecture is that we would not observe similar convergence for stock markets but
we leave that question for future research.
Similarly, we leave to future research to explore the convergence of other aspects of
financial development. While widely used in the literature and available for many countries and
17
years, the credit level is not a perfect measure of financial development. We want to know
whether new credit reaches more firms and households or it services mostly existing customers;
whether it goes mostly to households or firms; what are its uses; etc. Hopefully, future data sets
would allow us to explore the dynamics of these and other aspects of financial development.
18
References
Aghion, Philippe, Peter Howitt, and David Mayer-Foulkes. "The effect of financial development on convergence: Theory and evidence". The Quarterly Journal of Economics, 120 (1) (2005): 173-222. Antzoulatos, Angelos A., Ekaterini Panopoulou, and Chris Tsoumas. "Do Financial Systems Converge?." Review of International Economics 19.1 (2011): 122-136. Beck, Thorsten, Asli Demirgüç-Kunt, and Ross Levine. "Law, endowments, and finance." Journal of Financial Economics 70.2 (2003): 137-181. Beck, Thorsten, Asli Demirguc-Kunt, and Ross Levine. “Finance, inequality, and the poor." Journal of Economic Growth 12 (2007), 27-49.
Beck, T., Buyukkarabacak, B., Rioja, F., & Valev, N. (2012). Who gets the credit? And does it matter? Household vs. firm lending across countries. BE Journal of Macroeconomics, 12(1). Beck, Thorsten and Asli Demirgüç-Kunt. "Financial institutions and markets across countries and over time-data and analysis." World Bank Policy Research Working Paper Series, Vol (2009). Boyd, John H., Ross Levine, and Bruce D. Smith. "The impact of inflation on financial sector performance." Journal of Monetary Economics 47.2 (2001): 221-248. Bruno, Giuseppe, Riccardo De Bonis, and Andrea Silvestrini. "Do financial systems converge? New evidence from financial assets in OECD countries." Journal of Comparative Economics (2011). Büyükkarabacak, Berrak, and Neven T. Valev. "The role of household and business credit in banking crises." Journal of Banking & Finance 34.6 (2010): 1247-1256. Demirgüç-Kunt, Asli, and Leora Klapper. "Measuring Financial Inclusion: The Global Findex Database." World Bank Policy Research Working Paper 6025 (2012). Djankov, Simeon, Caralee McLiesh, and Andrei Shleifer. "Private credit in 129 countries." Journal of Financial Economics 84.2 (2007): 299-329. Durlauf, Steven N. "On the convergence and divergence of growth rates." The Economic Journal 106.437 (1996): 1016-1018. Goldberg, Linda S. "Financial sector FDI and host countries: New and old lessons" FRBNY Economic Policy Review, March, (2007).
19
Greenwood, Robin, and David Scharfstein. "The Growth of Modern Finance." Available at SSRN 2162179 (2012). Huang, Yongfu, and Jonathan Temple. "Does external trade promote financial development?" CEPR discussion paper no. 5150 (2005) Huybens, Elisabeth, and Bruce D. Smith. "Inflation, financial markets and long-run real activity." Journal of Monetary Economics 43.2 (1999): 283-315. Levine, Ross. "Finance and growth: Theory and evidence." Handbook of Economic Growth 1 (2005): 865-934. La Porta, Rafeal, Lopez de Silanes, Florencio, Shleifer, Andrei, and Vishny, Robert. (1997) “Legal determinants of external finance.” Journal of Finance, 52, 1131-1150. La Porta, Rafeal., López de Silanes, Florencio., Shleifer, Andrei., & Vishny, Robert. (1998) “Law and finance.” Journal of Political Economy, 106, 1113-1155. Pritchett, Lant (1997) “Divergence, big time." The Journal of Economic Perspectives 11.3 (1997): 3-17. Rioja, Felix, and Neven Valev. "Does one size fit all?: a reexamination of the finance and growth relationship." Journal of Development Economics 74.2 (2004): 429-447. Romer, Paul M. "Increasing returns and long-run growth." The Journal of Political Economy (1986): 1002-1037. Veysov, Alexander, and Mikhail Stolbov. "Do financial systems converge? A Comprehensive panel data approach and new evidence from a dataset for 102 countries." (2011).
Stulz, Rene M., and Rohan Williamson. "Culture, openness, and finance." Journal of Financial Economics 70.3 (2003): 313-349.
20
Table 1: Unconditional Convergence
(1) (2) (3) VARIABLES Fin System
Credit Bank Credit
Liquid Liabilities
CREDIT -0.029*** (0.00) BANK CREDIT -0.024*** (0.00) LIQUID LIABILITIES -0.021*** (0.01) Constant 4.05*** 4.14*** 3.35*** (0.21) (0.19) (0.27) Chi Square 64.87 59.02 24.91 Observations 842 840 809 Number of countries 143 143 140 Dependent variables are the average annual growth rate of financial system credit, bank credit and liquid liabilities. P-values calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with generalized least squares allowing for autocorrelation within countries as well as heterogeneous standard errors across countries.
21
Table 2: Baseline Regressions for all FD variables
(1) (2) (3) VARIABLES Fin System
Credit Bank Credit
Liquid Liabilities
CREDIT -0.05*** (0.01) BANK CREDIT -0.05*** (0.01) LIQUID LIABILITIES -0.05*** (0.01) GDP PER CAPITA 0.61*** 0.55*** 0.27** (0.16) (0.15) (0.13) INFLATION -0.13 -0.36** -0.28** (0.17) (0.17) (0.13) GOVERNMENT -1.25*** -0.58 -1.72*** (0.49) (0.46) (0.50) TRADE 0.60** 0.34 1.13*** (0.28) (0.27) (0.26) Constant 1.50 1.61 2.93* (1.66) (1.58) (1.54) Chi Square 48.51 45.12 63.94 Observations 802 800 770 Number of Countries 141 141 138 Dependent variables are the average annual growth rate of financial system credit, Bank credit and liquid liabilities. P-values calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with generalized least squares allowing for autocorrelation within countries as well as heterogeneous standard errors across countries.
22
Table 3: Private Credit with Institutions and Legal Origin
VARIABLES (1) (2) (3) (4) CREDIT -0.10*** -0.07*** -0.08*** -0.06*** (0.01) (0.01) (0.01) (0.01) GDP PER CAPITA 1.70*** 0.81*** 1.19*** 0.74*** (0.27) (0.25) (0.20) (0.20) INFLATION -0.23 -0.23 -0.05 -0.13 (0.16) (0.22) (0.17) (0.17) GOVERNMENT -3.72*** -2.47*** -1.26** -1.79*** (0.62) (0.65) (0.56) (0.49) TRADE 2.42*** 1.70*** 0.95*** 0.66** (0.51) (0.43) (0.34) (0.32) ICRG 0.26 (0.38) RULE of LAW 0.77*** (0.24) ORIGIN_UK 1.93* (0.99) ORIGIN_FR 2.44** (0.99) ORIGIN_GE 5.49*** (1.07) CATHOLIC -0.00 (0.01) MUSLIM -0.00 (0.01) Constant -6.07*** -3.00 -5.83** 2.11 (2.27) (2.26) (2.61) (1.96) Chi Square 133.15 64.55 99.12 60.81 Observations 511 511 737 800 Number of Countries 110 110 128 140 Dependent variable is the average annual growth rate of financial system credit. P-values calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with generalized least squares allowing for autocorrelation within countries as well as heterogeneous standard errors across countries.
23
Table 4: Testing the diminishing returns hypothesis – main effects
VARIABLES (1) (2) (3) (4) CREDIT -0.08*** -0.09*** -0.08*** -0.08*** (0.01) (0.01) (0.01) (0.01) GDP PER CAPITA 0.73*** 0.58** 0.62*** 0.73*** (0.21) (0.24) (0.15) (0.16) INFLATION -0.43 -0.40 -0.35 -0.33 (0.30) (0.38) (0.27) (0.29) GOVERNMENT -0.50 0.27 -0.41 -0.34 (0.50) (0.71) (0.38) (0.44) TRADE 1.05* 1.36** 1.05** 1.31*** (0.56) (0.56) (0.41) (0.38) OVERHEAD 9.79 (15.55) INTEREST MARGIN -12.84 (13.96) COST -0.72 (1.42) RETURN ON ASSETS 45.02*** (12.72) Constant -0.18 -0.95 0.83 -2.22 (2.13) (2.47) (1.75) (1.89) Chi Squared 51.93 71.96 55.83 119.08 Observations 310 301 314 315 Number of Countries 107 104 109 109 Dependent variable is the average annual growth rate financial system credit. P-values calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with generalized least squares allowing for autocorrelation within countries as well as heterogeneous standard errors across countries.
24
Table 5: Private Credit Convergence for Different Income Levels
(1) (2) (3) (4) VARIABLES Low Lower Middle Upper Middle High CREDIT -0.39*** -0.06*** -0.18*** -0.04*** (0.10) (0.02) (0.02) (0.01) GDP PER CAPITA -0.13 -0.59 3.08*** 0.34 (2.14) (0.69) (0.83) (0.49) INFLATION -1.57** 0.39 -0.51 -0.05 (0.62) (0.37) (0.32) (0.25) GOVERNMENT 1.64 -1.31 -4.68*** -0.24 (1.82) (1.05) (1.46) (0.74) TRADE -1.81 2.34*** 3.22*** 0.88*** (1.90) (0.89) (0.88) (0.31) Constant 15.09 1.21 -15.04* -1.19 (11.38) (4.35) (8.90) (5.00) Chi Square 27.42 20.40 75.15 25.99 Observations 122 216 189 275 Number of Countries 27 35 34 45
Dependent variable is the average annual growth rate of financial system credit. P-values calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with generalized least squares allowing for autocorrelation within countries as well as heterogeneous standard errors across countries.
25
Table 6: Private Credit Convergence for Different Financial Development Levels
(1) (2) (3) (4) (5) (6) VARIABLES Low Middle High Low Middle High CREDIT -0.86*** -0.23*** -0.02** -1.40*** -0.73*** -0.13*** (0.11) (0.04) (0.01) (0.25) (0.13) (0.03) CREDIT*CREDIT 0.04*** 0.01*** 0.00*** (0.01) (0.00) (0.00) GDP PER CAPITA 5.52*** 2.00*** 0.26 1.01*** 1.93*** 0.11 (1.74) (0.49) (0.30) (0.37) (0.48) (0.27) INFLATION -0.09 -0.41 -0.17 -1.32*** -0.31 -0.32 (0.46) (0.33) (0.25) (0.45) (0.33) (0.23) GOVERNMENT 1.65 -0.08 0.63 -1.25 0.40 1.28* (2.03) (1.17) (0.81) (1.08) (1.17) (0.76) TRADE 2.30 1.60* 0.21 -0.12 1.62* 0.19 (1.92) (0.87) (0.34) (0.94) (0.88) (0.30) Constant -33.35*** -10.80*** -1.34 13.43*** -4.63 2.18 (12.57) (3.81) (4.33) (3.32) (4.27) (3.85) Chi Squared 72.76 46.88 6.49 72.76 46.88 6.49 Observations 216 295 291 216 295 291 Number of Countries 47 48 46 47 48 46
Dependent variable is the average annual growth rate of financial system credit. P-values calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with generalized least squares allowing for autocorrelation within countries as well as heterogeneous standard errors across countries.
26
Table 7: Robustness Test: Fixed Effects and Robust Standard Errors
VARIABLES (1) (2) (3) CREDIT -0.17*** -0.18*** -0.19*** (0.04) (0.04) (0.04) GDP PER CAPITA 8.61*** 12.78*** 12.64*** (2.17) (3.63) (3.67) INFLATION -0.31 0.12 0.10 (0.61) (0.76) (0.78) GOVERNMNENT -4.32** -7.18*** -6.69*** (1.98) (2.48) (2.47) TRADE 2.48 3.63 3.45 (1.87) (2.69) (2.72) ICRG 0.17 (1.12) RULE of LAW 1.26* (0.65) Constant -54.22*** -85.78*** -89.27*** (14.77) (27.55) (27.76) Chi Squared 6.23 5.87 6.66 Observations 808 516 516 R-squared 0.10 0.13 0.14 Number of Countries 147 115 115
Dependent variable is the average annual growth rate of financial system credit. P-values are calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with fixed effects and robust standard error.
27
Table 8: Convergence relative to the Frontier (USA)
VARIABLES (1) (2) (3) (4) (5) CREDIT -0.06*** -0.06*** -0.06*** -0.03*** -0.05*** (0.01) (0.01) (0.01) (0.00) (0.01) GDP PER CAPITA 1.22*** 1.07*** 0.98*** 0.61*** 0.92*** (0.24) (0.29) (0.28) (0.17) (0.21) INFLATION 0.54*** 0.74*** 0.75*** 0.36** 0.47** (0.21) (0.23) (0.24) (0.18) (0.19) GOVERNMENT -1.09 0.23 0.04 0.41 -0.52 (0.81) (0.79) (0.80) (0.57) (0.68) TRADE -0.56 0.78 0.69 -0.40 -0.51 (0.44) (0.50) (0.50) (0.26) (0.33) ICRG 0.44 (0.40) RULE of LAW 0.51* (0.28) ORIGIN_UK 0.65 (0.65) ORIGIN_FR 0.50 (0.69) ORIGIN_GE 1.92** (0.81) CATHOLIC -0.00 (0.01) MUSLIM -0.01 (0.01) Constant -8.91*** -19.62*** -18.50*** -7.27*** -7.10*** (3.12) (3.25) (3.29) (2.43) (2.65) Chi Squared 83.45 97.07 91.29 60.76 78.73 Observations 804 512 512 739 802 Number of Countries 141 110 110 128 140
Dependent variable is the average annual growth rate of financial system credit. P-values calculated from robust standard errors are reported. *,**,*** indicate significance at the 10%, 5% and 1% level, respectively. All regressions are estimated with generalized least squares allowing for autocorrelation within countries as well as heterogeneous standard errors across countries.
28
Summary Statistic Bank Credit to
GDP Fin. System Credit to
GDP Liquid Liabilities to
GDP Credit
Growth Bank Credit
Growth LL
Growth GDP per
Inflation Government Trade Openness
ICRG RULE of LAW
Mean 37.42 40.73 57.33 4.9326 5.0328 2.9650 6541. 51.5714 15.8004 73.6444 2.9149 3.6616
Standard Deviation 35.75 38.28 35.29 14.3181 14.5072 6.6262 10431 703.1084 6.3926 46.6972 1.1595 1.4665
Minimum 0.83 0.84 1.21 -17.6568 -17.6568 -19.9553 58.4599 -9.0736 0.0000 0.0000 0.0000 0.0000
Maximum 261.45 265.90 115.2917 292.2205 292.2205 73.8810 105147 23773 69.5428 421.5666 5.3333 6.0000
Correlation
Bank Credit 1
Fin. System Credit 0.9033 1
Value Traded 0.5601 0.6208
Turnover 0.4095 0.4608
Liquid Liabilities 0.7447 0.6678 1
Credit Gr. -0.0003 -0.0165 -0.1497 1
Bank Credit Gr. 0.0089 -0.0297 -0.1511 0.9604 1
Value Traded Gr. -0.0851 -0.0894 -0.0517 -0.0271 -0.0228
Turnover Gr. -0.044 -0.0255 -0.0467 -0.0605 -0.06
Liq. Liab. Gr. 0.0267 -0.0012 -0.0117 0.5572 0.5394 1
GDP pc 0.657 0.7072 0.5898 -0.1037 -0.0923 -0.0971 1
Inflation -0.3465 -0.3515 -0.2867 -0.0226 -0.0337 0.004 -0.3054 1
Gov. Spending 0.3443 0.3501 0.1738 0.0352 0.0501 -0.0136 0.4592 -0.2508 1
Trade Open. 0.2711 0.2184 0.4147 0.0044 0.0038 0.0011 0.153 -0.1461 0.0043 1
ICRG 0.6019 0.6217 0.45 -0.1046 -0.1013 -0.0968 0.7884 -0.259 0.5387 0.1502 1
Rule of RULE of LAW 0.5546 0.5438 0.4304 0.0237 0.0369 -0.0038 0.6887 -0.255 0.5098 0.1925 0.8826 1