Master thesis
Abstract
The aim of this paper is to test whether there is a relationship between liberalization and the occurrence of banking crises. A dataset including developed as well as developing countries during 1981-2005 is used to analyze this. We made a distinction between OECD and non OECD countries and between different time periods (5 year periods). We find evidence that liberalization leads to an increase in banking crises. Liberalization leads to less GDP growth volatility at least for de facto, the two de jure measures show ambiguous results. During liberalization the quality of institutions does matter for GDP growth.
Key words: Financial liberalization; banking crisis; GDP growth
Supervisor Robert Inklaar
University of Groningen
Faculty of Economics and Business
Study International Economics and Business Name Jacco van Huizen (1881280)
Mail jacco.van.huizen@student.rug.nl Date August, 2011
1. Introduction
A lot of countries, developed as well as developing, liberalized their markets in the beginning of the 80’s. We define financial reform as measures aiming at the removal of non-competitive market forces in the financial sector, thereby increasing its level of liberalization (Shehzad and de Haan, 2009). Countries wanted to use liberalization to improve their prospects for economic growth. Foreign savings became available to entrepreneurs; these entrepreneurs would then invest in businesses, homes and infrastructure. Since the liberalization of the markets (1980’s) the events of banking crises increased rapidly. There are economists who argue that this is due to the liberalization (Tornell, Westermann, and Martinez, 2004). This paper discusses the role of financial liberalization on financial markets. In particular on systemic banking crisis (systemic banking crises are crises in which much or all bank capital has been exhausted) and GDP growth.
Financial liberalization can be divided in two parts de jure and de facto. The former liberalization concerns rules and regulation (authorities remove rules and laws to stimulate capital flows to the country); whereas the latter concerns the real capital flows to the country (Neumann and Penl, 2008). In other words, de jure means “concerning law” and de facto means “concerning fact”.
Banking crises can be divided into systemic and non-systemic crisis. The former are crises in which much or all bank capital has been exhausted, the latter being a crises limited to a small number of banks. When a country liberalizes its market the competition intensifies therefore a non-systemic crises is likely to occur. Hellman, Murdock and Stiglitz (2000) argue that increased competition erodes a bank’s franchise value, reducing its incentive to avoid risk. This paper focuses on systemic crises as this is a macroeconomic study and non-systemic crises are hard to capture with macroeconomic data. Shehzad and de Haan (2009) studied the effects of liberalization on banking crises they found evidence for non-systemic crises, but not for systemic crises.
becomes more volatile or the financial market becomes instable as it relies on foreign credit. Another important point seems to be the quality of the institutions when these are at a decent level the country benefits more.
Figure 1 summarizes the periods of capital mobility and the occurrence of banking crises for the period from 1800 to 2008. This figure comes from a study by Reinhart and Rogoff (2008) it shows the correlation between capital mobility and international banking crises. There is a high correlation between capital mobility and the occurrence of banking crises; this is not only the case since the 80’s but also historically. The figure plots a three-year moving average of the share of all countries experiencing banking crises on the right scale. On the left scale, the index (Obstfeld and Taylor, 2004) of capital mobility is graphed.
Figure 1
When we look at macroeconomic variables like for example GDP growth we conclude that the long run effect is positive. Comparing GDP growth of liberalized economies with closed economies liberalization seems beneficial. For example Thailand and India where Thailand is an open economy and India a closed one. Thailand’s GDP grew by 148% during 1980 - 2001 while India only grew by 99%. However, Thailand did experience several boom and bust periods whereas India faced a slow but safe growth.
more risk (risky loans can lead to defaults). Therefore credit availability plays an important role in the run up to a banking crisis. This research uses a large database (1981-2005) which includes developed as well as developing countries. Third this paper shows what happens to the volatility of GDP growth after liberalization. Finally, this article includes countries which did not face a banking crisis.
2. Literature review
Financial liberalization refers to the deregulation of domestic financial markets and the liberalization of the capital account. As discussed above there is an ongoing debate of the effects of liberalization of the financial market on the economy. Some (Levine, 2001; World Bank; International Monetary Fund) argue that it strengthens financial development and that it contributes to a higher long run growth. Other researchers (Rancière, Tornell and Westermann, 2006; Mehrez and Kaufmann, 2000; Minsky, 2008) argue that liberalization encourages excessive risk-taking, increases volatility which can lead to crises. In this literature review we theoretically explore some of the potential negative and positive effects of financial liberalization on economies. To that extend we develop hypotheses based on the impact of financial liberalization on the occurrence of a banking crises, GDP growth volatility and the quality of institutions.
Rodrik and Subramanian (2009) analyze the relationship between banking crises and financial globalization using data over the period 1970-2004 for 102 developing countries. Their findings suggest that financial globalization has not generated increased investment or higher growth in developing countries. Rodrik and Subramanian make a distinction between investment and saving constraint economies. In a savings constraint economy the domestic interest rate will decrease and there will be an increase in capital inflows, as firms will travel down their investment demand schedule. In an investment constraint economy the effect of liberalization is to boost consumption. The investment demand schedule is vertical therefore foreign savings substitute for domestic savings. Allowing capital inflows to an investment constraint economy (vertical investment schedule), does not affect investment because the equilibrium is primarily determined. Furthermore they argue that the exchange rate will appreciate which is good for importers, but bad news for exporters. They conclude that countries that have grown rapidly are countries that rely less on capital inflows.
Kaminsky and Reinhart (1999) focus on the link between currency and banking crises for 20 countries during 1970 to 1995. Their main finding is that during the 70’s there is no apparent link between the two types of crises, mostly due to the highly regulated markets. This changed in the 80’s and 90’s; in about half of the cases the banking crisis gets underway before the balance of payment crisis. They find evidence that financial liberalization has a significant role in the probability of a banking crisis. This may be due the fact that liberalization comes without an adequate regulatory and supervisory framework (Kaminsky and Reinhart, 1996). Furthermore they argue that the surge in credit that finances the import boom is prevalent in the periods prior to banking crises. The last finding is that most crises were in the early 80’s, when real interest rates in the United States were at their highest point since the 30’s (Calve, Leiderman, and Reinhart, 1993).
In contrast of what Rodrik and Subramanian argue, the IMF finds positive results of liberalization. In a speech during the IMF’s Annual Meetings in 1997, Fischer advocated an amendment to the IMF’s articles the purpose of which “would be to enable the Fund to promote the orderly liberalization of capital movements” (Fischer, 1997). He is aware of the risks involved in opening up the borders, but he is certain that the positive effects outweigh the negative ones.
According to the literature liberalization has an effect on banking crises. The first hypothesis is constructed to investigate the effect.
Hypothesis 1; Financial liberalization leads to an increase in banking crises.
capital inflows. This is in line with the theory of Minsky (2008) which argues that people get too greedy and think that the sky is the limit. When income and asset prices decline they cannot pay their debts which will lead to an increase in nonperforming loans and defaults. The expansion of credit is argued to be the cause of increased banking instability. The link between banking crises and liberalization is via credit growth; a country opens their border to foreign capital which will lead to an increase in credit supply. Due to this credit supply countries get more instable.
There are researcher (Büyükkarabacaka and Valev, 2010) which argue that liberalization leads to a higher credit growth which leads to more vulnerability and therefore to more crises periods. Therefore we construct a second hypothesis which argues that credit growth is the mediating instrument between liberalization and banking crises.
Hypothesis 2; Financial liberalization leads to a higher credit growth and therefore to more banking crises.
After liberalization the growth rate can become more volatile as explained previously by the example of Thailand and India. The results are ambiguous in one view liberalization strengthens financial development and contributes to higher GDP growth (Levine, 2001; Bekaert, Harvey and Lundblad, 2005; Bonfiglioli and Mendicino, 2006); some argue that liberalization does not affect growth (Grilli and Milesi-Ferretti, 1995; Kraay, 1998) and others argue that liberalization has a negative effect (Rodrik, 1998; Eichengreen and Leblang, 2003).
Figure 2
Thailand vs. India: Credit and Growth (1980-2002)
Levine (2001) does a literature study and discusses some conceptual issues and empirical findings associated with the linkages of liberalization and economic growth. In his view the answer to the question does liberalization increase growth is definitely yes. He argues that removing restrictions on international portfolio flows tends to enhance stock market liquidity and this accelerates economic growth. Allowing foreign banks into the country improves the efficiency of domestic banks through spillovers. And third, foreign banks have a higher productivity and replace (domestic) banks with a lower productivity.
conflict of interest; a finance minister his primary goal is to keep (foreign) investors happy and will pay less attention to development goals.
Hypothesis 3; Countries that liberalize their financial markets face a more volatile GDP growth.
Liberalization seems to have an indirect effect on GDP growth via the quality of institutions; researchers argue that good institutions promote economic activity, inventiveness, growth and development. Bad institutions form a barrier. This would mean that there is a positive indirect effect of liberalization via institutional quality. Prior studies (Shleifer and Vishny, 1993, Ehrlich and Lui, 1999) find support that corruption is harmful for economic growth. Although Bardhan (1997) agrees with the key arguments of Shleifer and Vishny, he argues that if pervasive and cumbersome regulations exist in a country, exactly the opposite happens corruption improves efficiency and increases growth. Using the International Country Risk Guide (ICRG) data, a number of studies (Knack and Keefer, 1995; Mauro, 1995; Dollar and Kraay, 2000) find evidence that some measure of the rule of law, property rights or corruption is significantly correlated with growth of per capita real income.
Barro (1999) does a panel study for around 100 countries from 1960-1990 and finds that the growth rate is improved by higher initial schooling and life expectancy, lower government consumption, better maintenance of rule of law, lower inflation and improvement of terms of trade. He finds different effects for political freedom, political rights and democracy. Political freedom has a little influence on growth, for political rights he finds that expanding at low levels of these rights stimulates growth and when a country reaches a moderate amount of democracy, a further expansion reduces growth. This suggests that there is a non-linear relationship between institutional quality and growth.
through indirect effects. They argue that the cost of crises is lower in countries with good institutions, as well as in the open (liberalized) ones.
Prasad and Rajan (2008) argue that the country’s institutions should first meet a certain (quality) threshold before opening their borders. If this threshold is not met the country should not open their borders, because they will not gain from it. The liberalization process can lead to more vulnerability. According to Prasad and Rajan the benefits of liberalization are better corporate governance, institution building and increasing financial sector development. Negative effects are real exchange rate overvaluation and loss of competitiveness. The government plays an important role during banking crises as it can provide liquidity to deposit-losing institutions to help stabilize them again as recently Freddie Mac, Fannie Mae, ING, AIB and many others (Demirgüç-Kunt and Servén, 2009).
Hypothesis 4; Countries with higher institutional quality have a higher GDP growth after financial liberalization.
3. Data
The data which is used in this research stems from the availability of the variables. The dataset consists out of the years 1981 – 2005 and covers the countries which are available in the dataset of Laeven and Valencia (2008, in total 102 countries, appendix A). The dataset includes developed as well as developing countries. Table 1 gives an overview of all variables.
The first hypothesis uses (systemic) banking crisis as a dependent variable. The independent variable is financial liberalization (de jure and de facto). Hypothesis 2 uses credit growth as a mediation variable to test whether liberalization leads to an increase in credit growth which might lead to an increase in banking crises. Hypothesis 3 uses the volatility of GDP growth as dependent variable and the fourth hypothesis uses GDP growth as dependent variable. The latter hypothesis uses an interaction term between institutional quality and financial liberalization to test the impact of institutional quality on growth. This paper uses several control variables which will be discussed below.
3.1 Dependent variables
The banking crises variable used in this article comes from the website of Laeven and Valencia. They provide a dataset with the most recent banking crises (1975-2008), this dataset is freely available. We will create a dummy variable for which 1 is a banking crisis and 0 when there is no banking crisis. Our dataset also includes countries which did not face a banking crisis.
3.2 Mediation variable
During financial liberalization countries get access to foreign credit more easily. Therefore available credit increases and probably the risk incentives as well, which will lead to a higher credit supply and demand. Banking crises occur due to the excessive risk taking of individuals and to high credit supply. Therefore credit growth is an important variable which can influence the amount of banking crises. During the recent crisis credit grew rapidly, a lot of people and companies were not able to repay their debt. Due to the interconnectedness of the financial markets a lot of countries were affected.
Credit growth will be measured by the indicator net domestic credit to private sector (% of GDP) annual growth rate. The data is available in the World Development Indicators (WDI) of the World Bank.
3.3 Independent variables
The data for this variable comes from an index of Abiad, Detragiache and Tressel (2010) from the IMF. They use seven variables which describe financial liberalization. Also Aizenman, Chinn and Ito (2008) created an index of financial liberalization. These datasets concerns de jure liberalization this research also needs to take into account de facto. To capture de facto effects we will use capital flows (FDI assets and FDI liabilities), from updated and extended version of the External Wealth of Nations Mark II database developed by Lane and Milesi-Ferretti (2007).
Financial liberalization is measured by two indexes. These two indexes are freely available via the websites of the authors (Abiad et al., 2010; Aizenman et al., 2008) both indexes allow for reversals in policies. The first index of de jure includes 7 variables to measure financial liberalization. For each of the variables a figure between the 0-3 is given (see Abiad et al. 2010 for a full explanation of the scores) fully liberalized = 3; partially liberalized = 2; partially repressed = 1; fully repressed = 0. The index measures liberalization during 1973-2005 and is scaled between 0 and 1. The variables are
- Credit controls and excessively high reserve requirements. - Interest rate controls.
- Entry barriers.
- Capital account restrictions.
- Prudential regulations and supervision of the banking sector. - Securities market policy.
The second index provided by Aizenman et al. (2008), measures financial openness/integration by using an index of capital account openness. This measurement is based on information regarding restrictions in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). More specific this is the first standardized principal component of the variables that indicate the
- Presence of multiple exchange rates.
- Restrictions on current account transactions. - Restrictions on capital account restrictions. - Requirement of the surrender of export proceeds.
This index is available for the period 1975-2008 for a large sample of countries. We will use both indexes of de jure liberalization, it is important to check if they are correlated with each other (if they do they measure the same and we can exclude one of them). We use both indexes to check for differences and for robustness purposes. Appendix G shows the different measures of de jure, de facto and the events of banking crises graphically. This figure shows that the two de jure measures are different.
The Chinn et al. (2008) index includes more countries and years. Chinn et al. (2008) only measure a country’s degree of capital account openness, one aspect of six policy dimensions on which the creation of the Abiad et al. (2010) index is based.
We will use capital flows as indicator for de facto liberalization. The indicator includes FDI assets, portfolio equity assets, FDI liabilities and debt liabilities as a percentage of GDP. Following previous studies like Neumann and Penl, 2008; Krugman, 1993 and Lane and Milesi-Ferretti, 2007.
3.4 Control variables Hypotheses 1 and 2;
inflation, the real interest rate, credit to the private sector, money and quasi money growth, institutional quality, exchange rate and GDP per capita. Inflation is used to measure bad macroeconomic policies. Credit to the private sector is used as a measure of external finance available to firms. Institutional quality is used to control for the differences between rich and poor countries. Institutional quality will be measured by an index created by the ICRG. They use the mean values of the three variables mentioned above and scaled these between 0 and 1. They give for each measure a number between the 0-3, where higher values indicate a higher quality of government, and scale them between 0 and 1. The variable corruption assesses the corruption within the political system. Law and order assess the strength and impartiality of the legal system. And bureaucracy quality is assessed by the strength and expertise to govern without drastic changes in policy or interruptions in government services. And finally, we include the level of real GDP per capita (in US$) to control for different stages of economic development.
Hypothesis 3;
For the third hypothesis we will use control variables following previous studies by Bekeart et al. (2006) and Ranciere et al. (2006). The control variables used in this hypothesis are government size, secondary school enrollment (human capital), population growth, trade openness, inflation, institutional quality, exchange rate, and credit to the private sector. Trade openness is included because a higher value of trade may affect the efficiency of an economy.
Hypothesis 4;
Table 1 summarizes the variables and Table 2 gives a list of all variables and their expected signs.
Table 1
List of variables, definitions and sources
Variable Definition Source
Banking crisis (dummy 0/1)
Systematically important financial institutions are in distress (0/1).
Laeven and Valencia (2008)
Financial
liberalization (de jure 1, index between 0-1)
Index created by Abiad which includes - Credit controls
- Interest rate controls - Entry barriers - State ownership
- Capital account restrictions
- Prudential regulation and supervision - Securities market policies
Abiad et al. (2010)
Financial
liberalization (de jure 2, index between 0-1)
Index created by Aizenman, Chinn and Ito which includes
- Multiple exchange rates - Restrictions on current account
transactions
- Capital account restrictions
- The requirement of the surrender of export proceeds
Aizenman et al. (2008)
Financial
liberalization (de facto, % of GDP)
An indicator of de facto liberalization measured by FDI assets, portfolio equity assets, FDI liabilities and debt liabilities.
Updated and extended version of the External Wealth of Nations Mark II database developed by Lane and Milesi-Ferretti (2007), Neumann and Penl (2008)
Credit growth (%) Domestic credit to private sector growth rate as % of GDP (annual growth).
World development indicators
Credit to private sector (% of GDP)
Domestic credit to private sector refers to
financial resources provided to the private sector.
World development indicators,
Büyükkarabacaka and Valev (2010)
indicators GDP volatility
(standard deviation of GDP per year per country)
The volatility is measured by the standard deviation of GDP. The annual growth of GDP per country minus the annual mean and then to avoid negative values squared and square rooted
World development indicators
Institutional quality (index between 0-1)
Indicator of quality of governance this indicator uses three variables
- Corruption in Government - Law and Order
- Bureaucratic Quality
International Country Risk Guide (ICRG), Bekaert, Harvey and Lundblad (2006)
Inflation (%) Inflation as measured by the annual growth rate of the GDP implicit deflator.
World development indicators, Ranciere et al. (2006) Secondary school enrollment (% of total)
Gross enrollment ratio. World development
indicators, Bekaert et al. (2006)
Interest rate (%) Real interest rate is the annual growth rate of the lending interest rate adjusted for inflation as measured by the GDP deflator.
World development indicators, Shehzad and de Haan (2009)
Population growth (%)
Annual growth of population. World development indicators, Ranciere et al. (2006)
Government size (% of GDP)
Government Consumption Share of PPP Converted GDP Per Capita at current prices.
Penn world table, Ranciere et al. (2006) Money and quasi
money growth (% of GDP)
Average annual growth rate in money and quasi money.
World development indicators, Shehzad and de Haan (2009)
Exchange rate (USD)
Purchasing power parity conversion factor is the number of units of a country's currency required to buy the same amount of goods and services in the domestic market as a U.S. dollar would buy in the United States.
World development indicators
GDP per capita
(USD) GDP per capita based on purchasing power parity (PPP).
World development indicators
Trade openness (% of GDP)
Trade is the sum of exports and imports of goods and services measured as a percentage of GDP.
Developed or developing country (dummy 0/1)
0 is a non OECD country and 1 is a developed country.
International Monetary Fund
Table 2
Variables and their expected signs Hypothesis 1 (Banking crisis) Hypothesis 2 (Credit growth) Hypothesis 3 (GDP volatility) Hypothesis 4 (GDP growth) 1. Credit to private sector + + + 2. GDP growth +/- + 3. GDP per capita +/- + +/- +
4. Real interest rate + +
5. Exchange rate +/- +/- +/-
6. Inflation + + + +/-
4. Methodology
We will use panel data analyses to test the hypotheses. This is the best method because we want to measure differences for multiple countries over several years.
Hypothesis 1 uses a probit panel data analysis because the dependent variable is a dummy. We should only test for multicollinearity, because the dependent variable is a dummy. In this model we use the lagged values (5 years) of the liberalization measures.
Hypothesis 2 tests whether there is an indirect effect of liberalization via credit growth on banking crises. To test this effect we will use the 2 stage least square regression model (2SLS). First we should test if Z (credit growth) mediates the X (financial liberalization) Y (banking crisis) relation. If credit growth mediates the financial liberalization-banking crisis relation the following conditions should hold
• Financial liberalization predicts banking crisis (test if X predicts Y) • Financial liberalization predicts credit growth (test if X predicts Z) • Credit growth predicts banking crisis (test if Z predicts Y)
• Test if financial liberalization still predicts banking crisis when credit growth is included in the model
The effect of financial liberalization on banking crisis should be significantly reduced by credit growth. The coefficient of financial liberalization (β1) is important in the first condition as well as in the second condition. The second condition shows the effect of financial liberalization on the mediation variable and the third shows the effect of the mediation variable on banking crisis. The last condition tests if credit growth reduced the effect of liberalization on banking crises. This model uses the lagged values (5 years) of the liberalization measures and credit growth.
Hypothesis 3 uses a panel dataset with an ordinary least squares regression (OLS). The volatility of GDP will be measured by the standard deviation of GDP growth per year per country as explained above. Furthermore the dependent variable must be homoskedastic and normally distributed. The equation used in this model is
Y it (Standard deviation GDP growth) = α + β1 * financial liberalization it + Β2 * control
Where I is country and T is year. In this model we also use the lagged values (5 years) of the liberalization measures.
Hypothesis 4 uses the same panel dataset. And the same tests, multicollinearity for all variables, heteroskedacity and normality checks for the dependent variable and the Durbin-Wu-Hausman test. This hypothesis tests whether institutional quality plays a role in GDP growth after financial liberalization. We will create three interaction variables between institutional quality and the three independent variables. Of interest are the interaction effects between institutional quality and the liberalization measures. The significance provides some information, but the main focus should be on the marginal effects. The equation for hypothesis 3 is;
Y it (GDP growth) = α + β1 * financial liberalization it + β2 * institutional quality it + β3 *
interaction variable it + Β4 * control variables it + it
Where I is country and T is year. In this model we also use the lagged values (5 years) of the liberalization measures.
4.1 Checks
Normal distribution is a continuous probability distribution to describe real valued random variables which tend to be clustered around a single mean value. In case of heteroskedasticity there is another estimator with smaller variance than OLS since least squares estimator cannot be said the best anymore but it is still linear and unbiased estimator. The reason why it is no more the best is that one of the least squares assumptions is violated; var(e) = σ2 will not be true. This will lead to another problem because standard errors of the least squares will be higher. This implies further problems in carrying out hypothesis tests and confidence interval calculations. They can be misleading because of the higher (not correct) standard errors.
parameter estimates. That may result in lack of statistical significance of individual independent variables while the overall model may be strongly significant. Multicollinearity may also result in wrong signs and magnitudes of regression coefficient estimates, and consequently in incorrect conclusions about relationships between independent and dependent variables.
Outliers can bias the result therefore we should deal with outliers. An outlier is an extreme score relative to the rest of the sample, they can be detected using scatter plots. For each variable we make a scatter plot to detect any outliers. Three countries (Congo Republic, Luxembourg and Zimbabwe) have several outliers in different variables therefore we exclude these countries from the sample. For example Luxembourg has many outliers not only for de facto, but also for money and quasy money and GDP per capita. These outliers are explained by the fact that Luxembourg is used as a tax haven, they have a big influence on the regression analysis. Furthermore the variable inflation has a lot of outliers to solve this problem we use the natural log of inflation (Micceri, 1989). There are different outliers for the other variables these outliers are rare events and bias the results of the regression analysis (regression analysis with outliers not presented). In total there are 12 outliers these are deleted (Judd and McClelland, 1989). Table 3 shows the descriptive statistics without outliers.
5. Empirical results
First we will provide an overview of the variables, after removal of outliers, which we will use in our analysis. This summary of the variables provide the variable names, the number of observations, the mean, the standard deviation, the minimum and maximum.
Table 3 Descriptive statistics
Variable Obs. Mean St. dev. Min max
Banking crisis dummy 2550 0,124 0,329 0 1
De jure 1 (Abiad et al.) 1640 0,452 0,286 0 1
De jure 2 (Aizenman et al.) 2072 0,391 0,336 0 1
De facto 2184 0,996 1,056 0,023 13,896
GDP growth 2390 3,002 6,241 -51,031 71,188
LN St. dev. GDP growth 2391 0,692 1,255 -7,193 4,613
Credit growth 2204 0,686 10,042 -104,586 125,361
GDP per capita 2550 6472,280 8144,736 0 47305,5
Real interest rate 1801 6,384 16,360 -98,145 88,114
Exchange rate 2372 0,572 0,291 0,112 2,080
LN Inflation 1963 2,091 1,480 -4,075 8,033
Money and quasy money 2009 43,505 36,295 0,918 242,239
Government size 2322 15,674 6,408 2,736 52,917 Secondary schooling 1963 66,220 33,952 2,569 160,347 Population growth 2539 1,557 1,361 -5,814 11,181 Trade openness 2334 71,118 38,244 6,320 280,361 Institutional quality 1835 0,557 0,236 0,042 1 Developed 2550 0,265 0,441 0 1
Interaction 1 (inst qual * de jure 1) 1398 0,319 0,262 0 0,964 Interaction 2 (inst qual * de jure 2) 1661 0,249 0,273 0 1 Interaction 3 (inst qual * de facto) 1736 0,561 0,563 0,005 6,007
Hypothesis 1
Hypothesis 1 tests the effect of liberalization on banking crises. We will test if there is an effect between financial liberalization and banking crisis via a probit regression analysis.
correlation above 0,8 this means that they measure different things. When the correlation was for example 0,95 then both measures of de jure would (almost) measure the same and we could exclude one of these de jure variables. There is no multicollinearity this is also supported by the variance inflation factor (VIF, not presented). The critical value of VIF is 10 a higher VIF value indicates multicollinearity. Because banking crisis is a dummy variable we do not have to check for heteroskedacity and normality.
Table 4 shows the direct effect on banking crisis, the model as a whole is significant at the 1% level. The de jure and de facto variables are the key variables to test hypothesis 1. The only significant liberalization measures are in the column of OECD countries (the first and second de jure measure at 5% level). The first measure of de jure is positive and the second measure is negative which is in line with our hypothesis. The second de jure is significant and negative which is not in line with our hypothesis (when a country liberalizes the probability of a banking crisis decreases). Column 2 shows that credit growth has a positive effect on banking crises. This also holds for non OECD countries, but not for OECD countries. GDP growth has a negative effect on banking crises which means that when GDP growth is high(er) the probability of a banking crisis is lower. GDP per capita is negative and significant at the 1% level for the total sample. Exchange rate is positive and significant for OECD countries at the 5% level. Inflation has a positive effect on banking crises and is significant in the total sample and for OECD countries both at the 5% level. Furthermore institutional quality is significant (total sample and non OECD countries at the 1% level) this coefficient is positive for the total sample and for non OECD countries.
Table 4
The effect of liberalization on banking crises
Banking crisis dummy Total sample OECD Non OECD
(2,46) (0,48) (2,51) GDP Growth -0,1152*** -0,1880* -0,1165***
(-6,74) (-1,76) (-6,59) GDP per capita -0,0001*** -0,0003 0,0000 (-2,86) (-1,46) (0,52) Real interest rate 0,0036 -0,0255 0,0016 (0,98) (-0,82) (0,39) Exchange rate -0,0234 5,8754** -0,1922
(-0,06) (2,42) (-0,34) LN Inflation 0,1404** 0,8616** 0,1036 (2,38) (2,13) (1,58) Money and quasy money 0,0008 0,0292* -0,0021
(0,32) (1,84) (-0,66) Credit to private sector -0,0018 -0,0086 0,0005 (-0,69) (-0,46) (0,16) Institutional quality 1,2780*** -2,8647 1,7061*** (2,70) (-0,89) (3,08) Constant -1,4530*** -4,7809 -1,6599*** (-3,82) (-1,40) (-3,63) Observations 816 170 646 Pseudo R2 0,15 0,46 0,15 LR chi2 87,80*** 32,61*** 77,49*** T-values in parentheses; * significant at 10%; ** at 5%; *** at 1%.
Hypothesis 2 tests the mediation effect of credit growth on banking crisis. The steps are explained in the methodology part. There are three independent variables that should be tested independently. As we are only interested in the mediation effect we will show the results of the binary mediation test. This test shows if the mediation effect is significant and what the effect is. We will start with de jure (Abiad et al., 2010).
The binary mediation test shows significant results at the 10% level. The total effect that is mediated is -0,19 (-19%) and the ratio of indirect to direct effect is -0,16 (16%). The second de jure measure does not show any significant results. The first two conditions hold for de jure 2 only the third and fourth condition are violated.
We will proceed with de facto variable. The de facto liberalization does not hold for any of the conditions therefore we cannot accept this hypothesis for de facto.
therefore we reject this hypothesis for OECD countries. For non OECD countries we find significant results for the first de jure at the 5% level. The proportion of total effect that is mediated is -0,16 (-16%) and the ratio of indirect to direct effect is -0,13. The second de jure measure does not show significant results the first condition holds but the other three do not hold. We did not find any significant results for de facto liberalization none of the conditions hold.
Hypothesis two is not accepted because the sign of the mediation effect is negative. Our hypothesis suggests that there is a positive effect of credit growth instead of a negative effect. The negative and significant mediation effect only holds for de jure 1 measure and for the total sample and non OECD countries. Therefore this hypothesis is not accepted. The effect is the opposite from what we expected for the total sample and for non OECD countries for de jure 1 measure.
Hypothesis 3
This hypothesis argues that countries which liberalize their financial market face a more volatile GDP growth. The volatility is measured by the standard deviation of GDP growth. The dependent variable must be normally distributed and homoskedastic. GDP growth volatility is not normally distributed; to solve this problem we will use the natural log (appendix C) of the GDP growth volatility. We should interpret the result now as percentage changes.
We should also check for heteroskedacity (appendix D). We do this by using the Breusch-Pagan test. The outcome of this test does not show heteroskedacity at the 10% level (chi2 of 3,34 and a prob of 0,0677). To make sure that we will not face problems of any kind of heteroskedasticity we use the robust function to correct for this.
The last check is the multicollinearity check this check includes all variables in the equation. The results are in appendix B it shows that for this regression there is no multicollinearity, the VIF scores do not show a sign of multicollinearity as well (not presented).
significant results for all three liberalization measures. However the results are positive and negative. The first de jure measure is significant and negative (5% level), but the second de jure measure is positive and significant (1% level). And the de facto measure is negative and significant at the 5% level. GDP per capita is positive and significant for OECD countries only at the 5% level. Population growth is significant and negative for non OECD countries at the 5% level. Trade openness is positive and significant at the 1% level for the total model, at the 10% level for OECD countries and at the 5% level for non OECD countries. This means that when a country is more open to trade they face a higher GDP growth volatility. Inflation is also positive and significant for column 2, 3 and 4 (respectively at the 1%, 5% and 1% level), which means that the higher inflation the more volatile GDP growth. Institutional quality is negative and significant for column 2 and 3 (Both at the 1 level) which means that higher institutional quality leads to lower GDP growth volatility. Credit to the private sector is negative and significant (5% level) for OECD countries. Exchange rate is positive and significant in all three columns (respectively at the 10%, 1% and 1% level) meaning that a higher exchange rate increases GDP growth volatility.
Table 5
The effect of liberalization on the volatility of GDP growth GDP growth volatility Total sample OECD Non OECD
(2,87) (1,86) (2,12)
LN Inflation 0,1148*** 0,1569** 0,1354***
(3,09) (1,97) (3,19) Institutional Quality -0,9288*** -1,8673*** 0,1583
(-2,69) (-3,13) (0,35) Credit to private sector -0,0012 -0,0040** 0,0007
(-0,86) (-2,05) (0,31) Exchange rate 0,4570* 1,0546*** 1,2099*** (1,89) (3,28) (2,91) Constant 0,6933*** 0,594207 -0,057784 (2,63) (1,25) (-0,16) Observations 981 390 591 R2 0,04 0,11 0,06 F-value 3,57*** 3,60*** 3,86*** T-values in parentheses; * significant at 10%; ** at 5%; *** at 1%. Hypothesis 4
This hypothesis argues that countries which liberalize with a higher institutional quality face a higher GDP growth than countries that liberalize with low quality institutions. This hypothesis is also an ordinary least square regression. Appendix E shows the normality check for GDP growth it shows that the dependent variable is normally distributed. The Breusch-Pagan test shows that the dependent variable is homoskedastic (Chi square is 0,42 and a p-value of 0,515, appendix F). Appendix B shows that there is multicollinearity between the interaction variables. The interaction variables are highly correlated because this is a variable created by two other variables. The problem of multicollinearity in an interaction model is overstated (Brambor, Clark, and Golder, 2006). Meaning that a high correlation between those two variables should not be a problem for our analyses. The other variables are not highly correlated so we will proceed with the regression analysis. The model is significant as a whole at 1 % level.
institutional quality lead to lower GDP growth. Since institutional quality and the liberalization measures are always larger or equal to zero the point estimate of the marginal effect is never positive. GDP per capita is significant and positive for OECD countries (5% level). Also Government size is significant for the total model, OECD countries and non OECD countries (respectively at the 1%, 1% and 10% level) the coefficients are negative. This means that a bigger government leads to a lower GDP growth. Population growth is negative and significant for the total model (10% level) and for non OECD countries (5% level). But the coefficient for OECD countries is positive and significant (1% level). Trade openness is positive and significant at the 1% level for all three models which means that when a country is more open to trade GDP increases. Inflation is significant for all three models the coefficient is negative respectively at the 1%, 5% and 1% level. Institutional quality is positive and significant at the 5% level for non OECD countries meaning that better institutions lead to higher GDP growth.
Table 6
The influence of institutional quality on the effect of liberalization on GDP growth
GDP growth Total sample OECD Non OECD
(-0,82) (-0,99) (-0,59) Interaction 2 (de jure 2 * inst qual) -2,9957* 4,1198 -6,1297
(-1,80) (0,90) (-1,47) Interaction 3 (de facto * inst qual) 0,3120 -1,0528 -3,5048
6. Robustness
We test the robustness of our results in a number of ways. First of all we make a distinction between developed (OECD) and developing (non OECD) countries. The different results are presented and explained in the analysis in section 5.
Splitting up between OECD and non OECD countries is important, but another important check is to divide the data in different time periods. Appendix G shows the different crises periods and the liberalization measures graphically. Based on this graph we make a distinction between every five years the results are presented in appendix H – J. The regression analyses includes the same (control) variables as the previous regression analyses. We present only the key variables for our analyses, these are for hypothesis 1 the liberalization measures and credit growth, for hypothesis 2 the mediation effect, for hypothesis 3 the liberalization measures and for hypothesis 4 the interaction effects. The results show that there are differences in time, there are variables which are negative in some periods and in other periods positive. The significant results for hypothesis 1 are all positive this holds for the total model and for non OECD countries, meaning that liberalization leads to an increase in banking crises. This is in line with our hypothesis. There are no results for OECD countries these countries do not have enough observations to analyze the hypothesis correctly. Hypothesis 2 does not have any significant results (not presented).
Hypothesis 3 shows significant and positive results for de jure 1 and de facto, but de jure 2 shows significant and negative effects. Also this hypothesis jumps from significant to insignificant results for the different periods. Hypothesis 4 shows positive and negative results as well, for example for the total model interaction 2 in the year 91-95 and 96-00 is negative and significant and in 01-05 positive and significant. This might seem strange but as explained in the introduction liberalizing the market can be a “trend” and can be seen as beneficial for every country which is not true in the real world.
7. Conclusion
This paper provides an empirical evaluation of the effects of liberalization on banking crises, credit growth, GDP growth volatility and GDP growth. The previous literature describes positive as well as negative effects of liberalization, therefore we construct these four hypotheses. Section 5 analyzed the hypotheses and presented the results, table 7 gives a short overview of the hypotheses, the results and their tests.
Table 7 Overview hypotheses
This paper examined the relation between financial liberalization and the occurrence of banking crisis, GDP growth volatility and the effect of institutional quality during liberalization on GDP growth. In the first model we investigate whether there is an effect of liberalization on banking crises. We find positive results for de jure 1 measure for OECD countries, but we don’t find any significant results for the total model or for non OECD countries. When we split the sample in different periods we find positive and significant results for the total sample and non OECD countries (appendix K) which is in line with our hypothesis.
Relation Expected sign Analysis Conclusion
Hypothesis 1 Liberalization leads to banking crises + financial liberalization Probit regression Hypothesis is accepted. Hypothesis 2 Liberalization leads
via credit growth to banking crises Liberalization + credit growth + banking crises Binary mediation Hypothesis is rejected. Hypothesis 3 GDP growth volatility increases with financial liberalization + Financial liberalization Ordinary least squares regression Hypothesis is partly accepted.
The second model investigates the mediation effect of credit growth, liberalization via credit growth on banking crises. Our hypothesis does not hold and we do not find any evidence that there is an indirect effect of liberalization on banking crises via credit growth. We do find a direct effect; however this effect is negative and only holds for OECD countries and the second de jure liberalization. This is the opposite of what we expected.
Furthermore this research does find little evidence that GDP growth becomes more volatile after financial liberalization for de jure 2 (non OECD countries). The de jure 1 shows negative signs for the total model and for non OECD countries. The de facto liberalization gives a negative sign this holds only for non OECD countries. When we split the sample in 5 year periods we still find negative results for de jure 1 and de facto liberalization. The de jure 2 gives for the total model positive significant results and for OECD countries negative results for the total model, but for the period 01-05 positive results. The hypothesis is accepted for the de jure 2 liberalization however the de jure 1 and the de facto liberalization show opposite results.
The final finding is that the quality of institutions does matter for GDP growth during liberalization. We only find a significant result for the second de jure liberalization of the total model; this effect is negative which is not in line with our hypothesis. If we make a distinction between time periods the results change from positive to negative (appendix K).
8. Limitations and future research
This study tries to correct for some limitation, however there are some shortcomings which should be named and dealt with in future researches. To start with the datasets used in this research are not balanced with a balanced dataset the results would be more robust.
Secondly, this analysis uses a dummy variable for banking crises periods. A dummy variable captures the event of a crisis but does not capture for example the severity. We now compare different crisis periods as if they are the same in reality this is not true. Therefore in future research you can use for example the severity of a crisis to correct for this. Thirdly, you can take a look at the microeconomic perspective rather than the macroeconomic perspective. With a microeconomic perspective you can make a distinction between systemic and non-systemic banking crisis. When using the microeconomic perspective we can distinguish between banks in the same country and use Mishkin (1997) and Minsky (2008) their views. According to Mishkin crises are failures of capital markets, asymmetric information problems and according to Minsky banks become too greedy and engage in excessive risk taking.
References
Abiad, A., Detragiache, E. and Tressel, T., 2010. A new database for financial reform. IMF Staff Papers, Palgrave Macmillan Journals, 57(2): 281-302.
Aizenman, J., Chinn, M. and Ito, H., 2008. A New Measure of Financial Openness. Journal of Comparative Policy Analysis, 10(3): 307-320.
Bardhan, P., 1997. Corruption and development: a review of issues. Journal of Economic Literature, 35: 1320-1346.
Barro, R.J., 1999. Determinants of Economic Growth: A Cross-Country Empirical Study. Cambridge (2nd ed.) MIT Press Books, The MIT Press.
Beck, T., Demirgüç-Kunt, A., and Levine, R., 2006. Bank Concentration, Competition, and Crises: First results. Journal of Banking and Finance, 30: 1581-1603. Bekaert, G., Harvey C.R. and Lundblad, C., 2005. Does Financial Liberalization Spur
Growth? Journal of Financial Economics, Elsevier, 77(1): 3-55.
Bekaert, G., Harvey C.R. and Lundblad, C., 2006. Growth volatility and financial liberalization. Journal of International Money and Finance, 25: 370-403.
Bonfiglioli, A., 2005. How Does Financial Liberalization affect Economic Growth? Seminar Papers 736, Stockholm University, Institute for International Economic Studies.
Bonfiglioli, A. and Mendicino, M., 2006. Financial liberalization, bank crises and growth: Assessing the links. Working paper series in economics and finance, 946.
Brambor, T., Clark, W.R. and Golder, M., 2006. Understanding Interaction Models: Improving Empirical Analyses. Political Analysis 14: 63-82.
Büyükkarabacaka, B. and Valev N., 2010. Credit Expansions and Banking Crises: The Roles of Household and Firm Credit. Journal of Banking & Finance, 34(6): 1247-1256.
Calve, G.A., Leiderman, L. and Reinhart, C.M., 1993. Capital inflows and real exchange rate appreciation: the role of external factors. IMF Staff Papers, 40: 108-51. Demirgüç-Kunt, A. and Detragiache E., 1997. The Determinants of Banking Crises in
Demirguc-Kunt, A. and Detragiache, E, 2002. Does deposit insurance increase banking system stability? An empirical investigation. Journal of Monetary Economics, Elsevier 49(7): 1373-1406.
Demirguc-Kunt, A. and Serven, L., 2009. Are all the sacred cows dead? Implications of the financial crisis for macro and financial policies. Working Paper 4807, World Bank.
Dollar, D. and Kraay, A., 2000. Property Rights, Political Rights, and the Development of Poor Countries in the Post-Colonial Period. World Bank Working Papers. Ehrlich, I. and Lui, F.T., 1999. Bureaucratic corruption and endogenous economic growth.
Journal of Political Economy, 107(6): 270-293.
Eichengreen,. B. and Leblang, D. 2003. Capital Account Liberalization and Growth: Was Mr. Mahathir Right. International Journal of Finance and Economics, 8: 205-224.
Fischer, S. and International Monetary Fund, 1997. Capital account liberalization and the role of the IMF. Washington D.C. International Monetary Fund.
Grilli, V. and Milesi-Ferretti, G.M., 1995. Economic Effect and Structural Determinants of Capital Controls. IMF Staff Papers, 42(3).
Hellman, T.F., Murdock K.C. and Stiglitz, J.E., 2000. Liberalization, Moral Hazard in Banking, and Prudential Regulation: Are Capital Requirements Enough? The American Economic Review, 90(1): 147-165.
Inklaar, R. and Yang, J., 2011. The impact of Financial crises and tolerance for uncertainty.
Judd, C.M. and McClelland, G.H., 1989. Data Analysis: A Model-Comparison Approach. San Diego, CA: Harcourt Brace Jovanovich.
Kaminsky, G.L. and Reinhart, C.M., 1996. Banking and Balance-of-Payments Crises: Models and Evidence. Working paper, Board of Governors of the Federal Reserve, Washington, D.C.
Kaminsky, G.L., Lizondo S. and Reinhart, C.M., 1998. Leading Indicators of Currency Crisis. IMF Staff Papers, 45(1).
Knack, S. and Keefer, P., 1995. Institutions and economic performance: cross-country tests using alternative institutional measures. Economics and Politics, 207–227. Kraay, A., 1998. In Search of the Macroeconomic Effects of Capital Account
Liberalization. World Bank, Mimeo.
Krugman, P., 1993. International Finance and Economic Development. Finance and Development: Issues and Experience, Cambridge University Press, 11-28.
Laeven, L.A. and Valencia F.V., 2008. Systemic Banking Crises: A New Database. IMF Working Paper No. 08/224.
Lane, P.R. and Milesi-Ferretti, G.M., 2007. The External Wealth of Nations Mark II. Journal of International Economics, 73: 223-250.
Levine, R., 2001. International financial liberalization and economic growth. Review of international economics, 9(4): 688-702.
Mauro, P., 1995. Corruption and growth. Quarterly Journal of Economics, 110: 681-712. Mehrez and Kaufmann, 2000. Transparency, Liberalization, and Banking Crises. Policy
Research Working paper 2286.
Micceri, T., 1989. The unicorn, the normal curve, and other improbable creatures. Psychological Bulletin, 105: 156-166.
Minsky H., 2008. Stabilizing an instable economy. New York: Mc Graw Hill.
Mishkin F.S., 1997. Understanding Financial Crises: a Developing Country Perspective. Working Paper 5600. National Bureau of Economic Research.
Neumann, R. and Penl R., 2008. Volatile capital flows: Interactions between de jure and de facto financial liberalization. Economics Bulletin, 6(2): 1-10.
Obstfeld, M. and Taylor, M.A., 2004. Global Capital Markets: Integration, Crisis, and Growth. Cambridge: Cambridge University Press.
Prasad, E.S. and Rajan, R.G., 2008. A Pragmatic Approach to Capital Account Liberalization. Journal of Economic Perspectives, 22(3): 149-172.
Reinhart, C. and Rogoff, K.S., 2008. Banking Crises: An Equal Opportunity Menace. NBER Working Papers 14587, National Bureau of Economic Research.
Rodrik, D., 1998. Who Needs Capital-Account Convertibility? Essays in international finance no. 207.
Rodrik D. and Subramaniam A., 2009. Why Did Financial Globalization Disappoint? IMF staff papers 56, 112-38.
Shehzad, C.T., and De Haan, J., 2009. Financial Liberalization and Banking Crises. CESifo Working Paper Series 2870.
Shleifer, A. and Vishny, R.W., 1993. Corruption. Quarterly Journal of Economics, 108: 599–617.
Stiglitz, J.E., 2010. Contagion, Liberalization, and the Optimal Structure of Globalization. Journal of Globalization and Development, 1(2).
Overview of appendixes
Appendix A Overview of countries in the sample Appendix B Correlation matrix
Appendix C Normality check of (natural log) standard deviation GDP growth with normal density plot
Appendix D Scatter plot residuals (hypothesis 3)
Appendix E Normality check GDP growth with normal density plot Appendix F Scatter plot residuals (hypothesis 4)
Appendix G Graphic overview of the measures of liberalization and banking crises Appendix H Overview of the results of the total model per 5 year period
Appendix A
Countries in the sample
Albania Ecuador Kuwait Romania
Algeria Egypt Kyrgyzstan Russia
Argentina El Salvador Latvia Senegal
Austria Equatorial Guinea Lebanon Sierra Leone
Azerbaijan Eritrea Liberia Slovakia
Bangladesh Estonia Lithuania Spain
Belarus Finland Madagascar Sri Lanka
Belgium France Malaysia Swaziland
Benin Georgia Mali Sweden
Bolivia Germany Mauritania Switzerland
Brazil Ghana Mexico Tanzania
Bulgaria Greece Mongolia Thailand
Burkina Faso Guinea Morocco Togo
Burundi Guinea-Bissau Mozambique Tunisia
Cameroon Guyana Nepal Turkey
Chad Hungary Netherlands Uganda
Chile Iceland Nicaragua Ukraine
China India Niger United Kingdom
Colombia Indonesia Nigeria United States
Congo, Dem Republic Ireland Norway Uruguay
Costa Rica Israel Panama Venezuela
Croatia Jamaica Paraguay Vietnam
Czech Republic Japan Peru Yemen
Denmark Jordan Philippines Zambia
Djibouti Kazakhstan Poland
Appendix C
Normality check of (natural log) standard deviation GDP growth with normal density plot
Appendix D
Scatter plot residuals (hypothesis 3)
Appendix E
Normality check GDP growth with normal density plot
Appendix F
Appendix G
OECD versus non OECD countries
Appendix H
Total model (five year period)
Banking crisis dummy 81-85 86-90 91-95 96-00 00-05
De jure 1 -40,4133 -0,6890 3,1309*** -1,7213 1,0437 (-0,00) (-0,43) (2,61) (-1,50) (0,53) De jure 2 -296,5805 0,2142 -0,7227 1,4924** 1,6920** (-0,00) (0,25) (-0,88) (2,21) (2,18) De facto -45,3682 1,0988*** -0,3896 -0,4209 -0,2619 (-0,00) (2,63) (-0,97) (-1,26) (-0,56) Credit growth 6,7748 0,0027 0,0051 0,0637*** 0,0248 (0,00) (0,06) (0,21) (3,05) (1,25) Observations 50 146 152 195 273 Pseudo R2 1,00 0,26 0,29 0,39 0,42 LR chi2 43,97*** 28,23*** 40,60*** 68,41*** 44,33***
Standard dev. GDP growth
R2 0,21 0,17 0,14 0,12 0,11
F-value 2,58*** 3,90*** 3,41*** 2,99*** 5,10***
GDP growth
Interaction 1 (de jure 1 * inst qual) -2,9830 -5,7825 9,7538 14,1369* 3,3830
(-0,26) (-0,94) (1,32) (1,94) (0,31)
Interaction 2 (de jure 2 * inst qual) -8,9059 -1,7804 -11,1273*** -17,3325*** 6,1924*
(-0,00) (-0,63) (-2,82) (-4,17) (1,94)
Interaction 3 (de facto * inst qual) 10,2183*** 1,9242 -0,8211 2,8895* -1,3613
(2,28) (0,74) (-0,37) (1,80) (-0,92) Observations 76 204 195 205 306 Pseudo R2 0,29 0,17 0,20 0,21 0,22 LR chi2 2,62*** 2,65*** 6,32*** 4,51*** 11,76*** T-values in parentheses; * significant at 10%; ** at 5%; *** at 1%. Appendix I
OECD countries (five year period)
Banking crisis dummy 81-85 86-90 91-95 96-00 00-05
De jure 1 129,3661 55,8907 (0,00) (0,00) De jure 2 -32,6889 -15,2651 (-0,00) (-0,00) De facto 1,81497 -3,72435 (0,00) (-0,00) Credit growth -0,29552 -0,5536 (-0,00) (-0,00) Observations 24 40 Pseudo R2 1,00 1,00 LR chi2 26,99*** 21,31***
Standard dev. GDP growth
GDP growth
Interaction 1 (de jure 1 * inst qual) 0,5510 -51,8367*** 21,7435 -22,8701 -29,3987
(0,02) (-4,50) (0,93) (-0,96) (-1,28)
Interaction 2 (de jure 2 * inst qual) 3,1308 28,6038*** -15,6723 12,7755 20,7981**
(0,00) (6,51) (-0,59) (1,43) (2,01)
Interaction 3 (de facto * inst qual) -2,7765 -10,1872 18,1901* -3,5713 3,8703
(-0,26) (-1,61) (1,88) (-0,72) (1,23) Observations 29 84 88 86 108 Pseudo R2 0,52 0,37 0,34 0,45 0,36 LR chi2 1,94 17,32*** 5,35*** 4,51*** 7,35*** T-values in parentheses; * significant at 10%; ** at 5%; *** at 1%. Appendix J
Non OECD countries (five year period)
Banking crisis dummy 81-85 86-90 91-95 96-00 00-05
De jure 1 -32,1831 -0,4731 2,9051** -1,8677 1,7569 (-0,00) (-0,29) (2,18) (-1,48) (0,89) De jure 2 -182,4209 -0,0556 -0,4640 1,4826** 1,3287 (-0,00) (-0,06) (-0,53) (2,01) (1,64) De facto -14,2257 0,9076** -0,1234 -0,2913 0,0393 (-0,00) (2,00) (-0,30) (-0,69) (0,07) Credit growth 0,1087 0,0061 -0,0071 0,0676*** 0,0548 (0,00) (0,14) (-0,29) (2,80) (1,37) Observations 40 121 128 155 202 Pseudo R2 1,00 0,22 0,30 0,38 0,41 LR chi2 40,03*** 21,89*** 34,21*** 57,27*** 39,10***
Standard dev. GDP growth
Interaction 1 (de jure 1 * inst qual) -15,3264 -8,2011 61,5279** -16,5273 58,2611**
(-0,69) (-0,43) (2,17) (-0,79) (2,18)
Interaction 2 (de jure 2 * inst qual) -20,4542 -4,7366 -31,9064** -16,4513 -5,0507
(0,00) (-0,64) (-2,25) (-0,77) (-0,38)
Interaction 3 (de facto * inst qual) 30,7521 8,8841 -24,0317** -3,3314 -28,2880***
Appendix K
