The Effect of Financial Liberalization on Bank Competition
Bas van Kesteren
Abstract
I.INTRODUCTION
There have been waves of liberalization measures on financial structures in numerous countries around the world over recent decades. However, humanity has witnessed these waves of financial liberalizationin addition to numerous occurrences of banking crises. Some researchers appear to be in agreement that financial liberalization contributed to the precedent incidents of financial volatility and that one of the main reasons is the intensified competition due to the financial liberalization. The intensified competition encourages riskier banking conduct, while governments aspired to attain more efficient markets. (Chen, 2001)
There is no disbelief in the notion that intensified competition can lead to riskier
banking conduct, but we do question if financial liberalization has this assumed positive effect on competition, seeing that the existing literature describes two opposing theories on the effect of financial liberalization on competition. The structure conduct
performance hypothesis indeed predicts that due to entry barriers liberalization more firms will enter the market, leading to more competition. Then again, the efficient-structure theory describes how financial liberalization leads to failure or merging activity from inefficient firms, consequently leading to less competition.
We will examine the effect of financial liberalization on different indicators of competition. Previous similar studies have used one indicator, mostly concentration, which is a widely criticized measure of competition. We will use a dataset on financial liberalization which is very recent and has not been used extensively. Most studies focus on just one or a few dimensions of financial liberalization, whereas our dataset uses 5 dimensions.
come up with general conclusions regarding the impact of financial liberalization on bank competition. Our study uses a multi-country and multi-year sample of 34 countries and 12 years.
The paper is organized as follows, in section II we will mostly discuss the existing literature on the two opposing theories on the effect of financial liberalization on bank competition. Section II will commence with a short discussion on financial
liberalization and the role of competition in linking financial liberalization and financial volatility. The end of section II will include a brief overview and discussion of the control variables we will use and a review of the critique on using concentration as a measure of competition. In section III we will discuss the data and the selection of the variables. Section IV will describe the methodology. Subsequently, the results of the empirical analysis are presented and discussed in section V. The paper will end with a conclusion and recommendations for further research in section VI.
II.FINANCIAL LIBERALIZATION AND BANK COMPETITION
II.1. Financial Liberalization and its Consequences
In numerous countries around the world financial liberalization has occurred in several ways. Official government policies have been implemented that focus on deregulating credit controls, deregulating interest rate controls, removing entry barriers for foreign financial institutions, privatizing financial institutions and removing restrictions on foreign financial transactions. (Demirgüç-Kunt and Detragiache, 1998; Hermes and Lensink 2008)
In the specific case of developing and emerging economies there has been a wave of financial liberalization policies as of the early 1990s. More and more, these policies are not forced from the outside world, whereas these policies used to be enforced
Policymakers, and particularly individuals in finance ministries across the developing and developed countries, show to have internalized the idea that financial liberalization measures are essential to improve the performance of the financial sector in terms of competitiveness. (Ghosh, 2005; Hermes and Vu, 2008)
Governments wish to improve competitiveness via the implementation of these financial liberalization policies in order to enhance the conditions for market competition. The governments wish to attain increasingly more efficient financial institutions by making the market less state-directed and thus exposing the market to increased competition. (Barajas and Steiner, 2000; Beck, 2008) Competitive pressure encourages banks to develop into more efficient institutions by means of reducing overhead costs, improving on overall bank management, improving risk management and offering new financial instruments and services. (Denizer et al., 2000; Hermes and Vu, 2008; Claessens et al., 2001)
Governments, however, cannot unthinkingly assume that financial liberalization only leads to increased competition as there are also forces resulting from financial
liberalization that lead to less competition. Consequently one could say that in most of the literature, on the effect of financial liberalization on bank efficiency and financial instability, financial liberalization measures may be falsely assumed to have a positive effect on competition. We will briefly discuss the role of competition within the literature on the effect of financial liberalization on bank efficiency and financial instability to point out the importance of the relationship between financial liberalization and bank competition.
Chen (2001) and other researchers argue that one of the main causes for banking
inflict restricted entry and thus restrict competition, existing banks have enhanced profit prospects and better capital cushions. As a result banks have fewer incentives to take excessive risk, which has positive consequences for financial stability. (Beck, 2008; Marcus, 1984; Keeley, 1990)
In the next subsection we will commence our main research with a review of the theory behind the increase in competition attributable to financial liberalization, the structure conduct performance hypothesis. After this we will discuss the efficient-structure theory, which gives an explanation for a decrease in competition in consequence of financial liberalization.
II.2. The Structure Conduct Performance Hypothesis
As stated before, governments wish to increase competition with their financial
liberalization measures. Governments and quite some researchers assume that the main effect from financial liberalization is in fact that competition increases.
In the existing literature the positive effect of financial liberalization on competition is based on the removal of entry barriers, a part of financial liberalization. (Chizzolini, 2002) This relationship is described within the structure conduct performance paradigm. This paradigm states that the elimination of barriers to entry would increase the number of competitors in the banking markets, thus increasing competition.
In fact the structure conduct performance paradigm has been the encouragement of many financial liberalization policies throughout industrialized countries and of deregulation in the banking industry too. Many predictions and empirical tests of the effects of financial liberalization, i.e. deregulation, in the banking industry are anchored in the structure conduct performance paradigm. (Berger et al., 2004)
dominant paradigm in industrial organization theory from 1950 till the 1970s, draws on disparities in market structure and business behavior to give an explanation of
differences in market performance. The features of market performance that attracted the majority of interest are output levels, the relation of price to marginal cost, consumers’ surplus and the amount of economic profit. As the name implies, the
structure conduct performance paradigm puts the emphasis on structure as a determinant of market performance.
The structure refers to market structure defined mainly by the concentration of market share in the market. Conduct refers to the behavior of firms. Conduct examines if the firms’ behavior is either competitive or collusive. This can be observed through for instance pricing, research and development, advertising, production, choice of
technology, entry barriers and predation. Finally, performance refers to efficiency. This is mainly defined by the extent of market power, which can be measured by pricing. The greater the market power, the higher is pricing, the lower is efficiency. The
important part of the structure conduct performance paradigm in terms of our research is derived from the first hypotheses of the paradigm. Structure influences conduct,
implying that a lower concentration leads to more competitive behavior of firms. We will come back to this after we discuss the other part of the structure conduct
performance paradigm which is important in terms of our research.
One of the main problems with the initial structure conduct performance paradigm was that the results of empirical investigations, conducted to test the relationships between structure, conduct and performance, on cross-sections of industries showed a very weak statistical relationship. The cause of this problem was endogeneity. The basic
what kind of entry barriers are created, what kind of entry barriers are in place and how larger firms predate small firms. The structure conduct performance approach now recognizes that market structure is also the product of economic forces. The paradigm now accedes that entry and exit affect market concentration. (Hannan, 1991; Martin, 2001)
In light of our particular research the important part of the structure conduct
performance hypothesis lies in the negative relationships between concentration and competition and the fact that entry and exit have a respectively negative and positive effect on concentration. Financial liberalization partly causes the removal of entry barriers, which decreases the concentration through entry from new competitors. This decreased concentration implies more competition as more firms will be striving for a greater share of the market.
II.3. The Efficient-Structure Theory
The assumption that financial liberalization has a positive effect on competition could very well be false, as there are also forces due to financial liberalization which have negative effects in terms of competition.
The efficient-structure theory argues that the increased competitive pressures on the banking sector from the financial liberalization forces inefficient banks to either improve their performance or exit the industry throughout merger or absolute failure, thus increasing bank concentration. (Demsetz, 1973; Winston, 1998) The liberalization of financial services in the European Union and the establishment of the Economic and Monetary Union (EMU) have already led to a wave of mergers in the European banking industry. (Bikker and Haaf, 2000)
continue to exist, whereas fragile performers are supposed to reduce in size, exit, or sell out. This relocation of market share from under-performers to more prosperous firms is an important element of the competitive process. (Stiroh and Strahan, 2002)
Financial liberalization measures are taken because governments wish to create increasingly efficient financial institutions. Increased efficiency in this case refers to improving on overall bank management, improving risk management and offering new financial instruments and services. Governments wish to attain these increasingly efficient financial institutions by making the market less state-directed and again by exposing the market to increased market competition. Winston (1998) argues that in state-directed markets the state protects weak firms. Consequently, the financial
liberalization which makes the market less-state directed also implies less protection for weaker firms.
Economic theory predicts that every profit-maximizing firm ought to minimize its costs at all times. The case of a state-directed firm is a different situation. Even though firms in partly state-directed markets can select their technologies and operating practices, the alternatives are made conditional on the state's power over prices, entrance and exit. This is also a situation without the challenges created by unobstructed competition from present firms and new entrants. Consequently, managers and other recruits in a state-directed market face a reasonably dissimilar set of incentives in searching for greater efficiency. Winston (1998) discusses three fundamental causes why an industry becomes increasingly efficient after liberalization from the decrease in state-direction. In a state-directed market competition amongst firms is limited. This lack of
Furthermore, particular policies can force firms to work in an inefficient way. Winston (1998) gives a couple of examples of policies which have forced particular firms within American industries other than the banking sector to operate inefficiently. Entry barriers prevented firms such as airlines from developing their networks optimally, exit barriers prevented firms such as railroads from detaching excess capacity and price regulations prevented firms such as natural gas pipelines from efficiently marketing their capacity during peak and off-peak periods.
The third cause described by Winston (1998) argues that regulations prevent firms from responding effectively to macro-economic instability, for example a recession or a great unforeseen change in prices or interest rates. An industry exposed to regulation may be to some extent insulated from these shocks. Liberalized markets give firms the ability to act in response more effectively to external turmoil.
To sum up, the intensified competition consequential from liberalization causes firms to make innovations in marketing, operations, technology, and governance that facilitate them to become more efficient, advance their service quality, launch new products and services, and become extra alert to consumers' preferences. Where weak firms could survive without these innovations in the past, following the liberalization the
competition in markets has become far more intense. Whereas regulation, ambiguity and further barriers to entry use to be able to shelter inefficient firms by limiting entry and exit and preventing the competition from pushing them out of the market or taking them over, liberalization has forced inefficient firms to either improve their
performance or leave the industry throughout merger or outright stoppage.
II.4. Previous Empirical Research
Only a few studies have examined the effect of financial liberalization on bank
competition. Most research has had its focus on the effect of financial liberalization on development and efficiency. Also, in many of these cases the focus is on deregulation, whereas financial liberalization covers more (e.g. privatization). An example is Barth et al. (2001), whom constructed a database on the regulation and supervision of banks in 107 countries. With the use of their database they focused on the relationship between bank performance and stability with differences in bank regulations and supervision. Their main findings were that countries with policies that promote private monitoring of banks have better bank performance and more stability and countries with more
generous deposit insurance schemes tend to have poorer bank performance and greater bank fragility.
One of the few related panel studies was conducted by Claessens and Laeven (2004). They related competitiveness with countries’ banking system structures and regulatory regimes for a panel of 50 countries for the period 1994-2001. They found that countries with systems that have greater foreign bank entry and fewer entry and activity
restrictions to be more competitive.
Another related study was conducted by Berger and Hannan (1989). They wrote one of the first papers on the price-concentration relationship. They used data from U.S. banks for the period 1983-1985. They found that the banks in the most concentrated markets of their sample pay lower money market deposit account rates than those paid in the least concentrated markets. Berger and Hannan (1989) state that the structure-performance hypothesis predicts this outcome as prices will be less favorable to consumers in concentrated markets because of the non-competitive behavior exhibited in such markets.
should force a kind of conduct or rivalry among member firms that would be reflected in a relatively large amount of mobility and turnover.” Empirical research by both Bodenhorn and Heggestad and Rhoades support this hypothesis. In light of our research, using this measure of competition, Bodenhorn finds that both entry and free banking are found to increase competitive behavior.
Molyneux and Forbes (1995) also use the same two theories as we do, the efficient structure hypothesis and the structure conduct performance hypothesis, in their research on the relationship between market structure and firm performance. Molyneux and Forbes use pooled data for European banks for the period 1986-1989. They find results which support the structure conduct performance hypothesis. The two theories which we use to link financial liberalization with competition can also explain the relationship between market structure and firm performance. Here the efficient structure hypothesis explains a positive relationship between firm profits and market structures with the gains made in market share by more efficient firms. The structure conduct performance hypothesis links profits and structure through the proposition that market concentration fosters collusion amongst firms.
Most of the little studies that have looked into the competition effects of financial liberalization policies focus on just one country. Denizer (1997) also observes
ambiguous theories on the effect of financial liberalization on competition. Eventually, Denizer finds that Turkey’s financial reforms have had a significant positive effect on competition. Denizer focuses on competition in the retail banking market. In particular, he analyzes the impact of new bank entry and sunk investments in the system that resulted from pre-1980 interest rate and regulatory policies on competition within Turkey.
“monopsony power” conduct parameters for the non-indexed local currency loan and deposit markets. These empirical findings gave us more reason to use multiple
competition indicators instead of just using concentration as sole indicator.
From these previous studies we can conclude that we will fill three research gaps within the existing literature. First of all we will use a more extensive measure of financial liberalization than previous similar studies which use only regulation. Secondly, we will examine the effect of financial liberalization on competition in a panel with multiple countries. Country specific studies leave open the possibility that in one country bank competition improves after liberalization, while in another country the opposite is found. Country-specific studies, therefore, may make it more difficult to come up with general conclusions regarding the impact of financial liberalization on bank competition. Lastly, we will use multiple competition indicators to research the effect. In the next subsection we will discuss why this study calls for the use of multiple indicators of competition.
II.5. Measuring Competition
In the literature competition in the banking industry has been measured in several ways. Researchers have not yet found the best way to measure competition, so still recent research into competition uses different measures. In order to make our research a bit more reliable and to walk in front of the critique on any single measure of competition we will use three kinds of measures. If we can find consistent results from these 3 different proxies for competition, these indicators of competition will become increasingly reliable as measures of competition.
consideration that banks with dissimilar ownership act differently and that banks might not contend directly with each other in the same line of business. We will take this criticism into account when conducting our analysis.
Another measures which we will use relates to the existence of scale and scope
economies. It is often assumed that unused scale economies would be exploited and, as a result, reduced under strong competition. Therefore, the existence of non-exhausted scale economies is an indication that the potential to reduce costs has not been
exhausted and, therefore, can be viewed as an indirect indicator of a lack of competition (Bikker and van Leuvensteijn, 2007).
A common method to represent economies of scale is the use of overhead costs
expressed as a fraction of total assets. Demsetz (1973) argues that more efficient banks have lower costs and gain greater market share. As overhead costs are general expenses of operating a business that are not directly related to the services provided, in the case of banks, they can be minimized by more efficient firms as the average costs will decrease with the size of the firm.
The third measure of competition that we will use is the interest income minus interest expense divided by interest-bearing assets, namely the net interest margin. Market power can be related to profit, in the sense that very high profits may be indicative of a lack of competition. A profit-maximizing monopolist operates with an interest gap that maximizes profit. On the other hand in the case of a perfectly competitive market, each bank operates with an interest gap that is at a point where all banks earn a normal rate of return on incentive. In the latter case there is no incentive for entry or exit.
Consequently, in the case of a high interest gap, the market moves towards a market supplied by a monopolist. (Martin, 2001)
earnings and variations in net interest income is a key determinant of changes in profitability for a greater part of banks worldwide.
The data on bank concentration, the net interest margin and overhead costs will be discussed further in section 3. Please note that all three proxies for competition are inverse indicators of competition.
II.6. Control Variables
In order to investigate the effect of financial liberalization on bank competition we will include several control variables in our empirical analysis.
To start with, Claessens and Laeven (2004) argue that the banking system is less probable to be more competitive when it is exposed to high inflation as prices of financial services such as interest rates will be less informative. This notion is best explained with an example of a customer who is considering a long-term relationship with a seller. In deciding whether to enter a relationship, potential customers use a firm's current price as a signal of the prices it will charge in the future. When inflation causes relative prices to vary, it reduces the information about future prices in current prices. The same holds for the information about future prices of financial services, less information will make more competition less probable as there is less information to base decisions on in the pursuit for increasing market share. With regards to one of our indicators of competition, Huybens and Smith (1999) stress that “inflation exacerbates informational asymmetries and therefore leads to larger interest margins.” As measure of inflation we will use the yearly average of the monthly consumer price index (CPI). We will use two indicators to control for the competition coming from interindustry. To control for the impact of the degree of competition banks face from capital markets, we use the size of the country's stock market capitalization to GDP. As a proxy for
annual life insurance premiums collected divided by GDP. The more developed other parts of the financial sector are, the more competitive pressure there will be on the banking system. We will use both annual life insurance premiums and the total value traded in stock markets as a fraction of the GDP to adjust for the country size. (Claessens and Laeven, 2004)
Table 8 exhibits the control variables and sums up the relationship between the control variables and our financial liberalization index. The rest of table 8 will be explained in subsection III.3.
III.DATA
In the process of gathering our data and selecting the variables we chose to try and create the largest panel of data possible in order to come up with general conclusions regarding the impact of financial liberalization on bank competition globally. We started with a panel of 42 countries, but we excluded 8 countries because of too much missing data. We ended up with a panel of 34 countries and 12 years, thus 408 observations. The result of the exclusion of the 8 countries with an excess of missing data resulted in the fact that for our main explanatory variable, FLI, we have a balanced dataset and the data for our 3 indicators of competition and our control variables are near-complete.
[Table 1 goes here]
income countries by the World Bank. In this paper we define the high income countries as developed countries and the rest as developing countries.
III.1. Financial Liberalization
To begin with, we will explain how the measure of financial liberalization we will use is constructed. The financial liberalization index that has been developed by Abiad et al. (2004) is used. This index shows the degree to which a country has put into practice financial liberalization policies in seven different areas in a particular year. The seven areas are directed credit (i.e. reserve requirements), interest rate controls, entry barriers (i.e. pro-competition measures), banking supervision, privatization, international capital flows and security markets. In each area, a country is given a score on a graded scale, with zero corresponding to being fully repressed, one to partially repressed, two to largely liberalized and three to fully liberalized. Abiad et al. (2004) made policy changes represent shifts in a country's score on the liberalization scale in a certain year. In some cases, for instance when all state-owned banks are privatized at the same time, or when controls on all interest rates are simultaneously eliminated, policy changes will be embodied by jumps of more than one unit in that dimension. Reversals, for example the imposition of capital controls or interest rate controls, are represented as shifts from a higher to a lower score.
The financial liberalization index for a particular year is the summation of the seven scores in that year. The higher the index, the higher the amount of policy areas for which the government has carried out significant liberalizations. The maximum score is 21, which represents a fully liberalized country in all seven areas.
sm flow priv bs bar i dc
Equation 1 summarizes the construction of the financial liberalization index. Where
FLIabiadis the financial liberalization index used by Abiad et al. (2004), which is the
summation of dc, directed credit (i.e. credit controls), i, interest rate controls, bar, entry barriers, bs, banking supervision, priv, privatization, flow, international capital flows and, sm, security markets.
However, we argue that only 5 of the 7 scores are of interest in the case of our particular research. We believe that our financial liberalization index should be made up from scores based on measures which relate to banks and have an effect on competition. The directed credit score includes reserve requirements, which relate to banks and have an effect on competition due to the fact that banking conduct is restricted. The same reasoning holds for interest rate controls. The entry barriers measure directly relates to the structure conduct performance paradigm, as liberalization of entry barriers can lead to more banks and thus more competition. The privatization measure also has a clear effect on the creation of firms. In a privatized country certain monopolies exist in certain industries, whereas in non-privatized countries in these industries it will be more likely that two or more firms will compete for the market share. We also believe that for the international capital flows measure the structure conduct performance hypothesis applies, as in financially repressed countries in terms of the international capital flows less foreign competitors can compete.
Consequently, our financial liberalization index differs from the index used by Abiad et al. (2004) in that the banking supervision measure (bs) and the security markets
measure (sm) are left out. Equation 2 shows our financial liberalization index, FLI, and how it is made up from 5 scores.
[Table 2 goes here]
Table 2 shows the correlation matrix for the 5 subscores which jointly make up our financial liberalization index, all the variables are positively correlated. In order to see if the subscores tell a similar story about financial liberalization we will conduct a factor analysis.
[Table 3 goes here]
The principal component analysis output in table 3 displays that we can reduce our 5 scores into one factor as the outcomes are all larger than 0,6. The Cronbach alpha confirms this conclusion, as a Cronback alpha of 0,675 implies that the 5 subscores all have 1 common factor, which we will call financial liberalization.
As an example for the use of the financial liberalization index we will use Italy.
European countries have implemented several regulatory changes affecting the banking industry, motivated by the need to attain the level of harmonization necessary for the establishment of a single, competitive market for financial services. This process concluded in the early 1990’s with the implementation of the Second Banking
so-called Single Banking License. Preceding this initiative, cross-border expansions were subject to the authorization and successive control of the host country, in addition to capital requirements, as if the branch represented the establishment of a new bank. On the contrary, under the new system, banks from European Union countries are allowed to branch freely into other European Union countries. The new legislation removed substantial entry barriers and exposed national banking markets to potential new entrants. (Angelini and Cetorelli, 2000) Italy implemented the Second Banking Directive in 1993. In the Abiad et al. (2004) dataset the entry barriers score (i.e. pro-competition measures) doubled from 1992 to 1993, thus moving from partially repressed to largely liberalized.
[Table 4 goes here]
Table 4 shows the descriptive statistics for the FLI, FLIper, FLIHand FLIu variables. The
FLIper variable is similar to our standard FLI variable as exhibited in equation 2. Only in
this case instead of looking at years separately we made periods from the averages of 3 succeeding years. The FLIHand FLIu variables are also similar to the FLI variable, only
here FLIH represents only the high income countries, indicated by a H in table 1. The
FLIuvariable represents the other countries, with a lower income. There is no data
missing for any of the 4 variables. The use of the FLIper, FLIHand FLIu variables will be
explained in the methodology section.
income countries have lower financial liberalization scores on average, as exhibited by the lower mean for FLIuin comparison with the mean for FLIH in table 4.
One can easily observe from table 4 that there are some normality problems, as all three Jarque-Bera statistics are significant. In order to analyze if the fact that our data is not normally distributed has implications for our models we will also conduct tests in our analyses with only the 2nd en 3rd quartile.
[Figure 1 goes here]
In figure 1 one can see that over time almost all countries have become increasingly liberalized. The exceptions are Argentina and Malaysia, which show a small decrease over time, but remain largely liberalized.
[Table 5 goes here]
Table 5 confirms what one can observe in graph 1. Here we estimated a panel least squares model with FLI as dependent variable and the years as sole explanatory variable. The outcome shows a significant positive relationship between time and financial liberalization.
III.2. Competition
The data on bank concentration is from the Financial Development and Structure database from the Worldbank website1. The bank concentration measure that we will use is defined as the ratio of the sum of the three largest banks' assets in a country to the total banking sector assets in the same country, more commonly known as the C3-statistic. (Beck et al., 2000)
The data for the net interest margin and overhead costs are also from the Financial Development and Structure database. Beck et al. (2000) have calculated yearly averages over all banks for the 34 countries in our dataset. We will use the overhead costs
expressed as a fraction of total assets to adjust for the aggregate bank size.
[Table 6 goes here]
Although we assumed that all our three indicators of competition are inverse indicators of competition, table 3 shows that concentration is both negatively correlated with the net interest margin variable and the overhead costs variable.
The fact that concentration acts differently as an indicator does not come as a surprise. As discussed in subsection II.4., Ribon and Yosha (2001) found a change in competition while concentration remained unchanged. This is also the critique on using
concentration as proxy as explained in subsection II.5., concentration measures the actual market shares not including implications of the competitive behavior of banks. If we look at the relationship the other way around we can also see reasons why we could
1 Available from:
dispute concentration as a proxy for competition. Factors other than competition may drive concentration. For instance, regulatory initiatives to increase capital may spark off a wave of mergers that considerably increases the level of concentration in the industry. Moreover, a banking system with high entry barriers, in which a small number of institutions dominate the industry, can nevertheless be characterized by competition. Another problem could be the use of the C3-statistic as concentration measure, as the 3 largest banks could potentially not be the full story in terms of concentration. We will continue using the 3 indicators as a proxy for competition, while keeping in mind that especially concentration as measure of competition could very well be unsuitable.
[Table 7 goes here]
Table 7 shows the descriptive statistics for the three indicators of competition. Our first observation is that the concentration has a clear upper bound at 1. We will take this upper bound into account; this will be discussed in section IV. Furthermore, there is some missing data, but not of a magnitude that we expect any problems from this.
III.3. Control Variables
As for the control variables which we will use in our analysis we have used different sources for the data.
Table 8 summarizes the control variables we used and the data source. We gathered data from 2 different sources, the IMF’s International Financial Statistics database and from Beck et al.. (2000)
In the case of the CPI data we have used the logarithm, in order to adjust for the great variability of the data for this variable.
[Table 9 goes here]
Table 9 shows the descriptive statistics for the control variables. First of all, again there are some missing observations. The LOGCPI variable misses relatively most
observations, 3,5%. We consider this a small amount, and thus we do not foresee any problems due to missing data.
[Tables 10, 11 and 12 go here]
In table 10 we correlated the variable FLI with the control variables. The matrix shows that on the whole correlation between the exogenous variables is low, which means that multicollinearity problems are not severe or non-existent . In tables 11 and 12 we did the same, but replacing FLI with respectively FLIHand FLIu. Again, we find no signs of
multicollinearity problems.
IV.METHODOLOGY
squares model using our multi-country and multi-year sample of 34 countries and 12 years.
Compit = β0 + β1FLIit + β2Xit + εit (Equation 3)
Equation 3 shows the econometric specification of the model. In this model Comp is one of the 3 variables which measures bank competition in country i and time t. β0 is the
constant, β1 is the coefficient for the variable FLI, which is the financial liberalization
index in country i and time t. β2 is the coefficient for the vector of control variables, X in
country i and time t. Lastly, ε is the random error in country i and time t.
We will also conduct similar research using periods instead of separate years and an analysis where we look at developed and developing countries seperately.
Compit = β0 + β1FLIper it + β2Xit + εit (Equation 4)
Equation 4 shows the econometric specification of the model with FLIper, instead of FLI.
Here we make use of periods to increase the variation. For instance, financial liberalization measures could be taken less frequently than yearly resulting in succeeding years with no change.
Equations 5 and 6 show the econometric specification of the models where we use FLIH
and FLIu, which are respectively the financial liberalization indices for the developed
and developing countries. We will make this distinction because the countries' general economic development is expected to affect the relationship between banking
competition and financial liberalization, this is in reaction to Demirgüç-Kunt et al. (2004), they found that banking system structure indicators have a less close relationship with competitiveness indicators in more developed countries. Lastly, as mentioned before, we will not only test the hypothesis that financial liberalization has an effect on competition, because of the existence of two opposing theories we will also test whether this relationship is positive or negative.
V.DISCUSSION OF RESULTS
We will commence our analysis with an estimation of a panel least squares model with concentration as the dependent variable. Please note that we will use the Akaike info criterion (from hereon: AIC) as our main goodness of fit test, in order to see which estimated model best explains the data. The AIC is a commonly used tool for model selection. Given a data set, a number of competing models can be ranked in accordance with their AIC, with the one having the lowest AIC being the best model. The reason for using the AIC is that the software we use to estimate models does not give an adjusted R-squared output for Tobit models, which we will use as discussed hereafter. Furthermore, with all the models we estimate we will start with only the main
explanatory variable, FLI. If FLI is significant we will start adding and removing the control variables until we get the model with the best fit.
The second column in table 13 shows the first model we estimated, a panel least squares model without fixed effects. We find a significant negative relationship between FLI and concentration, however we find that the explanatory power of this model is rather low with an adjusted R-squared statistic of no more than 0,05.
In an attempt to attain a model with better explanatory power we had a closer look at the data. As partially discussed in subsection III.2. the concentration data has an upper and a lower limit, of respectively 0 and 1. Whereas conventional regression models fail to account for the difference between limit and non-limit observations on the dependent variable, a Tobit model is able to account for an upper and lower bound. The third column in table 13 shows the outcome for our estimation of a Tobit model. Again we find a significant negative relationship between FLI and concentration, but the lower AIC implies that the panel least squares model in the second column of table 13 has a better goodness of fit.
Moreover, we will use a fixed effects model to assess country-specific differences with respect to the relationship between financial liberalization and concentration. The fourth column in table 13 shows the outcomes of the estimation of the panel least squares model with fixed effects methodology. Both the adjusted R-squared statistic and the AIC shows that this model has a much better goodness of fit. Again we observe a significant negative relationship between FLI and concentration.
decreased foremost due to less data from the exclusion of the 1st and 4th quartile. However, we will use the contradicting AIC and adjusted R-squared statistic as an indication that the normality problems within the data do not have big enough implications for our emperical analysis of the relationship between FLI and
concentration. In other words, as the robustness of the model when using only the 2nd and 3rd quartile of our explanatory variable does not differ much from the model when using the full dataset, we can conclude that the normality problems are not severe in our analysis.
As for the control variables, the only significant addition which gave our best fitting model a better goodness of fit was LOGCPI. We found that LOGCPI has a positive relationship with bank concentration.
[Table 14 goes here]
In the same manner as we came to the best model with concentration as the dependent model we estimated several models with overhead costs as the dependent variable. As you can see in table 14 the panel least squares fixed-effects model again gives the best model. The outcomes in the fourth column, where we use only the second and third quartiles due to normality problems, are insignificant. In contrast to the use of concentration, with overhead costs as a proxy for competition we find a significant positive relationship with FLI. Also, the explanatory power of this model is rather high, with an adjusted R-squared statistic of 0,78.
[Table 15 goes here]
Next, we use the net interest margin as dependent variable. Table 15 shows
contradicting results. Whereas the model with only the second and third quirtile of the dependent variable shows a significant negative relationship, the use of all four quartiles leads to a significant positive relationship. Again the AIC and the adjusted R-squared statistic give contrary results in terms of the model with the best goodness of fit, which makes the outcomes for these models less interpretable.
As with both the overhead costs and concentration as measure of competition, LOGCPI is again added as significant positive control variable.
In summary, we found that there is a negative relationship between concentration and
FLI and a positive relationship between overhead costs and FLI. The relationship between FLI and the net interest margin is unclear. As all three of our measures of competition are inverse indicators of competition, these results implies that we found the opposite relationships between FLI and competition.
Subsequently, we looked at our data in a different way. As discussed in section IV we will have a look at developed and developing countries separately and at our full dataset in periods of 3 years. We will continue with panel least squares fixed-effects models, as these resulted in the best models in terms of goodness of fit in our previous analyses. Furthermore we will not continue with testing of the models with only the 2nd and 3rd quartile, as this has not led to significantly better fitting models.
Again we commenced with concentration as the proxy for competition. As one can see in the second and third column of table 16, respectively we did find a significant relationship for the developing countries and not for the developed countries. The significant relationship that we find for the developing countries is negative. Also, the both the AIC and the adjusted R-squared statistic show a better fitting model than when we used the full dataset in table 13.
The fourth column in table 16 exhibits the model with periods instead of years. We find similar results as with the original model with the full dataset, only this model is worse in terms of goodness of fit, as can be seen by a higher AIC and a lower adjusted R-squared statistic.
As for the control variables, in the best model in terms of goodness of fit for the developing countries we included LOGCPI and STOCK. LOGCPI again shows a significant positive relationship with concentration in developing countries, the same holds for STOCK. LOGCPI is also the only included control variable in the period model, again it is significant and positive.
[Table 17 goes here]
The fourth column in table 17 shows the results for the model we estimated with the use of 3 year periods. In this case the model shows similar results as the model with years instead of periods, as one can observe in table 14, only here the model has a slightly better fit.
Only in the case of the developing countries we could add significant control variables. Again LOGCPI shows a significant positive relationship and just as with the full dataset
INS is included and also significantly positive. As for the model with periods, LOGCPI is again significantly positive. Here INS is also added as control variable, with a
significant positive relationship with overhead costs.
[Table 18 goes here]
Lastly, we estimated the same model as in the cases of tables 16 and 17, but now with the net interest margin as dependent variable. In table 18 one can see a significant negative relationship in case of developed countries and a significant positive relationship in case of the developing countries. This could explain the change from significant positive to significant negative in the previous analysis with the net interest margin as dependent variable as exhibited in table 15. In the analysis in table 15 with only the 2nd and 3rd quartile the relationship became negative. It could have been that the 1st and 4th quartile included more data from developing countries, which had been deleted.
Again, in all three cases LOGCPI is a significant positive control variable. Only in the case of the developing countries INS is added, which is significantly positive.
Overall, we did find a significant relationship between financial liberalization and all 3 of our proxies for competition. However, the use of competition gives the opposite results that the net interest margin and overhead costs give. We argue that these contradicting results are mainly due to the fact that concentration is a weak measure of competition for reasons described in subsection III.2.
When we look at only overhead costs and the net interest margin we see a significant positive relationship. Since both are inverse indicators of competition, the results imply a negative relationship between financial liberalization and competition. In light of our theoretical framework, this means that our results support the efficient structure hypothesis.
Whereas the same analysis using periods instead of years did not add much to our findings, the distinction between developing and developed country did generate some interesting results. We find evidence that there is a stronger relationship between financial liberalization and competition in developing countries than in developed countries. We argue that the explanation for this is that in more developed countries financial markets are more developed and consequently have better availability of finance. Access to finance makes competition more intense, as required finance is an important barrier to entry for potential new entrants. Financial liberalization has positive consequences in terms of availability of finance, as the increased competition from financial liberalization leads to more developed financial markets. Table 4 exhibits that developed countries have been close to being fully liberalized throughout the years in our dataset, whereas developing countries made relatively large steps between 1991 and 2002. Consequently we argue that the developing countries also made larger steps in terms of the development of their financial markets and thus again in terms of
cause the stronger relationship between financial liberalization and competition in developing countries.
Lastly, the only control variable which was consistently significant and positive throughout almost every analysis was inflation, indicated by LOGCPI. The positive relationship has to be translated to the negative relationship with competition, again because we use inverse indicators of competition. The control variables for interindustry competition were seldom significant in any of our analyses.
VI.SUMMARY AND CONCLUSION
Using a data sample of 408 observations from 34 countries in the period 1991-2002 we investigated the effect of financial liberalization on bank competition. Using a panel least squares fixed-effects model we came to several models, seeing that we used three indicators of competition.
Our paper contributes to the literature on financial liberalization and competition in the sense that we present one of the very few multi-country panel data regression analyses. Moreover, we use three proxies for competition and use a recent financial liberalization index constructed by Abiad et al. (2004).
Our empirical analysis produced multiple results. As in other previous research we found concentration to be a weak measure of competition. Concentration as dependent variable gave us contradicting results with regard to the other two measures of
competition. From the outcomes of the analyses with these other measures we find a significant negative relationship between financial liberalization and competition, which is in support of the efficient structure hypothesis. Furthermore, we found that the
VII.REFERENCES
Abiad, A. G., Oomes, N. and Ueda, K. (2004) The Quality Effect: Does Financial Liberalization Improve the Allocation of Capital? IMF Working Paper, No. WP/04/112. Acemoglu, D., Johnson, S. and Robinson, J.A. (2001) The colonial origins of
comparative development: An empirical investigation, American Economic Review, Vol. 91, pp. 1369-1401.
Angelini, P. and Cetorelli, N. (2000) Bank Competition and Regulatory Reform: The Case of the Italian Banking Industry, Banca d’Italia, No. 380.
Barajas, A. and Steiner, R. (2000) The impact of liberalization and foreign investment in Colombia's financial sector, Journal of Development Economics, Vol. 63 No. 1, pp. 157-197.
Barth, J. R., Caprio, G. and Levine, R. (2001) The Regulation and Supervision of Banks Around the World: A New Database, University of Minnesota Financial Studies
Working Paper No. 0006; World Bank Policy Research Working Paper No. 2588. Beck, T., Demirgüç-Kunt, A. and Levine, R. (2000) A New Database on Financial Development and Structure, World Bank Economic Review, Vol. 14, pp. 597-605. Beck, T. (2008) Bank Competition and Financial Stability: Friends or Foes? The World
Bank Development Research Group, Policy Research Working Paper, No. 4656. Berger, A. N. and Hannan, T. H. (1989) The Price-Concentration Relationship in Banking, Review of Economics and Statistics, Vol. 71, pp. 291-299.
Bikker, J.A. and Haaf, K. (2000) Competition, Concentration and their relationship: an empirical analysis of the banking industry, Paper for the Financial Structure, Bank
Behavior and Monetary Policy in the EMU Conference, October 5-6, Groningen. Bikker, J.A. and Leuvensteijn, M. van (2007) Competition and efficiency in the Dutch life insurance industry, Applied Economics, DNB Working Paper No. 47, De
Nederlandsche Bank.
Bodenhorn, H. (1990) Entry, Rivalry and Free Banking in Antebellum America, The
Review of Economics and Statistics, Vol. 72, No. 4, pp. 682-686.
Cetorelli, N. (2004) Real Effects of Bank Competition, FRB of Chicago Working Paper, No. 2004-03.
Chen, X (2001) Financial Liberalization, Competition and Sound Banking: Theoretical and Empirical Essays, Dissertation at Virginia Polytechnic Institute and State
University.
Chizzolini, B. (2003) Deregulation in the Banking Sector and its Consequences on Credit to Small and Medium Firms: The Literature and the Case of Italy, Paper
presented at the meeting “Innovare per competere: come finanziare l’innovazione?” held in Novara, 22-23 May.
Claessens, S., Demirguc-Kunt, A. and Huizinga, H. (2001) How does foreign entry affect domestic banking markets, Journal of Banking and Finance, Vol. 25, No. 5, pp. 891-912.
Claessens, S., and Laeven, L. (2004) What Drives Bank Competition? Some
International Evidence, Journal of Money, Credit, and Banking, Vol. 36, pp. 563-82. Demirgüç-Kunt, A. and Detragiache, E. (1998) Financial Liberalization and Financial Fragility, Paper presented to the Annual World Bank Conference on Development
Demirgüç-Kunt, A., Laeven, L. and Levine, R., (2004) Market Structure, Institutions, and the Cost of Financial Intermediation, Journal of Money, Credit and Banking, Vol. 36, No. 3, Part 2: Bank Concentration and Competition: An Evolution in the Making, A
Conference Sponsored by the Federal Reserve Bank of Cleveland, May 21-23, 2003, pp. 593-622.
Demsetz, H. (1973) Industry Structure, Market Rivalry and Public Policy, Journal of
Law and Economics, Vol. 16, No. 1 pp. 1-9.
Denizer, C. (1997) The Effects of Financial Liberalization and New Bank Entry on Market Structure and Competition in Turkey, World Bank Policy Research, Working Paper No. 1839.
Ghosh, J. (2005) The Economic and Social Effects of Financial Liberalization: A Primer for Developing Countries, DESA Working Paper, No. 4.
Gual, Jordi (1999) Deregulation, Integration and Market Structure in European Banking,
Journal of the Japanese and Internation Economics, Vol. 13, pp. 372-396.
Hannan, T. H. (1991) Foundations of the Structure-Conduct-Performance Paradigm in Banking,
Journal of Money, Credit and Banking, Vol. 23, No. 1, pp. 68-84.
Heggestad, A. A. and Rhoades S. A. (1976) Concentration and Firm Stability in Commercial Banking, The Review of Economics and Statistics, Vol. 58, No. 4, pp. 443-452.
Hermes, N. and Lensink, R. (2008) Does financial liberalization influence saving, investment and economic growth? Evidence from 25 emerging market economies, 1973-1996, in: B. Guha-Khasnobis and G. Mavrotas (eds.), Financial development,
Huybens, E. and Smith, B. D. (1999) Inflation, financial markets, and long-run real activity, Journal of Monetary Economics, Vol. 43, pp. 283-315.
Keeley, M.C. (1990) Deposit insurance, risk and market power in banking, American
Economic Review, Vol. 80, pp. 1183-1200.
Laeven, L. (2003) Does financial liberalization reduce financial constraints? Financial
Management, Vol. 32, No. 1, pp. 5-35.
Marcus, A.J. (1984) Deregulation and bank financial policy, Journal of Banking and
Finance, Vol. 8, pp. 557-565.
Martin, S. (2001) Industrial Organization: A European Perspective, Oxford University
Press, New York.
Molyneux, P. and Forbes, W. (1995) Market Structure and Performance in European Banking, Applied Economics, Vol. 27, No. 2, pp. 155-59.
Ribon, S. and Yosha, O. (1999) Financial Liberalization and Competition in Banking: An Empirical Investigation, Bank of Israel, Discussion paper No. 99.05.
Stiroh, K. J. and Strahan, P. E. (2002) Competitive Dynamics of Deregulation:
Evidence from U.S. Banking, Journal of Money, Credit and Banking, Vol. 35 No. 5, pp. 801-28.
Winston, Clifford (1998) U.S. Industry Adjustment to Economic Deregulation, Journal
VIII.APPENDIX VIII.1. Tables Argentina AustraliaH Brazil CanadaH Chile Colombia Costa Rica Ecuador Egypt FranceH GermanyH Guatemala India Indonesia IsraelH ItalyH JapanH KoreaH Malaysia Mexico Morocco Pakistan Peru Philippines SingaporeH South Africa Sri Lanka SwedenH Thailand Turkey UKH USAH Uruguay Venezuela
Table 1: Countries used in analysis. (H =High Income country according to World Bank.)
BS P ICF DC IR BS 1,000 P 0,343 1,000 ICF 0,460 0,441 1,000 DC 0,466 0,206 0,326 1,000 IR 0,183 0,331 0,366 0,312 1,000
Table 2: Correlation matrix for subscores of financial liberalization..
Components 1 d 0,636 i 0,696 b 0,660 p 0,618 f 0,769
Cronbach's Alpha N of Items
0,675 5
FLI FLIper FLIH FLIu Number of Countries 34 34 12 22 Mean 11,262 11,262 12,972 10,329 Median 12,000 11,666 13,000 11,000 Maximum 15,000 15,000 15,000 15,000 Minimum 1,000 2,333 7,000 1,000 Std. Dev. 2,949 2,877 1,873 3,013 Skewness -0,836 -0,778 -0,997 -0,581 Kurtosis 3,295 3,205 3,917 2,892 Jarque-Bera 48,956 13,974 28,902 15,021 Probability 0,000 0,000 0,000 0,001 Observations 408 136 144 264
Table 4: Descriptive Statistics for Financial Liberalization Indices.
FLI C (Prob.) -595,859 (0,000) YEAR (Prob.) 0,304 (0,000) N 408 Adj. R2 0,126 F-statistic 59,048 AIC 4,872
Table 5: OLS with FLI as dependent variable and years as explanatory variable.
CONC INTMAR OVERH CONC 1,000
INTMAR -0,119 1,000
OVERH -0,070 0,765 1,000
CONC OVERH INTMAR Mean 0,584 0,045 0,052 Median 0,564 0,036 0,040 Maximum 1,000 0,180 0,235 Minimum 0,200 0,003 0,010 Std. Dev. 0,199 0,028 0,037 Skewness 0,280 1,474 1,856 Kurtosis 2,290 5,319 6,631 Jarque-Bera 13,494 231,121 433,971 Probability 0,001 0,000 0,000 Observations 396 394 386
Table 7: Descriptive Statistics for indicators of competition.
Variable Theoretical Relationship
with Competition
Data Source
Inflation (CPI) Negative International Financial
Statistics (IMF) Non-Life Insurance
Premium Volume/ GDP
Positive Beck et al., 2000
Stock Market Total Value Traded/ GDP
Positive Beck et al., 2000
Table 8: Summary of variables used and their source.
LOGCPI INS STOCK
Mean 0,809 0,025 0,304 Median 0,783 0,011 0,142 Maximum 3,470 0,150 3,262 Minimum -1,732 0,001 5,40E-05 Std. Dev. 0,605 0,030 0,437 Skewness 0,532 1,688 3,000 Kurtosis 5,749 5,405 14,870 Jarque-Bera 142,776 283,599 2875,041 Probability 0,000 0,000 0,000 Observations 394 396 390
FLI LOGCPI INS STOCK FLI 1,000
LOGCPI -0,410 1,000
INS 0,406 -0,351 1,000 STOCK 0,282 -0,354 0,411 1,000
Table 10: Correlation matrix for FLI and control variables.
FLIH LOGCPI INS STOCK
FLIH 1,000
LOGCPI -0,321 1,000
INS 0,266 -0,097 1,000 STOCK 0,247 -0,184 0,212 1,000
Table 11: Correlation matrix for FLIH and control variables.
FLIu LOGCPI INS STOCK
FLIu 1,000
LOGCPI -0,248 1,000
INS 0,257 -0,115 1,000 STOCK 0,070 -0,184 0,255 1,000
OLS TOBIT FIXED FIXED23 C (Prob.) 0,184 (0,131) 0,161 (0,194) 0,835 (0,000) 0,730 (0,000) FLI (Prob.) -0,009 (0,023) -0,010 (0,016) -0,025 (0,000) -0,013 (0,000) LOGCPI (Prob.) 0,065 (0,000) 0,073 (0,000) 0,046 (0,005) INS (Prob.) 1,003 (0,005) STOCK (Prob.) N 382 371 382 198 Adj. R2 0,054 0,690 0,501 F-statistic 8,325 25,341 7,834 AIC -0,455 -0,415 -1,494 -2,773
Table 13: Dependent variable: concentration
OLS FIXED FIXED23
C (Prob.) -0,008 (0,192) 0,009 (0,187) 0,033 (0,000) FLI (Prob.) 0,003 (0,000) 0,0014 (0,009) 0,001 (0,157) LOGCPI (Prob.) 0,030 (0,0000) 0,014 (0,000) INS (Prob.) -0,016 (0,000) 0,317 (0,000) STOCK (Prob.) -0,006 (0,029) N 357 369 198 Adj. R2 0,430 0,781 0,649 F-statistic 68,295 37,579 15,009 AIC -4,825 -5,693 -7,513
Table 14: Dependent variable: overhead costs
OLS FIXED FIXED23 C (Prob.) -0,006 (0,425) 0,228 (0,016) 0,049 (0,000) FLI (Prob.) 0,003 (0,000) 0,002 (0,000) -0,001 (0,010) LOGCPI (Prob.) 0,040 (0,000) 0,014 (0,000) INS (Prob.) -0,329 (0,000) STOCK (Prob.) N 362 372 194 Adj. R2 0,457 0,854 0,777 F-statistic 102,361 61,136 25,961 AIC -4,336 -5,567 -7,720
Table 15: Dependent variable: net interest margin
FIXED(FLIu) FIXED(FLIH) FIXED(FLIper)
C (Prob.) 0,766 (0,000) 0,758 (0,000) 0,729 (0,000) FLI (Prob.) -0,025 (0,000) -0,014 (0,119) -0,020 (0,0257) LOGCPI (Prob.) 0,062 (0,000) 0,103 (0,0129) INS(Prob.) STOCK (Prob.) 0,096 (0,005) N 236 141 123 Adj. R2 0,753 0,726 0,658 F-statistic 30,801 31,947 7,715 AIC -1,853 -1,443 -1,353
FIXED(FLIu) FIXED(FLIH) FIXED(FLIper) C (Prob.) 0,013 (0,111) 0,005 (0,707) -0,008 (0,539) FLI (Prob.) 0,002 (0,009) 0,002 (0,050) 0,003 (0,010) LOGCPI (Prob.) 0,018 (0,000) 0,020 (0,000) INS (Prob.) 0,353 (0,003) 0,334 (0,041) STOCK (Prob.) N 238 141 121 Adj. R2 0,805 0,236 0,827 F-statistic 41,650 4,613 16,989 AIC -5,640 -5,872 -5,875
Table 17 Dependent variable: overhead costs
FIXED(FLIu) FIXED(FLIH) FIXED(FLIper)
C (Prob.) 0,355 (0,018) 0,037 (0,000) 0,012 (0,294) FLI (Prob.) 0,004 (0,000) -0,001 (0,0176) 0,003 (0,002) LOGCPI (Prob.) 0,015 (0,000) 0,004 (0,001) 0,015 (0,000) INS (Prob.) 0,267 (0,081) STOCK (Prob.) N 232 130 123 Adj. R2 0,813 0,788 0,920 F-statistic 41,293 37,956 41,311 AIC -5,164 -8,241 -6,115
Adj. R2 Adjusted R-squared AIC Akaike info criterion bar Entry barriers score bs Banking supervision score CONC Concentration - C3 statistic dc Directed credit score
FLI Financial liberalization index
FLIH Financial liberalization index for developed countries
FLIper Financial liberalization index for 3 year periods
FLIu Financial liberalization index for developing countries
flow International capital flows score i Interest rate controls score
INS Annual life insurance premiums as a fraction of GDP INTMAR Net interest margin
LOGCPI Logarithm of consumer price index LOGGDP Logarithm of real GPD per capita OVERH Overhead costs
priv Privatization score sm Security markets score STOCK Stock market capitalization
x23 Variable x without 1st and 4th quartile
VIII.2. Graphs 6 8 10 12 14 199219941996 199820002002 1 1 0 1 2 1 4 1 6 1992199419961998 20002002 2 2 4 6 8 10 1992199419961998 20002002 3 1 2 1 3 1 4 1 5 1 6 19921994199619982000 2002 4 11 12 13 14 15 1992 199419961998 20002002 5 8 9 1 0 1 1 1 2 1 3 1992 19941996199820002002 6 4 6 8 10 199219941996 199820002002 7 2 4 6 8 1 0 1 2 1 4 1992199419961998 20002002 8 4 6 8 10 12 1992199419961998 20002002 9 1 1 1 2 1 3 1 4 1 5 1 6 19921994199619982000 2002 10 11 12 13 14 1992 199419961998 20002002 11 6 8 1 0 1 2 1 4 1992 19941996199820002002 12 0 2 4 6 8 199219941996 199820002002 13 8 9 1 0 1 1 1992199419961998 20002002 14 9 10 11 12 13 14 1992199419961998 20002002 15 6 8 1 0 1 2 1 4 1 6 19921994199619982000 2002 16 11 12 13 14 1992 199419961998 20002002 17 1 0 1 1 1 2 1 3 1 4 1992 19941996199820002002 18 8 10 12 14 199219941996 199820002002 19 1 2 1 3 1 4 1 5 1 6 1992199419961998 20002002 20 0 4 8 12 1992199419961998 20002002 21 2 4 6 8 1 0 19921994199619982000 2002 22 6 8 10 12 14 16 1992 199419961998 20002002 23 8 1 0 1 2 1 4 1 6 1992 19941996199820002002 24 12 13 14 15 16 199219941996 199820002002 25 8 1 0 1 2 1 4 1 6 1992199419961998 20002002 26 4 6 8 10 12 1992199419961998 20002002 27 6 8 1 0 1 2 19921994199619982000 2002 28 6 8 10 12 14 1992 199419961998 20002002 29 9 1 0 1 1 1 2 1 3 1992 19941996199820002002 30 14 15 16 199219941996 199820002002 31 1 2 1 3 1 4 1 5 1 6 1992199419961998 20002002 32 8 10 12 14 1992199419961998 20002002 33 4 8 1 2 1 6 19921994199619982000 2002 34 FLI
