Competition in the banking industry

The banking system of a country plays a vital role in social welfare of the people in the country and of people of the world in general. It offers services to enterprises and consumers to undertake their business activities and to easily perform their day-to-day transactions. It is necessary to ensure an efficient functioning of the banking system; otherwise, a dull and bogus banking system brings about an ultimate threat of potential for financial instability. That is the reason why the competition in financial sector is of much importance. The importance is for many reasons; i.e. it relates to the efficiency, quality and innovation of the production of financial services. Most importantly, it helps in taking careful decisions in policy making for banks (Claessens and Laevens, 2003).

In recent years, a lot of research work has been carried out, investigating the nature of competition in the banking industry along with the degree of competition, factors affecting the competition and the effects of competition on other market factors on micro level as well as on macro economic level.

An explanation for the vast amount of studies on this topic is that competition can not be measured directly due to the lack of detailed information on prices and costs of the various banking products (Bikker et al., 2007). This topic has also gained popularity among bankers, economists and policy makers because of globalisation, liberalization of financial markets and banking harmonization all over the world, especially in the European Union.

Since early 90’s, there are a lot of regulatory changes observed in the banking industry in order to achieve the establishment of a single, competitive market in the financial sector of Europe. It was initially triggered with the implementation of the Second Banking Coordination Directive defining conditions for Single Banking License.

As a consequence, entry barriers have been removed substantially for the new entrants increasing competition, coupled with a significant consolidation process. The intuition behind this was Market Contestability; a market is contestable if there are no barriers to entry, exit is absolutely costless and the prices are highly elastic to demands for industry output. The key idea is that a firm may be compelled to be more competitive and efficient by the prospect of new entrants (Allen and Engert, 2007). Furthermore, costless exit means that if a firm enters into a new market and then decides to withdraw, it is required to recover sunk entry costs. These features insure that even if a market has a small number of active firms, it is still effectively contestable and competitive (Nathan A. and Neave E., 1989).

Moreover, the pro-competitive deregulation process has increased the level of competition (Cetorelli, 2004), particularly in non-traditional and non-interest bearing areas of banking activity (Goddard et al. 2001).

Trivieri F. (2005) documents that in the course of the 1990s, the Italian banking system underwent profound changes at normative and institutional levels, which led – among other things – to a significant relaxation of the entry barriers, to the liberalisation of bank branching, to the redefinition of ownership structure and to a large number of mergers and acquisitions.

The effects of these transformations and, in particular, of those linked to the process of consolidation have been studied by many authors (see, among others: Resti, 1997; Angelini and Cetorelli, 2000; Messori, 2001; Sapienza, 2002; Focarelli et al., 2002; Focarelli and Panetta, 2003).

According to European Central Bank 1999, 29 percent banks had been merged or shrunk between 1985 and 1997. In Italian banking industry, the Second Banking Directive was implemented in 1993, followed by a 20 percent reduction in the number of banks as a result of consolidation. It is observed that competition has been increased in recent years in European banking markets which is also generally true for Italy.

Angelini and Cetorelli (2000) cite that a rise in the competition is easily found in European banking markets during recent years. Danthine, Giavazzi, Vives and von Thadden (1999) report a somewhat generalized decrease in banks’ net interest margins across Europe during the 1990s. Consistent with the European evidence, a declining trend in bank margins is also observed across different markets in Italy.

This paper focuses only on the banking industry of Italy and analyzes the evaluation of competitive conditions, nature and the degree of competition in the Italian banking industry using firm-level balance sheet data.

In this paper, we explore more thoroughly the competitive nature and degree of competition in the Italian banking industry by adopting a methodology developed in empirical industrial organization and used extensively in banking. Further more, we will compare our results with previous results to find out that whether the degree of competition has been increased or it has been as same as it was in the past.

The setup of the remainder of this paper is as follows. Section 2 contains some important information about structure and features of a competitive banking industry which helps in understanding the competition more thoroughly. Next Section 3 introduces the original Panzar-Rosse model along with the previous studies in the field. Section 4 gives a brief explanation of the general Panzar and Rosse model. This section also shows the interpretation of the H-statistic along with the description of the testing hypothesis. Following Section 5 deals with the empirical model used in this study including long-run equilibrium test. This section also contains the banks’ data used for the empirical illustration for our theoretical findings. Finally in the last Section 6 empirical results and conclusion is discussed.

OPTIMAL COMPETITIVE STRUCTURE OF THE BANKING SYSTEM

According to Northcott C. (2004), competition improves efficiency and growth in the banking sector but market power or concentration is necessary for stability in the industry. Moreover, competitive environment promotes productive and allocative efficiency leading towards economies of scale while market power improves credit availability, stability, quality of banks’ loan portfolios, screening of loans and monitoring them.

As a result, market power should not be eliminated, but rather used to facilitate an environment that promotes competitive behaviour.

FEATURES OF A COMPETITIVE BANKING INDUSTRY

Concentration weakens competition by fostering collusive behaviour among firms. Increased market concentration was found to be associated with higher prices and greater than normal profits (Bain, 1951). Smirlock (1985) and Evanoff and Fortier (1988) argue that higher profits in concentrated markets could be the result of greater productive efficiency. Berger (1995) finds some evidence that the efficiency hypothesis holds in US banking. In Europe, on the other hand, structural factors appeared to be more important and the SCP hypothesis seemed to hold (Goddard et al., 2001).

If a well-developed financial system is provided then contestability improves with new entrants. Contestability is not necessarily related to concentration or the number of banks. Concentration and competition can exist together because of the presence of asymmetric information and branches and the effect and use of new technologies. (Northcott C, 2004)

LITERATURE REVIEW AND THEORETICAL ISSUES:

According to Bikker and Haaf (2000), initially the economic literature on the issue of competition in the industrial sector can be divided into two main categories; structural approach and non-structural approach. Structural approach can be further divided into two main paradigms.

First type of structural approach is Structure-Conduct-Performance (SCP) paradigm, which tells us that the degree of competition is determined by the structural characteristics of the market, such as, number of firms, size of the firms, etc. The SCP was developed in the early 1950’s by Mason (1939) and Bain (1951). Bain (1951) constructs the market power hypothesis that collusive behaviour is initiated by high concentration which results in large profits for firms. Later, Stigler (1964) and Demsetz propose efficiency hypothesis in contrast of marker power hypothesis stating that the efficiency of bigger firms may be the reason for high concentration instead of collusive behaviour of firms, while during 1980’s, Baumol, Panzar and Willig (1983) build contestability hypothesis. Their hypothesis states that if entry and exit barriers are relaxed then competition may be prevailed (Mkrtchyan A. 2005).

Second approach is Efficient-Structure-Hypothesis (ESH), which states that greater concentration in the industry not only increases the level of efficiency in the sector but also increases the degree of competition in that sector.

Non-structural approach is based on describing the nature of competition in the context of the studies of New Economic Industrial Organization (NIEO). It suggests non-structural models to analyse the competition in markets which do not rely on the markets’ structure. Particularly, Klein (1971), Baumol, Panzar, and Willig [1982] provide a theory that shows that market competitiveness can be inferred irrespective of the structure of the market.

NIEO studies include Iwata Model (1974), Brasnahan Model (1982), Rosse and Panzar (1977), Panzar and Rosse (1982), Panzar and Rosse Model (1987), etc. Non-structural method or firm’s input-output cost studies have gained more popularity than the structural approach among academics, researchers, analysts and policy makers. Particularly Panzar and Rosse model (1987) is the most widely used and is very popular model for competition.

Duncan (2003) mentions that the Panzar and Rosse (P-R) model provides a comprehensive and simple method to calculate the competition. It does not require intensive data as compared to other models and has been firmly related to theoretical side. The information required for this model is easily available as it calculates the sum of the factor prices’ elasticities estimated from a reduced form of revenue function.

The Rosse-Panzar test has been developed to examine competitive conditions in the light of the contestability theory (Rosse and Panzar, 1977; 1982; 1987). This approach measures the degree of competition by analyzing how each bank’s revenues react to changes in input prices. It has primarily emerged to test market conditions that encompass all spectrums of competitiveness away from the restrictions brought about by the structural concepts. Basically, it depends on the relationship between gross revenues of the firm and the change in its input prices by using a statistic – which is called the H-statistics – that measures the sum of elasticities of total revenue with respect to each input price. As this approach includes the revenue equation so for banks, mainly the revenues are interest revenue. In this approach, h-statistics is used to measure the degree of competition. The H-statistics will tell us the responsiveness of revenues to the changes in input prices. If h-statistics is less than or equal to zero then there will be monopoly, if it is between zero and one then there will be monopolistic competition and if it is equal to one then there will be perfect competition (Greenberg J. and Simbanegavi W.).

This approach is preferred when testing the data of different individual banks. Moreover, P-R approach yields similar results without any ambiguity as it has clearly defined hypotheses with specific interpretations.

PREVIOUS GENERAL STUDIES ABOUT BANK COMPETITION:

Rearrange the literature review according to the claessens and neave.

A great number of papers have been written on investigating competition in the banking industry using Panzar and Rosse model (1987). But the motivations for analyzing the nature of the competition are vastly varied like contribution of institutional and structural factors, growth, regions, stability, financing, efficiency, contestability, consolidation, cross-border capital flows, risks etc.

The summary of the previous works and their findings can be seen in the Appendix Table 1.

Panzar J. and Rosse J. (1987) develop test for Monopoly and use linear regression model to estimate the H-statistic for the newspaper industry, reporting that it is vague to conclude that the newspaper firms earn oligopoly profits. Looking at the cross-country studies carried out in the EU banking markets, one of the earliest analysis is undertaken by Molyneux et al. (1994) who test the Panzar-Rosse statistics on a sample of banks in France, Germany, Italy, Spain and the UK for the period 1986-89. Results indicate monopolistic competition in all countries except Italy where the monopoly hypothesis can not be rejected.

Shaffer and Disalvo (1994) use this test to analyze the data of a duopoly banking market in south central Pennsylvania to exercise the procedure for concentration and competitive conduct.

Waleed Murjan and Cristina Ruza (2002) examine the Arab Middle Eastern banking markets with this test concluding that the banking sector is more competitive in non-oil-producing countries than the banking industry in oil-producing countries.

Gelos and Roldos (2002) apply this method on 8 different countries of Latin America and Europe, finding that market contestability prevents the competitive pressure from declining which can happen because of the consolidation while Claessens and Laeven (2003) process the data of 50 countries obtaining the same results.

Bikker and Haaf (2002) assess the banking industry in 17 European countries and six countries that are outside of Europe comparing competitive conditions and market structure.

Goddard, J. and Wilson, J. (2006) report misspecification bias in the revenue equation for the banking sectors of 19 developed and developing countries. They suggested a dynamic revenue equation for unbiased estimation rather than fixed effects estimation which is severely biased towards zero.

Gilbert (1984) and Berger (1995) test the data for 8,235 banks in 23 developed nations producing the results that a higher degree of market power has less risk exposure.

Yuan Y. (2005) assesses the competition in Chinese Banking sector and comes up with the results that China already has had perfectly competitive condition before new foreign entrants and it still has the same situation.

Duncan D. (2003) presents the empirical assessment of the market structure of the Jamaican banking sector and competitive trends in the market finding monopolistic behaviour.

Al-Muharrami S. et al. (2006) take GCC Arab countries into observation and suggest that Kuwait, Saudi Arabia and the UAE operate under perfect competition; and Bahrain and Qatar operate under conditions of monopolistic competition.

Nathan A. and Neave E. (1989) exercise the test on Canadian financial industry and reject the hypothesis of monopoly power in Canada’s financial system.

PREVIOUS STUDIES ABOUT COMPETITION IN ITALIAN BANKING INDUSTRY:

A great number of studies on competition in financial sector of EU countries have been reported which also include Italy in general. But there are also some research-papers which are produced specifically for Italy. Some of them are:

Cetorelli N. and Angelini P. (2000) study the case of the Italian banking industry and cite that competitive conditions have improved substantially after 1992, and it is believed that the introduction of the Single Banking License in 1993 also helps fostering the competitive behaviour in Italian banking industry.

Dell’Ariccia G. and Bonaccorsi E. (2003) investigate the relationship between bank competition and firm creation. They document that the effects of competition in the banking sector on the creation of firms in the non-financial sector are less favourable to the emergence of new firms in industries where information asymmetries are greater.

Coccorese P. (2002) rejects the theory that competition can be easily reduced by the collusive behaviour of the firms, and comes up with the conclusion that strong concentration does not necessarily prevent competition among firms.

Trivieri F. (2005) compares the banks involved in the cross-ownership and banks that are not involved. He finds that Italian banks involved in cross-ownership are less competitive than the banks which are not involved in cross-ownership, hence proving cross-ownership decreases competition.

GENERALIZED PANZAR AND ROSSE (1987) APPROACH:

P-R model assumptions:

Firstly, there are some assumptions and conditions in which Panzar and Rosse model works. The model supposes that banks operate in long run equilibrium. Although Goddard Wilson (2006), documents that this condition is not needed any more if a correctly specified dynamic revenue equation is adopted which permits virtually unbiased estimation of the H-statistic. This eliminates the need for a market equilibrium assumption, but incorporates instantaneous adjustments as a special case. So in this paper long run equilibrium postulate holds. Another assumption is that the market participants affect the performance of the banks by their actions. Another postulate is that the price elasticity of demand is greater than unity. Moreover, the model posits that there is a homogenous cost structure. Furthermore, profits are maximised to obtain the equilibrium number of banks and the equilibrium output. In long rum equilibrium, it is known that banks maximise their profits when, marginal revenue equals to marginal cost (Bikker and Haaf, 2000). Trivieri F. (2005) also adds that the banks are treated as single product firms which mainly provide intermediation services.

EXPLANATION OF PR MODEL:

Claessens and Laeven (2003) cite that the Panzar and Rosse model studies the impact of changes in factor input prices reflected in equilibrium revenues by a specific bank.

Bikker and Haaf (2000) write that Panzar and Rosse model gives simple models for oligopolistic, competitive and monopolistic markets. This test works on the reduced form revenue equation and uses H-statistics. This H-statistics can tell us not only the nature of competition but also gives information about the degree of the competition. H-statistics if measures between 0 and 1, it is monopolistic competition, 0 is considered as monopoly and 1 as perfect competition. Here, a general banking market model is used, which determines equilibrium output and the equilibrium number of banks by maximising profits. The model is also able to allow for bank-specific variables in the equation.

According to Bikker and Haaf (2000), in the long run equilibrium, it is known that banks maximise their profits at the break-even point. The break-even point is where marginal revenue equals marginal cost. So, the bank i maximises its profits, where marginal revenue equals marginal cost:

(1)

Ri refers to revenues and Ci to costs of bank i (the prime denoting marginal), xi is the output of bank i, n is the number of banks, wi is a vector of m factor input prices of bank i, zi is a vector of exogenous variables that shift the bank’s revenue function, ti is a vector of exogenous variables that shift the bank’s cost function. Secondly, it means that in equilibrium at the market level, the zero profit constraint holds (Bikker and Haaf, 2000):

(2)

Variables marked with an asterisk (*) represent equilibrium values. Panzar and Rosse define a measure of competition H as the sum of the elasticities of the reduced-form revenues with respect to factor prices (Bikker and Haaf, 2000):

(3)

According to Khan, M. (2009), it measures the percentage change in (equilibrium) revenue due to a one percent change in all input factor prices (change in cost). From duality theory, it is known that one percent increase in factor prices will lead to one percent upward shift in cost function. The impact of this shift in cost function on the (equilibrium) revenue of the banks is directly related to the degree of competition in the banking sector.

Bikker and Haaf (2000) further explain that Panzar and Rosse prove that under monopoly or under perfectly collusive oligopoly, an increase in input prices will increase marginal costs, reduce equilibrium output and subsequently reduce revenues; hence H will be zero or negative. An increase in input prices raises both marginal and average costs by an equal proportion as the cost is homogeneous of degree one in input prices without altering the optimal output of any individual firm. Exit of some firms increases the demand faced by each of the remaining firms, thereby leading to an increase in prices and total revenues by as same amount as the rise in costs, resulting perfect competition where H-statistic is positive but not greater than unity. In this case marginal and average cost will be increased by the rise in input prices (Nathan A. and Neave H., 1989).

INTERPRETATION OF H-STATISTICS:

Panzar and Rosse prove that, under monopolistic competition, H is between zero and unity. H is a decreasing function of the perceived demand elasticity, so H increases with the competitiveness of the banking industry. As a result, this H-statistic can serve as a continuous interpretation of the competitiveness. Although this is not mentioned by Panzar and Rosse (1987) but with some assumptions this continuous interpretation is correct. So, the testable hypotheses are: The banking industry is characterised by monopoly for H=0, monopolistic competition for 0<H<1, and perfect competition for H=1 (Bikker and Haaf, 2000). The brief illustration of H-statistic can be found in the Appendix Table 2.

HYPOTHESIS TESTING;

Khan, M. (2009) mentions:

Two-sided Perfect Competition Test:

Maintaining the long run equilibrium postulate, if banks are operating under perfect competition, a one percent change in cost will lead to a one percent change in revenues. Output will not be changed if the demand function is perfectly elastic under perfect competition, output price and cost both will increase by the same extent. This implies that under perfect competition, H-statistic will be equal to one. Statistically, we will test the following hypothesis.

H0 : H = 1 Perfect competition prevails in the banking sector.

H1 : H ≠ 1 There is no perfect competition in the banking sector.

Two-sided Monopolistic Competition Test:

If banks are operating in monopolistically competitive environment, one percent increase in cost will lead to less than one percent increase in revenue as the bank faces fairly inelastic demand function. Statistically, we will test the following hypothesis.

H0 : 0 < H < 1 Banks are operating in a monopolistic competition environment.

H1 : H ≤ 0 or H ≥ 1 Banks are not operating in a monopolistic competition environment.

One-sided Monopoly Test:

Standard theory of market structure suggests that the sum of factor input price elasticities should be less than zero if the underlying market structure is monopoly. Statistically, we will test the following hypothesis.

H0 : H ≤ 0 Banks are operating in a monopoly condition.

H1 : H > 0 Banks are not operating in a monopoly condition.

(Khan M., 2009)

EMPIRICAL FRAMEWORK AND METHODOLOGY:

The test is robust with any definition of market whether it is within the national boundaries or it is the global international banking industry because there is no need to specify a geographic market. Before testing, it is commonly necessary to obtain a reduced form of revenue equation which consists of revenue as a dependent variable, factor input prices as independent variables and some controlled or firm’s specific factors. The basic equation is:

Total interest revenue = total cost + controlled variables + error term

The panel data is used in the paper which is the data collected over multiple time periods. It is the combination of cross-sectional and time series dimensions. Hence, it can be derived as:

Ci = a + Byi + Ei (4)

Ct = a + Byt + Et (5)

Where, C is the dependent variable, a is constant term, B is the coefficient of the independent term, y is the independent variable and E is the error term. Combining both the equations (4) and (5), the final basic equation can be given as:

Cit = a + Byit + Eit (6)

But Panzar and Rosse define the H as the sum of the elasticities of the reduced-form revenues with respect to factor prices, so the econometric model of the Panzar and Rosse statistic may be represented by the following equation:

(7)

For i = 1,…..I; t = 1,……T;

Where, R is a measure of gross revenue. W is a vector of factor prices (the H statistic is given by the sum of the estimated coefficients of the variables in this vector); S is a vector of scale variables; X is a vector of exogenous and bank-specific variables that may shift the cost and revenue schedule, ε indicates the error term; I is the total number of banks; T is the number of periods observed (Trivieri, 2005). To calculate the sum of elasiticities, it is necessary to estimate the log linear model instead of estimating a simple linear model that is the reason for taking the log of all the variables in equation (7).

The sign of the variables of different costs and bank specific variables are positive showing a direct relationship to revenues (Trivieri, 2005).

In this pooled regression, extra intercepts or ‘dummies’ for time are used, but dummies for individuals are not included because of the application of within-group-estimators. Because with-in-group estimator takes first difference and removes the individuals dummies variables by itself. Thus being a fixed effects model, it measures differences in intercepts for each group and the differences are calculated by a separate dummy variable for each group (Trivieri, 2005).

The use of fixed effects panel regression with time dummies allows calculating the relevant parameters of the empirical model. Furthermore, unobserved heterogeneity is controlled by the fixed effects too avoiding omitted variable problems (Trivieri, 2005).

In this paper, the intermediation approach developed by Sealey and Lindley (1977), is followed which tells that deposits, labour and capital are inputs for the banks. The empirical model applied in this paper is as:

LGIRTA =

B1LLABCOST + B2LCAPCOST + B3LFUNDCOST + B4LLTA + B5LBMIX

(8)

Where,

LGIRTA = Log of Gross Interest Revenues over Total Assets

LLABCOST = Log of Labour factor price

LCAPCOST = Log of Capital Cost

LFUNDCOST = Log of Funding Cost

LLTA = Log of Loans to Total Assets

LBMIX = Log of Loans to Banks and Clients over Total Loans

This paper addresses the banking industry of Italy. The data includes 480 banks approximately, of all sizes in Italy. The data contains two different samples. First sample consists of the data from 1995 to 1997, total 3 years, and the second sample contains data from 1997 to 2000, total 3 years. We make a comparison and inference between the results obtained by these two samples through our empirical model and find out the competitive behaviour of Italian financial market.

LONG RUN EQUILIBRIUM TEST:

An important underlying condition of the H-statistic for competition is the long run equilibrium. Panzar and Rosse (1987) cite that this postulate is crucial for the cases of perfect competition and monopolistic competition. Though, it is not a fundamental assumption in the case of monopoly because when H is less than or equal to zero then it is a long run assumption for monopoly (Trivieri, 2005).

Long run equilibrium test for the observations can be done with the prerequisite that: “competitive markets equalise the return rates across firms, so that in equilibrium these rates should not be correlated with input prices” (Trivieri, 2005).

In our empirical model as in Shaffer (1982), this test can be carried out by re-estimating the equation with the proxy for the return on assets, ROA, as dependent variable in the calculation of H. In this context, H = 0 implies that the data are in long run equilibrium (Trivieri, 2005). The intuition behind this theory is that, return on assets, ROA, should not be related to input prices.

De Bandt and Davis (1999), define the equilibrium condition as the state in which changes in banking sector are considered as gradual, long run equilibrium for the observations does not mean that competitive conditions remain the same and do not change through out the period of observations (Trivieri, 2005).

Although it is inappropriate to use Rosse-Panzar test which is based on a static equilibrium framework, but in the real financial market, the equilibrium adjustments are less than instantaneous, resulting disequilibrium on some points in time or frequently, or always. Moreover, when it is known that the adjustments towards equilibrium are partial and not instantaneous then using fixed effects estimation for the static revenue equation will result in biased H-statistics toward zero (Goddard J. and Wilson J., 2006).

For the long run equilibrium, we estimate the following equation:

LROA =

B1LLABCOST + B2LCAPCOST + B3LFUNDCOST + B4LLTA + B5LBMIX

(9)

DATA AND SAMPLE DESCRIPTION:

The empirical part of this paper uses an unbalanced panel data set on which the Panzar and Rosse methodology has been applied containing a range of Italian banking firms. The data and the samples used for the estimation of H indicator are provided by Dr. Leone Leonida, Queen Mary, University of London.

The data used in this paper are annual and refer to the period 1995-1997 (3 years) for the first sample. The first sample for the econometric analysis is made up of an unbalanced panel data of 480 financial institutions of all sizes, for a total of 1401 observations. The number of parameters is 487. The longest time series is 3 years long and the shortest time series is only 2 years long with 2 time dummies.

The second sample covers the period of 1998-2000 (3 years) having 1330 number of observation from 474 banks of all sized. The number of parameters is 481. The longest time series is 3 years long and the shortest time series is 2 years long depicting unbalanced panel data with 2 time dummies.

In the Appendix, Table 3 provides a summary of the definition of relevant dependent variable, independent variables, bank specific factors’ variables and control variables.

“LGIRTA” is the log of gross interest revenue over total assets, which is used as dependent variable, also used by De Bandt and Davis (2000), and Trivieri F. (2005). Trivieri (2005) points out that according to Vesala (1995) and De Bandt and Davis (2000) it is the most appropriate choice because it then represents a price equation and not the revenue equation. Moreover, our equation will be consistent with the conceptual structure used by the application of Panzar and Rosse’s statistic to the banking sector. The choice for taking only the interest part of the total revenue of banks is consistent with underlying notion of the P-R model that financial intermediation is the core business of most banks. However, Shaffer (1982) and Nathan and Neaves (1989) have included total revenue instead of only interest revenue because of the fact that banks have increased their non-interest activities and services which have started generating income other than interest. But still the effect of other income on banks’ revenues is negligible.

“LLABCOST” is the cost of personal expenditures which include wages, salaries, general and administrative costs (Murjan and Ruza, 2002).

“LCAPCOST” is the log of proxy for the capital cost which includes depreciation which is written down on intangible and tangible assets, like physical depreciation, depreciation of investment securities and other overheads, the sum of which is expressed as a ratio over total assets (Murjan and Ruza, 2002).

“LFUNDCOST” is the log of proxy for the cost of funding. If we treat the banks as firms to be able to apply Panzar and Rosse approach, then we can consider loans and investments as output produce by banks. Deposits and other fundings are regarded as the banks’ productive inputs. Interest expenses comprise of interest paid on deposits, interest paid on borrowing, and other interest paid, whereas interest bearing funds includes customers’ deposits, interbank deposits, subordinated debt, and other long-term debt (Murjan and Ruza, 2002).

“LLTA” is the log of loans to total assets. It is chosen to account for bank-specific risk. In particular, it reflects both credit and interest rate risk, given the special features of loans in the bank’s portfolio of assets. In principle, a high ratio is indicative of bank’s relative illiquidity and limited capacity of further lending, but positively associated with revenues (Murjan and Ruza, 2002).

EMPIRICAL RESULTS:

EMPIRICAL RESULTS FROM THE FIRST TEST:

The H-statistic for the first sample is H = 0.6255 which indicates monopolistic competition. The goodness of fit, R-squared (R^2) which shows the percent of variation in the dependent variable (LGIRTA) explained by the independent variables, is 0.8777 or 87.77 percent. The surety about the selected estimate, or standard errors for labour cost (LLABCOST), capital cost (LCAPCOST), cost of funding (LFUNDCOST), loans to banks and clients over total loans (LBMIX) and loans to total assets (LLTA) are 0.06517, 0.01071, 0.5707, 0.03993 and 0.02403, respectively. The distance of the observations from the mean or standard deviation is 0.04719 and the variance is 0.002227. Wald test for time, dummy and joint test, all are significant for our explanatory variables pointing out the importance of the variables included in the model. However, at the 99 percent level of confidence only cost of funding (LFUNDCOST) is significant explaining the dependent variable very well, while other variables having very high t-probabilities are not sufficiently explaining the dependent variable. The null hypothesis for H = 1 and the null for H = 0, both are rejected. For full results, consult Table 4 in the appendix

EMPIRICAL RESULTS FROM THE SECOND TEST:

The H-statistic for the first sample is H = 0.6373 which indicates monopolistic competition. The goodness of fit, R-squared (R^2) which shows the percent of variation in the dependent variable (LGIRTA) explained by the independent variables, is 0.0.8814 or 88.14 percent. The surety about the selected estimate, or standard errors for labour cost (LLABCOST), capital cost (LCAPCOST), cost of funding (LFUNDCOST), loans to banks and clients over total loans (LBMIX) and loans to total assets (LLTA) are 0.05139, 0.01355, 0.03358, 0.03045 and 0.04634, respectively. The distance of the observations from the mean or standard deviation is 0.05417 and the variance is 0.002934. Wald test for time, dummy and joint test, all are significant for our explanatory variables pointing out the importance of the variables included in the model. While, at the 99 percent level of confidence labour cost (LLABCOST), capital cost (LCAPCOST) and cost of funding (LFUNDCOST) all are significant explaining the dependent variable very well. However, other variables like loans to total assets (LLTA) and loans to banks and clients (LBMIX) have very high t-probabilities and are not sufficiently explaining the dependent variable. The null hypothesis for H = 1 and the null for H = 0, both are rejected.

For full results, consult Table 5 in the appendix.

DISCUSSION ON RESULTS:

In the results obtained from both the samples, the H-statistic values for the empirical model used happen to lie between 0 and 1. The results are statistically significant; most of the coefficients have the expected sign, although some are not statistically significant. Hence, it can be quoted that during the observed periods banks earn their revenues under monopolistic competition. The H-statistic for the first sample is H = 0.6255, and for the second sample is H = 0.6373. Therefore it is appropriate to reject the hypotheses that banks work under monopoly conditions, conjectural variations of short-run oligopoly or perfect competition.

In terms of other exogenous variables, the coefficient of cost of funding is positive and highly significant in both the results. Thus, cost of funding plays a pivotal role in determining the Italian banks’ level of revenues. Other independent variables have very minor impact on revenues.

CONCLUSION:

An important finding revealed by this study is that the banking industry of Italy operates and earns its revenues under the conditions of monopolistic competition. The slight difference in the results of the two samples is negligible. The reason could possibly be the effects of some minor dissimilarities between the two samples. During the observed period of 6 years in total of both the samples, the degree of competition has not been changed in Italy.

In the case of Italy, it had been earning revenues as if under monopoly or conjectural variations short-run oligopoly conditions in 1989 (Molyneux et al. 1994). The competitive conditions relatively unchanged until 1992, and started getting improved afterwards showing indications of monopolistic competition (Cetorelli N. and Angelini P. 2000). On the other hand, De Bandt and Davis (2000) reported monopolistic competition in Italy. The results of this study are in line and consistent with the results of the latest literature of the field as well as with the previous research work.