Review of Financial Economics 13 (2004) 41 – 63 www.elsevier.com/locate/econbase
Portuguese banking: A structural model of competition in the deposits market Ana Canhoto * FCEE, Lisboa, Universidade Cato´lica Portuguesa, Palma de Cima, 1649-023 Lisbon, Portugal
Abstract In this article, we specify a structural model of competition in the deposits market, in line with the New Empirical Industrial Organization (NEIO) approach, and estimate it using data from Portuguese banking in the early 1990s. The article contributes to the general debate on competition issues by the theoretical model selected and by the particular banking industry considered. In contrast with most related articles, our model assumes product differentiation and interest rates as the strategic variables. Besides, research on competition in Portuguese banking deserves a special attention given the deep deregulatory changes operated. Our results suggest that, in the early 1990s, the Portuguese deposits market was operating under conditions far from perfect competition, evidencing market power features. However, some progress towards more competition could be clearly detected following deregulation. On the other hand, differences in the size of the banking institutions have not dictated significant differences in terms of their competitive behavior. D 2003 Published by Elsevier Inc. JEL classification: G21; L13; D43 Keywords: Banking; Conduct; Competition; Deregulation; Portugal
1. Introduction The purpose of this article is to specify a structural model of competition in the deposits market and estimate it using data from the Portuguese banking industry in the early 1990s. This period witnessed unprecedented transformations in European banking. The establishment of a single market for financial services within the European Union triggered many countries to adopt programs of deregulation, procuring the required level of harmonization of banking activities (see, for instance, Dermine, 1993, 2002; Gual, 1999). Portuguese banking industry represents an outstanding illustration of this process. * Tel.: +351-21-721-4262; fax: +351-21-727-0252. E-mail address:
[email protected] (A. Canhoto). 1058-3300/$ - see front matter D 2003 Published by Elsevier Inc. doi:10.1016/j.rfe.2003.07.002
42
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
Indeed, for historical reasons, the regulatory framework in place by then in Portugal was particularly severe calling for far-reaching transformations. Empirical research on Portuguese banking deserves, therefore, special attention. This article intends to be a contribution to this research, focusing on the evaluation of the competitive conditions in the deposits market during the 1990–1995 period. Besides, the article also contributes to the general debate on competition issues through the specification selected for the theoretical model. In line with the New Empirical Industrial Organization (NEIO) approach, we consider a structural model of competitive behavior that differs from those used in most related empirical articles on banking competition as far as it assumes product differentiation and defines interest rates as banks’ strategic variables. Given the assumptions of the model, there is separation between the pricing decisions taken by banks in the credit and deposits markets. This allows us to focus, separately, on banks’ behavior in the deposits market. Our estimated conduct parameters suggest that, in the early 1990s, the Portuguese deposits market does not function under conditions of perfect competition and evidences market power features. However, some progress towards more competition could be clearly detected following deregulation. On the other hand, our results reveal that differences in the size of the banking institutions do not dictate significant differences in terms of their competitive behavior. The article is structured as follows. Next section presents a brief overview of the Portuguese banking sector, focusing on its historical path in the recent decades and reviewing the available empirical literature on competition in Portuguese banking. Section 3 introduces the general methodology used in the article and develops the theoretical model. Empirical literature on European banking markets that adopt this methodology is also considered here. Section 4 deals with the empirical implementation of the model, namely, sample selection, data issues, and specification of the functional forms to be estimated. Section 5 presents the empirical results and Section 6 concludes.
2. Portuguese banking: A brief overview Portuguese banking has undergone very rapid and deep transformations, particularly remarkable during the last two decades. Following the April 1974 political events and the 1975 wave of nationalizations, heavy governmental interventions were imposed on all the key sectors of Portuguese economy, which severely constrained their activity. Legal restrictions were particularly influential in the banking activity and consisted, for instance, on credit ceilings, deposits and loans interest rates administratively fixed and entry barriers imposed on the industry. These barriers kept the recently nationalized banks free from competitive pressures that could come from the entry of new private banks (domestic and foreign), as well as from the expansion of the branch network of already installed banks (Borges, 1993). In the mid-1980s, financial markets’ liberalization began. It was a slow and gradual process, intentionally designed to safeguard state-owned banks against competition, given their weak and vulnerable situation (Corkill, 1993). In 1984, the reopening of banking activity to private enterprises and the expansion of the branch network were allowed and, in 1989, the reprivatization process started. In August 1985, interest rate deregulation began, a process only fully completed in September 1988 for lending rates and in May 1992 for deposit rates. In January 1991, the Central Bank announced the removal of the limits on the growth of bank lending used, since 1977, as an instrument of direct monetary control.
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
43
By the early 1990s, Portuguese banking was fully operating under a framework guided by the principles of 1993’s single European market for financial services set in the Second Banking Directive and similar to that of other European countries (Dermine, 2002). The structure of Portuguese banking suffered deep changes during that period. The number of institutions operating in Portugal increased very rapidly following the removal of entry barriers. In 1983, there were only 16 banking institutions while between 1990 and 1995 this number expanded from 33 to 45. This notable expansion was due both to the creation of new domestic private banks (some of them with foreign capital) and to the expansion of the network of foreign credit institutions. This enlargement was strongly reflected in the number of branches that sprung from 2082 in 1990 to 3876 in 1995, a total growth of more than 86%. In general, newly created banks are relatively small: In 1995, all new domestic banks account for 18% of total assets of the banking sector, for 13% of deposits, for 15% of the number of employees and, worth mentioning, for 20% of branches. The weight of foreign banks ranges between 4% and 8%. In this new environment, banking competition emerges as a significant issue in empirical work. The impact of deregulation and liberalization on the competitive performance of European banking markets has become a prevalent subject of research.1 As far as Portugal is concerned, empirical work on banking competition is still an incipient field, as easily anticipated given the heavy regulatory apparatus that prevailed in the industry until recent years. We mention here the recent articles by Barros (1999), Barros and Leite (1996), Barros and Modesto (1999), Lopes (1994), and Pinho (2000). The specific questions addressed in these articles as well as the methodologies adopted are rather diverse. However, all of them converge to the overall conclusion of some degree of collusive behavior in Portuguese banking in the early 1990s. Lopes (1994) considers the case of lending operations in Portugal and investigates whether the gap between interest rates on two specific categories of loans (discount of commercial bills and loans and advances to nonfinancial firms) can be explained by market power differences. Using data from January 1990 to March 1993, the article concludes that there is some evidence of less competitive behavior in the loans and advances market than in the discount of commercial bills market, thus favoring stronger market power in the first one; however, the observed differential between the two interest rates should be ascribed to different risk premia and not to these market power differences. Barros and Leite (1996) discuss the impact of banking liberalization on competition in the Portuguese banking industry, using a sample of 15 representative banks, in 1991 and 1992. Price–cost margins of individual banks are used as an indicator of bank’s conduct and it is assumed that margins differ across banks according to their own characteristics. For empirical purposes, individual margins are calculated using an exogenous estimate of the marginal cost of real resources of loans and deposits, which is kept constant across banks. The authors conclude in favor of the exercise of market power in both the loans and deposits markets, with an increase in competition from 1991 to 1992 in the deposits market. Barros and Modesto (1999) discuss issues of regulatory intervention in the Portuguese banking sector, considering both loans and deposits markets. Relative to the competitive structure of the markets during the 1990–1995 period, the article reveals that the interest rate set by each bank is positively affected by the competitors’ rates. This result confirms the oligopolistic nature of markets and indicates that products are strategic complements.
1
Galli and Pelkmans (2000) discuss regulatory reform and competitiveness in Europe in the 1990s in several industries, including banking.
44
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
Barros (1999) considers a competition model introducing, explicitly, multibranch banks and the constraint of an equal pricing rule across branches of the same bank. The model is estimated using a sample of 15 Portuguese banks for the years 1991 and 1992 revealing evidence of Nash behavior for the whole sample; besides, collusion among banks that belong to the same economic group cannot be discarded. Finally, Pinho (2000) studies the impact of deregulation on price and nonprice competition in the Portuguese deposits market. He estimates a system of three reduced form equations representing optimal decisions in deposit rates, advertising expenditures, and branches. Results suggest that market concentration fell while competitiveness increased, during the deregulation period considered (1986– 1992).
3. Empirical assessment of competitive conditions in banking 3.1. Introduction The assessment of the competitive behavior in an industry can be achieved under alternative methodologies and assumptions. The basic idea is that, if there is some degree of market power, prices differ from perfectly competitive ones, i.e., from marginal costs. Estimation methods vary on the definition of the proper parameters to evaluate the extent of the deviation between prices and marginal costs. Traditionally, the discussion has been focused on the relationship between market structure and performance. This positive relationship received alternative interpretations in terms of the competitive evaluation of the industry concerned. The structure–conduct–performance paradigm (SCP) is, possibly, the more extensively investigated approach in this area.2 Recently, several theoretical shortcomings of the SCP approach have been mentioned in the literature and, nowadays, many empirical studies are following a new course. This novel approach to competition evaluation has emerged under the impulse of the NEIO. One of its key distinguishing features is the use of structural models, derived directly from the conditions for firms’ optimization behavior, assuming that performance measures (such as price–cost margins) are not observable. This approach, pioneered by Iwata (1974), was strongly enhanced by the 1982 papers of Bresnahan (1982) and Lau (1982) and, particularly, by Bresnahan’s (1989) outstanding survey on the NEIO. Conduct is one of the parameters estimated in the model and can be interpreted as the degree of market power actually exercised by firms, evaluated with reference to specific benchmark values, well established by market theory. In addition, under the heading of the NEIO, Panzar and Rosse (1987) developed an alternative approach to the evaluation of competition, based on the properties of the reduced form revenue equation of the firms in the market. The sum of the factor price elasticities estimated from this equation constitutes the so-called H statistic, reflecting firms’ competitive behavior. These methodologies have been extensively applied to European banking both on multicountry studies and single-country studies. Based on the Bresnahan’s methodology, Neven and Ro¨ller (1999) estimates a structural model for the loan market with data from six European countries between 1981 and 1989 while Bikker and Haaf (2000) examines competition in the markets for deposits and loans in nine countries, including Portugal, in the 1990s. Using this same methodology, Angelini and Cetorelli (2000) evaluates competition in Italian banking, Suominen (1994) extends the standard one product model to 2
For a detailed survey of the SCP, see Schmalensee (1989).
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
45
the case of two markets (deposits and loans) and applies it to the evaluation of banking competition in Finland and Toolsema (2002) considers a dynamic version of the model to analyze the consumer credit market in the Netherlands. Applying a similar methodology, Berg and Kim (1994, 1996) investigate the Norwegian banking industry estimating conjectural variation models. On the other hand, Bikker and Groeneveld (2000), Bikker and Haaf (2002), De Bandt and Davis (2000), and Molyneux, LloydWilliams, and Thornton (1994) are notable illustrations of multicountry studies designed under the Panzar–Rosse approach. Their results suggest that, in general, monopolistic competition is the proper characterization of the conditions under which European banks have been operating. Focusing on single countries under the Panzar–Rosse methodology, we can refer the articles by Coccorese (1998) on Italian banking, Vesala (1995) on Finnish banking, and Hondroyiannis, Lolos, and Papapetrou (1999) and Hempell (2002) on Greek and German banking markets, respectively. The main findings of these articles are summarized in Tables 1A and 1B. This brief overview of the literature uncovers some mixed evidence but, clearly, imperfect competition in European banking markets is the prevailing and stronger result. To our knowledge, there are no studies using a similar methodology focused exclusively on Portuguese banking. Our article on the Portuguese deposits market intends to be a contribution to fill this gap and it is strongly inspired by the NEIO conduct parameter method. Our conclusions are consistent with most of the evidence reported by other empirical studies on the competitiveness in European banking in the early 1990s. 3.2. Model specification The structural model of competition in deposits market estimated in the article is derived from a static partial equilibrium oligopoly model, consistent with the industrial organization approach to banking developed in Freixas and Rochet (1997). It was inspired by earlier models on bank’s behavior developed by Klein (1971) and Monti (1972) and, later, by Dermine (1984), Hannan (1991), and Hannan and Liang (1993). The model assumes n banks (i=1,. . .,n) operating in the markets for loans (L), deposits (D), and securities (S). The markets for loans and deposits are characterized by the crucial assumptions of competition in prices and product differentiation. In the market for government securities, we retain the assumption prevalent in the literature that each individual bank is small relative to the size of the market thus acting as a price-taker. The assumption of competition in prices means that each bank chooses their loan and deposit interest rates to satisfy its objective function, which we assume to be profit maximization. Given product differentiation,3 individual bank’s demand for loans and supply of deposits4 are bank specific and
3
The assumption of product differentiation is not often considered in the empirical literature on banking competition. For instance, Dick (2001) considers a discrete choice approach to analyze consumer behavior and competition in the banking industry and introduces his paper stating that ‘‘unlike the existing literature this paper estimates a structural model of demand which explicitly introduces product differentiation.’’ 4 Following Freixas and Rochet (1997, note 1, p. 85), we refer to a ‘‘demand for loans by borrowers’’ and a ‘‘supply of deposits,’’ in conformity with the more traditional view that banks buys funds from deposits and sells them to borrowers. Kim and Vale (2001) adopts a similar definition considering the ‘‘loan demand faced by a bank.’’ Alternatively, we could refer to a ‘‘demand of deposit services from depositors’’ (Dick, 2001) or a supply (by the bank) of loanable funds to creditors.
46
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
Table 1A Empirical studies on European Banking Bresnahan approach Authors
Countries
Period
Neven and Ro¨ller (1999)
Belgium, France, Spain, Germany, Denmark, UK, The Netherlands Belgium, France, Germany, Italy, Spain, The Netherlands, UK, Portugal, Sweden
1981 – 1990
1971/1983 – 1998
Finland
1986 – 1989
Finland
1960 – 1984
Angelini and Cetorelli (2000)
Italy
1983 – 1997
Toolsema (2002)
The Netherlands
1993 – 1999
Bikker and Haaf (2000)
Suominen (1994)
Results loan market (consumer and commercial loans) deposits market, loans market
deposit plus loan markets (two products model) quantity index (one product model) total assets
consumer credit market (static and dynamic models)
collusive cartel like conduct, reduction of collusion over time highly competitive conduct, exception: oligopoly in Portuguese deposits market monopoly power (both markets) nearly perfect competition rejection of perfect competition 1993: significant increase in competition perfect competition
Table 1B Empirical studies on European banking Panzar – Rosse approach Authors
Countries
Period
Results
Bikker and Haaf (2002)
23 European and non-European countries (including Portugal)
1988 – 1998
Bikker and Groeneveld (2000) De Bandt and Davis (2000)
15 European countries Germany, France, Italy, United States
1989 – 1996 1992 – 1996
Coccorese (1998)
Italy
1988 – 1996
Hempell (2002) Hondroyiannis et al. (1999) Molyneux et al. (1994)
Germany Greece Germany, France, UK, Spain, Italy
1993 – 1998 1993 – 1995 1986 – 1989
Vesala (1995)
Finland
1985 – 1992
monopolistic competition, Europe: large banks highly competitive (exceptions: Finland, Norway, Spain) monopolistic competition Germany and France: monopolistic competition (large banks), monopoly (small banks), Italy: monopolistic competition (all banks) monopolistic competition (exception: perfect competition 1992; 1994) monopolistic competition monopolistic competition Germany, France, UK, Spain: monopolistic competition, Italy: monopoly monopolistic competition (exception: 1989, 1990)
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
47
depend on the own interest rate and on the vector of the bank rivals’ rates.5 We assume, also, there is separability between bank’s demand for loans and supply of deposits. Therefore, the supply of deposits to bank i can be written as D ; zi Þ Di ¼ Di ðriD ; ri
where Di represents the amount of deposits received by bank i, r Di represents the deposit rate set by bank i, and riD represents the vector of deposit rates set by banks other than i, i.e., its rivals; zi is a vector of exogenous factors unrelated to the bank’s demand for loans. The amount of deposits supplied to bank i is supposed to increase with its own interest rate and decrease with the rate offered by rivals, i.e., BDi/Bri D > 0, and BDi/Brj D0 and d2Di/(driD)2 0 driD dDi D ¼ a2 < 0 drRi
where
dDi ¼ a3 þ a5 riD > 0 dEi
ci ¼
D drRi driD
dDi ¼ a4 > 0 dGNP
Concerning marginal (operating) costs in Eq. (4), we assume they can be written as a function of the amount of deposits received Di and of the input prices, basically the price of labor usage (wEi ) and the cost of capital plus materials (wKi ). The sign of parameter b1, which measures the impact of the amount of deposits on marginal costs, depends on the returns to scale characterization of the sample. Input price parameters are expected to be positive, as suggested by conventional production theory. Moreover, we add a scale variable defined as the number of employees by branch (Ei/BRi).12 Different values of this variable can capture differences in the efficiency of branch operation or (and) merely differences in the devising of the branch network itself. The efficiency argument suggests that more employees by branch may signal inefficiency and, therefore, higher marginal costs. The possibility of scale diseconomies (or 12
In the current specification, branches are considered as an exogenous variable in the specification of bank’s costs. Kim and Vale (2001) takes a different approach, introducing bank’s branching decisions as a strategic nonprice variable in banking conduct. Indeed, the purpose of the article is to empirically assess the role of the branch network as a nonprice strategic variable. In Pinho’s (2000) work, branching is also introduced as a nonprice competition variable, together with advertising.
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
51
economies) at the branch level would point to a positive (or negative) impact of the variable on marginal costs, i.e., to b4>0 (or b40). Also, larger banks, i.e., those with a higher number of employees, attract more deposit funds, dD/dE=a3+a5rD>0. The relationship between the amount of deposits and the expansion of economic activity, evaluated by the
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
57
Table 5C Portuguese deposit market: Estimation results Model
C1
Supply of deposits aCGD a0 a1 a2 a3 a4 a5 Demand of deposits b bCGD b0 b1 b2 b3 b4 Conduct c1 c2 Log likelihood Standard error supply Standard error demand
C2
C3
C4
Parameter
t statistic
Parameter
t statistic
Parameter
t statistic
Parameter
t statistic
1.009 0.264 12.272 19.358 0.007 0.082 0.107
4.61*** 0.54 2.99*** 2.87*** 1.88** 2.55** 2.58***
0.921 0.173 14.569 22.954 0.010 0.081 0.095
3.54*** 0.31 2.66*** 2.67*** 2.67*** 2.15** 2.17**
1.044 0.111 14.361 22.633 0.007 0.077 0.113
5.50*** 0.20 2.64*** 2.62*** 1.57 2.18** 2.51**
1.237 0.039 16.448 25.673 0.003 0.077 0.148
5.69 0.07 2.46 2.43 0.59 1.96 2.37
0.170 0.024
2.15** 0.46
0.224 0.071
1.92** 1.20 2.14** 1.55 2.22** 2.31** 2.68***
2.01
0.95 2.19** 2.29** 2.69
0.152 0.032 0.016 0.006 0.299
0.153
0.023 0.017 0.007 0.324
0.015 0.007 0.288
1.84 2.12 2.43
0.632 0.622
6.25*** 6.19***
0.643 0.635
6.80 6.78
0.632 0.622 520.82 0.231 0.027
6.00*** 5.94***
0.021 0.009 0.388
1.97** 2.05** 2.28**
0.635 0.629 519.56 0.262 0.033
6.48*** 6.44***
520.35 0.262 0.025
513.23 0.305 0.027
Supply:
D ¼ ða0 þ aCGD dumCGD Þ þ a1 rD þ a2 rRD þ a3 E þ a4 GNP þ a5 ðrD EÞ
ð3VÞ
Demand:
rD ¼ rS ð1 qÞ MCD
D a1 þ a2 ½dum1 c1 þ dum2 c2 þ a5 E
ð5WÞ
where
MCD ¼ b0 þ b1 D þ b2 wE þ b3 wK þ b4 ðE=BRÞ
ð4Þ
MCD ¼ ðb þ bCGD dumCGD Þ þ b1 D þ b2 wE þ b3 wK þ b4 ðE=BRÞ
ð4VÞ
or
*** Significant at the 1% level. ** Significant at the 5% level.
macroeconomic variable GNP shown in parameter a4, is positive and statistically significant. Finally, intercept parameters suggest that the supply curve of CGD is shifted outward relative to that of the other institutions, i.e., aCGD>0, a result in accordance with the weight of that institution in Portuguese banking. Table 6 also shows estimates of deposits supply elasticities. Own-price and cross-price elasticities are of
58
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
Table 6 Deposit demand: Slope and elasticities Model D
dD/dr = a1 + a2c + a5E dD/dE = a3 + a5rD Own-price elasticity (e) Rival-price elasticity (eRi) Total price elasticity (g) Labor elasticity
A1
B1
C1
0.156 0.018 2.457 3.743 0.031 0.011
0.650 0.019 2.601 3.939 0.129 0.011
0.162 0.018 2.441 3.716 0.032 0.010
opposite sign, as expected, and close to being equal in absolute value. This implies that, as far as the supply of deposits funds is concerned, the truly important price variable is the ratio between bank’s own rate and the rivals’ rate. Estimation of the deposits demand parameters shows that the CGD dummy, included in the intercept of the marginal cost specification, is never significant. The estimated parameter of the output variable in the marginal cost specification, b1, is also nonsignificant. The impact of the wage rate on marginal costs, measured by b2, appears negative, in contradiction with standard microeconomic theory, and exhibiting a relatively high t ratio. A plausible explanation for this somewhat puzzling result may be a high degree of substitution among factors in the production of deposit services.19 In addition, some explanation may lie on the quality of data used in the estimation, particularly in the case of the constructed input price proxies, that may contain some undetected inadequacy. Parameter b3 for the price of capital and materials is positive and statistically significant. The scale variable, defined by the number of employees by branch, has also a positive impact on the marginal cost of deposits as evidenced by estimated parameter b4. This may suggest an inverse relationship between the number of employees by branch and bank’s efficiency, measured by marginal costs of deposits. Alternatively, the number of employees by branch can be interpreted just as a size measure showing that some banks operate larger branches thus requiring a higher number of employees. In that case, a positive impact on marginal costs can be interpreted as decreasing returns to scale relative to this size measure. Let us consider now the results on the competitive behavior in the deposits market, assessed by estimated conduct parameters c. First, we note that, whatever the representation, there is clear evidence of some market power. Estimated c are all above 0.60 and significantly different from zero, thus discarding the case of Nash behavior (c=0) with evidence of less competition (c>0). This result is in line with most of the empirical literature on the competitive conditions in European countries reviewed above and summarized in Tables 1A and 1B. Indeed, in those articles that use the Panzar–Rosse methodology, monopolistic competition is clearly the predominant outcome. Using the Bresnahan approach, Neven and Ro¨ller (1999) rejects noncooperative Nash behavior in the consumer and commercial loans markets in a sample of European countries, finding evidence of a more collusive cartel-like behavior in the 1980s while Angelini and Cetorelli (2000) rejects perfect competition in Italian banking between 1983 and 1997 and Suominen (1994) uncovers monopoly power in Finnish deposits and loans markets in the late 1980s. Suggestively, Bikker and
19
Ro¨ller and Sickles (2000, p. 857) obtains a similar unexpected result in the estimation of the parameters of a structural model of competition for the European airline industry.
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
59
Haaf (2000) reveals a high degree of competition in the markets for deposits and loans in a group of European countries, including Portugal; overall, the hypothesis of perfect competition cannot be rejected in the article, but ‘‘the only exception is the Portuguese deposit market which is probably better characterized by oligopoly,’’ a result of utmost importance in our case. The possibility of a relationship between the pattern of conduct and banks’ size can be examined comparing the results of versions A and B of the model. In version B, estimated values of conduct parameters decrease with the size of the group considered (cs=0.63, cm=0.62, cl=0.58). This pattern of competitive behavior, closer to the Nash equilibrium in larger banks, would be consistent with our previous guess of a more competitive behavior in larger banks. However, the log likelihood ratio test shows that the restricted version of the model (version A) is not significantly different from the nonrestricted one (version B). The actual v2 statistic is 0.512, whereas the critical value, at the 5% significance level, is 3.841. Therefore, we conclude that size does not determine significantly different patterns of competitive conduct of Portuguese banks in the period under consideration. Relative to this issue, there is mixed evidence in the empirical literature. For instance, according to De Bandt and Davis (2000), large banks in Germany and France operate under monopolistic competition and smaller ones under monopoly while in Italy the result of monopolistic competition is not influenced by the size of the institutions. Similarly, Bikker and Haaf (2002) finds that, in Europe, large banks appear to operate in a highly competitive environment with the exceptions of Finland, Norway, and Spain. To discuss the effects of deregulation on the competitive behavior of Portuguese banking, we consider version C of the model. Setting 1993 as the beginning of the more deregulated period, we obtained conduct estimates decreasing from 0.63 to 0.62. Although this may seem a minor change in the value of the parameter, the log likelihood ratio test shows that restricted version A is significantly different from the nonrestricted version C, as the actual v2 statistic equals 16.79. This same result is confirmed by the clear rejection of the null hypothesis c1=c2 in version C of the model (t statistic=3.513). Therefore, we can conclude that the deregulation process that took place in Portuguese banking up to early 1990s had a significant impact on the strengthening of competition. The positive impact of deregulation on competition has been emphasized in the literature. For instance, and referring to European banking, Inzerillo, Morelli, and Pittaluga (2000) refers that, during the 1990s, Europe experienced an increase in competition in the banking markets that can be ascribed to three main causes: The process of globalization of the markets, major technological innovations and ‘‘an intense process of deregulation,’’ starting from the 1980s, more or less everywhere, aimed at the creation of a single financial market. Also, according to Vives (1991), ‘‘regulatory reform is a mixed process of deregulation and re-regulation which, when coupled with European integration, will translate in a substantial increase in competition in banking markets.’’ Neven and Ro¨ller (1999) address this question empirically for a sample of European countries concluding that ‘‘the behavior of banks has become less collusive over time and this observation can presumably be associated with the progressive deregulation that has taken place over the period.’’ Relative to the Italian banking industry, Angelini and Cetorelli (2000) finds that, in 1993, a relevant change in competitive conditions took place suggesting ‘‘the Italian banking industry has become more competitive in recent years.’’ Finally, following Section 3.2, we define mark-up in the deposits market as the ratio between the rate paid to depositors (rD) and the net return received by the bank for each unit of deposits, NR=rS(1q)MCD, i.e., mup=rD/NR=g/(1+g). Given this definition mark-up is expected to increase as market power decreases, approaching unity under the limit conditions of perfect competition. In that case, the net revenue derived form one unit of deposits would be fully channeled to reward
60
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
Table 7 Mark-up in the deposit market Model
Mark-up estimated in the model mup = g/(1 + g)
Mark-up Nash assumption mupc = 0 = e/(1 + e)
A1 B1 C1
0.030 0.114 0.031
0.711 0.722 0.709
depositors. Another convenient benchmark to assess estimated mark-up is the hypothetical one that would be obtained under the assumption of Nash conduct, i.e., c=0 such that mupc=0=e/(1+e). Table 7 shows both sets of results. The mark-up ratio obtained using estimated values for elasticities is always below unity, as expected, and relatively low. Under versions A and C of the model, for instance, figures suggest that the interest rate paid on deposits corresponds to only 3% of the net revenue obtained by the bank on an additional unit of deposits. If we compute mark-up using parameters c1 and c2 in separate, we conclude that mark-up increased from 0.013 in the period before 1993 to 0.052 in the second period. Under the hypothetical Nash case, mark-up would be much higher, around 70%, showing again that competition in the Portuguese deposits market, in the period under analysis, was very weak.
6. Conclusions In this article, we specified a structural model of competition in the deposits market, in line with the NEIO approach, and estimated it using data from Portuguese banking in the early 1990s. This period witnessed unprecedented transformations in European banking, with most countries adopting programs of deregulation, procuring the level of harmonization of banking activities required by the establishment of the single market for financial services. Portuguese banking represents an outstanding illustration of this process, as the regulatory framework in place by then was particularly severe, thus calling for farreaching transformations. Following the industrial organization approach to banking, we consider a static partial equilibrium model with product differentiation and competition in prices. The assumptions of no relation between the demand for loans and the supply of deposits and separability in operating costs, allow the study of bank’s behavior in the deposits market on its own. We estimate the model pooling time-series and cross-section data from a sample of 20 banking institutions with annual observations between 1990 and 1995. In the estimations, a special role is assigned to CGD, the leading Portuguese bank with an outstanding weight in Portuguese banking, particularly in the deposits. The results suggest that, in the early 1990s, the Portuguese deposits market was operating under conditions far from perfect competition, evidencing market power features. Indeed, estimated conduct parameters are all above 0.60 and significantly different from zero, thus discarding the case of Nash behavior with evidence of less competition. Next, we investigated the possibility of differences in conduct dictated by the size of the institutions, considering banks divided into three size classes. We estimated conduct parameters for each individual class introducing appropriate dummy variables. The statistical tests performed revealed that size does not determine significantly different patterns of competitive conduct in the period under consideration.
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
61
Finally, we investigated the effects of deregulation on the competitive behavior of Portuguese banking. The year 1993 was set as the beginning of the more deregulated period and we introduced properly defined time dummy variables. Conduct parameters for the prederegulation and deregulated periods were found statistically different and decreasing over time. This result can be interpreted as evidence that the deregulation process that took place in Portuguese banking had a significant impact on the strengthening of competition. This positive effect of deregulation on competition has been widely emphasized in the literature and is corroborated once more in the case of Portuguese banking, as shown in this article.
Appendix A The statistical banking data used in the article proceeds from different sources: statistical publications of the Bank of Portugal and of the Portuguese Banking Association as well as individual banks’ reports. Data reported on individual banks is that of the ‘‘global activity,’’ i.e., activity of the bank itself, in Portugal and abroad, not including the activity of the bank’s subsidiaries (‘‘group’’). Individual price variables—interest rate on deposits, price of labor, and price of capital usage—were replaced by proxies constructed in terms of ‘‘average’’ costs using the figures in bank’s balance sheet and loss and profit accounts. In this process, when stock variables are included in the definition, the end of period value Xt is replaced by a simple average, 0.5(Xt1+Xt). The particular procedure adopted in each case was highly dictated by data availability, requiring serious simplifying assumptions. The calculation of the price of labor relative to each individual bank, shortly termed the wage rate, is relatively standard. It is computed as the ‘‘average cost of labor usage’’ and is defined as the ratio between staff costs and the number of employees. The evaluation of the ‘‘price of capital’’ is a rather controversial topic. Assuming that we are approximating it by the ‘‘average cost of capital,’’ two problems remain—the definition of the costs incurred by the bank relative to physical capital usage and the selection of the more appropriate variable to ‘‘average’’ it. We handled these questions considering an enlarged definition that presumes the inclusion of the usage of materials and equipment together with physical capital. Costs are defined as that portion of operating costs related to the overall working of the institutions. It comprises, for instance, real estate rents and repairs, current materials expenses, and communication and advertising. The cost variable is averaged by the total amount of assets. This solution to the calculation of the ‘‘price of capital’’ avoids the traditionally discussed problem of evaluating physical assets in banking. Interest rates on deposits set by individual banks are proxied by the ‘‘average cost of deposits,’’ i.e., the ratio between interest costs paid by the bank and the amount of deposits held in each period. The implementation of this procedure requires some caution in the usage of the item ‘‘interest and equivalent costs,’’ as reported in bank’s loss and profit account. In fact, the definition of this item comprehends interest costs pertaining not only to customer’s deposits, but to all interest-earning banks’ liabilities. This broad interest cost definition calls for some adjustment in the method to construct the proxy. Given the limited information, it is not feasible to obtain a reliable split of total interest costs into the amount imputed exclusively to deposits from customers and the amount relative to other financial liabilities. Alternatively, we use an approximation where interest costs, taken as a whole, are averaged by a broader measure of ‘‘deposits’’ which go beyond customer deposits, including all interest-bearing liabilities. Note that, using this approach, interest rates can be computed only relative to an aggregate of all categories of deposits and not to single categories defined by maturity or the nature of the agent.
62
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
References Angelini, P., & Cetorelli, N. (2000). Bank competition and regulatory reform: The case of the Italian banking industry. Federal Reserve Bank of Chicago, Working Paper Series, Research Department WP. 99-32. Barros, F., & Modesto, L. (1999). Portuguese banking sector: A mixed oligopoly. International Journal of Industrial Organization, 17(6), 869 – 886. Barros, P. P. (1995). Post-entry expansion in banking: The case of Portugal. International Journal of Industrial Organization, 13, 593 – 611. Barros, P. P. (1999). Multimarket competition in banking, with an example from the Portuguese market. International Journal of Industrial Organization, 17, 335 – 352. Barros, P. P., & Leite, A. N. (1996). Competition in Portuguese commercial banking. Economia, 20(1), 7 – 30. Berg, S. A., & Kim, M. (1994). Oligopolistic interdependence and the structure of production in banking: An empirical evaluation. Journal of Money, Credit and Banking, 26(2), 309 – 322. Berg, S. A., & Kim, M. (1996). Banks as multioutput oligopolies: An empirical evaluation of retail and corporate banking markets. Proceedings of a Conference on Bank Structure and Competition, Federal Reserve Bank of Chicago ( pp. 183 – 201). Bikker, J. A., & Groeneveld, J. M. (2000). Competition and concentration in the EU banking industry. Kredit und Kapital, 33(1), 62 – 98. Bikker, J. A., & Haaf, K. (2000). Competition, concentration and their relationship: An empirical analysis of the banking industry. De Nederlandsche Bank, Research Series Supervision, 30. Bikker, J.A., & Haaf, K. (2002). Competition, concentration and their relationship: An empirical analysis of the Banking Industry. Journal of Banking and Finance, 26, 2191 – 2214. Borges, A. C. (1993). Portuguese banking in the single European market. In J. Dermine (Ed.), European banking in the 1990s. Blackwell. Bresnahan, T. F. (1982). The oligopoly solution concept is identified. Economics Letters, 10, 19 – 27. Bresnahan, T. F. (1989). Empirical studies in industries with market power. In R. Schmalensee, & R. Willig (Eds.), Handbook of industrial organization vol. 2 (pp. 1011 – 1057). Amsterdam: North-Holland. Cabral, L. M. B. (1995). Conjectural variations as a reduced form. Economic Letters, 49, 397 – 402. Coccorese, P. (1998). Assessing the competitive conditions in the Italian banking system: Some empirical evidence. Banca Nazionale del Lavoro, Quarterly Review, 205, 171 – 191. Corkill, D. (1993). The Portuguese economy since 1974. Edinburgh, UK: Edinburgh University Press. De Bandt, O., & Davis, E. P. (2000). Competition, contestability and market structure in European banking sectors on the eve of EMU. Journal of Banking and Finance, 24, 1045 – 1066. Dermine, J. (1984). Pricing policies of financial intermediaries. Studies in contemporary economics vol. 5. Berlin: SpringerVerlag. Dermine, J. (1993). European banking in the 1990s. (2nd ed.). Oxford: Blackwell. Dermine, J. (2002, October). European banking: Past, present and future. Second ECB Central Banking Conference, Frankfurt am Main. Dick, A. (2001). Demand estimation and market power in banking industry. Cambridge, MA: MIT Press. Farinha, L. A., & Santos, J. A. C. (2002). Switching from single to multiple bank lending relationship: Determinants and implications. Journal of Financial Intermediation, 11, 124 – 151. Freixas, X., & Rochet, J. (1997). Microeconomics of banking. Cambridge, MA: MIT Press. G., Galli, & J., Pelkmans (Eds.). (2000). Regulatory reform and competitiveness in Europe. Edward Elgar. Gual, J. (1999). Deregulation, integration and market structure in European banking. Journal of Japanese and International Economics, 13, 369 – 372. Hannan, T. H. (1991). Foundations of the structure – conduct – performance in banking. Journal of Money, Credit and Banking, 23(1). Hannan, T. H., & Liang, J. N. (1993). Inferring market power from time series data: The case of the banking firm. International Journal of Industrial Organization, 11. Hempell, H. S. (2002). Testing for competition among German banks. Deutsche Bundesbank, Economic Research Center, Discussion Paper. 04/02.
A. Canhoto / Review of Financial Economics 13 (2004) 41–63
63
Hondroyiannis, G., Lolos, S., & Papapetrou, E. (1999). Assessing competitive conditions in the Greek banking system. Journal of International Financial Markets, Institutions and Money, 9(4), 377 – 391. Inzerillo, U., Morelli, P., & Pittaluga, G. (2000). Deregulation and changes in the European banking industry. In G. Galli, & J. Pelkmans (Eds.), Regulatory reform and competitiveness in Europe vol. II. Edward Elgar. Iwata, G. (1974). Measurement of conjectural variations in oligopoly. Econometrica, 42(5), 947 – 966. Kim, M., Kristiansen, E. G., & Vale, B. (2001). Endogenous product differentiation in credit markets: What do borrowers pay for? Oslo, Norges Bank Working Paper (2001/8). Kim, M., & Vale, B. (2001). Non-price strategic behavior: The case of bank branches. International Journal of Industrial Organization, 93, 1583 – 1602. Klein, M. A. (1971). A theory of the banking firm. Journal of Money, Credit and Banking, 3, 205 – 218. Lau, L. J. (1982). On identifying the degree of competitiveness from industry price and output data. Economics Letters, 10, 93 – 99. Lopes, M. C. (1994). Market power measurement—an application to the Portuguese credit market. Investigaciones Econo´micas, XVIII(2), 391 – 399. Molyneux, P., Lloyd-Williams, D. M., & Thornton, J. (1994). Competitive conditions in European banking. Journal of Banking and Finance, 18, 445 – 459. Monti, M. (1972). Deposit, credit and interest rate determination under alternative bank objective functions. In K. Shell, & G.P. Szego (Eds.), Mathematical methods in investment and finance. Amsterdam: North-Holland. Neven, D., & Ro¨ller, L. H. (1999). An aggregate structural model of competition in the European banking industry. International Journal of Industrial Organization, 17, 1059 – 1074. Panzar, J. C., & Rosse, J. N. (1987). Testing for monopoly equilibrium. Journal of Industrial Economics, 35, 443 – 456. Pinho, P. S. (2000). The impact of deregulation on price and non-price competition in the Portuguese deposits market. Journal of Banking and Finance, 24, 1515 – 1533. Ro¨ller, L. H., & Sickles, R. C. (2000). Capacity and product market competition: Measuring market power in a ‘‘puppy-dog’’ industry. International Journal of Industrial Organization, 18, 845 – 865. Schmalensee, R. (1989). Inter-industry studies of structure and performance. In R. Schmalensee, & R .D. Willig (Eds.), Handbook of industrial organization vol. 2. Shaffer, S. (1993). A test of competition in Canadian banking. Journal of Money, Credit and Banking, 25(1), 49 – 61. Shapiro, C. (1989). Theories of oligopoly behavior. In R. Schamalensee, & R. D. Willig (Eds.), Handbook of industrial organization vol. II. Slade, M.E. (1986). Conjectures, firm characteristics and market structure. An empirical assessment. International Journal of Industrial Organization, 4, 347 – 370. Suominen, M. (1994). Measuring competition in banking: A two-product model. Scandinavian Journal of Economics, 96(1), 95 – 110. Tirole, J. (1989). The theory of industrial organization. Cambridge, MA: MIT Press. Toolsema, L. A. (2002). Competition in the Dutch consumer credit market. Journal of Banking and Finance, 26(11), 151 – 165. Vesala, J. (1995). Testing competition in banking: Behavioral evidence from Finland. Bank of Finland Studies Working Paper. No. E:1. Vives, X. (1991). Regulatory reform in European banking. European Economic Review, 35, 505 – 515.