Do Commercial Banks, Savings Banks, and Credit Unions Compete?

Transcription

1 Internatonal Journal of Busness and Economcs, 2006, Vol. 5, No. 1, Do Commercal Banks, Savngs Banks, and Credt Unons Compete? Luco Fuentelsaz Departamento Economía y Dreccón de Empresas, Unversty of Zaragoza, Span Jame Gómez Departamento Economía y Dreccón de Empresas, Unversty of Zaragoza, Span Vctora Lucea Departamento Economía y Dreccón de Empresas, Unversty of Zaragoza, Span Abstract Ths paper attempts to evaluate the extent to whch rvalry between the dfferent fnancal ntermedares that operate n the Spansh bankng system has been nfluenced by the recent lberalzaton process. Our theoretcal model s borrowed from populaton ecology and allows us to take nto account the relatonshp between the three types of fnancal ntermedares as well as the effect of other exogenous factors on the capacty of the envronment to sustan the growth of each populaton. Key words: bankng sector; competton; deregulaton; populaton ecology; Span JEL classfcaton: L19; G21 1. Introducton Do the dfferent types of fnancal ntermedares operatng n a bankng ndustry compete? Is there a clear overlap between the market segments where each type of frm operates? Has the recent wave of deregulaton ncreased rvalry between the dfferent groups that consttute the bankng system? These undoubtedly nterestng questons have receved relatvely lttle attenton n the lterature; we consder them n ths artcle. Tradtonally, bankng actvtes have been hghly regulated. For many years, bankng frms have had strong lmtatons n what refers to the choce of compettve varables such as nterest rates on loans and deposts or locaton decsons. In ths Receved November 7, 2005, revsed June 1, 2006, accepted June 12, Correspondence to: Departamento Economía y Dreccón de Empresas, Gran Vía 2, E Zaragoza, Span. E-mal: lfuente@unzar.es. The authors acknowledge fnancal support from the Spansh Mnstry of Scence and Technology and FEDER (projects SEC and SEJ ) and the Regonal Government of Aragón through the recognton of the Generés Research Group.

2 18 Internatonal Journal of Busness and Economcs context, t was argued that competton was usually scarce, wth a clear segmentaton of actvtes and most banks enjoyng certan monopoly power wthn ther market (product or geographcal) nches. Once most of the restrctons have dsappeared, fnancal ntermedares have redesgned ther strateges n order to face the new compettve envronment wth the subsequent effect on rvalry. Recent lterature has, n fact, documented ths ncrease n competton; see, for example, the analyses of Gual and Neven (1992) n Europe, Jayaratne and Strahan (1997) or Strahan (2003) n the US, or Sturm and Wllams (2004) n Australa. However, a great deal of research has focused the analyss of competton n only one group of ntermedares (usually commercal banks) forgettng that n many countres bankng actvtes are developed by a varety of fnancal ntermedares (such as banks, savngs and loans, savngs banks, credt unons, co-operatve banks, and buldng socetes among others) whose actvtes have presumably shown a degree of confluence as a result of deregulaton. Importantly, competton n the bankng ndustry takes place among a set of enttes wth dfferent objectves. Ths makes the analyss of the potental nteractons arsng among the dfferent types of ntermedares especally nterestng gven that for-proft frms compete wth mutual or not-for-proft frms (Emmons and Schmd, 2000). Ths s precsely the man queston we try to answer n ths paper. Gven that deregulaton has fnshed n most of the developed countres, we are nterested n knowng both (1) whether the dfferent types of ntermedares were competng before deregulaton and (2) how the new framework has affected competton between the dfferent bankng groups that operate n a gven country. Wth ths purpose n mnd, our emprcal analyss s conducted for the Spansh bankng system, a context especally sutable for ths type of research gven the presence of three types of fnancal nsttutons (.e., commercal banks, savngs banks, and credt unons) whch, n prncple, had very dfferent compettve profles a couple of decades ago and whose actvtes have progressvely converged. Our research evaluates to what extent the dfferent types of fnancal ntermedares take nto account not only the decsons of the members of ther same group but also competton comng from the rest of fnancal ntermedares at the tme of desgnng ther compettve strateges. Wth ths am, ths paper apples a socology-based model to evaluate the mpact of deregulaton n the ndustry as a whole, n ven smlar to Barron et al. (1998). To analyze ths phenomenon, we do not consder ndvdual compettors; nstead, we group the frms nto competng populatons as s common n populaton ecology. 2. Theoretcal Framework One mportant area of study n populaton ecology s the analyss of the nature of growth. The man assumpton n ths lterature s that the envronment has a lmted capacty of resources avalable to the development of an organzatonal form (e.g., a frm), whch lmts growth. When several organzatonal forms compete for the same bundle of resources, the nteractons takng place gve rse to a compettve

3 Luco Fuentelsaz, Jame Gómez, and Vctora Lucea 19 process whose fnal result s the selecton of the fttest frms, whle the rest ext the market. Although there are alternatve ways to model growth and the compettve nteractons among organzatonal forms (for example, Lotka-Volterra models) the use of partal adjustment models s frequent. The specfc model we use n ths research s largely based on Barron et al. (1998). It assumes that, for every nfntesmal perod and each populaton, the socal process under study (sze of the populaton n our case), y (, evolves toward an objectve value, y (, n accordance wth the followng expresson: [ y ( y ( ] dy ( = r, (1) dt where y ( s the sze of populaton at tme t, y ( s the envronment capacty where populaton develops, and r s the adjustment coeffcent that determnes the speed wth whch the system reduces the dfference between y ( and y (. Accordng to expresson (1), an ncrease n the value of y s proportonal to the dfference between the present value of the varable and ts target value. The speed wth whch that dfference s reduced depends on the adjustment coeffcent r. When r approaches zero, the system converges slowly. On the other hand, the larger the value of r, the qucker s the reducton. One nterestng aspect of ths type of model s that t represents competton between several populatons usng a system of nterdependent equatons. The most frequent strategy assumes that both the nteracton among competng populatons and the effect of exogenous factors affect equaton (1) through ther nfluence on the envronment capacty to mantan every populaton, (, n a lnear form. That s: y = + ( α j y j ( π x(, (2) j 1 where π x( represents the effect of exogenous varables x ( and the α are j competton coeffcents. As we can see n equaton (2), the envronment capacty for each populaton depends on exogenous factors and on the nteracton between the dfferent populatons. In ths way, the estmaton of the coeffcents α j allows us to obtan a measure of the ntensty and drecton of the nteractons. When α j and α are negatve, the populatons compete for the avalable resources. j Conversely, f both coeffcents are postve, the relatonshp s symbotc and an ncrease n y ( n one populaton has a postve effect on the other. If we substtute expresson (2) nto (1) and smplfy we obtan, for each populaton : y dy ( dt = r y α j y j ( + rπ x( r y(. (3) j In spte of ts smplcty, the model cannot be estmated n a drect way, gven the mpossblty of observng nfntesmal ncreases consdered on the left hand

4 20 Internatonal Journal of Busness and Economcs sde. An alternatve way to proceed s to consder the ntegral of equaton (3). To do ths, we need to wrte the partal adjustment model n matrx form. Thus, equatons (2) and (3) can be rewrtten (Tuma and Hannan, 1984) as: [ y ( y( ] dy( = R dt (4) y ( = Ay( + π x(, (5) where R s a dagonal matrx that contans the adjustment coeffcents for each of the J populatons, dy ( dt, y (, and y ( are J 1 vectors, A s a J J matrx wth zeros along the man dagonal, x ( s an ( M + 1) 1 vector of explanatory varables, and π s a J ( M + 1) matrx that contans the coeffcents that show the effect of the varatons of the explanatory varables that correspond to envronment capacty. Substtutng (5) nto (4) and smplfyng, we have: dy( dt = Cy( + Bx(, (6) where C RA R s a J J matrx and B Rπ s a J ( M + 1) matrx. If we ntegrate equaton (6) we obtan: y = Γy( t ) + Θ x( t ) + Θ Δx( ), (7) ( t where ΛΔ 1 Γ Ve t V (8) 1 1 Θ C ( Γ Ι C (9) 1 ) Θ C C ( Γ Ι) Ι B 2 t Δ (10) 1 C VΛV. (11) So, to estmate the coeffcents that appear n equaton (3) we need to follow a two-stage procedure. Frst, we need to estmate the ntegral equaton (7) to obtan matrxes Γ, Θ 1, and Θ 2. Second, we use equatons (9) to (11) to obtan the parameters of nterest. 3. Organzatonal Forms n the Spansh Bankng Sector As stressed n Secton 1, the man purpose of ths paper s to assess whether the transformatons takng place n the Spansh bankng sector n recent years have nfluenced the degree of rvalry among the dfferent types of populatons. Therefore, a frst key conceptual queston that arses before applyng the model proposed n Secton 2 s the dentfcaton of the organzatonal forms operatng n t. Accordng

5 Luco Fuentelsaz, Jame Gómez, and Vctora Lucea 21 to Hannan and Freeman (1977), a populaton may be defned as a set of organzatons that survve wthn the lmts of a system and that share a common form. Although the lterature does not provde us wth a lst of attrbutes n order to dfferentate among the dfferent organzatonal forms operatng n a market, prevous research (Barron et al., 1994, 1998) uses factors such as organzatonal orgn, governance system, or a frm s objectve functon that may be appled to defne populatons n our context. The Spansh bankng sector s comprsed of three types of fnancal ntermedares: commercal banks, savngs banks, and credt unons. In spte of the fact that, currently, the three groups are able to perform the same type of actvtes throughout the country, ths has not tradtonally been the case. Importantly, the dfferences between them not only have been condtoned by factors such as orgn, governance system, or objectve functons, but also by regulaton, whch has determned both the scope of operaton and the types of products and servces offered. For example, commercal banks have been tradtonally specalzed n wholesale bankng, whle savngs banks and credt unons have centered ther actvtes on retal bankng. To understand these dfferences, we must take nto account not only the regulatory envronment faced by the banks, but also ther dfferent foundatonal nature and objectves, gven that these crcumstances have hghly condtoned ther postonng. Banks were created as for-proft lmted frms, wth the am of performng the whole range of actvtes related to the provson of fnancal servces. Savngs banks, however, started to emerge n the mddle of the 19th century, and ther actvty was ntally orented towards famles and small busnesses. Somethng smlar occured n the case of credt unons, whch were founded under a clear mutual character. As a consequence, the product and geographc scope of savngs banks and credt unons had been much more lmted gven that they were tradtonally worred about the communtes they tred to satsfy. However, ths apparent constrant could have become, wth the passage of tme, a compettve strength gven that those bankng groups concentrated ther actvty n ther home markets, whch could have gven them a hgh reputaton among the customers they satsfed. It s frequently argued that, untl the early 1980s, savngs banks and credt unons benefted from some market power. Nevertheless, after deregulaton took place and some frms were nvolved n the subsequent branchng expanson, ths prvleged poston has been eroded, forcng the frms to redesgn ther strateges. The clearest mplcaton of ths regulatory change has been a new compettve landscape and a completely dfferent market structure. The consequences of the expanson of actvtes n both the geographc and product dmensons takng place after deregulaton could also be nterpreted from a socologcal perspectve. Populaton ecology assumes that, n order for competton between populatons to exst, some overlap between the market nches occuped by those populatons s needed. In ths context, a market nche s defned by a bundle of resources that condtons frm s growth and survval (Hannan and Freeman, 1977).

6 22 Internatonal Journal of Busness and Economcs Gven the specfc stuaton n the Spansh bankng sector, the overlap between the nches of the three populatons (and thus, ther growth) has been affected by the evoluton of regulaton. A plausble nterpretaton s that whle t was n place, every populaton had ts market segment clearly defned and the degree of overlap was lmted. Thus, each populaton chose ther strateges ndependently and the nteracton between groups was scarce. Therefore, the man mplcaton of deregulaton could have been to ncrease the set of avalable market segments and thereby ncrease market overlap. 4. Sample, Methodology, and Results As mentoned above, the objectve of ths paper s to analyze the compettve nteracton among the three groups of fnancal ntermedares that competed n Span from January 1, 1971, to December 31, For emprcal purposes we consder the total assets of each populaton as our dependent varable (Barron et al., 1998) n order to take nto account all the varous actvtes performed by the banks. We use publc nformaton provded by the Bank of Span, whch shows the quarterly evoluton of assets for the three populatons. Consequently, we have 130 observatons. The avalablty of a suffcently long tme horzon s crucal n ths type of study snce the consequences of the compettve process can only be observed n the long term. Addtonally, by gong back as far as 1971, we can study the deregulaton process from the begnnng. As a result, we mnmze the possble bas due to omsson of mportant ndependent varables. Fgure 1 presents the evoluton of the total assets of the three groups for the perod 1971 to 2003 (n constant 2003 prces). Although the trend s clearly ncreasng for the three groups, there are mportant changes wth reference to ther relatve market share. For example, banks consttute the man group when we measure sze by total assets, wth a market share above 50%. However, ts mportance has decreased snce the early 1980s, when the savngs banks start to sharply ncrease ther share, ganng more than 10 ponts n only a few years. Fnally, credt unons have mantaned a relatvely stable partcpaton, always bellow 4%, although ther relatve growth s greater than ether compettor (ther market share n terms of deposts or bank branches s hgher; currently they have above 10% of the network and receve 7% of total deposts). To better understand the evoluton of bankng assets, t s mportant to take nto account some of the prevously mentoned bankng laws. Regulaton has been especally mportant n drvng populaton behavor, wth the consequences for effcency. It s evdent that the present strategy s clearly condtoned by hstorcal condtons, and the startng pont (and startng resources and capabltes) s one of the man factors that determne future behavor. The resource-based vew of the frm (Barney, 1991; Peteraf, 1993), for example, establshes that to acheve sustanable compettve advantages, frms need to use resources that are non-mtable (Barney, 1991), and hstory s clearly dffcult to mtate.

7 Luco Fuentelsaz, Jame Gómez, and Vctora Lucea 23 Fgure 1. Evoluton of Total Assets of Banks, Savngs Banks, and Credt Unons, (Thousand Euros, Constant 2003 Prces) Total assets We estmate equaton (7) wth seemngly unrelated regresson (SURE). The man advantage of ths technque s that, n contrast to tradtonal regresson methods, t s possble to consder the exstence of correlaton between the dfferent equatons and the nterdependences arsng among them. Thus, ths method provdes us wth coeffcents that are consstent and more effcent than the ones resultng from separate estmaton of each equaton. Table 1 shows SURE estmates of the ntegral verson of the partal adjustment model proposed n Secton 2. Global ft, measured by the R-squared statstc, exceeds 0.99 n all the three models. Importantly, the results of the Breusch-Pagan test for ndependence ndcate that the absence of correlaton among the three equatons s rejected, and therefore the coeffcents presented n Table 1 are consstent and more effcent than those resultng from separate estmaton of each equaton. The model not only ncludes nteractons among the three populatons consdered but also takes nto account the nfluence of GDP on carryng capacty, y (, and explores the possblty of a structural change n the nteracton coeffcents. Gven that the lberalzaton took place progressvely along several years, t s dffcult to establsh the exact moment where ths structural change took place. However, there are several reasons to accept that t occurred around Begnnng then, savngs banks were allowed to open branches natonwde and many made wde use of ths possblty (Fuentelsaz and Gómez, 2001). As a consequence, the savngs bank branch network ncreased from 14,994 n 1989 to 20,871 at the end of In the same perod, commercal banks reduced ther network from 16, Year Banks Savngs Banks Credt Unons

8 24 Internatonal Journal of Busness and Economcs to 14,074 branches, whle the number of branches for the credt unons grew from 2,890 to 4,460. Ths was also the year n whch the Bank of Span establshed a calendar to defntely elmnate the nvestment coeffcent. Apart from ths, n 1987 all nterest rates and commssons became completely free. These measures completed the deregulaton process wth reference to bankng actvtes and branchng restrctons. The results of F-tests for the possblty that the structural change took place n 1985, 1989, and 1993 support ths concluson. Table 1. Interacton between Populatons wthn the Spansh Bankng Sector, ; Integral Equaton Estmates Banks Savngs Banks Credt Unons Constant (6.39) (5.24) (2.31) Banks (8.45) ( 3.69) ( 2.41) Savngs Banks (4.51) (41.17) (2.72) Credt Unons (4.67) (3.94) (26.38) GDP Growth ( 4.92) ( 4.82) ( 0.76) 1989 Dummy ( 1.76) (0.28) ( 2.69) Banks 1989 Dummy (1.38) (0.42) (2.45) Savngs Banks 1989 Dummy ( 0.56) ( 3.11) ( 0.27) Credt Unons 1989 Dummy (0.12) (2.35) ( 0.90) 2 R Notes:,, and denote sgnfcance at 1%, 5%, and 10% levels; t-ratos are shown n parentheses. As mentoned n Secton 2, the estmatons presented n Table 1 are not vald for our purposes and have to be transformed nto competton coeffcents n order to draw conclusons n terms of rvalry. Therefore, gven our nterest n examnng whether competton between the three groups of ntermedares has been affected by deregulaton, we focus attenton on the dfferental equvalents to the nteracton coeffcents presented n Table 1. Table 2 shows the nteractons among ntermedares n the Spansh bankng sector resultng from the converson of nteracton coeffcents n Table 1 to competton coeffcents (Tuma and Hannan, 1984). Panel A presents the results for the perod before deregulaton, hghlghtng the nteractons that were sgnfcant n the SURE estmaton. Although we focus attenton on competton coeffcents whose ntegral equvalent was sgnfcant, the lterature s not clear on whether ths

9 Luco Fuentelsaz, Jame Gómez, and Vctora Lucea 25 should be the case. Smlarly, Panel B centers on the post-deregulaton perod n order to show the competton coeffcents wth changed values (the hypothess that all nteracton coeffcents were the same for both the regulated and deregulated perods was rejected). Our orgnal nterest n ths artcle was to test whether deregulaton has modfed the compettve condtons among groups of ntermedares operatng n the Spansh bankng sector. The comparson of Panels A and B provdes us wth a frst answer. Snce only two competton coeffcents are sgnfcant n Panel B, deregulaton seems not to have sgnfcantly changed competton among groups. In fact, all coeffcents n Panel A are negatve, what leads us to conclude that the three groups of ntermedares were competng n the regulated perod. After deregulaton ths stuaton only changed for two relatonshps: (1) savngs banks began to experence a postve nfluence from credt unon growth and (2) the growth n bank assets ncreased the carryng capacty for credt unons. Table 2. Interacton between Populatons wthn the Spansh Bankng Sector, ; Dfferental Equaton Estmates Panel A: Regulated Perod ( ) Banks Savngs Banks Credt Unons Banks Savngs Banks Credt Unons Note: denotes sgnfcance n SURE estmaton. Panel B: Deregulated Perod ( ) Banks Savngs Banks Credt Unons Banks Savngs Banks Credt Unons Note: denotes sgnfcance n SURE estmaton. A close analyss of the patterns of competton reveals some dfferences n the ntensty of the nteractons. For example, the pressure exerted by savngs banks on banks s nearly double the one of banks on savngs banks (note that the assets of banks, savngs banks, and credt unons are sgnfcantly dfferent n magntude, whch condtons the nterpretaton of competton coeffcents). Interestngly, the non-sgnfcance of the correspondng competton coeffcents n Panel B means that ths relatonshp has been mantaned throughout the perod 1971 to Apart from the relatonshp between the two larger groups of ntermedares (banks and savngs banks), an nterestng queston that arses s whether credt unons compete wth the other ntermedares (see for example Emmonds and Schmd, 2000) and, therefore, have to be ncluded n the analyses performed. Gven the sgnfcance of competton coeffcents n the relatonshps before and after deregulaton, the

10 26 Internatonal Journal of Busness and Economcs concluson s that credt unons do affect the actvtes of both banks and savngs banks. Nevertheless, as mentoned before, t s necessary to underscore that ths relatonshp has not been compettve across all years n our sample. 5. Conclusons Ths research analyzes the compettve process among the fnancal ntermedares that have operated n the Spansh bankng system for the last 30 years. Ths long perod s especally nterestng because t ncludes a number of lberalzaton measures that have profoundly affected bankng actvtes and bankng frms. In fact, t has tradtonally been argued that rvalry before deregulaton was scarce and that the new rules have completely changed the compettve landscape wthn the ndustry. Our analyss departs from conventonal economc approaches by applyng a socologcal model borrowed from populaton ecology and performs SURE estmaton to consder the possble nterdependences among the dfferent types of frms that compete n the market. It s also mportant to hghlght that, contrary to prevous studes, we do not analyze ndvdual frms; nstead, we consder rval populatons that share some common elements. Ths homogenety wthn the groups (and heterogenety across groups) has lead some prevous work to consder them ndependently when analyzng rvalry n the bankng ndustry. Although most researchers agree that n recent years convergence has been evdent, they typcally consder only wthn-group rvalry when evaluatng competton durng the 1980s and early 1990s. Our emprcal analyss allows us to extract some nterestng conclusons concernng rvalry n the bankng ndustry. Frst, we confrm that deregulaton has changed compettve condtons wthn the ndustry. Our results show a clear structural change n compettve nteractons among the dfferent populatons that, n the case of Span, can be located around Second, t s also mportant to hghlght that ths structural change has not been as dramatc as s sometmes argued. Although t s true that there s a dfferent compettve pattern before and after 1989, the nteractons between the three consdered populatons were mportant even before Our results show, for example, that savngs banks and credt unons were ferce compettors before 1989, whle the effect of credt unons on savngs banks has been postve snce However, ths symbotc behavour after deregulaton s not surprsng gven that, before 1989, both populatons often competed for the same market nches wth no possblty of expandng actvty outsde ther home markets. Once the restrctons were lfted, both groups (especally the savngs banks, whose geographc expanson has been stronger) expanded ther market domans, dmnshng competton n the local markets where they tradtonally competed. Fnally, our results also suggest that we must be careful when we evaluate bankng strateges or effcency. Very often, these types of analyss are conducted only for a group of ntermedares wth the mplct argument that the frms have

11 Luco Fuentelsaz, Jame Gómez, and Vctora Lucea 27 dfferent objectves, hstory or market nches. To the extent that competton (and thus, bankng strateges or effcency) s affected by the nteracton among several populatons, the solaton of only one of these groups may bas the results obtaned. Of course, our analyss has been performed n a context that may be dfferent from other bankng systems. However, t s mportant to note that Span s one of the European countres where bankng actvtes had tradtonally been more clearly segmented. Thus, f even n cases such as Span the nteracton among populatons are sgnfcant, future bankng studes should nclude all relevant bankng ntermedares n order not to bas conclusons. Fnally, note that snce our analyss does not consder ndvdual banks, our results are not drectly comparable to studes that evaluate bankng behavour at the frm level, and the varables whch explan frm competton may be dfferent from the ones used n ths research. Therefore, our nvestgaton should be vewed as a complement and not a substtute of more tradtonal research, for example, n ndustral organzaton economcs. References Barney, J. B., (1991), Frm Resources and Sustaned Compettve Advantage, Journal of Management, 17, Barron, D. N., E. West, and M. T. Hannan, (1994), A Tme to Grow and a Tme to De: Growth and Mortalty of Credt Unons n New York, , Amercan Journal of Socology, 100(2), Barron, D. N, E. West, and M. T. Hannan, (1998), Deregulaton and Competton n the Fnancal Industry, Industral and Corporate Change, 7(1), Emmonds, W. R. and F. A. Schmdt, (2000), Bank Competton and Concentraton: Do Credt Unons Matter? Federal Reserve Bank of St. Lous, May/June, Fuentelsaz, L. and J. Gómez, (1998), Lberalzacón y ámbto geográfco de actuacón, Papeles de Economía Española, 74-75, Gual, J. and D. Neven, (1993), Deregulaton of the European Bankng Industry ( ), European Economy/Socal Europe, 3. Hannan, M. T. and J. Freeman, (1977), The Populaton Ecology of Organzatons, Amercan Journal of Socology, 82, Jayaratne, J. and P. E. Strahan, (1997), The Benefts of Branchng Deregulaton, FRBNY Economc Polcy Revew, December, Strahan, P. E., (2003), The Real Effects of U.S. Bankng Deregulaton, The Federal Reserve Bank of St. Lous, July/August, Sturm, J. E. and B. Wllams, (2004), Foregn Bank Entry, Deregulaton and Bank Effcency: Lessons from the Australan Experence, Journal of Bankng and Fnance, 28(7), Tuma, N. B. and M. T. Hannan, (1984), Socal Dynamcs: Models and Methods, Academc Press.