4 de Trabajo 4 - FBBVA · Todos los documentos de trabajo están disponibles, de forma gratuita y en for-mato PDF, en la web de la Fundación BBVA. Si desea una copia impresa, puede

4Documentosde Trabajo4Documentos

de Trabajo2007

José Manuel Pastor MonsálvezEmili Tortosa-Ausina

Social Capital andBank PerformanceAn International Comparison for OECD Countries

Gran Vía, 1248001 BilbaoTel.: 94 487 52 52Fax: 94 424 46 21

Paseo de Recoletos, 1028001 MadridTel.: 91 374 54 00Fax: 91 374 85 22

[email protected]

2007-04 9/2/07 11:17 Página 1

Social Capital and Bank PerformanceAn International Comparison for OECD Countries

José Manuel Pastor Monsálvez 1,3

Emili Tortosa-Ausina 2,3

1 U N I V E R S I T Y O F V A L E N C I A2 U N I V E R S I T Y J A U M E I

3 I N S T I T U T O V A L E N C I A N O D E I N V E S T I G A C I O N E S E C O N Ó M I C A S (Ivie)

� AbstractOver the last few years there has been a remarkableincrease in the number of published studies dealingwith social capital issues. A plethora of studies hasalso analyzed the efficiency of several banking indus-tries. In this article we merge the two literatures byanalyzing how social capital affects bank efficiencyfor a sample of financial institutions in OECD coun-tries. The analysis is performed using activity analy-sis techniques, and social capital is controlled for byentering the analysis as an environmental variable. Akey feature of our study is the higher complexity ofthe social capital measure used, compared to othersimpler measures hitherto considered in the litera-ture. Results suggest that disregarding the effect of so-cial capital can be irrelevant for some financial insti-tutions, yet the effect cannot be overlooked for othersthat operate in low-social-capital environments. Inthese cases, efficiency scores are biased downwards,and controlling for social capital enables these banksto move up in the efficiency rankings.

� Key wordsBanking firm, efficiency, environmental conditions,social capital.

� ResumenEn los últimos años ha habido un aumento notableen el número de artículos que tratan diversos aspec-tos relacionados con el concepto de capital social.Asimismo, existe gran cantidad de estudios que hananalizado la eficiencia de las empresas bancarias. Eneste documento de trabajo combinamos las dos lite-raturas, analizando cómo el capital social de las eco-nomías afecta a la eficiencia bancaria para unamuestra de instituciones financieras de distintos paí-ses de la OCDE. El análisis se realiza usando activityanalysis, y el capital social se controla considerándo-lo en el análisis como una variable ambiental. Unacaracterística importante de este trabajo es la mayorcomplejidad del indicador de capital social utilizado,comparada con otras medidas mucho más simpleshasta ahora consideradas en la literatura. Los resulta-dos sugieren que no considerar el efecto de capitalsocial puede ser irrelevante para algunas institucio-nes financieras. Sin embargo, el efecto no puede serpasado por alto para otras que funcionan en ambien-tes con bajo capital social y cuyos índices de eficien-cia están sesgados a la baja.

� Palabras claveEmpresa bancaria, eficiencia, condiciones ambienta-les, capital social.

Al publicar el presente documento de trabajo, la Fundación BBVA no asume res-ponsabilidad alguna sobre su contenido ni sobre la inclusión en el mismo dedocumentos o información complementaria facilitada por los autores.

The BBVA Foundation’s decision to publish this working paper does not imply any re-sponsibility for its content, or for the inclusion therein of any supplementary documents orinformation facilitated by the authors.

La serie Documentos de Trabajo tiene como objetivo la rápida difusión de losresultados del trabajo de investigación entre los especialistas de esa área, parapromover así el intercambio de ideas y el debate académico. Cualquier comenta-rio sobre sus contenidos será bien recibido y debe hacerse llegar directamente alos autores, cuyos datos de contacto aparecen en la Nota sobre los autores.

The Working Papers series is intended to disseminate research findings rapidly amongspecialists in the field concerned, in order to encourage the exchange of ideas and academ-ic debate. Comments on this paper would be welcome and should be sent direct to theauthors at the addresses provided in the About the authors section.

Todos los documentos de trabajo están disponibles, de forma gratuita y en for-mato PDF, en la web de la Fundación BBVA. Si desea una copia impresa, puedesolicitarla a través de [email protected].

All working papers can be downloaded free of charge in pdf format from the BBVAFoundation website. Print copies can be ordered from [email protected].

Social Capital and Bank Performance: An International Comparison for OECD Countries© José Manuel Pastor Monsálvez and Emili Tortosa-Ausina, 2007© de esta edición / of this edition: Fundación BBVA, 2007

EDITA / PUBLISHED BY

Fundación BBVA, 2007Plaza de San Nicolás, 4. 48005 Bilbao

DEPÓSITO LEGAL / LEGAL DEPOSIT NO.: M-9.310-2007IMPRIME / PRINTED BY: Rógar, S. A.

Impreso en España – Printed in Spain

La serie Documentos de Trabajo de la Fundación BBVA está elaborada con papel 100% reciclado,fabricado a partir de fibras celulósicas recuperadas (papel usado) y no de celulosa virgen, cumplien-do los estándares medioambientales exigidos por la legislación vigente.

The Working Papers series of the BBVA Foundation is produced with 100% recycled paper made from recov-ered cellulose fibre (used paper) rather than virgin cellulose, in conformity with the environmental stan-dards required by current legislation.

El proceso de producción de este papel se ha realizado conforme a las normas y disposicionesmedioambientales europeas y ha merecido los distintivos Nordic Swan y Ángel Azul.

The paper production process complies with European environmental laws and regulations, and has bothNordic Swan and Blue Angel accreditation.

La serie Documentos de Trabajo, así como información sobre otras publicaciones de laFundación BBVA, pueden consultarse en: http://www.fbbva.es

The Working Papers series, as well as information on other BBVA Foundation publications,can be found at: http://www.fbbva.es

C O N T E N T S

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2. On the Links between Social Capital and Bank Performance . . . . . 8

3. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.1. Measuring efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.2. Social capital as an environmental variable . . . . . . . . . . . . . . . . . . . . . 163.3. Comparing efficiency distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

4. Data and Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

5. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

About the Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

1. Introduction

SOCIAL capital—broadly, social networks, the reciprocities that arisefrom them, and the value of these for achieving mutual goals—has be-come an influential concept in understanding and debating the modernworld (Baron, Fiel and Schuller, 2000, p.1). In the specificities of the econom-ic arena, some authors argue that, although it is also becoming an influen-tial concept, some controversies still act as a major barrier to its general per-vasiveness. In particular, according to Durlauf (2002), different opinions areoffered on its precise meaning—i.e., whether it refers exclusively to a formof social networks, or whether it should be equated to trust and trustworthi-ness—, the notions that the social relations referred to by social capital arethemselves problematic, and that even when definitions are agreed upon,no consensus exists on how to measure social capital. Most of the problemsraised also deal with the way social capital shifts the focus of analysis fromthe behavior of individual agents to the pattern of relations between agents,social units and institutions, which constitutes a minor revolution in thefield of economics (Baron, Fiel and Schuller, 2000: 35).

In spite of these threatening points of view, and the fact that only a mi-nority (albeit growing) of economists are familiar with it, social capital ap-pears to be both quantifiable and, as shown by some recent empirical applica-tions (see, for instance Hall and Jones, 1999; Annen, 2003; Grafton, Knowelsand Owen, 2004; Beugelsdijk and Van Schaik, 2005) related to traditional eco-nomic variables like GDP growth rates (Fine and Green, 2000). Indeed,acording to Fukuyama (1995), if social capital can be set alongside the othercolumns of physical, financial and human capital, the result might be a moreintegral knowledge of both the economic and non-economic worlds. There-fore, many studies have used the concept as a productive resource emergingfrom the social links among individuals, granting them greater benefits and op-portunities than those enjoyed by members of other societies in which thoselinks do not exist, or are weaker. In such circumstances, social capital would beconceived as a set of intangible circumstances such as values, norms, attitudes,trust, social networks, etc., which allow a society to perform more efficiently.

Although Hanifan (1916) coined the concept of social capital almostone century ago, its use by economists has been intensified only over the

5

last few years to explain a variety of economic phenomena such as econom-ic growth (Zak and Knack, 2001; Iyer, Kitson and Toh, 2005), labor pro-ductivity (Hall and Jones, 1999), or the development of financial systems(Ongena and Smith, 2000; Guiso, Sapienza and Zingales, 2004). Most ofthese research studies suggest that social capital impacts positively on theeconomic phenomena under analysis, influencing and conditioning the po-tential performance of economies. Therefore, the results obtained in the cit-ed research studies reveal that social capital is an important issue to controlfor when forecasting the full potential of economies.

In the case of banking, a broad literature exists on a particulartype of social capital which originates and evolves between the financialinstitution itself and its customers. This type of capital emerges throughthe establishment of repeated and/or lasting transactional relations,which increase the knowledge the two parties have of each other, and gen-erates the trust and positive attitudes typically associated with social cap-ital. The repeated interaction between lender and borrower boosts theirconfidence in one another, thus strengthening social ties, and leading toan increase in social capital. The result is not only beneficial for the cus-tomer, but also for the financial institution, since the information advan-tages generated through transactional ties between lenders and borrow-ers lead to increased effectiveness in its accomplishment. In addition tothis, social ties generate social capital which factors in as an importantinput for future transactions not only in the credit decision-making pro-cess, but in many other decision-making processes. The increased levelof social capital improves the conditions under which financial contractsare fulfilled.

This approach, known as relationship banking, seeks to analyze whetherthe preferential treatment generated through repeated interaction affects avariety of aspects of banking activity. As such, it is regarded as an importantdeterminant of financial systems’ performance in developed countries and,consequently, a number of studies have analyzed a variety of related issuessuch as, how relationship banking affects the competitive conditions in theindustry (Degryse and Ongena, 2005), accessibility and/or credit rationing(Ferri and Messori, 2000), the cost of credit (Degryse and Ongena, 2005),or the amount of loan losses (Ferrary, 2003).

However, this paper seeks to analyze an economic phenomenon bare-ly explored so far in the literature: by adopting a different perspective tothat commonly considered in relationship banking studies, we analyze wheth-er the level of social capital in different countries has an impact on the ef-ficient provision of banking services. Therefore, we do not attempt to ana-

josé manuel pastor monsálvez and emili tortosa-ausina

6

lyze bank efficiency from a relationship banking point of view 1 but to ana-lyze whether the alleged positive impact of social capital on a variety ofeconomic phenomena also applies to bank efficiency.

Our sample is made up of the banking industries in 28 OECD coun-tries, containing data on commercial banks, savings banks and creditunions. Our results are manifold, but the most general findings suggest thatthe higher the level of social capital in the society, the higher the bank costefficiency measured. Put differently and from a social capital perspective,larger amounts of social capital available in each society lead to cost savingsin the intermediation process. We find it is essential to control for social cap-ital in order to unbiasedly assess the performance of each bank in oursample.

The study proceeds as follows. After this introduction, section 2 justifypresents the rationale for relating social capital to bank performance, alongwith its impact on efficiency. Section 3 describes the methodology employed,while section 4 and section 5 results present the data and results, re-spectively. Finally, section 6 details the most relevant conclusions.

social capital and bank performance: an international comparison for oecd countries

7

1. The existence of repeated and/or lasting ties between each bank and its clients may affect itsefficiency. However, this issue lies beyond the scope of our paper.

2. On the Links between SocialCapital and BankPerformance

THE literature on relationship banking 2 demonstrates that repeated andlasting interaction between the bank and the borrower facilitates monitor-ing and screening, and can overcome problems of asymmetric information(Boot, 2000), i.e., it has an impact on many aspects of banking activity. How-ever, banks not only benefit from the social capital generated throughouttheir relations with clients, but also from the existing social capital in the so-ciety in which the bank operates. This section explores the different ways bywhich the social capital in each society affects bank activity and bank perfor-mance.

The most direct way in which social capital impacts on banking activ-ity in general, and banking performance in particular, is through the in-crease in the confidence and trust of the individuals participating in differentbanking relationships in the institutions and systems that control the social,economic and political welfare in the society. Accordingly, for economieswith high levels of social capital, individuals have a mutual trust, either dueto the moral attitudes learned during their education, or because of existing so-cial and legal mechanisms that penalize default behaviour—in this case, breachof contracts—. As we shall see, this mutual trust has positive effects, not only forthe bank and its clients, but also for the society considered, as a whole.

Because of their very nature, banking contracts belong to a specialtype of contract where trust between parties is essential. Each depositor ex-pects the bank not only to pay interests in due time and form, but also thatthe money in her/his bank account will be refunded upon request. Like-wise, regarding the loan contracts, the bank expects the borrower to repaythe amount lent in accordance with to the terms stipulated in the agree-

2. See, for instance, the articles in the special issue of the Journal of Financial Intermediation devot-ed to the topic (volume 9, 2000), or the review by Ongena and Smith (2000).

8

ment. In both cases we can verify that each side trusts the other, i.e., bothparties expect that the other will repay the money under the terms stipu-lated in the agreement. Obviously, this type of fund exchange rests on aninstitutional and legal framework which prosecutes breach of contract; how-ever, this breach of contract is also penalized by the mutual trust betweenparties, which directly depends on the social capital in the society. Accordingto this reasoning, we may conclude that the higher the amount of socialcapital in the society, the higher its (positive) influence, not only on the ob-servance of bank contracts, but also on the number of relationships (Guiso,Sapienza and Zingales, 2004).

Defining social capital as people’s trust in institutions and the systemenables us to understand the reasons why in low-social-capital societieswhere the level of financial development is relatively low, financial transactionsare more intense within narrow subgroups, such as families and friends (Fu-kuyama, 1995; Banfield, 1958). In short, social capital is an essential inputfor the efficient performance of the banking system, since its existence has apositive impact on the legal enforceability of banking contracts, which con-tribute to easing the transmission of funds from the lender to the borrower,leading to an improvement in the environmental conditions in which banksoperate, which may ultimately influence their efficiency.

More specifically, multiple paths exist by which social capital may in-fluence banking activity and, ultimately, bank efficiency. Some of them em-phasize the positive impact of social capital on the environment in whichbanking activity takes place, by reducing certain costs. Others relate to theincreased turnover for banking firms due to the higher level of social capitalin the society. Explanations for some of these paths follow.

a) Social capital reduces information, transmission and monitoring cost: re-ducing information costs is one of the key factors related to the relevance ofsocial capital. As Levine (1997) suggests, information and transaction costsare lower in developed financial markets. The empirical evidence showsthat, in communities where access to financial markets is difficult, the neces-sary information is obtained through social capital in the guises of networksand internal alliances within the community (Morduch, 1999; Ferrary,2003). On the other hand, in developed financial systems, the task of re-ducing information asymmetries is performed by banks, which compile andprocess information. In social-capital-intensive economies, information re-quirements for both the lender and the borrower are lower, being replacedby the trust in markets and institutions. In addition, complementary mecha-nisms are in place provided by public authorities, such as socializing


9

information through the creation of national bad debtors’ (defaulters) files,and firms’ ratings which enable both parties to obtain information about eachother, facilitating a reduction of information asymmetries (Ferrary, 2003).According to these arguments, cæteris paribus, the higher the level of social capi-tal in the society, the lower the information, monitoring and transaction costsassociated to the bank operations which are necessary to fulfil contracts.

It is also argued (Vega-Redondo, 2006) that social capital and socialnetworks reduce information costs due to the existence of network econo-mies, which simplify the inherent complexity in the relationships among in-dividuals, and increase the fluency with which information is transmittedwithin the social network. Therefore, in segmented and isolated societies, inwhich there are no ties between individuals of social groups, the costs tobanks of obtaining information on their clients cannot be spread out to ob-tain information on other clients or groups of individuals, due to the ab-sence of relationships between them. In contrast, in cases where ties betweenindividuals are strong, the resources devoted by the bank to increasing itsknowledge about its clients pay dividends in terms of the information in gath-ers on other potential clients with whom the bank’s clients have ties, there-fore leading to cost reductions.

b) Social capital reduces risk premium, thus lowering financial and credit costs:another way by which social capital influences bank performance relates tothe reduction of the risk premium that participating parties mutually re-quire to offset any eventual breach of contract by either of them. Each partynot only establishes bonds of mutual trust—whose strength increases pro-portionally to the intensity of the relationship in terms of breadth, repeti-tion, and duration (relationship banking)—but also trusts that the legalenforceability of contracts is guaranteed by institutions and the system in gen-eral (social capital), which will stave off breach of contract—or at least theprobability of this happening—. Thus, depositors will trust the bank to ob-serve the deposit contract, paying back the deposited funds and the in-terests according to the terms of the contract and, therefore, will ask for alower risk premium to put the money in the bank, which will reduce the in-terest rate on deposits and, therefore, the financial costs for banks.

From the bank point of view, in social-capital-intensive societies, thebank trusts that the borrower will repay the amount lent in accordance withterms stipulated in the loan agreement, contributing to reducing the riskpremium, and therefore to reduce the cost of credit for the borrower. In ad-dition to this, we may expect this credit cost reduction to encourage the de-mand for credit.


10

Previous empirical evidence has been reported by Petersen and Rajan(1994, 1995) and, more recently, by Elyasiani and Goldberg (2004), whoanalyze how the ties (relationship social capital) between the bank and theborrower affect both supply and credit costs. Their results corroborate thehypothesis that the higher the level of social capital, the higher the supplyof credit and the lower its cost will be.

c) Social capital reduces loan losses: higher levels of social capital in thesociety not only lead to lower information costs, but also to higher qualityinformation collected to assess risk insolvency. Thus we may expect that thehigher the level of social capital, via improvements in the quality of informa-tion, the greater the decline of loan losses will be. On this point, Ferrary(2003) states that, to reduce risk evaluation uncertainty, the bank analystcomplements objective methods or, if these are unavailable, substitutesthem with information acquired through informal relations based on trust.The quality of risk evaluation and the reduction of asymmetric informationrisk will depend upon the quality of the social capital existing in the envi-ronment in which the analyst operates, as well as on his/her fitting into thesocial networks in this environment.

However, improvements in the quality of information are not the onlyways by which social capital affects loan losses, since the social costs (isola-tion and/or exclusion from the community) associated to failure to pay couldbe so damaging to the borrower that they discourage default behavior 3.

Several studies have analyzed the existence of an empirical link be-tween social capital and bad, or doubtful, bank loans. On this point, Karlan(2004) finds evidence that higher levels of social capital lead to higher re-payment and higher savings. Likewise, Ferri and Messori (2000) show thatbanks in the Northeast and Center of Italy with a strong orientation towardrelationship banking, and therefore, a higher level of social capital, have alower incidence of bad and doubtful loans.

d) Social capital increases loan supply and diminishes rationing: in somecases, reducing information asymmetries through information gatheringand processing is not profitable for the bank, since loan costs (basically ob-taining and processing information) are high compared to the expectedprofitability of the transaction, and the risks of the transaction are excessive.


11

3. Karlan (2004) observed direct evidence that members monitor loan performance through so-cial networks, that relationships deteriorate after default, and that through successful monitor-ing, individuals know who to punish and who not to punish after default.

In addition, not many individuals or firms have a sufficiently large volumeof assets to offer as collateral in the case of inability to pay back a loan. Inthese cases, banks, which have the ability to collect and process information,could decide not to make the loan, generating credit rationing.

In line with the above rationale, social capital contributes to reducinginformation asymmetries and costs, and thus its existence can make transac-tions profitable which otherwise would be loss-making. Therefore, social cap-ital contributes to increasing loan supply and reducing credit rationing 4.

However, although banks benefit from the social capital in the societythrough the increase in their turnover, small businesses and marginal com-munities, which usually have credit restrictions, will also benefit from theexistence of social capital, since it enables them to obtain credit more easily(less rationing), and a lower cost, since the risk premium also diminishes.

Petersen and Rajan (1994, 1995), Ferri and Messori (2000) andElyasiani and Goldberg (2004) corroborate that the higher the social capi-tal, the higher the credit supply will be.

e) Social capital encourages customers’ use of bank products: as commentedon earlier, all banking contracts can be considered as a confidence rela-tionship in which the principal trusts the agent. Thus, when someone ac-cepts a payment using a check or a credit card, the principal trusts that theagent is not out of funds. The fulfilment of the check contract depends onthe level of trust in the agent which, simultaneously, depends on the level ofsocial capital. Therefore, cæteris paribus, individuals living in low-social-capitalenvironments will use bank products to a lesser extent, due to its trust-inten-sive nature (Guiso, Sapienza and Zingales, 2004).

From the perspective of portfolio selection, assets depend not only ontheir intrinsic risk, but also on the probability of them being expropriatedand, therefore, on the required level of trust. Guiso, Sapienza and Zingales,(2004) use this rationale to conclude that when the level of social capital(and trust) is low, households will invest larger shares of their portfolios inlow-trust-intensive assets such as cash and smaller shares in high-trust-inten-sive assets such as bank deposits, firms’ shares, etc.

For the aforementioned reasons, since both traditional bank outputs(such as bank deposits and loans) and non-traditional bank outputs (usuallyproxied by fee-income) are high-trust-intensive contracts, we may expect


12

4. From the demand side, we must consider that the decrease in the cost of credit due to thedecrease in the premiums for risk associated to higher levels of social capital generates an in-crease of credit demand.

that their use will be higher for households and firms in high-social-capitalsocieties. On the other hand, in low-social-capital societies we may expectmore intense reliance on transactions within narrow subgroups, such as fam-ilies and friends (Guiso, Sapienza and Zingales, 2004).

Banks will also benefit from the level of social capital in the societydue to its positive influence on firms and households’ demand for bankproducts, which will enable them to issue more deposits, make more loansand raise more fee income in charges for the services rendered.

The empirical evidence on this matter is robust. Historically, trust cri-ses in the guises of panic runs have led to the use of cash to the detrimentof deposits (e.g., 1929 crisis). Guiso, Sapienza and Zingales (2004) resultalso show a positive and significant link between social capital and the use ofchecks and credit cards, and a negative and significant link between socialcapital and the use of cash.

In summary, in these lines we have described different ways by whichsocial capital impacts on banking activity. Social capital reduces monitoring,information, transactional, financial and loan loss provisions costs, and in-creases credit supply as well as any other types of banking products, in gen-eral high-trust-intensive products. All this implies that bank efficiency will behigher in high-social-capital economies.

It must be noted that the aforementioned effects have an impact onthe cost side of bank activity; in some cases due to the reduction in the quan-tity of inputs required for the bank to operate—cases a) and c)—; in othersreducing the price of a certain input, such as the financial costs for deposits—case b)—. In cases d) and e) it has a positive impact on bank outputs. In thatrespect, the appropriate measure of efficiency is cost efficiency.

Graphic 2.1 provides a graphic illustration of the ways by which socialcapital impacts on bank cost efficiency. The process of intermediation con-sists basically of obtaining a range of bank outputs (e.g., credits and loans,y1; deposits, y2; other earning assets, y3; fee-income, y4) using some inputs (xi,i = 1, ..., N) for which prices are charged (wi, i = 1, ..., N), in such a way thatcosts are raised (e.g., personnel expenses, w1x1; general expenses, w2x2; fi-nancial costs, w3x3). The process by which inputs turn into outputs is effi-cient when, through input prices and technology, the bank obtains outputsat the lowest costs.

The impact of social capital turns out to be crucial in this process, sincethe higher the social capital in the society: (a)The lower the informationcosts, leading to a reduction in the personnel expenses (w1x1) and generalexpenses (w2x2;), required to obtain such information; (b) the lower the riskpremium required by depositors, which reduces the financial costs for the


13

bank (w3x3) and, therefore, also reduces the cost of credit for the financier,leading to a beneficial effect on the credits and loans made by banks (y1);(c) the lower the number of bad and doubtful loans (loan losses), thereforenot only loan loss provisions will decrease, but also the personnel expenses(w1x1) and administration expenses (w2x2) involved in recovering bad loanswill be lower; (d) the higher the loan supply (y1), since it reduces asymme-tries and information costs; and (e) the higher the use of bank products, lead-ing to a positive effect on traditional bank outputs (credits, loans and depos-its, i.e., y1, y2, y3), and a positive effect on non-traditional bank outputs(usually proxied by fee-income, y4).


14

GRAPHIC 2.1: Social capital and cost efficiency

w1x1 = Personnel expenses

w2x2 = General expenses

w3x3 = Financial costs

Inputs/Costs

Outputs

y1 = Credits/Loans

y2 = Deposits

y3 = Other earning assets

y4 = Fee income

TECHNOLOGY

Social capital

a) � Information costs

b) � Risk premium

c) � Credit losses

d) � Credit availability

e) � Use of banking products

3. Methodology

3.1. Measuring efficiency

Efficiency can be measured through a variety of different methods, whichfall into two broad categories: parametric and nonparametric methods. Thebiggest advantage of nonparametric methods is their flexibility, since nofunctional form has to be specified for the frontier or the error term. Howev-er, due to their deterministic nature, they cannot disentangle inefficiencyfrom random error. On the other hand, parametric methods allow for thisdisentanglement, but on the downside, they have to previously specify afunctional form for the unobserved frontier, as well as making assumptionson the distribution of the error term and the inefficiency.

No consensus exists as to which method is most suitable to measureefficiency. Several monographs have provided painstaking descriptions ofthe techniques that have been developed over the last three decades or so(see, for instance, Fried, Lovell and Schmidt,1993a). However, because re-sults are usually non-coincidental (Ferrier and Lovell, 1990), more recentmonographs have dealt exclusively with one of the two techniques. For in-stance, Lovell and Kumbhakar (2000) deal exclusively with efficiency viaStochastic Frontier Analysis (SFA), one of the most popular parametrictechnique methods. On the other hand, Färe and Grosskopf (2004) exclusi-vely examine Data Envelopment Analysis (DEA), the most popular techni-que in the nonparametric field 5.

However, the evolution of parametric and nonparametric techniqueshas not been entirely balanced. Up to the early nineties, both groups of tech-niques made significant progress, but newer proposals have leaned towardsthe nonparametric field. For instance, Cazals, Florens and Simar (2002) pro-posed a nonparametric estimator which is more robust to outliers than DEAor FDH (Free Disposable Hull, a version of DEA in which the convexity as-

15

5. However, due to the intense use of nonparametric methods to measure efficiency in the Oper-ations Research literature, several other contributions focus exclusively on the nonparametricfield, essentially DEA. See, for instance, Cooper, Seiford and Tone (1999), Cooper, Seiford andZhu (2004), or Thanassoulis (2001).

sumption is dropped). Aragon, Daouia and Thomas-Agnan (2005) presenteda nonparametric estimator for the efficiency frontier based on conditionalquartiles of an appropriate distribution associated with the production pro-cess. More recently, Martins-Filho and Yao (2006) have proposed a nonpara-metric estimator of frontiers which envelope data and is more robust to outliersthan previously proposed methods.

Therefore, following the path cleared by the most up-to-date tenden-cies, we measure bank efficiency using nonparametric methods. Althoughwe do not adopt the most recent techniques considered in the above para-graph, due to difficulties in handling data for input or output prices, state-of-the-art techniques are used since social capital will enter the analysis follow-ing the recent contribution by Ruggiero (2004). In addition to this, wecombine the nonparametric DEA technique with nonparametric statistical meth-ods, which are essential to the studies by Cazals, Florens and Simar (2002), Ara-gon, Daouia and Thomas-Agnan (2005) and Martins-Filho and Yao (2006).

3.2. Social capital as an environmental variable

Since Farrell (1957), the literature on efficiency measurement using nonpara-metric methods has grown impressively. Later, Charnes, Cooper andRhodes (1978) introduced what is known as Data Envelopment Analysismentioned above to compute Farrell’s measure of efficiency under constantreturns to scale (CRS). Since then, a number of contributions have beenpublished, enabling efficiency measurement under variable returns to scale(VRS) (Banker, Charnes and Cooper, 1984), using nonradial measures(Färe and Lovell, 1978; Zhu, 1996), or controlling for non-discretionary in-puts (Banker and Morey, 1986a, 1986b; Ray, 1991; Ruggiero, 1996, 1998,2004; Fried, Schmidt and Yaisawarng, 1999).

Section 2 explained the multiple paths by which social capital can in-fluence bank efficiency. We may consider social capital as an input of theproduction process that is beneficial for banks. From the methodologicalpoint of view, the problem lies in considering the social capital in each econ-omy as an additional input to the optimization problem. However, in con-trast to what occurs for other inputs, banks cannot discretionarily modifythe social capital of the environment in which they operate; thus, it must betreated differently to the rest of the inputs—i.e., as a non-discretionary in-put—. Several methods exist to account for environmental, or non-discre-tionary, variables in DEA (Fried and Lovell, 1996; Rouse, 1996). They canbe classified into two- and three-stage methods.


16

Banker and Morey (1986a, 1986b) proposed the most straightforwardmethod for this, based in one stage only. Their procedure consists of treatingdiscretionary inputs separately from non-discretionary inputs, disallowing ra-dial reductions for the latter, in an attempt to restrain the comparison only tofirms, or decision making units (DMUs) under the same, or worse, environ-mental conditions. This is the most direct and easily interpretable method,and it has therefore been used intensively. However, it has certain disadvan-tages, since the direction of influence for each variable must be known a priori.

Ruggiero (1996) refined Banker and Morey’s (1986a, 1986b) method,removing the convexity constraint on the non-discretionary inputs. In sodoing, non-discretionary inputs are treated as factors that determine the po-sition of the frontier. In other words, they operate as constraints by generat-ing multiple frontiers, excluding DMUs operating under more favorable en-vironments according to the non-discretionary variable.

Two-stage methods are also used in the literature. The most widely-employed two-stage procedure aims to explain the efficiency scores obtainedin the first stage by means of an ex post regression including a set of control vari-ables which may include environmental variables 6. This type of method hassome disadvantages. One is that censored models (such as TOBIT) are need-ed in the second stage to control for the fact that efficiency scores are boundedbetween (0,1) 7. However, further trouble arises because of using paramet-ric methods in the second stage, due to the fact that parametric methods pre-suppose independence of observations, but efficiency scores obtained via nonpara-metric methods in the first stage are dependent in the statistical sense—they areobtained via linear programming—. Simmar and Wilson (2006) provide agood exposition of the econometric problems derived from combining thesetwo types of methodologies and propose an alternative, the bootstrap method,to overcome them. Balaguer-Coll, Prior and Tortosa-Ausina (2006) provide asimpler method based on nonparametric regression and bivariate density esti-mation.

Finally, Fried and Lovell (1996) propose a three-stage procedure. Inthe first stage, a DEA model is used, including both inputs and outputs. Inthe second stage either DEA or SFA models can be used to control for theeffect of environmental variables. To do this, the slacks obtained in the firststage are corrected by the effect of non-discretionary variables. Finally, in


17

6. This method was first applied by Timmer (1971).

7. Instead of adjusting the efficiency scores, other authors proposed an alternative two-stage ap-proach that consists of adjusting the residuals (slacks) obtained in the first stage (see McCartyand Yaisawarng, 1993; Fried, Lovell and Vanden Eeckaut, 1993b).

the third stage the corrected variables are used to obtain the environment-adjusted efficiency scores.

In our study, a single-stage method is used, considering the refine-ment proposed by Ruggiero (2004). As mentioned earlier, the biggest draw-back of this procedure is that the direction in which the environmental var-iable considered affects efficiency must be known a priori. In our case, thedirection in which social capital affects efficiency is well known a priori, andhas also been theoretically justified in section 2 8.

To illustrate the methodology, let us assume there exist r = 1, ..., Rbanking firms. Let us suppose that bank i produces P outputs yi = (yi1, ..., yiP)∈ R++

P using M inputs xi = (xi1, ..., xiM) ∈ RM++, paying for them prices wi =

(wi1, ..., wiM) ∈ RM++. Cost efficiency for bank i can be measured by solving the

following linear programming problem, which compares bank i with the re-maining R-1 banks in the sample (Charnes, Cooper and Rhodes, 1978):

Min SM

m = 1 wim xim

s.t.SR

r = 1 lr yrp ≥ yip p = 1, ..., P

SR

r = 1 lr xrm ≤ xim m = 1, ..., M (3.1)

SR

r = 1 lr = 1, lr ≥ 0, r = 1, ..., R

The solution to this problem x*i = (x*

i1, ..., x*i M) corresponds to the opti-

mal input vector, i.e., the one that minimizes the outputs’ production costs,given input prices, and has been obtained by comparing bank i to a linearcombination of inputs which produces the same, or more, of each outputusing the same, or less, of each input. Optimal costs will be C*

i = SMm = 1 wiM x*

im

which, by definition, will be the same, or less than the observed costs fo


18

8. However, analyses have been carried out to determine whether the influence of social capitalon efficiency is positive or negative. Following Lozano-Vivas, Pastor and Pastor (2002), to deter-mine whether an environmental variable—in our case social capital—should enter the analy-sis as an input or an output we simply reverse its character. Thus, if we consider that social capi-tal has a positive influence, we should model it as an input (Cooper and Pastor, 1996); on the ot-her hand, if we think that social capital has a negative influence it should be entered as an out-put in the model. Therefore, the sign of the influence of social capital on efficiency is deter-mined by comparing both assumptions—i.e., including it as an additional input or output inan intertemporal model. Later, the statistical differences between mean efficiency scores obtainedwhen controlling for social capital—considering it as an input and as an output—and notcontrolling for social capital are tested for. The results obtained suggest that differences betweenmeans are only significant when social capital enters as an input in the model. In other words,social capital has a positive effect on efficiency.

bank i , namely, Ci = SMm = 1 wim xim . Therefore, the cost efficiency score for

bank i, CEi, can be expressed as the ratio of its observed costs and its opti-mal costs, namely:

where CEi ≤ 1.Graphic 3.1 illustrates this concept for the two inputs case. The

isoquant curve is made up of a set of efficient banking firms, namely, B, F, Gand their linear combinations. Given the relative prices for inputs, the min-imum, or optimal, cost (C*

i ) is achieved via a linear combination of B and F;since firm i has Ci costs, total cost efficiency (CEi) can be represented as thedistance (ratio) between both isocost lines.

However, this approach ignores the role of non-discretionary inputs,i.e., it disregards the circumstance that banks operating in different econo-mies might face varying environmental conditions (i.e., varying social capi-tal levels), which might have an impact on efficiency. For instance, let us as-sume bank i operates in an unfavorable environment, i.e., in alow-social-capital economy. For this bank, operating with a cost level as lowas C*

i is unfeasible. Comparing this bank to other banks operating in more


19

CEi =C*

i =S

M

m = 1 wim x*

im (3.2)

Ci SM

m = 1 wim xim

x1

x2

i

Ci

Ci*

B

F

GIsoq(Y)

CEi

GRAPHIC 3.1: Cost efficiency in DEA models

favorable environments, i.e., in high-social-capital economies, is not appro-priate, since results are biased because banks operating in unfavorable envi-ronments are being penalized. Graphic 3.2 illustrates this circumstance forbank i.

Banking firm i can be compared with banks H, J and K, which operateunder environmental conditions at least as favorable as those for firm i , i.e.,with the same, or lower levels of social capital. Thus, the truly feasible opti-mal (i.e., minimal) costs for firm i are those represented by the C*

iKS isocost,and its cost efficiency score would be represented by the distance betweenthe isocost lines C*

iKS and Ci.Considering social capital as a non-discretionary input enables us to

decompose total cost efficiency for bank i as follows:

where the first term in square brackets is the pure cost efficiency (cost effi-ciency adjusted by social capital) for bank i, PCEi, which can be defined as theratio between optimal costs, represented by the frontier made up of banksand linear combinations of banks operating under environments with thesame or a lower level of social capital (C*

iKS ), and observed costs of bank i, (Ci).


20

CEi = C*

i = [C*iKS ] [ C*

i ] (3.3)Ci Ci C*

iKS

PCEi SEi

{ {x1

x2

i

Ci

Ci*

B

F

GIsoq(Y)

CEi

Isoq(Yks)

H

J

K

Ci*KS

PCE i

SEi

GRAPHIC 3.2: Cost efficiency and non-discretionary inputs in DEA models

The second term in square brackets represents the concept of social ef-ficiency for bank i , (SEi), and is defined as the ratio between the optimalcosts attainable for bank i when it operates in a favorable environment, withhigher levels of social capital (C*

i), and the feasible optimal costs for bank i(C*

iKS ):

Note that this indicator measures how the society penalizes the bank-ing firm (by increasing its costs) because of the low relative level of social cap-ital it faces.

In practical terms, the problem consists of computing the optimalcosts for bank i, (C*

iKS ), comparing bank i only with those facing similar ormore unfavorable environments—in terms of social capital—, which isachieved by solving a modified version of linear programming problem (1):

Min SM

m = 1 wim xim

s.t.SR

r = 1 lr yrp ≥ yip p = 1, ..., P

SR

r = 1 lr xrm ≤ xim m = 1, ..., M (3.5)

SR

r = 1 lr = 1, lr ≥ 0, r = 1, ..., R

lr = 0 if KSr > KSi

where an additional constraint has been inserted so as to restrain the scopeof comparison, considering only those banks facing the same or more unfa-vorable environmental conditions as those for the bank under analysis—inour case operating in societies with the same, or a lower, level of social capi-tal (KS)—. See Ruggiero (1996, 1998 and 2004).

3.3. Comparing efficiency distributions

Once efficiency scores have been computed with and without social capitalthere are several ways to present and compare results. Usually, conclusionsare highly descriptive, based only on what summary statistics reveal. How-


21

SEi = C*i (3.4)

C*iKS

ever, we may consider some additional methods which provide us with morethorough conclusions. In our case, results are obtained for several countriesand different types of financial institutions in each country. Therefore, wereinterpretations confined to simple summary statistics, the probability of miss-ing important information would be high.

Specifically, using kernel smoothing methods we will estimate nonpara-metrically the density functions corresponding to both CE and PCE indices.Several monographs cover this topic in depth (see Silverman, 1986) 9. Thekernel density estimator f for a univariate density f based on a sampleof R efficiency indices (either CE or PCE) is f (x) = (Rh)–1 ΣR

i = 1K ((CEi – x)/h),where r is the banking firm index, CEi represents its efficiency index, x is theevaluation point, h is the bandwidth and K is a kernel function satisfyingcertain properties 10. Additionally, we must control for the fact that effi-ciency indices are bounded between (0,1), otherwise estimation is incon-sistent (see Simar and Wilson, 2006). We use Silverman’s (1986) reflec-tion method, the basic idea of which consists of reflecting or mirroring theprobability mass lying beyond the unity—where, theoretically, no proba-bility mass should exist—. The kernel estimate disregarding the bound-ary condition can be shown to be biased an inconsistent (Simar and Wil-son, 1998) 11.

Given our overall nonparametric setting, we also consider nonpara-metric methods to explore the statistical differences between our efficiencyscores, since they focus on the entire distributions instead of confining thecomparison to summary statistics, such as the mean, in the case of the two-sample t -test, or the median, in the case of the Kruskal-Wallis test.

Once densities are estimated, the Li (1996) test enables us to ascer-tain whether the observed visual differences are statistically significant.These instruments, despite their usefulness, have rarely been employed byeconomists. The test is based on measuring the distance between two densi-ties f(x) and g(x) through the mean integrated square error, i.e.:


22

9. See also Wand and Jones (1995); Pagan and Ullah (1999); Härdle et al. (2004).

10. Further decisions relate to the choice of kernel and the choice of bandwidth. Regarding theformer, we considered the Gaussian method for its easiness to compute; regarding the latter, wechose the plug-in method suggested by Wand and Jones (1994). The details have been skippedfor the sake of brevity, since they are provided in the literature cited throughout the text.

11. Although we have used a variety of statistical procedures to perform all computations, theFEAR package by Paul W. Wilson for the statistical software R provides code for both computingefficiencies and densities, considering the reflection method. Seehttp://www.eco.utexas.edu/faculty/Wilson/Software/FEAR/fear.html.

I = I (f(x), g(x)) = ʃx (f(x) – g(x))2 dx = ʃx (f 2(x) + g2(x) – 2 f(x) g(x)) dx = (3.6)= ʃx (f(x) dF(x) + g(x) dG(x) – 2g(x) dF(x))

where F and G are two candidates for the distribution of X , with densityfunctions f(x) and g(x), which are estimated using kernel methods. Thus, fis the nonparametric kernel estimator referred to above. Since f = (Rh)–1 ΣR

i = 1

K ((x – CEi)/h), and g = (Rh)–1 ΣRi = 1K ((y – CEi)/h) a feasible estimator for I is:

In addition to this, the integrated square error is essential for estimat-ing the statistic in which the test is based, whose general expression corre-sponds to:

where

and h is the bandwidth. See Li (1996), Fan and Ullah (1999), or Pagan andUllah (1999) for full details. For an application, see the appendix in Kumarand Russell (2002). As mentioned above, economic applications of the test,despite its usefulness in nonparametric settings, are virtually non-existent.


23

I = ʃx (f (x) – g(x))2 dx = (3.7)

= 1S

R

i = 1 S

R

j = 1 [K (xi – xj)+ K (yi – yj)– K (yi – xj)– K (xi – yi)]R2h h h h hj ≠ i

s = 1

SR

j = 1 S

R

i = 1 [K (xi – xj)+ K (yi – yj)+ 2K (xi – yj)] (3.9)R2hp1/2 h h h

T = Rh1/2 I (3.8)s

4. Data and Variables

INTERNATIONAL comparisons of bank efficiency are complex, since theavailable information on the balance sheet and the profit and loss account ofeach bank must be homogeneous. We use data from the Bureau Van DijkBankScope data base which provides us with this type of information—i.e.,homogeneous data for all firms in our sample—. We use unconsolidated fi-nancial statements, or in their absence consolidated statements on individualcommercial banks, savings banks and credit unions for the 1993-2001 period.These data are available for all OECD countries. However, Turkey was ex-cluded from the study due to the lack of information about its social capital.

Because of the limitation of our database, we followed the intermedi-ation approach for the choice of inputs and outputs for banking firms. For agood exposition on the problematic of measuring bank activity, see [berger-humphreymeasurement]. Specifically, regarding the choice of outputs, weselected credits and loans (y1), deposits (y2), other earning assets (y3) andfee income (y4). The inputs selected were personnel expenses (x1), physicalcapital (x2), and loanable funds (x3). Input prices (w) are obtained as ratiosbetween the costs generated by each input category and their quantities.Thus, total costs are obtained by adding together personnel expenses(w1x1), general expenses (w2x2) and financial costs (w3x3).

The social capital indicators used in the literature so far vary greatly.Most of them aim to measure the trust among individuals in a given com-munity (see Iyer. Kitson and Toh, 2005). For instance, Guiso, Sapienza andZingales (2004) use electoral turnout and blood donation as primary mea-sures of social capital. Yet none of the indicators used so far is based on theconcept of social capital considered as an accumulated stock of trust. Incontrast, we consider the more complex measure proposed by Pérez et al.(2005, 2006). This indicator is, to date, the one that most rigorously ap-proaches the theoretical economic concept of social capital, and the onewith widest coverage in terms of time—covering the 1970–2001 period— andinternational scope—containing information for 28 countries—12.

24

12. These data are used and presented in Pérez et al. (2006) available athttp://w3.grupobbva.com/TLFB/tlfb/TLFBindex-pub.jsp. The fourth appendix in the workingpaper contains all data on social capital for OECD countries, and for the 1970-2001 period.

Contrary to other social capital indicators used in previous researchstudies, based on ad hoc measures bearing few links with the concept of cap-ital in the economic sense (i.e., an asset which accumulates over time), ourindicator of social capital stands alongside other measures of capital devisedby economists. It is based on a model which combines individual trust deci-sions (micro level) with the aggregate effect to co-operate conveyed in socialrelationship networks (macro level), within a context in which the econom-ic aspects are at the core of the analysis. The central role of economic as-pects arises because, on the one hand, economic relations are considered tobe one of the most important sources of interaction and trust creationamong individuals and, on the other, because social capital is considered cap-ital in the economic sense, i.e., it is produced through investment processes,and it depreciates over time.

Based on these premises, by Pérez et al. (2005, 2006) develop a theo-retical model enabling identification of all elements essential to a social cap-ital measure with solid fundamentals. The components contributing to so-cial capital are the various productive factors (physical capital, labor, andhuman capital) contributing to personal income generation, the degree ofconnectedness existing within the trust network—along with its dimension—,the reciprocity among agents, the level of inequality in society, the mar-ginal cost of investing in social capital along with its depreciation rate, thetemporal discount rate and, finally, the expectations of individuals as to so-ciety fitting (life expectancy for those cases in which the emigration rate islow). Considering this model, a social capital measure is proposed whichadopts proxies for variables that the theoretical model postulates as rele-vant.

Table 4.1 contains summary statistics (means) for the variables consid-ered, both for banks’ inputs, outputs and prices, and social capital. Thefirst column contains the number of observations, which add up to 36,664banking firms for the period under analysis (1993-2001). The last columncontains data on the level of per capita social capital, which reveals remark-able disparities across countries. For instance, Switzerland and Norway havethe highest endowments of social capital, with a value of 940 and 632, re-spectively. At the other extreme, Poland, Mexico and Spain have the lowestlevels of social capital—with 42.6, 58.82 and 71.88, respectively—.


25


26

TA

BL

E4.

1:Su

mm

ary

stat

istic

s fo

r ba

nk in

puts

, out

puts

and

soc

ial c

apita

l, av

erag

e 19

93–2

001

peri

od

# ob

s. 1

993-

2001

y1a

y2a

y3a

y4a

x1a

x2a

x3a

Soci

al

capi

talb,

c

Aus

tral

ia13

516

,150

,495

.72

13,9

88,5

37.5

84,

174,

667.

1936

6,27

7.32

195,

890.

7310

8,79

7.95

19,1

07,7

96.6

035

3.90

Aus

tria

951

629,

022.

8690

2,21

5.30

588,

367.

3810

,040

.88

10,6

07.4

612

,794

.62

1,18

6,94

8.07

566.

83

Bel

gium

372

8,24

4,55

1.60

14,9

49,1

00.1

09,

482,

655.

9418

3,79

9.22

160,

685.

9119

9,83

9.74

17,3

88,4

28.9

419

8.02

Can

ada

307

13,6

38,8

38.3

818

,422

,462

.52

9,61

8,34

5.19

590,

218.

7739

1,06

7.48

185,

957.

7222

,077

,679

.56

419.

93

Cze

ch R

epub

lic14

868

1,21

5.42

1,70

5,63

0.32

1,26

8,22

7.88

43,4

10.8

824

,408

.15

52,1

24.7

11,

873,

644.

3413

7.89

Den

mar

k73

11,

249,

376.

551,

947,

489.

711,

293,

250.

3619

,193

.08

26,6

47.5

919

,274

.17

2,31

0,23

1.30

180.

71

Finl

and

699,

781,

763.

1712

,943

,979

.32

9,91

6,11

5.86

78,3

76.0

877

,098

.21

109,

309.

3618

,156

,340

.88

171.

75

Fran

ce2,

105

4,88

6,45

8.61

7,88

7,73

7.73

6,29

3,56

2.94

153,

362.

9110

5,70

8.92

69,2

81.4

59,

957,

042.

3813

6.60

Ger

man

y15

,106

974,

451.

041,

459,

113.

1085

6,64

6.40

11,6

08.6

816

,206

.04

15,9

60.2

51,

790,

177.

8949

6.36

Gre

ece

135

4,31

3,53

6.62

7,76

8,15

0.26

5,91

6,58

5.88

139,

314.

3613

8,26

5.62

168,

326.

439,

813,

360.

3610

1.82

Hun

gary

7390

9,36

0.53

1,69

3,08

5.69

885,

810.

8238

,709

.49

27,9

78.0

548

,874

.59

1,79

1,04

1.16

91.9

3

Irel

and

821,

876,

542.

053,

708,

706.

773,

182,

055.

7816

,822

.30

3,80

4.26

25,1

14.0

34,

792,

237.

7239

9.09

Ital

y3,

238

1,18

7,72

8.74

1,22

9,13

9.46

651,

294.

0624

,978

.87

28,2

56.3

929

,560

.25

1,70

9,55

4.00

109.

32

Japa

n1,

846

9,13

5,63

1.86

12,4

42,5

97.9

25,

265,

944.

0353

,354

.49

69,4

59.6

118

8,77

5.90

14,1

71,0

61.2

757

7.32

Sout

h K

orea

140

24,1

90,7

82.9

730

,184

,682

.57

11,3

14,6

64.9

741

0,53

0.07

283,

969.

3668

0,12

9.43

34,8

83,1

50.4

666

4.44

Lux

embo

urg

912

1,27

2,40

5.31

4,54

8,17

7.09

4,01

9,63

5.07

36,3

32.0

316

,266

.90

18,0

73.6

05,

082,

317.

4369

0.27

Mex

ico

204

288,

209.

7550

7,66

2.19

82,1

23.7

516

,002

.54

984.

4323

,757

.70

507,

662.

1934

.89

Net

herl

ands

184

4,76

5,43

9.71

2,87

3,22

1.51

1,52

8,34

8.17

37,3

12.8

538

,296

.96

31,0

11.6

26,

076,

540.

0574

2.57

New

Zea

land

368,

285,

250.

179,

031,

495.

581,

800,

263.

6513

0,61

3.43

81,7

76.6

131

,040

.73

9,78

8,45

2.10

374.

59

Nor

way

278

3,02

8,27

6.62

2,47

6,97

2.77

672,

604.

1234

,767

.80

34,6

85.5

128

,312

.45

3,48

4,87

1.94

415.

81

Pola

n24

31,

150,

039.

492,

117,

517.

321,

094,

697.

0264

,739

.63

44,3

39.1

164

,853

.19

2,14

2,74

3.30

35.0

8

Port

ugal

225

5,88

7,76

3.37

7,77

0,86

8.82

2,82

4,45

5.80

235,

921.

3972

,569

.84

126,

800.

048,

923,

561.

2928

0.53

Slov

akia

9034

9,06

7.23

1,03

6,83

1.76

737,

925.

9720

,881

.22

13,1

52.2

542

,551

.81

1,05

0,41

2.13

69.7

5

Spai

n1,

101

4,24

7,37

5.29

6,12

4,71

1.73

2,78

3,05

2.68

56,1

58.8

084

,289

.01

122,

423.

296,

675,

055.

4414

1.66

Swed

en17

01,

240,

648.

492,

006,

126.

861,

464,

254.

3925

,598

.75

24,0

50.1

55,

895.

532,

513,

855.

1647

5.86

Switz

erla

nd1,

414

1,52

7,81

8.99

2,85

4,74

6.00

1,94

3,71

7.91

63,3

78.5

744

,288

.90

28,3

20.5

63,

304,

728.

561,

023.

80

Uni

ted

Kin

gdom

298

8,13

8,21

3.78

11,0

14,3

80.4

26,

720,

913.

6624

6,05

2.16

172,

142.

4393

,042

.26

13,1

86,8

37.4

640

3.01

USA

6,07

16,

277,

848.

946,

736,

987.

323,

070,

788.

8424

6,09

7.45

141,

061.

0911

3,96

9.25

9,36

2,45

3.93

506.

82

Tot

al36

,664

3,21

7,22

4.36

4,29

5,33

8.33

2,29

5,59

3.34

74,1

87.1

855

,939

.45

60,1

01.5

05,

315,

620.

0343

3.33

aT

hous

ands

US$

bV

olum

e in

dex

(no

units

), y

ear

1970

=100

. See

Pér

ez e

t al.

(200

5, 2

006)

.c

Dat

a on

soc

ial c

apita

l are

free

ly a

vaila

ble

at h

ttp:

//w

3.gr

upob

bva.

com

/TL

FB/t

lfb/T

LFB

inde

x_pu

b.js

p (a

ppen

dix

4).

5. Results

TABLE 5.1 reports results on average cost efficiency (CE) for each countryfor the sample period, showing a roughly stable path: there has been a mod-erate 3% decline. For the whole period (last column), efficiency averagesto 0.900, indicating that cost savings of up to 10% would be attainable if in-puts were used efficiently. Among the most efficient are the Japanese bank-ing system (averaging to 0.990) and the Swiss banking system (averaging0.988). On the other hand, the Norwegian and Italian banking systems arethe most inefficient, averaging to 0.832 and 0.726, respectively. The Spanishbanking system is one of the most efficient, averaging to 0.977—third onlyto Japan and Norway—. However, the 40% increase in standard deviation,from 0.090 in 1993 to 0.125 in 2001, contrasts sharply with the more stabletendency existing for mean efficiency. This marked increase of the disper-sion in cost efficiency could indicate that financial liberalization in theOECD countries has not entirely favored convergence among banking sys-tems’ efficiency.

However, average efficiencies in table 5.2 do not control for the effectof social capital on bank cost efficiency. In section 2 justify we described thevarious paths by which social capital affects efficiency. According to this ra-tionale, disregarding social capital would lead to biased efficiency scores, pe-nalizing those banks facing the most unfavorable environments in terms oflow levels of social capital. Table 5.2 reports measures for pure cost efficiency(PCE), which are adjusted for social capital. These measures were obtainedby solving the linear programming problem (3.1) for all banks, by compar-ing each bank with those facing similar, or more unfavorable environmentswith the same, or lower levels of social capital.

Pure cost efficiency (PCE) shows a similar evolution to that observedfor cost efficiency (CE), although its decline is slightly lower—averaging to2.5%—. If the entire period is analyzed, PCE averages to 0.912 for all 36,664banks in our sample, only slightly higher than CE . When social capital en-ters the analysis, relative positions vary, and the Polish banking system is themost efficient, followed by Japan. Note that Poland is the country with thelowest social capital levels throughout the analyzed period (table 4.1). Ac-cording to the cost efficiency measures which disregard social capital, Po-

27


28

TA

BL

E5.

1:C

ost e

ffic

ienc

y (C

E =

CC/C

),av

erag

es

1993

1994

1995

1996

1997

1998

1999

2000

2001

Tot

al

(ave

rage

)

Aus

tral

ia0.

894

0.88

40.

887

0.89

80.

896

0.88

90.

886

0.88

00.

853

0.88

4

Aus

tria

0.92

60.

906

0.90

40.

896

0.90

60.

851

0.86

20.

926

0.93

80.

900

Bel

gium

0.91

40.

913

0.90

30.

901

0.89

90.

869

0.85

80.

897

0.88

70.

894

Can

ada

0.97

10.

974

0.96

30.

951

0.91

50.

908

0.90

60.

944

0.94

40.

935

Cze

ch R

epub

lic0.

991

0.97

10.

980

0.94

30.

937

0.88

00.

879

0.93

20.

882

0.91

8

Den

mar

k0.

980

0.98

30.

985

0.98

70.

981

0.87

50.

886

0.98

60.

981

0.96

0

Finl

and

0.88

90.

858

0.85

80.

837

0.84

00.

819

0.83

90.

913

0.92

20.

859

Fran

ce0.

851

0.86

00.

873

0.86

90.

894

0.88

00.

884

0.90

30.

913

0.88

1

Ger

man

y0.

956

0.95

20.

948

0.94

70.

942

0.88

00.

879

0.95

10.

951

0.93

2

Gre

ece

0.99

50.

996

0.99

90.

997

0.99

30.

968

0.96

20.

841

0.83

40.

966

Hun

gary

—1.

000

0.96

70.

895

0.87

10.

876

0.82

70.

928

0.93

30.

893

Irel

and

0.97

40.

973

0.98

20.

952

0.96

00.

975

0.96

00.

976

0.96

70.

969

Ital

y0.

788

0.78

80.

766

0.71

80.

730

0.68

70.

689

0.72

70.

709

0.72

6

Japa

n0.

988

0.98

80.

985

0.98

60.

989

0.98

70.

982

0.99

30.

993

0.99

0

Sout

h K

orea

0.87

40.

863

0.86

70.

858

0.82

90.

856

0.89

90.

906

0.90

20.

870

Lux

embo

urg

0.96

90.

964

0.96

20.

952

0.94

90.

941

0.94

50.

960

0.96

70.

956

Mex

ico

0.75

70.

822

0.85

50.

822

0.97

00.

892

0.90

70.

993

1.00

00.

891

Net

herl

ands

0.97

90.

990

0.98

80.

981

0.96

80.

938

0.94

90.

951

0.94

70.

967

New

Zea

land

—0.

991

0.99

60.

979

0.98

20.

961

0.95

50.

956

0.95

50.

966

Nor

way

0.94

90.

921

0.90

20.

869

0.80

00.

777

0.78

70.

807

0.77

40.

832

Pola

n0.

997

0.99

80.

986

0.97

10.

966

0.89

80.

904

0.97

90.

962

0.95

3

Port

ugal

0.93

90.

943

0.96

80.

959

0.94

90.

899

0.84

80.

877

0.90

20.

921

Slov

akia

0.99

80.

997

0.99

90.

964

0.96

30.

940

0.92

30.

977

0.97

90.

959

Spai

n0.

983

0.98

60.

989

0.99

00.

987

0.96

20.

941

0.98

30.

975

0.97

7

Swed

en0.

978

0.97

60.

980

0.96

50.

973

0.95

90.

946

0.96

20.

991

0.98

0

Switz

erla

nd0.

843

0.87

50.

876

0.87

50.

863

0.84

10.

840

0.87

70.

858

0.86

0

Uni

ted

Kin

gdom

0.97

70.

974

0.97

20.

957

0.95

80.

916

0.93

90.

977

0.96

70.

959

USA

0.90

50.

883

0.88

60.

872

0.86

50.

843

0.81

20.

841

0.83

70.

862

Tot

al0.

926

0.92

20.

919

0.91

00.

908

0.86

60.

857

0.90

50.

900

0.90

0

Std.

dev

.0.

090

0.08

70.

093

0.10

90.

105

0.10

10.

102

0.11

70.

125

0.10

7


29

TA

BL

E5.

2:P

ure

cost

eff

icie

ncy

(soc

ial c

apita

l adj

uste

d) (

PCE

= C

* k/C

),av

erag

es

1993

1994

1995

1996

1997

1998

1999

2000

2001

Tot

al

(ave

rage

)

Aus

tral

ia0.

910

0.95

60.

957

0.95

00.

933

0.93

70.

942

0.90

40.

885

0.92

9

Aus

tria

0.92

80.

908

0.90

50.

895

0.90

60.

852

0.86

20.

926

0.93

80.

901

Bel

gium

0.93

40.

936

0.92

30.

922

0.91

00.

906

0.90

60.

908

0.89

50.

916

Can

ada

0.97

50.

983

0.96

80.

959

0.92

10.

934

0.95

40.

952

0.95

40.

950

Cze

ch R

epub

lic0.

992

0.98

70.

981

0.95

80.

954

0.91

60.

941

0.94

10.

907

0.94

3

Den

mar

k0.

981

0.98

40.

986

0.98

80.

989

0.98

70.

987

0.98

70.

983

0.98

6

Finl

and

0.90

90.

894

0.89

70.

912

0.89

30.

901

0.93

10.

965

0.95

60.

917

Fran

ce0.

898

0.90

90.

906

0.90

50.

912

0.91

20.

930

0.91

70.

935

0.91

4

Ger

man

y0.

956

0.95

20.

948

0.94

80.

942

0.88

10.

897

0.95

10.

951

0.93

4

Gre

ece

0.99

50.

997

1.00

00.

997

0.99

61.

000

1.00

00.

929

0.96

00.

990

Hun

gary

—1.

000

0.97

30.

944

0.89

60.

927

0.92

00.

946

0.98

90.

939

Irel

and

1.00

01.

000

0.98

80.

965

0.98

20.

988

1.00

00.

977

0.97

10.

983

Ital

y0.

831

0.86

10.

877

0.77

30.

749

0.75

40.

761

0.74

80.

788

0.78

3

Japa

n0.

988

0.98

80.

985

0.98

60.

989

0.98

70.

982

0.99

30.

993

0.99

0

Sout

h K

orea

0.88

40.

889

0.87

60.

868

0.83

70.

894

0.93

40.

906

0.90

20.

884

Lux

embo

urg

0.96

90.

964

0.96

30.

952

0.94

90.

941

0.95

10.

960

0.96

70.

957

Mex

ico

0.86

50.

876

0.90

00.

870

0.99

00.

994

0.99

40.

998

1.00

00.

945

Net

herl

ands

0.98

40.

993

0.99

10.

983

0.96

80.

939

0.94

90.

951

0.94

70.

969

New

Zea

land

—1.

000

1.00

01.

000

0.99

70.

988

0.97

10.

980

0.97

90.

986

Nor

way

0.94

90.

921

0.90

20.

869

0.80

00.

777

0.78

70.

815

0.79

80.

837

Pola

n0.

997

0.99

80.

999

0.99

00.

993

0.99

80.

995

0.99

20.

993

0.99

4

Port

ugal

0.96

20.

964

0.96

90.

961

0.95

50.

933

0.91

60.

901

0.92

40.

943

Slov

akia

1.00

00.

997

0.99

90.

968

0.97

40.

978

0.98

60.

978

0.98

40.

979

Spai

n0.

990

0.99

10.

991

0.99

20.

992

0.99

20.

987

0.98

60.

981

0.98

9

Swed

en0.

991

0.98

70.

989

0.98

60.

979

0.97

30.

990

0.97

30.

992

0.98

8

Switz

erla

nd0.

843

0.87

50.

876

0.87

50.

863

0.84

10.

840

0.87

70.

858

0.86

0

Uni

ted

Kin

gdom

0.98

00.

982

0.97

70.

967

0.96

30.

964

0.97

50.

979

0.97

20.

973

USA

0.91

20.

900

0.89

40.

874

0.87

10.

845

0.82

30.

842

0.83

80.

869

Tot

al0.

935

0.93

50.

932

0.91

80.

913

0.88

10.

884

0.90

90.

914

0.91

2

Std.

dev

.0.

073

0.06

80.

075

0.09

90.

103

0.09

60.

098

0.11

30.

107

0.09

7

land ranked twelfth. However, when social capital is controlled for—i.e., tak-ing into account that Polish banks operate in one of the most unfavorableenvironments in the terms described in section 2—their performanceranks in the top position.

Similar tendencies hold for Mexico and Finland which, facing low lev-els of social capital, and also having low average cost efficiency scores forbanks operating in their banking systems, move up six positions in the rank-ing for pure cost efficiency. Akin to Poland, when the negative effect that anunfavorable environment has on banks costs is controlled for, efficiency inMexico and Finland is enhanced by more than 6%.

In contrast, Austria and Netherlands, whose social capital levels rankamong the highest, move down in the pure cost efficiency ranking when thefavorable conditions provided by their financial systems are controlled for,simply because there are other countries which now overtake them. However,that does not imply that their performance deteriorates. For instance, in thecase of Austria, its average uncontrolled efficiency for1993 is CE 1993

Austria = 0.926and it remains virtually unchanged (PCE 1993

Austria = 0.928) after it is controlled for.Regarding dispersion measures, it should be emphasized that pure

cost efficiency is roughly 10% lower than cost efficiency. Therefore, discrep-ancies are greatly reduced when controlling for the differing circumstancesin which banks operate. Likewise, similarly to what occurs for cost efficiency,pure cost efficiency dispersion also increases by 33%, from 0.073 in 1993 to0.097 in 2001. Other studies have also obtained a significant reduction of thedisparities among countries when environmental variables were controlled for[(see Pastor and Serrano, 2005; Lozano-Vivas, Pastor and Pastor, 2002).

Graphic 5.1 shows densities for efficiency scores. The solid line ineach sub-graphic is the density for uncontrolled efficiency scores (CE), andthe dotted line represents densities controlling for social capital (PCE).Bandwidths for each density are provided in table 5.3. Each sub-graphic con-tains densities for each country in our sample. The scale of the Y-axis ingraphic 5.1: Densities for cost efficiency (CE) and cost efficiency adjusted bysocial capital (PCE), 1993–2001 each sub-graphic varies so as to show the ten-dencies for each country more precisely—otherwise it would be difficult tocompare results for Japan and South Korea, for instance, given their starkdifferences—. Probability mass tends to accumulate rightwards for both CEand PCE due to the usually large number of observations with a value of 1—i.e., efficient units—. In cases where efficiency is higher, probabilitymass tends to be more concentrated rightwards. However, densities revealfeatures concealed by summary statistics. For instance, in some banking sys-tems as efficient as the Swiss banking system there is also a large number of


30


31

Australia Austria

CanadaBelgium

Czech Republic Denmark

0.6 0.7 0.8 0.9 10

0

2

4

6

0.6 0.7 0.8 0.9 100.5

0.6 0.7 0.8 0.9 100.5

8

10

0

2

4

6

8

10

0

2

4

6

12

14 30

0

5

10

15

20

25

0

5

10

15

20

25

0

20

40

60

0.6 0.7 0.8 0.9 100.50.4

0.6 0.7 0.8 0.9 100.50.4 0.75 0.80 0.85 0.90 0.95 100

Cost efficiency (CE) Pure cost efficiency (PCE)

GRAPHIC 5.1: Densities for cost efficiency (CE) and cost efficiencyadjusted by social capital (PCE), 1993-2001


32

Finland France

GreeceGermany

Hungary Ireland

0.6 0.7 0.8 0.9 10

0

2

4

6

0.6 0.7 0.8 0.9 100.5

0.6 0.7 0.8 0.9 10

6

7

1

2

3

4

10

0

5

15

0

50

100

150

200

0

2

4

6

8

10

0

50

100

150

0.6 0.7 0.8 0.9 100.5

0.6 0.7 0.8 0.9 100.5 0.75 0.80 0.85 0.90 0.95 100


5

0.4

8

10

12

0.65 0.65

GRAPHIC 5.1 (continuation): Densities for cost efficiency (CE) and cost efficiencyadjusted by social capital (PCE), 1993-2001


33

Italy Japan

LuxembourgSouth Korea

Mexico Netherlands

0.6 0.7 0.8 0.9 10

0

100

200

300

0.8 0.9 10

0.6 0.7 0.8 0.9 10

2.5

3.0

0.5

1.0

1.5

3

1

2

4

0

10

20

30

40

0

5

10

15

20

0

10

20

30

0.75 0.80 0.95 1000.70

0.6 0.7 0.8 0.9 100.5 0.75 0.80 0.85 0.90 0.95 100


2.0

0.7

400

500

600

0.50.4

0.0

0.85 0.90 0.4 0.5

50

40



34

New Zealand Norway

PortugalPoland

Slovakia Spain

0.94 0.96 0.98 1.00

0.5

10

15

20

0.8 0.9 10

0.6 0.7 0.8 0.9 10

30

40

10

100

0

50

0

5

10

15

20

0

10

20

30

40

0

50

100

150

0.6 0.7 100.5

0.90 0.95 1.000.85 0.6 0.7 0.8 0.9 10


20

0.7

25

30

0.920.90

0.8 0.9 0.4 0.5

0.60.5

150

200

300

250

0.3

50

60


banks constituting an outstanding mode between 0.7 and 0.8. Multimodalityis also present for Austria or the USA, for instance. In other cases such asFinland, Italy or South Korea, the number of efficient banks is much lower,and the most perceptible modes are located well below the unity.

However, our main interest lies in comparing efficiency results whensocial capital is included. This effect is shown by the dotted line in each sub-graphic. The effect varies greatly across countries. In several cases, we observethat efficiency enhances ostensibly, as revealed by probability mass accumu-lating more tightly rightwards—suggesting a higher number of bankingfirms approaches the unity, i.e., the frontier—. This tendency is genera-lized, but is more apparent for countries with low levels of social capital such


35

Sweden Switzerland

USAUnited Kingdom

0.90 0.95 1.00

0

2

4

6

0.8 0.9 10

0.6 0.7 0.8 0.9 10

60

80

20

20

0

10

0

1

2

3

4

0.7 1.00.6


40

0.7

8

10

0.850.80

0.8 0.9 0.4 0.5

0.60.5

30

40

50

0

0.4

12

5

6

7


as Finland, Mexico or, more especially, Poland. However, the more strikingtendencies are those found for countries with the highest levels of social cap-ital, such as Austria, Germany, Japan, Netherlands, Norway, Switzerland, orthe USA. In these cases, densities are virtually the same, i.e., banks move upor down only marginally in the efficiency rankings.

These results are complemented via Li’s (1996) test, the summary ofwhich is reported in table 5.4. The summary is provided for the usual signif-icance levels. It broadly corroborates the analysis performed for densities.Accordingly, for cases in which controlled and uncontrolled densities dif-fered, we obtain statistically significant results. That is the case for Australia,


36

TABLE 5.3: Bandwidths for CE and PCE densities

hCE hPCE

Australia 0.0865 0.0493

Austria 0.0274 0.0275

Belgium 0.0274 0.0247

Canada 0.0203 0.0115

Czech Republic 0.0257 0.0141

Denmark 0.0057 0.0064

Finland 0.0585 0.0538

France 0.0218 0.0189

Germany 0.0083 0.0083

Greece 0.0038 0.0031

Hungary 0.0612 0.0398

Ireland 0.0035 0.0129

Italy 0.0515 0.0443

Japan 0.0004 0.0004

South Korea 0.0513 0.0664

Luxembourg 0.0060 0.0059

Mexico 0.0378 0.0279

Netherlands 0.0096 0.0096

New Zealand 0.0237 0.0096

Norway 0.0885 0.0836

Poland 0.0070 0.0018

Portugal 0.0201 0.0160

Slovakia 0.0184 0.0082

Spain 0.0026 0.0027

Sweden 0.0123 0.0069

Switzerland 0.0212 0.0211

United Kingdom 0.0099 0.0102

USA 0.0240 0.0248

Note: Bandwidths for CE and PCE densities have been estimated using Sheather and Jones (1991) plug-in methods.


37

TA

BL

E5.

4:D

istr

ibut

ion

hypo

thes

is te

stsa

(Li,

1996

) (1

993-

2001

)

10-P

erce

nt

5-P

erce

nt1-

Per

cent

Cou

ntry

Nul

l hyp

othe

sis

(H0)

bT

-test

sta

tistic

sp-

valu

esi

gnif

ican

ce le

vel

sign

ific

ance

leve

l si

gnif

ican

ce le

vel

(cri

tical

val

ue=1

.281

6)(c

ritic

al v

alue

=1.6

449)

(cri

tical

val

ue=2

.326

3)

Aus

tral

iaf(C

EA

ustr

alia)=

g (P

CE

Aus

tral

ia)

3.26

520.

0005

H0

reje

cted

H0

reje

cted

H0

reje

cted

Aus

tria

f(CE

Aus

tria) =

g(P

CE

Aus

tria)

0.02

800.

4888

H0

non

reje

cted

H0

non

reje

cted

H0

non

reje

cted

Bel

gium

f(CE

Bel

gium

) = g

f(PC

EB

elgi

um)

4.33

320.

0000

H0

reje

cted

H0

reje

cted

H0

reje

cted

Can

ada

f(CE

Can

ada )

= g

(PC

EC

anad

a )5.

2024

0.00

00H

0re

ject

edH

0re

ject

edH

0re

ject

ed

Cze

ch R

epub

licf(C

EC

zech

Rep

ublic

) = g

(PC

EC

zech

Rep

ublic

)4.

5698

0.00

00H

0re

ject

edH

0re

ject

edH

0re

ject

ed

Den

mar

kf(C

ED

enm

ark )

= g

(PC

ED

enm

ark )

26.3

440.

0000

H0

reje

cted

H0

reje

cted

H0

reje

cted

Finl

and

f(CE

Finl

and )

= g

(PC

EFi

nlan

d )4.

8706

0.00

00H

0re

ject

edH

0re

ject

edH

0re

ject

ed

Fran

cef(C

EFr

ance

) = g

(PC

EFr

ance

)32

.538

40.

0000

H0

reje

cted

H0

reje

cted

H0

reje

cted

Ger

man

yf(C

EG

erm

any )

= g(

PCE

Ger

man

y )U

nfea

sibl

e—

——

—

Gre

ece

f(CE

Gre

ece )

= g

(PC

EG

reec

e )5.

6958

0.00

00H

0re

ject

edH

0re

ject

edH

0re

ject

ed

Hun

gary

f(CE

Hun

gary) =

g(P

CE

Hun

gary)

3.44

670.

0003

H0

reje

cted

H0

reje

cted

H0

reje

cted

Irel

and

f(CE

Irel

and )

= g

(PC

EIr

elan

d )0.

8984

0.18

45H

0no

n re

ject

edH

0no

n re

ject

edH

0no

n re

ject

ed

Ital

yf(C

EIt

aly )

= g(

PCE

Ital

y )68

.126

80.

0000

H0

reje

cted

H0

reje

cted

H0

reje

cted

Japa

nf(C

EJa

pan )

= g

(PC

EJa

pan )

0.00

140.

4994

H0

non

reje

cted

H0

non

reje

cted

H0

non

reje

cted

Sout

h K

orea

f(CE

Sout

h K

orea

) = g

(PC

ESo

uth

Kor

ea)

0.99

550.

1598

H0

non

reje

cted

H0

non

reje

cted

H0

non

reje

cted

Lux

embo

urg

f(CE

Lux

embo

urg )

= g

(PC

EL

uxem

bour

g )0.

0902

0.46

41H

0no

n re

ject

edH

0no

n re

ject

edH

0no

n re

ject

ed

Mex

ico

f(CE

Mex

ico )

= g

(PC

EM

exic

o )17

.921

0.00

00H

0re

ject

edH

0re

ject

edH

0re

ject

ed

Net

herl

ands

f(CE

Net

herl

ands

) = g

(PC

EN

ethe

rlan

ds)

0.26

530.

3954

H0

non

reje

cted

H0

non

reje

cted

H0

non

reje

cted

New

Zea

land

f(CE

New

Zea

land

= f(C

EN

ew Z

eala

nd)

1.39

20.

0820

H0

reje

cted

H0

reje

cted

H0

reje

cted

Nor

way

f(CE

Nor

way) =

g(P

CE

Nor

way)

0.03

990.

4841

H0

non

reje

cted

H0

non

reje

cted

H0

non

reje

cted

Pola

ndf(C

EPo

land

) = g

(PC

EPo

land

)34

.643

70.

0000

H0

reje

cted

H0

reje

cted

H0

reje

cted

Port

ugal

f(CE

Port

ugal) =

g(P

CE

Port

ugal)

2.53

670.

0056

H0

reje

cted

H0

reje

cted

H0

reje

cted

Slov

akia

f(CE

Slov

akia) =

g(P

CE

Slov

akia)

5.84

140.

0000

H0

reje

cted

H0

reje

cted

H0

reje

cted

Spai

nf(C

ESp

ain )

= g

(PC

ESp

ain )

47.0

879

0.00

00H

0re

ject

edH

0re

ject

edH

0re

ject

ed

Swed

enf(C

ESw

eden

) = g

(PC

ESw

eden

)1.

0559

0.14

55H

0no

n re

ject

edH

0no

n re

ject

edH

0no

n re

ject

ed

Switz

erla

ndf(C

ESw

itzer

land

) = g

(PC

ESw

itzer

land

)0.

0007

0.49

97H

0no

n re

ject

edH

0no

n re

ject

edH

0no

n re

ject

ed

Uni

ted

Kin

gdom

f(CE

Uni

ted

Kin

gdom

)=

g(PC

EU

nite

d K

ingd

om)

3.82

150.

0001

H0

reje

cted

H0

reje

cted

H0

reje

cted

USA

f(CE

USA

) = g

(PC

EU

SA)

Unf

easi

ble

——

——

aT

he fu

nctio

nsf(·

)and

g(·)

are

(ker

nel)

dis

trib

utio

n fu

nctio

ns fo

r un

cont

rolle

d (C

E)an

d co

ntro

lled

(PC

E) e

ffic

ienc

y sc

ores

, res

pect

ivel

y.b

The

nul

l hyp

othe

sis

test

s fo

r th

e eq

ualit

y of

dis

trib

utio

ns H

0: f

(x) =

g(x

),∀

x, a

gain

st th

e al

tern

ativ

e, H

1:f

(x)≠

g(x)

, for

som

e x.

Belgium, Canada, Czech Republic, Denmark, Finland, France, Greece,Hungary, Italy, Mexico, Poland, Portugal, Slovakia, Spain and United King-dom—i.e., differences between efficiency scores obtained with and withoutsocial capital are statistically significant at 1% significance level—. The testalso corroborates the results for those countries whose banks’ efficiency scoresremained unaffected when controlling for social capital, showing no sta-tistical differences. That is the case for Austria, Ireland, Japan, Luxembourg,Netherlands, Norway and Switzerland. Of special note is the case of SouthKorea, New Zealand and Sweden for which densities revealed differences,although these are not significant. This relatively conflicting finding may beexplained by the fact that social capital per head in these countries is quitehigh, especially in South Korea (ranked 4 in social capital per head, see lastcolumn in table 4.1) and Sweden (ranked 9). In the case of New Zealand,although its ranking is lower (14), we must bear in mind that some evidencedoes exist on significant differences—at 10% level, with a p-value of 0.0820—.

Table 5.5 reports results for social inefficiency (SE), which measuresthe penalization—in terms of banking costs—which the society imposeson the banking firm according to the level of social capital present in eachcountry: the lower the level, the higher the penalization on bank costs. Forthe whole sample period, the value for social efficiency is 0.987, indicatingthat if social capital levels were similar to, or higher than those correspondingto the country with the highest level of social capital, banking firms’ costscould be reduced by 1.3%. Obviously, circumstances differ across countries.For instance, in Italy, the social inefficiency indicator is 0.926, with 7.4% ofcost savings, whereas in some other outstanding cases such as Finland, Mexi-co, or Hungary, cost savings are 6.3%, 5.7% and 4.9%, respectively.

To illustrate the magnitude of potential cost savings, table 5.6 reportssocial inefficiencies with respect to Gross Domestic Product (GDP). The lastcolumn presents the accumulated potential savings with respect to GDP in2001. Note that there are several countries for which substantial savings(over 3% of GDP) could be attained. That is the case for Italy (5.044%),Belgium (4.621%), or Finland (3.687%). Yet in other instances such as Swit-zerland (0.000%), Japan (0.004%) or the Netherlands (0.036%) the savingswould much lower.

In addition, to analyze whether the impact of social capital on effi-ciency differs according to the size of each bank, we ranked banking sizes,according to asset quartiles. Table 5.7 reports results for cost efficiency andpure cost efficiency. However, results do not suggest any particular tendencyfor this association. Likewise, we constructed banking firms’ quartiles accord-ing to the levels of social capital they face (see table 5.8). In this case, re-


38


39

TA

BL

E5.

5:So

cial

inef

fici

ency

(SE

= C

* /C

* k),

aver

ages

1993

1994

1995

1996

1997

1998

1999

2000

2001

Tot

al

(ave

rage

)

Aus

tral

ia0.

982

0.92

50.

927

0.94

50.

960

0.94

90.

941

0.97

40.

964

0.95

2

Aus

tria

0.99

80.

998

0.99

91.

001

0.99

90.

998

1.00

00.

999

1.00

00.

999

Bel

gium

0.97

90.

976

0.97

90.

977

0.98

80.

959

0.94

70.

987

0.99

10.

976

Can

ada

0.99

60.

991

0.99

50.

992

0.99

30.

972

0.95

00.

991

0.99

00.

984

Cze

ch R

epub

lic1.

000

0.98

40.

999

0.98

40.

982

0.96

10.

934

0.99

10.

973

0.97

4

Den

mar

k0.

999

0.99

90.

999

0.99

90.

992

0.88

70.

898

0.99

80.

998

0.97

4

Finl

and

0.97

80.

960

0.95

70.

918

0.94

10.

909

0.90

20.

946

0.96

50.

937

Fran

ce0.

948

0.94

50.

964

0.96

00.

980

0.96

60.

950

0.98

50.

976

0.96

4

Ger

man

y1.

000

1.00

01.

000

1.00

01.

000

0.99

90.

979

1.00

01.

000

0.99

7

Gre

ece

1.00

01.

000

0.99

91.

000

0.99

70.

968

0.96

20.

905

0.86

90.

976

Hun

gary

—1.

000

0.99

40.

948

0.97

20.

945

0.89

80.

981

0.94

40.

951

Irel

and

0.97

40.

973

0.99

40.

986

0.97

80.

987

0.96

01.

000

0.99

60.

986

Ital

y0.

948

0.91

50.

873

0.92

90.

975

0.91

20.

905

0.97

20.

900

0.92

6

Japa

n1.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

0

Sout

h K

orea

0.98

90.

971

0.99

00.

989

0.99

00.

957

0.96

21.

000

1.00

00.

984

Lux

embo

urg

0.99

91.

000

0.99

91.

000

1.00

01.

000

0.99

41.

000

1.00

00.

999

Mex

ico

0.87

50.

937

0.95

00.

945

0.98

00.

897

0.91

20.

995

1.00

00.

943

Net

herl

ands

0.99

40.

997

0.99

70.

998

1.00

00.

998

1.00

01.

000

1.00

00.

998

New

Zea

land

—0.

991

0.99

60.

979

0.98

50.

973

0.98

30.

975

0.97

50.

979

Nor

way

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

00.

990

0.97

00.

994

Pola

n1.

000

1.00

00.

987

0.98

10.

973

0.90

00.

908

0.98

70.

969

0.95

8

Port

ugal

0.97

60.

978

0.99

80.

998

0.99

40.

963

0.92

60.

973

0.97

60.

977

Slov

akia

0.99

81.

000

1.00

00.

996

0.98

90.

961

0.93

60.

998

0.99

60.

979

Spai

n0.

993

0.99

50.

998

0.99

80.

995

0.97

00.

953

0.99

60.

994

0.98

8

Swed

en0.

986

0.98

90.

991

0.97

90.

994

0.98

60.

955

0.98

90.

999

0.99

2

Switz

erla

nd1.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

01.

000

1.00

0

Uni

ted

Kin

gdom

0.99

70.

991

0.99

50.

990

0.99

50.

951

0.96

30.

998

0.99

50.

986

USA

0.99

20.

981

0.99

10.

997

0.99

40.

997

0.98

70.

999

0.99

90.

993

Tot

al0.

990

0.98

60.

986

0.99

10.

995

0.98

40.

969

0.99

50.

985

0.98

7

Std.

dev

.0.

040

0.04

20.

048

0.04

00.

023

0.04

00.

043

0.02

70.

056

0.04

2


40

TA

BL

E5.

6:P

oten

tial c

ost s

avin

gs a

ttri

buta

ble

to s

ocia

l cap

ital.

Per

cent

age

with

res

pect

to G

DP

1993

1994

1995

1996

1997

1998

1999

2000

2001

Acc

um. (

%)a

Aus

tral

ia0.

010

0.04

10.

064

0.09

00.

082

0.09

60.

100

0.09

20.

090

0.69

6

Aus

tria

0.04

60.

041

0.04

30.

066

0.02

00.

079

0.07

50.

026

0.00

70.

448

Bel

gium

0.14

20.

318

0.35

80.

513

0.65

10.

790

0.54

80.

655

0.27

54.

621

Can

ada

0.00

00.

001

0.00

10.

129

0.10

40.

064

0.22

10.

241

0.26

30.

973

Cze

ch R

epub

lic0.

001

0.07

10.

008

0.20

10.

234

0.24

90.

182

0.04

30.

029

0.96

7

Den

mar

k0.

099

0.12

30.

105

0.12

10.

173

0.19

10.

246

0.20

50.

236

1.55

8

Finl

and

0.07

00.

135

0.20

80.

494

0.77

80.

723

0.66

90.

407

0.11

73.

687

Fran

ce0.

235

0.37

60.

322

0.30

70.

348

0.23

90.

229

0.31

20.

245

2.79

0

Ger

man

y0.

026

0.03

30.

015

0.01

90.

032

0.04

70.

112

0.04

30.

036

0.40

5

Gre

ece

0.00

00.

009

0.00

00.

000

0.01

10.

035

0.04

80.

651

0.85

51.

583

Hun

gary

0.00

00.

000

0.00

00.

426

0.05

50.

073

0.30

60.

102

0.19

71.

059

Irel

and

0.05

90.

087

0.02

80.

051

0.05

70.

062

0.01

10.

001

0.00

90.

247

Ital

y0.

164

0.36

80.

854

0.83

20.

567

0.46

10.

418

0.37

20.

818

5.04

4

Japa

n0.

000

0.00

20.

000

0.00

00.

000

0.00

00.

000

0.00

00.

000

0.00

4

Sout

h K

orea

0.02

50.

061

0.05

10.

095

0.05

60.

328

0.39

90.

000

0.00

40.

904

Lux

embo

urg

0.08

30.

070

0.24

00.

040

0.15

40.

138

0.44

30.

007

0.00

01.

097

Mex

ico

0.00

10.

001

0.00

50.

003

0.00

20.

010

0.00

80.

051

0.00

00.

067

Net

herl

ands

0.00

80.

010

0.00

60.

006

0.00

00.

007

0.00

00.

000

0.00

00.

036

New

Zea

land

0.00

00.

008

0.00

40.

156

0.12

80.

226

0.07

20.

235

0.23

91.

168

Nor

way

0.00

00.

000

0.00

00.

000

0.00

00.

000

0.00

00.

023

0.28

60.

309

Pola

n0.

000

0.00

00.

026

0.03

40.

070

0.13

40.

111

0.01

30.

034

0.37

0

Port

ugal

0.04

60.

040

0.02

80.

056

0.10

30.

172

0.32

10.

534

0.28

81.

572

Slov

akia

0.00

00.

000

0.00

00.

036

0.03

90.

150

0.59

10.

040

0.02

50.

876

Spai

n0.

055

0.10

20.

110

0.14

30.

178

0.15

60.

256

0.13

40.

208

1.32

3

Swed

en0.

052

0.05

50.

087

0.08

10.

045

0.06

50.

045

0.10

20.

114

0.69

9

Switz

erla

nd0.

000

0.00

00.

000

0.00

00.

000

0.00

00.

000

0.00

00.

000

0.00

0

Uni

ted

Kin

gdom

0.00

00.

001

0.00

20.

006

0.00

20.

010

0.01

30.

011

0.01

60.

060

USA

0.02

40.

046

0.05

20.

019

0.04

80.

040

0.02

90.

020

0.01

80.

238

Tot

al0.

039

0.07

00.

091

0.09

50.

093

0.08

80.

091

0.07

20.

087

0.69

1

Std.

dev

.0.

040

0.04

20.

048

0.04

00.

023

0.04

00.

043

0.02

70.

056

0.04

2

aSu

m o

f pot

entia

l sav

ings

for

1993

–200

1 pe

riod

w.r

.t. 2

001

GD

P. R

esul

ts in

last

col

umn

are

expr

esse

d as

per

cent

age

for

the

sake

of c

lari

ty.


41

TA

BL

E5.

7:C

ost e

ffic

ienc

y(C

E)

and

cost

eff

icie

ncy

adju

sted

for

soc

ial c

apita

l (PC

E)

rela

tive

to a

sset

s

Ass

et q

uart

ile19

9319

9419

9519

9619

9719

9819

9920

0020

01T

otal

25%

0.94

00.

941

0.93

30.

920

0.91

40.

820

0.84

40.

887

0.88

60.

896

CE

50%

0.92

90.

931

0.92

50.

918

0.91

00.

852

0.83

90.

910

0.90

60.

901

75%

0.90

60.

903

0.90

30.

899

0.89

80.

881

0.84

60.

909

0.90

50.

894

100%

0.92

70.

915

0.91

60.

902

0.90

80.

913

0.89

70.

912

0.90

40.

910

25%

0.95

00.

950

0.94

40.

925

0.91

80.

844

0.87

30.

895

0.90

90.

909

PCE

50%

0.93

70.

938

0.93

50.

923

0.91

30.

864

0.86

20.

912

0.91

30.

909

75%

0.91

80.

919

0.91

60.

907

0.89

90.

889

0.88

20.

910

0.91

20.

905

100%

0.93

50.

933

0.93

40.

918

0.92

00.

925

0.91

70.

920

0.92

20.

924

Tot

al0.

926

0.92

20.

919

0.91

00.

908

0.86

60.

857

0.90

50.

900

0.90

0

Std.

dev

.0.

090

0.08

70.

093

0.10

90.

105

0.10

10.

102

0.11

70.

125

0.10

7

Not

e: ba

nks

wer

e ra

nked

by

tota

l ass

ets,

and

the

sam

ple

was

div

ided

into

qua

rtile

s. A

vera

ge v

alue

s fo

r ea

ch q

uart

ile w

ere

calc

ulat

ed fo

r th

e ba

nks

incl

uded

in e

ach

quar

tile.


42

TA

BL

E5.

8:C

ost e

ffic

ienc

y (C

E)

and

cost

eff

icie

ncy

adju

sted

for

soc

ial c

apita

l(PC

E)

rela

tive

to s

ocia

l cap

ital

Soci

al c

apita

l qua

rtile

1993

1994

1995

1996

1997

1998

1999

2000

2001

Tot

al

25%

0.91

30.

866

0.84

40.

831

0.84

90.

814

0.77

90.

795

0.78

60.

825

CE

50%

0.83

20.

944

0.91

20.

953

0.94

20.

884

0.88

00.

920

0.94

80.

910

75%

0.90

90.

890

0.89

30.

884

0.92

10.

882

0.88

10.

947

0.94

90.

910

100%

0.95

50.

950

0.94

60.

940

0.91

30.

870

0.84

80.

908

0.90

20.

921

25%

0.94

00.

913

0.91

60.

869

0.86

50.

866

0.84

70.

813

0.84

50.

868

PCE

50%

0.86

90.

959

0.93

30.

963

0.95

50.

951

0.94

10.

933

0.95

70.

938

75%

0.91

60.

907

0.90

20.

888

0.92

30.

885

0.90

10.

948

0.95

10.

917

100%

0.95

50.

951

0.94

60.

940

0.91

40.

871

0.85

40.

908

0.90

30.

922

Tot

al0.

956

0.95

20.

948

0.94

80.

942

0.88

10.

897

0.95

10.

951

0.93

4

Std.

dev

.0.

995

0.99

71.

000

0.99

70.

996

1.00

01.

000

0.92

90.

960

0.99

0

Not

e: ba

nks

wer

e ra

nked

by

soci

al c

apita

l, an

d th

e sa

mpl

e w

as d

ivid

ed in

to q

uart

iles.

Ave

rage

val

ues

for

each

qua

rtile

wer

e ca

lcul

ated

for

the

bank

s in

clud

ed in

eac

h qu

artil

e.

sults do clearly indicate that firms facing lower social capital levels experi-ence higher penalizations on their efficiency indices. For banking firms inthe first quartile, as a consequence of operating in unfavorable environ-ments, their costs are around 5% higher for the whole period, although thelevel reaches 7% in 2001. For banks in the second quartile, social penaliza-tion is lower (3%), yet it also reaches a high value (7%) in 2001.

To illustrate the existing relationship between social capital levels andpotential cost savings, graphics 5.2 and 5.3 display nonparametric regres-sions in which cost savings are related, in terms of total costs and in GDPterms for the whole period 13. In both cases we observe a negative relation-ship between potential cost savings and the level of social capital in the econ-omy.


43

3.5 4 4.5 5 5.5 6 6.5

10

9

8

7

6

5

4

3

2

1

0

In (social capital)

k(C

– C

)/C

**

7

GRAPHIC 5.2: Potential costs reduction (1993-2001)

13. We consider nonparametric methods to maintain consistency with instruments used through-out the article. The curves correspond to nonparametric regression estimations in which thedependent variable y corresponds to costs savings (C*

k – C*)/C, and the independent variable xcorresponds to the logarithm of per capita social capital. This type of regression consist of ex-plaining the value of the dependent variable yi as a function of xi as yi = m(xi) + ei, i = 1, ..., n,where ei is a random variable reflecting how y varies in the neighborhood of m(x), where m(x) isthe mean response curve. The idea of nonparametric regression consists of weighting the re-sponse variable in the vicinity of x, in such a way that yi observations are weighted depending ontheir distance from xi to x, i.e., we use the estimator mh (x) = n–1 Sn

i = 1Whi (x)Yi, where Whi(⋅) defines aweighted function depending on a smoothing parameter, or bandwidth, h, and the x1, x2, ..., xn ex-planatory variables. Most of the nonparametric regression techniques are based on weighted aver-ages of the response variable yi. We consider one of the most popular methods, the Nadaraya-Wat-son estimator, although the methodological variety is remarkable (see Härdle, 1990).


44

3.5 4 4.5 5 5.5 6 6.50

In (Social capital)

k(C

– C

)/G

DP

**

7

3.5

3

2.5

2

1.5

1

0.5

GRAPHIC 5.3: Potential costs reduction with respect to GDP (1993-2001)

6. Conclusions

THE aim of this study was to analyze how the level of social capital in eachsociety impacts on the efficiency with which their respective banking indus-tries operate. Therefore, it has sought to combine two rapidly expandingbranches of literature: the role of social capital when measuring economicperformance, and the efficiency of financial institutions.

The study is original in several aspects. First, to date there have beenno attempts to merge these two branches of economics. Second, in contrastto previous studies that analyze issues related to banking aspects of relation-ship social capital (relationship banking), we consider social capital as an en-vironmental variable which affects banking activity in general and bank per-formance in particular, i.e., the efficiency with which banks provide theirproducts and services. Finally, we consider a measure of social capital whichapproaches the theoretical concept of social capital more rigorously thanprevious measures hitherto used.

We have justified a series of ways by which social capital affects bank-ing activity in general and bank efficiency in particular. Specifically, weprovide rationale to indicate that the higher the social capital in the socie-ties, the lower the monitoring, information, transactional, financial, and in-solvency costs, and the higher the credit supply and intensity of use of bank-ing products for clients. All this suggests that bank efficiency can be higherin economies with higher social capital levels.

The analysis was approached from an international perspective, consid-ering all 28 OECD countries with the exception of Turkey. We used nonpara-metric methods to measure efficiency, and the social capital variable wasintroduced into the analysis as a non-discretionary output in the productionprocess. This strategy enables bank cost efficiency to be decomposed intotwo components. The first relates to pure cost efficiency, while the secondone relates to what has been defined as social efficiency, i.e., an indicator ofthe cost penalization experienced by banks operating in unfavorable en-vironments.

Results suggest that, indeed, banks conducting their activity in unfavor-able environments—i.e., in low-social-capital societies—, are penalized interms of costs. Considering the entire set of countries, the savings would be

45

roughly 1.3% of costs. Results are especially informative for countries withlow levels of social capital such as Italy, where increasing social capital wouldlead to cost savings of about 7.4%.

A comparison of the densities (estimated nonparametrically usingkernel methods) for efficiency scores obtained under the two scenarios con-sidered (without/with social capital) lends greater precision to the analysis.Specifically, for several countries with high levels of social capital, results arevirtually identical either with or without social capital. Yet at the other ex-treme there are several examples of countries for which densities differ greatlywhen social capital enters the analysis. The Li (1996) test reinforced the re-sults obtained via examination of densities, by indicating the instances inwhich the observed differences were statistically significant.

The most outstanding conclusion we may draw is that it is crucial toconsider the effects of social capital on the conditions prevailing in the envi-ronment in which banks operate when assessing how they perform. Al-though the global analysis yields interesting results, it conceals some peculiar-ities that would only be uncovered when assessing the tendencies for eachparticular country.


46

References

ANNEN, K. (2003): “Social Capital, Inclusive Networks, and Economic Performance”, Journal

of Economic Behavior & Organization, 50, 449-463.

ARAGON, Y., A. DAOUIA and C. THOMAS-AGNAN (2005): “Nonparametric Frontier Estimation: A

Conditional Quantile-Based Approach”, Econometric Theory, 21(2), 358-389.

BALAGUER-COLL, M. T., D. PRIOR and E. TORTOSA-AUSINA (2006): “On the Determinants of Lo-

cal Government Performance: A Two-Stage Nonparametric Approach”, European Economic

Review, forthcoming.

BANFIELD, E. C. (1958): The Moral Basis of a Backward Society, Free Press, New York.

BANKER, R. D., A. CHARNES and W. W. COOPER (1984): “Some Models for Estimating Technical

and Scale Inefficiencies in Data Envelopment Analysis”, Management Science, 30, 1.078-

1.092.

— and R. C. MOREY (1986a): “Efficiency Analysis for Exogenously Fixed Inputs and Out-

puts”, Operations Research, 34, 513-521.

— and R. C. MOREY (1986b): “The Use of Categorical Variables in Data Envelopment Analy-

sis”, Management Science, 32, 1.613-1.627.

BARON, S., J. FIELD and T. SCHULLER, Editors (2000): Social Capital: Critical Perspectives, Oxford

University Press, Oxford.

BERGER, A. N. and D. B. HUMPHREY (1992): “Measurement and Efficiency Issues in Commer-

cial Banking”, in Z. Griliches (editor), Output Measurement in the Service Sectors, pp. 245-

300, The University of Chicago Press, Chicago.

BEUGELSDIJK, S. and T. VAN SCHAIK (2005): “Social Capital and Growth in European Regions:

An Empirical Test”, European Journal of Political Economy, 21, 301-324.

BOOT, A. W. A. (2000): “Relationship Banking: What do We Know?”, Journal of Financial Inter-

mediation, 9, 7-25.

CAZALS, C., J.-P. FLORENS and L. SIMAR (2002): “Nonparametric Frontier Estimation: A Robust

Approach”, Journal of Econometrics, 106, 1-25.

CHARNES, A., W. W. COOPER and E. RHODES (1978): “Measuring the Efficiency of Decision Mak-

ing Units”, European Journal of Operational Research, 2(6), 429-444.

COOPER, W. W. and J. T. PASTOR (1996): “Non-Discretionary Variables in Dea: A Revision and

a New Approach”, Working paper, Universidad de Alicante.

— L. M. SEIFORD and K. TONE (1999): Data Envelopment Analysis: A Comprehensive Text with

Models, Applications, References and DEA-Solver Software, Kluwer Academic Publishers,

Boston.

47

COOPER, W. W., L. M. SEIFORD and J. ZHU, Editors (2004): Handbook on Data Envelopment Analy-

sis. International Series in Operations Research & Management Science, Kluwer Academic Pub-

lishers, Boston/Dordrecht/London.

DEGRYSE, H. A. and S. ONGENA (2005): “Distance, Lending Relationships, and Competition”,

Journal of Finance, 60(1), 231-266.

DURLAUF, S. N. (2002): “Symposium on Social Capital: Introduction”, The Economic Journal,

112, 417-418.

ELYASIANI, E. and L. G. GOLDBERG (2004): “Relationship Lending: A Survey of The Litera-

ture”, Journal of Economics and Business, 56, 315-330.

FAN, Y. and A. ULLAH (1999): “On Goodness-of-Fit Tests for Weekly Dependent Processes

Using Kernel Method”, Journal of Nonparametric Statistics, 11, 337-360.

FÄRE, R. and S. GROSSKOPF (2004): New Directions: Efficiency and Productivity, Kluwer Academic

Publishers, Boston/London/Dordrecht.

— and C. A. K. LOVELL (1978): “Measuring the Technical Efficiency of Production”, Journal

of Economic Theory, 19(1), 150-162.

FARRELL, M. J. (1957): “The Measurement of Productive Efficiency”, Journal of the Royal Statisti-

cal Society, Ser. A,120, 253-281.

FERRARY, M. (2003): “Trust and Social Capital in the Regulation of Lending Activities”, The

Journal of Socio-Economics, 31, 673-699.

FERRI, G. and M. MESSORI (2000): “Bank-Firm Relationships and Allocative Efficiency in North-

eastern and Central Italy and in the South”, Journal of Banking & Finance, 24, 1067-1095.

FERRIER, G. D. and C. A. K. LOVELL (1990): “Measuring Cost Efficiency in Banking”, Journal of

Econometrics, 46, 229-245.

FINE, B. and F. GREEN (2000): “Economics, Social Capital, and the Colonization of Social

Sciences”, in S. Baron, J. Field and T. Schuller (editors), Social Capital: Critical Perspectives,

chapter 4, pp. 78-93, Oxford University Press, Oxford.

FRIED, H. O. and C. A. K. LOVELL (1996): “Accounting for Environmental Effects in Data En-

velopment Analysis”, Paper presented at the Georgia Productivity Workshop, Athens.

— C. A. K. LOVELL and S. S. SCHMIDT, Editors (1993A): The Measurement of Productive Effi-

ciency: Techniques and Applications, Oxford University Press, Oxford.

— C. A. K. LOVELL and P. VANDEN EECKAUT (1993B): “Evaluating the Performance of U.S.

Credit Unions”, Journal of Banking & Finance, 17, 251-265.

— S. S. SCHMIDT and S. YAISAWARNG (1999): “Incorporating the Operating Environment into

a Nonparametric Measure of Technical Efficiency”, Journal of Productivity Analysis, 4, 419-

432.

FUKUYAMA, F. (1995): Trust: The Social Virtues and the Creation of Prosperity, The Free Press, New

York.

GRAFTON, R. Q., S. KNOWLES and P. D. OWEN (2004): “Total Factor Productivity, Per Capita In-

come and Social Divergence”, The Economic Record, 80(250), 302-313.

GUISO, L., P. SAPIENZA and L. ZINGALES (2004): “The Role of Social Capital in Financial Devel-

opment”, American Economic Review, 94(3), 526-556.


48

HALL, R. E. and C. I. JONES (1999): “Why Do Some Countries Produce so much More Output

Per Worker than others?”, Quarterly Journal of Economics, 114, 83-116.

HANIFAN, L. J. (1916): “The Rural School Community Center”, Annals of the American Academy

of Political and Social Science, 67, 130-138.

HÄRDLE, W. (1990): Applied Nonparametric Regression, Cambridge University Press, Cambridge.

— M. MÜLLER, S. SPERLICH and A. WERWATZ (2004): Nonparametric and Semiparametric Models,

Springer Series in Statistics, Springer-Verlag, New York.

IYER, S., M. KITSON and B. TOH (2005): “Social Capital, Economic Growth and Regional Devel-

opment”, Regional Studies, 39(8), 1015-1040.

KARLAN, D. S. (2004): “Social Capital and Group Banking”, BREAD Working Paper 062, Bureau

for Research in Economic Analysis of Development.

KUMAR, S. and R. R. RUSSELL (2002): “Technological Change, Technological Catch-Up, and

Capital Deepening: Relative Contributions to Growth and Convergence”, American Econom-

ic Review, 92(3), 527-548.

LEVINE, R. (1997): “Financial Development and Economic Growth: Views and Agenda”, Jour-

nal of Economic Literature, 35(2), 688-726.

LI, Q. (1996): “Nonparametric Testing of Closeness between Two Unknown Distribution

Functions”, Econometric Reviews, 15, 261-274.

LOVELL, C. A. K. and S. C. KUMBHAKAR (2000): Stochastic Frontier Analysis, Cambridge Univer-

sity Press, Cambridge.

LOZANO-VIVAS, A., J. T. PASTOR and J. M. PASTOR (2002): “An Efficiency Comparison of Euro-

pean Banking Systems Operating under Different Environmental Conditions”, Journal of

Productivity Analysis, 18(1), 59-77.

MARTINS-FILHO, C. and F. YAO (2006): “Nonparametric Frontier Estimation Via Local Linear

Regression”, Journal of Econometrics, forthcoming.

MCCARTY, T. A. and S. YAISAWARNG (1993): “Technical Efficiency in New Jersey School Districts”,

in H. Fried, C. A. K. Lovell and S. Schmidt (editors), The Measurement of Productive Efficiency:

Techniques and Applications, chapter 10, pp. 271-287, Oxford University Press, Oxford.

MORDUCH, J. (1999): “The Microfinance Promise”, Journal of Economic Literature, 37(4), 1569-1614.

ONGENA, S. and SMITH, D. C. (2000): “Bank Relationships: A Review”, in P. T. Harker and S.

A. Zenios (editors), Performance of Financial Institutions. Efficiency, Innovation, Regulation,

chapter 7, pp. 221-258, Cambridge University Press, Cambridge.

PAGAN, A. and A. ULLAH (1999): Nonparametric Econometrics. Themes in Modern Econometrics,

Cambridge University Press, Cambridge.

PASTOR , J. M. and L. SERRANO (2005): “Efficiency, Endogenous and Exogenous Credit Risk in

the Banking Systems of the Euro Area”, Applied Financial Economics, 15, 631-649.

PÉREZ, F., V. MONTESINOS, L. SERRANO and J. FERNÁNDEZ DE GUEVARA (2005): La medición del ca-

pital social: una aproximación económica, Fundación BBVA, Bilbao.

— V. MONTESINOS, L. SERRANO and J. FERNÁNDEZ DE GUEVARA (2006): “Measurement of Social

Capital and Growth: An Economic Methodology”, Documentos de Trabajo 4, Fundación

BBVA, Bilbao.


49

PETERSEN, M. A. and R. G. RAJAN (1994): “The Benefits of Lending Relationships: Evidence

from Small Business Data”, Journal of Finance, 49(1), 3-37.

— (1995): “The Effect of Credit Market Competition on Lending Relationships”, Quarterly

Journal of Economics, 110(2), 407-443.

RAY, S. C. (1991): “Resource-Use Efficiency in Public Schools: A Study of Connecticut Data”,

Management Science, 37(12), 1620-1629.

ROUSE, P. (1996): “Alternative Approaches to the Treatment of Environmental Factors in

DEA: An Evaluation”, Paper presented at the Georgia Productivity Workshop, Athens.

RUGGIERO, J. (1996): “On the Measurement of Technical Efficiency in the Public Sector”, Eu-

ropean Journal of Operational Research, 90(3), 553-565.

— (1998): “Non-Discretionary Inputs in Data Envelopment Analysis”, European Journal of

Operational Research, 111(3), 461-469.

— (2004): “Performance Evaluation When Non-Discretionary Factors Correlate with Techni-

cal Efficiency”, European Journal of Operational Research, 159(1), 250-257.

SEALEY, C. W. and J. T. LINDLEY (1977): “Inputs, Outputs and a Theory of Production and

Cost at Depository Financial Institutions”, Journal of Finance, 32(4), 1251-1266.

SHEATHER, S. J. and M. C. JONES (1991): “A Reliable Data-Based Bandwidth Selection Method

for Kernel Density Estimation”, Journal of the Royal Statistical Society, Ser. B,53(3), 683-690.

SILVERMAN, B. W. (1986): Density Estimation for Statistics and Data Analysis, Chapman and Hall,

London.

SIMAR, L. and P. W. WILSON (1998): “Sensitivity Analysis of Efficiency Scores: How to Boot-

strap in Nonparametric Frontier Models”, Management Science, 44(1), 49-61.

— and P. W. WILSON (2006): “Estimation and Inference in Two-Stage, Semi-Parametric Mod-

els of Productive Efficiency”, Journal of Econometrics, forthcoming.

THANASSOULIS, E. (2001): Introduction to the Theory and Application of Data Envelopment Analysis,

Kluwer Academic Publishers, Boston.

TIMMER, C. P. (1971): “Using a Probabilistic Frontier Production Function to Measure Tech-

nical Efficiency”, Journal of Political Economy, 79(4), 776-794.

VEGA-REDONDO, F. (2006): “Building Up Social Capital in a Changing World”, Journal of Econom-

ic Dynamics & Control, forthcoming.

WAND, M. P. and M. C. JONES (1994): “Multivariate Plug-In Bandwidth Selection”, Computation-

al Statistics, 9, 97-116.

— and M. C. JONES (1995): Kernel Smoothing, Chapman and Hall, London.

ZAK, P. J. and S. KNACK (2001): “Trust and Growth”, The Economic Journal, 111(470), 295-321.

ZHU, J. (1996): “Data Envelopment Analysis with Preference Structure”, Journal of the Operation-

al Research Society, 47, 136-150.


50

A B O U T T H E A U T H O R S *

JOSÉ MANUEL PASTOR MONSÁLVEZ graduated in economics from the

University of Valencia in 1990 and received his PhD in 1996 from

the same university, where he is currently a lecturer. He has received

scholarships from several institutions (Valencian Regional Govern-

ment, Fundación Caja de Madrid, FIES). He has been a visiting

scholar at Florida State University (1996-1997) and an external con-

sultant of the World Bank. His research interests include banking

and regional economics. He has co-authored several books and has

published more than thirty articles in academic journals. He parti-

cipates in the National Research Project Nuevos tipos de capital, creci-

miento y calidad de vida: instrumentos de medida y aplicaciones.

E-mail: [email protected]

EMILI TORTOSA-AUSINA graduated in economics from the University

of Valencia and holds a doctorate in economics from the University

Jaume I (Castellón), where he is currently a lecturer in applied eco-

nomics. He has also taught in the Department of Economic Analy-

sis, University of Alicante (1994-1995) and has received scholarships

from various institutions (Fundación Caja Madrid, among others).

He has recently held posts as visiting researcher in the Business Eco-

nomics Department of the Autonomous University of Barcelona,

the School of Economics of the University of New South Wales (Syd-

ney, Australia), and the Department of Economics of Oregon State

Any comments on the contents of this paper can be addressed to Emili Tor-tosa-Ausina at [email protected].

* The results presented here are part of the BBVA Foundation ResearchProgramme. Emili Tortosa-Ausina acknowledges support from the SpanishScience and Education Ministry SEJ2005-01163. Thanks are also due to theparticipants in the workshop organized by BBVA Foundation-Ivie in Castellóde la Plana, February 2006.

University (Corvallis, Oregon, USA). His specialised fields are ban-

king economics and the analysis of efficiency and productivity. He

has published various books in collaboration with others and his ar-

ticles have appeared in specialized journals. He has participated in

many Spanish and international congresses and is an associate rese-

archer of the National Research Project Reestructuración productiva y

movilidad en la Nueva Economía.

E-mail: [email protected]

D O C U M E N T O S D E T R A B A J O

NÚMEROS PUBLICADOS

DT 01/02 Trampa del desempleo y educación: un análisis de las relaciones entre los efectosdesincentivadores de las prestaciones en el Estado del Bienestar y la educaciónJorge Calero Martínez y Mónica Madrigal Bajo

DT 02/02 Un instrumento de contratación externa: los vales o cheques.Análisis teórico y evidencias empíricasIvan Planas Miret

DT 03/02 Financiación capitativa, articulación entre niveles asistencialesy descentralización de las organizaciones sanitariasVicente Ortún-Rubio y Guillem López-Casasnovas

DT 04/02 La reforma del IRPF y los determinantes de la oferta laboralen la familia españolaSantiago Álvarez García y Juan Prieto Rodríguez

DT 05/02 The Use of Correspondence Analysis in the Explorationof Health Survey DataMichael Greenacre

DT 01/03 ¿Quiénes se beneficiaron de la reforma del IRPF de 1999?José Manuel González-Páramo y José Félix Sanz Sanz

DT 02/03 La imagen ciudadana de la JusticiaJosé Juan Toharia Cortés

DT 03/03 Para medir la calidad de la Justicia (I): AbogadosJuan José García de la Cruz Herrero

DT 04/03 Para medir la calidad de la Justicia (II): ProcuradoresJuan José García de la Cruz Herrero

DT 05/03 Dilación, eficiencia y costes: ¿Cómo ayudar a que la imagen de la Justiciase corresponda mejor con la realidad?Santos Pastor Prieto

DT 06/03 Integración vertical y contratación externa en los serviciosgenerales de los hospitales españolesJaume Puig-Junoy y Pol Pérez Sust

DT 07/03 Gasto sanitario y envejecimiento de la población en EspañaNamkee Ahn, Javier Alonso Meseguer y José A. Herce San Miguel

DT 01/04 Métodos de solución de problemas de asignación de recursos sanitarios Helena Ramalhinho Dias Lourenço y Daniel Serra de la Figuera

DT 01/05 Licensing of University Inventions: The Role of a Technology Transfer OfficeInés Macho-Stadler, David Pérez-Castrillo y Reinhilde Veugelers

DT 02/05 Estimating the Intensity of Price and Non-price Competition in Banking:An Application to the Spanish CaseSantiago Carbó Valverde, Juan Fernández de Guevara Radoselovics, David Humphrey

y Joaquín Maudos Villarroya

DT 03/05 Sistemas de pensiones y fecundidad. Un enfoque de generaciones solapadasGemma Abío Roig y Concepció Patxot Cardoner

DT 04/05 Análisis de los factores de exclusión socialJoan Subirats i Humet (Dir.), Ricard Gomà Carmona y Joaquim Brugué Torruella (Coords.)

DT 05/05 Riesgos de exclusión social en las Comunidades AutónomasJoan Subirats i Humet (Dir.), Ricard Gomà Carmona y Joaquim Brugué Torruella (Coords.)

DT 06/05 A Dynamic Stochastic Approach to Fisheries Management Assessment:An Application to some European FisheriesJosé M. Da-Rocha Álvarez y María-José Gutiérrez Huerta

DT 07/05 The New Keynesian Monetary Model: Does it Show the Comovement between Output and Inflation in the U.S. and the Euro Area?Ramón María-Dolores Pedrero y Jesús Vázquez Pérez

DT 08/05 The Relationship between Risk and Expected Return in EuropeÁngel León Valle, Juan Nave Pineda y Gonzalo Rubio Irigoyen

DT 09/05 License Allocation and Performance in Telecommunications MarketsRoberto Burguet Verde

DT 10/05 Procurement with Downward Sloping Demand: More Simple EconomicsRoberto Burguet Verde

DT 11/05 Technological and Physical Obsolescence and the Timing of AdoptionRamón Caminal Echevarría

DT 01/06 El efecto de la inmigración en las oportunidades de empleode los trabajadores nacionales: Evidencia para EspañaRaquel Carrasco Perea, Juan Francisco Jimeno Serrano y Ana Carolina Ortega Masagué

DT 02/06 Inmigración y pensiones: ¿Qué sabemos?José Ignacio Conde-Ruiz, Juan Francisco Jimeno Serrano y Guadalupe Valera Blanes

DT 03/06 A Survey Study of Factors Influencing Risk Taking Behaviorin Real World Decisions under UncertaintyManel Baucells Alibés y Cristina Rata

DT 04/06 Measurement of Social Capital and Growth:An Economic MethodologyFrancisco Pérez García, Lorenzo Serrano Martínez, Vicente Montesinos Santalucía

y Juan Fernández de Guevara Radoselovics

DT 05/06 The Role of ICT in the Spanish Productivity SlowdownMatilde Mas Ivars y Javier Quesada Ibáñez

DT 06/06 Cross-Country Comparisons of Competition and Pricing Powerin European BankingDavid Humphrey, Santiago Carbó Valverde, Joaquín Maudos Villarroya y Philip Molyneux

DT 07/06 The Design of Syndicates in Venture CapitalGiacinta Cestone, Josh Lerner y Lucy White

DT 08/06 Efectos de la confianza en la información contable sobre el coste de la deudaBelén Gill de Albornoz Noguer y Manuel Illueca Muñoz

DT 09/06 Relaciones sociales y envejecimiento saludableÁngel Otero Puime, María Victoria Zunzunegui Pastor, François Béland,

Ángel Rodríguez Laso y María Jesús García de Yébenes y Prous

DT 10/06 Ciclo económico y convergencia real en la Unión Europea:Análisis de los PIB per cápita en la UE-15José Luis Cendejas Bueno, Juan Luis del Hoyo Bernat, Jesús Guillermo Llorente Álvarez,

Manuel Monjas Barroso y Carlos Rivero Rodríguez

DT 11/06 Esperanza de vida en España a lo largo del siglo XX:Las tablas de mortalidad del Instituto Nacional de EstadísticaFrancisco José Goerlich Gisbert y Rafael Pinilla Pallejà

DT 12/06 Convergencia y desigualdad en renta permanente y corriente: Factores determinantesLorenzo Serrano Martínez

DT 13/06 The Common Agricultural Policy and Farming in Protected Ecosystems:A Policy Analysis Matrix ApproachErnest Reig Martínez y Vicent Estruch Guitart

DT 14/06 Infrastructures and New Technologies as Sources of Spanish Economic GrowthMatilde Mas Ivars

DT 15/06 Cumulative Dominance and Heuristic Performancein Binary Multi-Attribute ChoiceManel Baucells Alibés, Juan Antonio Carrasco López y Robin M. Hogarth

DT 16/06 Dynamic Mixed Duopoly: A Model Motivated by Linux versus WindowsRamon Casadesus-Masanell y Pankaj Ghemawat

DT 01/07 Social Preferences, Skill Segregation and Wage DynamicsAntonio Cabrales Gaitia, Antoni Calvó-Armengol y Nicola Pavoni

DT 02/07 Stochastic Dominance and Cumulative Prospect TheoryManel Baucells Alibés y Franz H. Heukamp

DT 03/07 Agency RevisitedRamon Casadesus-Masanell y Daniel F. Spulber

4Documentosde Trabajo4Documentos

de Trabajo2007

José Manuel Pastor MonsálvezEmili Tortosa-Ausina

Social Capital andBank PerformanceAn International Comparison for OECD Countries

Gran Vía, 1248001 BilbaoTel.: 94 487 52 52Fax: 94 424 46 21

Paseo de Recoletos, 1028001 MadridTel.: 91 374 54 00Fax: 91 374 85 22

[email protected]

2007-04 9/2/07 11:17 Página 1

Documents

4 de Trabajo 4 - FBBVA · Todos los documentos de trabajo están disponibles, de forma gratuita y en for-mato PDF, en la web de la Fundación BBVA. Si desea una copia impresa, puede