Foreign Bank and Domestic

omtru

ouniric

highly protable. Moreover, when foreign banks operate in a host country with lower growth rates ofGDP, higher interest and ination rates, and more stringent regulatory compliance with Basel riskweights, their margins increase. Specically, changes in bank supervision of a parent banks ownershiprestrictiveness in the home country signicantly increases foreign bank margins, while supervisorychanges in regulatory compliance with Basel risk weights in the host country enhances foreign bank

ies havened

become dominant. Dietz et al. (2008) nd that according to 2006survey on global industry prot revenues and prots in the bank-ing industry amounted to $788 billion, which was higher thanprots in any other industry. The authors also indicate thatbetween 2000 and 2006 the prots of developed countries grewsignicantly faster than those in less developed countries. This

bank protability (e.g., Molyneux and Thorton, 1992; Ho andSaunders, 1998; Demirg-Kunt and Huizinga, 2000; Goddardet al., 2004). Jaffee (1989) nds that the interest spread was inu-enced by (i) the degree of market concentration that affected bankprotability; (ii) regulatory constraints that prohibited the bankfrom undertaking certain protable activities and increased thecost of providing permissible activities; (iii) higher credit risk;and (iv) exposure to interest-rate risk. Some researchers haverecently examined the impact of GDP on bank prots (Brock andSuarez, 2000; Staikouras and Wood, 2003; Williams, 2003; Claeys

Corresponding author. Tel.: +886 5 2721001x56541; fax: +886 5 2427172.E-mail addresses: [email protected] (S.-H. Chen), macrossgods@

Journal of Banking & Finance 35 (2011) 819839

Contents lists availab

k

w.yahoo.com.tw (C.-C. Liao).banks, thereby increasing their levels of nancial liberalizationand integration and boosting the banking industry worldwide.The globalization of banking has led to institutional and regulatoryimprovements and benetted both local and foreign banks. At thelocal level, globalization has increased the competitiveness ofbanking markets by reducing administrative costs, lowering netinterest margins, and driving down bank rates of return. At theinternational level, globalization has allowed foreign banksespe-cially those from more developed nancial systemsto expandinto emerging market economies, where they have sometimes

protability. Their ability to do so involves both internal and exter-nal determinants. Internal determinants encompass the manage-ment decisions made by each bank and are related to the bankslevel of liquidity, provisioning policy, bank size, capital adequacy,and expense management. External determinants encompass mac-roeconomic factors beyond the banks control, such as the legalenvironment, the state of the economy in which a nancial institu-tion operates, changes in national governance, and the impact ofglobalization.

Previous studies have evaluated how various factors impactG21G28

Keywords:Bank protabilityForeign bankBanking competitionPanzar and Rosse HGovernanceCountry riskRegulatory supervision

1. Introduction

Over the last few decades, countrnational nancial activities and op0378-4266/$ - see front matter Crown Copyright 2doi:10.1016/j.jbankn.2010.11.006margins.Crown Copyright 2010 Published by Elsevier B.V. All rights reserved.

e extended their inter-their doors to foreign

shows the role of protability in banking industry is essentiallyimportant and draws higher interest and concern from academicsand practitioners.

All bankswhether domestic or foreignseek to enhance theirJEL classication:E32L11

ket structure, macroeconomic condition, governance, and changes in bank supervision on foreign bankmargins. We nd that foreign banks are more protable than domestic banks when they operate in a hostcountry whose banking sector is less competitive and when the parent bank in the home country isAre foreign banks more protable than dhost-country effects of banking market s

Sheng-Hung Chen , Chien-Chang LiaoDepartment of Finance, Nanhua University, Chiayi, Taiwan

a r t i c l e i n f o

Article history:Available online 20 November 2010

a b s t r a c t

Using both bank- and c19922006, this paper emp

Journal of Ban

journal homepage: ww010 Published by Elsevier B.V. Allestic banks? Home- andcture, governance, and supervision

try-level data on banking sectors from 70 countries over the periodally investigates the joint home- and host-country effects of banking mar-

le at ScienceDirect

ing & Finance

elsevier .com/locate / jbfrights reserved.

measures, such as Panzar and Rosse H (PR H) statistics and the

The rest of this paper is organized as follows. Section 2 reviews

ankiand Vennet, 2008), while others have assessed the impact ofination on bank prots (Kosmidou et al., 2007; Pasiouras andKosmidou, 2007; Athanasoglou et al., 2008) and the ambiguouseffect of corporate income tax on net interest margin (Albertazziand Gambacorta, 2010). Recent literature has also analyzed theeffect of interest rates and banking market structure on such areasas bank protability (Saunders and Schumacher, 2000; Maudosand Fernndez de Guevara, 2004); credit risk (Angbazo, 1997;Maudos and Fernndez de Guevara, 2004); non-interest revenue(Spathis et al., 2002); and bank-specic factors like interest rates,loans to assets, and a banksmarket shares (Claeys andVennet, 2008).

To evaluate a banks performance and determine whether a po-sitive empirical relationship exists between market concentrationand bank protability, researchers frequently use the paradigmof StructureConductPerformance (Kosmidou et al., 2007;Pasiouras and Kosmidou, 2007). When such a relationship exists,it strongly implies that a bank will be able to gain a monopoly asits market concentration increases. Furthermore, Garca-Herreroet al. (2009) nd that better capitalized Chinese banks tend to bemore protable and indicate that a less concentrated bankingsystem increases bank protability. Delis and Tsionas (2009)show that there is a negative relationship between efciency andmarket power in line with the quiet life hypothesis. In addition,Maudos and Solisa (2009) nd that Mexican banking industry withhigh margins can be explained mainly by average operating costsand by market power.

Over the last few decades, many developing countries have lib-eralized their nancial policies and begun encouraging the entry offoreign banks into the domestic banking market. As a result, themarket structure of the banking industry has changed signicantly.The traditional banking industry perspective suggests that whenbanks enter a new market, it leads to more competition amongbanks already in the market, thereby benetting borrowers butultimately harming local banks monopoly rents. Thorne (1993)nds that the entry of foreign banks as a whole into a domesticmarket has a positive effect on the domestic market due to thespillover effect of the foreign banks know-how and expertise.Claeys and Hainz (2006) conduct a theoretical analysis thatshowed foreign bank entry would drive down a countrys averageinterest rate for new loans. Peria and Mody (2004) nd that theentry of foreign banks into Latin American markets decreased thelevel of bank concentration. Maudos and Fernndez de Guevara(2004) demonstrate that between 1993 and 2000, banking sectorsin Europe (including Germany, France, the United Kingdom, Italy,and Spain) proxied both the Herndahl index and the Lerner index,which measure the degree of concentration and market power.Their results showed a signicant positive impact of marketconcentration on the interest margin.

In regard to whether foreign banks or domestic banks performbetter in the same market, research has been contradictory. Somestudies have found no signicant differences between domesticand foreign banks (Vennet, 1996; Hasan and Lozano-Vivas, 1998;Crystal et al., 2001; Mian, 2003). Some studies have found thatdomestic banks perform better (Sturm and Williams, 2004) andthat foreign banks are disadvantaged when compared to domesticbanks in developed countries (Peek et al., 1999; Berger et al., 2000;Claessens et al., 2001; Sathye, 2001). One reason for this is that for-eign banks may lack knowledge of the specic market at the timeof entrance (DeYoung and Daniel, 1996; Mahajan et al., 1996;Berger et al., 2000; Kosmidou et al., 2004). Some studies havefound that foreign banks are more protable and more efcientthan domestic banks in emerging markets (Claessens et al., 2001;Demirg-Kunt and Huizinga, 2000; Bonin et al., 2005; Grigorian

820 S.-H. Chen, C.-C. Liao / Journal of Band Manole, 2006; Berger et al., 2009b). Others, however, havefound the opposite (Nikiel and Opiela, 2002; Yildirim andPhilippatos, 2007).related literature dealing with international studies of bank prot-ability. Section 3 describes the model for analysis used in the the-oretical framework of this paper. Section 4 presents the empiricalmodel used for estimation, explains the data collection process,and provides summary statistics. Section 5 discusses the resultsof our empirical models. Finally, Section 6 presents our concludingremarks.

2. Related literature

Previous literature related to bank protability can be classiedinto two major categories. The rst is a cross-country comparisonof bank protability, and the second is the impact of banking mar-ket structure on bank prot. We discuss our detailed ndings in thefollowing sections and summarize the important empirical resultsin Appendix A.

2.1. International comparison of bank protability

The empirical literature on bank prot focuses strongly on Euro-pean countries. One exception to this is Williams (1996), who con-structs a model that attempts to explain the performance ofJapanese nancial institutions in Australia. In general the modelperforms well for measures of size, but comparatively poorly formeasures of protability. Brock and Suarez (2000) report that bankspreads in the 1990s were inuenced by liquidity and capital riskat the bank level and by interest rate volatility, ination, andGDP growth at the macroeconomic level. Claessens et al. (2001)analyze 7900 banks from 80 countries for the period 19881995and found that foreign banks enjoyed higher prots than domesticbanks when operating in developing countries. They found theopposite effect, however, for banks operating in developedcountries.

Kosmidou et al. (2007) analyze 19 Greek bank subsidiariesLerner index, as well as static measures, such as the Herndhal Hir-schman Index (HHI) and the Concentration Ratio (CR), over theperiod 19922006. Second, we empirically examine whethercross-country differences in banking market structure, macroeco-nomic environment, institutional governance, banking competi-tion, and country risk between host country and home countryinuence foreign protability. Third, we use regulation and super-vision variables to explore the joint home- and host-country im-pacts of change in bank supervision on foreign bank protability.One possible reason for such contradictory results may lie in thefact that empirical analysis for the performance of foreign bankshas mainly concentrated on European Union and US banks operat-ing abroad. Another may be that most studies have compared theperformance of domestic banks with foreign banks for a singlecountry rather than analyzing how cross-country differences inbanking market structure inuence a banks protability. Forexample, Williams (1996, 1998a,b, 2003) studies Australia, MinhTo and Tripe (2002) studies New Zealand, Ursacki and Vertinsky(1992) focuses on Japan and Korea, Engwall et al. (2001) studiesthe Nordic countries, and Dietrich and Wanzenried (2009) studiesSwitzerland.

This paper contributes to previous studies in several ways. First,we empirically investigate the key factors in 70 countries that af-fect bank protability in foreign banks when compared to domesticbanks. To do so, we analyze the long-term relationship betweenbank protability and banking market structure using structural

ng & Finance 35 (2011) 819839operating in 11 nations during the period 19952001. Their nd-ings show that the protability of the parent bank and the operat-ing experience of the host nation had a robust and positive impact

countries (WEEC) over the years 19942001. They nd that capital

differ depending on the measure chosen to assess it. This is because

ankiindicators of competition measure different things. Berger et al.(2009b) teste indicators of market power tomeasure loan risk, bankrisk, bank equity capital, and the business environment. The resultssuggest that banks with a higher degree of market power have lessoverall risk exposure and that market power increases loan portfo-lio risk. In summary, these researchers indicate that bankingmarketstructure does have a substantial inuence on bank protability.

3. The model

3.1. Theoretical frameworkratio and market concentration were positive and signicant forboth the WEEC and accession countries. Claessens and Horen(2009) explore the performance of foreign banks in 51 countriesbetween 1999 and 2006 and nd that foreign banks from high in-come countries tended to perform better than domestic banks.They also indicate that foreign banks from developing countriestend to perform worse than domestic banks.

2.2. Banking market structure and bank protability

Banking structure clearly inuences protability persistence inboth domestic and foreign banks and plays a critical role in under-standing international competitive advantages. A number of stud-ies have identied determinants of bank protability in selectivecountries and regions, but the impact of banking market structureon bank protability across countries has seldom been studiedempirically, particularly in the context of international compari-son. Lately, numerous studies have evaluated the effect of marketstructure on bank protability in various countries, but interna-tional evidence based upon cross-country differences in the bank-ing market structure remains sparse.

Saunders and Schumacher (2000) assess the impact of marketstructure on bank prot and found that during the period 19881995 interest margins in six European countries and the UnitedStates were affected by the degree of bank capitalization, bankmarket structure, and volatility of interest rates. Corvoisier andGropp (2002) show that higher concentration might have resultedin less competitive pricing by banks located in the euro area for theperiod 19931999. Bikker and Haaf (2002) apply the PR H modelto measure banking competition in 23 countries. Their results sup-ported the conventional view that concentration impairs bankingcompetitiveness. However, Beck et al. (2003) conclude thathighly-concentrated banking systems are less likely to suffer fromcrises. Maudos and Fernndez de Guevara (2004) use the Lerner in-dex proxied as the degree of competition in banking markets andfound that a statistically signicant and positive correlation existedbetween concentration and bank interest margins in Europeanbanking markets for the period 19932000.

Additionally, Athanasoglou et al. (2008) utilize GMM (General-ized Method of Moments) technique to a panel of Greek banks forthe period 19852001 and found that concentration negatively af-fected the banks protability; however, this effect was relativelyinsignicant. To infer competitive behavior, Carb et al. (2009) em-ploy ve indicators: loan-deposit interest spread, Lerner index, abanks net income to the value of total assets, PR H statistics,and the HHI. They nd that the determination of competition canon the prots of Greek banks abroad. Claeys and Vennet (2008)investigate the determinants of bank interest margins by taking asample of 1130 banks from 31 Western and Eastern European

S.-H. Chen, C.-C. Liao / Journal of BBased on a theoretical perspective proposed by Ho andSaunders (1998) and Saunders and Schumacher (2000), this studydenes a bank as a risk-averse dealer in the credit market whosepurpose is the immediate provision of deposits and loans. Thus,the major portfolio risk comes from interest-rate uctuations, orvolatility, before any deposits or loans are made. Risk-averse banksfacing asymmetric arrival time for the demand for loans and thesupply of deposits select optimal loan and deposit rates, whichminimize the risk of excessive demand for loans or the insufcientsupply of deposits. Therefore, we set up rates using the followingstructure:

RD R a; RL R b 1

where RD is the rate set on deposits, RL is the rate set on loans, R isthe expected risk-free interest rate, and a and b are fees charged bythe bank in order to provide immediacy and to bear interest-raterisk. The fees a and b have to compensate the bank for bearing thisinterest-rate risk. The optimal fees a, b and thus the spreadsp = (a + b)3 are expressed as:

sp a b ch 12Rr2I Q 2

where ch measures the banks risk neutral spread and is the ratio ofthe intercept (c) and the slope (h) of the symmetric deposit and loanarrival functions of the bank. A large c and a small h will result in alarge ch and a large spread (sp). This means that if a bank faces rela-tively inelastic demand and supply functions in the markets inwhich it operates, it may be able to exercise monopoly power bydemanding a greater spread than it could get if banking marketswere competitive (low ch ratio).

The ratio of ch provides some measure of the producers surplusor monopoly rent element in bank spreads or margins. The secondterm is a rst-order risk-adjustment term that depends on threefactors: (i) R, the bank managements coefcient of absolute riskaversion; (ii) Q, the size of bank transactions; and (iii) r2, theinstantaneous variance of the interest rate on deposits and loans.Note that the second term implies that, the greater the degree ofrisk aversion, the larger the size of transactions. It also implies thatthe greater the variance of interest rates, the larger the bank mar-gins. This spread equation has an important implication: Even ifbanking markets are highly competitive, as long as a banks man-agement is risk-averse and faces transaction uncertainty, positivebank margins will exist as the price of providing deposit and loanintermediation.

3.2. Measuring banking market structure

This paper uses four measures proxies as the market structurein the banking industry. Two are structural: PanzarRosse Hstatistics and the Lerner index, and two are static: the HerndhalHirschman Index and the Concentration Ratio of the top four banks(CR4).

3.2.1. Structural measures3.2.1.1. PanzarRosse H statistics. Panzar and Rosse (1987) create aconvenient framework with which to assess banking market struc-ture. Their modelthe PanzarRosseuses bank-level data to mea-sure how a change in factor input prices is reected in equilibriumrevenues earned by banks. In a state of perfect competition, mar-ginal costs and total revenues increase proportionally to inputprices. In a state of monopoly market, however, an increase in fac-tor input prices raises marginal costs but reduces output and totalrevenues. Following Bikker and Haaf (2002), Bikker et al. (2007),and Bikker and Spierdijk (2008), this paper uses the PR H model

ng & Finance 35 (2011) 819839 821to measure the degree of competition in the banking industry byestimating, for each country, the following empirically reducedform of bank revenues:

ankilnpi;t a b ln PFi;t c ln PLi;t d ln PKi;t Xj

nj gOI ei;t

3where t is the number of periods observed, and i is the total numberof banks. ln denotes the natural logarithmic operator, p denotes abanks interest revenues, PF stands for annual funding rate calcu-lated as interest expense to total funds (IE/FUN) (as the proxy forunit price of fund), PL denotes price of personnel expenses calcu-lated as annual personnel expenses to total assets (PE/TA) (as theproxy for unit price of labor), PK is the price of physical capitalexpenditure calculated as non-interest expenses to xed assets(ONIE/FA) (as the proxy for unit price of capital), and OI is calculatedas the ratio of the income to total assets. In order to apply the PRapproach to our data, this paper uses the following empirical re-duced form equation of bank revenues, which is in line with Eq. (3):

lnpi;t a b ln PFi;t c ln PLi;t d ln PKi;t g1 ln LNASSTi;t g2 lnNONTAi;t g3 lnDPSFi;t g5 lnEQTAi;t g5OI ei;t

4where LNASST is the ratio of loans to total assets and represents thecredit risk on banks. NONTA is dened as the ratio of other non-earning assets to total assets. DPSF is the ratio of customer depositsto the sum of customer deposits and short-term funding. EQTA isgenerated by the ratio of equity to total assets ratio. Although thechoice of dependent and explanatory variables often varies, Eq.(4) is similar to other equations commonly used in the literature.

Themodel posits that banks use three input factors: deposits, la-bor, and physical capital. PF, PL and PK are the unit prices of thesethree inputs, or reasonable proxies. A number of bank-specic fac-tors are included into Eq. (4). The H statistic is calculated as thesum of the elasticities of a banks total revenue with respect to thatbanks input prices. Hence, based on Eq. (4), the H statistic equalsH = a + b + c. The estimated value of the H statistic ranges between1 and 1. Under perfect competition, a decrease in input prices re-duces marginal costs and revenues by the same amount as a costreduction (H = 1). A value of theH statistic between0 and1 indicatesmonopolistic competition. Values equal to or less than 0 are consis-tentwithmonopoly behavior, as a decrease in input prices decreasesmarginal costs but does not reduce revenues. (For a summary of pre-vious empirical ndings using H statistic, see Appendix B.)

3.2.1.2. Lerner index. The Lerner index can be characterized as thenegative inverse of demand elasticity. It represents the mark-upof price over marginal costs and is an indicator of the degree ofmarket power. For example, a high Lerner index indicates a strongdegree of monopoly in the banking market, while a low Lerner in-dex indicates a highly competitive market with less capacity to sethigh margins. In effect it is a level indicator of the proportion bywhich price exceeds marginal cost. Following the theoretical per-spective of Berger et al. (2009b), this paper uses the Learner indexas a proxy for market power. Its empirical expression is:

LERNER pMCp

5

where p is the average price of a bank (proxied as the quotient be-tween total revenues and total assets) and MC is the marginal costof total assets calculated from the estimation of a translogarithmiccosts function. Where the total costs depend on the prices of threeinputs and on the banks volume of production (total assets), thecosts function estimated is:

lnCostit bo b1 lnQit 12b2 lnQ

2it

X3k1

ckt lnWk;it

822 S.-H. Chen, C.-C. Liao / Journal of BX3k1

uk lnQit lnWk;it X3k1

X3j1

lnWk;it lnWj;it eit 6where Qit represents a proxy for bank output or total assets for banki at time t, andWk,it represent three input prices.W1,it,W2,it andW3,itindicate the input prices of labor, funds, and xed capital, respec-tively. The input prices are dened as: (W1,it) personnel costs/totalassets; (W2,it) interest expenses/total deposits; and (W3,it) otheroperating and administrative expenses/total assets.

MCTAit CostitQit

b1 b2 lnQit X3k1

u lnWk;it

" #7

Finally, the Lerner index is averaged over time for each bank i forinclusion in the regression model. The data for this measure of com-petition are at the bank level and are estimated country-by-countryand year-by-year.

3.2.2. Static measures3.2.2.1. Herndhal Hirschman Index (HHI). We dene the HHI as:

HHI X3i1

s2i 8

where si stands for the share of total volume of loans (deposits orassets) of the group of nancial institutions. The HHI is calculatedby summing the squared market shares of all banks in the industryon the assumption that competition takes place on a national scale.Only in the case of big banks and wholesale markets could a greaterthan national market be assumed. The fewer the rms in the bank-ing industry, and the more the industry is dominated by a smallnumber of them, the higher the value of the index.

3.2.2.2. Concentration Ratio. The CR describes the degree of bankingcompetition in a national banking sector. This indicator could bemeasured as the combined market share of the four biggest banks(CR4) in terms of total assets (deposits or assets). Williams (2003)suggests that CR may act as a barrier to entry when entering mar-kets where domestic banks are highly-concentrated. This implies anegative impact on prots.

4. Empirical model, data and summary statistics

4.1. Empirical model

Our baseline model aims to indentify cross-country determi-nants of bank performance using comprehensive panel data at boththe bank and country level. Following previous studies on empiricalspecications, suchas those conductedbyMaudosandFernndezdeGuevara (2004), Berger et al. (2009a), Claessens and Horen (2009),and Dietrich and Wanzenried (2009), we incorporate a number ofindependent variables in our empiricalmodel, including net interestmargins, return on assets (ROA), and return on equity (ROE). For adetailed description of these, see Appendix C.

Pi;j;t a0 a1FBi;j;t a2COSTi;j;t a3SIZEi;j;t a4LIi;j;t a5CRi;j;t a6IPi;j;t a7OPCi;j;t a8NOIi;j;t a9OBSi;j;t a10ELi;j;t a11MSi;j;t a12OOIi;j;t a13GDPHostj;t a14IRHostj;t a15IFHostj;t ei;j;t 9

where Pi,t in Eq. (9) stands for bank protability for bank i in coun-try j in year t and ei,j,t represents the error term. To represent depen-dent variables, we use three measures of a banks protabilityalternately: net interest margin (NIM), ROA, and ROE. NIM is thenet interest margin generated by the net interest income (=interestincome interest expense) divided by current assets. This ratio sug-

ng & Finance 35 (2011) 819839gests that the higher the net interest margin, the better the perfor-mance. ROA is dened as the net prot divided by total assets; itrepresents the earning performance of a bank based on total assets.

ankiROE is calculated as the return on equity, which is the net protafter tax divided by shareholder equity. It represents the earningperformance of a bank based on the shareholders stake.

As independent variables, we include bank nancial character-istics, foreign bank ownership, and comprehensive measures ofbanking market structure, macroeconomics, and governancebetween home and host country. FBi,j,t which indicates foreignownership, is the dummy variable. It is equal to one for foreign-owned banks whose shareholding is up to 50% or more. Domesti-cally-owned banks are equal to 0. The main control variables forbank nancial characteristics include: operation cost (COSTi,j,t),interest payments (IPi,j,t), liquidity ratio (Lii,j,t), opportunity cost(OPCi,j,t), credit risk (CRi,j,t), bank size (SIZEi,j,t), non-interest reve-nues (NOIi,j,t), capital strength (ELi,j,t), off-balance sheet activity(OBSi,j,t), market shares (MSi,j,t), and other operating income(OOIi,j,t). Three macroeconomic variables of the host country arealso used for regression: growth rate of GDP (GDPHostj;t , real interestrate (IRHostj;t , and ination rate (IFHostj;t .

4.1.1. The impact of banking competition on bank protabilityBased on the specications of Eq. (9), we investigate the impacts

of banking market structure on bank protability. In particular, weattempt to verify the explanatory power of banking competitionusing both structure measures (PR H statistics and Lerner index)and static measures (HHI and CR4). Hence, we incorporate themeasure of banking market structure into the following model:

Pi;j;t b0 b1FBi;j;t b2COSTi;j;t b3SIZEi;j;t b4LIi;j;t b5CRi;j;t b6IPi;j;t b7OPCi;j;t b8NOIi;j;t b9OBSi;j;t b10ELi;j;t b11MSi;j;t b12OOIi;j;t b13MSTj;t b14GDPHostj;t b15IRHostj;t b16IFHostj;t ni;j;t 10

where MSTj,t represents the proxy as banking market structure in acountry j in year t. It is measured and generated by four indicators:HHI, CR4, Lerner index and PR H statistic. We also use a foreignbanks ownership (FBi,j,t) and the same control variables used inEq. (9). We apply the Maximum Likelihood Estimation (MLE) meth-od for panel data model with Random Effects (RE) to estimatecoefcients.

4.1.2. Joint home- and host-country effects of banking competition,macroeconomic condition, governance, and supervision on bankprotability

Kosmidou et al. (2007) nd that the inuence of parent bankprotability had a signicant effect on the prots of subsidiaryGreek banks. To quantify home- and host-country effects on bankprotability, therefore, we incorporate the interaction term intothe following model with respect to banking competition, macro-economic condition (degree of development and country risk), gov-ernance, and bank supervision between a foreign banks homecountry and host country.

Pi;j;t c0 c1FBi;j;t c2Parent Performacei;m;t c3MSTj;tXh

c4hContoli;j;t c5FBi;j;t DEVHomei;j;t c6FBi;j;t DEVHosti;j;t

Xp

c7pFBi;j;t MacEconHostj;t c7pFBi;j;t MacEconHomej;t

Xq

c8qFBi;j;t RiskHostj;t c8qFBi;j;t RiskHomej;t

Xr

c9rFBi;j;t GOVHostj;t c9rFBi;j;t GOVHomej;t

S.-H. Chen, C.-C. Liao / Journal of BXs

c10sFBi;j;t SUPHostj;t c10sFBi;j;t SUPHomej;t

ni;j;t 11where Parent Performacei,m,t represents a parent banks performanceof foreign bank in home country m. We use two indicators to mea-sure parent banks performance in home country: protability leveland protability ranking. The protability level of parent bank in-cludes NIM, ROA, and ROE in response to foreign banks prot mea-sures. Protability ranking of parent bank indicates the comparativeprotability in the home country. This relative indicator is denedas the higher the protability for a parent bank, the larger the va-lue of ranking for the indicator.

Based on Claessens and Horen (2009), we also use where a for-eign bank comes from (either a developed country or a developingcountry) as a dummy variable. The variable is equal to one if theforeign bank comes from a developing country; if it comes froma developed country, it is equal to 0. FBi;j;t DEVki;j;t denotes a for-eign banks location: either the home country from which the for-

eign bank originates FBi;j;t DEVHomei;j;t

or the host country in

which the foreign bank operates and competes with domestic

banks FBi;j;t DEVHosti;j;t

. FBi;j;t MacEconHomej;t measures the indirecteffect of macroeconomic conditions in the home country on foreignbank protability. FBi;j;t MacEconHostj;t denotes the direct effect ofmacroeconomic conditions on the host country on all bank mar-gins, including both foreign and domestic banks. Our regressionsinclude three macroeconomic variables dened earlier as: growthrate of GDP per capita, ination rate, and real interest rate.

In addition, country risk plays an important role on bank prof-itability. Therefore, we use country-level variables, such as legalrisk (LR) and economic risk (ER), in our regression to control for riskdifferences across countries. FBi;j;t RiskHostj;t describes the hostcountry risks for a foreign bank, while FBi;j;t RiskHomej;t describesthe risks associated with the foreign banks home country. Further-more, cross-country differences in governance between homecountry and host country have potential inuences on bank perfor-mance. FBi;j;t GOVHostj;t measures the quality of government in thehost country, while FBi;j;t GOVHomej;t measures the quality of gov-ernment in the home country.

Following Kaufmann et al. (2007), we use three important gov-ernance indexes. The rst is Government Effectiveness (GE), whichwe dene as the quality of public services, the quality of the civilservice, the degree of independence from political pressure, thequality of policy formulation and implementation, and the credi-bility of the governments commitment to such policies. The sec-ond is Regulatory Quality (RQ), which we dene as the ability ofthe government to formulate and implement sound policies andregulations that permit and promote private sector development.The third is Control of Corruption (CO), which we dene as the ex-tent to which public power is exercised for private gain, includingboth petty and grand forms of corruption, as well as capture of thestate by elites and private interests. FBi;j;t GOVHomej;t stands forhome country governance, while FBi;j;t GOVHostj;t stands for hostcountry governance.

Following Barth et al. (2004), we next consider 10 dimensions ofbanking regulation and supervision (SUP) and investigate the homeand host country impacts of such supervision on bank protability.We include nine supervisory variables in our regression: 1. BankCapital Requirements (whether related parties can own capital ina bank); 2. Bank Ownership Restrictiveness (regulatory restrictive-ness of ownership by one nancial rm or bank); 3. Basel riskweights (whether the bank is risk-weighted in line with Basleguidelines); 4. Securities Activities; 5. Licensed/Certied Auditor(whether auditors are licensed or certied); 6. Auditors Report(whether an auditors report is given to a supervisory agency); 7.Consolidated Accounts (whether consolidated accounts covering

ng & Finance 35 (2011) 819839 823the bank and any one-bank nancial subsidiaries are required);8. Solvency Intervention (whether the law establishes pre-determined levels of solvency deterioration that forces automatic

actions such as intervention); 9. Onsite Inspections (frequency ofonsite inspections conducted in large and medium sized banks).

4.1.3. Joint home- and host-country effects of changes in banksupervision on bank protability

To further examine the home- and host-country effects ofsupervisory changes on bank protability, we use the rst differ-encing calculation (D) for bank supervision variables with twotime periods between t-1 and t and then estimate the followingregression:

Pi;j;t c0 c1FBi;j;t c2Parent Performacei;m;t c3MSTj;t

X11h1

c4hContoli;j;t c5FBi;j;t DEVHomei;j;t c6FBi;j;t DEVHosti;j;t

Xp

c7pFBi;j;t MacEconHostj;t c7pFBi;j;t MacEconHomej;t

Xq

c8qFBi;j;t RiskHostj;t c8qFBi;j;t RiskHomej;t

Xr

c9rFBi;j;t GOVHostj;t c9rFBi;j;t GOVHomej;t

Xs

c10sFBi;j;t DSUPHostj;t c10sFBi;j;t DSUPHomej;t

ni;j;t

12Basically, we follow the specications of Eq. (10). DSUPHomej;t repre-sents bank supervision changes in home country for foreign banks,

comes from the database of BankScope, which is produced by theBureau Van Dijk Corporation. Macroeconomic variables for a coun-try come from World Development Indicators (WDI); and the infor-mation on supervision comes from databases compiled by theWorld Bank. Cross-country differences in country condition andnancial development come from Global Insight Inc. Country-leveldata on institution quality comes fromWorldwide Governance Indi-cators (WGI), which was developed and ranked by Kaufmann et al.(2007). It covers more than 50 countries and can be downloadedfor free from: http://www.govidicators.org.

Table 1 summarizes statistics of the variables used for all banks,both foreign and domestic banks. Note that the mean of return onassets is 0.8480.797%, while the mean of return on equity is9.9548.994%; therefore, it is apparent that foreign banks on aver-age are more protable than domestic banks. The net interest mar-gin of foreign banks is slightly lower than that of domestic banks,but the gap between them is smaller. Compared to domestic banks,foreign banks exhibit a lower ratio of operating cost (0.041), liquid-ity (42.217%), and logarithm of bank size (3.187). However, theyalso have higher ratio of interest payment (4.312) and total capitalratio (18.757%).

5. Empirical results

To describe the correlation between bank ownership and per-formance, we investigate whether banking market structure,

Dom

N Mean Std. dev. N Mean Std. dev.

53,854,154,1

54,053,754,053,954,454,453,854,252,748,053,7

52,554,047,547,5

54,6

824 S.-H. Chen, C.-C. Liao / Journal of Banking & Finance 35 (2011) 819839while DSUPHostj;t describes bank supervision changes in home countryfor foreign banks.

4.2. Data and descriptive statistics

We collect data frommany different sources in order to conductan empirical analysis of 70 countries over the period 19922006.The bank-level data on the nancial statement report mainly

Table 1Summary statistics by bank ownership.

Variables All banks

N Mean Std. dev.

Dependent variableNet interest margin (NIM) 61,720 3.645 3.696Return on assets (ROA) 62,217 0.804 3.115Return on equity (ROE) 62,181 9.118 27.507

Independent variablesBank characteristicsOperation cost 62,098 0.038 0.059Interest payments 61,669 4.003 5.349Liquidity ratio 62,146 52.198 23.723Opportunity cost 61,862 0.029 0.043Credit risk 62,605 0.002 0.046Log (bank size) 62,546 3.164 0.975Non-interest revenues 61,731 0.628 0.407Capital strength 62,362 12.853 35.641Off balance sheet activity 60,451 0.264 4.862Market shares 55,584 0.013 0.043Other operating income 61,668 1.796 5.323

Banking market structurePR H statistics 60,069 0.435 31.287Lerner index 62,111 0.356 1.467HHI (assets) 54,914 0.306 6.477CR4 (assets) 54,914 0.492 14.996

Macroeconomic conditionGrowth rate of GDP 62,739 2.559 2.346

Ination rate 62,585 7.470 81.726 54,5Real interest rate 55,611 6.816 7.380 48,433 3.680 3.478 7887 3.407 4.92863 0.797 3.063 8054 0.848 3.44633 8.994 25.765 8048 9.954 37.149

70 0.038 0.059 8028 0.041 0.06286 3.957 5.226 7883 4.312 6.11994 53.684 23.067 8052 42.217 25.58401 0.029 0.040 7961 0.032 0.05847 0.002 0.041 8158 0.005 0.07013 3.161 0.954 8133 3.187 1.10836 0.615 0.424 7895 0.715 0.24570 11.972 32.666 8092 18.757 50.92796 0.251 5.175 7655 0.349 1.40866 0.012 0.040 7518 0.025 0.06080 1.707 5.227 7888 2.405 5.900

47 0.420 19.438 7522 0.513 72.60880 0.285 1.3267 8031 0.259 1.56534 0.291 6.066 7380 0.382 7.79234 0.333 11.774 7380 0.151 2.792

59 2.471 2.218 8080 3.151 3.005developing economy, governance, and supervision for foreignbanks between host country and home country play a partial orjoint role in driving this relationship. We start by using NIM,ROE, and ROA to estimate a baseline model that enables us to com-pare how both foreign ownership and banking competition affectbank performance. Next, we test whether banking competition,governance, and supervision from host country and home countryinuence (either individually or jointly) the association betweenforeign ownership and protability.

estic banks Foreign banks38 6.911 79.397 8047 11.260 95.95337 6.768 6.859 7174 7.136 10.219

ons.

rship

anki5.1. The effects of banking market structure: Are foreign banks moreprotable than domestic banks?

Before investigating the effect of banking market structure onbank protability, we rst use Eq. (9) to perform individual countryregression in order to examine whether foreign banks are moreprotable than domestic banks. This approach enables us toexplore whether huge differences exist between countries withrespect to the performance of foreign banks. Our results are sum-marized in Table 2. Note that the table divides the countries inour sample into 4 groups. The rst one (upper left quadrant) con-sists of countries in which the correlation of foreign ownership tobank protability is both positive and statistically signicant. Inthese countries foreign banks are, on average, more protable thandomestic banks. The second group (upper right quadrant) consistsof countries in which the correlation of foreign ownership to bankprotability is positive, but not statistically signicant. The thirdgroup (lower left quadrant) consists of countries in which the cor-relation of foreign ownership to bank protability is both negativeand signicant. In other words, domestic banks in these countriestend to outperform foreign banks. The fourth group (lower rightquadrant) consists of countries in which the correlation of foreignownership to bank protability is statistically negative, but

Table 2Impact of foreign ownership on protability: results from individual country regressi

Criteria Signicant

Foreign banks perform better than domestic ones AustriaEgyptItalyPanamaSingaporeTunisia

Domestic banks perform better than foreign ones CroatiaLuxembourg

The table provides summarized results from examining the impact of foreign owne

S.-H. Chen, C.-C. Liao / Journal of Binsignicant.Table 2 indicates that all four cases occur in our group of 47

countries. Foreign banks perform better than domestic banks in12 countries: Austria, Algeria, Egypt, Indonesia, Italy, Kenya,Panama, Poland, Singapore, Slovenia, Tunisia, and the UnitedStates. They perform worse in 4 countries: Croatia, Hong Kong,Thailand, and Luxembourg. Interestingly, this nding differsgreatly from that of Claessens and Horen (2009). This may bedue to the data period covered or to the control variables usedfor empirical specication. The majority of countries do not exhibita signicant difference between domestic and foreign banks.Among this group, ownership has a positive sign in 14 countriesand a negative sign in 15. These results reinforce those of previousstudies: When looking at aggregate data, there is no straightfor-ward relationship between bank ownership and performance.

Our next step was to investigate which bank and countryfactors have a signicant impact on the relative performance offoreign banks while quantifying the inuence of banking competi-tion on bank performance. As shown in Table 3, our empiricalresults from the estimation of Eq. (10) indicate that in the contextof international investigation, foreign banks correlate both posi-tively and signicantly to higher protability relative to domesticbanks.It is interesting to note that differences in banking market struc-ture across countries play a signicant role in protability whenwe use PR H statistics, Lerner index, HHI, and CR4 to correspondto different independent variables such as NIM, ROA, and ROE.From the perspective of structural measures like PR H statistics,Table 3 shows that a signicant negative relationship exists be-tween bank NIM and degree of competition in the banking indus-try. This nding indicates that bank net interest margin is morelikely to be squeezed when a bank operates in a competitive bank-ing industry in countries with free market access to foreign banksor new domestic banks. This result is especially robust when usingthe Lerner index (another structural measure). In contrast to alter-native measures such as the HHI index, we also nd the same re-sult. This indicates that a banks interest margin correlates bothsignicantly and positively to a monopolistic banking marketstructure, which is consistent with the CR4 measure.

In regard to bank nancial characteristics and macroeconomicvariables, the results from both NIM and ROA as dependent vari-ables indicate that banks with better protability correlate signi-cantly and negatively with operating cost. When we use ROE as adependent variable, however, the correlation is insignicant. Thisis in line with Maudos and Fernndez de Guevara (2004), whouse NIM as a dependent variable, as well as with Pasiouras and

Insignicant

Algeria Australia AlgeriaIndonesia Bangladesh ColombiaKenya Switzerland DenmarkPoland Cyprus GreeceSlovenia France MacauUnited States Honduras Norway

Netherlands ReunionPeru ZimbabweVenezuela

Hong Kong Bulgaria BoliviaThailand Canada Hungary

Ireland MexicoSouth Africa PortugalSlovakia SpainCzech IndonesiaRepublic of Korea

on protability for each country in the sample based on regression model (9).

ng & Finance 35 (2011) 819839 825Kosmidou (2007), who use ROA as a dependent variable. Bankswith better protability are positively and signicantly associatedwith implicit payments in the NIM model. This indicates thatcharging for a banks services through lower remuneration of lia-bilities would lead to higher interest margins for banks, which isconsistent with previous studies by Saunders and Schumacher(2000) and Maudos and Fernndez de Guevara (2004).

Liquidity ratios correlate signicantly and positively with NIM,ROA and ROE, which indicates that an increase in bank liquidity ra-tio tends to enhance a banks protability. Furthermore, banks withbetter protability are positively and signicantly correlated totheir opportunity cost. This result, which is similar to ndings bySaunders and Schumacher (2000), suggests that the opportunitycosts of reserves suggests the average return on earning assetsforegone by holding deposits in cash. As opportunity costs increasethe cost of funds beyond the observed rate, banks gain net interestmargins by compensating for these costs.

The relationship between bank protability and off-balancesheet is signicantly negative in all estimation results from themodels, which means the more off-balance activities a bank under-takes, the more its prots will decrease. The relationship betweensize (SIZE) and bank protability is negative on the net interestmargin. The negative coefcient indicates that larger banks tend

Table 3The impacts of banking market structure on bank protability: alternative measures of banking competition.

Independent variables Dependent variables: bank protability

Net interest margin (NIM) Return on assets (ROA) Return on equity (ROE)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Intercept 7.18*** 6.37*** 7.63*** 7.77*** 4.69*** 3.85*** 4.81*** 4.97*** 12.92*** 12.60*** 12.90*** 19.73***

(72.04) (57.07) (68.31) (71.51) (58.17) (43.06) (55.96) (59.50) (12.85) (12.55) (10.77) (15.22)Foreign banks ownership 0.34*** 0.37*** 0.42*** 0.45*** 0.19*** 0.18** 0.22*** 0.26*** 1.15** 1.57*** 2.00*** 1.11**

(4.25) (3.65) (4.77) (5.11) (3.38) (2.51) (3.77) (4.52) (2.26) (3.10) (3.81) (2.12)

Measures of banking market structurePR H statistics 0.01*** 0.02*** 0.35***

(5.08) (10.59) (7.93)Lerner index 0.05*** 0.06*** 15.98***

(8.70) (11.18) (15.94)HHI (assets) 0.76*** 1.00*** 0.39*

(3.78) (6.07) (1.95)CR4 (assets) 14.99*** 7.48*** 0.25***

(10.75) (6.16) (6.91)

Bank nancial characteristicsOperation cost 3.71*** 1.17*** 2.86*** 2.88*** 0.97*** 0.44* 1.07*** 1.09*** 1.21 3.44 3.54 5.27

(13.43) (4.36) (9.22) (9.29) (3.83) (1.83) (3.98) (4.06) (0.25) (0.72) (0.74) (1.09)Log (bank size) 0.35*** 0.23*** 0.40*** 0.41*** 0.20*** 0.13*** 0.21*** 0.22*** 0.34* 0.13 0.17 0.53***

(15.19) (9.11) (15.93) (16.50) (11.22) (6.49) (11.76) (12.11) (1.88) (0.70) (0.95) (2.82)Market shares 2.79*** 0.11 2.57*** 3.62*** 2.62*** 1.12* 2.62*** 3.04*** 33.13*** 43.69*** 48.04*** 30.19***

(5.26) (0.14) (5.01) (7.13) (6.13) (1.74) (6.62) (7.67) (7.03) (10.10) (10.25) (6.24)Credit risk 1.01* 1.95*** 1.09* 1.08* 1.11*** 1.47*** 1.06** 1.11*** 4.38 3.44 3.19 2.79

(1.78) (2.85) (1.75) (1.72) (2.79) (3.03) (2.55) (2.65) (1.16) (0.89) (0.81) (0.71)Interest payments 0.21*** 0.16*** 0.20*** 0.20*** 0.66*** 0.64*** 0.65*** 0.65*** 1.51*** 1.56*** 1.58*** 1.65***

(45.98) (33.25) (41.49) (41.26) (160.28) (150.96) (156.13) (156.38) (21.00) (22.60) (22.76) (23.55)Liquidity ratio 0.00*** 0.01*** 0.00*** 0.00*** 0.00*** 0.01*** 0.00*** 0.00*** 0.06*** 0.05*** 0.06*** 0.07***

(8.06) (9.18) (4.34) (4.27) (8.52) (9.48) (6.56) (6.29) (9.07) (6.78) (8.21) (9.79)Opportunity cost 9.18*** 7.89*** 9.13*** 9.23*** 1.37*** 0.77** 1.30*** 1.38*** 20.60*** 27.98*** 28.16*** 23.73***

(30.13) (19.98) (27.99) (28.33) (5.05) (2.23) (4.75) (5.03) (5.04) (7.05) (7.03) (5.85)Capital strength 0.00*** 0.00*** 0.00*** 0.00*** 0.01*** 0.01*** 0.01*** 0.01*** 0.02*** 0.02*** 0.02*** 0.02***

(8.27) (5.92) (8.10) (8.15) (16.85) (16.85) (16.51) (16.66) (4.37) (4.54) (4.18) (4.36)Non-interest revenues 6.56*** 6.03*** 6.52*** 6.51*** 4.39*** 3.82*** 4.51*** 4.48*** 12.44*** 11.46*** 13.01*** 14.57***

(102.90) (82.41) (93.65) (93.58) (77.00) (59.63) (76.56) (76.26) (13.96) (12.93) (14.49) (16.10)Off balance sheet activity 0.02*** 0.01*** 0.01*** 0.01*** 0.02*** 0.01*** 0.01*** 0.01*** 0.16*** 0.11*** 0.12*** 0.12***

(14.14) (6.23) (5.96) (5.91) (10.32) (8.25) (8.36) (8.38) (5.75) (4.17) (4.36) (4.22)Other operating income 0.11*** 0.13*** 0.15*** 0.15*** 0.70*** 0.66*** 0.68*** 0.68*** 1.90*** 1.81*** 1.85*** 1.90***

(33.55) (32.81) (40.42) (40.24) (225.87) (193.24) (216.69) (217.05) (35.48) (35.08) (35.70) (36.33)Macroeconomic conations in host countryGrowth rate of GDP 0.04*** 0.05*** 0.07*** 0.07*** 0.02*** 0.01** 0.00 0.00 0.39*** 0.06 0.19*** 0.09

(9.86) (11.34) (17.46) (17.48) (5.64) (2.35) (0.61) (0.54) (6.23) (1.05) (3.12) (1.48)Real interest rate 0.09*** 0.10*** 0.07*** 0.07*** 0.06*** 0.07*** 0.05*** 0.05*** 0.00 0.06*** 0.05** 0.04**

(50.85) (51.74) (40.12) (39.61) (40.02) (45.97) (37.83) (37.35) (0.08) (3.04) (2.57) (2.14)Ination rate 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.05*** 0.04*** 0.05*** 0.04***

(47.32) (52.34) (50.28) (51.41) (27.61) (28.42) (32.27) (32.72) (12.15) (15.51) (16.34) (15.64)

Observations 47,033 38,761 47,864 47,865 47,063 38,791 47,894 47,895 47,036 47,865 47,865 47,866Number of banks 4667 3847 4731 4731 4671 3851 4735 4735 4671 4735 4735 4735Log-likelihood 89,285 72,003 97,297 97,248 84,575 67,186 89,749 89,750 221,124 225,999 226,122 226,105Wald v2 17,300 *** 12,789*** 15,362*** 15,464*** 40,559*** 30,170*** 38,000*** 38,000*** 1741*** 2126 1879 1,922

Numbers in parenthesis are t-values.* Statistical signicant at level of 10%.

** Statistical signicant at level of 5%.*** Statistical signicant at level of 1%.

826S.-H

.Chen,C.-C.Liao/Journal

ofBanking

&Finance

35(2011)

819839

Table 4The effects of parent banks protability on foreign banks performance: protability level.


Net interest margins (NIMs) Return on assets (ROA) Return on equity (ROE)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Intercept 12.22*** 13.22*** 13.10*** 14.16*** 8.01*** 7.08*** 8.26*** 9.24*** 47.40*** 47.00*** 48.31*** 51.42***

(135.53) (130.96) (129.23) (127.42) (100.00) (78.05) (97.21) (98.72) (37.45) (37.48) (38.01) (36.55)Foreign banks ownership 0.54*** 0.66*** 0.58*** 0.47*** 0.31*** 0.36*** 0.32*** 0.23*** 1.61*** 1.98*** 2.02*** 1.79***

(8.58) (9.86) (8.78) (7.07) (6.91) (6.28) (6.96) (5.07) (3.55) (4.38) (4.36) (3.84)


(3.56) (11.87) 0.01* (9.25)Lerner index 0.21*** (1.85) 11.34***

(4.02) (12.61)HHI (assets) 1.55*** 1.55*** 4.24**

(10.06) (11.73) (2.16)CR4 (assets) 0.05*** 0.05*** 0.14***

(18.05) (19.00) (4.23)

Parent banks performanceProtability level 2.40*** 2.63*** 2.64*** 2.58*** 1.54*** 1.43*** 1.62*** 1.56*** 11.45*** 11.28*** 11.49*** 11.30***

(132.01) (128.17) (128.74) (124.88) (90.29) (74.16) (89.43) (85.66) (37.66) (37.79) (38.30) (37.34)

Bank nancial characteristicsOperation cost 6.79*** 7.52*** 7.51*** 7.27*** 2.47*** 1.92*** 3.00*** 2.78*** 18.69*** 22.58*** 22.38*** 21.48***

(32.88) (31.02) (30.99) (30.00) (12.18) (9.97) (13.41) (12.41) (4.32) (5.16) (5.11) (4.90)Log (bank size) 5.41*** 5.95*** 5.97*** 5.75*** 3.51*** 3.19*** 3.68*** 3.49*** 25.16*** 24.88*** 25.24*** 24.67***

(128.36) (124.96) (125.55) (118.08) (89.12) (72.33) (88.50) (82.19) (36.28) (36.57) (36.92) (35.45)Market shares 1.26*** 1.75*** 2.03*** 0.26 1.27*** 0.92* 1.76*** 0.30 14.11*** 28.95*** 28.05*** 21.72***

(3.17) (4.67) (5.26) (0.66) (3.76) (1.83) (5.63) (0.92) (3.37) (7.55) (7.02) (5.10)Credit risk 0.27 0.20 0.05 0.06 0.71** 0.92** 0.52 0.56* 2.53 1.37 0.82 0.98


(74.36) (68.89) (68.87) (67.92) (169.80) (144.35) (159.61) (160.92) (37.51) (36.49) (36.98) (37.16)Liquidity Ratio 0.00*** 0.01*** 0.01*** 0.00*** 0.00*** 0.00*** 0.00*** 0.00** 0.00 0.01 0.00 0.01

(6.89) (11.02) (11.03) (8.38) (3.33) (3.46) (5.44) (2.38) (0.13) (1.44) (0.06) (0.91)Opportunity cost 8.54*** 8.94*** 8.85*** 8.48*** 0.13 0.12 0.16 0.22 6.33* 12.20*** 11.66*** 9.86***



(27.68) (20.93) (21.78) (25.12) (17.05) (12.96) (19.08) (22.56) (6.08) (5.21) (3.69) (2.74)Off Balance sheet activity 0.02*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.02 0.02 0.02 0.03


(53.09) (50.84) (50.89) (49.62) (239.22) (192.00) (226.68) (228.16) (52.76) (52.07) (52.95) (53.02)Macroeconomic conations in host countryGrowth rate of GDP 0.05*** 0.08*** 0.08*** 0.09*** 0.00 0.01*** 0.02*** 0.02*** 0.39*** 0.12** 0.21*** 0.16***

(17.38) (24.66) (24.12) (26.66) (0.84) (3.10) (5.19) (8.29) (6.88) (2.18) (3.88) (2.85)Real interest rate 0.04*** 0.03*** 0.03*** 0.03*** 0.02*** 0.03*** 0.02*** 0.02*** 0.01 0.04* 0.03* 0.03

(30.09) (19.82) (20.22) (19.19) (18.67) (22.69) (18.37) (17.32) (0.64) (1.95) (1.73) (1.55)Ination rate 0.01*** 0.01*** 0.01*** 0.01*** 0.00*** 0.00*** 0.00*** 0.00*** 0.04*** 0.04*** 0.04*** 0.04***

(53.33) (51.57) (51.99) (51.23) (24.85) (24.66) (28.08) (27.10) (11.00) (14.82) (15.43) (15.08)

Observations 46,615 47,424 47,424 47,425 46,631 38,512 47,440 47,441 46,609 47,416 47,416 47,417Number of banks 4662 4726 4726 4726 4664 3845 4728 4728 4664 4728 4728 4728Log-likelihood function 74,308 83,349 83,306 83,197 72,306 56,839 78,457 78,348 213,175 217,912 217,989 217,987Wald v2 35,667*** 32,779*** 32,870*** 33,091*** 40,395*** 28,160*** 38,218*** 38,438*** 3886*** 4238*** 4085*** 4098***

Numbers in parenthesis are t-values.* Statistical signicant at level of 10%.** Statistical signicant at level of 5%.*** Statistical signicant at level of 1%.

S.-H.Chen,C.-C.Liao

/Journalof

Banking&

Finance35

(2011)819

839827

Table 5The effects of parent banks protability on foreign banks performance: protability ranking.


Net interest margins (NIMs) Return on assets (ROA) Return on equity (ROE)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Intercept 6.34*** 5.41*** 6.74*** 9.44*** 4.56*** 3.56*** 4.64*** 6.50*** 12.18*** 9.25*** 12.60*** 23.44***

(57.79) (43.98) (54.27) (72.92) (52.49) (36.44) (48.63) (64.39) (11.35) (9.08) (10.95) (9.18)Foreign banks ownership 0.70*** 0.74*** 0.71*** 0.38*** 0.24*** 0.28*** 0.26*** 0.10* 1.33** 2.97*** 1.70*** 2.54**

(7.73) (6.63) (7.48) (4.22) (4.11) (3.72) (4.38) (1.72) (2.57) (5.77) (3.23) (2.40)


(4.92) (10.56) (7.91)Lerner index 0.05*** 0.06*** 15.27***

(9.10) (11.23) (15.28)HHI (assets) 1.54*** 1.22*** 11.64***

(7.43) (7.07) (3.47)CR4 (assets) 0.16*** 0.11*** 0.38***

(41.27) (33.00) (5.15)

Parent banks performanceProtability ranking 0.30*** 0.32*** 0.28*** 0.44*** 0.05*** 0.09*** 0.05*** 0.19*** 0.22** 0.01*** 0.26** 2.55***

(20.95) (20.41) (17.78) (29.33) (4.32) (7.54) (4.25) (16.14) (1.98) (19.13) (2.05) (7.17)

Bank nancial characteristicsOperation cost 3.96*** 1.48*** 3.11*** 2.78*** 0.97*** 0.49** 1.08*** 0.87*** 1.06 2.09 3.98 3.45

(14.53) (5.56) (10.08) (9.15) (3.80) (2.03) (3.98) (3.25) (0.22) (0.44) (0.82) (0.37)Log (bank size) 0.54*** 0.42*** 0.57*** 0.36*** 0.22*** 0.18*** 0.24*** 0.15*** 0.21 0.22 0.18 3.42***

(21.08) (15.17) (20.67) (13.62) (11.90) (8.46) (12.40) (8.06) (1.11) (1.24) (0.93) (6.41)Market shares 4.50*** 1.87** 4.25*** 1.15** 3.01*** 1.87*** 3.00*** 0.83** 35.78*** 54.69*** 40.47*** 54.92***

(8.19) (2.29) (8.04) (2.27) (6.84) (2.85) (7.37) (2.04) (7.30) (12.52) (8.57) (7.18)Credit risk 1.62*** 2.61*** 1.57** 1.53** 1.19*** 1.65*** 1.13*** 1.21*** 4.66 4.85 2.72 2.97


(45.60) (33.08) (41.32) (39.94) (159.99) (151.00) (155.77) (159.72) (20.63) (20.35) (22.40) (9.35)Liquidity ratio 0.00*** 0.00*** 0.00 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** 0.06*** 0.02*** 0.06*** 0.11***

(4.08) (4.77) (0.87) (4.37) (7.36) (7.33) (5.45) (8.06) (8.10) (2.84) (7.50) (6.98)Opportunity cost 9.26*** 7.90*** 9.25*** 8.24*** 1.40*** 0.76** 1.34*** 0.48* 21.09*** 28.14*** 28.16*** 24.74***



(96.70) (78.64) (88.42) (93.89) (74.61) (57.88) (74.45) (78.13) (13.37) (11.07) (14.84) (7.08)Off balance sheet activity 0.02*** 0.01*** 0.01*** 0.01*** 0.02*** 0.01*** 0.01*** 0.01*** 0.16*** 0.12*** 0.12*** 0.04


(32.41) (31.65) (39.79) (36.99) (225.81) (193.58) (216.65) (221.73) (35.23) (32.26) (35.23) (15.04)Macroeconomic conations in host countryGrowth rate of GDP 0.04*** 0.06*** 0.07*** 0.09*** 0.02*** 0.01** 0.00 0.01*** 0.42*** 0.18*** 0.22*** 0.07

(9.63) (11.90) (16.93) (21.66) (5.84) (2.32) (0.91) (3.35) (6.47) (2.85) (3.45) (0.61)Real interest rate 0.08*** 0.10*** 0.07*** 0.06*** 0.06*** 0.07*** 0.05*** 0.05*** 0.00 0.04** 0.06*** 0.10**

(49.14) (50.91) (39.20) (35.87) (39.60) (45.54) (37.60) (34.41) (0.00) (2.03) (2.70) (2.35)Ination rate 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.00*** 0.05*** 0.04*** 0.04*** 0.04***

(47.89) (53.00) (50.61) (49.30) (27.67) (28.53) (32.32) (30.71) (12.22) (15.85) (16.07) (9.14)

Observations 47,033 38,761 47,864 47,865 47,049 38,777 47,880 47,881 47,022 47,865 47,852 16,699Number of banks 4667 3847 4731 4731 4669 3849 4733 4733 4669 4735 4733 1990Log-likelihood function 89,050 71,786 97,131 96,331 84,545 67,137 89,719 89,207 221,063 225,813 226,064 85,559Wald v2 17,770 13,224 15,694 17,298 40,569 30,221 38,009 39,036 1744 2497 1885 592

Numbers in parenthesis are t-values.* Statistical signicant at level of 10%.** Statistical signicant at level of 5%.*** Statistical signicant at level of 1%.

828S.-H

.Chen,C.-C.Liao/Journal

ofBanking

&Finance

35(2011)

819839

Table 6The impacts of home-country characteristics on foreign bank protability.

Independent variables Macroeconomic environment Country risk

Developing country GDP Real interest rate Ination rate level Legal risk Economic risk

Foreign bank ownership (FBO) 0.90* 0.56*** 0.64*** 0.56*** 4.75** 6.91**

(1.92) (7.76) (8.90) (8.27) (2.07) (2.52)Parent banks protability ranking 2.73*** 2.86*** 2.74*** 2.85*** 2.54*** 2.54***

(45.86) (142.27) (132.25) (141.63) (23.04) (23.03)FBO home-country characteristics 0.35* 0.01 0.00 0.00 2.50** 4.05**

(1.85) (0.67) (0.95) (1.45) (2.29) (2.33)Control variables Yes Yes Yes Yes Yes Yes

Observations 6496 50,416 47,068 50,255 1995 1995Number of banks 606 4741 4696 4733 313 313Log-likelihood function 13,815 91,378 83,848 91,072 4,489 4,489Wald v2 4058 30,948 28,369 30,679 1597 1597

Governance Bank supervision

Control of corruption Regulatory quality Government effectiveness Bank capital Ownership restrictiveness Basel risk weights

Foreign bank ownership (FBO) 0.58*** 0.47*** 0.48*** 0.91*** 0.58*** 0.51***


(116.67) (116.71) (116.71) (96.47) (96.43) (96.39)FBO home-country characteristics 0.04 0.13** 0.09 0.39*** 0.10* 0.15


Observations 37,028 37,028 37,028 27,943 27,943 27,943Number of banks 4753 4753 4753 4622 4622 4622Log-likelihood function 66,805 66,803 66,804 51,426 51,436 51,437Wald v2 21,229 21,232 21,231 16,724 16,703 16,702

Bank supervision

Securities activities Certied auditor Auditors report Consolidated accounts Solvency intervention Onsite inspections

Foreign bank ownership (FBO) 0.67*** 0.52*** 0.54** 0.66*** 0.65*** 0.74***


(96.38) (96.39) (96.40) (97.10) (96.31) (96.28)FBO home-country characteristics 0.02 0.12* 0.12 0.03 0.02 0.02***



Banking market structure

PR H-statistics Lerner index HHI (assets) CR4 (assets)

Foreign bank ownership (FBO) 0.5436*** 0.61*** 0.38*** 1.22***

(8.57) (8.90) (4.03) (7.98)Parent banks protability ranking 2.3968*** 2.63*** 2.86*** 2.63***

(132.12) (128.18) (142.07) (128.13)FBO home-country characteristics 0.004** 0.53*** 1.09*** 0.03***

(continued on next page)

S.-H.Chen,C.-C.Liao

/Journalof

Banking&

Finance35

(2011)819

839829

Table6(con

tinu

ed)

Ban

kingmarketstructure

PRH-statistics

Lerner

inde

xHHI(assets)

CR4(assets)

(2.09

)(4.66)

(3.03)

(4.06)

Con

trol

variab

les

Yes

Yes

Yes

Yes

Observation

s46

,615

47,424

50,839

47,425

Numbe

rof

banks

4662

4726

4778

4726

Log-likelihoo

dfunction

74,31

28

3,34

69

2,51

38

3,35

0Waldv2

35,659

32,785

31,203

32,787

Dep

ende

ntvariab

leisnet

interest

margin(N

IM).Numbe

rsin

parenthesisaret-values.

Con

trol

variab

lesforem

piricale

stim

ationinclude

intercep

t,op

erationcost,ban

ksize,m

arketshare,cred

itrisk,interest

paym

ents,liquidity,op

portunitycost,cap

italstrength,n

on-interestrevenues,off-balan

cesheet,an

dother

operatingincome.

*Statisticalsign

icantat

levelof

10%.

**Statisticalsign

icantat

levelof

5%.

***Statisticalsign

icantat

levelof

1%.

830 S.-H. Chen, C.-C. Liao / Journal of Bankito earn lower prots, while smaller banks tend to earn higher prof-its. In other words, smaller banks experience economies of scaleand scope, while larger banks experience diseconomies of scaleand scope. These ndings are consistent with studies conductedby Kosmidou et al. (2007) and Pasiouras and Kosmidou (2007).

Another nding in our study is that non-interest revenues aresignicantly negative. This means that operation income fromnon-interest operating will decrease income, and other operationactivities will decrease, too. Banks with higher protability arepositively and signicantly associated with the ratio of marketshare. More importantly, based on the results of the NIM andROA models, the relationship between bank protability and capi-tal ratio is signicantly positive. However, in the ROE model therelationship between bank protability and capital ratio is bothnegative and signicant.

In Eq. (10), we analyze parent bank performance in home coun-try in order to determine whether foreign banks are more prot-able than domestic banks. Table 4 presents the results of usingthe protability level of parent bank in home country. In columns(1)(4), the dependent variable is NIM. The coefcient on the prof-itability level of parent bank is 2.40, which is at a signicant levelof 1% when using PR H statistics. This implies that when a parentbank has high levels of protability at home, its foreign iterationswill, on average, perform better than domestic banks in the hostcountry. Alternative measures such as the Lerner index, HHI, andCR4 are signicant, which is similar to the ndings in Table 3.When using ROA and ROE as dependent variables, the coefcientsof protability level of parent bank maintain an economic signi-cance at the 1% level.

However, protability level of parent bank in home countrydoes not reect the relative performance of their counterparts inthe home country. Hence, we apply another comparative indicatoras protability ranking in home country. To do so, we dene theparent banks protability ranking as: the higher the protabilityin comparison to its counterpart in its home country, the higherthe ranking for parent banks. Table 5 presents the results of ourregression using protability ranking for parent banks in homecountry to explain their comparative performance. Parent bankswith higher protability rankings at home tend to experience high-er protability rankings in their foreign banks than do domesticbanks in the host country. There was almost no change in our esti-mates when we used alternative measures of banking marketstructure. Virtually all coefcients remain signicant at the 1%and 5% level; however, the estimates are much smaller than thosein Table 4.

5.2. Home-country impacts of macroeconomic condition, governance,and supervision on bank protability

After controlling for bank nancial characteristics and focusingon the results from NIM as a dependent variable, we next examinewhether home-country banking market structure, macroeconomicconditions, governance and supervision of foreign banks have anyspillover effects on NIM in the host country. To do so, we use inter-action terms for a foreign banks ownership with home-countrycharacteristics, including: developing country, macroeconomicconditions, country risk, governance, and bank supervision.

Table 6 presents the results of several robustness specications.Because some country indexes are highly correlated with eachother, we include the indexes individually in the models. Each col-umn of each panel reports coefcients from a single regression inwhich the dependent variable is NIM. The control variables arethe same as in Table 3; however, to conserve space we only report

ng & Finance 35 (2011) 819839select coefcients. The empirical results are consistent with ourprevious ndings. All major explanatory and control variablesmaintain their sign and signicance as before. Overall, we conrm

Table 7The Impacts of Host-Country Characteristics on Foreign Bank protability.

Independent variables Macroeconomic environment Country risk

Developing country GDP Real interest rate Ination rate level Legal risk Economic risk



(141.67) (141.56) (134.41) (141.49) (99.28) (99.01)FBO host-country characteristics 0.31** 0.11*** 0.02*** 0.01*** 0.04 0.99***



Governance Bank supervision

Control of corruption Regulatory quality Government effectiveness Bank capital Ownership restrictiveness Basel risk weights



(115.92) (116.10) (115.73) (93.80) (93.76) (93.79)FBO host-country characteristics 0.42*** 0.49*** 0.57*** 0.07 0.03 3.61***



Bank Supervision

Securities activities Certied auditor Auditors report Consolidated accounts Solvency intervention Onsite inspections



(93.72) (93.76) (93.80) (94.36) (93.57) (93.60)FBO host-country characteristics 0.07 0.16 0.03 0.23* 0.07 0.06



Banking market structure

PR H-statistics Lerner index HHI (assets) CR4 (assets)

Foreign bank ownership (FBO) 0.54*** 0.61*** 0.38*** 1.22***

(8.57) (8.90) (4.03) (7.98)Parent banks protability ranking 2.40*** 2.63*** 2.86*** 2.63***

(132.12) (128.18) (142.07) (128.13)FBO host-country characteristics 0.00** 0.53*** 1.09*** 0.03***

(continued on next page)

S.-H.Chen,C.-C.Liao

/Journalof

Banking&

Finance35

(2011)819

839831

Table7(con

tinu

ed)

Ban

kingmarketstructure

PRH-statistics

Lerner

inde

xHHI(assets)

CR4(assets)

(2.09

)(4.66)

(3.03)

(4.06)

Con

trol

variab

les

Yes

Yes

Yes

Yes

Observation

s46

,615

47,424

50,839

47,425

Numbe

rof

banks

4662

4726

4778

4726

Log-likelihoo

dfunction

74,31

28

3,34

69

2,51

38

3,35

0Waldv2

35,659

32,785

31,203

32,787

Dep

ende

ntvariab

leisnet

interest

margin(N

IM).Numbe

rsin

parenthesisaret-values.

Con

trol

variab

lesforem

piricale

stim

ationinclude

intercep

t,op

erationcost,b

anksize,m

arketshare,cred

itrisk,interest

paym

ents,liquidity,op

portunitycost,cap

ital

strength,n

on-interest

revenues,off-balan

cesheet,an

dother

operatingincome.

*Statisticalsign

icantat

levelof

10%.

**Statisticalsign

icantat

levelof

5%.

***Statisticalsign

icantat

levelof

1%.

832 S.-H. Chen, C.-C. Liao / Journal of Bankithat foreign banks tend to be more protable than domestic banksand that foreign bank whose parent bank has high levels of prot-ability tend to perform well in the host country.

Our results show that foreign banks from developing economiestend to have lower NIM in the host country. This indicates thatsuch banks do not compete well with domestic banks in regardto loan interest rate. In regard to macroeconomic factors, we ndno evidence that the growth rate of GDP, real interest rate, or ina-tion rate level experienced in the banks home country have anysignicant spillover effects on its foreign banks NIM.

We also nd that legal and economic risks in the home countrydo have economic and signicantly positive effects on their foreignbanks NIMs in the host country. This suggests that when a parentbanks legal and economic risks are high in its home country, it pre-fers to transfer its prots to its bank in other countries. When weanalyze the home-country effect of governance on foreign bankprotability, however, we nd a signicantly negative correlationbetween NIM and regulatory quality within the home country forforeign banks. No signicant effect exists regarding control of cor-ruption and effectiveness of government.

When we look at the home-country effect of banking supervi-sion on a foreign banks NIM, we nd that foreign banks whosehome country has strong restrictions regarding bank ownershipand requirements for licensed or certied auditors appear to havea higher NIM than domestic banks. This suggests that the betterthe quality of nancial regulation and supervision a home countryimposes on a foreign banks operations, the more likely the foreignbanks are to be more protable than domestic banks. Surprisingly,we nd that the interaction terms among foreign ownership, bankcapital, and onsite inspections in a home country have a signi-cantly negative coefcient. This suggests that foreign banks whosehome country implements more restrictive control on a parentbanks capital and onsite examination is harmful to a foreignbanks NIM in the host country.

When we look at the home-country effects of banking competi-tion on foreign bank protability, we nd that home-country bankcompetition, proxied as PR H statistics, Lerner index, HHI (assets),and CR4 (assets), has a signicant effect on a foreign banks NIM.This means that foreign banks are more likely to use their NIMwhen the banking market structure in their home country iscompetitive.

5.3. The host-country effects of macroeconomic conditions,governance, and supervision on bank protability

Similar to the specications in Table 6, we include the countryindicators individually in our empirical models because some in-dexes are highly associated with each other. Each column of eachpanel reports coefcients from a single regression in which thedependent variable is NIM. As shown in Table 7, we again nd thatforeign banks, on average, are more protable than domesticbanks. We also nd that foreign banks whose parent bank has highprotability rankings tend to perform better in the host country.

Table 7 shows the impact of host-country characteristics, suchas developing economy, macroeconomic environment, countryrisk, governance, banking regulation, and bank supervision, onbank NIM. Our results nd that foreign-owned banks operatingin developing host countries tend to be more protable thandomestic banks. Coefcients of both GDP and real interest ratesare signicantly negative, suggesting that foreign banks in a hostcountry with higher GDP and real interest rates tend to have lessNIM when compared to domestic banks. In addition, the regressioncoefcient on ination rate is positive and signicant. As a result, a

ng & Finance 35 (2011) 819839host country with high ination levels would largely magnify for-eign bank margins. When foreign banks operate in a host countrywith high levels of economic risk, it tends to harm their NIM. This

et in

ankiTable 8Joint effects of host- and home-country characteristics on foreign bank protability.

Dependent variables Dependent variable: n

PR H statistics

Foreign bank ownership (FBO) 3.96*** (4.33)Parent bank protability ranking in home country 1.80*** (23.34)

Home country characteristicsMacroeconomic conditionFBO foreign bank from developing country 0.12 (0.43)FBO legal risk 3.96 (1.51)FBO economic risk 4.36* (1.71)GovernanceFBO regulatory quality 0.00 (0.04)Bank supervisionFBO bank capital requirements 0.38* (1.90)FBO bank ownership restrictiveness 0.07 (0.93)FBO certied auditor 0.11 (0.76)FBO onsite inspections 0.01 (1.28)

S.-H. Chen, C.-C. Liao / Journal of Bsuggests a positive relationship between risk and return. However,this risk factor causes the opposite nding: Foreign banks operat-ing in countries with high levels of economic risk are less protablethan domestic banks.

When we focus on governance within the host country, we ndsome interesting results. Estimated coefcients on all governanceare negative and positive. This suggests that foreign-owned banksoperating in host countries with stricter controls over corruption,regulatory quality, and government effectiveness tend to degener-ate their margin. Nevertheless, foreign banks steadily demonstratebetter protability than domestic banks in all cases.

The results of bank supervision indicate that Basel risk weightsand consolidated accounts are signicantly and positively relatedto foreign bank NIM. However, they provide less support for ourexpectation that foreign banks in a host country with better banksupervision will exhibit greater margins than domestic banks.Again, we come back to the conclusion that foreign banks overallare more protable than domestic ones and that the protabilityof the parent strongly affects the performance of its foreign banks.Specically, foreign banks operating in a host country with less

Banking market structureFBO bank competition 1.94*** (3.20)

Host country characteristicsMacroeconomic conditionFBO foreign bank operating in developing country 0.47 (1.58)FBO GDP 0.06*** (4.78)FBO interest rate 0.04*** (6.35)FBO ination rate 0.09*** (11.25)FBO economic risk 0.23* (1.93)GovernanceFBO political instability 0.18 (1.40)FBO regulatory quality 0.03 (0.18)FBO government effectiveness 0.15 (0.75)Bank supervisionFBO basel risk weights 4.18*** (6.06)FBO consolidated accounts 0.09 (0.56)Banking market structureFBO bank competition 2.10*** (3.49)Observations 2197Number of banks 499Log-likelihood function 3947Wald v2 3871***

Numbers in parenthesis are t-values.We estimate the regression model specied as NIMijt = a0 + a1 FBOijt + a2 parentcountry characteristicsijt + a5 Rcontrolijt+eijt. Control variables include some bank naninterest payments, liquidity ratio, opportunity cost, capital strength, non-interest revenu* Statistical signicant at level of 10%.** Statistical signicant at level of 5%.*** Statistical signicant at level of 1%.terest margin (NIM)

Lerner index HHI (assets) CR4 (assets)

4.66*** (4.65) 8.57*** (8.60) 8.19*** (7.70)1.72*** (20.77) 2.03*** (24.73) 2.01*** (24.34)

0.30 (1.06) 0.30 (1.20) 0.29 (1.16)3.20 (1.16) 2.66 (1.06) 2.62 (1.03)3.76 (1.40) 3.20 (1.30) 3.29 (1.33)

0.15 (1.09) 0.19 (1.55) 0.19 (1.49)

0.41* (1.83) 0.31 (1.41) 0.42** (1.96)0.10 (1.13) 0.07 (0.88) 0.08 (0.96)0.16 (1.00) 0.16 (1.01) 0.17 (1.11)0.01 (0.73) 0.00 (0.11) 0.01 (0.53)

ng & Finance 35 (2011) 819839 833competition in the local banking industry tend to increase theirprotability. Note again that this result is extremely robust in allcases, regardless of which measure is used to gauge banking com-petition (i.e., PR H statistics, Lerner index, HHI, or CR4).

5.4. The joint effects of home- and host-country characteristics onbank protability

To further assess the joint effects of home- and host-countrycharacteristics and bank protability with respect to macroeco-nomic condition, governance, and supervision, we estimate ourregression models using signicant coefcients in both home andhost country estimation, individually. Table 8 shows the resultsof regression based on Eq. (11). Once again, the results conrm thatforeign banks are more protable, on average, than domestic banksand that the protability of the parent strongly affects the perfor-mance of its foreign iterations.

When analyzing home-country macroeconomic characteristicsusing PR H statistics as banking market structure, we nd thatthe regression coefcient on economic risk is signicant but

0.48*** (3.34) 6.37*** (14.45) 0.84*** (13.39)

0.41 (1.35) 0.07 (0.26) 0.05 (0.18)0.03** (2.29) 0.02* (1.70) 0.02* (1.76)0.05*** (7.98) 0.04*** (7.36) 0.04*** (7.44)0.12*** (47.73) 0.12*** (47.95) 0.12*** (47.82)0.09 (0.71) 0.17 (1.54) 0.17 (1.48)

0.03 (0.23) 0.17 (1.31) 0.14 (1.07)0.06 (0.30) 0.12 (0.60) 0.07 (0.38)0.15 (0.67) 0.17 (0.80) 0.16 (0.77)

3.72*** (4.69) 3.28*** (4.25) 3.32*** (4.27)0.30* (1.69) 0.25 (1.43) 0.25 (1.44)

3.67*** (3.39) 47.19*** (14.75) 0.11*** (13.84)

2380 2379 2380524 524 5244579 4488 45027160*** 8280*** 8140***

protibilityimt + a3 RFBOijt host country characteristicsijt + a4 RFBOijt homecial characteristics such as operation cost, log (bank size), market shares, credit risk,es, off-balance sheet activity, and other operating income.

tabi

et in

ankiTable 9Joint effects of supervision changing in host- and home-country on foreign bank pro

Dependent variables Dependent variable: n

PR H statistics

Foreign bank ownership (FBO) 5.84*** (12.23)Parent bank protability ranking in home country 0.66*** (7.90)

Home country characteristicsMacroeconomic conditionFBO foreign bank from developing country 0.10 (0.42)FBO legal risk 2.75 (1.35)FBO economic risk 2.38 (1.22)GovernanceFBO regulatory quality 0.04 (0.39)Bank supervisionFBO D bank capital requirements 0.62 (1.31)FBO D bank ownership restrictiveness 0.49*** (2.74)

*

834 S.-H. Chen, C.-C. Liao / Journal of Bnegative. This estimated coefcient is opposite to that of individualestimation. Also noted that the result is not as robust as when weuse different measures of banking market structure. The possiblereason for this phenomenon lies in the potentially higher correla-tion between country risk and banking competition. When a par-ent bank is subject to stringent bank capital requirements, it mayharm the protability of its foreign bank in the host country. Whenwe use the remaining measures for estimation, we nd that thecoefcient on bank capital requirement is both signicant and neg-ative. Clearly, the home-country impact of banking market struc-ture on foreign bank protability is quantitatively signicant.This suggests that banking competition in the home country is lesskeen when foreign banks operate at higher levels of protability inthe host country.

When we consider the effects of host-country characteristics onforeign bank protability, we nd that some parts of the estimatedresult are different from the individual estimations discussed ear-lier. For example, we observe that foreign banks operating in hostcountries that have higher real interest and ination rates display

FBO D certied auditor 0.98 (1.72)FBO D onsite inspections 0.03 (1.29)Degree of bank competitionFBO banking market structure 2.37*** (3.04)

Host country characteristicsMacroeconomic conditionFBO foreign bank operating in developing country 0.59** (2.17)FBO GDP 0.00 (0.14)FBO interest rate 0.05*** (7.90)FBO ination rate 0.03*** (4.35)FBO economic risk 0.27*** (6.69)GovernanceFBO political instability 0.20 (1.51)FBO regulatory quality 0.46*** (3.29)FBO government effectiveness 0.64*** (4.03)Bank supervisionFBO D basel risk weights 4.20*** (4.00)FBO D consolidated accounts 0.40 (1.45)Degree of bank competitionFBO banking market structure 2.47*** (3.18)Observations 3718Number of banks 556Log-likelihood function 7794Wald v2 2469***

Numbers in parenthesis are t-values.We estimate the following regression model specied as NIMijt = b0 + b1

try)jt + b4 RFBOijt D supervision (host country)jt + b5 RFBOijt host country charadenotes the operator of bank supervision changing between last and current calendar ycost, log (bank size), market shares, credit risk, interest payments, liquidity ratio, opporother operating income.* Statistical signicant at level of 10%.** Statistical signicant at level of 5%.*** Statistical signicant at level of 1%.lity.

terest margin (NIM)

Lerner index HHI (assets) CR4 (assets)

8.43*** (15.54) 10.70*** (17.00) 10.58*** (15.11)1.02*** (10.78) 1.26*** (12.78) 1.24*** (12.62)

0.29 (1.05) 0.25 (0.94) 0.24 (0.92)4.49* (1.76) 4.20* (1.66) 4.21* (1.66)4.03 (1.64) 3.82 (1.56) 3.83 (1.56)

0.44*** (3.24) 0.48*** (3.55) 0.47*** (3.52)

0.22 (0.36) 0.20 (0.32) 0.21 (0.34)0.62*** (2.74) 0.56** (2.48) 0.56** (2.46)

ng & Finance 35 (2011) 819839notably higher margins. While a host developing country plays norole in explaining foreign bank margins, the growth rate of GDP inthe host country seems to matter to foreign bank NIM. Indeed,incorporating foreign banks in countries with lower growth ratesof GDP benets foreign bank margins. In addition, bank supervisionon Basel risk weight in the host country appears to have a largelypositive effect on foreign bank NIM.

Interestingly, when we use PR H statistics as the measure ofbanking competition, the estimated coefcient on economic riskis signicant and positive. These results, which are opposite fromthe ones shown in Table 7, demonstrate the relationship betweenrisk and return. In other words, when foreign banks operate in ahost country with higher economic risks, they are more likely to in-crease their protability than domestic banks. Furthermore, whenforeign banks operate in a host country that implements compli-ance of consolidated accounts, they tend to be less protable thandomestic banks. Although this is statistically signicant only whenwe use the Lerner index as a measure of banking market structure,when we combine home and host country factors, the results show

0.93 (1.27) 0.88 (1.21) 0.89 (1.22)0.01 (0.34) 0.01 (0.23) 0.01 (0.23)

0.62*** (2.78) 4.47*** (8.45) 0.62*** (8.20)

0.22 (0.68) 0.08 (0.26) 0.05 (0.13)0.05*** (3.35) 0.05*** (3.53) 0.05*** (3.52)0.00 (0.52) 0.00 (0.22) 0.00 (0.28)0.00*** (2.64) 0.00** (2.08) 0.00** (2.12)0.36*** (7.47) 0.35*** (7.42) 0.36*** (7.41)

0.20 (1.29) 0.18 (1.19) 0.18 (1.19)0.49*** (2.93) 0.41** (2.42) 0.42** (2.49)0.87*** (4.57) 0.85*** (4.49) 0.84*** (4.44)

4.56*** (3.40) 4.50*** (3.36) 4.49*** (3.34)0.37 (1.09) 0.37 (1.08) 0.36 (1.07)

5.47*** (3.23) 33.53*** (8.87) 0.08*** (8.46)

3956 3956 3956581 581 5819251 9224 92272313*** 2525*** 2512***

FBOijt + b2 parent protibilityimt + b3 RFBOijt D supervision (home coun-cteristicsijt + b6 RFBOijt home country characteristicsjt + b7 Rcontrolijt + eijt. Dear. Control variables include some bank nancial characteristics such as operationtunity cost, capital strength, non-interest revenues, off-balance sheet activity, and

Table A1Empirical results from previous studies.

Explanatory variable Positive relation with Negative relation with No relation with

Default risk Angbazo (1997)N Athanasoglou et al. (2008)A Dietrich and Wanzenried (2009)A

Claeys and Vennet (2008)N

Maudos and Fernndez de Guevara (2004)N

Interest rate risk Claessens et al. (2001)N Angbazo (1997)N

Claeys and Vennet (2008)N

Maudos and Fernndez de Guevara (2004)N

Liquidity risk Pasiouras and Kosmidou (2007)A Angbazo (1997)N Kosmidou et al. (2007)A

Capital adequacy Angbazo (1997)N Claessens et al. (2001)N

Athanasoglou et al. (2008)A, Claeys andVennet (2008)N

Kosmidou et al. (2007)A

Pasiouras and Kosmidou (2007)A Williams (2003)A

Saunders and Schumacher (2000)N

Dietrich and Wanzenried (2009)A

Implicit interest payments Maudos and Fernndez de Guevara (2004)N Angbazo (1997)N

Saunders and Schumacher (2000)N

Opportunity cost of non-interest bearingreserves

Angbazo (1997)N

Opportunity cost of reserves Saunders and Schumacher (2000)N Maudos and Fernndez de Guevara(2004)N

Management quality Angbazo (1997)N Maudos and Fernndez de Guevara(2004)N

Differences in barriers to statewidebranching

Angbazo (1997)N

Productivity growth Athanasoglou et al. (2008)A

Claessens et al. (2001)N

Operating expenses management Athanasoglou et al. (2008)A Williams (2003)A

Size Bonin et al. (2005)A Kosmidou

Date post:	08-Oct-2015
Category:	Documents
Upload:	bambangsutrisno89
View:	16 times
Download:	0 times

Foreign Bank and Domestic

Documents