refreree report

Journal of Banking & Finance 34 (2010) 765–775

Contents lists available at ScienceDirect

Journal of Banking & Finance

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j b f

On the implications of market power in banking: Evidence from developing countries

Rima Turk Ariss *

Department of Economics and Finance, Lebanese American University, 13-5053, Chouran Beirut 1102 2801, Lebanon

a r t i c l e i n f o a b s t r a c t

Article history: Received 5 February 2009 Accepted 3 September 2009 Available online 6 September 2009

JEL classification: D4 G15 G21 L11 N20

Keywords: Bank efficiency Financial stability Lerner Market power

0378-4266/$ - see front matter � 2009 Elsevier B.V. A doi:10.1016/j.jbankfin.2009.09.004

* Tel.: +961 1 786456x1644; fax: +961 1 867098. E-mail address: [email protected]

1 While encouraging competition is important for c and Zingales, 2003), theory offers opposing arguments beneficial for economic activity (Gorton and Winton, 2

2 We thank an anonymous referee for pointing competition” hypothesis as an alternative to the ‘‘quest in driving banks into foreign markets.

This paper investigates how different degrees of market power affect bank efficiency and stability in the context of developing economies. It sheds light on the competition-stability nexus by documenting and analyzing the complex interactions between a tripod of variables that are central for regulators: the degree of market power, bank cost and profit efficiency, and overall firm stability. The results show that an increase in the degree of market power leads to greater bank stability and enhanced profit efficiency, despite significant cost efficiency losses. The findings lend empirical justification to the traditional view that increased competition may undermine bank stability, and may bear significant implications for stressed banking systems in developing economies.

� 2009 Elsevier B.V. All rights reserved.

1. Introduction

Over the past two decades, policymakers in various parts of the world have taken steps to liberalize financial markets, promoting foreign competition and deregulating interest rates.1 Heightened competitive pressures in banking encourage financial institutions to enter new markets where competition is low, or where efficiency gains may be materialized.2 Evidence on efficiency gains from enter- ing new markets is, however, mixed. Sengupta (2007) explores the impact of competition on firms’ access to credit by viewing bank competition as competition between different asymmetrically in- formed principals. He develops a model to explain the perceived bias (which is stronger in developing countries) of foreign and large domestic banks in lending to large businesses and neglecting small firms, while better informed local banks continue to find a market among small enterprises. Lensink et al. (2008) report that foreign

ll rights reserved.

redit market efficiency (Rajan as to whether competition is 003).

attention to the ‘‘dodging for market power” conjecture

ownership negatively impacts bank efficiency. They also provide evi- dence to suggest that the relative efficiency of domestic vs. foreign banks is dependent on host and home country conditions.

The recent global dimension of banking is modifying the pre- vailing structures, and may bear significant implications on the efficiency levels of the industry. While the literature on banking efficiency is vast, few studies have investigated the impact of the prevailing market structure on the efficiency in the delivery of financial services but only in the context of developed countries (Maudos and De Guevara, 2007; Koetter et al., 2008; Schaeck and Cihak, 2008; Delis and Tsionas, 2009). Other research has taken the opposite stand to consider bank efficiency as a determinant of the degree of market power (Casu and Girardone, 2006; De Guevara and Maudos, 2007).

In parallel, as banks expand the scope of their activities and identify new growth opportunities across national borders, they tend to gain market power, raising concerns among regulators about issues of moral hazard and excessive risk-taking.3 An intense merger activity is continuously taking place globally,

3 Berger et al. (2003) identify two dimensions of globalization, bank national and bank reach, to find that multinational firms rely on host banks with limited reach to deliver ‘‘concierge” service and expand further from their home nation. Their results suggest that the extent of globalization may remain limited.

766 R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775

notwithstanding little empirical evidence on economies of scale and scope, and the social costs associated with monopoly rents (De Nicoló, 2000).4 A current debate prevails on the implications of a higher degree of market power on bank stability. A voluminous body of literature has emerged to address the competition-stability nexus in banking, though no consensus has been reached yet. The seminal article by Keeley (1990) has shown that increased compe- tition and deregulation erode franchise value and increase the probability of bank failure. Recent research, however, demonstrates that less competitive markets are less stable (Boyd and De Nicoló, 2005).

This paper investigates the implications of market power on is- sues of bank efficiency and stability in developing countries where capital markets are relatively underdeveloped, and banks repre- sent the main providers of credit to the economy. Developing countries provide a fertile laboratory to examine issues of compe- tition because they are engaged in a process of deregulation, bank privatization and financial liberalization, while the industry is wit- nessing more consolidation. Changing banking structures, in turn, raise concerns about competitive conditions, the efficiency in the delivery of financial services, and overall bank stability.5 These is- sues are of particular importance in light of the adverse implica- tions of the recent financial crisis for developing countries (International Monetary Fund, 2009).

Related research which examines simultaneously the inter-relat- edness between bank competition, efficiency, and stability is by Schaeck and Cihak (2008) on European and US banking. The authors use a traditional Lerner index to establish that competition increases bank efficiency. They also estimate the Boone indicator (a country- level measure of the intensity of competition based on the idea that more efficient firms will gain market share in a competitive environ- ment) to show that competition increases bank soundness.

This study differs from previous work in terms of sample cover- age and methodology. First, no prior research has to our knowledge addressed the complex interaction between competition, efficiency and stability in the context of developing countries. Second, we investigate the inter-relatedness between key variables of interest – market power, efficiency and stability – at the bank level to lend more support to the analysis. Specifically, the Lerner index is a bank-level measure of the degree of competition, which is pre- ferred over nation-wide proxies such as traditional concentration ratios or the Panzar and Rosse (PR, 1987) H-statistic. Third, since no consensus prevails in the literature regarding how best to assess the degree of market power in banking (Carbó et al., 2009), we con- sider three different specifications of the Lerner index. In addition to the traditional price mark-up over marginal cost estimation (Berger et al., 2009), we employ a structural model to derive two other adjusted Lerner indices: an efficiency-adjusted Lerner and a funding-adjusted Lerner (Maudos and De Guevara, 2007; Koetter et al., 2008). The intuition is that both bank stability and efficiency may affect the degree of market power, resulting in an endogeneity bias in the traditional Lerner estimation. Thus, the three different Lerner specifications are likely to provide more robustness to the analysis. In addition, we run a series of sensitivity checks using other proxies of bank stability and by implementing alternative estimation procedures (cross-section vs. panel data, Tobit vs. logis- tic regressions).

4 Regulators play an important role in approving or obstructing mergers, taking prompt corrective action, chartering de novo bank entry (DeYoung, 2003), and maybe ensuring stability in developing countries.

5 As suggested by an anonymous referee, we acknowledge that banking sectors in developing countries are subject to random macro shocks (such as exchange rate fluctuations) and to systematic shocks (such as state-wide regulation on foreign ownership). It is however, very hard to totally disentangle these effects from the tripod analysis undertaken herein.

The findings indicate that banks with more market power en- dure significant cost efficiency losses, but they manage to improve their profit efficiency levels. As banks implement growth strategies (hoping to increase market power or to dodge competition), ensu- ing gains from revenue diversification outweigh their deteriorating cost efficiency. In parallel, banks with more market power achieve higher records of overall stability. The results lend empirical justi- fication across developing countries to the traditional view that in- creased competition may undermine bank stability. They also bear significant implications for stressed banking systems.

The rest of the paper is organized as follows: Section 2 reviews the relevant literature. It first presents the main arguments in favor of more competition in banking and the expected efficiency gains, followed by a discussion on the opposing views for the implications of market power on bank stability. Section 3 introduces the estima- tion procedure of each of market power, bank efficiency, and bank stability. Section 4 describes the data and methodology. Section 5 presents the results of the tripod estimation (market power, bank efficiency, and bank stability) and analyzes the empirical findings on the implications of market power in banking. Section 6 performs a series of sensitivity checks for robustness and Section 7 concludes.

2. Literature review

2.1. Market power and bank efficiency

The arguments in favor of greater competition, in principle, ap- ply to all industries and derive from applying classical industrial organization economics. Berger and Hannan (1998) argue that banks not exposed to competition tend to be less efficient than banks subject to more competition. When market power prevails, managers may pursue objectives other than profit maximization, and they do not have incentives to work hard to keep costs under control, thereby reducing cost efficiency. The authors find evidence that a ‘‘quiet life” effect prevails in US banking.

Compared to the voluminous body of literature on bank effi- ciency, research on the relationship between market structure and bank efficiency is limited for developed markets and practi- cally non-existent for developing countries.6 Casu and Girardone (2006) derive bank efficiency estimates using the non-parametric Data Envelopment Analysis methodology and include it as an exog- enous variable in estimating the PR H-statistic. Using a sample of banks in the European Union, they do not find a clear relationship between efficiency and competition. However, the PR H-statistic is contested as a continuous and long-run measure of competition (Shaffer, 2004). It is also calculated at the national level and cannot be used to assess firm-level decisions of banks.

Three studies use the Lerner index or a bank-level measure of competition to investigate the implications of market power on bank efficiency in the context of developed countries. Maudos and De Guevara (2007) find a positive relationship between market power and cost efficiency in European banking, thus rejecting the ‘‘quiet life” hypothesis. However, a comprehensive analysis of the relationship between market power and efficiency should consider both cost and profit efficiency. While cost minimization is a neces- sary condition to maximize profits, banks may also achieve higher profits by diversifying their revenue sources. Using US and Euro- pean samples, Schaeck and Cihak (2008) report that competition improves profit efficiency. A less innocuous problem arises from reverse causality between bank efficiency and the degree of market power. Under the Efficient Structure Hypothesis, most efficient banks survive competitive pressures and gain market share at

6 Berger and Humphrey (1997) provide a detailed review of the literature on banking efficiency, which is also updated by Berger and Mester (2003).

8

R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775 767

the expense of less efficient banks (Demsetz, 1973). Koetter et al. (2008) acknowledge that competition and efficiency are inter- twined. They use a structural model to find support for the predic- tions of the Efficient Structure Hypothesis rather than those of the ‘‘quiet life” hypothesis for US banks.

More recently, Delis and Tsionas (2009) provide an empirical framework for the joint estimation of efficiency and market power for a sample of European and US banks. The authors use a novel maximum localization technique to derive bank-specific estimates of market power, and report a negative relationship between mar- ket power and efficiency, in line with the predictions of the ‘‘quiet life” hypothesis.

2.2. Market power and bank stability

The traditional ‘‘competition-fragility” view contends that mar- ket power in banking may be desirable, despite possible ensuing efficiency losses. A bank with market power is likely to reduce the information asymmetry problem and develop on-going rela- tionships with individual firms (Petersen and Rajan, 1995). Incum- bent banks are prone to screen borrowers and differentiate between low- and high-quality debtors (Cetorelli and Peretto, 2000). This may improve loan portfolio quality and enhance bank stability. Besanko and Thakor (1993) find that banks which appro- priate informational rents from developing relationships with bor- rowers may have more incentives to limit their risk exposure. Keeley (1990) finds that increased competition has eroded the franchise value of US banks, leading to more risk-taking and a surge of bank failures in the 1980s.7 Carletti and Vives (2008) re- view the literature on competition and stability and show that, while banking is no longer an exception in the enforcement of competition policy rules in the European Union, market power may have a mod- erating effect on bank risk-taking incentives.

Recently, a counter trend has emerged both at the theoretical and empirical levels to support the ‘‘competition-stability” view and re- fute the traditional trade-off between market power and bank sta- bility. Caminal and Matutes (2002) show that monopoly banks incur monitoring costs and are inclined to originate risky loan port- folios; Beck et al. (2004) report that bank stability is enhanced in both more concentrated and competitive markets; and Allen and Gale (2004) argue that this relationship is complex and case depen- dent. Boyd and De Nicoló (2005) argue that the implications of mar- ket power have to be examined separately for the deposit and loan markets. Banks with more loan market power are in a position to charge higher rates for loan customers. This makes it harder for bor- rowers to repay loans, thereby exacerbating moral hazard incentives to shift into riskier projects and possibly resulting in a riskier set of bank clients due to adverse selection considerations.

A large body of empirical evidence employs concentration ratios to support the ‘‘competition-stability” view (see for example De Nicoló, 2000; De Nicoló et al., 2004; Boyd et al., 2006; Uhde and Heimeshoff, 2009). Alternatively, Schaeck et al. (2009) use the PR H-statistic to show that competitive banking markets are more sta- ble than monopolistic systems. Schaeck and Cihak (2008) employ another country-level measure of the intensity of competition (the Boone indicator) to establish that competition increases bank soundness through the efficiency channel.

Berger et al. (2009) show that two strands of the literature (the ‘‘competition-fragility” and the ‘‘competition-stability” views) need not necessarily yield opposing predictions regarding the ef- fects of competition on bank stability in the context of developed countries. While loan market power can result in riskier loan port- folios, banks may protect their overall franchise value using other

7 For a survey of the literature on financial stability and competition, see Carletti and Hartmann (2003).

means, such as increasing their equity capital or engaging in other risk-mitigating techniques.

3. Tripod estimation methodology: market power, bank efficiency and bank stability

3.1. Market power

This paper employs three different specifications of Lerner to investigate the implications of market power: a conventional Lern- er (Berger et al., 2009), an efficiency-adjusted Lerner (Koetter et al., 2008), and a funding-adjusted Lerner (Maudos and De Guevara, 2007).

The conventional Lerner indicator of market power is defined as:

ðPTA � MCTAÞ=PTA: ð1Þ

The Lerner index captures the essence of pricing power because it measures the disparity between price and marginal costs ex- pressed as a percentage of price. Ideally, output price or PTA should take into consideration the price of loans and deposits separately. However, the statistical data does not provide sufficient grounds to estimate separate prices or rates for loans and deposits. Loan rev- enues cannot be disentangled from those earned on other fixed-in- come investments, and deposit interest expenses cannot be isolated from interest which is paid on other liabilities. Consequently, the construction of the Lerner index rests on the estimation of price and marginal costs of a single indicator of total banking activity. Following the literature, total assets account for the aggregate prod- uct of the bank.8 Under the assumption that the heterogeneous flow of services produced by a bank is proportional to its total assets, PTA is calculated as the ratio of total revenues to total assets.

In order to derive MC, we estimate the following translog cost function for each country separately to reflect different technolo- gies, while capturing bank specificities using bank fixed effects:

ln Cost ¼ b0 þ b1 ln Q þ b2 2

ln Q 2 þ X2 k¼1

ck ln W k þ X2 k¼1

ak ln Zk

þ 1 2

X2 k¼1

X2 j¼1

hkj ln W k ln W j þ 1 2

X2 k¼1

X2 j¼1

jkj ln Zk ln Zj

þ 1 2

X2 k¼1

/k ln Q ln W k þ 1 2

X2 k¼1

gk ln Q ln Zk

þ 1 2

X2 k¼1

X2 j¼1

xkj ln W k ln Zj þ d1Trend þ 1 2

d2 Trend 2

þ d3 Trend � ln Q þ X2 k¼1

kk Trend � ln W k

þ X2 k¼1

qkTrend � ln Zk þ e; ð2Þ

where bank costs (Cost) are a function of output (Q for total assets), three input prices (W1 for the price of funds, W2 for the price of physical capital, and W3 for the price of labor), a vector of fixed net- puts (Z1 for fixed assets, Z2 for total nominal value of off-balance sheet items, and Z3 for equity capital), and technical change (Trend to capture movements in the cost function over time).9 Standard symmetry restrictions and input price homogeneity of degree one are required to estimate (2). We also scale cost and input prices by

See, for example, Angelini and Ceterolli (2003). 9 Data for developing countries is expected to be noisy. In estimating Eq. (2), we

exclude outliers on input prices at the 99th percentile. Appendix A provides information on descriptive statistics for variables entering Eq. (2).

12

768 R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775

W3, and netputs by Z3 to correct for heteroskedasticity and scale biases. Marginal costs MCTA are then computed as

10:

MC ¼ Cost

Q b1 þ b2 ln Q þ

X3 k¼1

/k ln W k þ d3 Trend " #

: ð3Þ

There are two potential problems associated with the estima- tion of the conventional Lerner index. First, the Efficient Structure Hypothesis postulates that efficiency (and stability) may be driving market structure, and reverse causality is likely to prevail between the variables of interest. Conventional Lerner indices implicitly as- sume full bank efficiency and fail to consider the possibility that banks may not exploit pricing opportunities resulting from market power. Following Koetter et al. (2008), we account for the endoge- neity bias by deriving efficiency-adjusted Lerner indices from a sin- gle structural model:

ðARTA � MCTAÞ=ARTA: ð4Þ AR denotes average revenues, or TR

^ =TA, and TR

^ ¼ TP

^ þTC ^

. The key to obtaining an efficiency-adjusted Lerner index is to estimate

expected profits TP ^

from an alternative profit function (defined be-

low), and to combine them with expected total costs TC ^

derived

from Eq. (2). Unlike the conventional Lerner of Eq. (1), such a struc- tural model allows for the simultaneous estimation of both bank efficiency and the degree of market power, thereby addressing endogeneity concerns.

The second issue associated with traditional Lerner calculation is that MC estimation following Eq. (2) is likely to reflect some form of monopoly power that has arisen in the deposit market, based on the bank’s ability to raise funds at a cheap cost. Typically, in pricing loans, bank managers cover their funding costs, factor in a risk pre- mium to reflect the uncertainty surrounding the loan contracting problem, and charge on top of that another premium to reflect the exercise of their market power. So, effectively, some form of deposit market power is already reflected in the pricing of loans. Maudos and De Guevara (2007) argue that including financial costs and consequently the price of deposits in the cost function cap- tures the effect of market power in banking and may bias the find- ings. By excluding funding costs, one is likely to obtain a ‘‘raw” or ‘‘clean” proxy of pricing power that is not distorted by market power which had previously originated in the deposit market while raising funds. More specifically, we estimate a variety of Eq. (2) by including only operating costs (the price of labor and the price of physical capital) in the translog cost function, and omitting financing costs (the cost of funds). After calculating an operating MC for each bank at each time period following Eq. (3) but including only two factor prices, we derive a funding-adjusted Lerner index from the structural model specified by Eq. (4).

The differences among the three Lerner specifications can be briefly summarized as follows. Contrary to the conventional Lerner index, the efficiency-adjusted Lerner accounts for the inter-relat- edness of competition and efficiency. Thus, it may provide a better basis to examining the implications of the degree of market power on issues of efficiency and stability. The funding-adjusted Lerner further accounts for market power that may have previously been exercised in the deposit market, and which is otherwise likely to bias the findings.

3.2. Bank efficiency

A voluminous body of literature has extensively investigated the concept of bank efficiency using theoretical and applied models.11

10 More details on the estimation of the Lerner index can be found in Berger et al. (2009).

11 See Berger and Mester (2003).

Cost and profit efficiency levels measure how well a bank is pre- dicted to perform relative to other banks in a particular sample or a peer group for producing the same output bundle under the same exogenous conditions. Following the intermediation ap- proach, banks are modeled as financial intermediaries that collect deposits and other liabilities and transfer them into interest-earn- ing assets such as loans and investments (Sealey and Lindley, 1977). Using parametric stochastic frontier analysis, cost and profit efficiency scores are estimated from the following equation:

ln A ¼ fðln Q; ln WÞþ ln e; ð5Þ

where A is either total operating costs or total profits, and Q and W denote bank output and input prices defined above. The underlying functional form used is the translog specification of Eq. (1) where the dependent variable is either bank profits of operating costs.12

The error term e is decomposed into v � u (v + u) for the profit (cost) model, where v and u are two components that are assumed to be multiplicatively separable from the rest of the function. While v is a two-sided disturbance that accounts for uncontrollable (random) factors, u is a one-sided non-negative inefficiency term. Using the maximum likelihood technique, Eq. (5) is estimated separately for each country with bank fixed effects to derive individual bank effi- ciency scores (Battese and Coelli, 1992). Following Berger and Mes- ter (2003), alternative profit efficiency is preferred over the standard profit function because of the international dimension of the sample.13

3.3. Bank stability

The Z-index assess overall stability at the bank level (Boyd et al., 2006; Berger et al., 2009). This proxy of bank stability combines indicators of profitability, leverage, and return volatility into a sin- gle measure. It provides information on the number of standard deviation units by which profitability would have to decline before bank capitalization is depleted. It is given by the ratio:

Z ¼ ROA þ E=TA

rROA ; ð6Þ

where ROA and E=TA are the average return on assets and equi- ty to total assets, respectively, over the sample period, and rROA is the standard deviation of return on assets. The bank stability indicator increases with higher profitability and capitalization levels, and decreases with unstable earnings reflected by a higher standard deviation of return on assets. Stated differently, an increase (decrease) in the Z-index indicates a decrease (in- crease) in overall bank risk exposure and more (less) bank stability.

Since it is difficult to assess and capture bank stability using a single measure, the sensitivity of the results is also checked using risk-adjusted measures of return for each bank following Mercieca et al. (2007) as:

RORROA ¼ ROA rROA

and RORROE ¼ ROE rROE

; ð7Þ

where RORROA and RORROE denote risk-adjusted ROA and ROE, respectively. Here again, higher values of risk-adjusted rates of re- turn indicate more bank stability.

Following the literature and in order to avoid taking the logarithm of a negative value, we add a constant to profits for all the banks in the sample (Berger and Mester, 2003).

13 Also, information on output prices that is necessary for estimating standard profit efficiency is not available.

Table 1 Country and bank representation in the sample. Source: BankScope.

Region Africa East/South Asia and Pacific Eastern Europe and Central Asia Latin America and Caribbean Middle East

No. of countries 14 8 20 14 4 No. of banks 98 156 292 233 42 Average total assets (USD mn) 212.99 1494.44 486.97 619.39 4385.28

15 The results of specification tests indicate that a cross-section analysis of banks is preferred over a panel specification.

16 In such a two-stage approach, efficiency scores are derived from a first-stage regression and then their determinants are identified. This methodology is preferred over a one-stage model where the exogenous variables of Eq. (8) enter as additional controls in the cost/ profit functions.

17 We also implement a non-linear logistic specification by transforming cost and profit efficiency scores into ln[Eff/(1 � Eff)]. However, the specification tests indicate that Tobit models are preferred over the logistic transformation.

18 In order to reduce scale bias, we consider the natural logarithm of the Z-index. 19 In the cost/profit functions from which efficiency scores are estimated, equity

capital enters as a fixed netput, and the Z-index also includes the ratio of equity to assets, implying that equity is considered twice. Using alternative measures of bank risk in Eq. (8), RORROA and RORROE, lends more support to the analysis.

20 Unlike the Lerner index, the estimation of concentration ratios and the PR H- statistic occurs at the country level.

21 We also test for the presence of endogeneity using an instrumental Tobit model

R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775 769

4. Data and methodology

4.1. Data

The data consists of bank-level financial statements for the years 1999–2005, retrieved from the BankScope database provided by Fitch-IBCA (International Bank Credit Analysis Ltd.) and which has comprehensive coverage in most countries. The sample in- cludes commercial banks in developing countries from five differ- ent world regions, including Africa, East/South Asia and Pacific, Eastern Europe and Central Asia, Latin America and Caribbean, and the Middle East.14 The original sample is filtered by excluding banks with less than three consecutive yearly observations, and banks for which data on the main variables are not available (such as loans, personnel expenses or net income). We also exclude obser- vations with negative values for loans, interest revenues and interest expenses. This reduces the sample to an unbalanced panel of 821 banks in 60 countries or a total of 4670 observations. Tables 1 and 2 provide descriptive statistics on the sample.

In terms of the number of banks, the Eastern Europe and Central Asia region dominates the sample, although the average size of banks over the period under study is largest in the Middle East, fol- lowed by the East/ South Asia and Pacific region.

In contrast to loans to assets ratios usually observed for banks in developed countries (generally above 60%), the highest average loans to assets is 51% for banks operating in Latin American and Caribbean and the lowest credit risk exposure in other emerging markets stands at less than 39% for Middle East. This might indicate that banks in developing countries have fewer incentives to engage in lending activities compared to developed nations. It could be that the legal setup and investor protection in developing countries do not encourage banks to increase portfolio risk. Credit bureaus, if existent, may not play an active role in developing countries, and the presence of investor protection laws does not guarantee effi- cient judicial enforcement. In this light, bank managers may prefer to generate income from other less risky uses of funds compared to loans. Figures in Table 2 indicate that banks in developing countries generally rely equally (if not more so) on other earning assets as alternative uses of funds. To illustrate, in the case of African, Middle Eastern and Eastern Europe and Central Asia banks, the share of other earning assets on average exceeds the proportion of loans on banks’ balance sheets, while the two ratios are almost at par with each other for East and South-East Asia banks.

Figures appearing in Table 2 also show that deposits and short- term funding represent the main sources of funds, and that banks in developing countries are on average well capitalized. Banks operating in the Middle East region have the lowest net interest margin and return on assets, while the highest records of profit- ability as measured by ROA are for banks in Africa.

4.2. Methodology

In order to investigate the implications of the degree of market power on bank efficiency and stability, we run several cross-sec- tion regressions following the baseline model:

14 The grouping of countries into different regions follows the World Bank classification.

Y ¼ fðMarket Power; Portfolio Characteristics; Regulatory EnvironmentÞ; ð8Þ

where the dependent variable Y measures each of bank cost effi- ciency, alternative profit efficiency and stability, all of which are calculated at the bank level to run cross-section regressions.15 Since bank cost and alternative profit efficiency scores are bound between zero and one, Tobit models are more appropriate because they bet- ter fit models where the dependent variable is derived from a first- stage regression (Greene, 2005).16 The results of specification tests also confirm that a Tobit specification is preferred to a conventional treatment of efficiency scores.17

In turn, the Z-index proxies bank stability, with larger values indicating more bank stability and less bank risk potential.18 As sensitivity checks, the two other risk-adjusted rates of returns indi- cators used are RORROA and RORROE.

19

The main independent variable in (8) is the degree market power measured by the Lerner index, or the mark-up of price over marginal costs, with higher values implying higher pricing power and less competitive market conditions. The Lerner index is the preferred measure of the degree of market power compared to other traditional indicators of market structure because it is ob- served at the bank level, similar to the unit of analysis of bank effi- ciency and stability to which they are related.20 The three different specifications of the Lerner index used include a conventional Lerner, an efficiency-adjusted Lerner, and a funding-adjusted Lerner that further accounts for market power arising in the deposit market. The last two adjusted Lerner measures derive from a structural model which, as explained in the previous section, better account for the inter-relatedness between market power and bank efficiency, thereby addressing endogeneity concerns.21

Bank portfolio characteristics include portfolio size (or bank size measured by the natural logarithm of total assets) and mix (or the bank’s credit exposure measured by the ratio of loans to as- sets). Previous studies have established that bank regulation, supervision, and the institutional framework affect banking system soundness (see for example Beck et al., 2004 and Barth et al., 2007). Eq. (8) controls for the Regulatory Environment with two indica- tors, foreign ownership and legal rights. Foreign bank ownership

for cost and profit efficiency, and a two-stage least-squares model for the Z-index. The specification tests indicate that exogenous models are more appropriate than instrumental models, probably because the above-mentioned Lerner indices already account for some of the endogeneity that is likely to be present in the models.

Table 2 Descriptive statistics by region (% of total assets). Source: BankScope.

Region Africa East/South Asia and Pacific Eastern Europe and Central Asia Latin America and Caribbean Middle East

Net loans 42.38 49.84 46.90 51.06 38.52 Other earning assets 53.27 48.70 48.27 44.81 59.34 Total deposits 75.69 81.46 72.73 71.68 83.15 Total equity 13.78 10.55 17.49 18.83 10.88 Net interest margin 5.96 2.99 4.57 7.92 2.55 Net income 1.87 1.18 1.48 1.43 1.12

770 R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775

is a dummy variable that is set to one when foreign shareholding exceeds 50% of total bank ownership. Legal rights represents an in- dex measuring the degree to which collateral and bankruptcy laws facilitate lending. It ranges from 0 to 10 with higher scores indicat- ing that collateral and bankruptcy laws are better designed to ex- pand access to credit (Djankov et al., 2007). All regressions include the natural logarithm of per capita Gross Domestic Product to con- trol for different levels of economic development across developing countries. Finally, bootstrapping the standard errors of all variants of Eq. (8) allows a better comparison of results across specifications.

5. Empirical results

5.1. Tripod estimation results

Table 3 presents the results of the country averages of the con- ventional Lerner index, the efficiency-adjusted Lerner index, the funding-adjusted Lerner index, cost efficiency, profit efficiency, the Z-index, and risk-adjusted rates of return over the period of 1999–2005.

The conventional Lerner figures show varying degrees of market power across countries, but the figures are generally closely aligned across all regions (around 30% price mark-up over marginal costs) except for Latin America and the Caribbean where the con- ventional Lerner is as low as 17%. The estimated efficiency and funding-adjusted Lerner indices also vary across countries and re- gions.22 In line of the findings of Koetter et al. (2008) for US banks, the magnitude of the efficiency-adjusted Lerner generally exceeds that of the conventional index. Similarly, the funding-adjusted Lern- er is also, on average, larger than the conventional Lerner, suggesting that the latter generally overestimates the degree of market power, and thus justifying the use of alternative Lerner specifications.

The measure of bank cost (profit) efficiency is the actual level of costs (profits) relative to an efficient cost (profit) frontier. The effi- ciency estimation results appearing in Table 3 are in line with those reported in the literature, with higher scores indicating bet- ter efficiency levels (see Berger and Humphrey, 1997). While cost efficiency levels are closely aligned across world regions, profit efficiency levels exhibit greater disparity. It should be noted, how- ever, that cross-country efficiency comparisons are to be treated with caution. It would be wrong to conclude that banks in the Mid- dle East are more profit efficient than banks in Latin America and the Caribbean. The reported efficiency averages per country or per region can only serve as reference, since a different frontier is estimated for each country.23

Measures of bank stability indicate that, on average, banks operating in Middle East developing countries appear to be ex- posed to the lowest risk potential compared to banks in other re-

22 As might be expected, correlation analyses show that various Lerner indices are highly correlated.

23 In the robustness section, we estimate cost and profit efficiency scores from a common global frontier while controlling for individual country effects. The main results are maintained.

gions. This finding is corroborated using the Z-index and risk- adjusted measures of returns. According to figures reported in Ta- ble 2, banks in the Middle East lend the lowest proportion of assets compared to banks in other regions, and they rely mostly on other earning assets as uses of funds. It could be the case that their earn- ings are likely to be more stable, resulting in higher Z-scores and risk-adjusted rates of return on average.24

The next section analyzes the implications of market power on bank efficiency and stability in a multivariate setting which con- trols for bank and country differences.

5.2. Implications of market power

Table 4 reports the results of the different estimations of Eq. (8) using bank cost and alternative profit efficiency as dependent vari- ables, and Table 5 considers the Z-index and RORROA measures of bank stability as exogenous variables. Each table includes the three different specifications of the Lerner index: a conventional Lerner (Model 1), an efficiency-adjusted Lerner (Model 2), and a fund- ing-adjusted Lerner (Model 3).

Following Berger et al. (2009), we include a quadratic term for the Lerner index to allow for a non-linear relationship between competition and each of bank efficiency and stability. In order to establish the sign of the relationship between the independent var- iable (Lerner index) and each of the dependent variables, the inflection point is calculated for every specification by setting the first-order derivative to zero and comparing its value to the empir- ical distribution of the Lerner index data. To illustrate, the inflec- tion point of Model 1 in Table 4 is �4.62, while the 1st percentile of the Lerner index data occurs at �0.43, implying that more than 99% of the degree of market power data lies above the inflection point. Given that the sign of the quadratic coefficient in Model 1 is negative, the resulting estimated function is a downward ori- ented or reverse parabola that decreases above the inflection point. Therefore, the empirical estimation supports a negative association between a bank’s degree of market power and its level of cost effi- ciency. A similar analysis for each estimated model reports the sign of the relationship between variables of interest (+/�).

The significant negative relationship between a bank’s degree of market power and cost efficiency holds across all three Lerner specifications. This suggests that banks with more market power operating in developing countries are not able to reduce costs and achieve lower cost efficiency levels compared to their peers. Hughes et al. (2003) argue that management may signal market power by maintaining large offices and other excessive spending, possibly driving significant cost efficiency losses. However, except for Delis and Tsionas (2009) who report similar negative associa- tion between market power and efficiency for a sample of EU and US banks, the results do not agree with those reported for developed countries. In the context of the EU, Maudos and De Guevara (2007) find that banks with more market power are able to achieve higher cost efficiency levels, and Casu and Girardone

24 Interestingly, none of the four developing countries in the Middle East has had to bail out banks following the recent global financial crisis.

Table 3 Tripod estimation results for the degree of market power, bank efficiency and stability. Source: Author’s calculations.

Country Conventional Lerner

Efficiency-adjusted Lerner

Funding-adjusted Lerner

Cost efficiency

Profit efficiency

Z-index Risk-adjusted ROA

Risk-adjusted ROE

East/South Asia and Pacific Bangladesh 21.79 70.91 67.51 82.56 33.62 45.48 2.45 2.55 Cambodia 46.75 57.40 56.53 82.54 44.51 40.64 3.61 4.67 India 24.96 62.37 56.64 84.21 34.79 24.90 3.28 3.26 Indonesia 23.47 64.56 60.06 82.15 48.10 23.78 2.55 2.95 Malaysia 44.28 78.48 77.20 76.91 51.31 40.55 1.68 1.44 Nepal 30.68 72.25 70.36 82.09 41.97 31.88 4.41 5.15 Pakistan 11.59 61.61 57.90 84.77 26.04 22.71 1.70 1.40 Vietnam 20.58 75.71 73.46 83.59 33.77 37.65 2.92 3.97 Average 28.01 67.91 64.96 82.35 39.26 33.45 2.83 3.17

Eastern Europe and Central Asia Albania 27.67 57.43 55.13 83.46 29.31 24.10 1.62 1.86 Armenia 34.46 32.77 31.74 84.95 53.56 15.65 2.40 2.57 Azerbaijan 30.72 33.92 32.02 84.88 52.84 20.27 1.90 1.23 Bulgaria 29.31 40.05 37.81 84.00 31.87 50.75 3.59 4.57 Croatia 29.04 45.53 41.66 83.34 29.85 36.06 2.68 3.06 Czech Republic 25.30 62.18 60.11 82.22 24.64 29.66 1.56 2.21 Georgia Rep. 41.30 28.78 26.57 83.97 59.90 26.63 3.13 2.78 Hungary 24.60 46.01 40.84 84.28 36.95 19.69 1.77 2.35 Kazakhstan 32.17 35.44 31.65 79.68 43.20 20.39 2.40 2.30 Latvia 30.35 57.76 56.99 81.89 33.01 21.38 1.95 2.17 Macedonia 47.57 44.79 42.29 82.72 52.57 51.36 3.56 4.05 Moldova Rep. of 33.03 22.17 20.10 84.80 57.11 20.25 2.45 3.09 Poland 23.34 50.95 45.93 82.96 29.50 20.81 1.63 2.00 Romania 15.26 28.55 22.26 84.78 42.59 16.12 0.47 0.73 Russian Fed. 3.49 36.62 33.93 83.55 39.87 27.37 2.00 2.18 Serbia 41.42 10.66 8.84 88.37 46.97 103.93 4.79 4.78 Slovakia 19.72 30.77 28.67 85.95 29.88 18.31 1.72 2.17 Slovenia 26.72 60.68 56.57 82.75 30.74 36.87 2.35 2.96 Ukraine 25.76 33.23 27.29 84.99 36.54 20.62 1.38 1.42 Uzbekistan 38.98 26.41 20.83 85.45 50.12 22.74 4.54 3.83 Average 29.01 39.24 36.06 83.95 40.55 30.15 2.39 2.62

Latin America and Caribbean Argentina 8.82 10.96 5.80 84.14 38.15 18.50 0.17 0.36 Bolivia 21.91 25.67 20.39 85.29 21.99 17.33 0.63 0.67 Brazil 23.00 46.17 35.11 78.21 52.99 20.88 2.46 2.33 Chile 20.71 46.22 45.05 81.18 33.69 25.69 1.12 0.97 Colombia 20.78 1.08 -11.15 86.88 37.51 11.71 1.13 1.12 Costa Rica 19.53 54.46 50.49 82.07 29.28 36.19 3.95 4.47 Dominican

Republic 11.41 37.19 30.28 84.29 44.89 20.27 2.09 2.45

Ecuador 19.53 15.39 14.65 86.33 28.12 27.45 2.49 2.10 El Salvador 26.98 42.25 38.62 82.64 31.82 34.65 1.61 1.75 Honduras 19.98 34.96 28.77 84.66 35.46 25.87 2.56 2.91 Paraguay 0.08 6.09 �18.05 88.99 27.78 16.14 1.92 1.89 Peru 22.35 18.02 13.74 84.82 23.32 27.00 1.72 1.51 Uruguay 6.07 53.06 37.83 89.57 25.75 12.52 0.49 0.70 Venezuela 28.95 18.16 13.03 84.56 47.72 18.78 2.66 2.32 Average 17.86 29.26 21.76 84.55 34.18 22.36 1.79 1.82

Middle East Iran 30.53 79.35 76.71 88.01 76.96 21.45 3.06 2.50 Jordan 32.27 58.50 54.84 82.68 40.28 34.84 3.15 2.45 Lebanon 10.31 71.23 66.95 84.78 23.99 41.71 2.98 3.40 Saudi Arabia 43.12 61.86 58.42 83.20 64.49 75.80 6.20 5.49 Average 29.06 67.74 64.23 84.67 51.43 43.45 3.85 3.46

Africa Angola 48.38 35.16 32.50 86.39 56.93 9.05 1.65 1.95 Burkina Faso 32.78 35.54 33.51 85.11 43.65 14.03 2.09 1.71 Cameroon 46.21 57.02 54.63 82.67 44.04 14.29 2.24 1.84 Congo 25.96 -33.17 -23.49 91.76 31.80 13.40 1.68 2.06 Ghana 26.85 29.72 25.25 83.95 51.93 27.78 5.63 4.88 Ivory Coast 32.90 27.25 25.06 85.27 25.89 12.50 0.67 0.60 Kenya 24.39 43.72 41.04 82.58 36.78 62.76 4.18 4.49 Mauritius 19.56 80.41 78.65 82.82 36.35 21.91 1.57 1.95 Mozambique 16.98 1.36 2.17 86.26 55.66 17.37 1.99 1.84 Nigeria 32.96 11.64 3.18 85.75 67.74 23.34 4.29 2.40 Senegal 30.07 38.47 37.11 85.35 42.02 26.36 3.83 3.06 Sudan 24.97 30.10 28.59 85.19 30.45 14.35 1.48 1.64

(continued on next page)

R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775 771

Table 3 (continued)

Country Conventional Lerner

Efficiency-adjusted Lerner

Funding-adjusted Lerner

Cost efficiency

Profit efficiency

Z-index Risk-adjusted ROA

Risk-adjusted ROE

Tunisia 25.80 57.71 55.09 80.66 32.40 36.19 2.83 2.29 Zambia 35.78 11.48 10.31 84.28 61.04 12.01 2.67 2.82 Average 30.26 30.46 28.83 84.86 44.05 21.81 2.63 2.39

Lerner index is the price mark-up over marginal cost expressed as a percentage of price (PTA � MCTA)/PTA. A higher Lerner index indicates a higher degree of monopoly power. Three varieties of the Lerner are reported, a conventional Lerner, an efficiency-adjusted Lerner, and a funding-adjusted Lerner. Cost and profit efficiency indices are estimated employing stochastic frontier analysis. Higher scores indicate better efficiency levels. Z-index = ((ROA + E/TA)/rROA) where ROA is return on assets, E/TA is equity to assets, rROA is the standard deviation of ROA. A larger Z-index indicates more stability and less bank risk. Risk-Adjusted ROA and ROE – calculated as ROA/rROA and ROE/rROE, respectively, are other indicators of bank stability. All figures except for the Z-index are in %.

Table 4 Market power and bank efficiency.

Dependent variable: cost efficiency Dependent variable: alternative profit efficiency

Model 1: conventional Lerner

Model 2: efficiency- adjusted Lerner

Model 3: funding- adjusted Lerner

Model 1: conventional Lerner

Model 2: efficiency- adjusted Lerner

Model 3: funding- adjusted Lerner

Lerner index �0.0351 (0.0122)***

�0.0365 (0.0061)***

�0.071 (0.0056)***

0.4241 (0.0714)***

0.0479 (0.0258)*

0.043 (0.0275)

(Lerner index)2 �0.0038 (0.0186)

�0.0125 (0.0153)

�0.0092 (0.0082)

0.0393 (0.1597)

�0.0208 (0.0093)**

�0.0156 (0.0157)

Loans to assets 0.0116 (0.0158)

0.0184 (0.0117)

0.0085 (0.0117)

�0.0562 (0.0335)*

�0.0833 (0.0416)**

�0.0808 (0.0448)*

Ln (total assets) 0.0012 (0.0014)

0.002 (0.0011)*

0.0032 (0.0010)***

�0.01 (0.0038)***

�0.0015 (0.0056)

�0.0012 (0.0047)

Foreign ownership

�0.01 (0.0064)

�0.0093 (0.0047)**

�0.0054 (0.0052)

0.0244 (0.0219)

0.0282 (0.0244)

0.0278 (0.0229)

Legal rights �0.0005 (0.0010)

�0.0004 (0.0011)

0.0018 (0.0011)

�0.0055 (0.0042)

�0.003 (0.0045)

�0.0033 (0.0052)

Ln (GDP pc) �0.0056 (0.0013)***

�0.0051 (0.0013)***

�0.0054 (0.0011)***

�0.0049 (0.0056)

�0.0189 (0.0075)**

�0.0186 (0.0078)**

Inflection point Sign of relationship

�4.62 –

�1.46 –

�3.86 –

�5.40 +

1.15 +

1.38 +

Marginal effects �0.0104 �0.0226 �0.0347 0.0273 0.0364 0.0311

Results from Tobit regression models to explain the implications of market power on bank cost efficiency and alternative profit efficiency. Bank cost efficiency is derived from a cost function, and profit efficiency is derived from an alternative profit function. Higher values of cost and profit efficiency scores indicate better cost and profit efficiency levels. The degree of market power is proxied by the Lerner Index or the price mark-up over marginal cost, with higher values indicating a higher degree of pricing power. Three different specifications of Lerner are included, a conventional Lerner, a funding-adjusted Lerner, and an efficiency-adjusted Lerner. The natural logarithm of total assets accounts for bank size, and the loans to assets ratio accounts for the portfolio mix of banks. Foreign bank ownership implies that total foreign shareholding exceeds 50% of total bank ownership (BankScope). Legal rights measures the degree to which collateral and bankruptcy laws facilitate lending (Djankov et al., 2007). The natural logarithm of per capita GDP accounts for differences in economic developments across countries. All models are run with bootstrapped standard errors (reported in parentheses) clustered by country. Data are for a cross-section of 821 banks from 60 developing countries.

* p < 0.1. ** p < 0.05.

*** p < 0.01.

772 R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775

(2006) conclude that the effect of competition on cost efficiency is not clear-cut. The findings are also not in line with those for the US reported by Koetter et al. (2008), namely that banks with more market power are also the most cost efficient. One should be cau- tious, however, before concluding that the results support the ‘‘quiet life” hypothesis. It is likely that the higher costs that are associated with more market power are eventually channeled to bank clients which, in turn, may feed into higher prices and possi- bly boost bank profit efficiency.

The results indicate that bank portfolio composition and size are not significant determinants of cost efficiency using the con- ventional Lerner, but the coefficient on total assets turns signifi- cantly positive when using the efficiency and funding-adjusted Lerner indices. This suggests that, when accounting for endogene- ity bias, larger banks are able to achieve higher cost efficiency lev- els. Foreign banks presence is associated with lower cost efficiency levels, but this finding is significant only when considering the effi- ciency-adjusted Lerner. The regulatory environment in terms of le- gal rights does not significantly affect cost efficiency, while a

higher level of economic development is significantly negatively associated with bank cost efficiency. It could be that growth strat- egies in developing economies outweigh cost efficiency consider- ations over the short-term.

While it seems that banks do not respond favorably to a higher degree of market power in terms of controlling costs more effec- tively, it is important to assess whether they are able to generate more revenue and/ or improve the performance of their lending activities. Table 4 also shows the implications of the degree of mar- ket power (using the three different Lerner specifications) on bank alternative profit efficiency. The corresponding inflection point for Model 1 is estimated at �5.40, and the sign of the quadratic term is positive, pointing to a direct association between market power and alternative profit efficiency. This significant positive relation- ship is persistent across all models of alternative specifications of the Lerner index. The findings provide evidence against the ‘‘quiet life” hypothesis. They are opposite to those reported by Schaeck and Cihak (2008) who establish a positive effect of competition on alternative profit efficiency for EU and US banking. In developing

Table 5 Market power and bank stability.

Dependent variable: Z-index Dependent variable: risk-adjusted ROA

Model 1: conventional Lerner

Model 2: efficiency- adjusted Lerner

Model 3: funding- adjusted Lerner

Model 1: conventional Lerner

Model 2: efficiency- adjusted Lerner

Model 3: funding- adjusted Lerner

Lerner index 0.5162 (0.1408)***

0.2948 (0.1176)**

0.3184 (0.1017)***

3.6224 (0.6644)***

1.4498 (0.3263)***

1.4766 (0.3795)***

(Lerner index)2 0.0412 (0.0673)

0.0718 (0.0966)

0.0983 (0.0896)

0.2979 (0.6073)

0.1068 (0.3259)

0.1887 (0.2760)

Loans to assets 0.4534 (0.1782)**

0.4475 (0.1537)***

0.4693 (0.1595)***

1.2634 (0.7101)*

1.1474 (0.6826)*

1.2382 (0.9213)

Ln (total assets) 0.0132 (0.0241)

0.0096 (0.0186)

0.012 (0.0229)

0.2514 (0.0743)***

0.2649 (0.0723)***

0.2782 (0.0778)***

Foreign ownership

�0.1123 (0.0825)

�0.12 (0.0977)

�0.1269 (0.0863)

�0.1625 (0.4025)

�0.1753 (0.4032)

�0.2008 (0.3955)

Legal rights 0.0585 (0.0169)***

0.0536 (0.0162)***

0.052 (0.0176)***

0.0523 (0.0925)

0.0418 (0.0765)

0.0354 (0.0708)

Ln (GDP pc) 0.0243 (0.0299)

0.0092 (0.0292)

0.0115 (0.0370)

�0.2148 (0.1202)*

�0.3285 (0.1086)***

�0.3186 (0.1046)**

Inflection point Sign of relationship

�6.26 +

�2.05 +

�1.62 +

�6.08 +

�6.79 +

�3.91 +

Marginal effects 0.1238 0.1449 0.1467 0.3443 0.2516 0.2386

Results from regression models to explain the implications of market power on the Z-index and risk-adjusted returns. The natural logarithm of the Z-index, a proxy for bank stability is calculated as the ratio (ROA + ETA)/rROA, where ROA is return on assets, ETA is equity to assets and rROA is the standard deviation of ROA. Higher Z-index values indicate more bank stability. The other dependent variable, risk-adjusted ROA, is another proxy for bank stability. It is calculated as ROAi=rROAi ; where ROAi is the bank’s average return on assets. Higher risk-adjusted ROA values indicate more bank stability. The degree of market power is proxied by the Lerner index or the price mark-up over marginal cost, with higher values indicating a higher degree of pricing power. Three different specifications of Lerner are included, a conventional Lerner, a funding-adjusted Lerner, and an efficiency-adjusted Lerner. The natural logarithm of total assets accounts for bank size, and the loans to assets ratio accounts for the portfolio mix of banks. Foreign bank ownership implies that total foreign shareholding exceeds 50% of total bank ownership (BankScope). Legal rights measures the degree to which collateral and bankruptcy laws facilitate lending (Djankov et al., 2007). The natural logarithm of per capita GDP accounts for differences in economic developments across countries. All models are run with bootstrapped standard errors (reported in parentheses) clustered by country. Data are for a cross-section of 821 banks from 60 developing countries.

* p < 0.1. ** p < 0.05.

*** p < 0.01.

25 Similar results (not reported) obtain when risk-adjusted ROE is used.

R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775 773

countries, banks with more market power are able to derive signif- icant revenue gains from diversified portfolios. Also, the results using the conventional Lerner show that banks that lend a higher portion of their assets are significantly less profit efficient, and that larger banks are marginally less profit efficient compared to their peers. This finding is in line with the literature that reports little or no benefit from consolidation and conglomeration (De Nicoló, 2000).

Table 5 shows the results of the implication of market power on bank stability, using the Z-index and risk-adjusted return on assets as proxies of overall bank stability.

The inflection point for Model 1 of Table 5 using the Z-index and the conventional Lerner specification is �6.26, which is far below the 1st percentile of the conventional Lerner index data. Since the sign of the quadratic term is positive, the estimated function is an upward parabola that rises after the inflection point, implying a direct relationship between the degree of market power and the Z-index. The results point to a significant and positive relationship between a bank’s degree of market power and its level of stability across all Lerner specifications. This suggests that banks with a lar- ger degree of market power also enjoy a higher level of overall bank stability and reduced risk potential. The findings for develop- ing countries thus support the traditional view on the trade-off be- tween bank competition and stability, and do not agree with those reported by De Nicoló et al. (2004), Boyd et al. (2006) , Schaeck et al. (2009).

Banks which lend a greater portion of their assets exhibit signif- icantly higher Z-indices, suggesting that firms that have a higher credit risk exposure (higher loans to assets ratios) are in fact ex- posed to a lower level of overall bank risk. It could be the case that banks in developing countries which are active in extending credit to the economy hedge their portfolios or hold more equity capital

in order to reduce their risk potential and ensure that their overall stability is safeguarded. This is crucial given the fact that capital markets are practically non-existent in developing economies, and banks represent the main source of credit for firms. Further, better protection of legal rights fosters bank stability, while foreign banks presence does not seem to have a significant effect on overall bank risk exposure.

In order to check the sensitivity of the Z-index results to other indicators of bank stability, Table 5 also shows the implications of market power on risk-adjusted ROA.25 Here again, a positive sign is consistently reported between different measures of market power and risk-adjusted ROA. This indicates that a higher degree of market power is significantly positively associated with larger risk-adjusted rates of returns. Thus, the sensitivity checks using proxies other than the Z-index support the positive association be- tween market power and bank stability. As markets become more concentrated and banks gain market power, financial conglomerates in developing countries are likely to benefit from greater overall financial stability and lower variability of returns. The coefficients on loan portfolio composition and bank size are positive and gener- ally significant, implying that banks that engage in more lending activities are able to achieve higher risk-adjusted rates of return, and the effect is more pronounced at larger banks.

To sum, the empirical analysis for developing countries shows that a higher degree of market power results in profit efficiency gains and enhanced bank stability, despite significant cost effi- ciency losses. In order to analyze the marginal effects of a higher degree of market power on key variables of interest, the bottom row of Tables 4 and 5 reports the computed market power

Table A.1 Descriptive statistics on variables entering the cost and profit functions. (Variables are in logarithmic format.) Source: BankScope.

Region Cost Profit Q Wl Wk Wf Z1 Z2 Z3

East/South Asia and Pacific 10.28 8.74 13.02 �4.67 �0.22 �2.97 8.25 10.84 10.43 Mean 1.65 1.69 1.55 0.65 0.60 0.57 1.62 2.10 1.28 Std. dev. 4.29 1.61 8.50 �8.46 �2.46 �6.55 3.26 3.74 7.31 Min.

13.48 13.33 16.81 �2.34 0.88 �1.02 11.55 14.64 13.44 Max. Eastern Europe and Central Asia 9.46 7.95 11.99 �4.08 �0.41 �3.16 8.23 9.38 10.01 Mean

1.40 1.60 1.50 0.73 0.74 0.74 1.38 2.37 1.20 Std. dev. 4.20 2.71 8.03 �7.32 �3.56 �6.93 3.22 1.39 6.85 Min.

13.37 13.32 16.08 �2.29 0.88 �0.69 11.49 14.65 13.38 Max. Latin America and Caribbean 10.24 8.31 12.20 �3.75 �0.20 �2.59 8.31 10.55 10.23 Mean

1.55 1.66 1.53 0.80 0.75 0.86 1.74 2.36 1.18 Std. dev. 5.30 0.69 7.86 �9.93 �3.23 �5.67 2.08 2.94 6.91 Min.

13.57 13.52 17.01 �2.27 0.88 �0.64 11.54 14.66 13.42 Max. Middle East 11.12 9.44 14.06 �4.70 �0.81 �3.22 9.47 11.43 11.06 Mean

1.52 2.11 1.66 0.49 0.65 0.63 1.36 1.80 1.15 Std. dev. 5.31 1.95 10.25 �6.89 �2.93 �4.68 6.37 2.56 7.39 Min.

13.54 13.89 17.48 �3.25 0.83 �1.62 11.55 14.56 13.42 Max. Africa 9.18 7.88 11.65 �3.95 �0.33 �3.34 8.17 9.74 9.48 Mean

1.12 1.57 1.18 0.77 0.78 0.88 1.40 1.82 1.08 Std. dev. 5.63 2.30 8.15 �8.18 �3.56 �10.3 4.03 3.99 6.83 Min.

12.10 11.40 14.73 �2.29 0.86 �1.17 11.18 13.31 12.58 Max. Total 9.94 8.31 12.36 �4.13 �0.34 �2.98 8.32 10.21 10.16 Mean

1.57 1.73 1.62 0.81 0.73 0.80 1.57 2.32 1.24 Std. dev. 4.20 0.69 7.86 �9.93 �3.56 �10.3 2.08 1.39 6.83 Min.

13.57 13.89 17.48 �2.27 0.88 �0.64 11.55 14.66 13.44 Max.

774 R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775

elasticities across all models. It is interesting to note that the abso- lute value of the cost efficiency elasticity of market power is well below the profit efficiency elasticity of market power for two Lern- er specifications, and that the two marginal effects are almost sim- ilar in absolute terms using the funding-adjusted Lerner. For example, using Model 2, a one percent increase in the degree of market power, on average, reduces bank cost efficiency by 2.26% and improves profit efficiency by 3.64% across developing coun- tries. Further, the positive relationship between all Lerner indices and bank stability is also translated into positive elasticities or marginal effects.

6. Sensitivity analysis

We run a series of sensitivity checks on the baseline model of Eq. (8).26 First, the logistic transformation is implemented instead of a Tobit model when bank efficiency scores are considered. We also estimate cost and profit efficiency scores from a global frontier while accounting for country differences instead of using the country-spe- cific estimates of efficiency. Using these different estimates, the pre- vious findings on the implications of market power on bank cost and profit efficiency levels are maintained for developing countries.

Other robustness checks consist of removing the quadratic term of the Lerner index from all specifications, and including control variables (other than legal rights) which are retrieved from the World Bank’s Doing Business for the business environment. The main previous results obtain.

Finally, a time dimension is introduced for the Z-index and for all other variables (except for Regulatory Environment), and panel regressions are run as sensitivity checks. The Z-index is allowed to vary across time for each bank following De Nicoló et al. (2004), and using the following equation:

Zit ¼ ROAit þ E=TAit jROAit � ROAij

; ð9Þ

where ROAit and E/TAit are the bank’s return on assets and equity to total assets at time t, respectively, and ROAi is the period average re-

26 The results are not reported in order to conserve space, but they can be obtained from the author.

turn on assets for bank i. Similarly, time-varying efficiency scores and Lerner indices are considered to estimate the baseline model of Eq. (8). The advantage of the panel dimension of the data is that it allows investigating the impact of lagged values of market power on bank efficiency and stability, thereby addressing resilient endo- geneity concerns, notwithstanding the questionable variability of the Z-index across time. We run the estimations using bank fixed effects and with robust standard errors clustered by country. The results (not reported) support the previous findings using the cross-section models. When banks in developing countries enjoy a higher degree of market power, they do not manage their costs effectively. Instead, they are able to achieve higher profit efficiency levels while at the same time deriving greater firm stability.27

7. Summary and conclusions

Most emerging countries have recently embraced financial lib- eralization as a means to achieve higher rates of economic growth. With the globalization of financial services worldwide, the bank model is shifting toward a universal banking system to provide a wide array of financial services (including commercial activities, investment banking and insurance underwriting) under the um- brella of the same financial conglomerate. As competitive condi- tions tighten and banks seek to increase their degree of market power, policymakers are concerned with the overall implications of changing banking structures, especially in light of the adverse implications of the late global financial turmoil onto developing countries.

The relationship between competition policies and financial stability is poorly documented for developing countries and no consensus prevails in the literature on the implications of market power on bank stability. This paper examines the impact of a high- er degree of market power on each of bank efficiency and stability. Using data from 821 banks in 60 developing countries over the per- iod 1999–2005, we compute proxies for the degree of market power, bank efficiency and bank stability, all of which are

27 We also account for endogeneity using instruments to explain the degree of market power, including activity restrictions, banking freedom indicators and the percentage of state-owned assets (Berger et al., 2009). The main findings obtain.

R. Turk Ariss / Journal of Banking & Finance 34 (2010) 765–775 775

estimated at the bank level. This tripod empirical approach may represent a more holistic way of analyzing the competition-stabil- ity nexus in banking. We find a significant negative association be- tween bank market power and cost efficiency, and a significant positive relationship between market power and each of bank profit efficiency and overall stability.

In developing countries, banks that command a high price mark-up over marginal costs may be reasonably adept at improv- ing their profit efficiency, but they do not do so well in terms of cost efficiency levels. As geographical and regulatory borders re- cede and the use of information technology intensifies, the global dimension of banking may evolve to create new opportunities for bank managers who ensure a wider spectrum of returns, while possibly passing on the resulting excessive costs to their clients.

Further, as banks gain market power, they also benefit from greater firm stability and reduced risk potential. This result sup- ports the traditional view that increased competition may under- mine bank stability. It can also provide a rationale for the intense merger activity that has taken place over the last two decades in the context of developing countries. More importantly, the finding may be relevant for policymakers in developing countries where the banking system is strained. The global dimension of the recent financial crisis has demonstrated that no country is immune to the turmoil hitting financial markets in developed countries. While antitrust laws in the US ensure that banking markets remain com- petitive (Berger et al., 2009) and competition policy is taken seri- ously in the EU (Carletti and Vives, 2008), the results suggest that increased market power in stressed banking systems of devel- oping countries may in fact be welcome due to the likely increase in bank soundness.

Acknowledgements

I am grateful to Philip Molyneux, Claudia Girardone, Iftekhar Hasan, Gianni De Nicoló, Ike Mathur (editor), and an anonymous referee for their helpful comments and suggestions, in addition to the participants at the Tor Vergata 2008 and the FMA 2008 con- ferences. Any remaining errors are my responsibility.

Appendix A

See Table A.1.

References

Allen, F., Gale, D., 2004. Financial intermediaries and markets. Econometrica 72, 1023–1061.

Angelini, P., Cetorelli, N., 2003. The effects of regulatory reform on competition in the banking industry. Journal of Money, Credit and Banking 35, 663–684.

Barth, J., Caprio, G., Levine, R., 2007. Bank regulation and supervision (updated dataset). Working Paper No. 2588, World Bank.

Battese, G., Coelli, T., 1992. Frontier production functions, technical efficiency and panel data: With application to paddy farmers in India. Journal of Productivity Analysis 3, 153–169.

Beck, T., Demirgüç-Kunt, A., Maksimovic, V., 2004. Bank competition and access to finance: International evidence. Journal of Money, Credit and Banking 36, 627– 648.

Berger, A., Hannan, T., 1998. The efficiency cost of market power in the banking industry: A test of the ‘‘quiet life” and related hypotheses. The Review of Economics and Statistics 80, 454–465.

Berger, A., Humphrey, D., 1997. Efficiency of financial institutions: International survey and direction for future research. European Journal of Operation Research 98, 175–212.

Berger, A., Mester, L., 2003. Explaining the dramatic changes in performance of US banks: Technological change, deregulation, and dynamics changes in competition. Journal of Financial Intermediation 12, 57–95.

Berger, A., Dai, Q., Ongena, S., Smith, D., 2003. To what extent will the banking industry be globalized? A study of bank nationality and reach in 20 European nations. Journal of Banking and Finance 27, 1–43.

Berger, A., Klapper, L., Turk-Ariss, R., 2009a. Bank competition and financial stability. Journal of Financial Services Research 35, 99–118.

Berger, A., Molyneux, P., Wilson, J., 2009b. Oxford Handbook of Banking. Oxford University Press, Oxford.

Besanko, D., Thakor, A., 1993. Relationship banking, deposit insurance and bank portfolio. In: Mayer, C., Vives, X. (Eds.), Capital Markets and Financial Intermediation. Cambridge University Press, Cambridge, UK, pp. 292–318.

Boyd, J., De Nicoló, G., 2005. The theory of bank risk taking revisited. Journal of Finance 60, 1329–1343.

Boyd, J., De Nicoló, G., Jalal, A., 2006. Bank risk taking and competition revisited: New theory and evidence. Working Paper WP/06/297, IMF.

Caminal, R., Matutes, C., 2002. Market power and bank failures. International Journal of Industrial Organisation 20, 1341–1361.

Carbó, S., Humphrey, D., Maudos, J., Molyneux, P., 2009. Cross-country comparisons of competition and pricing power in European banking. Journal of International Money and Finance 28, 115–134.

Carletti, E., Hartmann, P., 2003. Competition and financial stability: What’s special about banking? In: Mizen, P. (Ed.), Monetary History, Exchange Rates and Financial Markets: Essays in Honour of Charles Goodhart, vol. 2. Edward Elgar, Cheltenham, UK.

Carletti, E., Vives, X., 2008. Regulation and competition policy in the banking sector. Occasional Paper 159, IESE, Public-Private Sector Research Center.

Casu, B., Girardone, C., 2006. Bank competition, concentration and efficiency in the single European market. The Manchester School 74, 441–468.

Cetorelli, N., Peretto, P., 2000. Oligopoly banking and capital accumulation. Working Paper 2000-12, Federal Reserve Bank of Chicago.

De Guevara, J., Maudos, J., 2007. Explanatory factors of market power in the banking system. The Manchester School 75, 275–296.

De Nicoló, G., 2000. Size, charter value and risk in banking: An international perspective. Board of Governors International Finance Discussion Papers, No. 689.

De Nicoló, G., Bartholomew, P., Zaman, J., Zephirin, M., 2004. Bank consolidation, conglomeration, and internationalization: Trends and implications for financial risk. Financial Markets, Institutions and Instruments 13, 173–217.

DeYoung, R., 2003. The failure of new entrants in commercial banking markets: A split population duration analysis. Review of Financial Economics 12, 7–33.

Delis, M., Tsionas, E., 2009. The joint estimation of bank-level market power and efficiency. Journal of Banking and Finance 33, 1842–1850.

Demsetz, H., 1973. Industry structure, market rivalry and public policy. Journal of Law and Economics 16, 1–9.

Djankov, S., McLiesh, C., Shleifer, A., 2007. Private credit in 129 countries. Journal of Financial Economics 84, 299–329.

Gorton, G., Winton, A., 2003. Financial intermediation. In: Constantinides, G., Harris, M., Stulz, R. (Eds.), Handbook of the Economics of Finance, Corporate Finance. Amsterdam, North Holland.

Greene, W., 2005. Reconsidering heterogeneity in panel data estimators of the stochastic frontier model. Journal of Econometrics 126, 269–303.

Hughes, J., Lang, W., Mester, L., Moon, C., Pagano, M., 2003. Do bankers sacrifice value to build empires? Managerial incentives, industry consolidation, and financial performance. Journal of Banking and Finance 27, 417–447.

International Monetary Fund, 2009. Global Financial Stability Report, April. Keeley, M., 1990. Deposit insurance, risk, and market power in banking. American

Economic Review 80, 1183–1200. Koetter, M., Kolari, J., Spierdijk, L., 2008. Efficient competition? Testing the ‘‘quiet

life” of US banks with adjusted Lerner indices. Working Paper, Groningen University.

Lensink, R., Meesters, A., Naaborg, I., 2008. Bank efficiency and foreign ownership: Do good institutions matter? Journal of Banking and Finance 32, 834–844.

Maudos, J., De Guevara, J., 2007. The cost of market power in the European banking sectors: Social welfare cost vs. cost inefficiency. Journal of Banking and Finance 31, 2103–2125.

Mercieca, S., Schaeck, K., Wolfe, S., 2007. Small European banks: Benefits from diversification. Journal of Banking and Finance 31, 1975–1998.

Panzar, J., Rosse, J., 1987. Testing for ‘Monopoly’ equilibrium. Journal of Industrial Economics 35, 443–456.

Petersen, M., Rajan, R., 1995. The effect of credit market competition on lending relationships. The Quarterly Journal of Economics 110, 407–443.

Rajan, R., Zingales, L., 2003. The great reversals: The politics of financial development in the 20th century. Journal of Financial Economics 69, 5–50.

Schaeck, K., Cihak, M., 2008. How does competition affect efficiency and soundness in banking? New empirical evidence. Working Paper No. 932, European Central Bank.

Schaeck, K., Cihak, M., Wolfe, S., 2009. Are competitive banking systems more stable? Journal of Money, Credit and Banking 41, 711–734.

Sealey, C., Lindley, J., 1977. Inputs, outputs, and theory of production cost at depository financial institutions. Journal of Finance 32, 1251–1266.

Sengupta, R., 2007. Foreign entry and bank competition. Journal of Financial Economics 84, 502–528.

Shaffer, S., 2004. Comments on what drives bank competition: Some international evidence, by Claessens, S., Laeven, L.. Journal of Money, Credit and Banking 36, 585–592.

Uhde, A., Heimeshoff, U., 2009. Consolidation in banking and financial stability in Europe: Empirical evidence. Journal of Banking and Finance 33, 1299–1311.

gifford-how-to-referee.pdf

guidereferee.pdf

Page 1 of 2 8/28/2013

The University of British Columbia Department of Economics

Economics 560: Economics of Labour

Professor Nicole M. Fortin Fall 2013 Tuesday, Thursday: 14:00-15:30 Buchanan B218

Guidelines for your Referee’s Report

In the course of this class, you will be asked to write two (2) referee reports on assigned

papers that are at the “working paper” stage. As you pursue a career as a professional economist, writing referee reports will become part of your usual duties. As a student, it will give you the opportunity to study a paper in great detail, develop critical thinking skills, and learn about the fine craft of economic writing. Your referee reports should be 3-4 pages long and should include at the end a recommendation to the editor (that normally goes in the cover letter) as to whether the paper should be (1) accepted for publication as it stands, (2) accepted, subject to minor revisions, (3) returned to the author for major revisions, a judgement on publication to be made after resubmission, or (4) rejected. Let’s assume that the paper has been directed at a high level field journal such as the Journal of Labor Economics or the Journal of Human Resources of which I am a co-editor. A referee report normally begins with a short summary of the objectives of the paper, and of what the authors have accomplished in the paper. The key questions that you want to answer in this part of the report are: What did the authors view themselves as doing and what did they accomplish? This part should generally be no more than one half-page to one page long, given the editor has also read the paper. It should be quite neutral in tone as if you were recording the information for yourself. Your summary of the paper is a way of establishing your credibility to the authors and the editor, who want to know that you have carefully studied their paper and that you have understood the key points made by the paper and the nature of the contribution to knowledge. You are also providing the authors, and the editors, with an alternative introduction to and a summary of their work.

At this stage, you may also want to place the paper in the literature and may wish to indicate which parts (theoretical development, empirical results, methodology, policy implications) make a (i) very important contribution, or (ii) fairly important contribution or (iii) a not very important contribution to the literature. Note that it is the authors’ responsibility to establish the fact that an important contribution has been made. You may want to peruse a key paper that the authors cite as a source of debate or that provides the motivation for their paper. You may also wish to make a brief comment on the expositional qualities of the paper; that is, is the paper straightforward to read and self-contained, or is the exposition convoluted. You shouldn’t hesitate to make positive comments, if warranted, even if your judgement is ultimately going to be harsh.

The second part of the report, the critical analysis, is the most important one and should be 2 -3 pages long. In this part you discuss the merits of the paper itself and whether the paper does make a contribution to knowledge that is worthy of publication in this prestigious journal. You may want to organize your discussion by going from the big picture to the smaller details.

Page 2 of 2 8/28/2013

1) Overall view: The most important question to answer here is: Did the paper accomplish what it set out to do? Did the authors take the best approach, and were they successful in their approach? You may also want to evaluate the importance and originality of the question. Many papers are correct in their internal logic and fill some gap in the literature, but do not make an important advance in knowledge.

2) Main concerns: If there are some critical problems with the manuscript in that the authors’ analysis is incorrect in some manner then it is important to state these problems clearly in your evaluation. Here is a list of potential problems, luckily only a few might be present in the paper that you are evaluating. It could be that a) the description of the related literature is inappropriate to the actual material in the paper; b) the logical argument is not tight, including incorrect application of economic concepts or erroneous mathematical derivations; c) the econometric tools are being used inappropriately; d) there is only a loose link between the economic model and the empirical analysis; e) the coefficients of interest are not properly identified or not shown to be robust enough; f) the interpretation of empirical results is inappropriate given the available data and econometric strategy used or goes beyond what has actually been proven, g) the conclusions are incorrectly made or expressed; and h) the contribution to the literature is inaccurately described. You can also comment on the structure and organization of the paper and make suggestions as to whether the reader might be better served by alternative ways of presenting and discussing the theory and/or empirical results. Avoid making suggestions that the authors have not hope to being able to follow. If you are suggesting a “revise and resubmit” (options 2 and 3 in the opening paragraph above), you should point out some possible solutions to the problems that you identify.

3) Smaller points These are concerns that are usually easily corrected, but that are important nevertheless. You may want to mention a) areas where the author's line of thought is hard to follow or confusing; b) important mathematical derivations that are obscure; c) spelling and grammatical errors; d) missing data sources and poorly constructed Tables or Figures; e) references to the literature that are missing or incorrect.

how_to_referee.pdf

How  to  Write  a  Referee  Report  

Mar/n  Farnham  

Refereeing  is  essen/al  to  academic   research  

•  Journal  editors  are  the  ul/mate  gatekeepers   when  it  comes  to  publica/on   – Their  job  is  to  decide  what  papers  are  1)  credible;   2)  important;  and  3)  of  interest  to  their  readership  

– With  hundreds  of  papers  submiDed  each  year,   editors  need  help  making  judgments  on  which  get   published  

– Referees  are  unpaid,  specialized  assistants  to   editors  in  this  process  

Refereeing  is  essen/al  to  academic   research  

•  Credibility  is  obviously  very  important   – If  studies  are  published  that  use  flawed  methods   or  falsified  results,  the  credibility  of  the  en/re   field  suffers  

– Good  gatekeeping  by  an  editor  and  his/her   referees  benefits  all  of  us,  even  if  we  don’t  publish   in  that  journal  

– Bad  gatekeeping  reflects  badly  on  all  of  us    

Refereeing  is  essen/al  to  academic   research  

•  Since  publica/on  is  a  signal  of  quality,   refereeing  allows  us  to  beDer  select  our   reading  material  (i.e.,  it  saves  us  /me)   – Good  editors  (and  referees)  pick  up  good  new   work  by  young  unknown  scholars  that  we  might   not  otherwise  know  was  there  

– Without  expert  quality  control,  we’d  just  sit   around  reading  the  same  old  people  with  good   reputa/ons   •  Because  that’s  the  only  signal  of  quality  we’d  receive  

How  does  it  work?  

•  Your  first  referee  report  may  be  given  to  you   by  your  advisor   – It’s  good  training  for  you   – It’s  a  good  way  for  your  advisor  to  see  how  you   think  and  what  you  know  

– It’s  also  a  good  way  for  your  advisor  to  make  it   home  in  /me  for  dinner  

•  As  you  get  known  in  the  field,  editors  will   begin  to  send  you  requests  to  review   manuscripts  

Handling  requests  to  review  

•  If  asked  to  review  a  manuscript,  your   obliga/on  is  to   – Referee  the  work  if  you  are  qualified   – Tell  the  editor  if  you  are  not  qualified  to  referee   the  work  

– Suggest  appropriate  alterna/ve  reviewers  if  you   can’t  do  the  review  

– Agree  to  a  /me  by  which  you  will  complete  the   review  

– Complete  a  substan/ve  and  helpful  review  on   /me  

Is  it  ever  OK  to  say  “No!”?  

•  At  this  point  in  your  career,  only  if  you’re  not   qualified  to  comment  or  if  there’s  a  conflict  of   interest  

•  Later  on,  you  may  find  yourself  becoming   popular  with  editors   – Kees,  Daniel,  others  get  swamped  with  requests   – If  demand  for  your  services  gets  too  high,  you  can   start  refusing;  but  that’s  years  off  at  this  point  

– Then  you’d  start  refusing  to  referee  for  journals   that  are  of  less  interest  to  you  (and  that  you   intend  never  to  publish  in)  

Is  it  ever  OK  to  say  “No!”?  

•  Keep  in  mind  that  if  you  refuse  to  referee  for  a   journal  (more  than  once  or  twice)  you  may   begin  to  reduce  your  probability  of  gebng   published  at  that  journal.  

•  Some  journals  require  you  to  referee  if  you   submit  to  them   – Berkeley  Electronic  Press  requires  that  you  do  two   referee  reports  for  every  paper  you  submit  

– Or  else  pay  them  ~$250  per  submission.  

What  does  a  referee  do?  

•  Your  job  is  twofold.    You  1)  advise  the  editor;   and  2)  advise  the  author(s)   – Ul/mately,  your  job  is  to  tell  the  editor  whether  or   not  they  should  publish  it  and  why  or  why  not  

– But  along  the  way  you  do  a  valuable  assessment   of  the  paper  which  you  are  expected  to  share  with   the  authors  

– Generally,  a  referee  writes  two  documents   •  Comments  for  the  authors   •  A  review  and  recommenda/on  for  the  editor   •  There  can  be  substan/al  overlap  between  these  docs  

Organiza/on  of  the  report  to  the  editor  

•  There’s  no  one  way  to  do  it  but  the  following   is  how  I  tend  to  do  it  

1)  Summary  of  the  work   2)  Recommenda/on  to  the  editor  (publish,  

revise,  or  reject)   3)  Discussion  of  the  importance  of  the  work  and  

how  it  contributes  to  knowledge  

4)  Methodological  issues  (if  any)   5)  Sugges/ons  for  revision  (if  any)  

Summary  of  the  work  

•  The  editor  has  hopefully  read  the  paper,  but  it   may  have  been  awhile  

•  The  summary  reminds  them  what  the  main   findings  and  methods  of  the  paper  are   – Summarizing  the  work  is  also  useful  for  authors,   because  if  your  summary  doesn’t  match  their  idea   of  what  the  paper  is  about,  they’ll  know  they  need   to  communicate  beDer!  

Recommenda/on  to  the  editor  

•  Here  you  just  tell  the  editor  what  you  think   they  should  do  with  the  paper  

•  This  is  hard!   •  You  need  to  make  an  objec/ve  assessment   based  on     – The  quality  and  content  of  the  paper   – The  quality  and  typical  content  of  the  journal  (i.e.   does  the  paper  match?)  

– I’ve  rejected  a  paper  at  the  AER  that  I  then   accepted  at  the  JPubE  (with  virtually  no  changes)  

Discussion  of  importance  and  contribu/on  

•  This  can  be  easy  if  the  authors  have  done  a   good  lit  review  and  mo/va/on  of  the  paper   – However,  if  you  don’t  agree  with  their  assessment   of  how  the  paper  fills  a  gap  in  the  literature,  you   need  to  explain  why  you  don’t  agree   •  Don’t  take  what  the  authors  say  at  face  value.    Think   cri/cally  about  whether  their  paper  is  really  as  big  a   contribu/on  as  they  say  it  is  

– If  they  lack  good  mo/va/on  but  you  see  the  paper   as  filling  an  important  gap,  you  should  tell  the   editor—you  should  also  tell  the  author  how  to   beDer  mo/vate  their  paper  

Methodological  issues  

•  This  is  usually  the  meat  of  the  report   •  We  all  come  at  research  from  different  angles   – There  are  many  ways  to  skin  a  cat   – You  will  (hopefully)  have  ques/ons,  concerns,   sugges/ons,  etc.  about  the  methods  used  by  the   authors   •  Have  they  dealt  with  a  key  endogeneity  problem?   •  Have  they  made  an  unrealis/c  and  unnecessary   assump/on?  

•  Would  the  model  benefit  from  certain  changes   •  Is  the  dataset  up  to  the  task?  

Methodological  issues  

•  I  olen  ask  ques/ons  of  authors   – Did  you  try  this?   – Might  this  alterna/ve  hypothesis  explain  your   findings?  

– Can  I  see  evidence  in  support  of  this  assump/on?   – Can  you  do  the  following  test?  

•  Some/mes  I  ask  that  certain  things  be  added   to  the  paper  (or  dropped)  

Sugges/on  for  revision  

•  Referees  will  typically  suggest  things  that   ought  to  be  done  in  order  to  make  the  paper   acceptable  for  publica/on   – You  can  do  this  even  if  you’re  rejec/ng  the  paper   – Consider  the  author’s  next  submission.  This  could   be  useful  informa/on  

– If  you  plan  to  reject  the  paper  don’t  tell  the   author,  “I’d  recommend  this  for  publica/on  by  this   journal  if  you  did  the  following...”  

Notes  on  Wri/ng  

•  I  think  most  people  don’t  comment  much  on   wri/ng,  but  I  tend  to   – If  the  paper’s  badly  wriDen  I  tell  them  and  tell   them  why  and  how  it  could  be  improved  

– It’s  kind  of  embarrassing  to  do  this,  but  some/mes   it  needs  to  be  done  

– The  trick  here  is  to  try  to  be  nice.    It’s  hard  when   you’ve  spent  hours  trying  to  figure  out  what  the   hell  the  authors  are  trying  to  communicate  to  you.  

How  blind  is  the  process?  

•  It  used  to  be  double-­‐blind  and  some  journals   s/ll  try  to  do  this   – The  internet  makes  one  side  of  the  double-­‐blind   approach  virtually  impossible  

– If  you  get  a  manuscript  with  no  names  on  it,  but   it’s  posted  somewhere  online,  you’ll  easily  iden/fy   the  authors  

•  Most  journals  now  do  single-­‐blind  refereeing   – You  know  the  authors  but  they  (in  principle)  don’t   know  you  

Don’t  Google  authors  before  wri/ng  the   report  

•  Try  to  write  an  objec/ve  review  without   considering  who  the  authors  are   – If  you  find  they’re  grad  students  you  might  treat   them  differently  from  a  senior  prof  at  Harvard.      

– The  point  is  for  every  manuscript  to  get  a  fair   review  

– If  you  must,  Google  the  authors  once  you’re  done  

Things  to  avoid   •  Being  nasty.     – Small,  insecure  people  some/mes  write  nasty   reports.  They  can  be  devasta/ng  to  authors.  

– Pretend  the  authors  are  friends  of  yours.    Be   honest,  but  think  about  their  feelings.  

•  Telling  the  authors  to  write  a  new  paper   – Some  referees  love  to  tackle  the  paper  as  their   own  project  and  totally  revise  it  (or,  tell  the   authors  to  totally  revise  it)  

– Don’t  write  this  paper!    Write  that  paper  instead!   – Render  a  verdict  on  the  paper,  sugges/ng  modest   revisions  if  necessary.  

Things  to  Avoid  

•  Gratuitous  self-­‐cita/on   – If  you  really  want  the  authors  to  know  who  you   are,  this  will  give  you  away.  

– The  editor  is  unlikely  to  be  impressed.   – Certainly  cite  yourself  if  the  paper  would  benefit   from  the  authors  reading  your  work  

•  Going  into  too  much  detail  for  the  editor   – The  editor  is  busy.  Make  the  editor’s  report  one   page  or  less;  highlight  key  issues.    Make  a  strong   case  for  the  posi/on  you  take    

Things  to  Avoid  

•  Demanding  perfec/on   – The  authors  are  limited  to  about  20  pages  to  make   whatever  case  they’re  trying  to  make  

– Their  case  won’t  be  100%  water/ght   •  Models  could  be  tweaked   •  Alterna/ve  empirical  specifica/ons  could  be  tried   •  Other  datasets  could  be  explored  

– Set  a  reasonable  standard.    You  don’t  have  to  be   100%  convinced  by  what  they’ve  done.    You   should  find  their  argument  compelling,  but  it   doesn’t  need  to  be  water/ght.  

Things  to  Avoid  

•  Doing  it  at  the  last  minute   – The  best  report  will  be  one  that  involved  stewing   over  the  paper  for  some  /me.  

– Read  the  paper  soon  aler  you  get  it  from  the   editor.    Then  sit  on  it   •  Open  a  file  where  you  keep  notes   •  Write  down  thoughts  as  they  come  up  

– Reread  the  paper  again  later  when  you  go  to  do   the  review  in  earnest.    The  stewing  /me  will  pay   off.  

Keep  in  mind  that  you  can  benefit  from  this  

•  You’ll  learn  things  to  try  and  things  to  avoid  in   your  own  work  

•  You’ll  make  an  editor  happy  if  you  do  a  good   job  

•  You  may  even  pick  up  a  research  idea  along   the  way   – Naturally  you  can’t  poach  ideas  from  the   manuscript,  without  ci/ng  them  

– But  you’re  en/tled  to  new  research  ideas  that   stem  from  reading  it  

Instructions for Referee Report.pdf

Instructions for Referee Report Graduate students are required to write a referee report. A referee report is conducted for articles that are submitted to journals. Referees read the submitted article and analyze it for any mistakes and make suggestions on how to improve the paper. Referees write two essays. The first is a point by point analysis of the paper to be submitted to the authors of the article. The second is a confidential letter to the editor where the referee makes a recommendation regarding rejecting, accepting or allowing the review process of the article to continue. Important dates: November 4, 2011 – Submit to me a copy of an article that you would like to review. The article can be empirical or theoretical. It must cover a topic on either: welfare analysis with externalities or public goods, benefit cost analysis with environmental quality, valuation application or pollution regulation. Choose an article from a reputable journal. I suggest focusing on field journals in the natural resource economics, environmental economics or agricultural economics fields. November 7, 2011 – I will tell you if the article you propose to review is acceptable or not. December 2, 2011 – Submit a maximum three page referee report and one page letter to the editor to me. The letter to the editor needs to contain the following elements: (1) summary of the article in your own words (not the words of the authors); (2) where the article fits in the literature; (3) important comments regarding the article and (4) recommendation (accept, reject or allow for a revise and resubmit). The referee report must contain: (1) a summary of the article in your own words (not the words of the authors); (2) general comments (theoretical, empirical or conceptual comments, corrections or suggestions that make a significant improvement to the paper) and (3) specific comments (minor comments or corrections – grammatical or presentation corrections). I am attaching a sample letter and report below.

Letter to Editor: This article develops a model of gear regulation in coastal fisheries, specifically the use of light attraction, to derive the determinants of non-compliance of regulation. The theoretical model is used to estimate an empirical model of regulation violation in coastal fisheries in Ghana. Results show that the probability of detection, social pressure and personal characteristics of fishermen are significant determinants in violating regulations. The main contribution of this paper is in the empirical estimation of the determinants of violation in coastal fisheries. However, there are three significant flaws with regard to the theory, empirical estimation and connection between both in the paper. First, the paper attempts to develop a theoretical model that extends the original model of Akapalu (2008) by adding dynamics, an effort variable and a variable measuring the severity of violating the existing regulations. The dynamics is not added correctly since the fish stock evolution is not captured. The only dynamic component seems to deal with the time that a fisherman is caught violating the regulation but this can be simplified to a simple static problem. The effort variable is treated as an exogenous variable when in fact it should be an endogenous variable which implies that equation (1) is not really a profit function (it is a profit equation) and equation (3) is not a value function. One critical assumption in the model (which is not well explained) is that the severity of violations is proxied by the replacement cost of light attracting equipments which does not make sense intuitively. If large light gears that can be purchased in bulk are cheaper than small light gears purchased individually then the cost of light replacement does not reflect severity of the violation. Second, given the problems with the theory, the empirical estimation seems ad hoc with the sudden inclusion of a vector of socio-economic variables determining perception. Given the assumptions of the model, a logit and maximum entropy estimation technique should not be used. Instead, with the dynamic assumption that the probability of detection is increasing over time a duration model would be more appropriate to fit the theory. Lastly, the empirical specification and data does not fit with the theory. The theory relies on dynamic choices by the fishermen but the data and estimation is cross sectional. The paper itself is not written well and can be confusing from time to time, but more importantly, it missed important key points in the analysis. I believe this article is more suited to a journal focused on fishery management. I recommend this paper for rejection.

Letter to the author: This article develops a model of gear regulation in coastal fisheries, specifically the use of light attraction, to derive the determinants of non-compliance of regulation. The theoretical model is used to estimate an empirical model of regulation violation in coastal fisheries in Ghana. Results show that the probability of detection, social pressure and personal characteristics of fishermen are significant determinants in violating regulations. There are a number of issues regarding the theory, empirical estimation procedure and connection between both that the author needs to examine carefully. General Comments

1. First, the paper attempts to develop a theoretical model that extends the original model of Akapalu (2008) by adding dynamics, an effort variable and a variable measuring the severity of violating the existing regulations. The dynamics is not added correctly since the fish stock evolution is not captured. Here, X should follow some growth formulation (for example, logistic growth). The only dynamic component seems to deal with the time that a fisherman is caught violating the regulation. If this is the case, then the problem can be simplified to a simple static formula. The effort variable is treated as an exogenous variable when in fact it should be an endogenous variable which implies that equation (1) is not really a profit function (it is a profit equation) and equation (3) is not a value function. One critical assumption in the model (which is not well explained) is that the severity of violations is proxied by the replacement cost of light attracting equipments which does not make sense intuitively. For example, if large light gears that can be purchased in bulk are cheaper than small light gears purchased individually then the cost of light replacement does not reflect severity of violation.

2. Second, given the problems with the theory, the empirical estimation seems ad hoc with the sudden inclusion of a vector of variables determining perception which turns out to be one of the main focuses of the study. It may be helpful to provide a more in-depth conceptual discussion of the relationship between the perception variable and socio-economic variables of the individual.

3. There is a long discussion of the background of MELE but the only justification of its use is the presence of multicollinearity which is not presented in the paper. First, I do not think cost of light replacement captures the severity of the violation. Second, I am not fully convinced that there are any advantages to using MELE over OLS or GLS. The author may want to expound on showing that MELE is the appropriate estimation procedure.

4. The empirical results are not tested for robustness. Given the assumptions of the model, a logit and maximum entropy estimation technique should not be used. Instead, given the dynamic assumptions that the probability of detection is increasing over time a duration model would be more appropriate to fit the theory.

5. Lastly, the empirical specification and data does not fit with the theory. The theory relies on dynamic choices by the fishermen but the data and estimation is

cross sectional. Given the available data, I would recommend specifying a static model instead.

Specific Comments

1. Page 5, equation 1 – Effort is a choice variable by the fisherman not an exogenous variable. Thus, this is not a profit function but a profit equation.

2. Page 5, first line – One critical assumption is that the catchability coefficient is increasing in the cost of light attraction equipment which is not very plausible. The author would probably need to find a source or expound on this idea to justify the assumption.

3. Page 5, below equation 2 – did you mean “z(.) incorporates the disutility…” and not alpha?

4. Page 5, last paragraph – The expression for the fine can be simplified as fi = q(l)f_i instead of two separate components for Fi. Here, q can be increasing in l to remove the second component of the formulation of the fine.

5. Page 6, first paragraph- The idea that there is a probability distribution of detection which is a function of time does not make sense intuitively. What is the justification for this?

6. Page 6, equations (3) and (4) – These are not value functions since E is a choice variable. If it is dynamic, then X is also changing over time.

7. Page 6 middle of page – There are two definitions for pi(li). Which is the correct one?

RCER451.pdf

ref_report_help.pdf

EC2410 Urban Economics How to write a referee report

Matthew A. Turner Brown University

Spring 2015

As part of the course you are asked to write three referee reports. The object of these assignments is fourfold: to describe what the paper does; to describe whether and why the paper is important or interesting; to point out and describe problems in the paper; and finally, to recommend whether the paper should be published in its current form, whether the editor should ask for a revision, or whether the paper should be rejected.

You can say just about anything you want as long as you can back it up and you are polite about it.

The object of a referee report is to partly offer the editor your opinion on the paper and to partly save the editor from reading the paper closely. Given this, it is important that the report be shorter than the paper. For this class, don’t go above two pages if you can possibly help it. For real reports, try to keep them short, but longer reports can be OK for good, important papers that can still be improved in lots of ways.

Following is an outline that will work for most papers:

1. Describe what the paper does. If it is an empirical paper describe the estimating equation and data. If theory, sketch the model.

2. Describe what the paper finds.

3. Explain why the paper is important (or not).

4. Describe any problems with the paper that should be fixed before it is published, or that should disqualify the paper for publication. If it is an empirical paper, you will want to consider/address each of the following topics:

• Explanation of the identification strategy.

• Explanation of the Data and results.

• Discussion of Internal Validity.

• Discussion of External Validity.

1

referee.pdf

McDaniel College Prof. McIntyre WRITING A REFEREE REPORT Referee reports help journal editors determine whether or not a submitted paper is of “publication quality.” They are anonymous and are designed to help authors improve both the content and organization of their papers. A referee report typically has three sections. This first section is a brief summary of the paper you are refereeing. It should include what the authors are trying to show and what results they arrive at. The second section is the most important. This should include substantive comments on the paper. These are usually the “big criticisms” you have of the paper. You should point out any mistakes that you see and more importantly, areas that could be further developed or explained. It should be noted, however, that positive comments should be made. If the author makes a clever or interesting point it should be pointed out, especially if that point has implications beyond the scope of the paper itself. The third section is devoted to minor issues. Here you can point out typos, think-o’s, make suggestions regarding organization or for further references and praise minor points. (For the record, another purpose of the third section is to suggest where authors could trim the length of their papers. Most first manuscripts submitted to journals are considerably longer than what actually gets published, and authors are strongly encouraged to make their paper as short as possible throughout the editorial process. Given that I have mandated a page count for this project, I do not expect you to make suggestions for cuts unless you feel very strongly about something.) The good news for you is that referee reports are fairly short (about a page or so). This is almost surely a function of the fact that referees are not typically paid for their services. When you write your referee report, please remember to submit two (2) copies to me. I will keep one for grading and give one to the author of the paper you are refereeing. DO NOT put your name on the report itself. Put it on a title page that I can tear off. Remember, this is meant to be a double-blind process. The following page is an actual referee report I wrote. This will hopefully give you and idea of the style of these things. I have x’ed out the paper title and journal name to protect the author(s). You should note that I was pretty rough on the authors (I ultimately suggested that the paper be rejected) and that I do not want or expect you to try and be this harsh with your classmates. The idea here is to read and think critically about someone else’s research. IT IS NOT the goal of this assignment to pick your classmates’ papers apart, although that may end up happening.

Manuscript: xxxx Journal: xxxx Summary: Summary This paper develops and explores a series of empirical models of consumer confidence in Brazil. The author suggests the in addition to the expected fundamental drivers of consumer confidence and household spending, monetary policy credibility plays a role in determining confidence. Specific comments:

1. In general, my impression is that the author’s familiarity with the existing literature on consumer confidence is limited.

a. The theory behind the explanatory consumer confidence models could (and is) just as easily be used in models of household consumption. Accordingly, the author needs to explicitly state the relationship between consumer confidence and household consumption. If consumer confidence is just a coincident indicator for the economy and consumption, this should be a consumption paper. If there is something in consumer spending above and beyond what is found in consumption (predictive power, etc.), that needs to be spelled out.

b. The authors exaggerate the importance of consumer confidence for overall macroeconomic performance. Consumer confidence is not some sort of target variable for policymakers. The majority of existing literature on consumer confidence suggests that to a great extent, consumer confidence is primarily a coincident indicator, not a macroeconomic driver. This may be different in Brazil, and if so the author needs to clearly point out and show why/how this is the case.

c. The existing literature on consumer confidence suggests may have a small degree of predictive power for consumption spending. And in the U.S., the predictive power of consumer confidence is very sensitive to the measure of consumer confidence used.

2. The use of some measure of central bank credibility as a driver of consumer confidence is interesting, but for consumer confidence the issue here is price level and interest rate stability. The “CRED” variable is a proxy for these things. If this is not the author’s intention, this needs to be specifically explained.

3. The OLS and GMM results are similar to the point of redundancy. 4. Confidence intervals around the VAR impulse response functions would be very helpful.

It is unclear that any of the results are significant. Minor comments:

1. Figure 1 is largely unnecessary. 2. A more thorough description of Brazil’s CCI would be helpful.