Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil

The cyclicality of loan loss provisions under three different accounting models: the United... ISSN 1808-057X DOI: 10.1590/1808-057x201804490 O A e cyclicality of loan loss provisions under three di erent accounting models: the United Kingdom, Spain, and Brazil Antônio Maria Henri Beyle de Araújo Universidade Católica de Brasília, Departamento de Ciências Contábeis, Brasília, DF, Brazil Email: henri.beyle@uol.com.br Paulo Roberto Barbosa Lustosa Universidade de Brasília, Faculdade de Economia, Administração, Contabilidade e Gestão de Políticas Públicas, Departamento de Ciências Contábeis e Atuariais, Brasília, DF, Brazil Email: lustosa@unb.br Edilson Paulo Universidade Federal da Paraíba, Centro de Ciências Sociais Aplicadas, Departamento de Finanças e Contabilidade, João Pessoa, PB, Brazil Email: e.paulo@uol.com.br nd Received on 10.11.2016 – Desk acceptance on 11.07.2016 – 2 version approved on 08.08.2017 – Ahead of print on 11.06.2017 ABSTRACT A controversy involving loan loss provisions in banks concerns their relationship with the business cycle. While international accounting standards for recognizing provisions (incurred loss model) would presumably be pro-cyclical, accentuating the e ects of the current economic cycle, an alternative model, the expected loss model, has countercyclical characteristics, acting as a bu er against economic imbalances caused by expansionary or contractionary phases in the economy. In Brazil, a mixed accounting model exists, whose behavior is not known to be pro-cyclical or countercyclical.  e aim of this research is to analyze the behavior of these accounting models in relation to the business cycle, using an econometric model consisting of  nancial and macroeconomic variables.  e study allowed us to identify the impact of credit risk behavior, earnings management, capital management, Gross Domestic Product (GDP) behavior, and the behavior of the unemployment rate on provisions in countries that use di erent accounting models. Data from commercial banks in the United Kingdom (incurred loss), in Spain (expected loss), and in Brazil (mixed model) were used, covering the period from 2001 to 2012. Despite the accounting models of the three countries being formed by very di erent rules regarding possible e ects on the business cycles, the results revealed a pro-cyclical behavior of provisions in each country, indicating that when GDP grows, provisions tend to fall and vice versa. e results also revealed other factors in uencing the behavior of loan loss provisions, such as earning management. Keywords: provisions, cyclicality, economic cycles, credit operations, commercial banks. Correspondence address: Antônio Maria Henri Beyle de Araújo Universidade Católica de Brasília, Departamento de Ciências Contábeis SQSW 306, Bloco G, Apartamento 204 – CEP: 70673-437 Sudoeste – Brasília – DF – Brazil R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 97 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil 1. INTRODUCTION According to Longbrake and Rossi (2011), many of  e accounting principles in the United States and the rules that regulate the  nancial markets contributed those of the International Accounting Standards Board to deepening the crisis that a ected the global economy (IASB) adopt the incurred loss model, establishing that in the period between 2007 and 2009, fueling the debate a loss in credit operations will only be recognized in regarding the way the  nancial system accentuates the earnings if its occurrence (related to a loss event) is likely e ects of expansionary and contractionary phases in and its value can be estimated with certainty. the economy. At the heart of the claim lies the concern According to Betancourt and Baril (2009), an regarding pro-cyclicality, which represents a positive alternative to the incurred loss model is the expected correlation between a particular variable and economic loss model, which recognizes provisions based on activity (Bebczuk & Sangiácomo, 2010). expected future losses, independently of objective For Harrald and Sandal (2010), pro-cyclicality is evidence. This model uses as a reference the dynamic the way that the banking system interacts with the real provision adopted by the Banco de España, which economy and accentuates the e ects of an economic accepts the setting of a generic provision to protect cycle. It derives from incentives to accelerate lending banks against possible additional losses in a particular in periods of economic expansion and to reduce such economic cycle. activity in times of crisis. In Brazil, Resolution n. 2,682/1999 of the National According to Longbrake and Rossi (2011), the  nancial Monetary Council rules that the provision should be set system should ideally dampen and not amplify business based on the operation risk level and its value should be cycles. Pro-cyclicality becomes especially problematic reviewed at least monthly, depending on the delay in the when it accentuates the e ect of a fall in the economic payment of principal or interest.  is involves a mixed cycle and worsens a crisis.  us, countercyclical rules, system with characteristics of expected loss (provisions which reduce economic imbalances and the depth of based on future losses expectations) and incurred loss business cycles, are welcome. (provisions based on objective evidence represented by Also according to these authors, factors that explain delays in payment). the in uence of the  nancial system on the economic The three accounting models would cause different environment include the accounting rules that set loan loss behaviors of the provision in relation to economic provisions.  e expression “loan loss provision” (“provisão cycles. In order to confirm this assumption, the para créditos de liquidação duvidosa” in Portuguese) is the relationship between the accounting criteria for setting term used by the Brazilian Central Bank for the estimated provisions in banks and economic cycles should be losses from loans granted by  nancial institutions. In investigated. international accounting standards, this provision is Studies such as those by Bikker and Metzemakers represented by impairment losses derived from loans (2004) and by Boutavier and Lepetit (2007) have already and receivables. been conducted to investigate the relationship between For Bikker and Metzemakers (2004), the provision loan loss provisions and economic cycles. Besides being leaves room for subjective judgment and allows for restricted to certain aspects of the provision, these studies discretion when establishing levels considered as used relatively di erent models and variables, thus leaving “adequate”. The study by these authors identified a room for new analyses on the topic. direct relationship between the provision set by banks ere are few studies involving the Brazilian model, and economic cycles, revealing that income smoothing, which is presumably countercyclical since most of capital management, and tax rules also determine the the value of the provisions is the result of classifying level of provision. operations by level of risk at the time of lending and Betancourt and Baril (2009) observe that at the not migration between the risk categories due to credit beginning of a downward economic cycle there is little defaults. provision recognized in banks earnings. As the crisis e aim is to analyze whether the accounting models worsens the provisions grow, deteriorating the equity involved in setting loan loss provisions in banks are pro- situation of banks and reducing levels of lending when the market needs funds the most.  is pro-cyclical behavior cyclical, countercyclical, or acyclical, using an econometric is adopted by banks that follow international accounting model with a consistent theoretical framework and standards for provisions. empirical construction. 98 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo A panel data econometric model was specified in three countries, signaling that the empirically con rmed which the loan loss provision expense in relation to pro-cyclical behavior is independent of the provision total assets is regressed against the variation in Gross model adopted. One probable assumption regarding Domestic Product (GDP), under the control of the this  nding is that the banks may be using discretion unemployment rate and accounting variables that affect and engaging in earnings management via the provision the provision. An identical specification is run for and more than compensating for the expected e ect of samples from three different provision accounting the models, but this is a hypothesis to be investigated in systems (United Kingdom, Spain, and Brazil) with future research. the aim of verifying whether the covariance between Besides this introduction, this paper presents the the provision and GDP is significantly positive literature review in section 2; section 3 formulates (countercyclical), negative (pro-cyclical), or without the three research hypotheses; section 4 details the significance (acyclical). methodology; section 5 presents, analyzes, and The results indicated a negative and significant discusses the results; and section 6 offers some final association between the provision expense and GDP in the remarks. 2. CONSIDERATIONS REGARDING ECONOMIC CYCLES AND PRO-CYCLICALITY A classic de nition of “economic cycle”, or “business e ects are damaging to the economy. Pro-cyclicality derives cycle”, is that of Burns and Mitchell (1946): from incentives for the acceleration of lending in periods of expansion and the reduction in such activity in crises. Business cycles are a type of  uctuation found in the aggregate According to Bebczuk et al. (2010), pro-cyclicality is economic activity of nations that organize their work mainly in a synchronized movement in the growth of credit and business enterprises: a cycle consists of expansions occurring production, both in time t. Excessive cyclicality in lending at about the same time in many economic activities, followed can lead to problems such as exacerbating the economic by similarly general recessions, contractions, and revivals cycle, an increase in systemic risk, and the inadequate which merge into the expansion phase of the next cycle; in duration business cycles vary from more than one year to allocation of resources for loans. ten or twelve years. For Gonzales (2009), pro-cyclicality is present when the actions of banks tend to reinforce the movements of For Harrald and Sandal (2010), pro-cyclicality is the the underlying business cycles, indicating a correlation way in which the banking system interacts with the real between the aforementioned actions and the current economy, accentuating the e ects of an economic cycle. economic environment. Pro-cyclical measures are, e pro-cyclicality cost depends on the extent to which the therefore, those that contribute to strengthening the e ects of the cycle are accentuated and the fact that these current economic cycle. 3. LOAN LOSS PROVISIONS IN BANKS Hendriksen and Van Breda (1999) state that an 3.1. Provision Models important element in evaluating receivables is the Banks’ provision rules can be: (i) backward-looking, treatment of uncertainty with regards to their payment based on losses incurred in operations and considering by the client.  us, revenue should be measured by the the events and conditions prior to the balance sheet; and value that it is expected will be received. In the quality of (ii) forward-looking, set by considering expected losses the value reserve, the provision adjusts the gross value of and based on trend analyses (Bouvatier & Lepetit, 2012). loans according to their credit risk.  is adjustment has According to Cortavarria, Dziobek, Kanaya, and Song a direct e ect on the income calculation, a ecting the amount to be paid out to shareholders. (2000), provisions can be generic and speci c. Generic R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 99 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil provisions are possible or latent losses not yet identi ed management being practiced by  nancial institutions in Brazil, based on the use of loan loss provisions. Macedo (ex-ante) and applicable to all operations. Specific and Kelly (2016) identi ed that from 2006 to 2012 the provisions re ect individually identi ed losses related income level obtained by  nancial institutions in uenced to an observable event (ex-post). these provisions. Bortoluzzo, Sheng, and Gomes (2016) found similar results by analyzing the behavior of 123 3.2. Earnings Management via Provisions Brazilian banks between 2001 and 2012. Silva (2016) Accounting rules allow for the use of discretion showed that between June 2009 and December 2014 and judgment in the  nancial statements preparation Brazilian and Luso-Spanish banking institutions used (Cohen & Zarowin, 2007).  is enables the use of business provision expenses to manage earnings. Cursio and knowledge and expertise to choose the most appropriate Hasan (2015) found that from 1996 to 2006 earnings accounting methods, estimates, and disclosure.  e management was shown to be an important determinant problem is when managers  nd incentives to convey of the provisions of  nancial intermediaries that operate information to their own bene t, choosing accounting in the Euro zone. methods and accounting estimates that do not adequately re ect the underlying economic conditions. 3.3. Capital Management via Provisions For Cohen and Zarowin (2007), this accounting In the capital management hypothesis, banks set higher information discretion can be both a value maximizer and provisions when their regulatory capital is relatively low. opportunistic, leaving room for “earnings management”. According to Bikker and Metzemakers (2004), this occurs Earnings management can be achieved via the because the Basel Accord allows for the accumulated manipulation of accruals, which according to Martinez provisions in the balance sheet, although limited to a (2001, p. 29) represent “the di erence between net income percentage of risk-weighted assets, to comprise banks’ and net cash  ow”, or via real transactions, such as a regulatory capital.  is allowance occurs only in relation reduction in spending on research and development. to level 2 capital, which is the lowest quality regulatory In the case of provision, earnings management occurs capital and whose value cannot exceed the value of level through accruals manipulation. For Martinez (2001), 1 capital, taken as main capital. when it is le to managers to make accounting choices, they may not limit themselves to accounting facts alone 3.4. Studies Relating the Provision to Economic and use exogenous variables, which will also in uence Cycles the decision.  us, provision will also consider aspects that extrapolate credit quality. Table 1 lists studies involving the pro-cyclicality of the provision in banks. Studies have revealed indications of earnings 100 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 1 Summary of the main articles involving the analysis of the pro-cyclicality of the loan loss provision in  nancial institutions. Author/Period Methodology Results Estimation of an equation in which the provision Strong evidence that the relationship between the is a function of: (i) speci c variables of the provision and the earnings of banks from the G-10 Cavallo and Majnoni banks, (ii) macroeconomic variables, and (iii) countries presents a positive sign. The same sign is Period: 1988 to 1999 institutional variables, with the use of three different negative for banks from countries outside the G-10, which Publication: 2001 techniques (OLS cross section, pooled cross on average present much smaller provisions in good section, and panel  xed effect estimation). periods and are forced to increase them in bad periods. To identify the relationship between the provision Increases in the provision depend a lot on the business and the business cycle, three macroeconomic Bikker and Hu cycle. In bad times, the provisions increase. Banks variables (real GPD growth, unemployment, and Period: 1979 to 1999 contribute signi cantly more to the provisions in in ation) and three speci c variables of the banking Publication: 2002 years of relatively higher earnings (as a precaution sector (loans, net  nancial intermediation earnings, or as a way of smoothing income), leading the and defaults) were used as explanatory variables sector to be less pro-cyclical than it should. for the provision in the linear regression. In the linear regression, the provisions are a result of Many banks delay the recognition of provisions as Laeven and Majnoni the earnings before income tax and of the provisions much as possible, recording them when an economic Período: 1988 to 1999 for losses; of the real growth in loans; of the real contraction has just taken hold and thus increasing the Publication: 2003 growth in GDP per capita; and of the year dummies. impact of economic cycles on earnings and on capital. The provisions of banks are usually substantially greater in Based on the models by Cavallo and Majnoni (2002) periods of lower GDP growth, re ecting the growing risk of Bikker and Metzemakers and by Laeven and Majnoni (2003), the authors their credit portfolios when the economic cycle changes and Period: 1991 to 2001 worked with additional variables such as GDP enters into a downturn. This effect is mitigated by an increase Publication: 2004 growth and dummies for the countries. The use of the in the provision in periods of higher earnings and in those in provision for capital management was also tested. which the growth in the volume of lending becomes greater. There is a positive correlation between the provision and The model is based on the assumption that the GDP. The empirical tests do not support the pro-cyclicality of provision (dependent variable) is a linear function of the provision in banks. The results were different depending Handorf and Zhu two variables: (i) the initial value of the accumulated on the size of the institutions. Medium-sized banks tend Period: 1990 to 2000 provision, net of the falls or losses ex-post of the current to use information on projected losses when de ning the Publication: 2006 period; and of management’s expectations in relation value of the provision (countercyclical attitude). Smaller to future falls, based on current available information. banks and much bigger banks tend towards provision practices that consider current losses (pro-cyclical attitude). The model evaluates whether the evolution of the Setting provisions to cover expected future losses in provision explains the changes in banks’ behavior in loans (“non-discretionary provisions”) causes major relation to lending during the economic cycle. The uctuations in credit. The non-discretionary component linear regression model estimated the discretionary Bouvatier and Lepetit of the provision ampli es the credit cycle: in an economy and non-discretionary components of the provision. Period: 1992 to 2004 rising phase, banks tend to underestimate credit risks, The authors introduced the lagged dependent variable Publication: 2007 indicating the existence of greater incentives for the as an explanatory variable, in order to consider granting of new loans since the costs of lending are a dynamic adjustment in the provision. Capital undervalued. Provisions with managerial purposes management, earnings management, and the signaling (discretionary provisions) do not produce the same effect. of equity robustness are also important variables. The effects of business cycles on the performance of While economic growth is the main driver in the banks’ loans portfolios in developing countries were performance of banks’ credit portfolios, the effects caused by Glen and analyzed using linear and non-linear models. The the interest rates are second order. The relationship between Mondrágon-Vélez provision was the proxy for the portfolio’s performance. the provisions and economic growth is highly linear only Period: 1996 and 2008 GDP and other macroeconomic variables were considered in economic conditions of extreme stress. Greater loss Publication: 2011 as explanatory variables. The interest rate on loans and provisions are related to the private sector leverage level, a set of variables with the individual characteristics to the bad quality of a loans portfolio, and to the absence of the banks of each country were also considered. of penetration and capitalization of the  nancial system. GDP = gross domestic product. Source: Elaborated by the authors. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 101 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil 4. RESEARCH HYPOTHESES e hypotheses to be tested in this study and economic activity, preventing possible contractions in formulated in an alternative form are: the economy from being enhanced, or the e ects of favorable economic cycles from being ampli ed.  ey H : Provisions based on the incurred loss model present are countercyclical in nature. a negative relationship with the variables that represent H : Provisions based on the mixed model adopted in economic activity, contributing towards enhancing Brazil present a positive relationship with the variables possible contractions experienced by the economy, or that represent economic activity, preventing possible towards amplifying the e ects of favorable economic contractions experienced by the economy from being cycles.  ey are pro-cyclical in nature. enhanced, or the e ects of favorable economic cycles H : Provisions based on the expected loss model present a positive relationship with the variables that represent from being ampli ed.  ey are countercyclical in nature. 5. METHODOLOGY such as Cavallo and Majnoni (2001), Bikker and Hu (2002), 5.1. Presentation of the Model and Discussion of the Variables Laeven and Majnoni (2003), Bikker and Metzemakers (2004), Handorf and Zhu (2006), Bouvatier and Lepetit A linear regression model was de ned to evaluate the (2007), and Glen and Mondragón-Vélez (2011). relationship between the variables of interest in order to e model is estimated using the panel data method. identify how the provisions are associated with the variables e empirical tests use the model below, elaborated based that characterize business cycles. Most of the studies relating provisions and economic cycles use this type of model, on the aims and hypotheses of the research: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = 𝛽𝛽 +𝛽𝛽 ∆𝐺𝐺𝐺𝐺𝑃𝑃 +𝛽𝛽 𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 𝑃𝑃 +𝛽𝛽 𝑈𝑈 𝐸𝐸𝑃𝑃 �� � � � � � � �� +𝛽𝛽 ∆𝑙𝑙𝑙𝑙𝑙𝑙𝑃𝑃𝐸𝐸𝑈𝑈 +𝛽𝛽 𝑙𝑙𝑃𝑃𝐸𝐸𝑈𝑈 +𝛽𝛽 𝑈𝑈𝑈𝑈 +𝛽𝛽 𝑆𝑆𝑆𝑆𝑆𝑆𝑈𝑈 +𝜀𝜀 � �� � �� � �� � �� �� where: The aim of the model is not to capture a possible PROV = loan loss provision over banks’ average total cause and effect relationship between dependent it assets i in time t. variables and independent variables, but rather, to ΔGDP = real variation in GDP in time t. identify how the dependent variable behaves in relation UNEMP = unemployment rate in time t. to variations in each independent variable, especially EAR = earnings before income tax, pro t sharing, and the macroeconomic ones. it loan loss provisions over banks’ average total assets i in Despite the accounting variables being endogenous, time t. their inclusion in the model is warranted because the ΔlnLOAN = variation in the loan balances of banks i provision is in uenced by prominently accounting factors, it in time t. such as receivables, earnings, and equity. Moreover, these LOAN = balance of credit operations over total assets variables work as control variables, helping to more it of banks i in time t. adequately capture the relationship between the dependent NE = equity over total assets of banks i in time t. and the independent variables, taken as critical to the it ▪ SIZE = size of banks i in time t. study (GDP and unemployment rate). it 102 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Most of the models that study the provision-economic in the balances of credit operations in real terms. In the cycle relationship use the provision as a dependent Brazilian case, the de ator used was the General Market variable. In this study, it represents the net amount of Price Index (IGPM). For Spain and the United Kingdom, provision expenses set in the period, obtained by the the Consumer Price Index (CPI) was used.  e variation di erence between the set and the reverted provisions. in the balances of loans was calculated by the di erence e independent variables can be classi ed into: (i) in the natural logarithms in the following way: ln(credit macroeconomic variables for the countries (real GDP operations /IGPM ) – ln(credit operations /IGPM ). it it it-1 it-1 growth and unemployment rate); (ii) accounting variables  e loans over total assets variable considers the for the banks (earnings before income tax, pro t sharing, exposure of the banks to credit risks, indicating the relative and loan loss provisions over average total assets, variation size of the loans portfolio. It represents the division of the in loan balances, loan balances over total assets, and equity balance of receivables from credit operation accounts by over total assets); and (iii) a control variable (bank size). the value of total assets. e real variation in GDP is a critical variable in  e equity over total assets variable, inserted into the model and considered a more useful indicator for the model to indicate a possible use of the provision as representing the business cycle. Its value represents the a capital management tool, is the result of dividing net percentage variation in GDP at constant prices. equity by total assets. e unemployment rate is another representative metric of  e SIZE variable, de ned as the natural logarithm it the current phase of the economic cycle. In Brazil, it indicates of total assets, de ated by the IGPM or the CPI, aims to the joblessness rate for 10-year-olds upwards, considering only control the size e ects of the institutions. Larger banks, the main metropolitan regions of the country. or those participating in conglomerates, are expected to e earnings over total assets variable, which indicates set more robust provisions. a possible use of the provision for earnings management, is the result of dividing the value of earnings before income 5.2. Summary of the Expected Results tax, pro t sharing, and loan loss provisions by the average Table 2 summarizes the expected results in relation value of total assets. to the behavior and to the sign of the coe cients of the e growth in lending variable, which indicates the explanatory variables, based on the formulated hypotheses. evolution of the banks’ credit risk, represents the variation Table 2 Summary of the expected results in relation to the model explanatory variables. Variable Expected Behavior Sign In countries with provision accounting procedures that are considered pro-cyclical, such as the United Kingdom, a positive variation in - GDP should make the level of the banks’ provision decrease. In countries with provision accounting criteria that are considered countercyclical, such as Spain, a positive variation in GDP + GDP Growth should make the level of the banks’ provision increase. In Brazil, a country with provision accounting criteria that is considered mixed, a positive variation in GDP should make the level of the banks’ provision increase, considering the prevalence of ex-post criteria over ex- ante criteria, which presupposes a prevalently countercyclical nature. In countries with pro-cyclical provision accounting criteria such as the United Kingdom, a positive variation in unemployment + should make the level of the banks’ provision increase. In countries with countercyclical provision accounting criteria, such as Spain, a positive variation in unemployment - Unemployment rate should make the level of banks’ provision decrease. In Brazil, a country with provision accounting criteria that is considered mixed, a positive variation in the unemployment rate should make the level of the banks’ provision decrease, given the prevalence of ex-post criteria over ex-ante criteria, which presupposes a prevalently countercyclical nature. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 103 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil Table 2 Cont. Variable Expected Behavior Sign Considering that banks use the provision as an earnings management tool, Earnings/Assets a positive variation in earnings before income tax, pro t sharing, and loan + loss provisions should make the level of the bank’s provision increase. Considering that banks use the provision as a capital management tool, a negative variation in net equity should make the level of the NE - banks’ provision increase. A bank would be likely to record a higher provision when its regulatory capital was below that required. Considering that a greater share of credit operations in total investments represents an increase in credit risk, a positive Loans/Assets + variation in the relationship between loans and total assets should make the level of commercial banks’ provisions increase. Considering that a positive variation in the volume of credit operations over time Variation in loans represents an increase in credit risk, a positive variation in the volume of credit + operations should make the level of commercial banks’ provisions increase. NE = net equity; GDP = gross domestic product. Source: Elaborated by the authors. studied the risk of a spurious regression was removed. 5.3. Tests Applied and Robustness Procedures e Pearson correlation matrix (tables 3, 4, and 5) proved Adopted that a high correlation (above 0.8) did not exist between the e Im, Pesaran, and Shin – I.P.S, ADF-Fisher, and independent variables in the countries analyzed. PP-Fisher unitary root tests indicated that in the countries Table 3 Pearson Correlation Matrix – Brazil. LOAN ∆lnLOAN EAR NE ∆GDP UNEMP SIZE LOAN 1.000000 - - - - - - ∆lnLOAN 0.095587 1.000000 - - - - - EAR 0.233714 0.024260 1.000000 - - - - NE -0.147020 -0.051504 0.139431 1.000000 - - - ∆GDP -0.011816 0.007476 0.025810 0.017003 1.000000 - - UNEMP -0.057126 0.039506 0.120247 0.045110 0.040903 1.000000 - SIZE -0.097314 0.005905 -0.107682 -0.592854 -0.009121 -0.234355 1.000000 LOAN = share of credit operations in the total assets of Brazilian banks; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over average total assets of the Brazilian banks; NE = equity over total assets of banks i in time t; ∆GDP = real growth in gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. Table 4 Pearson correlation matrix – Spain. LOAN ∆lnLOAN EAR NE ∆GDP UNEMP SIZE LOAN  1.000000 - - - - - - ∆lnLOAN 0.179954  1.000000 - - - - - EAR -0.088175 0.062070  1.000000 - - - - NE -0.113306 -0.026035  0.241673  1.000000 - - - ∆GDP  0.121625 0.119785  0.082047 0.004057  1.000000 - - UNEMP  -0.191935 -0.091091 -0.126566 0.003399 -0.739170  1.000000 - SIZE -0.062023 0.011395 -0.070449 -0.646019 -0.070423 0.059811 1.000000 LOAN = share of credit operations in the total assets of Brazilian banks; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over average total assets of the Brazilian banks; NE = equity over total assets of banks i in time t; ∆GDP = real growth in gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. 104 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 5 Pearson correlation matrix – United Kingdom. LOAN ∆lnLOAN EAR NE ∆GDP UNEMP SIZE LOAN 1.000000 - - - - - - ∆lnLOAN 0.136788 1.000000 - - - - - EAR 0.053496 0.026894 1.000000 - - - - NE 0.114561 -0.055089 0.015586 1.000000 - - - ∆GDP -0.036011 0.110507 -0.035462 -0.003076 1.000000 - - UNEMP 0.012340 -0.155514 -0.030084 0.050401 -0.607156 1.000000 - SIZE -0.279305 0.042143 -0.041180 -0.448594 -0.060661 0.097252 1.000000 LOAN/ASS = share of credit operations in the total assets of Brazilian banks; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over average total assets of the Brazilian banks; NE = equity over total assets of banks i in time t; ∆GDP = real growth in gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. e risk of multicollinearity was analyzed by applying Banks, Multiple Banks with Commercial Portfolio, and Savings the variance in ation test among the explanatory variables, Banks that do not form part of a conglomerate). which did not reveal multicollinearity problems. Ninety-eight institutions were initially considered. e Chow test was carried out to test the existence of Together, these institutions controlled 84.1% of the total assets st individual heterogeneity and con rm whether the use of of the National Financial System on December 31 2012. panel data would apply to the study. In the three countries, Banco Plural, Morgan Stanley, Banco BM&F, Banco the option was to use the regression with individual e ects. Opportunity, BNY Mellon, Western Union, and Banco en the Hausman test was carried out to de ne the best Petra were excluded from the sample as they did not panel data method for the regression estimation. For Brazil, present a credit operations balance in the period analyzed. the result indicated the random e ects model. For Spain Data was extracted from the semi-annual  nancial and the United Kingdom, it was the  xed e ects model. st statements of the chosen institutions on June 30 and In order to analyze the existence of autocorrelation st December 31 between 2001 and 2012, obtained from between the residuals of the regression, the Durbin- the Banco Central do Brasil website. Watson test was used. In the three countries, the value of the statistic was between d and d , indicating that the 5.4.2. In relation to the banks that operate in Spain. l u test was inconclusive. Adopting a conservative position, The sample considered all of the Spanish banks the null hypothesis of the inexistence of autocorrelation associated with the Asociación Española de Banca (AEB). is rejected; that is, the model residuals appear to be Fi y-eight institutions, including  nancial conglomerates autocorrelated. and individuals banks, were considered in the study. In light of the possibility of the existence of cross-sectional autocorrelation of the residuals, the use of the SUR cross-sectional 5.4.3. In relation to the banks that operate in the United standard errors method (PCSE) in the model estimation arose as Kingdom. an alternative to the problem, enabling the generation of robust The selection of banks that operate in the United parameters even in the presence of residues autocorrelation. Kingdom was based on the publication “List of Banks as e overall signi cance of the model was proven using Compiled by the Bank of England on 31 March 2013” , which the F test. covers the list of banks under the supervision of the Bank of England, available from the website www.bankofengland. 5.4. Sample De nition and Description of the co.uk.  e list is comprised of 153  nancial institutions, with Data Source 45 banks being selected to compose the sample (29.41% of the supervised institutions).  e sample prioritized st 5.4.1. In relation to the banks that operate in Brazil. institutions closing their  nancial year on December 31 and which presented their information in identical monetary In Brazil, the sample considered the institutions that bases (pounds sterling).  e information was obtained formed part of Consolidated Banking I, formed by Banking from the Company Check website (www.companycheck. Conglomerate I (composed of at least one institution of the Commercial Bank or Multiple Bank with Commercial Portfolio co.uk), which provides the annual  nancial statements of type) and Independent Banking Institutions I (Commercial banks operating in the United Kingdom. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 105 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil 6. RESULTS set, on average, provisions in percentages equal to 0.83%, 6.1. Descriptive Statistics of the Dependent Variable 0.2668%, and 0.3607% of total assets, respectively.  e standard deviations show a large variability in the e descriptive statistics of the dependent variable are provision, perhaps caused by the reversion mechanism, found in Table 6. From the second semester of 2001 to and indicate that the frequency distributions of the the second semester of 2012, the commercial banks that operated in Brazil, in Spain, and in the United Kingdom provision in the three countries have a similar behavior. Table 6 Descriptive statistics of the dependent variable in the model – Brazil, Spain, and the United Kingdom, from 2001 to 2012. Provisions/Total Assets Brazil Spain United Kingdom Mean 0.0083 0.0027 0.0036 Median 0.0044 0.0008 0.0013 Maximum Value 0.1643 0.0858 0.1432 Minimum Value -0.0519 -0.0337 -0.0407 Standard Deviation 0.0146 0.0071 0.0102 Source: Elaborated by the authors. and the United Kingdom, which were around 82% and 6.2. Descriptive Statistics of the Independent 71%, respectively. Variables e average share of own resources over the total investments of the commercial banks has shown to be 6.2.1. Brazil. quite signi cant, at around 23%, and more comfortable According to Table 7, on average only 34.34% of than that of the banks that operate in Spain (average of the resources were used for credit operations, a lower 20%) and much higher than that of the banks that operate percentage than those of the banks that operate in Spain in the United Kingdom (average of 13.8%). Table 7 Descriptive statistics of the independent variables in the model – Brazil, 2001-2012. MEASURE LOAN ∆lnLOAN NE EAR ∆GDP UNEMP SIZE Mean 0.3434 0.1272 0.2320 0.0225 0.0167 0.0855 13.999 Median 0.3309 0.1268 0.1551 0.0184 0.0196 0.0830 13.922 Maximum Value 1.0344 5.6920 0.9999 0.3568 0.0474 0.1300 20.8069 Minimum Value 0.0000 -5.2298 -0.1038 -0.1831 -0.0279 0.0470 8.2451 Standard Deviation 0.2356 0.5020 0.2154 0.0326 0.0174 0.0222 2.3498 LOAN = share of credit operations in the total assets of the Brazilian banks; ∆lnLOAN = real growth in the balances of credit operations recorded in the balance sheets of the Brazilian banks; NE = share of equity over total assets of the Brazilian banks; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of the Brazilian banks; ∆GDP = real growth in the gross domestic product of Brazil; UNEMP = unemployment rate in Brazil; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market prices index. Source: Elaborated by the authors. 6.2.2. Spain. in Spain.  e average return on assets was around 0.53%, Table 8 presents the descriptive statistics of the below the returns of the banks that operate in Brazil independent variables in relation to the banks that operate (2.25%) and in the United Kingdom (1.91%). 106 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 8 Descriptive statistics of the independent variables in the model – Spain, 2001-2012. MEASURE LOAN ∆lnLOAN NE EAR ∆GDP UNEMP SIZE Mean 0.8195 0.0534 0.1998 0.0053 0.0078 0.1362 14.0939 Median 0.8793 0.0354 0.0808 0.0036 0.0140 0.1130 13.9648 Maximum Value 1.0766 8.0520 0.9999 0.3045 0.0201 0.2510 20.9620 Minimum Value 0.0000 -8.0301 -0.0315 -0.1705 -0.0278 0.0810 0.0810 Standard Deviation 0.1905 0.4701 0.2745 0.0242 0.0131 0.0527 2.4123 LOAN = share of credit operations in the total assets of the Spanish banks; ∆lnLOAN = real growth in the balances of credit operations recorded in the balance sheets of the Spanish banks; NE = share of equity over total assets of the Spanish banks; EAR = share of earnings before income tax, pro t sharing, and loan loss provision over the average total assets of the Spanish banks; ∆GDP = real growth in the gross domestic product of Spain; UNEMP = unemployment rate in Spain; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. 6.2.3. United Kingdom. e descriptive statistics of the independent variables related to the banks that operate in the United Kingdom are presented in Table 9. Table 9 Descriptive statistics of the independent variables in the model – United Kingdom, 2001-2012. MEASURE LOAN ∆lnLOAN NE EAR ∆GDP UNEMP SIZE Mean 0.7076 0.1129 0.1380 0.0191 0.0150 0.0600 13.2893 Median 0.7453 0.0900 0.0935 0.0086 0.0224 0.0533 12.8845 Maximum Value 1.3041 6.5087 0.9885 5.5899 0.0395 0.0807 20.6667 Minimum Value 0.0000 -2.0549 0.0000 -0.1625 -0.0517 0.0467 3.98689 Standard Deviation 0.2353 0.4501 0.1439 0.2472 0.0242 0.0140 2.65957 LOAN = share of credit operations in the total assets of the British banks; ∆lnLOAN = real growth in the balances of credit operations recorded in the balance sheets of the British banks; NE = share of equity over total assets of the British banks; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of the British banks; ∆GDP = real growth in the gross domestic product of the United Kingdom; UNEMP = unemployment rate in the United Kingdom; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. 6.3. Test of the Hypotheses independent variables are associated with 38.87% of the behavior of the dependent variable.  e F statistic, with 6.3.1. H1: Banks that operate in the United Kingdom a p-value of 0.0000, con rms the statistical signi cance e coe cient of determination (R ) indicates that the of the model (Table 10). Table 10 Coef cients of Determination and F Statistic of the model – United Kingdom. Variables Value R 0.3887 R Adjusted 0.3191 F Statistic 5.5863 P-Value (F) 0.0000 Source: Elaborated by the authors. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 107 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil The macroeconomic variables and the specific loan loss provisions over total assets” variable was also accounting variables of the banks, except GDP, do not not found to be signi cant, indicating that variations in have signi cant e ects on the provision in the United earnings would not explain the behavior of the provision Kingdom.  e “loans over total assets” and “growth in and going against the expectations of the research. lending” variables were not found to be signi cant, thus Earnings management via provisions does not appear going against the expectations presented in Table 2. to be a common practice among banks in the United e “earnings before income tax, pro t sharing, and Kingdom. Table 11 Regression results – United Kingdom. Variable Coef cient Standard deviation T statistic Prob. C 0.0114 0.0074 1.5429 0.1236 LOAN -0.0009 0.0023 -0.3715 0.7105 ∆lnLOAN -0.0006 0.0019 -0.3013 0.7633 EAR -0.0022 0.0053 -0.4124 0.6803 NE -0.0052 0.0072 -0.7189 0.4726 ∆GDP -0.0553 0.0140 -3.9372 0.0001 UNEMP 0.0505 0.0270 1.8662 0.0627 SIZE -0.0007 0.0005 -1.2307 0.2191 C = constant in the regression; LOAN = balances of credit operations of banks i in time t; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of banks i in time t; NE = equity over the total assets of banks i in time t; ∆GDP = real growth of gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of banks i in time t. Source: Elaborated by the authors. e “net equity over total assets” variable was not and Metzemakers (2004). found to be signi cant, thus going against the capital  e “size” control variable was also not found to be management hypothesis. signi cant. At a 1% level of signi cance, the coe cient of the “GDP 6.3.2. H2: Banks that operate in Spain. growth” variable was found to be signi cantly negative, With regards to the banks that operate in Spain, indicating pro-cyclical behavior of the banks in the United the coe cient of determination (R ) indicates that the Kingdom, in line with hypothesis H1. independent variables are associated with 38.31% of e “unemployment rate growth” variable was only the dependent variable behavior.  e F statistic, with a found to be signi cant at a 10% level of signi cance, p-value of 0.0000, con rms the statistical signi cance of thus going against the expectations.  is result, however, the model (Table 12). supports the  ndings of previous studies, such as Bikkers Table 12 Coef cients of determination and F statistic of the model – Spain. Variables Value R 0.3831 R Adjusted 0.3436 F Statistic 9.7147 P-Value (F) 0.0000 Source: Elaborated by the authors. e “loans over total assets”, “earnings over total assets”, signi cance, as can be observed in Table 13.  e other “GDP growth” and “size” variables have signi cant e ects variables were not signi cant. on loan loss provisions in Spain, at di erent levels of 108 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 13 Regression results – Spain. Variable Coef cient Standard deviation t-statistic Prob. C -0.0149 0.0074 -2.0191 0.0438 LOAN 0.0031 0.0017 1.8418 0.0658 ∆lnLOAN -0.0004 0.0003 -1.2059 0.2281 EAR -0.1004 0.0170 -5.9089 0.0000 NE -0.0020 0.0052 -0.3825 0.7022 ∆GDP -0.0534 0.0192 -2.7727 0.0057 UNEMP 0.0040 0.0049 0.8224 0.4111 SIZE 0.0011 0.0004 2.5368 0.0113 C = constant in the regression; LOAN = balances of credit operations of banks i in time t; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of banks i in time t; NE = equity over the total assets of banks i in time t; ∆GDP = real growth of gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of banks i in time t. Source: Elaborated by the authors. e “loans over total assets” variable was not found to signi cant, thus going against the expectation with regards be signi cant, at a 10% level of signi cance.  e positive to the use of the provision for earnings management. sign of the coe cient supports the expectation presented Contrary to the expectation of hypothesis H2, at a 1% in Table 2, indicating that the provision tends to rise when level of signi cance the coe cient of the “GDP growth” the share of loans over total assets increases. variable was found to be signi cantly negative, indicating Contrary to expectations, the “growth in lending” pro-cyclical behavior. variable was not found to be signi cant, indicating that  e “unemployment rate” variable was not found the variation in the volume of lending does not explain to be signi cant, going against expectations.  e “size” the behavior of the provisions. control variable was found to be signi cant, with a positive The “earnings before income tax, profit sharing, sign, indicating that the larger the bank, the higher the and loan loss provisions over total assets” variable was provision tends to be. not found to be signi cant in Spain, at a 1% level of 6.3.3. H3: Banks that operate in Brazil. signi cance.  e negative sign of the coe cient indicates e coe cient of determination (R ) indicates that the growth of the provision whenever earnings decrease and independent variables of the model are associated with vice-versa, thus going against expectations.  ere is no 15.65% of the dependent variable behavior (Table 14). evidence of earnings management practices via provisions e F statistic, which presents a p-value equal to 0.0000, in the Spanish banks. con rms the statistical signi cance of the model. The “net equity over total assets” variable is not Table 14 Coef cients of determination and F statistic of the model – Brazil. Variables Value R 0.1565 R Adjusted 0.1531 F Statistic 45.1995 P-Value (F) 0.0000 Source: Elaborated by the authors. According to the data in Table 15, the macroeconomic studied in Brazil, with the exception of the unemployment variables and the speci c accounting variables of the banks rate and size, have signi cant e ects on the provision. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 109 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil Table 15 Regression results – Brazil. Variable Coef cient Standard deviation t-statistic Prob C -0.0086 0.0080 -1.0795 0.2805 LOAN 0.0268 0.0026 10.4186 0.0000 ∆lnLOAN -0.0014 0.0007 -1.8981 0.0579 EAR 0.0868 0.0170 5.1125 0.0000 NE 0.0138 0.0045 3.0866 0.0021 ∆GDP -0.0477 0.0186 -2.5637 0.0104 UNEMP -0.0173 0.0185 -0.9356 0.3496 SIZE 0.0004 0.0004 0.9305 0.3523 C = constant in the regression; LOAN = balances of credit operations of banks i in time t; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of banks i in time t; NE = equity over the total assets of banks i in time t; ∆GDP = real growth of gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of banks i in time t. Source: Elaborated by the authors. e “loans over total assets” variable was found to sign of the coe cient would go against the expectation of be signi cant at a 1% level of signi cance. As expected, use of the provision for capital management; it is supposed when a bank increases its share of credit operations in that the banks raise the provision whenever their capital total investments, the percentage of the provision also levels are more comfortable. It is assumed that the fact that increases, con rming the claim of growth in risks in the banks in Brazil have historically presented a higher phases of economic expansion.  is prudent behavior of reference equity than that required by the Central Bank the banks contributes to attenuating the e ects caused and the absence itself of the generic provision instrument by possible pro-cyclical behaviors. in the country’s regulation explain this result. e “growth in lending” variable was only found Regarding hypothesis H3, at a 1% level of signi cance to be significant at a 10% level of significance. The the coe cient of the “GDP growth” variable was shown to coe cient presented a negative sign, indicating that the be signi cantly negative, indicating pro-cyclical behavior. provisions usually decrease when the balance of credit  is appears to indicate the absence of an e cient risk operations increases.  is  nding would go against the evaluation mechanism with a forward-looking character. initial expectation, which indicated a probable positive With regards to the “growth in the unemployment relationship between growth in the lending volume and rate” variable, this was not found to be signi cant, thus the provision. going against the expectation presented in Table 2.  e e “earnings before income tax, pro t sharing, and same occurs with the “size” control variable. It cannot be loan loss provisions over total assets” variable was also claimed that larger banks in Brazil, or those belonging to found to be signi cant at a 1% level of signi cance, conglomerates, set more robust provisions than the rest. indicating that variations in earnings are related with As a sensitivity analysis, the lagged GDP and the behavior of the provision.  e positive sign of the unemployment rate variables were added to the models. coe cient indicates growth in provision levels whenever In Brazil and the United Kingdom, these variables were earnings increase and vice-versa, thus supporting the not found to be signi cant, indicating that the e ect expectation presented in Table 2.  e result indicates the of their behavior on the provision calculation occurs use of the provision for earnings management, a practice in the same accounting period. With regards to Spain, which ameliorates the pro-cyclical e ects of the provision. the unemployment rate variable was not found to be e “net equity over total assets” variable was shown significant, but the GDP variable was shown to be to be signi cant at a 1% level of signi cance.  e negative signi cant at a 1% level of signi cance. 7. FINAL REMARKS Studies such as those by Bikker and Hu (2002), Laeven the incurred loss model leads to pro-cyclical behavior and Majnoni (2003), and Bikker and Metzemakers (2004) of the provision and the expected loss model results in reveal that the choice between di erent accounting models countercyclical behavior of the provision.  is study would determine the behavior of the provision with proposed to investigate whether the incurred loss models, relation to the business cycles.  e expectation is that which are conceptually pro-cyclical, would lead banks to 110 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo setting pro-cyclical provisions, and whether the expected di culty itself of predicting future economic scenarios, loss models, which are conceptually countercyclical, would among other factors, could be leading to provisions of a lead banks to recognizing countercyclical provisions. di erent nature from that which the legislator intended Two countries were intentionally chosen to represent when establishing more speci c criteria for setting them. these accounting models: Spain (expected loss) and Only in relation to the United Kingdom was the hypothesis the United Kingdom (incurred loss). A mixed model con rmed that the accounting model adopted, based adopted in Brazil was also analyzed.  e prevalence of essentially on the IASB rules, has in fact caused provisions typical procedures of expected loss models reinforced that are pro-cyclical in nature. the hypothesis that the Brazilian model would be Also in relation to the United Kingdom, only the GDP countercyclical. growth variable was found to be signi cant, at a 5% level of Based on the linear regression models used in previous signi cance. It is assumed that the variables that represent studies, especially those by Cavallo and Majnoni (2001), credit risk, such as “loans over total assets” and “growth Bikker and Hu (2002), Laeven and Majnoni (2003), and in lending” were not found to be signi cant due to the Bikker and Metzemakers (2004), variables were selected characteristics of the accounting model itself adopted in to compose an econometric model that identi ed the that country, which only recognizes the provision when relationship between the provision and the economic objective evidence of loss is identi ed.  us, the impact cycles in the three countries selected. of the two variables would only occur indirectly (a greater Two variables were inserted into the model in order share of loans over assets or growth in the volume of to identify the relationship between the provision lending can lead to a higher probability of occurrence of and economic cycles: the real variation in GDP and objective evidence of loss in the future).  ese variables the unemployment rate. Contrary to expected, the can, therefore, change without the banks in the United unemployment rate was not found to be statistically Kingdom having adjusted their provisions. In the United signi cant in Brazil or in Spain. In United Kingdom this Kingdom, there was also no evidence found that the banks variable was found to be signi cant, at a 10% signi cance are using the provision for earnings management, thus level, presenting a positive coe cient, which indicates that contradicting the study by Silva (2016), or for capital the provision tends to grow when the unemployment rate management. increases. It bears mentioning that studies such as those With regards to earnings management, this is by Bikker and Metzemakers (2004) have already revealed assumed to be a reality for commercial banks in Brazil, that the unemployment rate would not be the best proxy thus supporting the results of the studies by Macedo to evaluate the relationship between the provision and and Kelly (2016), Bortoluzzo, Sheng, and Gomes (2016), economic cycles, a fact that is supported by this study. and Silva (2016).  e “earnings before income tax, pro t e variable unanimously taken in the studies involving sharing, and loan loss provisions” variable was found to be cyclicality is the variation in GDP.  e  ndings of this signi cant, at a 1% level of signi cance, having presented research are consistent with these studies, as GDP was a positive coe cient indicating that the provision tends also found to be a statistically signi cant variable at a to grow when the banks’ earnings increase. In relation to level of signi cance of practically 1% for each accounting Spain, despite the signi cance at a 1% level of signi cance, model that was the object of this research. this variable presented a negative sign, signaling that the With regards to the sign of the coe cient of the variable provision decreases when earnings increase and thus that re ects GDP, two of the formulated hypotheses were contradicting the study by Silva (2016). not supported.  e biggest surprise was in Spain, whose Earnings management practices, which are accounting model prescribes setting forward-looking reprehensible in some aspects, can represent a counterpoint provisions, including provisions of a generic nature, to setting pro-cyclical provisions, easing their e ects on which should cause countercyclical provisions.  e sign the current phase of the business cycle. of the coe cient of the variable, which would presumably Capital management practices were also not found to be positive, was found to be negative, indicating pro- be a reality for the commercial banks in Brazil and Spain. cyclicality. In Brazil, despite the general rule being to set In Brazil, despite the signi cance at a level of 1%, the forward-looking provisions, and although backward- aforementioned variable presented a positive coe cient, looking previsions ultimately prevail when the delays in indicating that the greater the share of net equity over total the payments of principal and interest are present, the assets, the greater the provision set by commercial banks. study also revealed pro-cyclical behavior. It is inferred  is situation may be indicating that the comfortable level that the subjective character of the provision and the of capitalization presented by the banks in Brazil (23.20% R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 111 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil of total assets, on average) is inducing an increase in the It was also found that in Brazil and the United Kingdom provision level whenever net equity presents a positive the e ect of economic variations on the provision behavior variation, which represents a conservative procedure takes place in the same accounting period. In the case of that also serves as a counterpoint to setting pro-cyclical Spain, however, a variation in GDP in the previous period provisions. In Spain, the “net equity over total assets” was found to be a signi cant variable. variable was not found to be signi cant, indicating that With relation to the econometric model chosen for the provision level is not a ected by the behavior of the the research, it was found that the peculiar nature of equity of commercial banks that operate in that country. the provision rules in the United Kingdom, in relation The “size” variable was found to be statistically to those in Brazil and Spain, may be determining the signi cant only in relation to the banks that operate in di erence in the levels of signi cance of the explanatory Spain. In the other countries, the size of the banks does variables for the model, when this is applied to commercial not appear to in uence the provision level. banks in the country that adopts the IASB accounting e results support the previous studies by Bikker standards.  us, it is assumed that adaptations are needed and Hu (2002), Laeven and Majnoni (2003), and Bikker to adjust the model used to the speci cities of countries and Metzemakers (2004), in the sense that the provision that use relatively di erent accounting rules for setting behavior strongly depends on the economic cycle, the provision. indicating that the provision generally grows in poor As an intentional sample is concerned, which considers economic times. Another similarity between the results of only one country for each accounting model, the results this research and the aforementioned studies is the  nding may not be generalized. For future research, studies with that the e ect of pro-cyclicality is usually mitigated by an di erent econometric models are suggested, incorporating increase in the provision in periods of higher earnings variables not considered in this research, especially (earnings management practices). those related to the delay in the payment of principal A possible assumption regarding the fact that and interest and to the accounting recognition of losses. the research revealed pro-cyclical behavior when Evaluating the question of cyclicality by considering countercyclical behavior was expected is that the banks di erent accounting scenarios in the same country (for may be using discretion and managing their earnings example, studying cyclicality before and a er the advent via the provision and more than compensating for the of Resolution n. 2,682/1999) could also prove to be an expected e ect of the models, which is a hypothesis to interesting experiment for re ecting on the provision be investigated in future studies. behavior in di erent regulatory environments. REFERENCES Bebczuk, R., Carrera, J., & Sangiácomo, M. (2010). A provisioning behaviour and procyclicality. Journal of new look into credit procyclicality: International International Financial Markets, Institutions & Money, panel evidence. In BIS Consultative Council for the 209, 1-17. doi: 10.1016/j.int n.2004.03.004. Americas Conference on “systemic risk, bank behavior Bortoluzzo, A. B., Sheng, H. H., & Gomes, A. L. P. and regulation over the business cycle”. Buenos Aires, (2016) Earning management in Brazilian  nancial Argentina. institutions. Revista de Administração da Universidade Betancourt, L., & Baril, C. P. (2009, October/November). de São Paulo, 51(2), 182-197. ISSN 0080-2107. doi: Accounting for loan losses: Is Dynamic provisioning 10.5700/rausp1233. the answer? Bank Accounting Finance, 22(6), 9-16. Bouvatier, V., & Lepetit, L. (2007). Banks’ procyclical Retrieved from http://connection.ebscohost.com/c/ behavior: Does provisioning matter? Journal of articles/44680540/accounting-loan-losses-dynamic- International Financial Markets, Institutions & Money, provisioning-answer. 18, 513-528. doi: 10.1016/j.int n.2007.07.004. Bikker, J. A., & Hu, H. (2002). Cyclical patterns in pro ts, Bouvatier, V., & Lepetit, L. (2012, July). Provisioning rules provisioning and lending of banks and procyclicality and bank lending: A theoretical model. Journal of of the new basel capital requirements. Banca Nazionale Financial Stability, 8(2), 25-31. Retrieved from http:// del Lavaro Quarterly Review, 55(221), 143-175. ssrn.com/abstract=1148314. Retrieved from https://www.dnb.nl/binaries/ot039_ Burns, A. F., & Mitchell, W. C. (1946). Measuring business tcm46-146052.pdf. cycles. NBER Book Series Studies in Business Cycles, Bikker, J. A., & Metzemakers, P. A. J. (2004, March). Bank New York: National Bureau of Economic Research. 112 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Cavallo, M., & Majnoni, G. (2001, June). Do banks openview/ee52c0978ed6e519591 513944a8a54/1?pq- provision for bad loans in good times? Empirical origsite=gscholar&cbl=26518. evidence and policy implications. Working Paper Harrald, P., & Sandal, T. (2010, October). Addressing pro- Nº 2.619. World Bank Policy Research. Retrieved cyclicality. Asia-Risk, 70-72. Retrieved from www.risk. from https://openknowledge.worldbank.org/ net. handle/10986/19607. Healy, P. M., & Wahlen, J. M. (1999, December). A Cohen, D. A, & Zarowin, P. (2007, August). Earnings review of the earnings management literature and its management over the business cycle. New York implications for standard setting. Accounting Horizons, University. Stern School of Business. Retrieved 13(4), 365-383. Retrieved from https:// sher.osu. from http://w4.stern.nyu.edu/emplibrary/ edu/~young.53/Healy-Wahlen. EM_08_23_07FINAL.pdf. Hendriksen, E. S., & Van Breda, M. F. (1999). Teoria da Cortavarria, L., Dziobek, C., Kanaya, A., & Song, I. contabilidade. São Paulo, SP: Atlas. (2000, December). Loan review, provisioning, and Laeven, L., & Majnoni, G. (2003, April). Loan loss macroeconomic linkages. WP/00/195. IMF Working provisioning and economic slowdowns: too much, too Paper. International Monetary Fund. Retrieved late? Journal of Financial Intermediation, 12(2), 178- from http://www.imf.org/external/pubs/ /wp/2000/ 197. Retrieved from http://www.sciencedirect.com/ wp00195.pdf. science/article/pii/S1042957303000160. Curcio, D., & Hasan, I. (2015). Earnings and Longbrake, W. A., & Rossi, C. V. (2011, July). Procyclical capital management and signaling:  e use of versus countercyclical policy e ects on  nancial markets. loan-loss provisions by European banks.  e Study prepared for the Anthony T. Clu Fund.  e European Journal of Finance, 21(1), 26-50. doi: Financial Services Roundtable. Financing America’s 10.1080/1351847X.2012.762408. Economy. Retrieved from http://www.fsround.org/fsr/ Glen, J., & Mondragón-Vélez, C. (2011, January). Business pdfs/clu /CountercyclicalPE.pdf. cycle e ects on commercial bank loan portfolio Macedo, M. A. S., & Kelly, V. L. A. (2016, May/August). performance in developing economies. Review of Development Finance, 1(2), 150-165. Retrieved from Gerenciamento de resultados em instituições http://papers.ssrn.com/sol3/papers.cfm?abstract_ nanceiras no Brasil: uma análise com base em id=1754672. provisões para crédito de liquidação duvidosa. Revista Evidenciação Contábil & Finanças, 4(2), 82-96. Gonzales, J. E. N. (2009, July).  e fundamentals of procyclicality of the  nancial system. Economic Martinez, A. L. (2001). “Gerenciamento” dos resultados Newsletter. Bangko Sentral NG Pilipinas, 9(3). contábeis: estudo empírico das companhias abertas Retrieved from http://www.bsp.gov.ph/downloads/ brasileiras (Ph.D. thesis). Faculdade de Economia, EcoNews/EN09-03.pdf. Administração e Contabilidade, Universidade de São Paulo, São Paulo, SP. Gray, R. P., & Clarke, F. L. (2004). A methodology for calculating the allowance for loan losses in commercial Silva, C. A. M. (2016). Utilização da provisão para créditos banks. Abacus, 40(1), 321-341. de liquidação duvidosa para  ns de gerenciamento de Handorf, W. C., & Zhu, L. (2006, spring). US Bank resultado nas instituições  nanceiras brasileiras e luso- loan-loss provisions, economic conditions, and espanholas (Master’s thesis). Faculdade de Economia, regulatory guidance. Journal of Applied Finance, 16(1), Administração e Contabilidade, Universidade de 97-114. Retrieved from http://search.proquest.com/ Brasília, Brasília, DF. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 113 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Quantitative Finance arXiv (Cornell University)

The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil

Quantitative Finance , Volume 2019 (1907) – Jul 17, 2019

Loading next page...
 
/lp/arxiv-cornell-university/the-cyclicality-of-loan-loss-provisions-under-three-different-4v90LU8TPT
ISSN
1808-057X
eISSN
ARCH-3346
DOI
10.1590/1808-057x201804490
Publisher site
See Article on Publisher Site

Abstract

ISSN 1808-057X DOI: 10.1590/1808-057x201804490 O A e cyclicality of loan loss provisions under three di erent accounting models: the United Kingdom, Spain, and Brazil Antônio Maria Henri Beyle de Araújo Universidade Católica de Brasília, Departamento de Ciências Contábeis, Brasília, DF, Brazil Email: henri.beyle@uol.com.br Paulo Roberto Barbosa Lustosa Universidade de Brasília, Faculdade de Economia, Administração, Contabilidade e Gestão de Políticas Públicas, Departamento de Ciências Contábeis e Atuariais, Brasília, DF, Brazil Email: lustosa@unb.br Edilson Paulo Universidade Federal da Paraíba, Centro de Ciências Sociais Aplicadas, Departamento de Finanças e Contabilidade, João Pessoa, PB, Brazil Email: e.paulo@uol.com.br nd Received on 10.11.2016 – Desk acceptance on 11.07.2016 – 2 version approved on 08.08.2017 – Ahead of print on 11.06.2017 ABSTRACT A controversy involving loan loss provisions in banks concerns their relationship with the business cycle. While international accounting standards for recognizing provisions (incurred loss model) would presumably be pro-cyclical, accentuating the e ects of the current economic cycle, an alternative model, the expected loss model, has countercyclical characteristics, acting as a bu er against economic imbalances caused by expansionary or contractionary phases in the economy. In Brazil, a mixed accounting model exists, whose behavior is not known to be pro-cyclical or countercyclical.  e aim of this research is to analyze the behavior of these accounting models in relation to the business cycle, using an econometric model consisting of  nancial and macroeconomic variables.  e study allowed us to identify the impact of credit risk behavior, earnings management, capital management, Gross Domestic Product (GDP) behavior, and the behavior of the unemployment rate on provisions in countries that use di erent accounting models. Data from commercial banks in the United Kingdom (incurred loss), in Spain (expected loss), and in Brazil (mixed model) were used, covering the period from 2001 to 2012. Despite the accounting models of the three countries being formed by very di erent rules regarding possible e ects on the business cycles, the results revealed a pro-cyclical behavior of provisions in each country, indicating that when GDP grows, provisions tend to fall and vice versa. e results also revealed other factors in uencing the behavior of loan loss provisions, such as earning management. Keywords: provisions, cyclicality, economic cycles, credit operations, commercial banks. Correspondence address: Antônio Maria Henri Beyle de Araújo Universidade Católica de Brasília, Departamento de Ciências Contábeis SQSW 306, Bloco G, Apartamento 204 – CEP: 70673-437 Sudoeste – Brasília – DF – Brazil R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 97 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil 1. INTRODUCTION According to Longbrake and Rossi (2011), many of  e accounting principles in the United States and the rules that regulate the  nancial markets contributed those of the International Accounting Standards Board to deepening the crisis that a ected the global economy (IASB) adopt the incurred loss model, establishing that in the period between 2007 and 2009, fueling the debate a loss in credit operations will only be recognized in regarding the way the  nancial system accentuates the earnings if its occurrence (related to a loss event) is likely e ects of expansionary and contractionary phases in and its value can be estimated with certainty. the economy. At the heart of the claim lies the concern According to Betancourt and Baril (2009), an regarding pro-cyclicality, which represents a positive alternative to the incurred loss model is the expected correlation between a particular variable and economic loss model, which recognizes provisions based on activity (Bebczuk & Sangiácomo, 2010). expected future losses, independently of objective For Harrald and Sandal (2010), pro-cyclicality is evidence. This model uses as a reference the dynamic the way that the banking system interacts with the real provision adopted by the Banco de España, which economy and accentuates the e ects of an economic accepts the setting of a generic provision to protect cycle. It derives from incentives to accelerate lending banks against possible additional losses in a particular in periods of economic expansion and to reduce such economic cycle. activity in times of crisis. In Brazil, Resolution n. 2,682/1999 of the National According to Longbrake and Rossi (2011), the  nancial Monetary Council rules that the provision should be set system should ideally dampen and not amplify business based on the operation risk level and its value should be cycles. Pro-cyclicality becomes especially problematic reviewed at least monthly, depending on the delay in the when it accentuates the e ect of a fall in the economic payment of principal or interest.  is involves a mixed cycle and worsens a crisis.  us, countercyclical rules, system with characteristics of expected loss (provisions which reduce economic imbalances and the depth of based on future losses expectations) and incurred loss business cycles, are welcome. (provisions based on objective evidence represented by Also according to these authors, factors that explain delays in payment). the in uence of the  nancial system on the economic The three accounting models would cause different environment include the accounting rules that set loan loss behaviors of the provision in relation to economic provisions.  e expression “loan loss provision” (“provisão cycles. In order to confirm this assumption, the para créditos de liquidação duvidosa” in Portuguese) is the relationship between the accounting criteria for setting term used by the Brazilian Central Bank for the estimated provisions in banks and economic cycles should be losses from loans granted by  nancial institutions. In investigated. international accounting standards, this provision is Studies such as those by Bikker and Metzemakers represented by impairment losses derived from loans (2004) and by Boutavier and Lepetit (2007) have already and receivables. been conducted to investigate the relationship between For Bikker and Metzemakers (2004), the provision loan loss provisions and economic cycles. Besides being leaves room for subjective judgment and allows for restricted to certain aspects of the provision, these studies discretion when establishing levels considered as used relatively di erent models and variables, thus leaving “adequate”. The study by these authors identified a room for new analyses on the topic. direct relationship between the provision set by banks ere are few studies involving the Brazilian model, and economic cycles, revealing that income smoothing, which is presumably countercyclical since most of capital management, and tax rules also determine the the value of the provisions is the result of classifying level of provision. operations by level of risk at the time of lending and Betancourt and Baril (2009) observe that at the not migration between the risk categories due to credit beginning of a downward economic cycle there is little defaults. provision recognized in banks earnings. As the crisis e aim is to analyze whether the accounting models worsens the provisions grow, deteriorating the equity involved in setting loan loss provisions in banks are pro- situation of banks and reducing levels of lending when the market needs funds the most.  is pro-cyclical behavior cyclical, countercyclical, or acyclical, using an econometric is adopted by banks that follow international accounting model with a consistent theoretical framework and standards for provisions. empirical construction. 98 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo A panel data econometric model was specified in three countries, signaling that the empirically con rmed which the loan loss provision expense in relation to pro-cyclical behavior is independent of the provision total assets is regressed against the variation in Gross model adopted. One probable assumption regarding Domestic Product (GDP), under the control of the this  nding is that the banks may be using discretion unemployment rate and accounting variables that affect and engaging in earnings management via the provision the provision. An identical specification is run for and more than compensating for the expected e ect of samples from three different provision accounting the models, but this is a hypothesis to be investigated in systems (United Kingdom, Spain, and Brazil) with future research. the aim of verifying whether the covariance between Besides this introduction, this paper presents the the provision and GDP is significantly positive literature review in section 2; section 3 formulates (countercyclical), negative (pro-cyclical), or without the three research hypotheses; section 4 details the significance (acyclical). methodology; section 5 presents, analyzes, and The results indicated a negative and significant discusses the results; and section 6 offers some final association between the provision expense and GDP in the remarks. 2. CONSIDERATIONS REGARDING ECONOMIC CYCLES AND PRO-CYCLICALITY A classic de nition of “economic cycle”, or “business e ects are damaging to the economy. Pro-cyclicality derives cycle”, is that of Burns and Mitchell (1946): from incentives for the acceleration of lending in periods of expansion and the reduction in such activity in crises. Business cycles are a type of  uctuation found in the aggregate According to Bebczuk et al. (2010), pro-cyclicality is economic activity of nations that organize their work mainly in a synchronized movement in the growth of credit and business enterprises: a cycle consists of expansions occurring production, both in time t. Excessive cyclicality in lending at about the same time in many economic activities, followed can lead to problems such as exacerbating the economic by similarly general recessions, contractions, and revivals cycle, an increase in systemic risk, and the inadequate which merge into the expansion phase of the next cycle; in duration business cycles vary from more than one year to allocation of resources for loans. ten or twelve years. For Gonzales (2009), pro-cyclicality is present when the actions of banks tend to reinforce the movements of For Harrald and Sandal (2010), pro-cyclicality is the the underlying business cycles, indicating a correlation way in which the banking system interacts with the real between the aforementioned actions and the current economy, accentuating the e ects of an economic cycle. economic environment. Pro-cyclical measures are, e pro-cyclicality cost depends on the extent to which the therefore, those that contribute to strengthening the e ects of the cycle are accentuated and the fact that these current economic cycle. 3. LOAN LOSS PROVISIONS IN BANKS Hendriksen and Van Breda (1999) state that an 3.1. Provision Models important element in evaluating receivables is the Banks’ provision rules can be: (i) backward-looking, treatment of uncertainty with regards to their payment based on losses incurred in operations and considering by the client.  us, revenue should be measured by the the events and conditions prior to the balance sheet; and value that it is expected will be received. In the quality of (ii) forward-looking, set by considering expected losses the value reserve, the provision adjusts the gross value of and based on trend analyses (Bouvatier & Lepetit, 2012). loans according to their credit risk.  is adjustment has According to Cortavarria, Dziobek, Kanaya, and Song a direct e ect on the income calculation, a ecting the amount to be paid out to shareholders. (2000), provisions can be generic and speci c. Generic R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 99 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil provisions are possible or latent losses not yet identi ed management being practiced by  nancial institutions in Brazil, based on the use of loan loss provisions. Macedo (ex-ante) and applicable to all operations. Specific and Kelly (2016) identi ed that from 2006 to 2012 the provisions re ect individually identi ed losses related income level obtained by  nancial institutions in uenced to an observable event (ex-post). these provisions. Bortoluzzo, Sheng, and Gomes (2016) found similar results by analyzing the behavior of 123 3.2. Earnings Management via Provisions Brazilian banks between 2001 and 2012. Silva (2016) Accounting rules allow for the use of discretion showed that between June 2009 and December 2014 and judgment in the  nancial statements preparation Brazilian and Luso-Spanish banking institutions used (Cohen & Zarowin, 2007).  is enables the use of business provision expenses to manage earnings. Cursio and knowledge and expertise to choose the most appropriate Hasan (2015) found that from 1996 to 2006 earnings accounting methods, estimates, and disclosure.  e management was shown to be an important determinant problem is when managers  nd incentives to convey of the provisions of  nancial intermediaries that operate information to their own bene t, choosing accounting in the Euro zone. methods and accounting estimates that do not adequately re ect the underlying economic conditions. 3.3. Capital Management via Provisions For Cohen and Zarowin (2007), this accounting In the capital management hypothesis, banks set higher information discretion can be both a value maximizer and provisions when their regulatory capital is relatively low. opportunistic, leaving room for “earnings management”. According to Bikker and Metzemakers (2004), this occurs Earnings management can be achieved via the because the Basel Accord allows for the accumulated manipulation of accruals, which according to Martinez provisions in the balance sheet, although limited to a (2001, p. 29) represent “the di erence between net income percentage of risk-weighted assets, to comprise banks’ and net cash  ow”, or via real transactions, such as a regulatory capital.  is allowance occurs only in relation reduction in spending on research and development. to level 2 capital, which is the lowest quality regulatory In the case of provision, earnings management occurs capital and whose value cannot exceed the value of level through accruals manipulation. For Martinez (2001), 1 capital, taken as main capital. when it is le to managers to make accounting choices, they may not limit themselves to accounting facts alone 3.4. Studies Relating the Provision to Economic and use exogenous variables, which will also in uence Cycles the decision.  us, provision will also consider aspects that extrapolate credit quality. Table 1 lists studies involving the pro-cyclicality of the provision in banks. Studies have revealed indications of earnings 100 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 1 Summary of the main articles involving the analysis of the pro-cyclicality of the loan loss provision in  nancial institutions. Author/Period Methodology Results Estimation of an equation in which the provision Strong evidence that the relationship between the is a function of: (i) speci c variables of the provision and the earnings of banks from the G-10 Cavallo and Majnoni banks, (ii) macroeconomic variables, and (iii) countries presents a positive sign. The same sign is Period: 1988 to 1999 institutional variables, with the use of three different negative for banks from countries outside the G-10, which Publication: 2001 techniques (OLS cross section, pooled cross on average present much smaller provisions in good section, and panel  xed effect estimation). periods and are forced to increase them in bad periods. To identify the relationship between the provision Increases in the provision depend a lot on the business and the business cycle, three macroeconomic Bikker and Hu cycle. In bad times, the provisions increase. Banks variables (real GPD growth, unemployment, and Period: 1979 to 1999 contribute signi cantly more to the provisions in in ation) and three speci c variables of the banking Publication: 2002 years of relatively higher earnings (as a precaution sector (loans, net  nancial intermediation earnings, or as a way of smoothing income), leading the and defaults) were used as explanatory variables sector to be less pro-cyclical than it should. for the provision in the linear regression. In the linear regression, the provisions are a result of Many banks delay the recognition of provisions as Laeven and Majnoni the earnings before income tax and of the provisions much as possible, recording them when an economic Período: 1988 to 1999 for losses; of the real growth in loans; of the real contraction has just taken hold and thus increasing the Publication: 2003 growth in GDP per capita; and of the year dummies. impact of economic cycles on earnings and on capital. The provisions of banks are usually substantially greater in Based on the models by Cavallo and Majnoni (2002) periods of lower GDP growth, re ecting the growing risk of Bikker and Metzemakers and by Laeven and Majnoni (2003), the authors their credit portfolios when the economic cycle changes and Period: 1991 to 2001 worked with additional variables such as GDP enters into a downturn. This effect is mitigated by an increase Publication: 2004 growth and dummies for the countries. The use of the in the provision in periods of higher earnings and in those in provision for capital management was also tested. which the growth in the volume of lending becomes greater. There is a positive correlation between the provision and The model is based on the assumption that the GDP. The empirical tests do not support the pro-cyclicality of provision (dependent variable) is a linear function of the provision in banks. The results were different depending Handorf and Zhu two variables: (i) the initial value of the accumulated on the size of the institutions. Medium-sized banks tend Period: 1990 to 2000 provision, net of the falls or losses ex-post of the current to use information on projected losses when de ning the Publication: 2006 period; and of management’s expectations in relation value of the provision (countercyclical attitude). Smaller to future falls, based on current available information. banks and much bigger banks tend towards provision practices that consider current losses (pro-cyclical attitude). The model evaluates whether the evolution of the Setting provisions to cover expected future losses in provision explains the changes in banks’ behavior in loans (“non-discretionary provisions”) causes major relation to lending during the economic cycle. The uctuations in credit. The non-discretionary component linear regression model estimated the discretionary Bouvatier and Lepetit of the provision ampli es the credit cycle: in an economy and non-discretionary components of the provision. Period: 1992 to 2004 rising phase, banks tend to underestimate credit risks, The authors introduced the lagged dependent variable Publication: 2007 indicating the existence of greater incentives for the as an explanatory variable, in order to consider granting of new loans since the costs of lending are a dynamic adjustment in the provision. Capital undervalued. Provisions with managerial purposes management, earnings management, and the signaling (discretionary provisions) do not produce the same effect. of equity robustness are also important variables. The effects of business cycles on the performance of While economic growth is the main driver in the banks’ loans portfolios in developing countries were performance of banks’ credit portfolios, the effects caused by Glen and analyzed using linear and non-linear models. The the interest rates are second order. The relationship between Mondrágon-Vélez provision was the proxy for the portfolio’s performance. the provisions and economic growth is highly linear only Period: 1996 and 2008 GDP and other macroeconomic variables were considered in economic conditions of extreme stress. Greater loss Publication: 2011 as explanatory variables. The interest rate on loans and provisions are related to the private sector leverage level, a set of variables with the individual characteristics to the bad quality of a loans portfolio, and to the absence of the banks of each country were also considered. of penetration and capitalization of the  nancial system. GDP = gross domestic product. Source: Elaborated by the authors. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 101 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil 4. RESEARCH HYPOTHESES e hypotheses to be tested in this study and economic activity, preventing possible contractions in formulated in an alternative form are: the economy from being enhanced, or the e ects of favorable economic cycles from being ampli ed.  ey H : Provisions based on the incurred loss model present are countercyclical in nature. a negative relationship with the variables that represent H : Provisions based on the mixed model adopted in economic activity, contributing towards enhancing Brazil present a positive relationship with the variables possible contractions experienced by the economy, or that represent economic activity, preventing possible towards amplifying the e ects of favorable economic contractions experienced by the economy from being cycles.  ey are pro-cyclical in nature. enhanced, or the e ects of favorable economic cycles H : Provisions based on the expected loss model present a positive relationship with the variables that represent from being ampli ed.  ey are countercyclical in nature. 5. METHODOLOGY such as Cavallo and Majnoni (2001), Bikker and Hu (2002), 5.1. Presentation of the Model and Discussion of the Variables Laeven and Majnoni (2003), Bikker and Metzemakers (2004), Handorf and Zhu (2006), Bouvatier and Lepetit A linear regression model was de ned to evaluate the (2007), and Glen and Mondragón-Vélez (2011). relationship between the variables of interest in order to e model is estimated using the panel data method. identify how the provisions are associated with the variables e empirical tests use the model below, elaborated based that characterize business cycles. Most of the studies relating provisions and economic cycles use this type of model, on the aims and hypotheses of the research: 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 = 𝛽𝛽 +𝛽𝛽 ∆𝐺𝐺𝐺𝐺𝑃𝑃 +𝛽𝛽 𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 𝑃𝑃 +𝛽𝛽 𝑈𝑈 𝐸𝐸𝑃𝑃 �� � � � � � � �� +𝛽𝛽 ∆𝑙𝑙𝑙𝑙𝑙𝑙𝑃𝑃𝐸𝐸𝑈𝑈 +𝛽𝛽 𝑙𝑙𝑃𝑃𝐸𝐸𝑈𝑈 +𝛽𝛽 𝑈𝑈𝑈𝑈 +𝛽𝛽 𝑆𝑆𝑆𝑆𝑆𝑆𝑈𝑈 +𝜀𝜀 � �� � �� � �� � �� �� where: The aim of the model is not to capture a possible PROV = loan loss provision over banks’ average total cause and effect relationship between dependent it assets i in time t. variables and independent variables, but rather, to ΔGDP = real variation in GDP in time t. identify how the dependent variable behaves in relation UNEMP = unemployment rate in time t. to variations in each independent variable, especially EAR = earnings before income tax, pro t sharing, and the macroeconomic ones. it loan loss provisions over banks’ average total assets i in Despite the accounting variables being endogenous, time t. their inclusion in the model is warranted because the ΔlnLOAN = variation in the loan balances of banks i provision is in uenced by prominently accounting factors, it in time t. such as receivables, earnings, and equity. Moreover, these LOAN = balance of credit operations over total assets variables work as control variables, helping to more it of banks i in time t. adequately capture the relationship between the dependent NE = equity over total assets of banks i in time t. and the independent variables, taken as critical to the it ▪ SIZE = size of banks i in time t. study (GDP and unemployment rate). it 102 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Most of the models that study the provision-economic in the balances of credit operations in real terms. In the cycle relationship use the provision as a dependent Brazilian case, the de ator used was the General Market variable. In this study, it represents the net amount of Price Index (IGPM). For Spain and the United Kingdom, provision expenses set in the period, obtained by the the Consumer Price Index (CPI) was used.  e variation di erence between the set and the reverted provisions. in the balances of loans was calculated by the di erence e independent variables can be classi ed into: (i) in the natural logarithms in the following way: ln(credit macroeconomic variables for the countries (real GDP operations /IGPM ) – ln(credit operations /IGPM ). it it it-1 it-1 growth and unemployment rate); (ii) accounting variables  e loans over total assets variable considers the for the banks (earnings before income tax, pro t sharing, exposure of the banks to credit risks, indicating the relative and loan loss provisions over average total assets, variation size of the loans portfolio. It represents the division of the in loan balances, loan balances over total assets, and equity balance of receivables from credit operation accounts by over total assets); and (iii) a control variable (bank size). the value of total assets. e real variation in GDP is a critical variable in  e equity over total assets variable, inserted into the model and considered a more useful indicator for the model to indicate a possible use of the provision as representing the business cycle. Its value represents the a capital management tool, is the result of dividing net percentage variation in GDP at constant prices. equity by total assets. e unemployment rate is another representative metric of  e SIZE variable, de ned as the natural logarithm it the current phase of the economic cycle. In Brazil, it indicates of total assets, de ated by the IGPM or the CPI, aims to the joblessness rate for 10-year-olds upwards, considering only control the size e ects of the institutions. Larger banks, the main metropolitan regions of the country. or those participating in conglomerates, are expected to e earnings over total assets variable, which indicates set more robust provisions. a possible use of the provision for earnings management, is the result of dividing the value of earnings before income 5.2. Summary of the Expected Results tax, pro t sharing, and loan loss provisions by the average Table 2 summarizes the expected results in relation value of total assets. to the behavior and to the sign of the coe cients of the e growth in lending variable, which indicates the explanatory variables, based on the formulated hypotheses. evolution of the banks’ credit risk, represents the variation Table 2 Summary of the expected results in relation to the model explanatory variables. Variable Expected Behavior Sign In countries with provision accounting procedures that are considered pro-cyclical, such as the United Kingdom, a positive variation in - GDP should make the level of the banks’ provision decrease. In countries with provision accounting criteria that are considered countercyclical, such as Spain, a positive variation in GDP + GDP Growth should make the level of the banks’ provision increase. In Brazil, a country with provision accounting criteria that is considered mixed, a positive variation in GDP should make the level of the banks’ provision increase, considering the prevalence of ex-post criteria over ex- ante criteria, which presupposes a prevalently countercyclical nature. In countries with pro-cyclical provision accounting criteria such as the United Kingdom, a positive variation in unemployment + should make the level of the banks’ provision increase. In countries with countercyclical provision accounting criteria, such as Spain, a positive variation in unemployment - Unemployment rate should make the level of banks’ provision decrease. In Brazil, a country with provision accounting criteria that is considered mixed, a positive variation in the unemployment rate should make the level of the banks’ provision decrease, given the prevalence of ex-post criteria over ex-ante criteria, which presupposes a prevalently countercyclical nature. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 103 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil Table 2 Cont. Variable Expected Behavior Sign Considering that banks use the provision as an earnings management tool, Earnings/Assets a positive variation in earnings before income tax, pro t sharing, and loan + loss provisions should make the level of the bank’s provision increase. Considering that banks use the provision as a capital management tool, a negative variation in net equity should make the level of the NE - banks’ provision increase. A bank would be likely to record a higher provision when its regulatory capital was below that required. Considering that a greater share of credit operations in total investments represents an increase in credit risk, a positive Loans/Assets + variation in the relationship between loans and total assets should make the level of commercial banks’ provisions increase. Considering that a positive variation in the volume of credit operations over time Variation in loans represents an increase in credit risk, a positive variation in the volume of credit + operations should make the level of commercial banks’ provisions increase. NE = net equity; GDP = gross domestic product. Source: Elaborated by the authors. studied the risk of a spurious regression was removed. 5.3. Tests Applied and Robustness Procedures e Pearson correlation matrix (tables 3, 4, and 5) proved Adopted that a high correlation (above 0.8) did not exist between the e Im, Pesaran, and Shin – I.P.S, ADF-Fisher, and independent variables in the countries analyzed. PP-Fisher unitary root tests indicated that in the countries Table 3 Pearson Correlation Matrix – Brazil. LOAN ∆lnLOAN EAR NE ∆GDP UNEMP SIZE LOAN 1.000000 - - - - - - ∆lnLOAN 0.095587 1.000000 - - - - - EAR 0.233714 0.024260 1.000000 - - - - NE -0.147020 -0.051504 0.139431 1.000000 - - - ∆GDP -0.011816 0.007476 0.025810 0.017003 1.000000 - - UNEMP -0.057126 0.039506 0.120247 0.045110 0.040903 1.000000 - SIZE -0.097314 0.005905 -0.107682 -0.592854 -0.009121 -0.234355 1.000000 LOAN = share of credit operations in the total assets of Brazilian banks; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over average total assets of the Brazilian banks; NE = equity over total assets of banks i in time t; ∆GDP = real growth in gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. Table 4 Pearson correlation matrix – Spain. LOAN ∆lnLOAN EAR NE ∆GDP UNEMP SIZE LOAN  1.000000 - - - - - - ∆lnLOAN 0.179954  1.000000 - - - - - EAR -0.088175 0.062070  1.000000 - - - - NE -0.113306 -0.026035  0.241673  1.000000 - - - ∆GDP  0.121625 0.119785  0.082047 0.004057  1.000000 - - UNEMP  -0.191935 -0.091091 -0.126566 0.003399 -0.739170  1.000000 - SIZE -0.062023 0.011395 -0.070449 -0.646019 -0.070423 0.059811 1.000000 LOAN = share of credit operations in the total assets of Brazilian banks; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over average total assets of the Brazilian banks; NE = equity over total assets of banks i in time t; ∆GDP = real growth in gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. 104 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 5 Pearson correlation matrix – United Kingdom. LOAN ∆lnLOAN EAR NE ∆GDP UNEMP SIZE LOAN 1.000000 - - - - - - ∆lnLOAN 0.136788 1.000000 - - - - - EAR 0.053496 0.026894 1.000000 - - - - NE 0.114561 -0.055089 0.015586 1.000000 - - - ∆GDP -0.036011 0.110507 -0.035462 -0.003076 1.000000 - - UNEMP 0.012340 -0.155514 -0.030084 0.050401 -0.607156 1.000000 - SIZE -0.279305 0.042143 -0.041180 -0.448594 -0.060661 0.097252 1.000000 LOAN/ASS = share of credit operations in the total assets of Brazilian banks; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over average total assets of the Brazilian banks; NE = equity over total assets of banks i in time t; ∆GDP = real growth in gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. e risk of multicollinearity was analyzed by applying Banks, Multiple Banks with Commercial Portfolio, and Savings the variance in ation test among the explanatory variables, Banks that do not form part of a conglomerate). which did not reveal multicollinearity problems. Ninety-eight institutions were initially considered. e Chow test was carried out to test the existence of Together, these institutions controlled 84.1% of the total assets st individual heterogeneity and con rm whether the use of of the National Financial System on December 31 2012. panel data would apply to the study. In the three countries, Banco Plural, Morgan Stanley, Banco BM&F, Banco the option was to use the regression with individual e ects. Opportunity, BNY Mellon, Western Union, and Banco en the Hausman test was carried out to de ne the best Petra were excluded from the sample as they did not panel data method for the regression estimation. For Brazil, present a credit operations balance in the period analyzed. the result indicated the random e ects model. For Spain Data was extracted from the semi-annual  nancial and the United Kingdom, it was the  xed e ects model. st statements of the chosen institutions on June 30 and In order to analyze the existence of autocorrelation st December 31 between 2001 and 2012, obtained from between the residuals of the regression, the Durbin- the Banco Central do Brasil website. Watson test was used. In the three countries, the value of the statistic was between d and d , indicating that the 5.4.2. In relation to the banks that operate in Spain. l u test was inconclusive. Adopting a conservative position, The sample considered all of the Spanish banks the null hypothesis of the inexistence of autocorrelation associated with the Asociación Española de Banca (AEB). is rejected; that is, the model residuals appear to be Fi y-eight institutions, including  nancial conglomerates autocorrelated. and individuals banks, were considered in the study. In light of the possibility of the existence of cross-sectional autocorrelation of the residuals, the use of the SUR cross-sectional 5.4.3. In relation to the banks that operate in the United standard errors method (PCSE) in the model estimation arose as Kingdom. an alternative to the problem, enabling the generation of robust The selection of banks that operate in the United parameters even in the presence of residues autocorrelation. Kingdom was based on the publication “List of Banks as e overall signi cance of the model was proven using Compiled by the Bank of England on 31 March 2013” , which the F test. covers the list of banks under the supervision of the Bank of England, available from the website www.bankofengland. 5.4. Sample De nition and Description of the co.uk.  e list is comprised of 153  nancial institutions, with Data Source 45 banks being selected to compose the sample (29.41% of the supervised institutions).  e sample prioritized st 5.4.1. In relation to the banks that operate in Brazil. institutions closing their  nancial year on December 31 and which presented their information in identical monetary In Brazil, the sample considered the institutions that bases (pounds sterling).  e information was obtained formed part of Consolidated Banking I, formed by Banking from the Company Check website (www.companycheck. Conglomerate I (composed of at least one institution of the Commercial Bank or Multiple Bank with Commercial Portfolio co.uk), which provides the annual  nancial statements of type) and Independent Banking Institutions I (Commercial banks operating in the United Kingdom. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 105 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil 6. RESULTS set, on average, provisions in percentages equal to 0.83%, 6.1. Descriptive Statistics of the Dependent Variable 0.2668%, and 0.3607% of total assets, respectively.  e standard deviations show a large variability in the e descriptive statistics of the dependent variable are provision, perhaps caused by the reversion mechanism, found in Table 6. From the second semester of 2001 to and indicate that the frequency distributions of the the second semester of 2012, the commercial banks that operated in Brazil, in Spain, and in the United Kingdom provision in the three countries have a similar behavior. Table 6 Descriptive statistics of the dependent variable in the model – Brazil, Spain, and the United Kingdom, from 2001 to 2012. Provisions/Total Assets Brazil Spain United Kingdom Mean 0.0083 0.0027 0.0036 Median 0.0044 0.0008 0.0013 Maximum Value 0.1643 0.0858 0.1432 Minimum Value -0.0519 -0.0337 -0.0407 Standard Deviation 0.0146 0.0071 0.0102 Source: Elaborated by the authors. and the United Kingdom, which were around 82% and 6.2. Descriptive Statistics of the Independent 71%, respectively. Variables e average share of own resources over the total investments of the commercial banks has shown to be 6.2.1. Brazil. quite signi cant, at around 23%, and more comfortable According to Table 7, on average only 34.34% of than that of the banks that operate in Spain (average of the resources were used for credit operations, a lower 20%) and much higher than that of the banks that operate percentage than those of the banks that operate in Spain in the United Kingdom (average of 13.8%). Table 7 Descriptive statistics of the independent variables in the model – Brazil, 2001-2012. MEASURE LOAN ∆lnLOAN NE EAR ∆GDP UNEMP SIZE Mean 0.3434 0.1272 0.2320 0.0225 0.0167 0.0855 13.999 Median 0.3309 0.1268 0.1551 0.0184 0.0196 0.0830 13.922 Maximum Value 1.0344 5.6920 0.9999 0.3568 0.0474 0.1300 20.8069 Minimum Value 0.0000 -5.2298 -0.1038 -0.1831 -0.0279 0.0470 8.2451 Standard Deviation 0.2356 0.5020 0.2154 0.0326 0.0174 0.0222 2.3498 LOAN = share of credit operations in the total assets of the Brazilian banks; ∆lnLOAN = real growth in the balances of credit operations recorded in the balance sheets of the Brazilian banks; NE = share of equity over total assets of the Brazilian banks; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of the Brazilian banks; ∆GDP = real growth in the gross domestic product of Brazil; UNEMP = unemployment rate in Brazil; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market prices index. Source: Elaborated by the authors. 6.2.2. Spain. in Spain.  e average return on assets was around 0.53%, Table 8 presents the descriptive statistics of the below the returns of the banks that operate in Brazil independent variables in relation to the banks that operate (2.25%) and in the United Kingdom (1.91%). 106 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 8 Descriptive statistics of the independent variables in the model – Spain, 2001-2012. MEASURE LOAN ∆lnLOAN NE EAR ∆GDP UNEMP SIZE Mean 0.8195 0.0534 0.1998 0.0053 0.0078 0.1362 14.0939 Median 0.8793 0.0354 0.0808 0.0036 0.0140 0.1130 13.9648 Maximum Value 1.0766 8.0520 0.9999 0.3045 0.0201 0.2510 20.9620 Minimum Value 0.0000 -8.0301 -0.0315 -0.1705 -0.0278 0.0810 0.0810 Standard Deviation 0.1905 0.4701 0.2745 0.0242 0.0131 0.0527 2.4123 LOAN = share of credit operations in the total assets of the Spanish banks; ∆lnLOAN = real growth in the balances of credit operations recorded in the balance sheets of the Spanish banks; NE = share of equity over total assets of the Spanish banks; EAR = share of earnings before income tax, pro t sharing, and loan loss provision over the average total assets of the Spanish banks; ∆GDP = real growth in the gross domestic product of Spain; UNEMP = unemployment rate in Spain; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. 6.2.3. United Kingdom. e descriptive statistics of the independent variables related to the banks that operate in the United Kingdom are presented in Table 9. Table 9 Descriptive statistics of the independent variables in the model – United Kingdom, 2001-2012. MEASURE LOAN ∆lnLOAN NE EAR ∆GDP UNEMP SIZE Mean 0.7076 0.1129 0.1380 0.0191 0.0150 0.0600 13.2893 Median 0.7453 0.0900 0.0935 0.0086 0.0224 0.0533 12.8845 Maximum Value 1.3041 6.5087 0.9885 5.5899 0.0395 0.0807 20.6667 Minimum Value 0.0000 -2.0549 0.0000 -0.1625 -0.0517 0.0467 3.98689 Standard Deviation 0.2353 0.4501 0.1439 0.2472 0.0242 0.0140 2.65957 LOAN = share of credit operations in the total assets of the British banks; ∆lnLOAN = real growth in the balances of credit operations recorded in the balance sheets of the British banks; NE = share of equity over total assets of the British banks; EAR = share of earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of the British banks; ∆GDP = real growth in the gross domestic product of the United Kingdom; UNEMP = unemployment rate in the United Kingdom; SIZE = size of the commercial bank, represented by the natural logarithm of total assets de ated by the general market price index. Source: Elaborated by the authors. 6.3. Test of the Hypotheses independent variables are associated with 38.87% of the behavior of the dependent variable.  e F statistic, with 6.3.1. H1: Banks that operate in the United Kingdom a p-value of 0.0000, con rms the statistical signi cance e coe cient of determination (R ) indicates that the of the model (Table 10). Table 10 Coef cients of Determination and F Statistic of the model – United Kingdom. Variables Value R 0.3887 R Adjusted 0.3191 F Statistic 5.5863 P-Value (F) 0.0000 Source: Elaborated by the authors. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 107 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil The macroeconomic variables and the specific loan loss provisions over total assets” variable was also accounting variables of the banks, except GDP, do not not found to be signi cant, indicating that variations in have signi cant e ects on the provision in the United earnings would not explain the behavior of the provision Kingdom.  e “loans over total assets” and “growth in and going against the expectations of the research. lending” variables were not found to be signi cant, thus Earnings management via provisions does not appear going against the expectations presented in Table 2. to be a common practice among banks in the United e “earnings before income tax, pro t sharing, and Kingdom. Table 11 Regression results – United Kingdom. Variable Coef cient Standard deviation T statistic Prob. C 0.0114 0.0074 1.5429 0.1236 LOAN -0.0009 0.0023 -0.3715 0.7105 ∆lnLOAN -0.0006 0.0019 -0.3013 0.7633 EAR -0.0022 0.0053 -0.4124 0.6803 NE -0.0052 0.0072 -0.7189 0.4726 ∆GDP -0.0553 0.0140 -3.9372 0.0001 UNEMP 0.0505 0.0270 1.8662 0.0627 SIZE -0.0007 0.0005 -1.2307 0.2191 C = constant in the regression; LOAN = balances of credit operations of banks i in time t; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of banks i in time t; NE = equity over the total assets of banks i in time t; ∆GDP = real growth of gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of banks i in time t. Source: Elaborated by the authors. e “net equity over total assets” variable was not and Metzemakers (2004). found to be signi cant, thus going against the capital  e “size” control variable was also not found to be management hypothesis. signi cant. At a 1% level of signi cance, the coe cient of the “GDP 6.3.2. H2: Banks that operate in Spain. growth” variable was found to be signi cantly negative, With regards to the banks that operate in Spain, indicating pro-cyclical behavior of the banks in the United the coe cient of determination (R ) indicates that the Kingdom, in line with hypothesis H1. independent variables are associated with 38.31% of e “unemployment rate growth” variable was only the dependent variable behavior.  e F statistic, with a found to be signi cant at a 10% level of signi cance, p-value of 0.0000, con rms the statistical signi cance of thus going against the expectations.  is result, however, the model (Table 12). supports the  ndings of previous studies, such as Bikkers Table 12 Coef cients of determination and F statistic of the model – Spain. Variables Value R 0.3831 R Adjusted 0.3436 F Statistic 9.7147 P-Value (F) 0.0000 Source: Elaborated by the authors. e “loans over total assets”, “earnings over total assets”, signi cance, as can be observed in Table 13.  e other “GDP growth” and “size” variables have signi cant e ects variables were not signi cant. on loan loss provisions in Spain, at di erent levels of 108 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Table 13 Regression results – Spain. Variable Coef cient Standard deviation t-statistic Prob. C -0.0149 0.0074 -2.0191 0.0438 LOAN 0.0031 0.0017 1.8418 0.0658 ∆lnLOAN -0.0004 0.0003 -1.2059 0.2281 EAR -0.1004 0.0170 -5.9089 0.0000 NE -0.0020 0.0052 -0.3825 0.7022 ∆GDP -0.0534 0.0192 -2.7727 0.0057 UNEMP 0.0040 0.0049 0.8224 0.4111 SIZE 0.0011 0.0004 2.5368 0.0113 C = constant in the regression; LOAN = balances of credit operations of banks i in time t; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of banks i in time t; NE = equity over the total assets of banks i in time t; ∆GDP = real growth of gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of banks i in time t. Source: Elaborated by the authors. e “loans over total assets” variable was not found to signi cant, thus going against the expectation with regards be signi cant, at a 10% level of signi cance.  e positive to the use of the provision for earnings management. sign of the coe cient supports the expectation presented Contrary to the expectation of hypothesis H2, at a 1% in Table 2, indicating that the provision tends to rise when level of signi cance the coe cient of the “GDP growth” the share of loans over total assets increases. variable was found to be signi cantly negative, indicating Contrary to expectations, the “growth in lending” pro-cyclical behavior. variable was not found to be signi cant, indicating that  e “unemployment rate” variable was not found the variation in the volume of lending does not explain to be signi cant, going against expectations.  e “size” the behavior of the provisions. control variable was found to be signi cant, with a positive The “earnings before income tax, profit sharing, sign, indicating that the larger the bank, the higher the and loan loss provisions over total assets” variable was provision tends to be. not found to be signi cant in Spain, at a 1% level of 6.3.3. H3: Banks that operate in Brazil. signi cance.  e negative sign of the coe cient indicates e coe cient of determination (R ) indicates that the growth of the provision whenever earnings decrease and independent variables of the model are associated with vice-versa, thus going against expectations.  ere is no 15.65% of the dependent variable behavior (Table 14). evidence of earnings management practices via provisions e F statistic, which presents a p-value equal to 0.0000, in the Spanish banks. con rms the statistical signi cance of the model. The “net equity over total assets” variable is not Table 14 Coef cients of determination and F statistic of the model – Brazil. Variables Value R 0.1565 R Adjusted 0.1531 F Statistic 45.1995 P-Value (F) 0.0000 Source: Elaborated by the authors. According to the data in Table 15, the macroeconomic studied in Brazil, with the exception of the unemployment variables and the speci c accounting variables of the banks rate and size, have signi cant e ects on the provision. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 109 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil Table 15 Regression results – Brazil. Variable Coef cient Standard deviation t-statistic Prob C -0.0086 0.0080 -1.0795 0.2805 LOAN 0.0268 0.0026 10.4186 0.0000 ∆lnLOAN -0.0014 0.0007 -1.8981 0.0579 EAR 0.0868 0.0170 5.1125 0.0000 NE 0.0138 0.0045 3.0866 0.0021 ∆GDP -0.0477 0.0186 -2.5637 0.0104 UNEMP -0.0173 0.0185 -0.9356 0.3496 SIZE 0.0004 0.0004 0.9305 0.3523 C = constant in the regression; LOAN = balances of credit operations of banks i in time t; ∆lnLOAN = variation in the balances of credit operations of banks i in time t; EAR = earnings before income tax, pro t sharing, and loan loss provisions over the average total assets of banks i in time t; NE = equity over the total assets of banks i in time t; ∆GDP = real growth of gross domestic product in time t; UNEMP = unemployment rate in time t; SIZE = size of banks i in time t. Source: Elaborated by the authors. e “loans over total assets” variable was found to sign of the coe cient would go against the expectation of be signi cant at a 1% level of signi cance. As expected, use of the provision for capital management; it is supposed when a bank increases its share of credit operations in that the banks raise the provision whenever their capital total investments, the percentage of the provision also levels are more comfortable. It is assumed that the fact that increases, con rming the claim of growth in risks in the banks in Brazil have historically presented a higher phases of economic expansion.  is prudent behavior of reference equity than that required by the Central Bank the banks contributes to attenuating the e ects caused and the absence itself of the generic provision instrument by possible pro-cyclical behaviors. in the country’s regulation explain this result. e “growth in lending” variable was only found Regarding hypothesis H3, at a 1% level of signi cance to be significant at a 10% level of significance. The the coe cient of the “GDP growth” variable was shown to coe cient presented a negative sign, indicating that the be signi cantly negative, indicating pro-cyclical behavior. provisions usually decrease when the balance of credit  is appears to indicate the absence of an e cient risk operations increases.  is  nding would go against the evaluation mechanism with a forward-looking character. initial expectation, which indicated a probable positive With regards to the “growth in the unemployment relationship between growth in the lending volume and rate” variable, this was not found to be signi cant, thus the provision. going against the expectation presented in Table 2.  e e “earnings before income tax, pro t sharing, and same occurs with the “size” control variable. It cannot be loan loss provisions over total assets” variable was also claimed that larger banks in Brazil, or those belonging to found to be signi cant at a 1% level of signi cance, conglomerates, set more robust provisions than the rest. indicating that variations in earnings are related with As a sensitivity analysis, the lagged GDP and the behavior of the provision.  e positive sign of the unemployment rate variables were added to the models. coe cient indicates growth in provision levels whenever In Brazil and the United Kingdom, these variables were earnings increase and vice-versa, thus supporting the not found to be signi cant, indicating that the e ect expectation presented in Table 2.  e result indicates the of their behavior on the provision calculation occurs use of the provision for earnings management, a practice in the same accounting period. With regards to Spain, which ameliorates the pro-cyclical e ects of the provision. the unemployment rate variable was not found to be e “net equity over total assets” variable was shown significant, but the GDP variable was shown to be to be signi cant at a 1% level of signi cance.  e negative signi cant at a 1% level of signi cance. 7. FINAL REMARKS Studies such as those by Bikker and Hu (2002), Laeven the incurred loss model leads to pro-cyclical behavior and Majnoni (2003), and Bikker and Metzemakers (2004) of the provision and the expected loss model results in reveal that the choice between di erent accounting models countercyclical behavior of the provision.  is study would determine the behavior of the provision with proposed to investigate whether the incurred loss models, relation to the business cycles.  e expectation is that which are conceptually pro-cyclical, would lead banks to 110 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo setting pro-cyclical provisions, and whether the expected di culty itself of predicting future economic scenarios, loss models, which are conceptually countercyclical, would among other factors, could be leading to provisions of a lead banks to recognizing countercyclical provisions. di erent nature from that which the legislator intended Two countries were intentionally chosen to represent when establishing more speci c criteria for setting them. these accounting models: Spain (expected loss) and Only in relation to the United Kingdom was the hypothesis the United Kingdom (incurred loss). A mixed model con rmed that the accounting model adopted, based adopted in Brazil was also analyzed.  e prevalence of essentially on the IASB rules, has in fact caused provisions typical procedures of expected loss models reinforced that are pro-cyclical in nature. the hypothesis that the Brazilian model would be Also in relation to the United Kingdom, only the GDP countercyclical. growth variable was found to be signi cant, at a 5% level of Based on the linear regression models used in previous signi cance. It is assumed that the variables that represent studies, especially those by Cavallo and Majnoni (2001), credit risk, such as “loans over total assets” and “growth Bikker and Hu (2002), Laeven and Majnoni (2003), and in lending” were not found to be signi cant due to the Bikker and Metzemakers (2004), variables were selected characteristics of the accounting model itself adopted in to compose an econometric model that identi ed the that country, which only recognizes the provision when relationship between the provision and the economic objective evidence of loss is identi ed.  us, the impact cycles in the three countries selected. of the two variables would only occur indirectly (a greater Two variables were inserted into the model in order share of loans over assets or growth in the volume of to identify the relationship between the provision lending can lead to a higher probability of occurrence of and economic cycles: the real variation in GDP and objective evidence of loss in the future).  ese variables the unemployment rate. Contrary to expected, the can, therefore, change without the banks in the United unemployment rate was not found to be statistically Kingdom having adjusted their provisions. In the United signi cant in Brazil or in Spain. In United Kingdom this Kingdom, there was also no evidence found that the banks variable was found to be signi cant, at a 10% signi cance are using the provision for earnings management, thus level, presenting a positive coe cient, which indicates that contradicting the study by Silva (2016), or for capital the provision tends to grow when the unemployment rate management. increases. It bears mentioning that studies such as those With regards to earnings management, this is by Bikker and Metzemakers (2004) have already revealed assumed to be a reality for commercial banks in Brazil, that the unemployment rate would not be the best proxy thus supporting the results of the studies by Macedo to evaluate the relationship between the provision and and Kelly (2016), Bortoluzzo, Sheng, and Gomes (2016), economic cycles, a fact that is supported by this study. and Silva (2016).  e “earnings before income tax, pro t e variable unanimously taken in the studies involving sharing, and loan loss provisions” variable was found to be cyclicality is the variation in GDP.  e  ndings of this signi cant, at a 1% level of signi cance, having presented research are consistent with these studies, as GDP was a positive coe cient indicating that the provision tends also found to be a statistically signi cant variable at a to grow when the banks’ earnings increase. In relation to level of signi cance of practically 1% for each accounting Spain, despite the signi cance at a 1% level of signi cance, model that was the object of this research. this variable presented a negative sign, signaling that the With regards to the sign of the coe cient of the variable provision decreases when earnings increase and thus that re ects GDP, two of the formulated hypotheses were contradicting the study by Silva (2016). not supported.  e biggest surprise was in Spain, whose Earnings management practices, which are accounting model prescribes setting forward-looking reprehensible in some aspects, can represent a counterpoint provisions, including provisions of a generic nature, to setting pro-cyclical provisions, easing their e ects on which should cause countercyclical provisions.  e sign the current phase of the business cycle. of the coe cient of the variable, which would presumably Capital management practices were also not found to be positive, was found to be negative, indicating pro- be a reality for the commercial banks in Brazil and Spain. cyclicality. In Brazil, despite the general rule being to set In Brazil, despite the signi cance at a level of 1%, the forward-looking provisions, and although backward- aforementioned variable presented a positive coe cient, looking previsions ultimately prevail when the delays in indicating that the greater the share of net equity over total the payments of principal and interest are present, the assets, the greater the provision set by commercial banks. study also revealed pro-cyclical behavior. It is inferred  is situation may be indicating that the comfortable level that the subjective character of the provision and the of capitalization presented by the banks in Brazil (23.20% R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 111 The cyclicality of loan loss provisions under three different accounting models: the United Kingdom, Spain, and Brazil of total assets, on average) is inducing an increase in the It was also found that in Brazil and the United Kingdom provision level whenever net equity presents a positive the e ect of economic variations on the provision behavior variation, which represents a conservative procedure takes place in the same accounting period. In the case of that also serves as a counterpoint to setting pro-cyclical Spain, however, a variation in GDP in the previous period provisions. In Spain, the “net equity over total assets” was found to be a signi cant variable. variable was not found to be signi cant, indicating that With relation to the econometric model chosen for the provision level is not a ected by the behavior of the the research, it was found that the peculiar nature of equity of commercial banks that operate in that country. the provision rules in the United Kingdom, in relation The “size” variable was found to be statistically to those in Brazil and Spain, may be determining the signi cant only in relation to the banks that operate in di erence in the levels of signi cance of the explanatory Spain. In the other countries, the size of the banks does variables for the model, when this is applied to commercial not appear to in uence the provision level. banks in the country that adopts the IASB accounting e results support the previous studies by Bikker standards.  us, it is assumed that adaptations are needed and Hu (2002), Laeven and Majnoni (2003), and Bikker to adjust the model used to the speci cities of countries and Metzemakers (2004), in the sense that the provision that use relatively di erent accounting rules for setting behavior strongly depends on the economic cycle, the provision. indicating that the provision generally grows in poor As an intentional sample is concerned, which considers economic times. Another similarity between the results of only one country for each accounting model, the results this research and the aforementioned studies is the  nding may not be generalized. For future research, studies with that the e ect of pro-cyclicality is usually mitigated by an di erent econometric models are suggested, incorporating increase in the provision in periods of higher earnings variables not considered in this research, especially (earnings management practices). those related to the delay in the payment of principal A possible assumption regarding the fact that and interest and to the accounting recognition of losses. the research revealed pro-cyclical behavior when Evaluating the question of cyclicality by considering countercyclical behavior was expected is that the banks di erent accounting scenarios in the same country (for may be using discretion and managing their earnings example, studying cyclicality before and a er the advent via the provision and more than compensating for the of Resolution n. 2,682/1999) could also prove to be an expected e ect of the models, which is a hypothesis to interesting experiment for re ecting on the provision be investigated in future studies. behavior in di erent regulatory environments. REFERENCES Bebczuk, R., Carrera, J., & Sangiácomo, M. (2010). A provisioning behaviour and procyclicality. Journal of new look into credit procyclicality: International International Financial Markets, Institutions & Money, panel evidence. In BIS Consultative Council for the 209, 1-17. doi: 10.1016/j.int n.2004.03.004. Americas Conference on “systemic risk, bank behavior Bortoluzzo, A. B., Sheng, H. H., & Gomes, A. L. P. and regulation over the business cycle”. Buenos Aires, (2016) Earning management in Brazilian  nancial Argentina. institutions. Revista de Administração da Universidade Betancourt, L., & Baril, C. P. (2009, October/November). de São Paulo, 51(2), 182-197. ISSN 0080-2107. doi: Accounting for loan losses: Is Dynamic provisioning 10.5700/rausp1233. the answer? Bank Accounting Finance, 22(6), 9-16. Bouvatier, V., & Lepetit, L. (2007). Banks’ procyclical Retrieved from http://connection.ebscohost.com/c/ behavior: Does provisioning matter? Journal of articles/44680540/accounting-loan-losses-dynamic- International Financial Markets, Institutions & Money, provisioning-answer. 18, 513-528. doi: 10.1016/j.int n.2007.07.004. Bikker, J. A., & Hu, H. (2002). Cyclical patterns in pro ts, Bouvatier, V., & Lepetit, L. (2012, July). Provisioning rules provisioning and lending of banks and procyclicality and bank lending: A theoretical model. Journal of of the new basel capital requirements. Banca Nazionale Financial Stability, 8(2), 25-31. Retrieved from http:// del Lavaro Quarterly Review, 55(221), 143-175. ssrn.com/abstract=1148314. Retrieved from https://www.dnb.nl/binaries/ot039_ Burns, A. F., & Mitchell, W. C. (1946). Measuring business tcm46-146052.pdf. cycles. NBER Book Series Studies in Business Cycles, Bikker, J. A., & Metzemakers, P. A. J. (2004, March). Bank New York: National Bureau of Economic Research. 112 R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 Antônio Maria Henri Beyle de Araújo, Paulo Roberto Barbosa Lustosa & Edilson Paulo Cavallo, M., & Majnoni, G. (2001, June). Do banks openview/ee52c0978ed6e519591 513944a8a54/1?pq- provision for bad loans in good times? Empirical origsite=gscholar&cbl=26518. evidence and policy implications. Working Paper Harrald, P., & Sandal, T. (2010, October). Addressing pro- Nº 2.619. World Bank Policy Research. Retrieved cyclicality. Asia-Risk, 70-72. Retrieved from www.risk. from https://openknowledge.worldbank.org/ net. handle/10986/19607. Healy, P. M., & Wahlen, J. M. (1999, December). A Cohen, D. A, & Zarowin, P. (2007, August). Earnings review of the earnings management literature and its management over the business cycle. New York implications for standard setting. Accounting Horizons, University. Stern School of Business. Retrieved 13(4), 365-383. Retrieved from https:// sher.osu. from http://w4.stern.nyu.edu/emplibrary/ edu/~young.53/Healy-Wahlen. EM_08_23_07FINAL.pdf. Hendriksen, E. S., & Van Breda, M. F. (1999). Teoria da Cortavarria, L., Dziobek, C., Kanaya, A., & Song, I. contabilidade. São Paulo, SP: Atlas. (2000, December). Loan review, provisioning, and Laeven, L., & Majnoni, G. (2003, April). Loan loss macroeconomic linkages. WP/00/195. IMF Working provisioning and economic slowdowns: too much, too Paper. International Monetary Fund. Retrieved late? Journal of Financial Intermediation, 12(2), 178- from http://www.imf.org/external/pubs/ /wp/2000/ 197. Retrieved from http://www.sciencedirect.com/ wp00195.pdf. science/article/pii/S1042957303000160. Curcio, D., & Hasan, I. (2015). Earnings and Longbrake, W. A., & Rossi, C. V. (2011, July). Procyclical capital management and signaling:  e use of versus countercyclical policy e ects on  nancial markets. loan-loss provisions by European banks.  e Study prepared for the Anthony T. Clu Fund.  e European Journal of Finance, 21(1), 26-50. doi: Financial Services Roundtable. Financing America’s 10.1080/1351847X.2012.762408. Economy. Retrieved from http://www.fsround.org/fsr/ Glen, J., & Mondragón-Vélez, C. (2011, January). Business pdfs/clu /CountercyclicalPE.pdf. cycle e ects on commercial bank loan portfolio Macedo, M. A. S., & Kelly, V. L. A. (2016, May/August). performance in developing economies. Review of Development Finance, 1(2), 150-165. Retrieved from Gerenciamento de resultados em instituições http://papers.ssrn.com/sol3/papers.cfm?abstract_ nanceiras no Brasil: uma análise com base em id=1754672. provisões para crédito de liquidação duvidosa. Revista Evidenciação Contábil & Finanças, 4(2), 82-96. Gonzales, J. E. N. (2009, July).  e fundamentals of procyclicality of the  nancial system. Economic Martinez, A. L. (2001). “Gerenciamento” dos resultados Newsletter. Bangko Sentral NG Pilipinas, 9(3). contábeis: estudo empírico das companhias abertas Retrieved from http://www.bsp.gov.ph/downloads/ brasileiras (Ph.D. thesis). Faculdade de Economia, EcoNews/EN09-03.pdf. Administração e Contabilidade, Universidade de São Paulo, São Paulo, SP. Gray, R. P., & Clarke, F. L. (2004). A methodology for calculating the allowance for loan losses in commercial Silva, C. A. M. (2016). Utilização da provisão para créditos banks. Abacus, 40(1), 321-341. de liquidação duvidosa para  ns de gerenciamento de Handorf, W. C., & Zhu, L. (2006, spring). US Bank resultado nas instituições  nanceiras brasileiras e luso- loan-loss provisions, economic conditions, and espanholas (Master’s thesis). Faculdade de Economia, regulatory guidance. Journal of Applied Finance, 16(1), Administração e Contabilidade, Universidade de 97-114. Retrieved from http://search.proquest.com/ Brasília, Brasília, DF. R. Cont. Fin. – USP, São Paulo, v. 29, n. 76, p. 97-113, jan./abr. 2018 113

Journal

Quantitative FinancearXiv (Cornell University)

Published: Jul 17, 2019

There are no references for this article.