Latin American Bank Capital Buffers and the Business Cycle: Are They Pro-Cyclical?

Transcription

1 Latin American Bank Capital Buffers and the Business Cycle: Are They Pro-Cyclical? Oscar A. Carvallo V. Deputy Manager for Financial Research CEMLA V Reunión de Estabilidad Financiera Banco Central de la República Argentina July 23 y 24, 2015

2 Preliminary comments This is a regional question but to answer we have to think about it globally This presentation is based on: Carvallo and Jimenez (2015): Bank Capital Buffers and Procyclicality, Chapter on CEMLA 2014 Publication: Latin American and Caribbean Research Network 2014: Monetary Policy and Financial Stability, on print. Carvallo O., A. Kasman and S. Kontbay-Busun (2015): The Latin American bank capital buffers and business cycle: Are they pro-cyclical? Int. Fin. Markets, Inst. and Money 36. Carvallo, Ortiz and Jimenez (in progress): Capital buffers around the World: Cyclicality and Determinants. Opinions and views are own and do not reflect CEMLA s positions on matters

3 The Problem When banks fail to accumulate capital buffers in times of economic booms, they could be trapped with an insufficient level of capital during an economic downturn Under these circumstances banks are forced to deleverage assets and reduce their lending to the market in order to meet the regulatory minimum capital requirements in busts since it is costlier to raise capital through new equities Hence, the cyclical behavior of capital buffers amplifies the impact of shocks on economic stability through reduced lending (Repullo and Suarez, 2009; Borio and Zhu, 2008). Procyclicality

4 The Regulatory Response Aiming to prevent these destabilizing cyclical impacts of capital buffers fluctuations, Basel III requires banks to increase capital buffers during economic booms: through a "mandatory capital conservation buffer" of 2.5% and through a "discretionary counter-cyclical buffer" of up to another 2.5% in times of credit booms. These proposals were calibrated against BIS membership, but are buffers procyclical, globally?

5 Some work on global evidence Ayuso, et al. (Spain, 2004), Lindquist (Nroway, 2004), Bikker and Metzemakers (WW, 2004), Stoltz and Wedow (Germany, 2005), Garcia-Suaza, Gomez-Gonzalez, Pabon and Tenjo-Galarza (Colombia, 2012), Tabak (Brasil, 2011) and Shim (US, 2012) find evidence in favor of counter-cyclical fluctuation of capital buffers in advanced economies. On the other hand, Jakipii and Milne (RAM, EU15, EU25, 2009) study commercial, savings, and co-operative banks separately, as well as small and large banks, and find that the capital buffers of different banks exhibit different cyclical behaviors. Their results show that the capital buffers of commercial, savings, and large banks fluctuate counter-cyclically, while those of co-operative and smaller banks fluctuate pro-cyclically. Fonseca and Gonzales (2010) find differentiated patterns in the levels of capital buffer holdings across and within developed and developing countries. Carvallo, Kasman and Kontbay-Busun 2015: Latam different cyclical patterns, country regressions with bank specific variables

6 First: Global View

7 Bankscope data: Bank specific variables Varible Obs Mean Std. Dev. Banks (N) BUF 40, ,214 Size 160, ,216 Return On Avg Assets 99, ,866 Loan Loss Res / Gross Loans 67, ,564 Equity / Total Assets 99, ,872

8 Capital Buffers globally BUF AFRICA EUREAST EURWEST FAREASTCA MIDEAST NORTHAMER OCEANIA SOUTHAMER Region Capital buffer (BUF): Following Jokipii and Milne (2009), capital buffer is defined as the amount of actual capital ratio of banks exceeding the regulatory minimum level of capital. The actual capital ratio of banks is measured by the variable Total Capital Ratio, from Bankscope. The minimum capital ratio for each country is obtained from the World Bank s Banking Regulation Survey for 2000, 2003, 2007 and 2012.

9 Bankscope data: Bank specific variables Specialization Freq Percent Cum Banks (N) Commercial 226,493 71% 71% 17,447 Cooperative 46,432 15% 86% 3,575 Islamic 2,473 1% 86% 191 Real Estate 5,679 2% 88% 439 Savings 37,948 12% 100% 2,920 Total 319, % 24,572

10 Bankscope data: Bank specific variables RegionN Freq Percent Cum Banks (N) AFRICA 10,091 3% 3% 781 EUREAST 25,730 8% 11% 1,981 EURWEST 91,660 29% 40% 7,055 FAREASTCA 28,191 9% 49% 2,178 MIDEAST 4,050 1% 50% 313 NORTHAMER 139,895 44% 94% 10,763 OCEANIA 1,676 1% 94% 129 SOUTHAMER 17,732 6% 100% 1,372 Total 319, % 24,572

11 Environmental variables Global Financial Development Database of the World Bank: Financial development and competition variables. World Bank s Banking Regulation Surveys for 2000, 2003, 2007 and 2012.

12 Environmental variables The Boone (2001, 2008) Indicator Boone s model is based on the idea that competition enhances the performance of efficient banks and weakens the less efficient ones. This effect is stronger the higher the competition in the market is.

13 Environmental variables Boone Argentina Bolivia Brazil Colom bia Cos ta Rica Dom inican Republic Ecuador El Salvador Honduras Jam aica Panam a Paraguay Peru Trinidad and Tobago Venezuela From: Kasman and Carvallo 2014: Financial Stability, Competition and Efficiency in Latin American and Caribbean Banking, JAE Volume XVII, issue November, 2014.

14 Environmental variables OSP AFRICA FAREASTCA MIDEAST NORTHAMER OCEANIA SOUTHAMER EUREAST EURWEST Region Official Supervisory Power: Whether the supervisory authorities have the authority to take specific actions to prevent and correct problems.

15 Environmental variables NORTHAMER OCEANIA PMI AFRICA EUREAST EURWEST FAREASTCA MIDEAST SOUTHAMER Region Private Monitoring Index: Measures whether there incentives/ability for the private monitoring of firms, with higher values indicating more private monitoring.

16 Environmental variables ORBA AFRICA EUREAST EURWEST FAREASTCA MIDEAST NORTHAMER OCEANIA SOUTHAMER Region Overall Restrictions on Banking Activities: Sum of i)securities Activities, defined as the extent to which banks may engage in underwriting, brokering and dealing in securities, and all aspects of the mutual fund industry; ii)insurance Activities, which is the extent to which banks may engage in insurance underwriting and selling: and iii)real Estate Activities, defined as the extent to which banks may engage in real estate investment, development and management.

17 Other variables BACC LFBEO Bank Accounting Limitations on Foreign Bank Entry/Ownership Whether the income statement includes accrued or unpaid interest or principal on nonperforming loans and whether banks are required to produce consolidated financial statements. Whether foreign banks may own domestic banks and whether foreign banks may enter a country's banking industry. FID Funding with Insured Deposits The degree to which moral hazard exists. FOB Foreign-Owned Banks The extent to which the banking system's assets are foreign owned.

18 Variables and working hypotheses Business Cycle (CYCLE): The business cycle is approximated by the GDP growth variable. Testing for cyclical fluctuation Bank Size (SIZE): The natural logarithm of total assets is used to control for bank size in the regression. Associated Hyp..: too big to fail Profitability (ROA): The return on assets is used to control the impact of profit on buffer. Source of internal funds for buffers. Loan loss reserve (LLR): The loan loss reserve ratio is a quality indicator of the loan portfolios of banks as it measures the amount of reserves banks hold to cover potential losses in their loan portfolios. Effect of risk on buffers.

19 Variables and working hypotheses Competitiveness (Boone): banks with market power try to preserve the charter value of their institutions (Keeley 1990), become more risk adverse and maintain more buffers. L.BUF: the lags values of buffers indicate adjustment costs. Financial development influence the access to capital resources. The regulatory variables are included to capture incentive schemes across counties: Private monitoring and official supervisory power potentially complement and substitute, the stringency of capital regulations support buffer formation. Restrictions on bank activities can have a ambiguous effect. Partial adjustment model. To control for endogeneity and individual effects use Arellano and Bover (1995)/Blundell and Bond (1998) system GMM (xtabond2).

20 Global Results

21 Global Results: Buf determinants Variable Model1 (YD) Model2 (YD) Model3 (YD) Model4 (YD) Model5 (YD) GDP -.108** ** -.106*** -.053* BUF L1..414***.465***.569***.453***.522*** BUF L * -.062*** Size -.709*** -.577*** -.552*** -.619*** -.642*** Return On Avg Assets.633***.44***.65***.473*** Loan Loss Res / Gross Loans.25* ***.069* Boone indicator 2.31* 2.35** 2.26*** 2.51** 1.73* Overall Capital Stringency *.22**.159*.147* Official Supervisory Power -.175*** -.153* -.138** Cost of Funding -.003*** ***.001*** Money and quasi money.007***.006*** Constant 20.5*** 19.4*** 13.4*** 11.9*** 11.5*** N 8,210 5,064 4,462 5,685 4,939 j Hansen Hansenp AR AR1p AR AR2p legend:*p<.05;**p<.01;***p<.001

22 Global Results: Savings are + cyclical Variable Model1 (YD-ISP) Model2 (YD-ISP) Model3 (YD-ISP) Model4 (YD-ISP) Model5 (YD-ISP) GDP -.118*** -.059* -.082** -.108*** -.056* BUF L1..415***.465***.57***.46***.532*** BUF L * -.058*** Size -.691*** -.565*** -.53*** -.582*** -.609*** Return On Avg Assets.667***.436***.674***.47*** Loan Loss Res / Gross Loans.297**.128* ***.076* Boone indicator 2.29* 2.32** 2.25** 2.35* 1.5* Overall Capital Stringency *.205*.162*.15* Official Supervisory Power -.173*** -.148* -.127** Cooperative Bank * GDP Savings Bank * GDP.323*** ***.259*** Cost of Funding -.003*** ***.001*** Money and quasi money.007***.007*** Constant 20.3*** 19.5*** 11.1*** 13.4*** 10.6*** N 8,210 5,064 4,462 5,685 4,939 j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

23 Global Results: - procyclic. seems to be a large bank phenom in a global perspective Variable Model1 (Size-Int) Model2 (Size-Int) Model3 (Size-Int) Model4 (Size-Int) Model5 (Size-Int) GDP BUF L *** 0.535*** 0.442*** 0.503*** BUF L ** Size Country * GDP *** *** *** *** *** Return On Avg Assets 0.650*** 0.410*** 0.640*** 0.468*** Loan Loss Res / Gross Loans 0.282* *** 0.063* Boone indicator 2.287* 2.225** 2.312*** 2.593* 2.042** Overall Capital Stringency Official Supervisory Power Cost of Funding *** Money and quasi money Constant 9.170*** *** 2.401*** 3.931*** 3.275*** N 8,210 5,064 4,462 5,685 4,939 j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001 *Size-Country: taking into account only 25% of the largest banks in each country.

24 Global Results: Buf determinants Business Cycle (CYCLE): Overall negative cyclical fluctuation, BUT lots of caveats Bank Size (SIZE): hold less capital buffers and are exhibit more (-) cyclical fluctuation. Posibly too big too fail effect Profitability (ROA): Positively related to buffers. Possible explanation: source of internal funds for buffers (Kasman & Carvallo 2013) Loan loss reserve (LLR): Risk effect of risk on buffers The higher the cost of funds the higher the buffers Boone: level of competition varies negatively with buffer s formation: charter value hypothesis.

25 Global Results: Buf determinants Financial development matters for buffer formation (+) Official supervisory power negatively related to buffers.the stringency of capital regulations support buffer formation. Restrictions on bank activities have no ambiguous effect. Cooperative banks hold less capital buffers, Savings less: more stable funding/conservative stance? Savings behave with regard to cycle (-) cyclical fluctuation seems to be a large bank phenomenon from a global perspective

26 Regional Results: More Heterogeneity

27 Regional Results: overall same results across regions but important exceptions Variable Model1 M1- AFRICA M1- EUREAST M1- EURWEST M1- FAREASTCA M1- MIDEAST M1- SOUTHAMER GDP -.108** ***.116***.115*** -.202** -.271*** BUF L1..414***.331***.42***.687***.578***.712***.537*** Size -.709*** *** -.337*** -1.15*** *** Return On Avg Assets.633***.584**.859***.724*** 2.61*** *** Loan Loss Res / Gross Loans.25* ***.0637*.78***.198***.604*** Boone indicator 2.31* 6.2*** 2.95*** *** *** Overall Capital Stringency * *** -.51*** 3.38* -.849*** Official Supervisory Power -.175*** ** -.289*** *** Constant 20.5*** *** 10.3*** 20.3*** ** N 8, , j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

28 And their cyclical pattern LATAM: But savings are not different from peers regarding cycle: Variable Model1-ISP M1ISP- AFRICA M1ISP- EUREAST M1ISP- EURWEST M1ISP- FAREASTCA M1ISP- MIDEAST M1ISP- STHAM GDP -.118*** *** *** -.202** -.273*** BUF L1..415***.331***.42***.69***.578***.712***.535*** Size -.691*** *** -.281*** -1.14*** *** Return On Avg Assets.667***.584**.859***.611*** 2.68*** *** Loan Loss Res / Gross Loans.297** *** ***.198***.597*** Boone indicator 2.29* 6.22*** 2.81*** 6.29* 9.73*** *** Overall Capital Stringency * *** -.522*** 3.38* -.886*** Official Supervisory Power -.173*** ** -.258*** *** Cooperative Bank * GDP *** -.342** Savings Bank * GDP.323*** ** -.174* Constant 20.3*** *** 9.19*** 19.5*** *** N 8, , j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

29 No difference across region with regard to size and cyclical fluctuation LATAM: but large banks are not different from peers regarding cycle: Variable Model1-ISC M1ISC- AFRICA M1ISC- EUREAST M1ISC- EURWEST M1ISC- FAREASTCA M1ISC- MIDEAST M1ISC- STHAM GDP ***.188***.297*** * BUF L1..419***.33***.437***.721***.595***.68***.594*** Size Country * GDP -.206*** -.385* -.133*** -.3*** -.328*** -.104** -.157* Return On Avg Assets.65***.58**.753***.689*** 2.62*** *** Loan Loss Res / Gross Loans.282* *** ***.202***.611*** Boone indicator 2.29* 6.68*** 2.02*** ** * Overall Capital Stringency ** *** -.591*** 3.27* -.464* Official Supervisory Power *** -.162*** *** Constant 9.17*** *** 3.27*** ** N 8, , j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

30 Some global Cases: Latin countries are not alone

31 So heterogeneity is not a Latam only feature Variable Malaysia Brasil México India Indonesia UK Spain GDP -.596***.792*** *** -1.82***.7***.273*** BUF L1..651***.242***.695***.7***.687***.642*** 1.07*** Size Country * GDP -.129* -1.42*** -.216*** *** -.224*** -1.1*** -.347*** Return On Avg Assets 9.61*** -.339***.134***.552***.916*** 1*** -.748*** Loan Loss Res / Gross Loans 1.46*** -.174***.155***.0382***.119*.174*** Constant -8.63*** 9.6*** 2.15*** 2.11*** 12.9*** 3.47***.303** N j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

32 An Exercise with Latam

33 And cyclical effects Variable Argentina Brasil México Panamá Perú Venezuela GDP *** ** *** BUF L1..493***.242***.695***.315*** *** Size Country * GDP *** -.216*** -.29*** Return On Avg Assets 2.19* -.339***.134***.657*** *** Loan Loss Res / Gross Loans -1.54* -.174***.155*** 1.59*** *** Constant *** 2.15*** 2.86*** *** N j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

34 Also regarding forms of organization Variable Argentina Brasil México Panamá Perú Venezuela GDP *** -.168*** *** *** BUF L1..439***.256***.693***.356*** *** Return On Avg Assets ***.45*** Loan Loss Res / Gross Loans ***.126***.606*** *** Cooperative Bank * GDP ***.139*** 1.55*** *** Savings Bank * GDP *** Constant *** 2.18*** 2.27*** *** N j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

35 Stretching results using EQTA MCR, when high correlation with Buffers

36 Same differential results with regard to Size Variable Brasil Panamá Perú Venezuela Chile Colombia CostaRica Ecuador DomREp Guatemala Honduras GDP *** ** *** *** BUF2 L1..62***.252*** 1.12***.56***.996***.418***.523***.53***.655***.715***.564*** Size Country -2.84*** -1.37*** 1.45* ** ** -1.67* *** Return On Avg Assets.336***.145** 1.13***.533*** *** 2.21***.071*.569***.515***.414*** Loan Loss Res / Gross Loans ***.284***.568*** **.809***.278***.145***.193*** Constant * -4.81*** ** *** *.741*** N j Hansen Hansenp AR AR1p AR AR2p Correlation BUF-BUF legend: * p<.05; ** p<.01; *** p<.001

37 Size and Cyclical patterns Variable Brasil Panamá Perú Venezuela Chile Colombia CostaRica Ecuador DomREp Guatemala Honduras GDP.298*** *** *** *** BUF2 L1..62***.26*** 1.08***.563***.968***.432***.519***.538***.65***.721***.565*** Size Country * GDP -.588*** -.14***.231** * -.229** -.175* *** Return On Avg Assets.339***.157** 1.14***.512*** *** 2.28***.0813**.584***.494***.409*** Loan Loss Res / Gross Loans ***.273***.554*** **.82***.277***.147***.2*** Constant *** ** *** **.346* N j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

38 Type of banks Variable Brasil Panamá Perú Venezuela Chile Colombia CostaRica Ecuador DomREp Guatemala Honduras GDP *** ** *** *** BUF L1..628***.26*** 1.09***.56*** 1***.381***.414***.528***.632***.724***.57*** Return On Avg Assets.345***.148** 1.18***.536*** *** 1.69***.133***.598***.53***.399*** Loan Loss Res / Gross Loans ***.259***.572*** ***.872***.23***.165***.205*** Cooperative Bank -4*** 4.33*** -6.95*** ** 3.01*** Savings Bank * 42.3*** -157 Constant *** ** -2.36** -2.89*** ** N j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

39 Type and cyclical patterns Variable Brasil Panamá Perú Venezuela Chile Colombia CostaRica Ecuador DomREp Guatemala Honduras GDP *** ** -.298** -.282*** *** BUF2 L1..629***.257*** 1.09***.561*** 1.01***.408***.666***.528***.625***.724***.57*** Return On Avg Assets.344***.173** 1.17***.543*** *** 1.37***.133***.602***.53***.399*** Loan Loss Res / Gross Loans ***.264***.574*** ***.872***.256***.165***.205*** Cooperative Bank * GDP -.696***.558*** -1.09*** ** Savings Bank * GDP *** Constant *** ** *** ** N j Hansen Hansenp AR AR1p AR AR2p legend: * p<.05; ** p<.01; *** p<.001

40 Conclusions So, are Latam banks capital buffers procyclical?

41 Yes with lots of qualifications Like global peers, differentiated patterns exist regarding type and size of banks within their national markets. This affect both the levels and the cyclical fluctuation of buffers. Other critical factors: Competitiveness and market power within national borders. This factors have tend to converge in the region but differences remain. It affects potentially the internal/external funding mix, moral hazard issues involved Regulatory incentives: they may interact to competition to result in different incentive structures. Financial depth and development of capital markets. Market liquidity. Regulatory calibration should be sensitive to this issues, globally, regionally and domestically.

42 Thanks!