Determinants of Leverage Ratio in Brazil

New Capital Restriction under Basel III: Determinants of Leverage Ratio in Brazil

Douglas da Rosa München Marcos Soares da Silva

Abstract

This article empirically investigates the main determinants of the Leverage Ratio of Brazilian banks, examining how this ratio reacts to macroeconomic variables (business cycle and monetary policy) and how it is affected by bank performance indicators. We estimate a dynamic panel of 131 financial conglomerates for 2008-2014. We find that the adjustment cost is relatively moderate, except for the retail segment that would need up to 12 quarters to absorb a financial shock. These results also suggest that capital restrictions that the leverage limit imposes have a counter-cyclical pattern. Therefore, they serve as instruments to limit capital based on risk-weighted assets. JEL Classification: E41, E52, C32 Keywords: Leverage, Capital Ratios, Procyclicality, Global Financial Crises.

                                                              Monitoring Department, Banco Central do Brasil. Address: Setor Bancário Sul Quadra 3 Bloco B Ed. Sede, Brasília-DF CEP 70074-900 phone number: +55 61 3414-2293 Email: [email protected]   Research Department, Banco Central do Brasil. Address: Setor Bancário Sul Quadra 3 Bloco B Ed. Sede, Brasília-DF 70074-900 phone number: +55 61 3414-4150 Email: [email protected] 

1

 

1. Introduction Leverage is the degree to which a financial institution funds its assets with debt rather than equity. Still, the risks involved in financial intermediation may cause unexpected losses that are usually supported by capital. For this reason, banking regulators should adopt mechanisms that limit the degree of leverage of such entities such as to prevent banks of taking excessive risks [Keeley (1990), Wang (2013), Jarrow (2013), Dermine (2015)]. Such rule is intended to increase the cost of bankruptcy, thus improving the resilience of the financial system. Leverage limits were originally treated in the first Basel Accord (BCBS, 1988), which recommended the adoption of minimum capital to cover credit risks. In Brazil, this matter was regulated by Resolution No. 2,099, dated August 17, 1994, released by the National Monetary Council (CMN). This rule introduced minimum capital requirement based on the risk of active credit operations of the bank. Later, in 1996, the “Amended of Banking Supervision Committee Basel 1988 Agreement” incorporated regulatory minimum capital to cover market risks. For details on the Basel I Agreement implementation process in Brazil, see Puga (1999) and Silva and Divino (2012). A new capital accord, known as Basel II, was released in 2004, with the primary intention of introducing more accurate measures of the risks incurred by banks, including a minimum capital requirement for operational risk. This document breaks a new ground by adding to the principles of banking regulation good practices of risk management, as well as market discipline. The Communiqué n° 12,746, dated December 09, 2004, of the Central Bank of Brazil, introduced in Brazil the Basel II rules, which was implemented over the period of 2006 up to 20111. With the advent of the subprime crisis in 2008, the Financial Stability Board and the G20 proposed a set of measures to protect the banking system against financial crises. These discussions contributed to a comprehensive reform that culminated in the latest Capital Accord, named Basel III (BCBS, 2011), which involves greater focus on liquidity

                                                             1

The regulatory framework relative to Basel II was set by the Resolution n° 3,444, dated February 28, 2007, and Resolution n° 3,490, dated August 28, 2007. 2

   

and quality of core capital of banks. Additionally, it introduces the additional core capital buffer and the countercyclical capital buffer2. The high leverage of financial institutions is seen as a relevant factor for explaining the last global financial crisis. Papanikolaou and Wolff (2014) find empirical evidence that the increased exposure in off-balance sheet operations can hide relevant risks and weaken the financial stability. On the other hand, the adoption of deleveraging policy could be beneficial from a microprudential perspective, but would be detrimental to the banking system if the asset sales volume is large enough to trigger a price depreciation process. Therefore, Basel III recommends financial institutions to adopt operational limits3. In the financial intermediation activity, the degree of leverage is markedly higher than the level observed in other economic sectors. While commercial firms, for example, operate with an average leverage of 0.6, banks can build passive 12 times greater than their equity. Examining a panel of 14 developed countries, Brei and Gambacorta (2014) investigate how leverage behaves over the business cycles. They show that the leverage ratio is more counter-cyclical than regulatory capital ratios that are based on risk. Monitoring decisions on loans made by dominant banks, while considering restrictions on leverage and risk-based capital, are investigated in a theoretical study carried out by Balasubramanyan (2014). His model shows that, under certain conditions, dominant banks will monitor their portfolios when operating under capital constraint based on leverage criteria. However, such banks do not exercise monitoring actions when the capital constraints are that rely on risk approach, which suggests that the adoption of a capital limit based on leverage criterion is an active constraint on the solution of the bank's profit maximization problem. The dynamics of leverage cycle is evaluated in another theoretical work proposed by Aymanns and Farmer (2015). According to the authors, the effectiveness of the application of leverage limits, in order to prevent turbulence in the banking system,                                                               Resolution n° 4,192, dated March 1, 2013, is the benchmark for determining capital, according to the concept of Basel III in Brazil.  3  In Brazil, this matter is regulated by Circular n° 3,748, dated February 27, 2015, the rule that established the Leverage Ratio (LR).  3 2

   

depends on the level of the appropriate parameters. Moreover, a lower limit of leverage contributes to smoothening the credit cycle, regardless of the degree of banks’ risk aversion levels. The relationship between leverage and systemic risk is studied by Tasca et al. (2014). The results of the model proposed by the authors concern to banking regulators as it identifies a critical threshold of risk diversification separating: (1) an "insurance scheme" in which the risk reduction occurs with increasing degree of leverage; and (2) "a risky scheme" in which the risk mitigation strategy cannot be implemented via increased leverage. Keima and Joinville (2014) examine how the application of a leverage limit could affect the stability of the financial system. To this end, these authors develop an economic model of partial equilibrium with the explicit inclusion of a competitive banking sector in order to study the portfolio adjustment process of the banks for different capital restriction levels. The simulation exercises conducted show that banks that originally adopt the strategy of operating with low-risk loan portfolio tend to refocus their portfolio to higher risk levels as the leverage constraint is active. Therefore, limiting the degree of leverage is characterized by being punitive for banks specialized in low default risk business. On the other hand, banks that adopts excessive risk-taking present margin to include into their portfolio less risky assets. Thus, after the implementation of a regulatory limit on leverage, the portfolios of both these kinds of banks become more correlated, weakening the stability of the financial system whenever the effect derived from failures in asset risk assessment model is large enough. This paper aims to empirically identify the key determinants of the leverage ratio of Brazilian banks by examining how this variable reacts to macroeconomic conditions (business cycle and monetary policy) and how it is affected by the operating performance of banks. To achieve that, we estimate a dynamic panel of 131 financial conglomerates and isolated financial institutions from 2008 to 2014 on a quarterly basis. The estimation strategy uses a dynamic panel that seeks to assess the cost of adjustment of banks to the new restriction of capital represented by the leverage limit set under the Basel III rules. This paper presents the following contributions to the literature of Banking Economics. First, to the best of our knowledge, it is the first work that assesses the impacts 4

   

of leverage restriction within the Brazilian banking system, exploring the understanding of the adjustments of financial institutions to the new regulatory environment. Second, a proxy of the proposed leverage ratio in Basel III standards, as defined by Circular n° 3,748, dated February 27, 20154, was calculated with retroactive effect5. The remainder of this paper is organized as follows. Section 2 presents stylized facts related to the prudential regulation of capital, especially regarding the limit of the degree of leverage of financial institutions. Section 3 develops the econometric estimation strategy, including references to data description. In Section 4, we discuss the estimation results. Finally, section 5 is devoted to the conclusions and recommendations of the research agenda. 2. Prudential Regulation of Capital Limiting the leverage of banks is an additional instrument to capital requirements based on risk-weighted assets, according to the proposal made by the Basel Committee on Banking Supervision (BCBS, 2011). From the committee viewpoint, adopting a leverage limit, together with the capital requirements recently reviewed, would promote greater resilience of banks and of the financial system given that such restriction would act as a supplementary tier of protection against possible errors in assessing risks. Still, imposing limitations on the leverage of banks can reduce the likelihood of credit booms, thus softening potential cyclicality in capital requirements of banks. The Basel guideline suggests that the leverage measure of each jurisdiction should be simple, transparent and easy to quantify6. Circular n° 3,748, dated February 27, 2015, defines the methodology for determining the Leverage Ratio that should be adopted as a minimum requirement. It also regulates the reporting and disclosing process of relevant information. Thus, banks are compelled to send such data on a monthly basis to the Central Bank of Brazil, from October 2015. The Leverage ratio (LR) is defined as:

                                                              This regulatory instrument came into force from 1 October 2015.  The historical series has built a quarterly basis over the period 2008-2014. 6  Comuniqué no 20,615, dated February 17, 2011, states that a minimum Leverage Ratio requirement will be required, starting in January 2018. Despite not yet being decided, a 3% limit is expected. 5 4 5

   





(1)

where "Tier I" is the sum of the Common Equity Tier l and Additional Tier 1 Capital. The components of Tier I and Total Exposure are defined in detail in the appendix 1. In this study, we define a proxy to measure Leverage Ratio using information available in the Accounting Plan for Institutions of the National Financial System (COSIF) and in the Operational Limits Document (DLO). The latter instrument contains detailed information that financial institutions declare to Central Bank of Brazil their limits of total regulatory capital, common equity tier 1, additional common equity tier l, credit risk, operational risk and market risk. Table 1 shows the components of the LR as defined in Equation 2, in comparison with regulatory parameters of Circular n° 3,748, dated February 27, 2015. In this work, a leverage measure proxy was rebuilt retroactively 2008-2014, on a quarterly basis, making use of the available information in the documents mentioned above, according to the concept of leverage approved by the Central Bank of Brazil. We consider as on-balance sheet items the total assets, excluding derivatives, which have specific treatment in items 3 to 9 of Table 1. There are not the subject of the calculation of the approach of LR items I, II, and IV of § 4, Art. 5 of the Circular n° 3,748, dated February 27, 2015. We evaluate the leverage measure by including the appropriate treatment for regulatory adjustments applied in the calculation of Additional Tier 1 Capital, except deferred tax liabilities associates. Derivative financial instruments that are not included in the balance sheet are partially considered such as (1) replacement cost associated with all derivatives transactions; (2) add-on amounts for PFE associated with all derivatives transactions and (3) adjusted effective notional amount of written credit derivatives.

6

   

Table 1. Common model for the Leverage Ratio (LR) Comparison with Annex II of the Circular 3,748, dated February 27, 2015 Item

Instruction

Considered in the calculation of proxy LR

Items in the balance sheet 1 2

On-balance sheet items (excluding derivatives and SFTs, but including collateral) (Regulatory adjustments in determining Basel III Tier 1 capital)

On-balance sheet items (excluding derivatives) According to Resolution n° 4,192, dated March 1, 2013

Derivative exposures 3 4 5 6 7 8 9

Replacement cost associated with all derivatives transactions Add-on amounts for PFE associated with all derivatives transactions Gross-up for derivatives collateral provided (Deductions of receivables assets for cash variation margin provided) (Exempted CCP leg of client-cleared trade exposures) Adjusted effective notional amount of written credit derivatives (Adjusted effective notional offsets and add-on deductions for written credit derivatives)

According to Circular n° 3,748 According to Circular n° 3,748 NA NA NA According to Circular n° 3,748 NA

Securities Financing Transactions (SFTs) 10 11 12 13

Gross SFT assets (Netted amounts of cash payables and cash receivables of gross SFT assets) CCR exposure for SFT assets Agent transaction exposures

Included in item 1 NA NA NA

Other off-balance sheet exposures 14 15

Off-balance sheet exposure at gross notional amount (Adjustments for conversion to credit equivalent amounts)

According to Circular n° 3,748 Included in item 14

Total Exposure and Basel III Tier 1 Capital 16

Basel III Tier 1 Capital

17

Total exposure

According to Resolution n° 4,192, dated March 1, 2013 Sum of accounts 1 to 15

Leverage Ratio (LR) 18

Basel III leverage ratio

Item 16 / item 17

Sources: Circular n° 3,748, dated February 27, 2015 and authors NA – not applicable or unavailable data

Exposures related to SFTs are included in the balance sheet items, without specific treatment under items 11, 12 and 13. Off-balance sheet exposure at gross notional amount was computed considering the adjustments for conversion to credit equivalent amounts, through the use of information declared by the financial institutions in the Operational Limits Document (DLO). Nevertheless, the Leverage ratio measure that we compute can 7

   

be considered very close to that recommended by Basel III, given that, to the Brazilian jurisdiction, banking regulation provides a declaration of non-accounting information about exposures and risks to which are subject financial institutions since 2008. Table 2 shows descriptive statistics of the Leverage Ratio for the entire National Banking System (NBS) and for clusters of banks according to their size. Table 3 shows the same statistics for ownership of capital and economic activities predominantly exploited by the financial institution. They calculate the relationship between capital and exposure, expressed as a percentage, as defined in Equation 1. Hence, the lower the value of this ratio is, the higher the “degree of leverage” is. For example, a financial institution with a Leverage Ratio of 20 would have an exposure of 5 times its capital, while another bank with a Leverage Ratio of 8 would have an exposure of 12.5 times its capital. The latter represents the classical degree of leverage. Table 2. Descriptive Statistics: Leverage ratio - LR (%) by size of financial institution Statistics Minimum Maximum Mean Standard deviation 10th percentile 25th percentile 50th percentile 75th percentile 90th percentile

NBS 3.0 93,8 20.9 20.0 6.1 9.4 14.2 21.5 48.3

Large 4.6 9.2 6.9 1.7 4.9 5.6 7.2 7.8 8.5

Medium 3.0 78.4 11.3 12.7 5.2 6.2 9.2 12.2 14.8

Small and Micro 5.0 93.8 25.1 21.2 8.8 12.3 17.1 27.1 51.6

Sources: authors

In the period, the median financial institution is exposed 7.0 times its capital. The banks at the 10th percentile and the 90th percentile exposure have 16.3 and 2.1 times, respectively, higher than the level of capital I. The largest financial institutions have a higher degree of leverage. While the large banks can leverage 13.9 times its capital, small and micro-sized banks operate with a leverage of 5.9. However, there are small and microsized banks that operate under a high degree of leverage (20.1 times the capital). The sample includes a medium-sized bank whose total exposures account for 33.1 times their capital, as well as a large bank whose total exposure is equivalent to 21.8 its capital. We observe in Table 3 that state-owned banks, which are predominantly large corporations, have a degree of leverage of 13.3 times their capital, in median terms. The less leveraged state-owned bank has exposures corresponding to 3.9 times the capital 8

   

level I, while the most leveraged reaches 34.5 times its capital. Domestic private and foreign financial institutions have a leverage of 6.4 and 6.7 times their levels of capital, respectively. Table 3. Descriptive statistics: Leverage ratio (LR) by ownership and economic activity Statistics Minimum Maximum Mean Standard deviation 10th percentile 25th percentile 50th percentile 75th percentile 90th percentile

StateOwned

Domestic Private

Foreign

Retail

5.2 12.6 8.1 2.5 5.5 6.7 7.3 9.1 11.5

3.0 93.8 22.8 20.8 6.3 10.0 15.3 22.9 48.5

4.6 90.4 20.9 20.2 7.1 9.4 14.8 20.0 47.4

4.5 86.7 16.7 17.4 6.0 8.8 11.8 15.9 24.9

Wholesale and Middle Market 3.0 51.4 16.5 12 6.1 8.0 13.6 19.5 30.7

Treasury and Asset Management 5.0 93.8 28.8 26.8 9.4 12.9 17.9 29.8 78.4

Sources: authors

By economic activity, we observe that wholesale banks have a degree of leverage slightly lower than that of retail banks. Figure 1 shows the trajectory of how evolved the median Leverage Ratio of the banking system during the interval of 2008 to 2014. In the same panel, we show the median trajectory of Tier 1 Ratio, noting that the latter measure is a limit of regulatory risk-weighted capital. These two variables appear to be weakly correlated. Indeed, the Pearson correlation test does not allow rejection of the null hypothesis that these variables have zero correlation. This result suggests that the leverage ratio as a supplementary measure of prudential regulation effectively incorporates additional information not captured by the limits of capital based on risk-weighting.

9

   

Figure 1. Basel III capital ratios: 2008-2014 Capital Limits: Basel III (Median Banking System) 20 18 16 14 12

Dec 2008

Dec 2009

Dec 2010

Dec 2011

Leverage Ratio

Dec 2012

10 Dec 2014

Dec 2013

Tier 1 Capital Ratio

Source: authors

In the second half of 2008, there is a move to reduce the leverage of Brazilian financial institutions, indicating a possible adjustment to the current economic conditions. Between 2010 and mid-2013, the bank leverage increases while the risk-weighted capital ratio keeps relatively stable. Acknowledging that the time span of the time series is relatively short to feature a full economic cycle, the pattern identified until the present moment suggests that these variables do not maintain a co-integration process, or share a common stochastic process. Figure 2. Leverage ratio (LR): 2008-2014 Leverage Ratio: 2008-2014 100 90 80 70 2014

60 50 40 30 20 10 0 0

10

20

30

40

50

60

70

80

90

100

2008

Source: authors

 

10

   

Figure 2 presents the Leverage Ratio observed for the same financial institution at the end of 2008 and 2014. The diagonal line indicates the situation where the financial institution did not change its degree of leverage between the dates considered. Financial institutions located above (below) the diagonal reduced (increased) their level of operational leverage. About 60% of financial institutions increased their degree of leverage between 2008 and 2014. Figure 3. Loans growth rate Loans, 12 months growth rate 50% 40% 30% 20% 10%

Dec 2009

Dec 2010

Dec 2011

State-Owned Banks

Dec 2012

Dec 2013

0% Dec 2014

Private Banks

  Source: authors

As indicated in Figure 3, the increasing of the degree of leverage in NBS could be explained by government-driven credit expansion over 2009-2014. 3. Econometric specification and data The estimation is accomplished using the estimator proposed by Arellano and Bover (1995) and Blundell and Bond (1998). This allows the use of conditions of additional moments from the original estimator of Arellano and Bond (1991), in order to cope with the problem of weak instruments. Similar estimation strategy was adopted by Brei and Gambacorta (2014) for an empirical study on the cyclical pattern of bank leverage. In order to identify determinants of leverage of banks in Brazil, the following econometric specification will be estimated. 11

   

ln RA ) = γln RA ,

)+ ∑

∑



β

α +



(2)

denotes the “Leverage Ratio” of bank at time , as defined in section 2 of

where this paper;

is a vector of variables that characterize each bank;

variables that capture the macroeconomic conditions of the country; variable used to control potential effects of economic changes;

is a vector of is a dummy is unobservable

individual factors; the last term is the error term that, by hypothesis, is identically and e independently distributed with zero mean and constancy variance ε , ~ IID 0, σ

.

To control the effects of economic environment conditions, the following variables were used: (1) Selic interest rate7 at the end of the quarter; (2) Volatility of the Reference rate swaps8 at the end of the quarter. Thus, it is expected to capture the process of adjustment of expectations of interest rate; and (3) real growth of gross domestic product at market prices (GDP)9, in order to assess business cycle effects on the level of the adjustment process of leverage of banks. To characterize the profile of the banks, the following indicators are used: (1) Probability of Default (PD), measured by the likelihood of a default over a particular time horizon; (2) Cost of Funding (CF) represented by the average funding rate of interestbearing liabilities; (3) Liquidity, measured by highly liquid assets held by financial institutions in order to meet short-term obligations; (4) Efficiency Ratio, represented by administrative expenses as a percentage of revenue; (5) Assets Growth Rate; (6) Size, measured by the logarithm of adjusted assets; (7); Total Capital Ratio (total capital calculated by dividing tier 1 plus tier 2 capital by the risk-weighted assets); (8) Profit Margin, calculated as net income divided by revenue; (9) Return on Adjusted Assets (ROAA). The latter indicator is calculated by taking net income and dividing it by average total assets, adjusted for the last six months from the reporting date, considering the following settings: prepaid expenses; securities available for sale and hedge cash flow.

                                                             7

Interest rate - Over / Selic - (% pm) - Brazil's Central Bank, Bulletin, Section financial and capital markets (BCB Boletim / M Finan..) - BM12_TJOVER12 8 Reference rate - swaps – average DI prefixed - 180 days - (% pa) - Securities, Commodities and Futures Exchange (BM & FBovespa) - BMF12_SWAPDI18012 9 GDP - market prices - var. real trim. - Ref. 2010 - (%) - Brazilian Institute of Geography and Statistics, National Accounts System Reference 2010 (IBGE / SCN 2010 Quarterly) - SCN104_PIBPMG104 12

   

The following variables are used to control financial institutions by category of ownership, and predominant economic activity: (1) state-owned banks; (2) domestic private banks; (3) foreign banks; (4) wholesale and Middle Market Banking; (5) Retail Banking; (6) treasury and asset management banks. Equation 2 was estimated using the GMM-system estimator in one stage proposed by Blundell and Bond (1998). The set of instruments used in the estimation process consists of lagged-dependent and predetermined variables (lagged covariates). In determining the number of lags of instrumental variables, we try to work around the following problems: (1) to avoid correlation with the error term (ε , ); (2) to control the excessive use of available instruments. Accordingly, the instruments matrix was collapsed so that only the second, third and fourth lagged covariates are used as instrumental variables. The proposed estimator of Arellano and Bover (1995) and Blundell and Bond (1998) requires stationary data to produce consistent estimates. To verify this condition, we perform the unit root tests proposed by Levin et al. (2002), Maddala and Wu (1999), and Im et al. (2003)10. At the 5% level of significance, the hypothesis that the panel series have a unit root is rejected. The results are shown in Table 4. Table 4. Unit root tests Variable LR GDP Selic Swaps Prob. Default Cost Liquidity Efficiency Assets growth Total capital Profit margin ROAA

LLC t-value -5.7*** -19.1*** -18.7*** -23.1*** -7.6*** -15.1*** -12.3*** -7.7*** -5.2*** -13.3*** -8.6*** -5.0***

Trend 0 1 1 1 1 0 0 0 1 1 1 1

Lag 1 1 1 1 1 1 1 1 3 2 2 0

IPS t-value -8.2*** -6.0*** -16.8*** -13.4*** -4.8*** -8.9*** -14.3*** -7.2*** -4.6** -6.2*** -8.3*** -13.3***

Trend 1 1 1 1 1 1 1 1 1 1 1 1

Lag 0 0 1 0 0 0 1 1 2 0 0 0

Fisher t-value -2.8*** -4.7*** -19.8*** -8.2*** -8.2*** -4.1*** -8.8*** -2.1** -5.6*** -4.3*** -5.8*** -8.3***

Trend 1 1 1 1 1 1 1 1 1 1 1 1

Lag 0 0 0 0 0 0 3 2 2 4 4 4

Note: * p < 0.10; ** p < 0.05; *** p < 0.01.

In the Hansen test of overidentifying restrictions, the null hypothesis is that there is no correlation between the instruments and the error term of the regression. In all                                                              10

Further information can be found in the appendix 2. 13

   

specifications the estimated model does not reject the null hypothesis, which ensures the validity of the instruments used. For consistency, the Arellano and Bond (1991) estimator require that the error term be serially uncorrelated for order higher than 1. 4. Results Table 5 shows the result of three model specifications. The specification (1) is an auto-regressive process for the entire sample, including control variables as covariate in order to control for characteristics of banks, such as ownership, economic activity and size. In the specification (2) we add some macroeconomic variables, while in the specification (3) we include microeconomic variables at the bank’s level.

14

   

Table 5. Regression results of the determinants of the Leverage Ratio for the Brazilian banking industry in the period 2008-2014 Variable Leverage Ratio\(t-1)

(1)

(2)

(3)

0.7063*** (0.0195)

0.7105*** (0.0195) 1.4268 (1.5594) 1.1313*** (0.3829) 0.0003 (0.0002)

-66.6131*** (24.8397) -3.7427** (1.5260) -2.0888** (1.0847) -4.7570*** (1.2613) 13.6705*** (1.3507) Yes Yes Yes

-66.3266*** (24.8873) -3.7152** (1.5295) -2.0571** (1.0867) -4.7371** (1.2629) 11.1271*** (2.5934) Yes Yes Yes

0.4910*** (0.0173) 1.7324* (1.0723) 0.8983*** (0.3149) 0.0614* (0.0408) -0.0003 (0.0163) -0.0077*** (0.0020) -0.4689*** (0.1502) -0.0026* (0.0037) -0.0515*** (0.0041) -8.4834*** (0.3570) 0.1166 (0.1402) -1.3996 (1.9612) 0.2724*** (0.0391) 37.4798** (18.8566) 10.8819*** (1.2472) -0.9037 (0.8752) -1.3852 (1.0044) 68.2897*** (3.5617) Yes Yes Yes

Selic Swap GDP Prob. Default Cost of Funding Liquidity Efficiency Assets Growth Rate (t-1) Log (Asset) Total Capital Ratio Profit Margin ROAA State-owned Foreign Wholesale Treasury and Asset Management Constant Dummy Time Dummy Season Fixed effect

Abond-Order 2 0.4261 0.3812 0.4906 Sargan (p-value) 0.2795 0.2894 0.2484 Wald Test (p-value) 0.0000 0.0000 0.0000 Observations 3,210 3,210 3,210 Note: * p