bank competition and the limits of creditor protection … generated by an increase in creditor...

Bank Competition and the Limits of Creditor ProtectionReforms

Leonardo Alencar Rodrigo Andrade Klenio Barbosa

The Central Bank of Brazil The Central Bank of Brazil Insper

August 2017

XII Annual Seminar on Risk, Financial Stability and Banking

Sao Paulo - Brazil

1 / 45

Creditor Protection and Credit Markets

Better legal conditions for repayment of loans: Aghion et al. (1992)

increases credit supply for firms and alleviates credit rationing

better credit terms in debt contracts: lower loan interest rates

Law, Economics and Finance: La Porta et al. (1998)

Investor Protection and Development of Credit Markets

Evidence:

La Porta et al. (1997,1998): cross country (49 countries)

Demirguç-Kunt and Maksimovic (1998)

Djankov et al. (2008)

Araujo et al. (2006, 2012), Coelho et al. (2012), Assunção et al. (2014)

Lilienfeld-Toal et al. (2012), Campello et al. (2016), Alencar et al. (2016)

2 / 45

Creditor Protection and Credit MarketsSource: World Bank (2005)

3 / 45

Creditor Protection and Credit MarketsSource: World Bank (2005)

Better Creditor Protection is associated withLower Interest Rate Spread and Higher Credit Volume

4 / 45

Credit Markets and Bank LoansKey Features of the Banking Industry

Bank loans account for a large share of corporate external financing:

France (37%), Germany (12%), Japan (42%), U.S. (26%), U.K. (5%)

Mayer (1988), Rajan and Zingales (1995), Tirole (2006).

High concentration rates and mild competition in the banking industry

Bikker et al. (2002), Claessens et al. (2004), Bikker et al. (2012)

Evidence is consistent with imperfect competition (monopoly, oligopoly,and no long-run perfect competition) for a large set of countries.

Lack of intense competition is explained by:

Technological entry barriers, activity restrictions, restrictions foreign bankentry, and regulations on capital adequacy

Claessens and Laeven (2004), Bartha, Caprio and Levine (2004)

5 / 45

This Paper

Research Question:

Can the lack of competition in the financial sector hamper the effects of amore effective creditor protection on credit markets?

Economic Foundations: Banking oligopoly price theory

A bank, in an imperfect competition market, wants to retain the extramargin generated by an increase in creditor protectionMargin-volume trade-off: little incentive to reduce loan rates to boostdemand for credit (unless competition is prevalent)The lower is the competition in the loan market, the lower will be thereduction (increase) in loan rates (credit supply).

Empirical Investigation:

Using loan-contract level data from Brazil (BCB), we investigate whether(and to what extent) competition in credit lines shapes the effects of anincrease in creditor protection on loan interest rates, spread and size.

6 / 45

Why Brazil?

Brazil is fertile ground for that research question for a number of reasons:

Corporate Bankruptcy Reform in 2005:

it improved creditor protection in corporate debt transactions and thebankruptcy system’s efficiency (Lisboa et al., 2005; Araujo et al, 2006).

Diverse degree of competition in credit markets:

it depends on local conditions (Coelho et al., 2013), set of product offered bycreditors (Barbosa et al., 2015), and on credit lines (Andrade, 2015)

Good database:

loan-contract level data: rates and size of corporate and consumer loansbank level data: financial statements and balance sheetsmarket power indicators at credit line levelSource: The Central Bank of Brazil (SCR)

Bank loan accounts for a large share credit in Brazil:

free lending funds correspond to 50% of total credit: BCB (2016)

7 / 45

Empirical Strategy and Results

Approach: Modified differences-in-differences estimationTraditional groups:

treated group: corporate loanscontrol group: consumer loans (non-payroll attached)

New interaction: triple difference modeltreatment-status dummy with market power indicator.

Results:Potential Effect: Brazilian Bankruptcy Reform (BBR) could have:

↓ loan rates 736 basis points (before BBR 33% a.a) wrt to the control group.↓ spreads 600 basis points (before BBR 13% a.a) wrt to the control group.↑ loan size BRL 9,000 wrt to the control group.

Lack of Competition: it hampers 18%-33% (in avg.) of the BBR effect.

Bottom line:Creditor protection reforms have to be combined with a reduction in entrybarrier and competition-enhancing policies in the banking industry.

8 / 45

The Plan of the Talk

Introduction

Institutional Background

The Model

Database and Descriptive Statistics

Empirical Strategy

Estimation Results and Findings

Robustness Tests

Final Remarks and Further Steps

9 / 45

The Brazilian Corporate Bankruptcy Reform

Institutional aspects:

New Bankruptcy Legislation (BBR): Federal Law no. 11.101 (2005)It was issued in February 2005, became legally effective in June 2005BBR was inspired in the U.S. Bankruptcy Codes

Previous legislations:

Until 2005: Federal Law no. 7.661 (1945)Preference to labor demands and taxes at expenses of creditorsSystem punished firms under financial distress (transaction cost)

Main features of the new legislation:

to increase creditor’s recovery rateto replace the norms that used to see bankruptcy as a penalty for companythat did not fulfill its debt contract, but it has good prospectus.to encourage creditors-borrowers cooperation (extrajudicial recovery)

⇒ Aim: continuity of business enterprisers with profitable projects

10 / 45

Recovery Rate in Brazil: Before and After BBRSource: World Bank

Figures and Tables

Figure 1: Recovery Rate in Brazil over time

05

1015

20R

ecov

ered

cen

ts o

n 1

US$

2002 2004 2006 2008 2010 2012year

Notes: Data from the World Bank Doing Business database.

38

11 / 45

The ModelOverview

The theoretical model is based on two key ingredients:

1 Brazilian Bankruptcy Reform increased bank’s profit

it increased recovery rate of secured credit

it turned reorganization more efficient: lower probability of default

2 Banks have some market power in the credit market

Current version: no entry (no evidence that BBR leaded to banking entry)

Work in progress: switching cost and heterogonous entry cost.

References: Klemperer (1987), Kim et al. (2003), Gehrig et al. (2007)

12 / 45

Comparative Statics and Testable ImplicationsFigures and Intuition

BBR leaded to a higher κ ∈ (p, δ)

13 / 45

Comparative Statics and Testable ImplicationsFigures and Intuition

BBR leaded to a higher κ ∈ (p, δ)

↑ Creditor protection (κ): ↓ Effective lending costIncentive to protect higher margin vis-a-vis volume

Competition pressure: puts loan rates down

14 / 45

Database and Descriptive Statistics

Sources:

Credit Information System (SCR): The Central of Brazil (BCB)

Our key source of information (contract-level data)

information on contract loans above BRL 5,000.00 (USD 2,270.00)

not public available (restricted access)

Monthly Banking Accounting Data (Cosif): BCB

Macroeconomic-financial data (controls): IPEA and BCB

Swap Pré-DI rates (for loan spreads): BM&FBovespa

15 / 45

Credit Information System (SCR)

Period: July 2004 to December 2007 (monthly data)

Available information: New loan contracts issued every month

contracted interest rate, loan size, maturity, contract characteristics (bank,credit line, credit risk, collateral)

Data restriction:

Free lending funds: no compulsory destinationOwn credit operations: excluded credit operations of intermediariesPrefixed loan rates: loans with pre-determined fixed credit termsExclusion: real estate and mortage loans, BNDES loans, non-payrollattached loan

Our observation: Bank-level data - BCB restriction

collapsed in the following dimensions: credit line, risk, w/out collateral

16 / 45

Credit lines

Credit contracts are classified in 10 different credit linesTable 3: Markets or Credit Lines

Description Number

Overdraft - Consumers 1Leasing and Goods Financing - Consumers 2Vehicle Financing - Consumers 3Loans and Other Credit Lines - Consumers 4Working Capital; Overdraft and Supplier Financing - Firms 5Commercial Papers Discount - Firms 6Leasing and Goods Financing - Firms 7Vehicle Financing - Firms 8Loans and Other Credit Lines - Firms 9Trade Finance: Import and Export - Firms 10

37

Each line is also divided in collateralized and non-collateralized lines.

17 / 45

Outcome Variables

Loan interest rate:

bank-month average contracted interest rate of individual credit operationsweighted by the size (value) of the credit operations (CAop).

Yblrct =

∑Ii CAopiblrctRiblrct∑I

i CAopiblrct(1)

Spread over interbank rate:

bank-month average spread over the interbank rate, IRTS (same maturity)

Sblrct =1 + Yblrct

1 + IRTSWMatblrct, (2)

Loan size:

bank-month average loan size (accounting value)

CAopblrct =

∑Ii CAopiblrct

I. (3)

18 / 45

Competition Indicators

Proxies for competition:

Relevant market: competition in each credit line

Baseline: credit line, credit risk, collaterized, month, year (finest criteria) .

HHIlrct =B∑1

( ContractedCreditblrct∑B1 ContractedCreditblrct

)2(4)

Extensions: credit line and time (coarser criteria).

Other proxies:

C4: the sum of market share of the first big lenders in a given market

MS: the market share of the bank in a given market

H − Statistics: Panzar-Rosse competition measure in a given credit line

19 / 45

Descriptive StatisticsOutcome variables

Table 7: Descriptive Statistics - Dependent Variables

Variables Before Bankruptcy Reform After Bankruptcy ReformObs Mean Sd.Dv. Min Max Obs Mean Sd.Dv. Min Max

Yb,l,r,c,t 6655 0.3614 0.2602 0 1.8127 22590 0.3633 0.2837 0 1.8279Sb,l,r,c,t 6620 0.1529 0.2174 -0.1648 1.3905 22402 0.1933 0.2433 -0.1646 1.5238

Table 8: Descriptive Statistics - Dependent Variables (Control group)


Yb,l,r,c,t 1782 0.4395 0.3115 0 1.7724 7184 0.4711 0.3583 - 1.8261Sb,l,r,c,t 1781 0.22 0.2637 -0.1647 1.3202 7142 0.292 0.3101 -0.1644 1.5189

Table 9: Descriptive Statistics - Dependent Variables (Treated group)


Yb,l,r,c,t 4873 0.3329 0.2323 0 1.8127 15406 0.3131 0.2241 0 1.8279Sb,l,r,c,t 4839 0.1281 0.1919 -0.1648 1.3905 15260 0.147 0.1875 -0.1646 1.5238

41

More Tables

20 / 45

Empirical Strategy

Modified Differences-in-Differences approach: (triple difference)

Yblrct = β0 + β1Λlrct + β2dmLawt + β3TblrctdmLawt + β4ΛlrctTblrct + β5ΛlrctdmLawt +

+ β6ΛlrctTblrctdmLawt +

C∑c=1

ϕcBankControlsbt +M∑

m=1

µmMacroControlst +

+F∑

f=1

φf dmYeart +H∑

h=1

φhdmMontht + ηb,l,r,c + εb,l,r,c,t,

dmLawt: BBR dummy variable (0 before May/2005, and 1, otherwise)

Tblrct: treated-control dummy (1 for treated group, and 0 for control)

treated group: corporate loans (without subsided loans)control group: consumer credit loans (without payroll attached credits)

Market-power “hampering” dummy: ΛlrctTblrctdmLawt (effect β6)

21 / 45

Parallel Trends: Treated and Control Groups

40

45

50

55

60

65

70

30

32

34

36

38

40

42

44

46

48

50

jul/0

4

ago/

04

set/

04

out/

04

nov/

04

dez/

04

jan/

05

fev/

05

mar

/05

abr/

05

mai

/05

jun/

05

jul/0

5

ago/

05

set/

05

out/

05

nov/

05

dez/

05

jan/

06

fev/

06

mar

/06

abr/

06

mai

/06

jun/

06

Cons

umer

cre

dit -

Loan

Inte

rest

Rat

es (%

a.a

.)

Corp

orat

e Lo

nas -

Inte

rest

Rat

es (%

a.a

.)

Time

Loan Interest Rates : Corporate Loans versus Consumer CreditBefore BBR (June/2005)

Corporate Loans Consumer Credit

38

39

40

41

42

43

44

45

46

47

48

22

23

24

25

26

27

28

jul/0

4

ago/

04

set/

04

out/

04

nov/

04

dez/

04

jan/

05

fev/

05

mar

/05

abr/

05

mai

/05

jun/

05

jul/0

5

ago/

05

set/

05

out/

05

nov/

05

dez/

05

jan/

06

fev/

06

mar

/06

abr/

06

mai

/06

jun/

06

Cons

umer

cre

dit -

Spre

ad (%

a.a

.)

Corp

orat

e Lo

nas -

Spre

ad (%

a.a

.)

Time

Loan Spread: Corporate Loans versus Consumer CreditBefore BBR (June/2005)


30

32

34

36

38

40

42

44

46

48

50

60

65

70

75

80

85

jul/0

4

ago/

04

set/

04

out/

04

nov/

04

dez/

04

jan/

05

fev/

05

mar

/05

abr/

05

mai

/05

jun/

05

jul/0

5

ago/

05

set/

05

out/

05

nov/

05

dez/

05

jan/

06

fev/

06

mar

/06

abr/

06

mai

/06

jun/

06

Cons

umer

Cre

dit (

thou

sand

s of B

RL)

Corp

orat

e Lo

ans (

thou

sand

s of B

RL)

Time

Loan Size: Corporate Loans versus Consumer CreditBefore BBR (June/2005)


22 / 45

Estimation Results and Findings

Econometric specifications:

Model (1): Dif-in-Dif variables, month-year dummies, outliers excluded.

Outliers are treated: Hadi (1994) algorithm for loan rates and spreads (1%).

Model (2): Dif-in-Dif variables, month-year dummies, with outliers

Model (3): Model (1) with controls (bank-specific/macro variables)

Model (4): Model (3) with symmetric sample (11 months before and after)

Outlier Sample

Estimations

23 / 45

Main ResultsPotential and Net of BBR, and the Hampering Effect

Potential Effect of BBR: β3 under perfect competition HHI = 0 Estimations

Loan Interest Rates: drop ∼= 700 basis-points (∼=36% a.a. → 29% a.a.)Spread: drop ∼= 600 basis-points (∼=19.06% a.a. → 13.06% a.a.)Loan Size: increase ∼= 9,000 BRL

Net Effect of BBR: β3 + β6 × HHIt≥June/2005

Loan Interest Rates: drop ∼= 481 basis-points (∼=36% a.a. → 31.31% a.a.)Spread: drop ∼= 436 basis-points (∼=19.06% a.a. → 14.7% a.a. )Loan Size: increase ∼= 7,356 BRL

Lack of Competition - Hampering Effect: β6 × HHIt≥June/2005

Loan Interest Rates: ∼= 219 basis-points (0.1083× 0.202 = 2.19%)Spread: ∼= 164 basis-points (0.0812× 0.202 = 1.64%)Loan Size: ∼= 2,140 BRL (−8, 140× 0.202 = 1, 644 BRL)

Bottom Line: Lack of Competition hampers 18.3%-31.3% of the BBR.

24 / 45

Main ResultsPotential and Net of BBR, and the Hampering Effect

Potential Effect of BBR: β3 under perfect competition HHI = 0 Estimations

Loan Interest Rates: drop ∼= 700 basis-points (∼=36% a.a. → 29% a.a.)Spread: drop ∼= 600 basis-points (∼=19.06% a.a. → 13.06% a.a.)Loan Size: increase ∼= 9,000 BRL

Net Effect of BBR: β3 + β6 × HHIt≥June/2005

Loan Interest Rates: drop ∼= 481 basis-points (∼=36% a.a. → 31.31% a.a.)Spread: drop ∼= 436 basis-points (∼=19.06% a.a. → 14.7% a.a. )Loan Size: increase ∼= 7,356 BRL

Lack of Competition - Hampering Effect: β6 × HHIt≥June/2005

Loan Interest Rates: ∼= 219 basis-points (0.1083× 0.202 = 2.19%)Spread: ∼= 164 basis-points (0.0812× 0.202 = 1.64%)Loan Size: ∼= 2,140 BRL (−8, 140× 0.202 = 1, 644 BRL)

Bottom Line: Lack of Competition hampers 18.3%-31.3% of the BBR.

25 / 45

Hampering Effect (HF): Across Credit Lines

The most competitive credit line: Vehicle Financing (HHI = 0.14)

Loan Interest Rates: ∼= 152 basis-points (HF: 21.7%)Spread: ∼= 117 basis-points (HF: 19.5%)Loan Size: ∼= 1,139 BRL (HF: 12.7%)

The least competitive credit line: Loans (HHI = 0.33)

Loan Interest Rates: ∼= 357 basis-points (HF: 50.0%)Spread: ∼= 267 basis-points (HF: 44.5%)Loan Size: ∼= 2,686 BRL (HF: 30%)

Robustness Tests

Invariance of banking competition to BBR Invariance

Falsification test

Placebo tests Placebo

Random assignment of competition measures Random

Cluster error at credit line, collateralized and credit risk level

Other proxies for bank competition Other Proxies

Coarse definitions of market power

Other proxies for market power

26 / 45

Conclusions and Final Remarks

Take-Home Message:

2005 Brazilian Bankruptcy Reform (BBR) leaded to lower interest rates (↓4% a.a.) and larger size (↑ 7,356 BRL) of corporate loans.Lack of competition in corporate credit market hampered around 18-31percent of BBR effects on loan interest rates and size (30-50 percent inless competitive credit lines).

Policy Implication:

Potential effects of creditor protection reforms cannot be achieved withoutcompetition-enhancing policies in the banking sector

Further Steps:

Model: Switching cost and heterogonous entry cost.More investigation on the invariance of banking competition to BBREstimation with exclusion of outliers in loan size regressions

27 / 45

Additional Material

28 / 45

Descriptive StatisticsDif-in-Dif dummy variables

Table 10: Descriptive Statistics - Control Variables and Dummies

Before BBR After BBRControl Variables Obs. Mean Sd. Dv. Obs. Mean Sd. Dv.Market Share - Credit Portfolio 6620 0.0215 0.0337 22402 0.0258 0.0398HHICreditLine,Risk,Collateral 6620 0.2524 0.1433 22402 0.2753 0.1605C4CreditLine,Risk,Collateral 6620 0.6683 0.2046 22402 0.6899 0.2036MkSCreditLine,Risk,Collateral 6620 0.0841 0.1473 22402 0.0776 0.1448Interest Rate Term Structure 6620 0.1787 0.0112 22402 0.1407 0.0266Overnight Interbank InterestRate 6620 0.1749 0.0138 22402 0.1457 0.0289

Volatility of OvernightInterbank Interest Rate 6620 0.0018 0.0009 22402 0.0013 0.0012

Gross Domestic Product 6620 169.6801 5.8324 22402 206.1822 18.0957Industrial Production Index 6620 110.6983 6.4425 22402 118.5726 7.9421Inflation Index 6620 2,399 47.8 898 22402 2,605 69.8346Basel Capital Index 6614 0.2376 0.23 22400 0.2028 0.1835Liquidity Index 6108 0.2791 0.1485 20815 0.2772 0.1458Total Monthly Revenue over NetCapital 6620 0.0575 0.1573 22402 0.0585 0.0734

Total Defaulted CreditOperation 6200 0.0186 0.0561 21696 0.0303 0.1605

Mean Maturity 6620 346.057 368.5779 22402 414.375 430.8747Net Capital 6620 2.7821 3.9043 22402 4.3234 6.5374DummiesDummy of BBR 6620 0 0 22402 1.0000 0Dummy of Treated Group 6620 0.731 0.4435 22402 0.6812 0.466Dummy of Public Bank 6620 0.1718 0.3772 22402 0.2006 0.4005Dummy for CollateralizedOperations 6620 0.731 0.4435 22402 0.6812 0.466

InteractionsDummy of BBR * Dummy of TreatedGroup 6620 0 0 22402 0.6812 0.466

HHICredit Line, Risk,Collateral * Dummy of Treated Group 6620 0.179 0.1557 22402 0.1856 0.1834

HHICredit Line, Risk,Collateral * Dummy of BBR 6620 0 0 22402 0.2753 0.1605

HHICredit Line, Risk,Collateral * Dummy of BBR * Dummy of TreatedGroup

6620 0 0 22402 0.1856 0.1834

42

29 / 45

Descriptive StatisticsControl variables

Table 10: Descriptive Statistics - Control Variables and Dummies

Before BBR After BBRControl Variables Obs. Mean Sd. Dv. Obs. Mean Sd. Dv.Market Share - Credit Portfolio 6620 0.0215 0.0337 22402 0.0258 0.0398HHICreditLine,Risk,Collateral 6620 0.2524 0.1433 22402 0.2753 0.1605C4CreditLine,Risk,Collateral 6620 0.6683 0.2046 22402 0.6899 0.2036MkSCreditLine,Risk,Collateral 6620 0.0841 0.1473 22402 0.0776 0.1448Interest Rate Term Structure 6620 0.1787 0.0112 22402 0.1407 0.0266Overnight Interbank InterestRate 6620 0.1749 0.0138 22402 0.1457 0.0289

Volatility of OvernightInterbank Interest Rate 6620 0.0018 0.0009 22402 0.0013 0.0012

Gross Domestic Product 6620 169.6801 5.8324 22402 206.1822 18.0957Industrial Production Index 6620 110.6983 6.4425 22402 118.5726 7.9421Inflation Index 6620 2,399 47.8 898 22402 2,605 69.8346Basel Capital Index 6614 0.2376 0.23 22400 0.2028 0.1835Liquidity Index 6108 0.2791 0.1485 20815 0.2772 0.1458Total Monthly Revenue over NetCapital 6620 0.0575 0.1573 22402 0.0585 0.0734

Total Defaulted CreditOperation 6200 0.0186 0.0561 21696 0.0303 0.1605

Mean Maturity 6620 346.057 368.5779 22402 414.375 430.8747Net Capital 6620 2.7821 3.9043 22402 4.3234 6.5374DummiesDummy of BBR 6620 0 0 22402 1.0000 0Dummy of Treated Group 6620 0.731 0.4435 22402 0.6812 0.466Dummy of Public Bank 6620 0.1718 0.3772 22402 0.2006 0.4005Dummy for CollateralizedOperations 6620 0.731 0.4435 22402 0.6812 0.466

InteractionsDummy of BBR * Dummy of TreatedGroup 6620 0 0 22402 0.6812 0.466

HHICredit Line, Risk,Collateral * Dummy of Treated Group 6620 0.179 0.1557 22402 0.1856 0.1834

HHICredit Line, Risk,Collateral * Dummy of BBR 6620 0 0 22402 0.2753 0.1605

HHICredit Line, Risk,Collateral * Dummy of BBR * Dummy of TreatedGroup

6620 0 0 22402 0.1856 0.1834

42

30 / 45

Market ConcentrationHHI Structural change analysis: Corporate credit lines

31 / 45

Market ConcentrationHHI Structural change analysis: Consumer credit lines

Descriptive Statistics

32 / 45

Outlier Sample

Table 11: Detected outliers on the sample


Yb,l,r,c,t 58 2.1549 0.3496 1.8367 3.2441 464 2.6882 0.9501 1.8421 7.9072Sb,l,r,c,t 35 1.8047 0.3271 1.381 2.6625 308 2.1301 0.6095 1.4214 6.0057

43

Specifications

33 / 45

Main ResultsLoan Interest Rates

Table 12: Main Results - HHI and Bankruptcy Reform Effect on Mean Interest Rate

(1) (2) (3) (4)R-sq: within 0.0311 0.0402 0.0465 0.0453Test F F(20,26743) F(33,23859) F(33,23555) F(32,10522)

42.88 30.3 34.83 15.61Independent Variablesβ0 0.2052*** 0.7833 0.4531 0.3938

[0.009] [0.686] [0.477] [1.121]Market Share - Credit Portfolio 4.0087*** 5.2588*** 2.0405*** 5.1715***

[0.245] [0.366] [0.265] [0.600]HHICreditLine,Risk,Collateral 0.0713*** 0.1536*** 0.0881*** 0.0826***

[0.023] [0.034] [0.024] [0.030]Dummy of BBR 0.0824*** 0.1060*** 0.0863*** 0.0678***

[0.008] [0.013] [0.009] [0.012]Dummy of BBR * Dummy of Treated Group -0.0707*** -0.0836*** -0.0736*** -0.0708***

[0.009] [0.013] [0.009] [0.012]HHICreditLine,Risk,Collateral * Dummy of Treated Group -0.0575** -0.0998** -0.0621** -0.0708***

[0.029] [0.044] [0.031] [0.036]HHICreditLine,Risk,Collateral * Dummy of BBR -0.0960*** -0.2204*** -0.1091*** -0.1328***

[0.023] [0.035] [0.024] [0.025]HHICreditLine * Dummy of BBR * Dummy of Treated Group 0.0925*** 0.2020*** 0.1083*** 0.1307***

[0.029] [0.044] [0.031] [0.032]Interest Rate Term Structure -0.1267 -0.0990 -0.0941

[0.235] [0.163] [0.200]Overnight Interbank Interest Rate 1.1106*** 1.0320*** 0.7836**

[0.304] [0.211] [0.314]Volatility of Overnight Interbank Interest Rate 8.6189*** 5.5937*** 4.2250

[1.558] [1.084] [3.292]Gross Domestic Product 0.0045*** 0.0035*** 0.0021

[0.001] [0.001] [0.002]Industrial Production Index -0.0024** -0.0014* 0.0019

[0.001] [0.001] [0.002]Inflation Index -0.0005** -0.0003** -0.0003

[0.000] [0.000] [0.000]Basel Capital Index 0.0326* 0.0275** 0.0293*

[0.018] [0.013] [0.016]Liquidity Index 0.0034 -0.0120 -0.0594***

[0.020] [0.014] [0.022]Total Monthly Income Over Net Capital -0.0318** -0.0305*** 0.0037

[0.015] [0.010] [0.011]Total Defaulted Credit Operation 0.024 0.0234 0.0110

[0.010] [0.007] [0.011]Mean Maturity -0.0001*** -0.0001*** -0.0001***

[0.000] [0.000] [0.000]Net Capital 0.0121 0.0072 0.0110

[0.000] [0.001] [0.002]Dummy of Public Bank 0.0278 0.0225 -0.0121

[0.172] [0.119] [0.102]Year Fixed Effect Yes Yes Yes YesMonth Fixed Effect Yes Yes Yes YesThe econometric models use differences-in-differences method with panel data. We estimate the Market Power and Bankruptcy Reform effects on the MeanInterest Rate of new contracted collateralized loans to firms. (1) The estimations include only fixed-effects of year and month. The Brazilian BankruptcyReform is effective from June/2005. We exclude the outlier observations from the sample. (2) The estimations include controls, fixed-effects of year andmonth. The Brazilian Bankruptcy Reform is effective from June/2005. We keep the outlier observations on the sample. (3) The estimations include controls,year fixed-effect and month fixed-effect. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlier observations from the sample.(4) The estimations include controls, fixed-effects of year and month. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlierobservations. We symmetrically trunked the monthly data to keep the same number of panels before and after BBR.(a) All estimations include intercept.(b) Robust Standard Deviation in brackets, ∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01.

44

34 / 45

Main ResultsSpread over Interbank Rate

Table 13: Main Results - HHI and Bankruptcy Reform Effect on Mean Spread over IRTS

(1) (2) (3) (4)N Obs 29,022 26,132 25,800 11,790R-sq: within 0.0371 0.047 0.0523 0.0581Test F F(20,26550) F(33,23859) F(33,23555) F(32,10522)

51.110 35.680 39.390 20.270Independent Variablesβ0 0.0882*** 0.7508 0.4398 0.3763


[0.209] [0.320] [0.231] [0.509]HHICreditLine,Risk,Collateral 0.0524*** 0.1200*** 0.0655*** 0.0734***

[0.019] [0.080] [0.021] [0.025]Dummy of BBR 0.0705*** 0.0915*** 0.0744*** 0.0584***

[0.007] [0.011] [0.008] [0.010]Dummy of BBR * Dummy of Treated Group -0.0664*** -0.0722*** -0.0638*** -0.0610***

[0.008] [0.012] [0.008] [0.008]HHICreditLine,Risk,Collateral * Dummy of Treated Group -0.0402 -0.0675* -0.0376 -0.0626**

[0.025] [0.038] [0.027] [0.030]HHICreditLine,Risk,Collateral * Dummy of BBR -0.0853*** -0.1815*** -0.0868*** -0.1146***

[0.020] [0.030] [0.021] [0.022]HHICreditLine * Dummy of BBR * Dummy of Treated Group 0.0758*** 0.1601*** 0.0812*** 0.1121***

[0.025] [0.030] [0.021] [0.027]Interest Rate Term Structure -1.1738*** -1.1413*** -1.0699***

[0.205] [0.142] [0.170]Overnight Interbank Interest Rate 0.9837*** 0.9081*** 0.6389***

[0.266] [0.184] [0.266]Volatility of Overnight Interbank Interest Rate 7.6038*** 4.8968*** 3.4584


[0.001] [0.001] [0.001]Industrial Production Index -0.0022** 0.0031*** 0.0017

[0.001] [0.001] [0.001]Inflation Index -0.0005** -0.0003** -0.0003

[0.000] [0.000] [0.000]Basel Capital Index 0.0220 0.0179 0.0243*

[0.016] [0.011] [0.014]Liquidity Index 0.0082 -0.0061 -0.0503***

[0.018] [0.012] [0.019]Total Monthly Income Over Net Capital -0.0276** -0.0274*** 0.0040

[0.013] [0.009] [0.009]Total Defaulted Credit Operation 0.0210 0.0203 0.0094

[0.008] [0.005] [0.005]Mean Maturity -0.0001*** -0.0001*** -0.0001***

[0.000] [0.000] [0.000]Net Capital 0.0111 0.0067 0.0094

[0.001] [0.001] [0.002]Dummy of Public Bank 0.0241 0.0197 -0.0151

[0.150] [0.104] [0.086]Year Fixed Effect Yes Yes Yes YesMonth Fixed Effect Yes Yes Yes YesThe econometric models use differences-in-differences method with panel data. We estimate the Market Power and Bankruptcy Reform effects on the MeanInterest Rate of new contracted collateralized loans to firms. (1) The estimations include only fixed-effects of year and month. The Brazilian BankruptcyReform is effective from June/2005. We exclude the outlier observations from the sample. (2) The estimations include controls, fixed-effects of year andmonth. The Brazilian Bankruptcy Reform is effective from June/2005. We keep the outlier observations on the sample. (3) The estimations include controls,year fixed-effect and month fixed-effect. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlier observations from the sample.(4) The estimations include controls, fixed-effects of year and month.. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlierobservations. We symmetrically trunked the monthly data to keep the same number of panels before and after BBR. (a) All estimations include intercept.(b) Robust Standard Deviation in brackets, ∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01.

45

35 / 45

Main ResultsLoan Size (with outliers)

Table 1: Table 9a: Main Results - HHI and Bankruptcy Reform Effect: Credit Volume as DependentVariable

The econometric models use Diff-and-Diff Methodology with panel data. We estimated the Market Power and Bankruptcy Reform effects on the Mean of Credit Volumeof new contracted collateralized loans to firms. The HHI Index is used as Market Power measure.

(1) The estimations include only year fixed-effect and month fixed-effect. The effect of Brazilian Bankruptcy Reform is considered from June/2005. The outliersobservations of interest rate are excluded.

(2) The estimations include controls, year fixed-effect and month fixed-effect. The effect of Brazilian Bankruptcy Reform is considered from June/2005. The outliersobservations of interest rate are kept on the sample.

(3) The estimations include controls, year fixed-effect and month fixed-effect. The effect of Brazilian Bankruptcy Reform is considered from June/2005. The outliersobservations of interest rate are excluded.

(4) The estimations include controls, year fixed-effect and month fixed-effect. The effect of Brazilian Bankruptcy Reform is considered from June/2005. The sample istrunked to keep a symmetric number of panels before and after BBR. The outliers observations of interest rate arent excluded.

(1) (2) (3) (4)

R-sq: within 0.0287 0.0574 0.0566 0.0686Test F F(20,26743) F(33,23555) F(33,23859) F(32,10522)

39.470 43.490 43.370 24.220

Independent Variables

β0

24.5279*** -2.6759 6.2030 165.5696[2.503] [132.135] [130.580] [281.560](0.000) (0.984) (0.962) (0.557)

Market Share - Credit Portfolio-393.9715*** -619.7218*** -635.4845*** 46.0910

[71.901] [73.354] [69.795] [150.675](0.000) (0.000) (0.000) (0.760)

HHICreditLine,Risk,Collateral

-0.3497 -0.2849 -0.1236 1.9331[6.620] [6.654] [6.506] [7.459](0.958) (0.966) (0.985) (0.796)

Dummy of BBR-7.7331*** -5.8297** -5.7529** -1.3178

[2.325] [2.523] [2.492] [3.097](0.001) (0.021) (0.021) (0.671)

Dummy of BBR * Dummy of Treated Group8.5815*** 9.2331*** 9.1169*** -2.1521

[2.617] [2.568] [2.540] [2.453](0.001) (0.000) (0.000) (0.380)

HHICreditLine,Risk,Collateral * Dummy of Treated Group44.5477*** 47.3016*** 46.8313*** 46.5028***

[8.459] [8.465] [8.327] [8.941](0.000) (0.000) (0.000) (0.000)

HHICreditLine,Risk,Collateral * Dummy of BBR3.3293 3.6863 3.5448 -1.5168***[6.756] [6.728] [6.595] [6.395](0.490) (0.550) (0.540) -(0.240)

HHICredit Line, Risk, Collateral * Dummy of BBR * Dummy of Treated Group-4.2503 -8.4554 -8.1399 13.9792*[8.530] [8.485] [8.359] [7.938](0.618) (0.319) (0.330) (0.078)

Interest Rate Term Strutucture-25.9957 -25.3527 184.3105***[45.047] [44.663] [50.374](0.564) (0.570) (0.000)

Overnight Interbank Interest Rate170.4905*** 169.0018*** -9.3177

[58.437] [57.928] [78.767](0.004) (0.004) (0.906)

Volatility of Overnight Interbank Interest Rate877.8463*** 852.1144*** 507.8274

[300.255] [296.653] [827.032](0.003) (0.004) (0.539)

Gross Domestic Product0.5987*** 0.5854*** 0.7017*

[0.187] [0.185] [0.399](0.001) (0.002) (0.078)

Industrial Production Index-0.1281 -0.1227 -0.0752[0.208] [0.206] [0.392](0.539) (0.552) (0.848)

Inflation Index-0.0476 -0.0497 -0.1383[0.049] [0.049] [0.117](0.334) (0.307) (0.238)

Basel Capital Index-18.7869*** -18.6701*** -10.2363**

[3.537] [3.512] [4.013](0.000) (0.000) (0.011)

Liquidity Index0.4676 0.3044 -0.0930[3.914] [3.880] [5.643](0.905) (0.937) (0.987)

Total Monthly Income over Net Capital-1.4979 -1.7507 6.9413***[2.797] [2.776] [2.641](0.592) (0.528) (0.009)

Total Defaulted Credit Operation-0.5248 -0.5189 -0.0141[1.833] [1.822] [1.855](0.775) (0.776) (0.994)

Mean Maturity-0.0009 -0.0008 0.0022[0.001] [0.001] [0.002](0.434) (0.492) (0.199)

Net Capital4.7196 4.6068 11.4326[0.185] [0.182] [0.567](0.000) (0.000) (0.000)

Dummy of Public Bank5.5030 5.6466 -4.1140

[32.916] [32.721] [25.565](0.867) (0.863) (0.872)

Year Fixed Effect Yes Yes Yes YesMonth Fixed Effect Yes Yes Yes Yes

(a) All estimatives considered a intercept term

(b) Robust Standard Deviation in braquets, ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

1

Main Results

36 / 45

Invariance of Banking Competition to BBR

Potential Concerns:

The increase in banking rents/ profits created by BBR may have leaded tobanking entry in more protected credit lines

It could have created more competition in those credit lines

Two strategies to deal with them:

No structural changes: Number of banks, HHI in credit lines

Pre-treatment market power

37 / 45

Invariance of Banking Competition to BBRNumber of Banks: Corporate credit lines

38 / 45

Invariance of Banking Competition to BBRNumber of Banks: Consumer credit lines

39 / 45

Invariance of Banking Competition to BBRConstant Market Power Test

We use pre-treatment level of competition.

HHI is freezed at the one month before the BBR became effective.

Λ̃lrct =

{Λlrct, if t < May/2005

ΛlrcMay/2005, if t ≥ May/2005

Results are similar (quali-quantitatively)

40 / 45

Constant Market Power TestLoan Interest Rates

Table 24: Robustness - Constant Market Power - Interest Rate as Dependent Variable


42.900 29.900 34.460 20.830Independent Variablesβ0 0.2204*** 0.6545 0.4314 2.0394***


[0.245] [0.363] [0.264] [0.321]HHIConsCreditLine,Risk,Collateral -0.0622 0.2015*** -0.0342 -0.0294

[0.039] [0.061] [0.044] [0.023]Dummy of BBR 0.0948*** 0.0938*** 0.0945*** -0.0213*

[0.009] [0.014] [0.010] [0.012]Dummy of BBR * Dummy of Treated Group -0.0813*** -0.0680*** -0.0788*** 0.0045

[0.009] [0.014] [0.010] [0.012]HHIConsCreditLine,Risk,Collateral * Dummy of Treated Group 0.0636 -0.1611** 0.0484 0.0880***

[0.045] [0.070] [0.049] [0.029]HHIConsCreditLine,Risk,Collateral * Dummy of BBR -0.1553*** -0.1768*** -0.1504*** 0.0112

[0.027] [0.039] [0.027] [0.023]HHIConsCreditLine,Risk,Collateral * Dummy of BBR * Dummy of Treated Group 0.1449*** 0.1439*** 0.1381*** -0.0412

[0.032] [0.047] [0.033] [0.028]Interest Rate Term Structure -0.1693 -0.1262 -0.9474***

[0.233] [0.162] [0.193]Overnight Interbank Interest Rate 1.0833*** 1.0043*** -0.5140

[0.302] [0.210] [0.428]Volatility of Overnight Interbank Interest Rate 8.8718*** 5.8852*** 4.2519***


[0.001] [0.001] [0.001]Industrial Production Index -0.0027** -0.0014* -0.0006

[0.001] [0.001] [0.001]Inflation Index -0.0005** -0.0003* -0.0007***

[0.000] [0.000] [0.000]Basel Capital Index -0.0005** 0.0309** 0.0209

[0.000] [0.013] [0.017]Liquidity Index 0.0355* -0.0156 0.0259*

[0.018] [0.014] [0.016]Total Monthly Income Over Net Capital -0.0001 -0.0293*** -0.0903***

[0.020] [0.010] [0.016]Total Defaulted Credit Operation -0.0328** 0.0248 0.0285

[0.014] [0.007] [0.010]Mean Maturity 0.0249** -0.0001*** -0.0001***

[0.010] [0.000] [0.000]Net Capital -0.0001*** 0.0071 0.0054

[0.000] [0.001] [0.001]Dummy of Public Bank 0.0211 0.0177 0.0748

[0.170] [0.118] [0.156]Year Fixed Effect Yes Yes Yes YesMonth Fixed Effect Yes Yes Yes YesThe constant Market Power test simulates an absolute non-endogeneity condition. We maintain the HHI Index on the same values calculated before theBrazilian Bankruptcy Reform. We expected the estimated coefficient for the Market Power and its interaction to be statistically significant; we also expectsimilar level of the estimated coefficients, mainly for model (3). (1) The estimations include only fixed-effects of year and month. The Brazilian BankruptcyReform is effective from June/2005. We exclude the outlier observations from the sample. (2) The estimations include controls, fixed-effects of year andmonth. The Brazilian Bankruptcy Reform is effective from June/2005. We keep the outlier observations on the sample. (3) The estimations include controls,year fixed-effect and month fixed-effect. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlier observations from the sample.(4) The estimations include controls, fixed-effects of year and month. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlierobservations. We symmetrically trunked the monthly data to keep the same number of panels before and after BBR. (a) All estimations include intercept.(b) Robust Standard Deviation in brackets, ∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01.

56

41 / 45

Constant Market Power TestSpread over Interbank Rate

Table 25: Robustness - Constant Market Power - Spread over IRTS as Dependent Variable


53.070 36.070 40.370 20.830Independent Variablesβ0 0.1016*** 0.6402 0.4226 2.0394***


[0.209] [0.318] [0.230] [0.321]HHIConsCreditLine,Risk,Collateral -0.0631* 0.1590*** -0.0417 -0.0294

[0.033] [0.053] [0.038] [0.023]Dummy of BBR 0.0801*** 0.0809*** 0.0815*** -0.0213*

[0.007] [0.012] [0.008] [0.012]Dummy of BBR * Dummy of Treated Group -0.0749*** -0.0583*** -0.0679*** 0.0045

[0.008] [0.012] [0.008] [0.009]HHIConsCreditLine,Risk,Collateral * Dummy of Treated Group 0.0645* -0.1201** 0.0581 0.0880***

[0.038] [0.061] [0.043] [0.029]HHIConsCreditLine,Risk,Collateral * Dummy of BBR -0.1328*** -0.1435*** -0.1220*** 0.0112

[0.023] [0.034] [0.024] [0.023]HHIConsCreditLine,Risk,Collateral * Dummy of BBR * Dummy of Treated Group 0.1189*** 0.1086*** 0.1058*** -0.0412

[0.028] [0.042] [0.029] [0.028]Interest Rate Term Structure -1.2106*** -1.1647*** -0.9474***

[0.203] [0.141] [0.193]Overnight Interbank Interest Rate 0.9637*** 0.8873*** -0.5140

[0.264] [0.184] [0.428]Volatility of Overnight Interbank Interest Rate 7.8161*** 5.1395*** 4.2519***


[0.001] [0.001] [0.001]Industrial Production Index -0.0024** -0.0013* -0.0006

[0.001] [0.001] [0.001]Inflation Index -0.0005** -0.0003* -0.0007***

[0.000] [0.000] [0.000]Basel Capital Index -0.0005** 0.0208* 0.0209

[0.000] [0.011] [0.017]Liquidity Index 0.0243 -0.0092 0.0259*

[0.016] [0.009] [0.016]Total Monthly Income Over Net Capital 0.0052 -0.0264*** -0.0903***

[0.018] [0.009] [0.016]Total Defaulted Credit Operation -0.0285** 0.0217 0.0285

[0.013] [0.006] [0.010]Mean Maturity 0.0219** -0.0001*** -0.0001***

[0.009] [0.000] [0.000]Net Capital -0.0001*** 0.0066 0.0054

[0.000] [0.001] [0.001]Dummy of Public Bank 0.0178 0.0153 0.0748

[0.149] [0.103] [0.156]Year Fixed Effect Yes Yes Yes YesMonth Fixed Effect Yes Yes Yes YesThe constant Market Power test simulates an absolute non-endogeneity condition. We maintain the HHI Index on the same values calculated before theBrazilian Bankruptcy Reform. We expected the estimated coefficient for the Market Power and its interaction to be statistically significant; we also expectsimilar level of the estimated coefficients, mainly for model (3). (1) The estimations include only fixed-effects of year and month. The Brazilian BankruptcyReform is effective from June/2005. We exclude the outlier observations from the sample. (2) The estimations include controls, fixed-effects of year andmonth. The Brazilian Bankruptcy Reform is effective from June/2005. We keep the outlier observations on the sample. (3) The estimations include controls,year fixed-effect and month fixed-effect. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlier observations from the sample.(4) The estimations include controls, fixed-effects of year and month. The Brazilian Bankruptcy Reform is effective from June/2005. We exclude the outlierobservations. We symmetrically trunked the monthly data to keep the same number of panels before and after BBR. (a) All estimations include intercept.(b) Robust Standard Deviation in brackets, ∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01.

57

Robustness

42 / 45

Falsification TestsPlacebo tests

False Months: BBR was implemented

We replaced event-dummy dmLawt by 6 month, 9 months (before andafter the law)

Estimated coefficients for any placebo and its interactions (including β3,β6) are not significant (unstable point estimations)

Robustness

43 / 45

Falsification TestsRandom assignment of competition measures

We randomly switched the HHI between observations (400 draws )Estimated 400 regressions to obtain a distribution of estimatedcoefficientsWe expect that the estimated coefficients (mainly, β6) remain statisticallysignificant in the simulated confidence interval (90%-95%).

Do list: to increase the number of simulations.Robustness

44 / 45

Other proxies for bank competition

Coarse definitions of market power

HHI is computed at credit line level (credit line, month-year)

Results are similar (hampering effect 37%)

Other proxies for market power/competiton

C4lrct: the sum of market share of the first four big lenders

MSblrct: the market share of the bank

H − Statisticslt: Panzar-Rosse competition measure

Results are similar (hampering effect: 30%-60%)

Robustness

45 / 45

bank competition and the limits of creditor protection … generated by an increase in creditor...

Documents