The Distributive Impact of Reforms in Credit Enforcement: Evidence from Indian Debt Recovery Tribunals

Size: px

Start display at page:

Download "The Distributive Impact of Reforms in Credit Enforcement: Evidence from Indian Debt Recovery Tribunals"

Posy Booker
5 years ago
Views:

1 The Distributive Impact of Reforms in Credit Enforcement: Evidence from Indian Debt Recovery Tribunals Stockholm School of Economics Dilip Mookherjee Boston University Sujata Visaria Boston University March 10, 2009

2 Motivation The usual argument: Weak enforcement of credit contracts restricts the functioning of credit markets Borrowers cannot credibly commit to repay loans lender risks rise cost of credit increases credit access decreases partial equilibrium argument

3 Motivation, contd. Our argument: strengthening credit enforcement has ambiguous effects: not always a Pareto improvement due to general equilibrium (GE) effect

4 This Paper PE effect: stronger enforcement every borrower is more creditworthy than before demand increases upward sloping supply of credit (or other input) GE effect: interest rate (or input price) increase

5 This Paper PE effect: stronger enforcement every borrower is more creditworthy than before demand increases upward sloping supply of credit (or other input) GE effect: interest rate (or input price) increase Borrower heterogeneity PE effect more important for large firms Possible redistribution: small (poorer) borrowers may become more credit-constrained, while large (wealthier) borrowers gain average effects may conceal these distributional changes

6 This Paper Model with PE and GE effect Evidence from firm-level panel data examine the effects of Indian debt recovery tribunals reform increasing enforcement leads to credit reallocation from small to large borrowers

7 Borrowers with heterogenous (collaterizable) fixed assets W Borrower invests in project at scale γ upfront investment costs γ I return of y f (γ), with borrower-specific shock y {y s, y f } probability of success (y = y s ) is e, given endogenous project scale γ Contract {γ, T s, T f }: T k repaid in state k {s, f }

8 Enforcement and default Default: lender can seize θ fraction of assets W Enforcement institution represented by θ, incorporating delays and/or uncertainties in the collection process; these are affected by judicial reforms such as DRT Reputational loss due to default d Entrepreneur will not default in state k iff payment = T k θ W + d = cost of default

9 Individual Demand with θ π W = 0 W > 0 Aggregated Demand and Supply with θ π Supply π 2 π 1 γ 2 γ 1 γ 1 γ 2 L d (θ, π), L s (π)

10 Results θ δ = θ 0 < ɛ < δ < θ δ = 0 W L W L γ H γ F 6 γ F γ γ H W W L W L W W L W L W A: No GE-effect: B: Weak GE effect C: Strong GE effect: γ F Figure: Impact of changing liability law with GE effects.

11 Description and data Specification Results Context for Empirical Work: 1993, national law to set up debt recovery tribunals (DRTs) across country DRT: specialized court which only processes debt cases, follows new streamlined procedure Suit can be filed by lenders in a DRT in a given state if claims exceed Rs 1 million

12 DRTs, continued Introduction Description and data Specification Results What happens in DRTs? Speed: Fast processing of cases Defendants given less time to respond Defendants not allowed to file counter-claims increased cost of appeals Power: DRTs allowed to make interim orders defendant cannot transfer or dispose assets DRTs allowed to obtain arrest warrants Effect: Visaria (2009): decreased time taken to process cases increased repayment on loans subject to DRT limit

13 Description and data Specification Results DRTs, continued DRTs set up in different states at different times between Timing of establishment of DRTs across states not related to state economic condition, prior cases pending in courts, or political variables (Visaria 2009)

14 Description and data Specification Results Exogenous introduction of DRTs (Visaria 2009) Dependent variable: Does a state have DRT in a specific year? Explanatory variables include: Judicial data Economic Variables Political Variables observations from for 23 states

15 Description and data Specification Results Exogenous introduction of DRTs (Visaria 2009) Judicial variable cases pending in the state High Court number of judges in the state High Court per capita Economic variables level and growth rate of state-level GDP Political variables nature of the dominant political party in the state government dominant part is ally of the party in power at national level

16 Table 15: Predicting the pattern of DRT incidence (1) (2) (3) (4) (5) (6) (7) (8) GDP per capita 0.07 (0.05) GDP per capita growth rate (0.00) (0.00) (0.04) (0.00) Cases pending per capita (0.00) (0.01) (0.11) (0.03) Number of judges per capita (4.14) (7.38) (62.35) (5.65) State government Congress & allies (0.11) (0.13) (0.91) (0.13) Janata & allies (0.12) (0.14) (0.86)* (0.16) Communist party (0.12) (0.13) (0.85) (0.15) Regional party (0.11) (0.13) (0.82) (0.17) Center s ally (0.06) (0.08) (0.61) (0.07) Regression OLS OLS OLS OLS OLS OLS Probit Region FE Observations R squared Source: Visaria, Sujata (2009) Legal Reform and Loan Repayment: The Microeconomic Impact of Debt Recovery Tribunals in India, American Economic Journal: Applied Economics, Forthcoming. Notes: The dependent variable takes value 1 if state i had a functional DRT in year t, and 0 otherwise. Year dummies in all columns not reported. Observations correspond to 8 years of data ( ) for 23 states. Union territories are excluded. GDP growth rates are not available for In column (8), the group variable is DRT region. Standard errors in parentheses are clustered by DRT region. * significant at 10%; ** significant at 5%; *** significant at Psuedo R squared.

17 Description and data Specification Results Data: PROWESS accounting database of large and medium sized Indian firms headquarter location financial information

18 Empirical specification Introduction Description and data Specification Results Dependent variable y ijt for borrower i located in state j in time t: 1. borrowing 2. fixedassets 3. wage bill 4. profits Regress y ijt on DRT dummy D jt for state j at time t interaction of DRT dummy with fixed-asset size class of firm i at some previous time t time*assetsize and borrower FE; cluster at state-time level

19 Description and data Specification Results Empirical specification OLS specification K K y ijt = α 0 + α 1 D jt + α 2k size ik + α 3k (D jt size ik ) k=2 k=2 +state j + year t + industry i + year t assetsize i + year t assetsize 2 i + ɛ ijt y ijt : credit, assets, profit, for firm i in state j in year t. D jt : value 1 in years when the firm s state has a DRT and 0 otherwise size ik : size class indicator: use 2 or 4 classes

20 Description and data Specification Results Empirical specification First DRT occurs in 1994, last one in 1999, we use for the regressions Potential endogeneity of asset size: we use only pre-drt asset size measures Use HQ location to determine which state-drt applies

21 Table 1: Summary Statistics, Whole Sample Quartile 1 Quartile 2 Quartile 3 Quartile 4 Assets (as of 1990) (74.43) (0.79) (1.31) (3.29) (130.42) Borrowing (380.24) (36.24) (59.43) (79.16) (710.61) Secured Borrowing (262.03) (31.17) (54.87) (80.20) ) Long term Borrowing (339.67) (27.73) (42.84) (63.57) (641.47) Net Fixed Assets (549.38) (35.98) (49.00) (88.39) ) Profits (PBDIT) (156.85) (11.43) (14.93) (27.00) (297.05) Salaries (30.74) (4.38) (8.60) (13.03) (52.22) N All values are reported in tens of millions of rupees. Standard deviation in parentheses.

22 Table 2: Dependent Variable: Borrowing Average DRT*Small DRT*Large DRT*Quartile 1 DRT*Quartile 2 DRT*Quartile 3 DRT*Quartile 4 Ordinary Least Squares Borrower Fixed Effects (4) (5) (6) (10) (11) (12) *** ** *** ** ** ** *** ** ** * Year Dummy controls Yes Yes Yes Yes Yes Yes Size x Year Dummy controls Yes Yes Yes Yes Yes Yes Size^2 x Year Dummy controls Yes Yes Yes Yes Yes Yes N R p-values in italics. Standard errors clustered at the state x industry level. OLS regressions include size dummies, industry dummies, state dummies.

23 Table 3: Dependent Variable: Secured Borrowing Average DRT*Small DRT*Large DRT*Quartile 1 DRT*Quartile 2 DRT*Quartile 3 DRT*Quartile 4 Ordinary Least Squares Borrower Fixed Effects (4) (5) (6) (10) (11) (12) *** *** ** * *** *** *** ** * Year Dummy controls Yes Yes Yes Yes Yes Yes Size x Year Dummy controls Yes Yes Yes Yes Yes Yes Size^2 x Year Dummy controls Yes Yes Yes Yes Yes Yes N Rp-values 2 in italics. Standard errors clustered 0.68 at the state 0.68 x industry 0.68level p-values in italics. Standard errors clustered at the state x industry level. OLS regressions include size dummies, industry dummies, state dummies.

24 Table 4: Dependent Variable: Long-term Borrowing Average DRT*Small DRT*Large DRT*Quartile 1 DRT*Quartile 2 DRT*Quartile 3 DRT*Quartile 4 Ordinary Least Squares Borrower Fixed Effects (4) (5) (6) (10) (11) (12) *** ** *** ** *** ** Year Dummy controls Yes Yes Yes Yes Yes Yes Size x Year Dummy controls Yes Yes Yes Yes Yes Yes Size^2 x Year Dummy controls Yes Yes Yes Yes Yes Yes N R p-values in italics. Standard errors clustered at the state x industry level. OLS regressions include size dummies, industry dummies, state dummies.

25 Table 5: Dependent Variable: Net Fixed Assets Average DRT*Small DRT*Large DRT*Quartile 1 DRT*Quartile 2 DRT*Quartile 3 Ordinary Least Squares Borrower Fixed Effects (4) (5) (6) (10) (11) (12) *** *** *** *** *** ** *** ** *** ** DRT*Quartile Year Dummy controls Yes Yes Yes Yes Yes Yes Size x Year Dummy controls Yes Yes Yes Yes Yes Yes Size^2 x Year Dummy controls Yes Yes Yes Yes Yes Yes N R p-values in italics. Standard errors clustered at the state x industry level. OLS regressions include size dummies, industry dummies, state dummies.

26 Table 6: Dependent Variable: Profits before Depreciation, Interest & Taxes Ordinary Least Squares Borrower Fixed Effects (4) (5) (6) (10) (11) (12) Average DRT*Small DRT*Large DRT*Quartile 1 DRT*Quartile 2 DRT*Quartile 3 DRT*Quartile *** *** ** 8.21 *** *** *** *** *** ** *** Year Dummy controls Yes Yes Yes Yes Yes Yes Size x Year Dummy controls Yes Yes Yes Yes Yes Yes Size^2 x Year Dummy controls Yes Yes Yes Yes Yes Yes N R p-values in italics. Standard errors clustered at the state x industry level. OLS regressions include size dummies, industry dummies, state dummies.

27 Table 7: Dependent Variable: Salaries Ordinary Least Squares Borrower Fixed Effects (4) (5) (6) (10) (11) (12) Average ** *** DRT*Small DRT*Large DRT*Quartile 1 DRT*Quartile 2 DRT*Quartile 3 DRT*Quartile *** *** *** *** *** *** * Year Dummy controls Yes Yes Yes Yes Yes Yes Size x Year Dummy contr Yes Yes Yes Yes Yes Yes Size^2 x Year Dummy con Yes Yes Yes Yes Yes Yes N R p-values in italics. Standard errors clustered at the state x industry level. OLS regressions include size dummies, industry dummies, state dummies.

28 Description and data Specification Results Continuous DRT variable DRT dummy variable we also consider continuous definition: percentage of filed cases in the DRT that had been disposed by that year. Idea The effective quality of DRT is relevant Similar results

29 Description and data Specification Results Potential other explanation What if DRT less effective for smaller firms? legal suit data as a proxy for effectiveness Data: Random sample of 49 debt recovery suits filed in Maharashtra by a large private bank cases filed in civil courts before establishment of DRT filed in DRT The evidence: DRT increases speed for settlement No adverse effect of DRT for smaller claim sizes

30 Description and data Specification Results Alternative explanations, cont. 1. Other channel for GE effect: labor market wage bill decreased; not consistent with increased wages 2. Incomplete contracts: Bankruptcy law provides insurance Preliminary evidence that interest rates increase (using other data); work in progress

31 Description and data Specification Results Conclusions project financing model of firms GE effect in the supply of funds (or other inputs) Changing credit enforcement may lead to redistribution of credit Possible that productive efficiency decreases

32 Description and data Specification Results Conclusions evidence for redistributive effects of Indian Debt Recovery Tribunal reform financial (i.e. credit) and real decisions (fixed assets, profits, wage bill) affected work in progress: better differentiate alternative channels consistent with redistributive effects

The Distributive Impact of Reforms in Credit Enforcement: Evidence from Indian Debt Recovery Tribunals

The Distributive Impact of Reforms in Credit Enforcement: Evidence from Indian Debt Recovery Tribunals Work in Progress: Very Preliminary Ulf von Lilienfeld-Toal, Dilip Mookherjee & Sujata Visaria May