140 likes | 238 Views
Credit Scoring of Bank-affiliated Captive Finance Companies Gabriela P ásztorová CERGE-EI Bratislava Economic Meeting 8 June 201 2. Outline. Topic : Credit Scoring of Bank-affiliated Captive Finance Companies Finding s :
E N D
Credit Scoring of Bank-affiliated Captive Finance Companies Gabriela Pásztorová CERGE-EI Bratislava Economic Meeting 8 June 2012
Outline Topic: Credit Scoring of Bank-affiliated Captive Finance Companies Findings: • consumer loans from bank-affiliated captive finance companies and parent individual lending institutions differ in default rates • the default rate differential between lending institution types is mainly due to differences in application characteristics. Contribution: - default rate analysis on the unique dataset of consumer loans containing loans from bank-affiliated captive finance companies
Motivation Bank-affiliated captive finance companies Independent lending institutions • Direct consumer loans, revolving credit including credit cards • Strict credit scoring mechanism • Low interest costs • Officer - no extrabonuses for selling products • Loans for specific product purchase - increase the sales of themanufacturers • Strict redit scoring of the parent independent lending institution • Low interest costs • Dealer – extra commissionfor the sale of the manufacturer’s product THE DIFFERENCE BETWEEN DEALER’S MOTIVATION IS A SOURCE OF MORAL HAZARD
Focus of interest Research questions Assuming the same credit scoring mechanism and same interest costs, do loans from bank-affiliated captive finance companies and parent individual lending institutions have the same default rate? Are individual lending institutions consistently granting riskier installment loans through bank-affiliated captive finance companies and if so, is it the issue of the bank or economy?
Methodology – 1/3 Probit model Carey et al. (1998), Barron et al. (2008), Bertola et al. (2002) 0 if the borrower’s loan does not default 1 if the borrower’s loan defaults Xijapplication characteristics (age, education, income, employer …) Dijdummy variable on loans provided by bank-affiliated captive finance company standard normal distribution Focus of interest:
Methodology – 2/3 Propensity score matching (Rubin, 1974; Rosenbaum and Rubin, 1983, Heckman and Vytlacil, 2005) probability of default if the borrower applies for a loan in the bank-affiliated captive finance company) probability of default if the borrower applied for a loan in the independent lending institution) Treatment effect on treated: Matching assumptions: 1. Unconfoundedness 2. Common support Focus of interest
Methodology – 3/3 Oaxaca-Blinder decomposition (Oaxaca, 1973; Blinder, 1973; Fairlie, 1984) - decomposition of the mean outcome differential • quantifying the individual contributions of different observables, and the individual contributions of estimated coefficients from a nonlinear regression
Data • consumer loan data from a Czech commercial bank -application data - performance indicator data - payment time series data • Over 5,000 individuals who were granted a consumer loan between January 2001 and February 2006 • Loan repayment monitored till October 2008 • 4,2 % default rate • Loans both from the independent lending institution and the bank-affiliated captive finance company
Data Figure 1. Distribution of the loan amount (0 – 100 000 CZK) Direct consumer loans Captive loans Source: Author’s calculations Table 1. Sample statistics on the share of defaultedand captive loans Source: Author’s calculations
Findings – 1/3 Table 2. Estimation results of the probit model (compared to linear probability model) Source: (1) Author’s computations, 2001-2008. (2) In case of LPM model, the estimates denote the LPM coefficients, and in case of probit model, the estimates denote the calculated average marginal effects for factor levels (dy/dx) expressing the discrete change from the base level. (3) * represents statistically significant at 10%, ** statistically significant at 5%, and *** statistically significant at 1%. .
Findings – 2/3 Table 3. Test of the balancing property Table 4. Estimation results of the ATT with the stratification matching Source: Author’s calculations. Bootstrapped standard errors.
Findings – 3/3 Table 5. Default rate decomposition results by CAPTIVE Source: Source: Author’s computations, Number of obs (A) = 372, Number of obs (B) = 4902
Summary • loans from bank-affiliated captive finance companies are less likely to be repaid than loans from independent lending institutions • after controlling for observable application characteristics the effect of being granted a loan through a bank-affiliated captive finance company results in significantly different default rates than the effect of loans granted from an independent lending institutions. • the default rate differential between lending institution types is mainly due to differences in application characteristics • Why borrowers with worse application characteristics are eventually given a loan may be explained by the dealers’ financial incentives from selling the product.