Skip to main content

Table 8 Regression results recovery rate in the overall sample

From: Fintech platforms: Lax or careful borrowers’ screening?

  (1) (2) (3)
Beta Zero-inflate Beta Zero-inflate Beta Zero-inflate
Not verified − 0.042*** (0.0043) 0.1169*** (0.0100) − 0.0054*** (0.0058) 0.138*** (0.0146) − 0.0412*** (0.0055) 0.0896*** (0.0130)
ln(Loan Amount) 0.0003 (0.0030) 0.0445*** (0.0738) − 0.0049 (0.0057) − 0.0531*** (0.0110) 0.0088 (0.0099) − 0.0183 (0.0230)
Term 0.028*** (0.0042) 0.172*** (0.0102) − 0.0061*** (0.0005) 0.171*** (0.0149) − 0.00407*** (0.0055) 0.259*** (0.0135)
Interest rate − 0.0006 (0.0004) − 0.026 (0.0009) − 0.024*** (0.0042) − 0.0225*** (0.0014) − 0.00126*** (0.0005) − 0.0334*** (0.0012)
Revolving utilitation    − 0.0539*** (0.0139) − 0.338*** (0.0364)   
Months since last delinquent    0.0006*** (0.0001) 0.00215*** (0.0003)   
Total account    0.00137*** (0.0002) − 0.0015*** (0.0005)   
Bankcard balance > 75%    6.94e−05 (9.06e−05) 0.0009*** (0.00024)   
Debt to income ratio    0.00182*** (0.0002) 0.00850*** (0.0006)   
Mortgage account    − 0.0053*** (0.0013) − 0.0070** (0.0035)   
ln(Annual Income)      0.0282*** (0.0106) − 0.0423* (0.0245)
Employment length      0.00100 (0.0007) 0.00764*** (0.00165)
Loan to annual income      − 0.263*** (0.0482) 1.008*** (0.105)
Loan purpose: credit card      0.0264*** (0.0098) 0.135*** (0.0242)
Loan purpose: debt consolidation      0.0243*** (0.00870) 0.0775*** (0.0217)
Loan purpose: small business      − 0.146*** (0.0190) − 0.0550 (0.0492)
Home mortgaged      − 0.00809 (0.103) − 0.0492** (0.299)
Home owned      0.0432 (0.103) − 0.0950* (0.299)
3-digit zip      Yes Yes
Year      0.191*** (0.00458) 0.716*** (0.0112)
Observations 291,664   145,646   177,963  
AIC − 105,632   − 62,897   − 73,825  
BIC − 105,516   − 62,630   − 73,381  
  1. The table reports results from beta regression and the logit model on RR with indicators and continuous explanatory variables. Both in the beta models and zero-inflated are reported the coefficients. The standards errors are in parentheses. All models are estimated with intercepts. The primary independent variable is associated with the verification process. Other control variables are inserted, like loan contract information and borrower’ characteristics. For brevity, only the loan’s significance is exposed. The borrowers’ state is based on the first three-digit ZIP code, captured into ten dummy variables building on the classification of the United States. The estimated goodness of fit is shown
  2. ***, ** and * denotes significance at levels 1%, 5%, and 10% levels, respectively