Skip to main content

Table 5 Regression results with the dependent variable Prudence grade

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

  (1) (2) (3)
Low-Prundence Medium Low-Prundence Medium Low-Prundence Medium
Not verified 0.875*** 0.663*** 0.719*** 0.523***   
  (0.00441) (0.00578) (0.00352) (0.00444)   
Income div. 5000 0.929*** 0.908***     
  (0.00550) (0.00852)     
Income div. 10,000 0.998 0.974**     
  (0.00648) (0.0101)     
Amount div. 5000    1.128*** 1.162***   
    (0.00681) (0.0110)   
Amount div. 10,000    0.814*** 0.745***   
    (0.00609) (0.00892)   
Length of title      0.999 0.934***
      (0.00622) (0.00904)
Suspect empl      1.052*** 1.020**
      (0.00586) (0.00895)
ln(Loan amount) 0.00168*** 0.00166***    0.00309*** 0.00248***
  (0.000111) (0.000179)    (0.000248) (0.000321)
ln(Loan amount)2 1.460*** 1.468***    1.420*** 1.449***
  (0.00519) (0.00847)    (0.00614) (0.0100)
Debt to income ratio 1.039*** 1.044*** 1.042*** 1.044*** 1.041*** 1.046***
  (0.000279) (0.000426) (0.000287) (0.000439) (0.000345) (0.000530)
Months since recent inq 0.988*** 0.972*** 0.990*** 0.974*** 0.989*** 0.973***
  (0.000392) (0.000631) (0.000388) (0.000626) (0.000480) (0.000767)
Revolving utilitation 0.0422*** 0.0440*** 0.0491*** 0.0523*** 0.0496*** 0.0592***
  (0.000449) (0.000676) (0.000511) (0.000794) (0.000652) (0.00112)
Account open 0.954*** 0.967*** 0.948*** 0.963*** 0.959*** 0.977***
  (0.000712) (0.00106) (0.000707) (0.00104) (0.000872) (0.00127)
Delinquencies in last 2y 0.622*** 0.625*** 0.607*** 0.610*** 0.616*** 0.620***
  (0.00268) (0.00403) (0.00261) (0.00389) (0.00321) (0.00479)
Purpose: credit card      0.328*** 0.245***
      (0.00438) (0.00533)
Purpose: debt consol      0.555*** 0.485***
      (0.00488) (0.00626)
Purpose: home      0.671*** 0.585***
      (0.00835) (0.0109)
Purpose: vacation      0.947* 0.950
      (0.0297) (0.0451)
Purpose: small business      1.388*** 1.464***
      (0.0319) (0.0422)
Home mortgaged    1.221*** 1.085*** 1.085*** 0.908***
    (0.00609) (0.00859) (0.00675) (0.00886)
Home owned    1.175*** 1.164*** 1.113*** 1.094***
    (0.00886) (0.0135) (0.0105) (0.0157)
Year      0.794*** 0.745***
      (0.00454) (0.00672)
3-digit zip code      Yes Yes
Observations 1,558,546 1,558,546 1,558,546 1,558,546 1,053,512 1,053,512
  1. The dependent variable is created by comparing the LC grade and the new classification of FICO scores based on the percentiles. This table displays odds ratios from Gologit regressions that m-1 models, where m is the number of clusters. The dependent variable is an indicator of Prudence'screening by platforms, and the highest Prudence is the reference category. Prudence degree takes value 1 (Low Prudence) if LC rating grade has been underestimating borrowers'credit risk; Prudence degree takes value 2 (Medium) if the assessing of the borrower riskiness is the same in LC rating grade and FICO score; Prudence degree takes value 3 (High Prudence) if LC rating grade has been overestimating borrowers'credit risk.Imprudent models are displayed in columns 1, 3 and 5, and neutral groups in Columns 2,4 and 6. The variables with superscript α meet the odds assumptions and are the same in all categories (e.g. debt-to-income ratio, home mortgaged, home owned). For variables violating proportional odds assumptions refer to coefficients for responses of 2,3 vs 1 group (Low Prudence models) and category 3 vs 1, 2 in Neutral models