52_Degryse

							Distance and Information Asymmetries
         in Lending Decisions

                                       by

             Sumit Agarwal and Robert Hauswald
                         (& sons)

                                 Discussant

                               Hans Degryse
        CentER – Tilburg University, TILEC, K.U. Leuven and CESIfo
            TILEC-AFM Chair on Financial Market Regulation

     Conference on the “Changing Geography of Banking”, Ancona, September 2006


                                                                                 1
                    General Issue
• Does distance still matter?
   – Petersen and Rajan (JF2002): US
      • No, as distance between lender and borrower has quadrupled
        from 1970’s to 90s
      • Yes, as loan rate decreases in distance
   – Degryse and Ongena (JF2005): Belgium
      • Yes, distance between lender and borrowers did not increase
        substantially over from 1975-97
      • Yes, as loan rate hinges on distance to lender and distance to
        competitor
   – This paper: US
      • Loan pricing: when controlling for score, no
      • Loan volume and switching: yes                                   2
General Issue: evidence




                          3
 General Issue: why should distance matter?

• What does distance capture?
  – Petersen and Rajan (JF2002): US
     • No, as distance between lender and borrower has
       quadrupled from 1970’s to 90s
     • Yes as loan rate decreases in distance
  – Degryse and Ongena (JF2005): Belgium
     • Yes; distance between lender and borrowers did not
       decrease dramatically
     • Yes, as loan rate hinges on distance to lender and
       distance to competitor
                                                            4
            This paper’s findings
•   Loan rate determinants
    –   decreases in borrower-lender distance (DL) and
        increases in borrower-competitor distance (DC)
    –   Distance loses statistical significance when introducing
        the internal score of lender
    –   Interaction between score, and DL and DC shows that
        higher scored firms that are closer pay higher rates,
        hinting at asymmetric information being the driver
    –   Nice treatment of potential selection issues
•   Decision to offer credit
    –   Decrease in DL and increase in DC
    –   Increase in score; distance remains significant
                                                              5
           This paper’s findings
•   “Switchers away from Bank”
    –   Increase in DL and decrease in DC
    –   Interaction of score, and DC and DL suggest
        that higher scored firms are more likely to
        switch when DC is large, suggesting
        asymmetric information
•   delinquency: further away borrowers are
    more delinquent

                                                      6
  Comments:Setting more inclined
  towards asymmetric information?
“transportation costs” vs. “asymmetric information”

• Degryse and Ongena (JF2005): Belgian
  sample: 44% borrowed from closest branch

• US: “borrowers do not turn to the closest
  branch but prefer to borrow from further
  away”
                                                      7
Comments:Setting more inclined
towards asymmetric information?
1. US versus Belgium: absolute differences in distance

2. US versus Belgium: huge relative differences in DL
   versus DC




=> how would results be in a restricted sample with
    “similar” absolute and/or relative differences
                                                         8
         Comments: asymmetric
            information?
• Inclusion of “score” in loan pricing model
  – Removes statistical significance of distance as
    standard errors increase
  – Magnitude of coefficients and economic
    significance, however, often remains
• Main Bank and Duration of Relationship
  (Months on Books)
  – Relatively short duration: median of 30 months
  – both have a negative coefficient in loan rate
    regression => asymmetric information would
    suggest a positive coefficient                 9
           Comments: Is it asymmetric
                information?
•   Suppose branch managers know that the bank applies the following pricing
    model:

     Loan rate = f (Score)

     => Branch managers reflect local market power in score => borrowers where the branch
        manager asserts to enjoy a lot of market power get low score

        (see Ioannidou and Ongena (2006): competing bank credit scores get adjusted to allow
        loan officers to give borrowers lower rate to attract borrowers)

     => distance and competitive setting gets reflected into score

•   On p. 8-9 the authors mention: our bank estimates that, on average, 20 to 30% of
    our bank’s scores consists of soft information

•   Suggestion:
        orthogonalize score to investigate the separate impact of distance and
        check in how far “internal score” differs from “external score”
                                                                                       10
       Comments: Technology and
          potential selection
• Loans are from January 2002-April 2003, a recent
  period when new technologies may already be in
  place
   – Is the bank only granting loans via its branches, or are
     also loans through other channels like
      • Internet
      • Online banking
      • Other stores/shops, available
      => i.e. do you cover all the loan applications and approvals of the
        bank?
      => if not, potential selection issues

   – Similar issues and question for rival banks?
                                                                       11
         Minor Comments
– Do “competing branches” include own bank’s
  branches?
– “Switching away from bank” versus “switching
  towards bank”
– Should “months on books” and “main bank” and
  some other variables still play a role when
  internal score is good proxy of private
  information?


                                            12
                 Summary
• Interesting topic? Yes!

• Interesting paper? Yes!

• Do I recommend that you read it? Yes!

• Am I convinced? May be … Not Yet

                                          13