Docstoc

IIIb_Egert_discussion

Document Sample
IIIb_Egert_discussion Powered By Docstoc
					Discussion
Session IIIb: Retail Finance
in the New EU Member States



       Balázs Égert
OECD, Economics Department
De Haas, Ferreira and Taci: What determines banks’
  customer choice? Evidence from transition
  countries

Seč and Zemčik: The impact of mortgages, house
  prices and rents on household consumption in
  the Czech Republic



   TWO VERY INTERESTING PAPERS
   UNIQUE DATASETS
    Results De Haas, Ferreira and Taci
   Small and private domestic banks tend to lend to
    SMEs
   ‘Green field’ foreign banks are found to lend to
    foreign-owned private firms (as opposed to domestic
    banks overtaken by foreign investors)
   Foreign banks appear
       to service households
      to shift lending away from state-owned firms to private
       sector customers
   Better legal protection
        fosters mortgage lending
   Small and private domestic private banks tend to
    lend to SMEs

Because
   Large banks have economies of scale in lending
    activities

     Small banks are constrained by regulatory
      limits (portfolio diversification)
But also because
   Small banks tend to be domestically owned
       They can lend at higher prices and are not
        competitive (they are riskier, no access to the internal capital market of
           the international banking network that has direct access to world capital markets)

          SMEs      are too risky for large banks; => small
           domestic banks are left with SMEs => therefore
           themselves more risky => higher costs of capital => not
           competitive enough for large firms
          Does not have the technology of processing
           hard information (lack of capital)
Therefore, it would be nice to see an interaction of
          BANK SIZE and OWNERSHIP

 Small banks * domestic banks
 Large banks * domestic banks

 Small banks * foreign owned banks
 Large banks *foreign owned banks
    Or, even better, let the data determine what
                  is small and large

-    Nonlinear models à la Hansen (1999)
-    Multiple regimes for the interaction term
     (nonlinearity in the interaction term)
-    Bank size as the threshold variable





             n 1

             j  X j  1  X n  
         
                                                       if    T
               j 1
      Y     n 1
             j  X j   2  X n                if    T
         
             j 1




    where rho is the threshold variable
    T denotes the threshold that separates the two regimes
 ‘Greenfield’ foreign banks tend to lend to
  foreign-owned private firms (as opposed to
domestic banks overtaken by foreign investors)


  Because they do not have the local
              knowledge

 But isn’t this related to the specific
        features of the sample?
Very few countries where foreign banks
    have been present for a long time
              Asset share of foreign-owned banks (in per cent)
 100



                                                                 1998
                                                                 2005
  50




   0




             serbia
              russia




             bosnia
          bulgaria
           estonia




            croatia
        czech rep




       macedonia




          slovenia
            poland




           belarus
           ukraine
          hungary




          slovakia




          romania
            albania
              latvia




         moldova
         lithuania




       kazakhstan




 Source: EBRD transition indicators
                         Generally, there is an oversampling of
                         banks from countries with less reformed
                         banking sectors (if outliers are dropped)
banking sector reforms




                         4.00




                         2.00
                                0%       10%    20%     30%     40%    50%        60%

                                     number of banks included / total number of
                                                banks in a country
   Foreign banks tend
      to service households
     to shift lending away from state-owned firms to private
      sector customers

    This is not very surprising

      Foreign banks seek to increase market shares in
      household lending
     Bank privatisation usually goes hand in hand with
      overall restructuring
           Privatisation of noncorporate private sector => so foreign
            banks cannot lend to state-owned firms because there are not
            any left.
                             4
  large scale privatisaton



                             3


                             2


                             1


                             0
                                 0.00   1.00        2.00       3.00   4.00

                                           banking sector reform

Source: EBRD transition indicators
   Better legal protection
       fosters mortgage lending


    Again, the overall context should not be ignored

     Bank privatisation is related to
           Privatisation of noncorporate private sector

           Financial market developments

           Improvement of the overall legal environment
                     4




competition policy
                     3


                     2


                     1


                     0
                         0.00         1.00          2.00        3.00    4.00

                                         banking sector reform

                           4
nonbank fin institutions
 securities markets and




                           3


                           2


                           1


                           0
                               0.00     1.00          2.00       3.00   4.00

                                             banking sector reform
    Results Seč and Zemčik
   House price increases       WEALTH EFFECT

    higher consumption of homeowners
    no impact on renters

   Rises in rents               INCOME EFFECT

    higher consumption of homeowners
    Lower consumption of renters

   Changes in mortgage payments do not influence
    consumption
Aggregate house prices in CR
 20
                           nominal house price growth
                           real house price growth
 15


 10


 5


 0


 -5
      2000   2001   2002      2003      2004         2005
Large regional differences?
 Changes in aggregate house price
 do not seem very important

 Are there substantial regional
 differences?
(the authors find that data is not driven by
  cross-section heterogeniety)
Wealth effect: how important?
       Real house prices and household consumption
  20
          real growth of final consumption expenditure of households

  15
          real house price growth


  10


   5


   0


  -5
            2002            2003           2004            2005
Wealth effect: how important?
        Wealth effect and household consumption
   20
           real growth of final consumption expenditure of households
   15      wealth effect

   10


   5


   0


   -5
            2002           2003            2004           2005
Wealth effect: how important?

   How does this effect compare to other countries?
    For example, results are very controversial for the UK
    • Campbell and Cocco (2007, JME)              YES
    • Attanasio et al. (2005, IFS)                NO
    • Cristini and Sanz (2008)                    results are not robust
                                                  to alternative
                                                  specifications
Wealth effect: how important?
   Why should the wealth effect be important if it is not
    for instance in the UK?
   How do house prices affect homeowners’
    consumption?
       Home equity withdrawal? (anecdotal evidence for CR?)
       Expectations? (if income is controlled for, savings should decrease)
   Different effects across age groups (young vs. old)
   How does the wealth effect work for members of real
    estate cooperatives?
           Estimation issues
•The authors argue that regional and time effects are
not important => surprising because those effects
usually explain a very important part of variation of
the data
  •Regional fixed effects
  • Time fixed effects (instead of a time trend)
(R-squared is fairly low now)

•Why not use GMM if endogeniety may be an issue
(small T, large N)?
Thank you

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:7
posted:6/25/2011
language:English
pages:24