PE and entry

1
11th Conference of the ECB-CFS NetworkSession
IVA Retail Finance and SMEs Discussion by
Alexander Popov(European Central Bank)October
21, 2008
2
Hempel and Fischer, Rate Dispersion in Retail
Banking Do Local Factors Still Matter?

• Motivation
• cross-border trade in retail banking still small
  compared to what is expected no pan-European
  model in place even for the largest European
  banks
• while banking competition has intensified within
  countries, no evidence is available on the effect
  of local market power
• Main idea characteristics of national markets
  used in recent studies to explain this
  phenomenon insufficient -gt go in the direction
  of Guiso et al (2004) and Huang (2006)
• Main hypothesis
• Local factors matter for retail financial
  integration
• Draw policy conclusions is interest rate
  dispersion higher across regions within a
  country, or across countries?
• Main findings
• Local market power of banks matters for
  short-term deposits and credit to enterprises
• 1 st. dev. increase in competition leads to
  widening of interest rate spread by between 16
  and 19 basis points.
• Local factors do not matter for consumer loans
  and retail mortgages

3
Hempel and Fischer, Rate Dispersion in Retail
Banking Do Local Factors Still Matter?

• General remarks
• Choice of sample/country
• Focusing on Germany only prevents from evaluating
  the relative importance of local vs. national
  markets
• Regional banks only but market behavior will
  matter depending on whether dominant bank in
  local market is local or national
• Rate calculations yield curve?
• 10 rate on a 3-month deposit is not the same as
  a 10 rate on a 12-month deposit (in loans to
  enterprises, for example, you are using gross
  interest rates spreads).

4
Hempel and Fischer, Rate Dispersion in Retail
Banking Do Local Factors Still Matter?

• Specific remarks
• Construction of indices
• How are the HH and Lerner index calculated? By
  assets? Clarification needed!
• If by assets, then you are mis-measuring true
  concentration.
• For example 5 banks, all equally big asset-wise,
  but one does most of the retail banking and the
  others do mostly commercial banking
• Index can still not capture true local
  competition
• For example, imagine region Bundesregion 17
• Three equally big cities Bad Schon, Bad
  Sehrschon and Bad Wunderbahr
• A different bank has monopoly in each city HH
  will tell you that there is triopoly while in
  fact there are 3 local monopolies!
• Empirical methodology
• Income per capita, unemployment and insolvency
  rates may be collinear with bank sector
  concentration. (Black and Strahan (QJE 1996),
  Beck et al (WB 2007)). Regression without those?
  Correlation matrices?
• Omitted variable bias industrial concentration
  (Cetorelli (Fed Chicago 2002))

5
Hempel and Fischer, Rate Dispersion in Retail
Banking Do Local Factors Still Matter?

• Specific remarks (continued)
• Demand considerations
• Proxies for demand human wealth, non-human
  wealth and demographics. Only first one present.
• Instrument
• Number of banks in 1986 not a good instrument in
  a dynamic market (not a good analogue to Guiso et
  al. (2004) procedure where banking structure is
  fixed over time). First stage regressions?
• Interpretation of results
• Bank local power matters for short-term deposits
  and credit to enterprises
• Political economy explanation for fact 2?
  (Incumbents teaming with banks to prevent new
  entrants by charging them high rates?)
• But why would enterprises be restricted in taking
  credit from local sources only? Wouldnt they
  take trade credit if loan rates became too high?
  Or VC?
• No effect on consumer loans counterintuitive?
  Why are consumers more mobile in their decisions
  than enterprises?

6
Carbo-Valverde et al. Bank Lending, Financial
Constraints and SME Investment

• Main idea examine the sensitivity of investment
  to bank loans and trade credit
• Thus, circumvent econometric challenges
  associated with studying the sensitivity of
  investment to cash flows (Fazzari et al vs.
  Kaplan and Zingales)
• Focus on SMEs as the most obvious firms to whom
  bank and trade credit matter
• Investigate causality channels
• Main hypothesis
• Desired loans gt actual loans for constrained
  firms gt Investment not sensitive to bank loans
  for those
• Desired loans gt actual loans for constrained
  firms gt Investment sensitive to bank loans for
  those

7
Carbo-Valverde et al. Bank Lending, Financial
Constraints and SME Investment

• Data
• 30,897 Spanish SMEs (less than 250 employees)
  from Amadeus
• 1994-2002
• Total of 278,073 observations
• Main findings
• Investment is sensitive to bank loans for
  unconstrained firms
• Investment is sensitive to trade credit for
  constrained firms
• Main contribution
• Moves away from the investment-cash flow
  sensitivity literature
• Uses Granger predictability tests

8
Carbo-Valverde et al. Bank Lending, Financial
Constraints and SME Investment

• General remarks
• Size
• Why SMEs? Sounds sexy, but more theoretical
  motivation needed
• Some descriptive stats how important are SMEs
  to the Spanish economy?
• Definition of SME broad non-monotonic
  relationship to size? Split the sample
• Defining constraints
• Non-direct measure of constraints can generate
  large classification errors and weaken the
  validity of existing tests for liquidity
  constraints (Japelli (1990))
• Firms may have preferences over investment
  sources (Hines and Thaler) can the Euler
  equation method account for a possible pecking
  order?

9
Carbo-Valverde et al. Bank Lending, Financial
Constraints and SME Investment

• Specific remarks
• Data issues
• Customary to exclude certain industries subject
  to specific regulations (utilities, financial
  sector)
• Industry characteristics use a Rajan-Zingales
  criterion to exclude industries for which
  external finance doesnt matter.
• Empirical procedure
• The power of a Granger test diminishes quickly as
  the sample is reduced. 3 lags in a 9-period
  panel?
• Robustness tests
• Is investment still non-sensitive to bank loans
  for constrained firms once you take out the
  sub-sample of fully-constrained firms (2,426)?

10
Schaeck and Berger Small and medium-size
enterprises, Banking Relationships, and the Use
of Venture Capital

• Main idea examine the determinants and
  performance effect of VC on firms
• Test whether VC is a substitute for multiple
  lending relationships (firms get VC to avoid the
  rent-extracting behavior of the main bank)
• Focus on SMEs, but distinguish by age and growth
  potential
• Use an econometric technique (matching) to
  circumvent the fact that venture capitalists are
  targeting more productive firms
• Main findings
• Firms with multiple banking relationships are
  less likely to use VC
• Venture capital backed firms grow faster and
  spend more on RD
• Main contribution
• Look at SMEs
• Link between banking relationship and VC
• European data (Italy, Germany, and UK)

11
Schaeck and Berger Small and medium-size
enterprises, Banking Relationships, and the Use
of Venture Capital

• General remarks
• Three papers 1) firm-level determinants of VC
  2) VC vs. multiple bank relationships 3)
  VC-backed firms performance
• Survey data 247 firms in UK, 162 in Italy and
  152 in Germany. Small size!
• How representative in terms of size and banking
  relationships? (reported age and VC use only)
• How representative is the data regionally? As
  pointed out, VC-using firms tend to be clustered.
  
• How comparable is the environment in terms of
  economic dynamics? For example, share of new
  firms is 3.5 in Italy, 12.3 in Germany and 15
  in UK (Rajan et al (JFE 2006)).
• SMEs how important?
• Page 1 23 million SMEs account for 99 of all
  companies in Europe (Observatory for European
  SMEs (2004))
• If true, then we need to change the definition!
• Negates contribution 1
• Given that 99 of firms are SMEs, doesnt it make
  sense to distinguish among other lines for
  instance, YIC vs. non-YIC?

12
Schaeck and Berger Small and medium-size
enterprises, Banking Relationships, and the Use
of Venture Capital

• Specific remarks
• Section on facts and figures added value?
• Account for the banking structure local factors
  matter (Hempel and Fischer). Maybe VC is used
  rather than multiple banking relationships
  because only 1 bank is available locally?
• Too little variability in unobservable country
  factors once Italy is dropped (3 of firms
  obtaining VC so it should be).
• In particular, accounting for difference in the
  supply of VC and its determinants becomes tricky.
• Supply of VC use variation in the prudential
  regulation of institutional investors across
  countries and over time but 2 countries?

13
Schaeck and Berger Small and medium-size
enterprises, Banking Relationships, and the Use
of Venture Capital

• Selection issues
• Only observe firms with banking relationships if
  they 1) desired bank credit, and 2) were not
  denied bank credit
• Only observe firms that get VC if they 1) desired
  VC, and 2) were not denied VC.
• 4 selection criteria!
• Increase in employment is an indicator variable
  no information about economic significance
• Instrument dummy1 if bank on firms board can
  also be correlated with the decision to use VC.