Are Competitive Banking systems more stable?

1
Are Competitive Banking systems more stable?

• Discussion
• By Erlend Nier

2
Summary

• Two views
• Competition fragility (C-F) view
• Competition stability (C-S) view
• Question is competition associated with increase
  or decrease in likelihood of systemic crises?

3
Summary

• Analyse this using cross-country regressions
• In sample of 38 countries over 1980-2003
• Measuring competition by Panzar- Rosse
  H-statistic, Claessens and Laeven (2004)
• Using account of systemic crises provided by
  Demirguc-Kunt and Detragiache (2005).
• Find evidence in favour of competition stability
  (C-S) view.

4
Comments

• This is an interesting, if somewhat surprising
  result.
• Most of the theory, eg Keeley (1990) argues that
  an increase in competition (as preceded by
  financial liberalisation) would tend to increase
  fragility.
• There is evidence that a number of financial
  crises (eg US 1980s, Nordic countries 1990s) were
  preceded by financial liberalisations and
  associated increases in competition.
• Authors may need to do more to fully convince the
  reader.

5
Comments

• Simultaneity most theory views both competition
  and stability as outcomes.
• Sf (government safety nets, strength of
  institutions, financial market development, etc)
• Cf (entry regulation, size of market,
  technology, financial market development, etc)
• Setup here does not fully recognise this.
• Relationship might not be causal,
• There may be an endogeneity bias.
• Difficult to identify policy measures, that could
  have a reliable effect.

6
Comments

• Is there an independent mechanism that links
  competition and stability (C-S)?
• Eg, does competition foster efficiency and hence
  stability?
• This could be be more fully described, or even
  analysed econometrcically.
• Efficiency measures are available across
  countries, eg from Demirguc-Kunt, Laeven and
  Levine (2003)
• Crisis regressions control for a number of
  variables,
• some of which might have an independent effect on
  stability,
• but some through competition, eg activity
  restrictions.

7
Comments

• More convincing identify the C-S mechanism using
  instrumental variable techniques
• instrument H-statistic by employing variables
  that have a direct effect on competition, but not
  on stability.
• eg entry restrictions, as available from Barth et
  al (2004), activity restrictions, etc
• In the second step, control more fully for
  variables that have a impact on stability,
  including strength of institutions, eg as
  measured by rule of law, Demirguc-Kunt and
  Detragiache (2002)

8
Comments

• Is it competition from banks or competition from
  financial markets, or both?
• Crisis regressions control for legal origin,
  since this is seen as important determinant of
  financial market development (amongst other
  things).
• But more direct measures of financial market
  development are available (eg stock market cap,
  credit to GDP, etc, from Beck et al 1999) and
  could be used in the first or second step.

9
Comments

• Robustness results hold for crisis definition
  presented by Demirguc-Kunt and Detragiache (2005)
• but not for Caprio and Klingebiel (2003).
• Some differences are Italy 1990-95 and US
  1980-92.
• Claessens and Laeven point out that their
  H-statistic might be biased downward for
  countries with many banks, such as Italy, US.
• Could exclude these countries as a robustness
  check when using D-K and D (2005).

10
Comments

• H-statistic assumes long-run equilibrium. But
  crises, consolidations and changes in regulation
  make this a strong assumption.
• Could also have a difficulty if competitive
  conduct in crisis times is different from conduct
  in stable times.
• Could run a test excluding the sample period over
  which H was measured 1994-2001.

11
Comments

• For further research reconciling the evidence.
• Could the short-run effect of financial
  liberalisations and competition be different from
  its long-run effects?
• This paper suggests that in the long-run,
  efficiency benefits create stability.
• But in the short run competition can
  destabilise.
• Need for time series cross-section data on
  competition to analyse this more fully.