Comments on Credit Constraints as a Barrier to the Entry and Post-Entry Growth of Firms

1
Comments on Credit Constraints as a Barrier to
the Entry and Post-Entry Growth of Firms

• Simon Johnson
• MIT Sloan

2
Summary Assessment

• Growth is largely about entry
• Confirm earlier results
• Availability of external finance constrains entry
• Nature of constraint and effect varies
  systematically across firms
• Most novel here Size of firms may matter in this
  regard

3
A (Potential) Debate

• The Finance View
• Access to credit has a first-order effect on
  ability to become an entrepreneur (and therefore
  growth)
• The Property Rights View
• Only potential entrepreneurs who perceive their
  property rights to be secure will make
  investments

4
An Econometric Problem

• If you do not control for property rights
  (instrumented) then the finance variables may
  actually be picking up broader institutional
  effects
• Not a problem if you can instrument convincingly
  for finance variables
• But this is hard
• Is there plausible evidence that property rights
  matter for entry and post-entry growth of firms?
• See work by John McMillan

5
The main finding

• Effect of credit constraint depends on size of
  firms
• Relaxing this constraint helps small firms more
  than large firms
• it may even discourage entry by large firms
  (p. 2)

6
Question 1

• What exactly is entry by large firms?
• Is this entry, as in start-up?
• Or entry into the database?
• How does the database handle
• Entry by foreign companies
• Mergers and acquisitions
• This entire measurement issue needs to be at
  least 1/3 of the paper
• By the way, how big were Microsoft or Intel or
  Google or Whole Foods when they first started up?!

7
Question 2

• Rajan and Zingales (1998) breakthrough paper for
  methodology of studying financial development
• Why 1,000 papers using this method, and almost
  no work checking if nonUS financial dependence of
  sectors is really similar/correlated with the US?
• Exception Subramanian et al (India)
• 1980s vs. 1990s why such differences?
• RZ AJ robustly, financial development matters
  for composition of output (types of sectors) but
  NOT for GDP per capita
• AJM may matter for vertical integration, when
  contracting costs are high (CAUTION new result)

8
Question 3

• Djankov et al Doing Business Indicators have
  been a breakthrough for quantifying and focusing
  on barriers to entry
• But just putting these in a regression (not
  instrumented) may not tell us very much
• Likely correlated with broader institutions
• Probably there is some form of seesaw effect
• If you push down on one end (lower some cost for
  new entrants to become large firms) then the
  other end will pop up (some other costs/barriers
  emerge)

9
Question 4

• If you have a potentially endogenous rhs variable
  or other issues is it still just OK to find a
  possibly more exogenous variable as an
  instrument?
• Dont we need more theory and evidence (also from
  outside the regression framework being used) to
  support the exclusion restriction?
• There is a great deal of confusion over the
  other channels issue, but this is a potentially
  important point
• What are the deep determinants (of growth) and
  how are the effects of these manifest?

10
The Limits to Knowledge (Today)

• What are the robust facts?
• Financial development matters
• For the composition of output agreed
• For GDP per capita contested
• Legal origin matters
• for financial development probably
• For GDP per capita no
• What are the implications for Entry?
• Composition vs. level of output?!
• Property rights security of investment cannot
  safely be omitted from the regression

11
Policy Implications

• We need rules, not discretion, for
• Updating our stock of useful knowledge based on
  new research
• Apply the same rules across different kinds of
  research (but you cant randomize everything!)
• Sometimes the new findings are sufficiently
  convincing and important that we should change
  our views
• Resist fashion, if you can (hence the need for
  rules)
• Emphasize robustness, replication, and data
  revisions
• First, do no harm
• Agreed, but what if you are in a crisis?