11/4/2009

What happens in a world where people can't borrow (or there is some 'credit constraint')?

Banerjee + Newman (1993)
  - Poverty begets poverty---a poverty trap
  - Compare two nations that are identical, apart from the extent of their
    credit constraints.  Model says they will diverge in economic growth.
  - Two nations, both identical (incl.) credit constraints, except for initial
    levels of inequality will diverge in economic growth
  - Inequality in countries can get worse over time
  
Intuition
  - Imagine two production technologies:
    - One costs $200, deprecates fully over each period, plus one worker makes $1000
      of output.  This technology has a fixed cost.
    - The other is a subsistence model---only one worker, again.

Imagine two people in this society and have total of $400 net worth at the beginning

Consider four cases, varying initial wealth and level of credit constraints
  - Equal, no constraints: each starts w/ $200, can both buy machines, and make
    a profit of $800 each
  - Equal w/ constraints: same thing---neither borrowed anyway
  - Unequal, no constraints: one has $100, the other has $300, but can borrow/lend
    to each other.  Again, total profit of $1600.  Vary how much rich guy/poor
    guy 
  - Unequal, constraints: $300 guy buys machine, but cna't lend more than $99.
    The other guy has to work and gets $10 at end of period.  It will take
    10 periods until the poor guy can buy a machine, and until then, the
    gap between both of them widens.  In this example, there's no long-run steady
    state of a trap.  In the real world, there's issues w/ saving money, depreciation,
    etc.

Takeaways
  - persistence of underdevelopment---small frictions can have big consequences.
  - colonialism or environment/agriculture might have left some regions w/ 
    inequality poorer if access to credit is imperfect.

Are credit constraints even plausible?
  - banks might lack information on the poor
  - they could just charge higher interest. why would someone take on a high-interest loan?
    - they are sitting on a fast, high-return opportunity
    - they have no plans to repay the loan
    -> the information assymetry is high here
  - poor don't have much collateral to hold against their loans.  this makes it
    hard to enforce the collection

Evidence of credit constraints
  - Do firms want to borrow more than they are able to?  Instead of asking,
    which is error-prone, the following study studied return on capital:
  - McKenzie + Woodruff in 2008: give cash or equipment to random microenterprises
    in Sri Lanka
  - Equipment selected by enterprise owner, purchased by researchers
  - Estimate effect of capital input on profit to estimate return on capital.
  - If market interest rate is lower than internal return on capital,
    then there's a constraint.
  - Turns out that rate of return is higher (50-60%) than interest rates (12-18%).
  - This means that firms would want to borrow, but can't for some reason.  Otherwise
    banks would reduce their limitations and increase interest rates.

Why does group lending succeed
  - bank can hold collateral against a group: social capital/reputation---the group
    will chastize you for not repaying it.
  - risk goes down in group loan: other people in group will repay it so that they
    don't get dinged.
  - lending larger ammounts at a time to a group reduces the fraction of overhead.

Group lending not always beneficial
  - think of it game theoretically---once person A defaults, person B can
    pick up the slack or give up as well

There is correlation between microcredit presence and improved economic/social
  outcomes.  But this might not be causality.
  - Villages might be predisposed to entrepreneurialism and have let microcredit in
  - Microcredit might have set up in a successful area
  - Easier to set up any firm, including a microcredit organization, in areas
    that are easier to have trade
  
Why it's hard to set up randomized evaluation?
  - hard to find a place w/out it
  - hard to study place that doesn't have it that microlending organization
    won't extend to.

So Banerjee + Duflo did a randomized study in Hydrabad.  Randomized startup
  of an MFI called Spandana.  Group loans w/o requiring empowerment training
  or requiring money go into business development, like Grameen does.

52 groups get MFI built, and 52 don't, randomly selected from 104 attractive regions.
  - there may be spillover into the comparison non-treatment group, so
    the observed effect size may be smaller than the true one.

Did Spandana entry actually increase total MFI borrowing in these areas?
  - yes.  people in 52 treatment slums got 8.3% more loans from some MFI.

Within 15-18 months of loan coming in
  - There was an increase in new businesses
  - There was an increase in profit in already-existant businesses
  - No measurable increase in per-capita expenditure or nondurable goods, but
    increase in durable goods (for starting businesses) was measurable.
  - In 15-18 months, there is no measurable effect of women's empowerment,
    child health, or education.  But this is probably far too short a timespan
    to see a noticable difference.