SelfDiscovery, Lending and Microenterpreneurship

1
Self-Discovery, Lending and Microenterpreneurship

• Mikhail Drugov and Rocco Macchiavello
• University of Oxford, Nuffield College
• Nuffield Economic Theory Workshop
• 29th April 2008

2
Motivation

• Each of us has much more hidden inside us than
  we have had a chance to explore. Unless we create
  an environment that enables us to discover the
  limits of our potential, we will never know what
  we have inside of us.
• Muhammad Yunus, the founder of Grameens Bank
• Can this discovery explain some puzzles or
  puzzles of microfinance?

3
Motivation cont.

• There are still substantial gaps in our
  understanding of the poor Banerjee and Duflo
  (2007)
• We focus on three aspects
• Use of moneylenders who charge (very) high
  interest rates even when MFI is available
• Contemporaneous borrowing and saving (at
  unfavourable interest rates)
• Rejection of bigger loans (and failure to scale
  up)

4
Motivation cont.

• Use of moneylenders
• Madestam (2007) moneylenders have an advantage
  in monitoring
• Karlan and Mullainathan (2006) moneylenders are
  more flexible
• Contemporaneous borrowing and saving
• contemporaneous borrowing and saving (evidence
  from Fincas clients, mostly women operating
  small informal businesses) This behaviour is more
  likely in poorer towns and where moneylenders
  charge higher interest rates (Basu (2008) ,
  Karlan)
• substantial evidence of time inconsistent
  preferences in borrowing and saving decisions
  (Ashraf et al. (2003, 2006), Anderson and Baland
  (2002))

5
Motivation cont.

• Rejection of bigger loans (and failure to scale
  up)
• often the poor do not respond to immediate
  availability of profitable investments
  opportunities (e.g. Duflo, Kremer and Johnson
  (2007))
• not uncommon that microfinance clients turn down
  offers of bigger loans, even when they still have
  
• high interest rate debts to repay, and/or
• when slightly smaller loans showed very high
  return rates (see e.g. Banerjee and Duflo (2007)
  
• failure of micro-entrepreneurs to scale-up
  (Fisher (2008), Stiglitz et al. (2007))

6
Preview of Results

• We build a (simple) non-behavioral model that
  explains the three puzzles simultaneously. It
  also shows that all of them are likely to happen
  to the same borrowers those with low confidence
• Moneylenders can enforce contracts with
  successful entrepreneurs and thus finance the
  experimentation initially
• Contemporaneous borrowing and savings may be
  used to ease the separation of different types of
  borrowers
• Rejection of bigger loans bigger loans seem to
  be better but they are more costly to experiment
  with, and these costs have to be financed in the
  future

7
Literature

• Banerjee and Mullainathan (2007)
• Basu (2008)
• Baland, Guirkinger, and Mali (2007)
• Giné and Klonner (2007)
• Bénabou and Tirole (2002, 2004)
• Bergemann and Hege (1998, 2005)

8
Model Set Up

• Two periods, discount factor ?
• In each period, an agent may undertake a project
  that
• needs investment of 1
• succeeds and yields r if
• the agent does not divert the investment earning
  ?1
• the agent exert the effort that costs her e
• otherwise, it fails and yields 0
• There are two types of agent Good with eG0 and
  Bad with eBe Probability(Good)? initially the
  agent is uninformed (and lenders as well)
• If the agent does not undertake the project, she
  takes her outside option u
• The agent has no wealth and protected by limited
  liability

9
Parametric Assumptions
10
Timing of Events and Contractual Structure
Period 1
Period 2
Invest and exert effort ? r
Invest and exert effort ? r
Further funding
Borrow and start a project
Divert ? 0
Outside option u
Divert ? 0
t
Conditional on 1st and 2nd period outputs and
messages 2nd period transfers
Agent learns type and sends message
Conditional on output and messages 1st period
transfer and 2nd period financing
Invest and exert effort ? r
Borrow and start a project
Competitive one-period lending with prior ?
Outside option u
Divert ? 0
Outside option u
11
Benchmark Self-Financed Agent

• When the agent invests her own money, no moral
  hazard arises
• Investing, she gets
• Taking outside option, she gets

Experiment (invest)
Take outside option
12
The First Best

• In period 1 invest, both types repay
• In period 2 the bad type takes outside option,
  the good type invests and repays
• What (cost-minimizing) contracts implement it?

13
Optimal Allocation

• When d is small, FB is very costly as it demands
  a high rent of ?egtr-1 for both types and period
  2 production is not important
• Then, implement the second best
• In period 1 invest, only the good type repays
• In period 2 the bad type takes outside option,
  the good type invests and repays
• Proposition 1. The optimal allocation is
• The first best whenever possible
• Otherwise, the second best whenever possible
• Otherwise, outside option
• See next figure

14
Equilibrium Characterization
Credit Constrained Agents
1st Best - Competitive Lenders
No Project Is Optimal
2nd Best -Competitive Lenders
15
Comparative Statics, Increase in u
16
I. Demand for Moneylenders

• Unless d is very low, the good type pays a
  positive interest rate in period 2
• Other lenders know that the agent which takes a
  loan in period 2 is of the good type and are
  willing to offer her a loan at zero interest rate
• The commitment of the agent to stay with her
  lender is crucial
• If she cannot commit, lenders may refuse to lend
  in period 1 (when the prior is not very good)
• The agent may have no choice but to apply to a
  moneylender who, through private/informal
  enforcement, can enforce the contract in period 2

17
Equilibrium Characterization No Agents
Commitment
1st Best - Moneylender
Credit Constrained Agents
1st Best - Competitive Lenders
2nd Best Money- Lender
No Project Is Optimal
2nd Best -Competitive Lenders
18
II. Contemporaneous Borrowing and Saving

• Because of the agents linear utility, the
  distribution of consumption across periods is
  indeterminate
• To pin it down, introduce u(c)c? and take ??1
• Focus on the case dd? (FB, TGd? and TBd(?-u))
• max ?u(c1,G)du(c2,G)(1-?)u(c1,B)-edu(c2,Bu)
  
• s.t. c1,Gdc2,G d? BCG
• c1,Bdc2,B d(?-u) BCB
• c2,G ? ICG
• Solution c1,Bc2,Bud?/(1d) smoothing
• c1,G0 and c2,G? no smoothing

19
II. Contemporaneous Borrowing and Saving cont.

• Which indirect mechanism implements this
  allocation?
• in period 1
• the agent borrows 1, obtains r-1 and repays D1
  (cannot default)
• consumes c1,i and saves
• in period 2
• either repays D2, borrows 1, repays B2, and
  consumes
• or defaults on D2, consumes outside option u and
  savings
• (and is denied access to credit)
• It is possible to show that when B2?1,r-?

20
III. Rejection of Bigger Loans

• Call the project studied so far street vendor V
• Introduce shop keeper S
• needs kgt1 of investment
• needs ke of effort
• the agent may divert k?
• yields ar, 1ltaltk, if completed
• more profitable than street vendor ar-kgtr-1
• Make the same assumptions
• ar-kgtk? the project yields enough to solve MH of
  the good type
• ar-kltk(?e) the project does not yield enough to
  solve MH of the bad type
• argtk(?e) once initial investment is sunk, it is
  optimal to complete the project for any
  type

21
III. Rejection of Bigger Loans cont.

• In period 2, the type of the agent is known ?
  optimal to finance a shop
• In period 1, start with either vendor or shop
• Proposition 3. It is possible that the following
  conditions are satisfied simultaneously
• Both VS and SS are feasible
• VS is preferred to SS
• E(c1VS) lt E(c1SS)
• E(u1VS) lt E(u1SS)
• ibVS gt ibSS

22
III. Rejection of Bigger Loans cont.
Vendor - Shop
Shop - Shop
23
III. Rejection of Bigger Loans cont.

• Intuition
• when ? is low, experimenting with a shop is very
  costly ? VS is preferred
• in SS, the binding constraint is MH of the bad
  type in period 1 ? the good type smoothes her
  consumption
• in VS, the binding constraint is MH of the good
  type in period 1
• ? the good type cannot smooth her consumption
  (c1,G0 and c2,G?)
• Expected consumption (and utility) in period 1 is
  higher in SS
• For the interest rate it is similar to
  contemporaneous borrowing and saving

24
Empirical Implications

• ? is usually unobservable
• Proposition 4. The model implies the following
  testable predictions in a cross-section of small
  entrepreneurs
• 1. Higher interest rates on loans are
• a) negatively correlated with future scaling-up,
• b) positively correlated with default rates and
  drop-out rates,
• c) positively correlated with contemporaneous
  borrowing and saving at unfavorable terms.
• 2. Micro-entrepreneurs financed by moneylenders
• a) pay higher interest rates on loans,
• b) are less likely to scale-up, and more likely
  to drop-out and default,
• c) are more likely to hold savings paying
  interest rates lower then the one paid on loans.

25
Conclusion

• A non-behavioural model that explains some
  puzzling features of microlending
• Use of moneylenders who charge (very) high
  interest rates even when MFI is available
• Contemporaneous borrowing and saving (at
  unfavourable interest rates)
• Rejection of bigger loans (and failure to scale
  up)
• Uncertainty and experimentation are crucial
• A lot of future work
• Effect of market structure on the financing of
  the exploration
• Group exploration
• Self- as opposed to market discovery