Credit and microfinance
Discussion about features of credit
markets in LDCs
Formal and informal sources of credit
Get around the informational constraints
What is credit?
agreement for repayment at a
specified date or dates
repayment includes an interest rate
purpose of loan is usually specified to
Sources of demand for credit
1 Fixed Capital
2 Working Capital
3 Consumption Credit
1 Need for cash because of shock (need for
2 Consumption smoothing to cope with
3 Exceptional expenditures (wedding, funeral…)
1. involuntary default
-issue of monitoring, fungibility
2. voluntary or strategic default
enforcement mechanisms (formal,
Where does credit come from (1)?
1. Institutional or Formal Lenders
rural development banks
advantage of moving capital across
enforcement and informational issues
(relative to informal credit)
transactions costs higher
An example: r=10%, 2 projects
Initial Return Final Bank Investor
1 100,000 15% 115,000 10,000 5,000
2 100,000 20% 120,000 10,000 10,000
Same project 2, but project 1 risky
100,000 - 230,000 with prob 0.5
- 0 with prob 0.5
(1) 0.5 (230,000-110,000)+0.5 x 0 = 60,000
(1) 0.5 x 10,000 +0.5 x 0 = 5,000
Limited liability, discrimination against the
Rational for use of collateral
Where does credit come from (2)?
can take different forms of collateral (eg labor)
often have better information than formal lenders
India (1951) 7.2 percent of all borrowing from
governments, banks and cooperatives
India (1981) 61.2 percent formal, 24.3 percent from
moneylenders (Bell, 1993)
Thailand (1975) 90 percent of total rural credit informal
Thailand (1993) 50 percent informal
Nigeria (1989) 7.5 percent in villages from formal
Characteristics of rural credit
High interest rates
Chambar region of Pakistan: 18-200%, average is 79%
(formal sector = 12%)
Little use of collateral
Simultaneous existence of formal and informal
Chambar region in Pakistan –Aleem (1993):
10 out of 14 moneylenders lend 75% of their funds to old
Validity dependant on the stationarity of the
Moral Hazard (use of the loan, reimbursement
Need for auditing in case of non repayment.
Understanding features of
informal credit market
Benchmark: competitive equilibrium with perfect
Competitive equilibrium with imperfect
Interactions between external and internal
Assume no voluntary default
Imperfect enforcement and long-term
The question of voluntary default
2 types of borrowers: 1 and 2
Returns of the projects in case of success:
Success probability: (1)> (2)
Same expected returns, one safe borrower (1)
and a riskier one (2)
Reservation utility: W(1)=W(2)=W
(t) R(t)-i i
1- (t) 0 0
For the borrower: U(i, t)=(t)[R(t)-i]
For the lender: (i, t)= (t).i
No voluntary default
Competitive equilibrium with perfect
ρ is the opportunity cost of funds (return on the
risk-free capital market)(R>ρ≥1)
In equilibrium, i(t) is such that:
There is no i(t)≠ i1(t) s.t. Π(i1(t), t)≥ρ and that borrower of
type t would prefer to i1(t)
For each type t, there is an equilibrium with
lending characterized by the solution to:
Max (t )(R(t ) i )
s.t. i. (t ) and (t )(R(t ) i ) W
If R W , there will be positive lending
t , i1 (t )
and t , U (i1 (t ), t ) R
i1 (1) i1 (2)
Each agent can be identified
Different interest rates
Lower interest rate for the safe type
Taking into account default probability can yield to
high interest rates
Competitive equilibrium with imperfect
information = adverse selection
The lender cannot distinguish the two types of risks but knows
the proportion p1 of type 1 risks in the population. The interest
rate is therefore unique and note that i:
U (i,1) (1)R(1) i (2)R(2) i U (i,2)
(i,1) (1).i (2).i (i,2)
The participation constraint is
(t )R(t ) i) W
U (i, t )
Obviously we have 0
Hence we can define the highest interest rate at
which borrowers are willing to borrow. Call it i*(1)
for type 1 borrower. It is defined by equation:
R (1).i * (1) W
hence i * (1)
Define i*(2) in the same way.
i*(1)<i*(2), so that safer borrowers drop out of the
pools of borrowers first when the interest rate
If i i * (1), then E (i ) p(1) (1)i [1 p(1)] (2)i.
If i * (1) i i * (2), type1 borrowersdrop out and lender income falls.
E (i ) (2).i
If i i * (2), E (i ) 0
Expected lender’s income as a
function of i
The competitive equilibrium with adverse selection
is then defined as an interest rate i2 such that
E (i2 )
and there doesn't exist any i i2 s.t. E (i ) and
U(i,t) U (i2 , t ) t and U (i, t ) W t
If R-ρ>W, there will be lending in equilibrium.
If E(i * (1)) p1 (1)i * (1) (1 p1 ) (2)i * (1)
Then, equilibrium interest rate will be i2 / (2)
And only type 2 borrowers will want a credit.
If E(i * (1))
Then the interest rate will be given by
i2 /p(1) (1) [1 p(1)] (2)
Which is less than i*(1), and all potential borrowers
will demand loans.
Interactions between outside and local
Local lenders: well informed, high h.
If R- h W, then
i3(1) = h /(1) and i3(2) = h /(2)
i3(1) i*(1) et i3(2) i*(2)
Outside lenders are uninformed. Then cannot distinguish
between type 1 and type 2, but low l. Offer a unique i.
If i i*(1) : type 1 borrowers drop out of the outside market
as soon as i > i3(1) .
If no local lenders:
ĩ = l / (p1 (1) + (1- p1) (2) ) but ĩ > i3(1)
With incomplete information
Credit depends on the characteristics of
other potential borrowers in her locality
Adverse selection can lead to credit
rationing even in the absence of govt
interest rate controls
Lenders with access to cheap funds may
not be able to penetrate local markets
Voluntary default and enforcement
Lack of formal legal enforcement
Why do borrowers repay?
Construction of repayment incentives in
the absence of legal sanctions
Low income before the harvest (Yg); high
income after the harvest (Yh).
UA U (Yg) + U (Yh)
With borrowing allowing consumption smoothing
ULR U (Yg+L) + U (Yh-(1+r)L)
L>0 st ULR > UA since U concave.
If voluntary default
ULD U (Yg+L) + U (Yh)
A = 0
LR = rL
LD = -L
For the borrower ULD > ULR , No lending will happen.
Lender/borrower Repay :Yes Repay : No
Lend : Yes ULR , LR ULD , LD
Lend :No UA , A UA , A
Repeated game: constant probability to play during
the next period
For the borrower:
If default : ULD at this period, UA during the following ones.
If repay, ULR at each period
Repayment will occur if
U (Yh) - U (Yh- (1-r)L) < t=1, t (ULR - UA )
ULD – ULR < (/(1-))(ULR - UA )
Role of the size of the loan
Upper limit on the size of self-enforcing loan contracts.
Improving the probability of reimbursement in a
start with a small loan and work your way to bigger loans
interest rate below the next best alternative source of credit
credible threat to cut people off
If other sanctions are possible (social or economic), with a utility
ULD U (Yg+L) + U (Yh) -D
(similar role for collateral)
It allows to reach a greater L with a smaller
If new sources of credit appear such that UA < UN < ULR (profit
attracts other lenders; it generates more default)
The previous reimbursement constraint might not hold anymore.
The exclusivity in the relationship is crucial for the threat of
exclusion from future loans to be credible.
Limits on the size of the loans.
If imperfect information st voluntary and unvoluntary
default cannot be distinguished, L if smaller and the
probability of market collapse if higher.
Market is segmented.
Better information (notably within the family)
Better enforcement capacity because of the wider availability of
Depends on the environment (availability ofcollateral,
Very segmented activities
Credit + insurance (Udry, Nigeria): 97% of the transactions
occur within the village or within the family.
Interlinked credit : obtained from the landowner or the trade
Some formal institutions attempt to cumulate the
advantages of the formal system (spatial
diversification, lower refinancing cost..) and from
the informal one (information, sanctions):
Group lending: Members of the group are jointly
responsible for reimbursing loans and collectively
sanctioned in case of default (exclusion from future
Ex: Grameen Bank
Very low default rate … two years after the due date
Reality of sanctions?
Disputed impact on poverty.
Group lending with joint
Recap: problems faced by a potential lender are:
Adverse selection (what is the risk of a potential borrower?)
Moral hazard (will the borrower se the loan so as to be able to
Auditing costs (assessing the reality of a project failure)
Enforcement (how to enforce repayment in case of voluntary
Using local information and social capital among
borrowers, group lending with joint liability may be able
to deal with those problems.
Output Y takes two values: YH > YL =0
YH is obtained with probability p
Each project requires 1 unit of capital and repayment is
ρ>1 (principal + interests)
u is the opportunity cost of labour
Project are socially profitable, i.e.
pY H u
In the case of individual liability, interest rate (gross) is r.
In the joint liability case, if one member of the group
default, the other has to pay an additional cost c.
Dealing with adverse selection
Two types of borrower: safe (a), risky (b).
p varies with the type so that pa> pb
The expected payoff for a borrower of type i paired with a borrower of
type j in a joint liability contract is:
EUij (r, c) pi p j (Yi H r ) pi (1 p j )(Yi H r c)
Hence, the net expected gain for the risky borrower to have a safe
EU ba (r , c) EU bb (r , c) pb ( pa pb )c
And the expected loss for a safe borrower of having a risky partner:
EU aa (r , c) EU ab (r , c) pa ( pa pb )c
As a result, risky borrower are never ready to pay
enough to compensate safe borrowers to be paired with
→ under joint liability, group formation should display
positive assortative matching.
This allows the bank to screen the borrowers by offering
two different contracts: low interest rate and high joint
liability (chosen by safe borrowers groups) and high
interest rate and low joint liability (chosen by risky
Dealing with Moral Hazard
Peer monitoring works because group members have an incentive
to take remedial action against a partner who mis-use her loan.
Borrowers can choose a level of effort that affect probability of
success at a cost. This level of effort is not observable by the bank,
but is observable by other group members.
Interest rate is like a tax on success, since you repay only when
output is high: hence, the higher r, the lower the effort.
Under joint liability, if borrowers do not cooperate, the same level of
effort as under individual liability is chosen. Nevertheless, if they
cooperate, they can choose higher p and benefit from lower r.
This is true even if monitoring the others actions is costly.
If group members face a lower cost of verifying
each other’s output than the bank, then the bank
can avoid the cost of performing its own audit every
time a borrower claims she has low output by
inducing her partner to undertake liability for her.
The partner has an incentive to audit the borrower
because he is partly liable for her repayment.
Only when the whole group announces its inability
to repay will the bank have to audit.
Issue arises from the lender’s limited ability to apply
sanctions in case of voluntary default.
But the group and the community can apply social
sanctions (as already discussed).
This compensates the detrimental effect of joint
liability on enforcement due to the fact that
moderately successful borrower might prefer to
default on her own loan if the partner is defaulting
rather than incurring all the cost.
Difficulties in implementation:
Optimal group size?
Big groups allow to diversify risk within the
But big groups dilute the monitoring capacity.
The cost of group lending
meeting time cost
one person’s risk is all’s risk
if one borrower really needs to default, then the
others may also prefer to default
possible excessive pressure to undertake safe
Joint liability in real life:
About two millions borrowers.
lends to individuals in self-selected group of 5 from the same
village. If any individual within a group defaults, the group is
responsible. No collateral is required. The groups meet weekly to
check in with one another.
Starts with a small saving. Then two members receive a loan, to be
repaid weekly, for one year. If the repay, two more members get a
average loan size of around $100
Repayment rates may average around 97-98 percent
94 percent of borrowers are women
technical assistance includes vocational training, productive inputs.
Impose a code of conduct that dictates “life improving” changes
interest rates average around 20 percent (real interest of 12
3 sets of criteria:
the ability to earn revenues that exceed
the ability to survive without subsidized
the ability to effect meaningful changes in
the living standards of clients
Grameen Bank Balance Sheet Data from Morduch
millions of US $
Variable 1987 1989 1991 1992 1993 1994
Avg. annual loans outstanding 18.15 42.03 62.99 91.31 169.77 249.04
Annual disbursement 35.77 64.67 100.03 164.59 321.96 380.98
Borrowers (millions) 0.329 0.648 1.042 1.385 1.683 1.861
Default Rate 3.7 3.1 7 3.7 3.7 3.7
Income from lending 2.45 5.07 9.09 13.51 27.31 41.13
Investment income 1.98 2.85 4.62 3.85 2.99 1.44
Income grants 0.11 1.86 1.98 1.61 1.94 2.17
Total income 4.84 9.35 14.92 19.99 34.29 50.43
Total expenses 4.83 9.28 14.92 20.13 34.04 49.88
Reported profit 0.02 0.07 0.32 -0.15 0.25 0.54
Revised profit -0.1 -1.79 -1.66 -1.76 -1.7 -1.63
Total subsidies 4.65 9.86 15.39 17.6 23.51 27
Subsidy per $ outstanding % 26 23 24 19 14 11
The Grameen controversy
Wall Street Journal Critique:
repayment rates of 95% not true
2 northern districts 50+ late by more than 1 year
19% of total portfolio over 1 year overdue
using 2 year cutoff, 10% overdue
competition more default
implementation mistakes, e.g. “group funds”
responses: Grameen Managing Director
fixing things, 95% repayment rate in 6 months
15% having difficulties due to “external factors”
“We built our system on the faith that the poor
always pay back. Some times they take longer
than the originally scheduled time period,
sometimes natural disasters like flood, drought
or political unrest, rules and procedures of the
bank make it difficult or impossible to pay back,
but given the opportunity, they pay back. Non-
repayment is not a problem created by the
borrowers, it is created by the factors external to
“Current experience, supplemented with a host of
short-term studies, attests to great possible
achievements. The successful programs are
clearly successes. Visits to areas served by
programs show what cannot be seen in books of
accounts – earnings from microfinance participation
are funding new houses, further education for
children, new savings accounts, new businesses.
In the process, lives are being changed and poor
households are lifting themselves up.”
+ Yunus’ Nobel price.
participatory rural assessment
program evaluation regressions:
welfare= controls + receipt of credit
participants differ in systematic, unobservable
can distort measured impact by up to
Fixing selection bias
two stage correction (Heckman)
estimate determinants of participation
use these estimates to correct for
selection bias in welfare outcomes
importance of first stage variables
(influence participation but no
independent effect on outcomes from
need control with no spillovers
Evaluation difficulties: Grameen
Pitt & Khandker (1998) estimate that household
consumption increases by 18 taka for every 100 take
lent to a woman.
Strong identification hypotheses: relies on the fact that
observable characteristics of individuals affect outcome
differently for treated (eligible with access) and non
treated groups, and assign all differences to the
treatment. No non-linearities in the impact of Xs on
Morduch (1998), with the same data, in a simple DD
framework: hh with access have a lower consumption
(by 7%) than those without.
Both papers find impact on consumption smoothing.
Banerjee et al.(2009) find that 15 months after lending,
no effect on avg monthly expenditure but increase in
expenditures in durable goods and in number of new
businesses; heterogenous effect