ROSCAs and Credit Cooperatives September 2007 () ROSCAs September - - PowerPoint PPT Presentation

ROSCAs and Credit Cooperatives

September 2007

() ROSCAs September 2007 1 / 24

There is a wide spectrum of indigenous …nancial institutions , ! family and friends , ! ROSCAs , ! credit cooperatives , ! moneylenders Understanding ROSCAs and credit cooperatives helps to undertand costs and bene…ts of group lending , ! also emphasizes link to savings constraints (not just credit problems)

() ROSCAs September 2007 2 / 24

Rotating Savings and Credit Associations (ROSCAs)

Also known as tontines (Africa), hui (Taipei), tanda (Mexico) Very common throughout the world and very important , ! 40% of micro…nance borrowers in Indonesia , ! funds involved equal 10% of GDP in Ethiopia (1977) , ! 1/2 of rural residents in Cameroon, Cote d’Ivoire, Congo, Liberia, Togo and Nigeria , ! 1/5 of Taiwanese population (1977-95) Several alternative structures: , ! pre-determined order ROSCA , ! random ROSCA , ! bidding ROSCA

() ROSCAs September 2007 3 / 24

Characteristics

"Local" institution (neighboourhood, workplace) Varying membership: e.g. 5 to 100 in Bangladesh (Rutherford 1997) Varying pot size (e.g. $25 to $400 in Rutherford’s survey) Indivisible goods (e.g. school fees, rent, medical costs, equipment)

() ROSCAs September 2007 4 / 24

A Simple Model of a Random ROSCA

Number of individuals = n Preferences at each date U = v(c) without indivisible good v(c) + θ with indivisible good where v(c) =

• c

if c c ∞ if c < c y = monthly income B = cost of indivisible good T = planning horizon t = acquisition date (endogenous)

() ROSCAs September 2007 5 / 24

Problem faced by individual outside ROSCA

Constrained maximization problem: max

c,t

tc + (T t) (y + θ) subject to c

• c

t(y c)

• B

Constrained maximum is where c = c and t = B y c Utility of agent is UA =

• B

y c

• c + (T

B y c ) (y + θ)

() ROSCAs September 2007 6 / 24

c t c Savings Constraint

() ROSCAs September 2007 7 / 24

c t c Savings Constraint Increasing Utility

() ROSCAs September 2007 8 / 24

c t c Savings Constraint t* Constrained Maximum

() ROSCAs September 2007 9 / 24

Problem faced by ROSCA participant

Let n = number of periods that cycle lasts If agent ends up being the ith receiver of the pot, her utility is ui = ic + (n i)(c + θ) + (T n)(y + θ) = nc + θ(n i) + (T n)(y + θ) Her (ex ante) expected utility of joining the ROSCA is then UR = 1 n

n

∑

i=1

ui = nc + θ

• n n + 1

2

• + (T n)(y + θ)

() ROSCAs September 2007 10 / 24

Optimal design of ROSCA so that c = c and n = B y c and so UR =

• B

y c

• c + (T

B y c ) (y + θ) + θ

• n n + 1

2

• =

UA + θ

• n n + 1

2

• Example illustrates the "early pot motive":

, ! even though saving pattern is unchanged, ROSCA participation gives each member the chance of receiving pot early

() ROSCAs September 2007 11 / 24

Enforcement

What stops a member who has received the pot early from reneging? , ! Kenya: place "least trustworthy" at end of cycle ) requires ex ante screening , ! ban past absconders , ! social sanctions Lack of alternative ways of saving keeps ROSCAs intact , ! most common response when asked why join , ! key feature of ROSCAs: do not require a place to store money e.g. Anderson and Baland (2003)

() ROSCAs September 2007 12 / 24

Limits of ROSCAs

In‡exible pot size , ! adding members ) hard to manage Do not introduce new funds into system from outside , ! "bidding ROSCAs": pot goes to member that bids the most , ! but problematic "bidding wars" during downturns

() ROSCAs September 2007 13 / 24

Credit Cooperatives

Modi…cation of ROSCA that allows some participants to mainly save and others to mainly borrow Old idea going back to 1850s rural Germany (Friedrich Rai¤eissen) , ! by 1910, there were 15,000 institution serving 2.5 million people (9%

• f German banking market)

Spread to Madras and Bengal (India) in the 1890s. By 1946 membership exceeded 9 million

() ROSCAs September 2007 14 / 24

Key features

Credit cooperatives di¤er from ROSCAs in several ways: , ! members do not have to wait their turn to borrow , ! participants (savers and borrowers) are all shareholders , ! key decisions (interest rates, loan size) determined democratically In the Rai¤eisen model (Prinz, 2000): , ! members were from same local parish , ! unlimited liability: defaulters lose all current assets , ! low income borrowers could not be discriminated against , ! cooperative performed other functions (e.g. purchasing of inputs) , ! extended short and long term loans

() ROSCAs September 2007 15 / 24

Credit Cooperatives as a Vehicle for Saving

Saver-borrowers each with initial wealth w Two ways of saving: (1) inside the cooperative yields gross interest θ (2) another commercial bank in the city yields θ δ Each member has access to a project with unit cost and

• utput =

y with probability e with probability 1 e where ε < e < 1 and cost of e¤ort = Ce , ! assume that θε < θ δ In case of failure borrowers loses wealth invested in cooperative, plus a non-monetary sanction Gross interest rate on loans = r Timing: (1) borrowers decide how much to invest, and (2) given investment, how much e¤ort to provide

() ROSCAs September 2007 16 / 24

Optimal choices

Given wi borrower chooses e¤ort e to maximize payo¤ UB = e (y + θwi r) + (1 e)(H) Ce = e (y + θwi r + H C) H , ! it follows that e(wi) = 1 if y + θwi r + H C ε if y + θwi r + H < C , ! probability of default is reduced if wi and/or H are high enough , ! critical wealth level: wc

i = C + r y H

θ

() ROSCAs September 2007 17 / 24

Savings in cooperative, wi, chosen to maximize e(wi) (y + θwi r) (1 e(wi))H Ce(wi) + (θ δ) (w wi)

• r

e(wi) (y + θwi r + H C) H + (θ δ) (w wi) , ! it follows that the optimal level of saving in the cooperative is w

i =

w if w wc if w < wc since θε < θ δ

() ROSCAs September 2007 18 / 24

Implications

Investing in the cooperative acts as a "commitment device" for the borrower , ! induces the borrower to minimize default probability and take advantage of higher (risk-adjusted) return on savings high social sanctions, H, and su¢ciently high relative return δ allow cooperative to mobilize savings higher return may also increase overall savings, w

() ROSCAs September 2007 19 / 24

Credit Cooperatives as a form of Peer Monitoring

Based on Banerjee, Besley and Guinnane (1994)

Individual has an investment opportunity with cost F and

• utput =

y with probability e with probability 1 e Competitive outside lender ) gross lending rate, R, must satisfy eR = r where r is opportunity cost of funds Bene…t from "shirking" depends on monitoring m : Bene…t from shirking = B(e, m) = 1 m

• a 1

2e2

• ,

! assume outsider lender can only monitor at minimal level m

() ROSCAs September 2007 20 / 24

Borrower chooses e to maximize UB = e(y RF) + 1 m

• a 1

2e2

• ,

! yields (constrained) optimal e¤ort e = m(y RF)

() ROSCAs September 2007 21 / 24

Simple (2 member) Credit Cooperative

Now suppose borrower forms a cooperative with another individual This "insider" plays three roles: , ! partial lender, provides F b , ! guarantor: promises w rb to outside lender in case of default ) lender faces no risk and can o¤er rate r , ! monitor: can optimally adjust monitoring, m, subject to cost C(m) Why would the insider do this? , ! must receive a big enough share of pro…ts, α

() ROSCAs September 2007 22 / 24

Optimal choice of e¤ort by borrower is now e(m) = m(1 α)(y rb) (IC) If cost of monitoring is C(m) = c 2m2 , ! insiders payo¤ is UI = e(m)α(y rb) (1 e(m))w C(m) r(F b) = m(1 α)(y rb) [α(y rb) + w] c 2m2 w r(F b) ) optimal monitoring level m(α, w) = 1 c (1 α)(y rb) [α(y rb) + w]

() ROSCAs September 2007 23 / 24

Implications

In a credit cooperative, each non-borrowing member acts as a lender, guarantor and monitor , ! as guarantor, she reduces interest rate paid by borrower ) increased incentive to provide e¤ort , ! as monitor, the fact that insider has stake in project ) increased incentive to monitor ) induces more e¤ort Because of improved incentives, a well-designed credit cooperative can increase overall output , ! must choose the share, α, appropriately

() ROSCAs September 2007 24 / 24