Rotating Savings and Credit Associations (Roscas) are found throughout the world, particularly in developing countries, and are either random Roscas, where in each period members put a fixed sum of money into a `pot', which is randomly allocated to one member, who is then excluded from the draw for subsequent periods, or bidding Roscas, in which the pot is allocated in the present period to whichever member pledges the highest future contributions one-time side-payments to other members.

In Discussion Paper No. 443, Research Fellow Tim Besley, Stephen Coate and Glenn Loury use a simple general equilibrium model to analyse the economic performance of Roscas. They find them to be inefficient means of buffering against risk, since even bidding Roscas permit individuals to obtain the pot only once, and risks in LDCs are commonly faced by many individuals at the same instant. Roscas may play a role as means of saving to buy an indivisible good or to finance a major event, but they are less appropriate than for saving for old age, which is more often effected through intra-familial transfers.

In a closed economy without access to external funds, savings lie idle that could be employed by some individuals to enjoy the services of the indivisible durable good. Roscas make these joint savings work, but the resulting allocations will differ from those that would emerge from a fully functioning credit market. Besley, Coate and Loury use a simple two-good model with indivisibilities to characterize the consumption path and the accumulation of indivisible goods under these alternative institutional arrangements. They also compare the welfare properties both Roscas and competitive credit markets under criteria of ex post Pareto efficiency and ex ante expected utility. Neither type of Rosca is ex post efficient and randomization is preferable to bidding for homogenous agents. Moreover, a perfect credit market will always dominate a bidding Rosca, but not necessarily a random Rosca.

Their results show that Roscas are inefficient compared with credit markets, because they do not allow flexibility of intertemporal consumption, and it is optimal to give the indivisible good to more people at the beginning of the asset accumulation phase. Random Roscas dominate bidding Roscas when individuals are identical, since the bidding equilibrium constrains all individuals to have identical utility levels. Credit markets are efficient but do not allow agents to equate their marginal utility of consumption; and as the value of the durable good increases, this latter effect may dominate so that a market allocation is inferior ex ante to a random Rosca. The period of accumulation will also increase with the efficiency of financial intermediation.

Important outstanding issues include the sustainability of Roscas, since there is a strong incentive for those who win the pot early to stop contributing. This is not a serious problem in practice, however, since primitive societies are comparatively rich in the social sanctions with which to enforce such contracts, although thir ability to do so may deteriorate in the process of economic development.

The Economics of Rotating Savings and Credit Associations
Tim Besley, Stephen Coate and Glenn Loury

Discussion Paper No. 443, August 1990 (AM)