Bank-Insured RoSCA for Microfinance: Experimental Evidence in Poor Egyptian Villages

How to cite

Karlan, Dean; Osman, Adam; El-Komi, Mohamed; El-Gamal, Mahmoud. 2016. "Bank-Insured RoSCA for Microfinance: Experimental Evidence in Poor Egyptian Villages." AEA RCT Registry. AEARCTR-0001814. https://doi.org/10.1257/rct.1814-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Dean Karlan* , Adam Osman , Mohamed El-Komi , Mahmoud El-Gamal

* Primary investigator

Registry ID
AEARCTR-0001814
DOI
https://doi.org/10.1257/rct.1814-1.0
First registered
Last updated
Status
Completed
Trial dates
2009-09-01 – 2010-02-28
Countries
Egypt (North Africa)
Registry record
http://www.socialscienceregistry.org/trials/1814

Abstract

Microfinance institutions (MFIs) have continued to grow over the past few decades, both in numbers of clients and portfolio sizes. The growth of these MFIs has enabled greater access to credit in many of the world's less developed nations. However, recent studies have shown that very many of the poor - especially Muslims - remain unbanked. Confounding this problem in many Muslim countries is the poor's propensity to reject microfinance, when available, on religious grounds. In this paper we develop an alternative microfinance model which aims to establish credit unions for the poor in which the bank plays the role of a guarantor in the familiar rotating savings and credit association (RoSCA). We test the performance of this model against a stylized sequential Grameen-style microcredit provision in a "laboratory experiment in the field" conducted in poor Egyptian villages. Our model of bank-insured RoSCAs is shown to solve coordination-failure problems that may otherwise prevent the spontaneous development of informal RoSCAs in practice. Empirically, our bank-insured RoSCA model generated significantly higher take up and repayment rates than the Grameen model. This suggests that this model, by overcoming the religious barriers to credit, can be a useful alternative to Grameen-style microfinance.

Intervention

Each session is randomly assigned to be either a Grameen-style game or a RoSCA game. In each game, players are partnered up and Player 1 is endowed with a specific amount of initial capital. Players have the opportunity to make a capital investment, but must take out a loan to do so. If Player 1 takes out a loan, their default or repayment behavior affects the options available to Player 2 in the future, and vice versa.

In the Grameen-style game, players borrow from and are punished by the bank. However, after the first period, players can borrow from each other to repay the bank.

In the RoSCA game, if the players mutually decide to play the game and invest capital, they pay a small fee to the bank upfront which will guarantee the RoSCA against default. Player 1 receives the first RoSCA loan. If they default, they receive a loan from the bank to repay the RoSCA (which they can either repay or default). If Player 1 repays the RoSCA, Player 2 will receive a loan from the RoSCA, and will face the same choices as Player 1.

The games differ because in the Grameen-style game, bank loans occur first, with player support secondary. In the RoSCA game, this pattern is reversed, with intra-player support first, and bank loans secondary. For more detail on the games, please refer to El-Gamal et al. (2015).

Experimental design

We conduct laboratory experiments in the field, using very poor subjects in rural Egypt who are eligible for microloans, and compare the take up and repayment rates of randomly assigned subjects under two microfinance designs - a sequential Grameen-style microlending structure as well as our proposed alternative, a bank-insured fixed-order RoSCA. Subjects participated in experimental games, modelled after a Grameen-style or RoSCA loan structure. Subjects who chose to participate in the games kept the money in their possession at the end of the game. After the game we administered, subjects answered few survey questions to gather demographic data for use in our analysis.

Design details

Randomization method
Done in office by computer
Randomization unit
Session
Clusters
25 experimental sessions
Observations
354 subjects
Arms
RoSCA: 15 sessions Grameen-style: 10 sessions
Primary outcomes
Proportion of participants who play, Take-up rates, Repayment rates

Materials

Papers from this trial

  1. El-Gamal, Mahmoud, Mohamed El-Komi, Dean Karlan, Adam Osman. "Bank-Insured RoSCA for Microfinance: Experimental Evidence in Poor Egyptian Villages. Science Direct 2014; .doi.org/10.1016/j.jebo.2014.02.025. http://www.sciencedirect.com/science/article/pii/S0167268114000742

Cited by

No citation data available for this registration yet.

Keywords