Reply to Jonathan Morduch's "Does Microfinance Really Help the Poor? New Evidence from Flagship Programs in Bangladesh"

This paper addresses the concerns raised by Jonathan Morduch in his paper, "Does Microfinance Really Help the Poor? New Evidence from Flagship Programs in Bangladesh", about the methods and results of M. Pitt and S. Khandker (PK) in their study of the impact of group-based credit programs on poor households in Bangladesh.

Murdoch's concerns are:

• "Mis-targeting" of eligibility biases the results;
• Village fixed effects are incorrectly specified and unduly parsimonious;
• "New Evidence" fails to find positive program effects;
• The higher positive impact of female credit on household consumption may simply reflect the diminishing marginal returns;
• The linear functional form is not flexible enough.

The paper argues that Morduch has misunderstood and mischaracterized PK's methods, and that he has applied incorrect methods to collect his new evidence. The author responds as follows to Morduch's criticisms:

• Credit programs are aware of the differential cultivatibility of plots of land and make adjustments for land quality in judging eligibility;
• PK allow for program availability to be responsive to the unobservables of only the target group;
• Morduch fails to set out a clear framework justifying his differences-in-differences estimate;
• While declining returns may exist, the differences in loan sizes that Morduch cites are an unrealistic explanation for differential returns based on gender;
• PK include the most "crucial" interactions in the consumption equation. It is the land eligibility rule that is the source of identification of this model.