Association for Public Policy Analysis & Management

Cash transfer programs (CTPs) have increasingly become one of the central policy tools for alleviating chronic poverty and deprivation for the world’s most impoverished households. A recent review by the World Bank found that about 150 countries around the world have some type of cash assistance program, reaching approximately 800 million people. The emerging literature from several rigorous impact evaluations (IE) have shown positive impacts on a broad range of household living conditions including schooling, health seeking behavior and household production capacity. This evidence further strengthens the argument for CTPs as a preferred social protection tool that as well act as a conduit for a sustainable pathway out of poverty.

A major gap in the evidence literature, however, is the fact that most of the IEs typically report only average treatment effects (ATE) based on intent-to-treat, or local average treatment effects (LATE) based on treatment on the treated. The most that these IEs do is to explore heterogenous treatment effects across pre-defined variables such as among female headed households or among small households. As is well known, ATEs or LATEs could easily mask many important distributional characteristics of impacts, and does not allow for learning about which types of households are able to make the most from CTPs, and whether indeed some households are ‘harmed’ by CTPs because of lack of capacity to handle the ‘shock’. An understanding of this dynamics would help with better policy in targeting and in overall program design.

This paper therefore seeks to fill this crucial evidence gap by examining various conceptions of consumption mobility using data from an IE of the Malawi social cash transfer program (MSCT). The data consists of three waves of household level micro data with the first collected immediately prior to the start of the implementation of the cash transfers in 2013. The other two follow-ups surveys were conducted about 12 months and 22 months after the start of the cash transfers. The MSCT is targeted at ultra-poor labor constrained households and provides about 20 per cent of baseline consumption to households every month.

Consistent with much of the literature, the MSCT showed significant positive impacts on consumption and other living conditions. This paper uses the consumption mobility patterns within the control group to predict likelihoods of observing the consumption levels among the treatment group in the follow-up surveys. Double and triple difference specifications are used to identify which baseline characteristics of households are favorable to high mobility. Similar strategies are employed to also analyze which post-treatment behavioral responses are favorable to high mobility. Household baseline productive efficiency and social integration and found to be favorable to high mobility while having a chronically ill household member is one of the strongest predictors of low mobility. Households venturing into livestock production post-cash transfer also appear to have done much better than those venturing to non-farm household enterprises. Giving due consideration to some of these dynamics could be useful in improving the effectiveness of CTPs as a vehicle for inclusive economic growth.