UC Berkeley

This dissertation examines the impacts and functioning of a community led public works intervention in Yemen, the Labor Intensive Public Works Program (LIWP). LIWP was designed using the "twin-track" approach of combining short-term relief with long-term investments. The program transfers funds to poor rural households by creating temporary employment opportunities in projects that benefit the local community. Income from program wages provides short term protection against negative consumption shocks, while the public works projects themselves provide medium to long term benefits for the community. The first chapter describes an unexpected effect of the intervention on decreasing local prices and explains how this effect can occur as a result of the existing informal institution of store credit in rural communities. The second chapter examines heterogeneity in program impacts related to the type of project chosen by the community and the correlates of this choice. The third chapter provides a detailed evaluation of the impact of the LIWP intervention on households in communities where projects took place.

In Chapter 1, I show the surprising empirical finding that in villages with random assignment to participate in LIWP, the price of staple goods in local stores rose less over time, and connect this to the institution of store credit in isolated villages. I find that there is an increase in prices in communities with 5 or more stores, and a significant relative decline of about 10% in villages with 4 or fewer stores. I develop a simple model of credit as insurance based on qualitative interviews in the field which shows that villagers are willing to pay a premium to local store owners over a more distant and anonymous market because their local store owner can be relied on to let them buy on credit in the future. In this way, the cost of zero-interest credit is internalized in prices charged, however this informal contract is only sustainable if there are sufficient dynamic incentives, which will be more likely in less competitive markets, consistent with the empirical findings.

In Chapter 2, which is based on joint work with Alain de Janvry and Elisabeth Sadoulet, we examine the choice of project by communities in the LIWP intervention. While LIWP was designed with the short-term purpose of providing cash for work to underemployed villagers, the community as a whole decided on the type public infrastructure to be constructed using the labor paid for by LIWP. We characterize the various projects in terms of skill intensity based on administrative data about relative pay receipts to unskilled and skilled workers and show that the skill-intensity of project choice is correlated with demographic features of the village which imply greater political power for older men. This result indicates that the choice of projects by the community is influenced by the relative distribution of benefits. To identify the channel at play, we show that in the immediate aftermath of the 2011 economic and political crisis in Yemen, there was a higher preference for skill-intensive projects. This is consistent with a model in which older, skilled men who prior to the crisis would not have self-selected as participants in the LIWP program due to the relatively low wages paid compared to alternative employment options, became potential participants due to the crisis. As a consequence, they pushed for skilled labor-intensive projects in which they could capture a larger share of the benefits.

In Chapter 3, which is based on joint work with Alain de Janvry, Elisabeth Sadoulet, and Daniel Egel, we use a randomized control trial to measure improvements in household welfare brought about by the program intervention. We find positive program impacts on household income and durable good ownership, reflecting the short term benefits of wages from work in the LIWP project. We also find surprisingly large program effects on decreasing the average outstanding debt. By analyzing the calorie content of staple grains consumed per person, we find a program impact equivalent to 11-13% higher calorie consumption in treated compared to untreated communities. We also summarize community perceptions of the project benefits and find that completed water related projects reduced water scarcity and fetching time.