Poverty Graduation Programs: Psychological Implications and Sources of Heterogeneous Impacts

Poverty graduation programs in low-income countries provide an integrated package of interventions to poor rural households. The program traditionally includes the transfer of a productive asset, training on agricultural technologies and/or business administration, and building beneficiary self-confidence and other psychological assets through soft-skills or coaching interventions. Multiple studies have found participants to benefit substantially from the program and positive effects have been shown to sustain for several years after the end of the program. Nevertheless, there are two pressing questions that are still to be answered regarding the functioning of the program. First, it is not clear why these programs work. Its multifaceted nature presents a challenge to understanding which are the necessary components for it to be cost-effective. Second, there is evidence that the program effects are highly heterogeneous, and some participants do not benefit much from participating. Understanding the possible causes of the differentiated effects is important to design better interventions. This dissertation is an attempt to contribute to the answering of these questions. A central component of graduation programs is the life-skills coaching module. This is an expensive component that both program implementors and participants have claimed to play a central role in the success of these programs. Nevertheless, there have been few attempts to understand its role in the success of the intervention. In Chapter 1, I analyze how including a life-skills coaching module to Haku Wiñay, a program in Peru that had all of the other components of the graduation intervention but this one, affects a series of psychological variables. I find evidence that the additional program increases the belief of having control over the outcomes of events that affect participant's lives. At the same time, the evaluation reveals that individuals with initial levels of psychological assets above the median increase their annual income as a consequence of the whole program. The two results of this chapter may indicate that income could be positively affected through an increase in locus of control. The next round of data collection will allow to test this hypothesis. Chapter 2 stems from the results of Chapter 1 that reveal that participants that come into the program better psychologically equipped outperform those that start the program less equipped, and from the evidence from other studies highlighting the heterogeneity of impacts of graduation programs. In this chapter, with the aid of simulations generated by an infinite horizon model parametrized to mimic a rural economy where households may be stuck in a low equilibrium, I characterize the importance of two possible sources of heterogeneity - covariate shocks and initial psychological assets - in the impacts of a graduation program that transfers both physical and psychological assets. This simulation analysis suggests that a possible avenue for having graduation programs that generate benefits for a larger portion of the participants is by expanding pre-intervention activities to include an assessment of psychological wellbeing status. By doing this, the program may shift some of the psychological asset-building activities, such as the coaching component, from those that are well endowed towards those lacking these assets. At the same time, expanding the program towards insurance literacy and the offering of such products may be beneficial. Additionally, some of the program funds destined to each household may go directly towards insurance in order reduce the possibility of falling to a low equilibrium, reducing the probabilities of benefiting from the program in the face of a shock. Lastly, Chapter 3 looks at whether the coaching program embedded in the Haku Wiñay intervention generates differentiated effects for female-headed households compared to male-headed households when considering the psychological variables and a series of agricultural practices. This chapter is based on the same data from Chapter 1, and continues to explore possible sources of heterogenous impacts. The chapter shows that female household heads have pre-treatment psychological variables that are lower compared to those of male household heads, which may make them more receptive to the coaching component in particular since the intervention was modestly fine-tuned to speak to the needs of women. I show that certain psychological variables for female household heads do increase relative to the male household heads, although the differences are not statistically significant. I additionally find that female-headed households are more likely to grow vegetables, prepare and use organic fertilizer, and cultivate pastures as a consequence of participating in the coaching program relative to male-headed households, although the effect is only significant for the first variable, and marginally significant for the second. Nevertheless, given the small number of women that are household heads, the significance tests for the effects on this subgroup are underpowered