UC Davis

This dissertation is comprised of two essays on development economics and one essay on methodological issues related to the impact evaluation of development interventions. The first two chapters explore the implications of farmland consolidation on rural development in Colombia. In the first essay, I estimate the effect of a previous consolidation event on the structural transformation of rural economies. In the second, I evaluate the impacts of potential consolidation policies on aggregate welfare. Finally, the last chapter analyzes how attrition affects the internal validity of field experiments and provides recommendations for current empirical practice.

The consolidation of farmland is accelerating in many developing countries. An important question that arises in this context is, what are the impacts of this consolidation on rural development? In the first chapter, I examine the effect of land consolidation on the structural transformation of rural Colombia. To motivate empirical work, I present a conceptual framework where the impact of consolidation on sectoral employment and wages depends on the strength of the pull response in the nonfarm sector relative to the push response in the farms. I examine this question by assembling a novel dataset of rural counties and leveraging quasi-experimental variation in response to a trade shock that changed the scale of production during the nineties. I find that counties with an increase in large-farm consolidation experienced a reallocation of labor from the agricultural to the nonagricultural sector. Yet, this labor reallocation led to a decline in workers’ income over the medium term due to a sizeable increase in unemployment rates. These findings shed light on the implications of structural change within rural economies and the potential distributional impacts of consolidation across producers and workers.

Building on the first essay, the second chapter evaluates the impacts of large-farm consolidation on the aggregate welfare of rural populations. In particular, I develop a quantitative model of rural economies and conduct counterfactual experiments to evaluate the impacts of potential consolidation policies. This model features several empirical patterns connecting aggregate income with the distribution of farm sizes. The scale of operation affects agricultural productivity, while local wages and employment are determined by the concentration of profits through non-homothetic consumption growth. In line with previous work, I find that large-farm consolidation increases the welfare of farmers due to gains in agricultural productivity. However, since the demand for rural labor decreases substantially, workers are adversely affected, and aggregate social welfare declines. I show that these effects vary by type of consolidation and are exacerbated when the rise of large operations is driven by merging the smallest farms. These findings shed light on the distributional impacts of consolidation and speak to the trade-off between productivity, employment, and social welfare inherent in land policies.

The third chapter shifts focus to methodological issues on the impact evaluation of development interventions. Randomized control trials are an increasingly important tool of applied economics. Non-response on outcome measures at endline, however, is an unavoidable threat to their internal validity. In this chapter, we approach the problem of testing attrition bias in field experiments with baseline outcomes. We differentiate between two internal validity questions. First, does the difference in mean outcomes between treatment and control respondents identify the average treatment effect for the respondent subpopulation? Second, is this estimand equal to the average treatment effect for the study population? For each object of interest, we establish identifying assumptions and propose procedures to test its sharp implications. We also document that the most widely used test in the field experiment literature, the differential attrition rate test, is not a valid test of internal validity in general and provide a Stata package to implement the procedures that we propose. These findings have public policy implications since some agencies use attrition rates to evaluate the reliability of research studies (e.g., What Works Clearinghouse, US Department of Education).