Research

Weekly markets are an age-old way to connect geographically separated producers and consumers, and they remain widespread in rural areas of low-income countries. How do these gatherings shape development around them? To address long-standing data gaps, I combine historical sources with novel satellite-based methods to map marketplaces and measure local population density. I focus on Kenya over the last five decades and establish three stylized facts. First, while rural population quadrupled, two thirds of weekly markets operating in 1970 no longer do so today. Second and despite many markets no longer operating, population concentrated on average around markets that were active in 1970. Third, markets further from large cities saw the most population concentration relative to their surroundings. To rationalize these findings, I extend a model of rural-urban trade with markets as population-independent locations that aggregate otherwise sparse supply and demand and enable economies of scale in transportation. The model explains when new markets emerge, why some markets decline, and which complementary policies catalyze markets for local development.

Effective targeting of social policies and their rigorous evaluation requires relevant and accurate data. With the majority of the world's poor depending on agriculture and informal businesses for their livelihoods, information on these sectors is particularly valuable. I use high-frequency satellite imagery to develop a novel method mapping rural periodic markets across large geographies and tracking activity within them in real-time. I show that the method accurately detects existing markets and that measured activity not only correlates with alternative indicators, but also expands their temporal and geographical detail. Focusing on Kenya and Ethiopia, I present an application of the method to the effects of lockdowns and violent conflict on market activity. 

Many believe that high transport costs constrain market integration, technology adoption, and agricultural productivity in much of rural Sub-Saharan Africa. However, previous evidence is limited, especially in poor and remote rural areas, where the costs of data collection are prohibitively high. We use high-resolution satellite imagery and machine learning to measure discontinuities in land use at river crossings in northern Mozambique, where inadequate road infrastructure creates discrete jumps in travel costs but agricultural fundamentals vary smoothly across space. Better-connected banks have on average 1.2 percentage points more land under cultivation than their worse-connected counterparts, driven by river crossings with relatively high average market access where the difference rises to 4.1 percentage points. In the most remote areas, land use does not vary at river crossings. The results suggest that transport costs are a binding constraint in areas within reach of existing markets, but not in more remote ones. We also show that our approach, which uses freely-available imagery and analysis tools, detects local land use patterns substantially more accurately than off-the-shelf alternatives.

The image above shows the output from the land classification at 10m resolution we use in this paper. Green areas are classified as agricultural fields, red areas as natural vegetation.

Climate change is expected to affect the intensity and frequency of more extreme weather events and health is a primary channel linking climate change to human welfare, especially for countries and populations already facing poverty and malnutrition. We assess the effect of strengthened community-health services in rural Uganda on reducing the adverse effects of negative rainfall shocks on infant mortality. 

Investments in roads in underdeveloped rural areas are often assumed to stimulate agricultural production, ultimately allowing farmers to earn higher incomes. However, robust causal evidence on this is scarce. In part, this is because of a lack of data: few countries collect sufficiently comprehensive administrative data to permit detection of effects over large areas, and extensive direct data collection in remote rural areas is prohibitively expensive. This impact evaluation will measure the effect of a program of road improvements on agricultural intensification in rural Mozambique, combining freely available remote sensing data with high-frequency measurements of the state and usage of infrastructure on the ground. We will focus on the construction and maintenance of rural roads of various sizes in 12 priority districts in the provinces of Nampula and Zambézia. We will distinguish the causal effects of the intervention from other background trends that may also influence agricultural intensification by exploiting sharp changes in travel times generated by repairs and upgrades to roads and river crossings. We will leverage remote sensing data as well as advances in machine learning to measure changes in land use that would indicate agricultural intensification, such as increased conversion of land to fields, shorter fallowing periods, higher dry season vegetation indices, or changed land clearing practices. The approaches we use to measure agricultural intensification have the potential to be extended to other contexts where administrative data is lacking and data collection is costly. 

Small-scale farmers in low-income countries can potentially benefit from integration into global value chains, but lasting business relationships between producers and intermediates are often challenging to establish. We partner with a major exporter of vanilla - a crop underlying large price and weather risk - to understand farmers' motivations to engage in or walk away from pre-arranged buying agreements. Building on the detailed price, quantity and quality data collected by our partner, we focus in particular on the role of outside options and expected future engagements. 

The COVID-19 pandemic led governments around the world to impose unprecedented restrictions on economic activity. Were these restrictions equally justified in poorer countries with fewer demographic risk factors and less ability to weather economic shocks? We develop and estimate a fully specified model of the macroeconomy with epidemiological dynamics, incorporating subsistence constraints in consumption and allowing preferences over “lives versus livelihoods” to vary with income. Poorer countries’ demography pushes them unambiguously toward laxer policies. But because both infected and susceptible agents near the subsistence constraint will remain economically active in the face of infection risk and even to some extent under government containment policies, optimal policy in poorer countries pushes in the opposite direction. Moreover, for reasonable income-elasticities of the value of a statistical life, the model can fully rationalize equally strict or stricter policies in poorer countries. 

Smallholder farmers in developing countries are often poorly integrated into broader agricultural markets. It is widely believed that their participation in the market economy would pave the way to agricultural intensification, specialization and development of institutional capacity to address farmers’ specific needs. However, research aiming to understand this transition faces an intrinsic data problem: few developing countries have the institutional capacity to systematically collect information on the agricultural sector, such as on input choices, production, or trading of outputs. I develop a novel method to detect and monitor rural marketplaces using satellite imagery. Using daily images since 2016, I exploit the facts that (i) markets typically occur at a regular periodicity and (ii) have a distinct reflectance pattern in high-resolution optimal imagery. The algorithm extracts the market’s perimeter and measures activity within it based on the density of pixels with characteristic reflectance patterns. I detect up to 80% of known markets from ground-truthed reference locations. The activity indicators display intuitive variation, including the effects of Covid-19-related lockdowns. The method will allow me to provide a previously unavailable mapping of rural marketplaces across countries, with a current focus on East Africa. 

Early reports suggest the fatality rate from COVID-19 varies greatly across countries, but non-random testing and incomplete vital registration systems render it impossible to directly estimate the infection fatality rate (IFR) in many low- and middle-income countries. To fill this gap, we estimate the adjustments required to extrapolate estimates of the IFR from high-income to lower-income regions. Accounting for differences in the distribution of age, sex and relevant comorbidities yields substantial differences in the predicted IFR across 21 world regions, ranging from 0.11% in Western Sub-Saharan Africa to 1.07% for high-income Asia Pacific. However, these predictions must be treated as lower bounds in low- and middle-income countries as they are grounded in fatality rates from countries with advanced health systems. To adjust for health system capacity, we incorporate regional differences in the relative odds of infection fatality from childhood respiratory syncytial virus. This adjustment greatly diminishes but does not entirely erase the demography-based advantage predicted in the lowest income settings, with regional estimates of the predicted COVID-19 IFR ranging from 0.37% in Western Sub-Saharan Africa to 1.45% for Eastern Europe. 

It is widely understood that the same policy interventions can have different outcomes depending on the social structures they are implemented in. Much less clear though is what fundamentally determines communities’ social capital, or the way individuals in a society interact. This study analyzes how irrigation practices, a key feature of agricultural organization in India, affect different indicators of social capital. To this end, an informative household survey is combined with detailed historical agricultural statistics on irrigation and grain cultivation. The depth of the datasets allows for fine-grained analysis of effects of different kinds of irrigation on groups with varying degrees of involvement in the agricultural production process. I show a significant negative influence of specific types of irrigation on the prevalence of conflict and an increased likelihood for attendance of public meetings, yet it is not possible to establish a broad and consistent relationship between agricultural indicators and social capital, which contradicts previous research. Furthermore, specific groups with varying incentive structures based on their land-holding status and agricultural engagement appear to be affected by different kinds of irrigation. The results underline the need for a multidimensional analysis of both social capital and agricultural organization, as well as their interaction. I do not find evidence supporting the rice theory of culture.