I am an applied economist focusing on program evaluations in two areas: urban economics issues and customs, trade facilitation and governance. My projects involve work on combining new and existing data sources to answer policy questions and analyze customs and border crossing, cities, and networks of trade and transport. I rely combination of administrative, survey, and webscrapped data, satellite imagery and mobile phone records.
This paper examines how the provision of information to customs inspectors and their monitoring impact tax revenue collection and fraud detection, conducting two separate but highly complementary analyses. First, we examine the returns to the third-party provision of information, in the form of valuation advice for a subset of high-risk import declarations. Second, we document the results of a nationwide risk management randomized control trial to increase revenue collection by increasing information provision and monitoring of customs inspectors’ actions. The treatment consisted in the provision by the customs risk management unit of detailed comments on a subset of randomly selected high-risk declarations and a set of randomly selected high-risk declarations were flagged as being subject to an ex-post audit. We find that the provision of extensive comments helped reduce fraud but the resulting improvements in tax collection are small and only valid for declarations that do not benefit from third-party valuation advice already. We find no impact of monitoring on customs performance. The lack of impact of the monitoring treatment and the relatively limited impact of extensive comments may in part be due to the timing of the randomized control trial, which was implemented just before Madagascar’s general election, which led to significant turnover among senior customs management, which undermined the credibility of potential punitive actions. Moreover, the absence of well-defined sanctions may have limited the deterrence effect of monitoring.
Many government agencies have multi-dimensional missions, in which achieving one objective can reduce attainment of another organizational objective. This presents
particular challenges to government analytics. Incomplete measurement of objectives risks encouraging attainment of measured objectives while unknowingly impairing other objectives. This chapter showcases how government analytics can be applied in such contexts, using the example of customs agencies. Customs agencies typically have three core objectives: facilitating trade, collecting revenue, and ensuring the security and safety of the goods entering or exiting the country. Attaining one objective (e.g. greater safety of traded goods) can come at the expense
of another (e.g. facilitating trade). This puts a premium on effective measurement of all dimensions of a customs mission, which requires triangulating different data sources. This chapter showcases how this can be done, deriving indicators for (1) trade facilitation (e.g. costs of the process, in particular time delays); (2) revenue collection (e.g. trade volume and revenue collected based on the assessed value); and (3) safety (e.g. number of goods in infraction seized). The chapter also underscores how a wider use of the customs database itself could help measure performance, combining it with other data collection methods such as time release studies (TRS) and exciting developments in GPS tracking data.
with June Ghimire, Jonna Bertfelt and Bryan Mthiko
In this project, the team is studying two trade facilitation reforms in Malawi: the streamlining of the procedures at the border and the connection of several agencies to the ASYCUDA database, and the change in declaration filed for goods for warehousing. The trade facilitation interventions are anticipated to augment revenue collection by the Malawi customs authorities, reduce clearance time for goods at the borders, decrease trade related costs, and increase trade volumes. The data system supporting the evaluation is low cost and based on making better use of existing customs data. While customs data are routinely collected, they are seldom used for monitoring the performance of the border clearance process and of customs agents at the border at a high frequency. The data system will combine transaction level records of customs data and risk management advice combined with surveys of traders. The survey of traders will provide an understanding of total trade costs related to crossing borders, including transport costs and bribes/ unofficial fees. This survey is intended to cover a wider range of trade costs at a low cost and hence be able to be repeated more frequently than the Time Release Studies usually used for monitoring trade facilitation objectives.
with A. Arun
Transport infrastructure has the potential to bolster economic development by connecting several pillars of a nation’s economy. Several studies have independently
documented the gains and losses from investments in transport infrastructure, such as direct and indirect outcomes, wider economic benefits, and complementary measures. This article presents a programmatic approach for evaluating the impact of transport infrastructure investments with respect to their short and long run goals. It builds on existing data and reviews newly available high-frequency geospatial data and innovations on outcome measurements. The program exposition is substantiated with examples of projects currently evaluated under the ieConnect for Impact program at the World Bank
with an Björkegren, Geetika Nagpal, and Nick Tsivanidis
In this project, we are studying two questions related to public transit: does the introduction of a public option crowd out the private sector, and what are the distributional impacts of
new system across commuters and informal incumbents? What are the optimal regulations for the network of public and paratransit in this context? We use the Lagos Bus Reform Intervention to study the two questions based on a large scale data collection on paratransit frequencies and fares, e-ticketing data and cell phone home registry.
with Simon Alder, Kevin Croke, Rob Marty, Ariana Vaisey.
This project studies the impact of the 20 years Road Sector Development project on land use in Ethiopia. The team uses satellite imagery to look into changes in cropland and urbanization following the opening of the newly built or rehabilitated roads. Given the large number of roads built or rehabilitated, we are able to provide estimates for different road types.
with L. Zavala
This research project measures the impact of a major reform toward regional integration – the 2018 Customs Union between Honduras and Guatemala. In the first stage, we leverage our engagement with the Honduran authorities to estimate the Union’s effect on key trade outcomes at the firm level. The granularity of our data allows us to utilize identifying variation across products and destinations within a firm; to consider the impact on both imports and exports; and to examine heterogeneity by pre- existing trade barriers. In the second stage, we will establish a similar partnership with the Guatemalan authorities. Through this joint collaboration, we will develop the first database of firm-to-firm trade across borders. In addition to providing novel outcomes for analysis, these data will allow us to measure the impact of the Customs Union on both member countries. The project will advance our understanding of regional integration using novel network data, rigorous evaluation methods, and a deep engagement with government partners
with Rob Marty.
Household surveys give a precise estimate of poverty; however, surveys are costly and are fielded infrequently. We demonstrate the importance of jointly using multiple public and private sector data sources to improve predictions of levels and but also changes in wealth for a large set of countries. We train models using 63,854 survey cluster locations across 59 countries, relying on data from satellites, Facebook Marketing information, and OpenStreetMaps. The model generalize previous studies to a wide set of countries and explains 61% of the variation in levels of wealth at the survey cluster level and 68% of the variation at the district level, and the model explains 5% and 10% of the variation of changes in wealth at the cluster and district levels. Models perform best in lower income countries and in countries with higher variance in levels and changes in wealth. Features from OpenStreetMaps, nighttime lights, and land cover data are most important in explaining levels of wealth, and features from nighttime lights are most important in explaining changes in wealth
In this paper, I look at the impact of inter-university research partnerships on the
production of research outputs. Using an original data set of scientic publications,
I analyze the network of research in Spain based on the network of Spanish coauthors.
I show how the growth in research productivity of Spanish institutions
before the crisis was linked to the increase in universities’ budgets and in interuniversity
In this ongoing impact evaluation, we study the impacts of transport investments on economic activity in Pakistan.