This paper provides the first systematic empirical examination of bureaucratic nepotism and anti-nepotism
an entire modern bureaucracy. By linking confidential information on family ties and administrative
records for the universe of civil servants in Colombia, I uncover three sets of empirical findings.
First, using a novel
methodology of family network reconstruction, I provide evidence on the pervasiveness of close family
connections in the
public administration and demonstrate its negative relationship with the performance of public sector agencies.
by further exploiting within-bureaucrat variation in family connections generated by the turnover of top
non-elected bureaucrats, I show that family connections to public sector managers and advisors distort
allocation and compensation of workers at lower levels of the hierarchy. Connected bureaucrats receive higher
and are more likely to be hierarchically promoted but are negatively selected in terms of public sector
education, and records of misconduct. Third, I evaluate an anti-nepotism legislation reform by exploiting a
discontinuity in the set of family connections restricted by this law. I prove the limited effectiveness of this
and show how bureaucrats strategically responded to this policy change by substituting margins of favoritism and
reshuffling posts within the public administration.
As part of its Cold War counterinsurgency operations in Southeast Asia, the US government conducted a
"Secret War" in Laos from 1964-1973. This war constituted one of the most intense bombing campaigns in human
a result, Laos is now severely contaminated with UXO (Unexploded Ordnance) and remains one of the poorest
the world. In this paper, we document the negative long-term impact of conflict on Laotian economic development,
using highly disaggregated and newly available grid-cell data on bombing campaigns, satellite imagery, and
development outcomes. We find that bombings have a negative, significant and economically meaningful impact on
lights, expenditures and poverty rates. Almost 50 years after the conflict officially ended, bombed regions are
poorer and growing at slower rates than unbombed areas. A one standard deviation increase in the total pounds of
dropped is associated with a 33% decrease in nightlights. To tackle the potential endogeneity of bombing, we use
instruments the distance to the Vietnamese Ho Chi Minh Trail as well as to US military airbases outside Laos
before the conflict started. Using census data at the village and individual levels, we show the deleterious
of bombing and UXOs in terms of health, education, structural transformation, and rural-urban migration.
How can clientelism impose restrictions on bureaucratic state capacity? We develop a model of the politics of
state capacity building undertaken by incumbent
parties that have a comparative advantage in clientelism rather than in public goods
provision. The model predicts that, when challenged by opponents, clientelistic incumbents
have the incentive to prevent investments in state capacity. We provide empirical support for
the model's implications by studying policy decisions by the Institutional Revolutionary Party
(PRI) that affected local state capacity across Mexican municipalities and over time. Our
difference-in-differences and instrumental variable identification strategies exploit a
national shock that threatened the Mexican government's hegemony in the early 1960s. The
intensity of this shock, which varied across municipalities, was partly explained by severe
droughts that occurred during the 1950s.
We examine the long-term impact of violence on educational attainment, with evidence from Colombia's La
Violencia. Individuals exposed to violence during, and especially before, their schooling years experience
a significant and economically meaningful decrease in years of schooling. This impact has consequences
beyond human capital accumulation: exposed cohorts engage in activities with less human capital content.
Violence thus influenced aggregate development - particularly the process of structural transformation, in
which some sectors gain prominence as income increases. The effects result not so much from the direct
destruction of physical infrastructure, but from affected households' responses to the hardships of
This survey article reviews the literature on the multifaceted consequences of historical conflict. We revisit
key topics, which are especially relevant for the current Ukrainian context. 1) The negative long-term impact of
bombing campaigns and political repression against civilians. 2) The interplay between forced migration,
conflict. 3) The role of gender and war, with a special focus on sex ratios and conflict-related sexual
conclude with an empirical investigation of the ongoing Russian-Ukrainian conflict, including historical
such as ethnic populations, historical political repression and voting outcomes.
• "Media, Secret Ballot and the Process of Democratization in the United States"
with Leopoldo Fergusson
& BK Song Accepted at the Journal of Historical Political Economy.
Abstract · Paper
Can the media determine the success or failure of institutional reforms? We study the adoption
of secret voting in the US and the role of media in this arguably crucial step to improve
democracy. Using a difference-in-difference identification strategy and a rich dataset on
local newspapers, we find that areas with high levels of media penetration exhibited multiple
improvements in democratization outcomes following the adoption of the electoral reform.
Specifically, the press contributed to the decrease in partisan attachment and support for
dominant parties. It undermined the unintended consequence of the manipulation of electoral
boundaries and the fall in turnout. We consider multiple concerns about our identification
strategy and address the potential endogeneity of newspapers using an instrumental variable
approach that exploits the introduction of wood-pulp paper technology, and woodland coverage
at the county level in 1880. Exploring the heterogeneous effects of our results, we argue that
the media mattered through the distribution of information to voters, and the increase of
public awareness about political misconduct.
Tax evasion lies at the core of the relationship between citizens and the state: it reflects the level of
trust in the state and compliance with society’s implicit ‘social contract’. However, empirically
analyzing it is challenging, with few direct and reliable measures. This has hampered the advancement of
the theoretical and empirical literature, which is especially underdeveloped in the case of indirect tax
evasion. We conduct list experiments on a large sample of households to estimate the incidence of
value-added tax (VAT) evasion, as well as the extent of social desirability bias in respondent answers.
Around 20% of respondents engage in evasion and, surprisingly, they are not ashamed to recognize this
openly. Evasion is more prevalent in places with more informality and less physical presence of the state,
as well as among poorer, less educated individuals, and those who disregard the rule of law.
• "I Sell My Vote, and So What? Incidence, Social Bias, and Correlates of Clientelism in Colombia"
with Leopoldo Fergusson &
Economía Journal   19(1), 181-218. Brookings Institution Press, 2018.
Paper · Journal
Exchanging one’s vote for particularistic benefits—practices usually grouped under clientelism—is often
thought to weaken programmatic links between citizens and politicians and disincentivize public good
provision, as well as undermine voter autonomy and the ideal role of elections. However, empirically
analyzing this key phenomenon for the working of democracies entails formidable challenges. We conduct
list experiments on a large sample of households to estimate the incidence of clientelistic vote buying,
as well as the extent to which respondents refrain from openly recognizing this behavior. Nearly one out
of every five respondents engage in clientelism, and, surprisingly, they do not feel ashamed to admit it.
Guided by the existing literature and systematically verifying the sensitivity of the results to model
specification, we examine the robust correlates of clientelism and discuss the implications of our key
The extent to which authoritarian regimes use coercive, relative to redistributive, strategies
to manage social dissent exhibit significant variation across the territory they govern. We
argue that the incidence of different authoritarian tactics to deal with dissent depends on
the coercive capacity of the state, which autocrats often inherit from the past. Where
autocrats facing increasing discontent can rely on their capacity to coerce regime dissidents,
they are more likely to eschew redistributive strategies. In contrast, dissent increases the
likelihood of redistribution where autocrats lack readily-available tools for coercion. We
provide empirical support for this argument primarily using a difference-in-differences
identification strategy that exploits three sources of variation. First, we use a land reform
that between 1910 and 1992 redistributed more than 50% of Mexico's agricultural land. Second,
we exploit a wave of dissent around the 1960s. Finally, we use municipal data on the
availability of loyal semi-formal militias to coerce dissidents. Our results indicate that,
when confronted with dissent, the PRI regime redistributed relatively less land in
municipalities with more rural militia presence. We also show that, in those municipalities,
events expressing social discontent were more successfully deterred. The study sheds light on
how state coercive capacity shapes authoritarian strategies.
• "The Roots of Violence and Racism in the US: Historical Lynchings and Today's Police Brutality"
with Thorsten Rogall,
What explains the prevalence of police brutality against the black minority in the US? This paper shows that
historical lynchings are strongly associated with today's police violence. To this end, we combine historical
county-level data on the number of blacks lynched in the Deep South with recent geolocated micro-level data on
police violence. In terms of magnitudes, lynchings can account for about 10 percent of all police violence. The
relationship is robust to various controls, passes several placebo checks, and confirmed using an instrumental
variables strategy. In terms of mechanisms, lynchings created a persistent culture of both violence and racism.
particular, the effects are especially strong in counties that continuously experienced lynchings, as opposed to
where all lynchings were concentrated in one event. Furthermore, newspaper circulation is a strong enabler of
persistence and in-migration (especially of young men) mutes the effects. We also rule out typical alternative
• "Low Entrepreneurial Intent: A Legacy of South Africa's Mining Monopsony"
with Neil Lloyd, July
This project investigates the hypothesis that the low levels of self-employment and informal sector activity
in contemporary South Africa can be linked to the legacy of labour recruitment in South Africa’s gold mining
industry. At the outset of the Witwatersrand gold rush (1886) labour was largely sourced from outside of South
Mozambique, Botswana, even China. However, during the early 1900's a series of events culminated in the
recruitment to African men within South Africa. These events included the collapse of the Witwatersrand
Association (WNLA) as a centralized recruitment agency, the decision by the Transvaal government to
30,000 Chinese workers, and a recession in the Cape province which led the governor to encourage recruiters
their operations to the Cape's `native reserves'. We combine historical recruitment data with contemporary,
survey data to estimate the long run impact of the centralized mining recruitment on contemporary
research design exploits the historical borders of districts selected for recruitment, as well as the
location of private sector recruiters prior to the re-establishment of a monopsony in 1918.
• "Digging for Votes: Electoral effects of disclosing illegal mining"
• "Decolonizing the State: Evidence from the Tanzanian Civil Service, 1956-76"
with Jeremy Bowles,