Google Scholar   |   CV


Family matters: The double-edged sword of police-community connections

Full Paper

Scholars and policymakers frequently advocate recruiting “embedded” bureaucrats with strong ties to citizens to improve service delivery. Yet, officials who are too embedded in their community are often blamed for corruption, favoritism, and ineffectiveness. We argue that this ambiguity stems from a mismatch between individual and community-level effects of embeddedness. While personal ties increase engagement between directly connected citizens and bureaucrats, a community-level increase in bureaucrats’ personal ties alienates unconnected citizens and undermines claims of impartiality. We test this argument on public safety provision in the Philippines. We measure family networks in 286 villages, locate police officers within those networks, and analyze citizen survey responses. Citizens are more willing to trust and engage with officers to whom they are more closely related. However, in villages where officers are highly embedded, unconnected citizens evaluate their performance more poorly. Consequently, village-level officer embeddedness is associated with higher rates of feuds and disputes.

Journal of Politics (Accepted)
with Matthew Nanes and Michael Davidson

Deadly Populism: How Local Political Outsiders Drive Duterte’s War on Drugs in the Philippines

Full Paper

Around the world, populists have won elections on the strength of crowd-pleasing, but norm-defying, policy proposals. Although effective at mobilizing support at election time, these policies are often difficult to implement in practice because populists lack allies throughout the political system. Examining President Rodrigo Duterte’s brutal “War on Drugs” in the Philippines, we find that mayors excluded from existing establishment patronage networks filled this critical implementation gap for the Duterte administration. Employing regression discontinuity and difference-in-differences approaches, we demonstrate that outsider mayors received 40 percent lower public works appropriations, and, in turn, executed Duterte’s drug war much more aggressively. Outsider-led municipalities had 40 percent more anti-drug incidents and 60 percent more extra-judicial killings by police. The results illustrate an important trade-off between patronage politics and corruption (politics-as-usual), and violent democratic backsliding.

Journal of Politics (Accepted)
with Nico Ravanilla and Renard Sexton

Brokers, social networks, reciprocity and clientelism

Full Paper

Although canonical models of clientelism argue that brokers use dense social networks to monitor and enforce vote buying, recent evidence suggests that brokers can instead target intrinsically reciprocal voters and reduce the need for active monitoring and enforcement. Combining a trove of survey data on brokers and voters in the Philippines with an experiment-based measure of reciprocity, and relying on local naming conventions to build social networks, we demonstrate that brokers employ both strategies conditional on the underlying social network structure. We show that brokers are chosen for their central position in networks and are knowledgeable about voters, including their reciprocity levels. We then show that, where village social networks are dense, brokers prefer to target voters that have many ties in the network because their votes are easiest to monitor. Where networks are sparse, brokers target intrinsically reciprocal voters whose behavior they need not monitor.

American Journal of Political Science (Accepted)
with Nico Ravanilla and Allen Hicken

Self-enumerated field surveys on sensitive topics

Full Paper

Research on sensitive topics uses a variety of methods to combat response bias on in-person surveys. Increasingly, researchers allow respondents to self-administer responses using electronic devices as an alternative to more complicated experimental approaches. Using an experiment embedded in a survey in the rural Philippines, we test the effects of several such methods on response rates and falsification. We asked respondents a sensitive question about reporting insurgents to the police alongside a non-sensitive question about school completion. We randomly assigned respondents to answer these questions either verbally, through a “forced choice” experiment, or through self-enumeration. We find that self-enumeration significantly reduced nonresponse compared to direct questioning, but find little evidence of differential rates of falsification. Forced choice yielded highly unlikely estimates, which we attribute to non-strategic falsification. These results suggest that self-administered surveys can be effective for measuring sensitive topics on surveys when response rates are a priority.

Journal of Experimental Political Science (Accepted)
with Matthew Nanes

Digit ratio (2D:4D) and social integration: an effect of prenatal sex hormones

Full Paper

The position people occupy in their social and professional networks is related to their social status and has strong effects on their access to social resources. While attainment of particular positions is driven by behavioral traits, many biological factors predispose individuals to certain behaviors and motivations. Prior work on exposure to fetal androgens (measured by second-to-fourth digit ratio, 2D:4D) shows that it correlates with behaviors and traits related to social status, which might make people more socially integrated. However, it also predicts certain anti-social behaviors and disorders associated with lower socialization. We explore whether 2D:4D correlates with network position later in life and find that individuals with low 2D:4D become more central in their social environment. Interestingly, low 2D:4D males are more likely to exhibit high betweenness centrality (they connect separated parts of the social structure) while low 2D:4D females are more likely to exhibit high in-degree centrality (more people name them as friends). These gender-specific differences are reinforced by transitivity (the likelihood that one’s friends are also friends with one another): neighbors of low 2D:4D men tend not to know each other; the contrary is observed for low 2D:4D women. Our results suggest that biological predispositions influence the organization of human societies and that exposure to prenatal androgens influences different status seeking behaviors in men and women.

Network Science 5:4 (December 2017)
with Jaromír Kovárík, Pablo Brañas-Garza, Shannon Carcelli, Michael Davidson, and James Fowler

Alliance networks and trade: The effect of indirect political alliances on bilateral trade flows

Full Paper

How does the network of international political alliances influence trade flows? Previous work suggests that alliances matter in predicting trade outcomes because governments align trade policies with national security interests and firms take political relations between states into account when assessing risk. However, work to date investigates only the relationship between direct political alliances and trade, which ignores the complexity of international alliance structures. In this article, I argue that states and firms not only consider direct political relationships when shaping international trade, but also focus crucially on indirect alliance relationships. I find that higher levels of trade result when states have more shared alliances and when they are in the same alliance community. Once these indirect relationships are accounted for, the apparent association of dyadic alliances with trade is drastically reduced. Joint membership in an alliance community predicts an increase in trade that is more than twice the increase associated with a dyadic alliance. This effect is magnified when considering highly ‘central’ states in the alliance network. States trade significantly more with central states in their own alliance community and less with central states in other communities

Journal of Peace Research 53:3 (May 2016)

Using networks to combine ‘big data’ and traditional surveillance to improve influenza predictions

Full Paper

Seasonal influenza infects approximately 5–20% of the U.S. population every year, resulting in over 200,000 hospitalizations. The ability to more accurately assess infection levels and predict which regions have higher infection risk in future time periods can instruct targeted prevention and treatment efforts, especially during epidemics. Google Flu Trends (GFT) has generated significant hope that ‘‘big data’’ can be an effective tool for estimating disease burden and spread. The estimates generated by GFT come in real-time – two weeks earlier than traditional surveillance data collected by the U.S. Centers for Disease Control and Prevention (CDC). However, GFT had some infamous errors and is significantly less accurate at tracking laboratory-confirmed cases than syndromic influenza-like illness (ILI) cases. We construct an empirical network using CDC data and combine this with GFT to substantially improve its performance. This improved model predicts infections one week into the future as well as GFT predicts the present and does particularly well in regions that are most likely to facilitate influenza spread and during epidemics.

Nature, Scientific Reports 5 (2015)
with Michael Davidson and Jennifer Radin


Ongoing Projects and Working Papers

Sustained government engagement improves subsequent COVID-19 pandemic surveillance in conflict zones

Full Paper

Citizen information sharing is crucial to a government’s ability to efficiently respond to an outbreak of a pandemic disease, such as COVID-19. In conflict zones, however, citizens and local leaders often lack trust in state institutions and are unwilling to cooperate, risking costly delays and information gaps. We report results from a randomized experiment in the Philippines regarding an ongoing government effort to provide services and build trust with rural communities in a heavily conflict-affected region. We find that the outreach program increased by 20% the probability that local leaders provide time-sensitive risk information critical to the regional Covid Task Force. The effect appears to be driven by leaders who at baseline were skeptical about government capacity to deliver services and fairness in distribution, had neutral or positive attitudes towards rebels, and ambivalent feelings towards government. These findings highlight the important role that government efforts to build connections with conflict-impacted communities can play in determining public health outcomes during times of national emergencies.

Revise and Resubmit, American Political Science Review
with Nico Ravanilla and Renard Sexton

Civilian Social Networks and Credible Counterinsurgency: Evidence from the Philippines

Full Paper

How do civilian social networks influence the effectiveness of counterinsurgency? I posit that counterinsurgents’ ability to win civilian “hearts and minds” and elicit useful information about insurgents is driven in large part by information about counterinsurgent credibility that flows through civilian social networks. Civilian perceptions of local military control, as well as the government’s commitment to economic service provision, are shaped not only by what they observe in their own village, but also by the experiences of their friends and family members in the surrounding area. To test this argument, I use newly collected data on 1) family ties between over 55 million individuals in 42,000 Philippine villages, 2) village-level insurgent presence (from military intelligence reports) and 3) counterinsurgency-related development projects. I find that counterinsurgency efforts were significantly more effective at reducing insurgent presence when civilians in targeted villages had family ties to other nearby villages that received development projects. Counterinsurgency efforts were less effective when civilians in targeted villages had family ties to other nearby villages affected by insurgents.

Winner of the APSA Southeast Asian Politics Group Best Paper Award

Fire Alarms for Police Patrols: Experimental Evidence on Co-Production of Public Safety

EGAP Metaketa IV

Public safety provision requires joint action by citizens and the government. Authorities rely on citizen information sharing to allocate scarce resources efficiently. In the Global South, search costs like excessive travel time to a police station discourage citizens from reporting information. We experimentally evaluate an emergency hotline intended to reduce these costs in a rural province of the Philippines where authorities struggle to combat both crime and insurgency. We compare the hotline against both a control condition and a typical community policing initiative designed to increase citizen trust in the police. Relative to control, the hotline significantly increased reporting by citizens, even after accounting for the impact of typical community policing activities. However, while treated areas experienced less insurgent activity, the hotline had no perceptible impact on ordinary crime. Evidence suggests that the police acted only on information concerning their own priorities while ignoring other tips. Our findings demonstrate the impact of search costs on citizen-police cooperation, while cautioning that citizen engagement alone may not automatically improve public safety.

with Matthew Nanes and Nico Ravanilla

Clash of Clans: Local Elections, Political Networks, and Conflict in the Philippines

How do local electoral politics influence counterinsurgency effectiveness? Previous research has largely conceptualized counterinsurgency as being conducted by a unified national government motivated to expel insurgents. In reality, certain local elected officials benefit from an alliance with insurgents, undermining national government attempts to establish military control. Specifically, I argue that the control of political clans over local politics inhibits national government attempts to effectively conduct counterinsurgency. Political clans rely on insurgent presence to strengthen their hold on local power and, in return, use their political influence to subvert national government attempts to eliminate insurgent presence. While clans are able to slow national government attempts to win military control from insurgents, clans’ strong grip on power also results in reduced violent incidents due to lower levels of contestation. I test this theory using village-level data from the Philippines on 1) political clan influence, 2) village level electoral returns, 3) insurgent presence, and 4) conflict-related violence from 2009-2015.


De-escalating conflict by targeting at-risk youth and village elites: A field experiment in the Philippines

Pre-Analysis Plan

What can governments do to re-incorporate conflict-affected regions? Employing a randomized field experiment in a conflict-impacted region of the Philippines (Bicol), we test whether state security forces-sponsored monthly meetings between civilian agencies and local elites from conflict-afflicted villages can facilitate better public service delivery, align incentives by providing opportunities for credit claiming by local leaders, and incorporate these local elites into mainstream politics.

with Nico Ravanilla and Renard Sexton