Working paper · July 2026 draft
Which Principals Matter? Accountability When Local Officials Answer to Both Bureaucrats and Politicians
Matthew Nanes, Nico Ravanilla, and Dotan Haim
In brief
When a local official answers to both a national agency and a patronage-based mayor, whose pressure actually works? Village officials in the Philippines received identical accountability letters differing only in who signed them. Only the national agency's letter changed behavior, and only citizens in those villages rated their government fairer and more effective.
Abstract
Government service provision often relies on local officials who implement policy while answering to overlapping administrative and political authorities. Yet we know little about which principals induce responsiveness in such settings. We study this question using a randomized evaluation of a local security reform in the Philippines. Village officials received an identical accountability letter that varied only in the identity of the principal issuing it: a national agency with formal administrative authority or the municipal mayor whose influence rests on informal patronage. Officials increased participation only in response to the national agency's letter. Citizens in villages where officials received the agency letter also reported higher perceptions of government fairness and effectiveness. These findings provide evidence on accountability among local implementers with one foot in politics and the other in bureaucracy, and motivate a discussion of when similar officials in other settings respond to accountability from one principal rather than another.
Research annotationQuestion, variables, design, findings, and mechanism at a glance▾
- Research question
- When a local official answers to both an administrative and a political principal, which one induces compliance?
- Independent variable (X)
- The identity of the principal signing an otherwise identical accountability letter to a village captain: the Department of the Interior and Local Government, a national agency with formal administrative authority, or the municipal mayor, whose influence rests on informal patronage. The letter was light touch by design, altering no formal responsibility, material incentive, or program resource, and merely announcing that the sender had asked the implementing NGO for monthly updates on the village team.
- Dependent variable (Y)
- Meeting-level compliance recorded by observers at each problem-oriented policing meeting, namely attendance out of the six assigned team members, delay in minutes past the scheduled start, and a zero to ten quality rating; and citizens' evaluations of government trust, fairness, effectiveness, and respect for human rights from a panel survey.
- Identification strategy
- Preregistered randomized evaluation embedded in a local security reform in Sorsogon Province, Philippines. The 198 participating villages were randomly assigned in equal numbers, 66 each, to the agency letter, the mayor letter, or no letter, with delivery timed between the third and fourth of seven monthly meetings so that meetings four through seven identify the effect. Because the reporting described in the letter ran through the implementing NGO rather than either principal's own channels, and the letters were otherwise identical, both principals were placed in the same informational position. Estimation is OLS with standard errors clustered by village, with no covariates beyond cross-randomized arms for the administrative outcomes and baseline controls for the survey outcomes.
- Main findings
- Only the administrative principal moved behavior. The agency letter raised attendance by 0.224 officials (SE 0.099) and delayed meeting starts by 6.2 minutes (SE 2.9), both significant at 5 percent, which is about a fifth of a standard deviation in attendance and closes roughly a sixth of the gap to full attendance. The mayor's letter produced no detectable effect on any of the three (0.111, -2.44, and 0.062, all insignificant). Neither letter improved meeting quality. On citizen attitudes, the agency letter raised perceived government fairness by 0.177 (SE 0.060), effectiveness by 0.117 (SE 0.057), and respect for human rights by 0.141 (SE 0.071), while the mayor's letter moved nothing. A placebo check finds no relationship in the three meetings held before the letters were delivered, and the mayor null does not vary with the mayor's prior vote share in the village or with lame-duck status.
- Mechanism
- Compliance concentrates on the dimension a principal can most readily verify. Attendance was named in the letter and is directly observable, and it moved together with longer delays, consistent with teams waiting to assemble their members; the ambiguous quality rating, which officials do not fully control, did not move. That the mayor null is unchanged by measures of individual mayoral strength suggests the difference lies in the type of authority rather than in the standing of particular mayors.
- Why it matters
- Bridges two literatures that developed in isolation, one on street-level bureaucrats accountable to administrative principals and one on brokers accountable to political patrons, by studying officials who are both at once and hold the same authority constant while varying only who is watching. The finding that formal administrative authority outperforms patronage-based political authority for eliciting compliance, and that citizens notice, speaks directly to how service delivery reforms should be routed in patronage democracies.
- Speaks to
- street-level bureaucracy and frontline complianceprincipal-agent models of accountabilityclientelism, brokers, and patronagepolicing and local security reformfield experiments on governance
Cite
@misc{ravanilla_accountability,
title = {Which Principals Matter? Accountability When Local Officials Answer to Both Bureaucrats and Politicians},
author = {Matthew Nanes and Nico Ravanilla and Dotan Haim},
howpublished = {Working paper},
}