## Notes from 01 July 2025 [[2025-06-31|← Previous note]] ┃ [[2025-07-02|Next note →]] Today I came across a fascinating paper by [Wayne Aaron Sandholtz]([Wayne Aaron Sandholtz](https://waynesandholtz.com/)) (2024), _The Politics of Public Service Reform_. It dives into a large-scale education reform in Liberia — the _[Partnership Schools for Liberia](https://www.lgtvp.com/en/portfolio/graduated-portfolio/leap)_ (PSL) - where some public schools were randomly assigned to private operators. The program clearly worked on technical grounds: students learned more and teachers showed up to work more often. But here's the paradox: despite these improvements, the incumbent party lost votes in the next presidential election. In fact, vote share dropped by 3 percentage points. So we get this uneasy tension: better services, worse electoral outcomes. Sandholtz unpacks this using two potential explanations. First, maybe voters didn’t like the reform? But no - surveys showed popular support, and in fact areas with the biggest gains in school quality gave _more_ votes to the incumbent. So the issue wasn’t public backlash. The second explanation hits harder: it was the teachers. The reform lowered their job satisfaction and made them less likely to support or campaign for the government. Electoral participation by teachers dropped by 25%. And critically, vote losses were concentrated in places where teachers were most politically disengaged - _not_ where student outcomes had worsened (they hadn’t). This adds an important layer to the political economy of public service reform. Teachers weren’t opposing the goals, but they were clearly reacting to how the reform affected them. Sandholtz points out that even if the public benefits, antagonizing an organized group like service providers can cancel out political gains. This challenges the logic where voters are supposed to reward improved performance... by showing how provider interests can disrupt that route. One last point I found interesting: the paper finds _no necessary trade-off_ between student learning and teacher disengagement. Those two effects weren’t correlated across locations. So it’s plausible that with a different policy design (one that built more buy-in from teachers) the government could’ve had the learning gains _without_ the electoral cost. I liked this paper a lot. It hits right at the heart of something I keep coming back to - how politically feasible reforms often hinge not just on what works, but on who’s organized and who feels heard. And the PSL case also intersects with another theme I care about: public-private partnerships in service delivery. Definitely one to revisit.