Leveraging unprecedented austerity cuts, this paper studies a large-scale reorganization that closed over 70% of London police stations while keeping force-wide police strength constant. Combining geocoded police records with victimization survey data, I show that greater distance to stations increases violent crime, weakens police performance, reduces crime reporting and confidence in policing, and lowers house prices. Near surviving stations, officer concentration generates off setting local gains, reducing violent crime and raising nearby house prices. A spatial model combining both margins implies a net increase in violent crime, with effects varying sharply across neighborhoods. Alternative closure policies could have reduced aggregate crime.

Understanding parental responses to non-test score attributes is crucial to designing effective school choice systems. We study the impact on parental school choice of qualitative information sessions that mainly offer information about school attributes other than test scores. The outflow to private education is reduced by 17%, with larger responses among relatively advantaged students. Parents respond by increasing take-up of state school offers, intensifying competition for seats at oversubscribed schools. The intervention does not affect parental demand for measurable school attributes. We conclude that low-cost interventions providing qualitative information can improve state schools' finances and peer quality.

Many individuals experience crime as victims. However, many more individuals witness crime. We study how both experiences, being a witness to crime and being a victim of crime, impact crime-related perceptions, spanning confidence in institutions of the justice system, perceptions of neighborhoods, and personal safety. Combining waves of the Crime Survey for England and Wales from 2001-18 we document: (i) witnessing crime is twice as prevalent as victimization; (ii) while victimization rates have fallen by 32% over our study period, rates of witnessing crime have remained unchanged over time; (iii) violent crime constitutes a greater share of all crime over time; (iv) rates of witnessing violent crime have risen by 28%, which has 10 times as many witnesses per incident than other crimes. We then show that crime-related perceptions negatively associate with both experiences, with witnessing crime having stronger negative associations with con dence in institutions and views of local neighborhoods, while victimization is more strongly associated with feelings of personal safety. At a final stage, we consider implications for measuring the social cost of crime if witnessing experiences are also accounted for. We show: (i) the aggregate social costs of crime based only on victims have fallen by 17% over our study period; (ii) the aggregate social costs borne by witnesses have risen by 36%; (iii) once victim and witness costs aggregated, the total social costs of crime have remained relatively unchanged over time. By incorporating witnessing experiences into measuring the costs of crime, our analysis shed lights on a phenomenon observed in many countries: the general perception that crime is rising, a view unanchored from persistent declines in the actual incidence of crime. 

We study the role of middle managers in shaping public-sector performance, focusing on Borough Commanders who lead territorial divisions within the Lon don Metropolitan Police Service. We reconstruct their careers and assignments and link them to administrative data on police outcomes and workforce. Exploit ing manager mobility across divisions, we find that individual managers account for approximately one quarter of the variance in police productivity, measured by crime cleared per worker. More effective managers increase clearances across offense types, raise both police-issued sanctions and court convictions, and expand proac tive enforcement without increasing frontline staffing. These gains emerge rapidly after managers arrive and largely disappear after they leave. Effective managers also shape organizational responses to policy shocks: they mitigate increases in violent crime following police station closures and limit deterioration in response times following body-worn camera adoption. Finally, residents exposed to more effective managers report greater trust in the police and feel safer.

We study how immigration affects natives' access to public schools  by examining competition for primary-school places in England after the 2004 EU enlargement. The post-enlargement inflow was large and concentrated among Polish migrants, who were disproportionately likely to satisfy faith-based admissions criteria and therefore intensified competition for high-performing, state-funded Catholic schools. To address endogenous migrant location choices, we construct an  instrument for  post-2004 Polish settlement patterns using the location and capacity of post-World War II Polish resettlement camps digitized from archival records. Combining this historical variation with administrative data on the universe of pupils in public primary schools, we estimate the causal effects of local exposure to non-native pupils at school entry. We find that higher non-native exposure displaces native pupils from Catholic schools toward non-faith schools and toward closer schools with slightly weaker baseline performance and a less advantaged pupil intake. It also reduces natives' test scores at the end of primary school. Using sibling-priority rules and within-school designs with  family fixed effects, we show that these attainment effects are driven primarily by displacement across schools rather than  adverse within-school interactions.

Estimation of unintended costs of crime is scarce, although essential. This paper investigates a prominent indirect cost of crime, i.e. the effect of exposure to crime on achievement of pupils enrolled in primary school. I employ novel geo-referenced data on the universe of crimes from police records in London. The analysis takes advantage of the very fine spatial variation in crimes and in pupils' residences, precisely measuring the exposure to crime in the surroundings of pupils' homes. By exploiting the within-school variation in crime,  I find that crime occurring where pupils live has a significant negative impact on pupils' achievement at the final exams of primary school. The heterogeneity analysis shows that high-ability and wealthy students are those who suffer the most from exposure to crime. I find evidence of decreasing marginal sensitivity to crime, as pupils living in less criminal areas suffer the most from exposure to crime. Overall, I interpret this evidence as consistent with a story of scarring effect and adaptation. 

Resources and the justice system, with Sarah Cattan, Rocco d’Este, Magdalena Domínguez, Elisa Facchetti, Ilina Logani, Rocco Macchiavello, Ottavia Pesenti, Imran Rasul, Luca Rizzotti and Joe Tomlinson, Institute for Fiscal Studies, 2026

The value of parental and children experience in choosing secondary schools, with Lorenzo Neri and Marco Ovidi, 2020

Evaluation of the Meet The Parents project.