Variation in School-level Child Maltreatment Reporting

Involvement with the child welfare system varies dramatically across places. While some of this variation may reflect differences in underlying maltreatment risk, it may also result from differences in institutional responses to similar circumstances. To disentangle these explanations, this study uses Michigan public school data linked to Child Protective Services reports from 2017–2024. 

School-level reporting rates range from virtually none to more than half of students reported between kindergarten and fifth grade. The authors estimate school-specific reporting tendencies using value-added methods and document substantial variation in reporting behavior after accounting for differences in student risk. Using student movers, they show that these estimates are forecast-unbiased. These estimates imply that school-specific reporting practices account for roughly 40 percent of the variation in reporting rates across schools. 

 Differences in skill explain more than twice as much variation as differences in preferences, and higher-reporting schools tend to be more skilled at identifying high-risk cases. Counterfactual simulations imply that improving detection skill would yield significant welfare gains, while recent policy proposals aimed at reducing or equalizing reporting rates generally lower welfare by increasing the number of high-risk cases that go unreported.