The Art of Not Being Raided: Beliefs, Learning, and Consumption Under Immigration Enforcement

Immigration enforcement affects millions of individuals who make daily economic decisions under uncertainty about state action. Using data from 2014 to 2018, we study how households learn about and respond to enforcement risk by combining daily bank account transaction data with arrest-level records of Immigration and Customs Enforcement (ICE) operations. We show that, during that period, ICE enforcement followed predictable weekday patterns and that communities with large immigrant populations learned these patterns over time: Consumption is depressed not only on raid days but also seven days before and after—the same weekday, and bounces back in between—, indicating that behavior responds to beliefs about enforcement risk, not just to realized enforcement. To distinguish permanently foregone consumption from activity shifted to safer days, we estimate a structural model of consumption with pent-up demand and Bayesian learning. Roughly half of the immediate decline in consumption is recovered through subsequent rebound, while the remainder reflects genuinely foregone activity. Counterfactual exercises quantify the welfare costs of living under enforcement risk and the value of information. Eliminating enforcement risk entirely would increase Hispanic foreign born consumption by 3.6 percent. However, when enforcement is removed and immigrants must learn that their environment has changed, only 42 percent of the potential gains materialize within the first year. The remaining losses reflect learning frictions: even after state action ceases, beliefs must adjust.