Objectives. Determine whether the local supply of journalism leaves a measurable imprint on the dark figure of crime, and whether that imprint differs across crime types.
Methods. We develop a rational-choice model of intimidation and reporting that delivers both a testable cross-crime ranking and a reduced-form specification. We take our model to a novel dataset combining administrative crime records and a unique outlet-level mapping of local journalism in Argentina. Measuring media intensity by the number of local outlets, we estimate this model-implied reduced form by Poisson pseudo-maximum-likelihood (PPML), with a set of placebo and robustness tests. To further assess whether this relationship can be interpreted causally, we complement our empirical analysis with a matching design.
Results. A denser supply of local outlets is associated with more reported property crime (about +1.5% per outlet for theft and +3% for robbery, the latter near +30% across the bottom-to-top quartile contrast), no detectable change for assault, and a negative point estimate on reported sex offenses. The cross-crime ranking that the theoretical model predicts (robbery above theft, both positive; assault near zero; sex offenses directionally negative) is stable across matching estimators, an alternative journalist-based measure, and a month-by-month placebo, and is sharpest where outlets produce genuinely local content.
Conclusions. Local information environments shape what enters the official record of crime. Comparisons of recorded crime across regions that ignore media supply can misrepresent true victimization, and information-based interventions are better judged by their effect on offending, detection, and reporting than on aggregate counts. Because the design is cross-sectional and covers a single country and year, identification rests on selection on observables, and replication is warranted.