Threshold Events and Identification: A Study of Cash Shortfalls

Abstract

Threshold events are discrete events triggered when an observable continuous variable passes a known threshold. We demonstrate how to use threshold events as identification strategies by revisiting the evidence in Rauh (2006) that mandatory pension contributions cause investment declines. Rauh's result stems from heavily underfunded firms that constitute a small fraction of the sample and that differ from the rest of the sample in important ways; that is, the control group differs from the treated group. To alleviate this issue, we use observations near funding thresholds and find causal effects of mandatory contributions on receivables, R&D, and hiring, but not on investment. We also provide useful suggestions and diagnostics for analyzing threshold events.