An extensive literature has examined the economic effects of non‐violent political instability events. Nonetheless, the issue of whether economies react differently over time to such events remains largely unexplored. Using synthetic control methodology, which constructs a counterfactual in the absence of political instability, we estimate the output effect of 38 regime crises in the period 1970–2011. A crucial factor is whether crises are accompanied by mass civil protest. In the crises accompanied by mass civil protest, there is typically an immediate fall in output which is never recovered in the subsequent five years. In crises unaccompanied by protest, there are usually no significant output effects. It is unclear, however, whether mass civil protest causes the greater fall in output or is simply an indicator of a more severe political crisis.