We derive the efficiency loss from using grouped data to estimate coefficients of variables that vary across groups but not individuals within a group (e.g., state unemployment rates) when micro data are unavailable on the dependent variable. We present an empirical example of our theoretical...