Note(s): This table presents the results of 2SLS regressions. We use local officials' age (⁠$AGE$⁠) and the mean value of official turnover of other cities in the same province (⁠$TURNOVER_PEER$⁠) as instrumental variables for local official turnover. In the first stage, we regress endogenous variables $ΔlnSALEt×TURNOVERt$ and $ΔlnSALEt×Dt×TURNOVERt$ on $ΔlnSALEt×AGE$⁠, $ΔlnSALEt×TURNOVER_PEER,ΔlnSALEt×Dt×AGE$⁠, $ΔlnSALEt×Dt×TURNOVER_PEER$ and control variables used in the baseline model. In the second stage, we replace $ΔlnSALEt×TURNOVERt$ and $ΔlnSALEt×Dt×TURNOVERt$ by their predicted values from the first stage and re-run the baseline regression. Panel A reports the estimation results based on the turnover of city heads and Panel B reports the estimation results based on the turnover of city mayors. The subscript $t$ denotes the time index, while prefecture-city and firm indices are omitted for brevity. The dependent variable is $ΔlnXOPR$⁠, the log-change in operating costs. $ΔlnSALE$ is the log-change in sales. $D$ is a dummy variable that takes the value of one if sales decrease, and zero otherwise. $TURNOVER_HEAD$ is a dummy variable that takes the value of one if the city head is changed in year $t$⁠, and zero otherwise. $TURNOVER_MAYOR$ is a dummy variable that takes the value of one if the city mayor is changed in year $t$⁠, and zero otherwise. All variables are defined in the Table A1. Regressions control for both firm and year fixed effects. Robust t-statistics reported in parentheses are based on standard errors clustered by firm. In this table, *, **, and *** denote statistical significance at the 10, 5, and 1% levels, respectively