4.4. Endogeneity Concerns
To address the endogeneity concerns, we apply a two-stage least squares (2SLS) instrumental variable approach. In the first-stage regression, we regress the continued increase of public health threats on two instrument variables (IMMIGRANT and EMIGRANT). Here, IMMIGRANT is the six-day mean value of the ratio of the daily provincial immigrants to the national immigrants; and EMIGRANT is the six-day mean value of the ratio of the daily provincial emigrants to national emigrants. The daily mobility data is collected from the Baidu Migration website. Based on Jia et al. [10] and Kraemer et al.’s [9] findings, we argue that when the provincial immigrant (emigrant) rate is increased, the new COVID-19 cases are more (less) likely to be confirmed for the out-in (in-out) human mobility, enhancing (decreasing) the likelihood to face the continued increasing public health threats. We report the first-stage regression results in Column (A) of Table 7, where we regression CIPHT on all two instruments and the control variables added in the second-stage regression. We find that the coefficient of EMIGRANT is negative and significant, which consistent with our conjecture. Columns (B) and (C) of Table 7 report the second-stage regression results. We find that the coefficient of Predicted_CIPHT, estimated from the first-stage regression, is negatively and significantly associated with CAR [−1, 1] and CAR [−2, 2], respectively. Accordingly, the robust results based on the 2SLS approach mitigate endogeneity concerns and strengthen the main inference that continued increasing public health threats significantly influence market performance.