3. Research Design

3.1. Sample Selection
To test our hypotheses, we use the Chinese setting from 10 January to 31 March 2020. The financial and executive data is collected from the CSMAR database; the COVID-19 data is collected from the National Health Commission of the People’s Republic of China; the stock price is collected from the RESSET database; and the province-level data is collected from the National Bureau of Statistics. Panel A in Table 1 shows the sample selection process. After dropping the sample with missing data, the final sample contains 178,805 firm-day observations.
Panel B of Table 1 shows the sample distribution by month. There are 237 firm-day observations for firms located in provinces face a continued increase in public health threats (CIPHT = 1) in January. The sample group of CIPHT = 1 increased dramatically during February, while there is a declining trend in the group of CIPHT = 1 during March. Panel C of Table 1 shows the sample distribution by industry. Firms from the manufacturing industry dominate the full sample, which is consistent with China’s industrial structure. The proportion of cross-industry distribution in sub-samples is similar to the full sample.

3.2. The Measure of Continued Increasing Public Health Threats
We attempt to evaluate public health threats using a continued increase of provincial COVID-19 cases to ascertain the investors’ reaction to the local increasing health threats. Specifically, we generate an indicator variable (CIPHT) to represent the regional increasing health threats. Here CIPHT equals one if there have been provincial new COVID-19 cases for at least six consecutive days including the current day and zero otherwise. The information on daily province-level new confirmed COVID-19 cases is collected from the National Health Commission of the People’s Republic of China. The COVID-19 disclosure news started from 11 January and contained data from 10 January. The daily based distribution of CIPHT = 1 sample is listed in Appendix B.

3.3. The Measure of Market Reaction
We apply the cumulative abnormal return to represent the short-window market reaction to the continued increase of public health threats. Particularly, following the prior studies [38,50,51,52,53], we compute two measures of the firm’s cumulative abnormal return (CAR) with a three-day [−1, 1] window and a five-day [−2, 2] window based on the market model as follows:Firm Return = β0 + β1Market Return + ε(1)  where Firm Return is the firm’s daily stock return, and Market Return is the daily stock market return. Similar to the prior studies [51,54], we estimate the value of the constant term (β0) and the systematic risk of the stock (β1) based on model (1) over the period from current day 200 to current day 60 ([−200, −60]) and day 0 is the date of the current day. Then we get the abnormal returns by calculating the residuals of model (1) with the estimated value of the constant term and systematic risk of the stock. Finally, we generate two types of cumulative abnormal returns around the three-day and five-day short windows (CAR [−1, 1] and CAR [−2, 2]). These two short-window abnormal return measures capture investors’ risk assessment of expected costs of the continued increasing public health threats.

3.4. Empirical Model
We describe the regression model for the main test of H1. The regression models for cross-sectional tests are described in Section 5. To test H1, we apply the multiple regression model as follows:CAR = β0 + β1CIPHT + β2PRO_CASE + β3SIZE + β4ROA + β5CURR + β6R&D + β7LOSS+ β8LEV + β9OPCF + β10TURN + β11CEO_AGE+ β12CEO_COM + β13CEO_TEN+ β14CEO_DUA + Week FE + Industry FE + Province FE + ε(2)  where CAR refers to our two types of accumulative abnormal return (CAR [−1, 1] and CAR [−2, 2]), CIPHT is an indicator variable that equals one if there have been provincial new COVID-19 cases for at least six consecutive days including the current day and zero otherwise. Based on H1, we suppose a negative coefficient of CIPHT.
Model (2) contains several determinants of accumulative abnormal return. Considering that provincial accumulated COVID-19 cases would affect the investors’ risk assessment, we add PRO_CASE into our model. PRO_CASE is the six-day mean value of the provincial ratio of the daily accumulated confirmed COVID-19 cases to the resident population. Moreover, along with prior studies, we control the firm attributes that will affect abnormal return [39,55,56]. SIZE is the natural logarithm of total assets; ROA is the return on assets; CURR is the current ratio; R&D is the ratio of R&D expenses to sales; LOSS is an indicator variable that equals one if the firm suffered a loss and zero otherwise; LEV is the leverage ratio of total liabilities to total assets; OPCF is the ratio of the firm’s operating cash flow to total assets; TURN is the asset turnover ratio. In addition, following prior studies [55,57,58,59], we add CEO attributes that will affect the market reaction. CEO_AGE is the age of the firm’s CEO; CEO_COM is the ratio of the firm’s CEO compensation to the net income; CEO_TEN is the tenure of the firm’s CEO that is defined as days of CEO’s tenure divided by 365; CEO_DUA is an indicator variable that equals one if the firm’s CEO holds a concurrent post in other work units and zero otherwise. Finally, we add week fixed effects, industry fixed effects, and province fixed effects. Appendix A presents detailed variable definitions.

3.5. Descriptive Statistics
Panel A of Table 2 presents the descriptive statistics on all variables in the model (2) for the full sample. The median values of the CAR [−1, 1] and CAR [−2, 2] are −0.004 and −0.005, which suggests that during the COVID-19 outbreak period, more than half of the assessments of the firm performance are negative. The mean value of CIPHT is 0.360, which shows that there is 36.0 percent of the observations show the continued increasing public health threats. Panel B of Table 2 reports mean difference test between sub-samples (CIPHT = 0 vs. CIPHT = 1). The mean values of CAR [−1, 1] and CAR [−2, 2] are significantly lower for firm-days with continued increasing public health threats (−0.001 and −0.001) than for those without continued increasing public health threats (0.000 and 0.000). This result provides preliminary support on the negative relationship between continued increasing public health threats and abnormal return. Regarding the determinations of accumulative abnormal return, we find that firms facing continued increasing public health threats are located in the provinces where have more confirmed accumulated COVID-19 cases; have the larger size and better performance; have lower R&D ratio and leverage ratio; and hire CEOs with younger age, higher compensation, longer tenure, and a higher likelihood to hold a concurrent post.
Table 3 reports the correlation matrix of the variables in the main tests—Pearson correlations in the lower diagonal and Spearman correlations in the upper diagonal. For the Pearson correlation, the two measures of cumulative abnormal return (CAR [−1, 1] and CAR [−2, 2]) are significantly and negatively correlated with provincial continued increasing public health threats (CIPHT), which is consistent with H1 that increasing provincial threats positively affect the risk assessment of the investor. Given that the results in Table 3 are pairwise univariate correlations, we focus the primary analyses based on the multivariate tests in the next section. For multivariate tests, we calculate the VIF for each variable in the regression model (2) and find that all VIF values of variables, exclude the fixed effects, are less than 4. Thus, our multivariate analyses are not subject to a multicollinearity problem.