4.2.2. Placebo Test

We further perform a placebo test to exclude the effect of the pilot CET policy on a firm's market value from the interference of other non-observable omitted variables. This involved random selection of certain regions as virtual "pilot regions" to enable the comparison of the differences of the effects between the real treatment group and the randomly generated group [61]. More specifically, if there are n firms in 2013 located in the area where the pilot CET policy is launched, keeping the time of the pilot CET policy constant, we randomly select n sample firms from the entire sample of firms as the

treatment group to conduct a counterfactual test. We repeat 500 estimates based on the benchmark regression results in column (3) of Table 4. Figure 3 illustrates the probability density distribution of the regression coefficients for the placebo test. Based on random samples, the estimated coefficients are distributed centrally around 0, while the benchmark regression result (0.016) falls almost outside the possible range estimated from the virtual pilot regions shown in Figure 3. Hence, the results indicate that the pilot CET policy has no policy effect when randomly set up. Therefore, the placebo test reveals that the benchmark regression results of the DID method are reliable.

**Figure 3.** Results of the distribution of the DID estimator for the placebo test.
