The initial sample used in this study comprises 15,238 observations for A-share firms listed on the Shanghai
and Shenzhen stock markets. We screened the sample as follows: (1) we remove 2541 observations for firms
that had been listed for less than 2 years; (2) we remove 119 financial companies; and (3) we remove 1881
observations with missing values. A total of 10,724 observations remain. To reduce the influence of extreme
values, we exclude the top and bottom 1% of Roe and Ret values, which leaves 10,321 observations. The data
used is from the CSMAR and CCER databases. We use SAS and STATA software to process the data. The
procedure for sample selection is listed in Table 1.