“…Control variables. This study controlled for the following variables (Balkin et al , 2000; Artz et al , 2010; Chen, 2014; He and Shen, 2019; Jia et al , 2019; Iqbal et al , 2020): firm size , measured by the natural logarithm of the total number of employees; firm age , measured by the number of years from the firm establishment date to the observation date; capital structure , measured by the ratio of total liabilities to total assets at year's end; ownership , if the firm is a state-controlled firm, the value is 1, otherwise it is 0; assets turnover , measured by the ratio of sales revenue to the average of beginning total assets and ending total assets; intangible assets , measured by the ratio of intangible assets to total assets; subsidies , measured by the natural logarithm of the amount of government subsidies (including fiscal appropriations, fiscal interest subsidies and tax rebates); knowledge capital , measured by the ratio of the number of science and technology personnel to the total number of employees; R&D , measured by the ratio of the R&D spending (including expended and capitalized R&D spending) to the total assets of the firm; attainment discrepancy , measured by the extent to which the firm's actual performance falls short of the aspiration, the actual performance is measured by ROA and the level of aspiration is measured by ROA of the previous year; TMT size , measured by the total number of members on the TMT; market competition , measured by the Herfindahl-Hirschman Index, which reflects the market shares of listed companies in the industry. To control endogeneity, the independent variables, moderating variables and control variables were all those of period t , while the dependent variable was that of period t +1.…”