*3.3. Models*

This study used a fixed-effects panel regression model to test our hypothesis. The dependent variable was treated with a one-period lag to address potential endogeneity due to reverse causality. Therefore, the following regression models were used to test the effect of the executive's environmental protection background on green innovation, the moderating role of media attention and board independence. In addition, we performed data analysis based on stata15 statistical software, using a fixed-effects model commonly used in the previous literature for testing [67]:

$$\text{GRI}\\
 mno\_{i,t+1} = a0 + a\_1 E P\_{i,t} + a\_k \sum \text{Control}\_{it} + \sum \text{Ind} + \sum \text{Year} + \varepsilon\_{i,t} \tag{1}$$

$$\text{Gignuo}\_{l,l+1} = \beta\_0 + \beta\_1 \text{EP}\_{l,l} + \beta\_2 \text{MA}\_{l,l} + \beta\_3 \text{EP}\_{l,l} \times \text{MA}\_{l,l} + \beta\_4 \text{BI}\_{l,l} + \beta\_5 \text{EP}\_{l,l} \times \text{BI}\_{l,l} + \beta\_k \sum \text{Contrad}\_{l,l} + \sum \text{Ind} + \sum \text{Year} + \varepsilon\_{l,l} \tag{2}$$

where *i* and *t* denote firm and year; GRInno is the level of green innovation, respectively; EP is the executive's environmental protection background; *MA* and *BI* are the moderating variables, which refer to the media attention and board independence; and ∑*Industry* and ∑*Year* represents industry fixed effect and year fixed effect, respectively.
