**5. Research design**

#### **5.1 Samples**

Our sample consists of all the commercial banks, 30 in this case, listed on Dhaka Stock Exchange (DSE). We hand-collected necessary information on CSR spending by banks through the CSR reports published by Bangladesh Bank covering the period 2007–2020. Simultaneously, we checked the annual reports of sampled banks to check for any inconsistencies in CSR disclosures. For firmspecific data, we studied the annual reports of selected banks and obtained the required information.

#### **5.2 Board structure and control variables**

We used four board elements: Board size, board independence, board gender, and frequency of board meetings as our main variables of interest. Board size was measured by the total number of directors on the board. Board independence represented the number of independent directors relative to the board size. Board gender was indexed by the percentage of female members on the board. Finally, the frequency of board meetings was measured by the number of meetings held annually. Also, we adopted some control variables, such as firm size, leverage, profitability, and firm age, following previous literature to get robust estimates [26, 49, 50]. We proxy for firm size with the total assets (*TA*), leverage with the liability to asset ratio (*LEV*), profitability with the return on assets ratio (*ROA*), and firm age with the number of years from the year of listing to the reporting period in our study. Besides, we controlled both year and firm effects in the OLS model. **Table 2** presents the definition and acronyms of all the variables used in the study.

#### **5.3 Regression model**

We used the following regression model to examine the relationship between board structure and CSR spending by banks.


**Table 2.**

*Definition of variables.*

#### 0 1 2 3 4 56 7 8 CSREXP BDSIZE BDIND BDGEND BDMEET =β +β +β +β + β +β +β +β +β +ε *it it it it it TA ROA LEV AGE it it it it it*

where CSREXP is the corporate social responsibility (CSR) expenditure, BDSIZE is the size of the board, BDIND is the board independence, BDGEND is the board gender, BDMEET is the number of total board meetings, TA is the total assets, ROA is the return on assets, Lev is the gearing based on debt-to-equity ratio. εit is the disturbance term, and β1… β8 are the coefficients of the variables.
