where SM is the firm's SM score, size is LNSALES, financing constraints are measured by CASH/TA and TD/TA, performance is measured by return on assets (ROA) and total return on shareholders (TRS), growth opportunities are measured by Q and DESM = [[epsilon].
The correlations between the two measures of DESM (IASM) and SM are significant.
Analysis of the Effect of Internal Governance Mechanisms on the DESM
where DESM is a function of managerial ownership incentives (LNDELTA), board monitoring (BRDSIZE, PIND, DUALITY), institutional ownership (INSTOWN), controls (CASH/TA, TD/TA, LNSALES, ROA, TRS), and dummy variables for year.
While a negative value of DESM (a negative residual) suggests an underinvestment in SM, a positive value of DESM (a positive residual) implies an overinvestment.
Columns 7 and 8 include estimated results for regression Equation (3) replacing DESM with IASM as a proxy for the deviation from expected investment in SM.
First, to ensure that causality runs from internal governance to DESM, we reestimate regression Equations (3) and (4) replacing contemporaneous internal governance variables with their lagged values and also using contemporaneous internal governance variables while including a lagged dependent variable in our regression equation.
Our analysis to this point has treated the internal governance variables as exogenous, but if any one of these variables is determined simultaneously with DESM, it violates the least squares regression assumption that the regressors are uncorrelated with the error term.
For example, the negative and significant estimated coefficients for DESM in Columns 2 and 3 suggest that there is an inverse relation between deviations from expected SM and the level of CEO incentives and board independence.
Table IV reports the results of FD3SLS estimates of the system of four Equations (5)-(8) using the consumer and industrial subsamples and replacing DESM with the deviation from expected investment in managing stakeholder strengths (DESMSTR) or concerns (DESMCON), or deviation from expected investment in social issue participation (DESIP).
Table V presents the results of the FD3SLS estimates of the system of four Equations (5)-(8) using the consumer and industrial subsamples and replacing DESM with measures of deviation from expected investment in the five primary stakeholder dimension categories of: 1) employee relations (DEEMP), 2) diversity issues (DEDIV), 3) product issues (DEPRO), 4) community relations (DECOM), and 5) environmental issues (DEENV).