Literature Review Women on boards and in TMTS and firm performance 47 least squares 3SLS regression by Carter D Souza Simkins and Simpson 2010 Carter et al 2010 use single OLS regression equation and a 3SLS regression analysis both accounting for firm and time fixed effects 3SLS estimation is considered to be advantageous compared with 2SLS becau se it accounts for both potential endogeneity and cross correlation between equations Carter et al 2010 Carter et al 2003 consider other explanatory va riables such as firm size board size CEO chair du ality or the percentage of insiders on the board and the approximation of Tobin s Q in their 2SLS model Bøhren and Strøm 2010 examine the interaction of the four board mechanisms gender mix emplo yee directors director independence and multiple directorships with firm value Amongst others they control for the effects of firm characteristics inclu ding unobservable fixed and random effects and of potential endogeneity between board mechanis ms and firm value They use instrumental variab les and 2SLS regression Endogenous variables they specify in the estimation in addition to performan ce are board independence CEO director exported CEO imported CEO board size gender and board age dispersion Two ownership variables network employee directors risk and firm size constitute the exogenous variables Bøhren Strøm 2010 Huang and Kisgen 2013 use a difference in diffe rences approach in order to mitigate endogeneity issues comparing activity before and after tran sitions from a male to a female executive with a control sample of male to male transition firms Panel data regressions with firm fixed effects with a female executive dummy variable are conduc ted in addition as a robustness check Chapple and Humphrey 2014 handle the endogeneity prob lem through forming and comparing portfolios of firms with gender diverse boards to those without Portfolio formation is the method of choice as the interest is focused on the market level impact of gender diversity They apply OLS and firm fixed ef fects regression followed by an Arellano and Bond dynamic panel model The authors further state that the firm specific characteristics were averaged out by applying this approach and the precision of estimates from regression analysis was improved The authors claim that the heterogeneity issue was thus eliminated and the omitted variables problem reduced Chapple Humphrey 2014 Deszö and Ross 2012 handle the issue of reverse causation by controlling for prior firm performance through adding lagged values of Tobin s Q to the regres sion If the positive association between female leadership and firm performance was driven by re verse causality it should then disappear However two new problems may emerge adding lagged values to a panel data regression may result in difficulties with autocorrelation and other control variables related to firm policies might also be en dogenous Deszö Ross 2012 Deszö and Ross 2012 as well as further recent studies Campbell Minguez Vera 2010 Chapple Humphrey 2014 He Huang 2011 Liu et al 2014 try to over come these challenges by using generalized me thod of moments GMM estimators as proposed by Arellano and Bond 1991 Arellano and Bo ver 1995 Blundell and Bond 1998 or Wintoki 2012 Strøm et al 2014 address the problem of reverse causation of female leadership in financial performance regression by applying the Heckman 1978 model for an endogenous dummy variable They solve the sample selectivity problem by the inverse Mill s ratio IMR test Women on boards and in TMTs and firm performance

Vorschau DIRK-Forschungsreihe Band 21 Workforce diversity and personal policies Seite 47
Hinweis: Dies ist eine maschinenlesbare No-Flash Ansicht.
Klicken Sie hier um zur Online-Version zu gelangen.