Purpose: Using a dataset of Taiwanese listed firms from 2002 to 2015, we empirically examine the determinants to voluntarily appoint independent directors.
Design/methodology/approach: We employ the panel estimation to exploit both the cross-section and time-series nature of the data. Further, we use Tobit regression, generalised linear model in the additional analysis, as well as the two-stage least squares to mitigate for a possible endogeneity issue.
Findings: The main findings show that Taiwanese firms with large board size tend to voluntarily appoint independent directors, and firms that already have independent supervisors more willingly to accept additional independent directors onto the board. Furthermore, ownership concentration and institutional ownership are positively associated with the voluntary appointment of independent directors. On the contrary, firms controlled by family members are generally reluctant to voluntarily appoint independent directors.

Research implications/limitations: Our findings are important for managers, shareholders, creditors and policy makers. In particular, when considering the determinants of the voluntary appointment of independent directors, our results indicate that independent supervisors, outside shareholders and institutional investors are significant factors in influencing effective internal and external corporate governance mechanisms. This research work on focuses on the voluntary appointment of independent directors. It would be interesting to compare the effectiveness of voluntary appointment with mandatory appointment within Taiwan as well as with other jurisdictions.

Originality/value: This study incrementally contributes to the corporate governance literature in several ways. First, this study extends the earlier research by using a more comprehensive dataset of non-financial Taiwanese firms, as well as employing alternative methodologies to investigate the determinants of voluntary appointment of independent directors. Second, prior studies tend to neglect the possible issue of using a censored and fractional dependent variable, the proportion of independent directors, which might yield biased and inconsistent parameter estimates when using OLS regression estimation. Finally, this study addresses the relevant econometric issues by using the Tobit, generalised linear model as well as the two stage least squares for a possible endogeneity concern.