

臺大管理論叢
第
27
卷第
2
期
25
have weaker performance when they have higher proportions of directors serving as central
government officers or as legislative body representatives. Second, we test whether the
documented association varies with transitions between ruling parties or with an individual
director’s political affiliation. We find that firm performance is negatively related to directors
with a KMT affiliation but is not related to directors with a DDP affiliation. In addition, this
negative association is moderated when the ruling party switches from KMT to DDP. Third,
to resolve the potential endogeneity of political connections, we test our main model using a
two-stage instrumental variable approach as well as a difference-in-difference design (DID).
The DID design considers the changing status of the board of director’s connections after
election of new directors. Regardless of the approach taken, the findings are qualitatively
similar. Fourth, we consider firms’ differential incentives of having political connections,
including their future growth opportunities, financing needs, and preferences for tax
avoidance. We partition the sample based on these incentives and find that the negative
association between firm performance and political connections are consistent across
subsamples with high and low incentives. Finally, we find that our results exist both before
and after the 2008 financial crisis, after removing government-owned enterprises, and that
the results are not sensitive to directors’ level of professional knowledge.
4. Research Limitations/Implications
Using Taiwanese sample firms as the context for research, we provide consistent
evidence that political connections are negatively associated with firm performance. The
findings are informative for capital market participants in making their investment and
lending decisions. The results also provide implications for firms’ corporate governance
practices. In particular, firms should consider the possible negative impact of political
connections when hiring or electing top managers and directors.
However, our findings are subject to several limitations and should be interpreted with
some caveats. First, this study does not consider the potential differential effects of industrial
life cycles as well as the macroeconomic environment. Second, while we consider firms’
incentives to become politically connected, the analysis may be incomplete. For example, we
do not include firms that use political connections to conduct fraud or to avoid regulatory
investigations. Third, due to data limitations, our measures of political connections include
only those available from public sources.