Page 38 - 34-3

Divestiture from China by Taiwanese Listed Electronic Information Firms: Effects of Host Country
Performance and Selection Bias

5. Originality and Contribution

The present study has three main contributions. First, it applies firm behavior theory
and prospect theory to investigate the association between host country performance and

divestiture decisions. It uses a spline function to assess the effects of gains and losses
relating to host country performance on divestiture decisions. By employing spline
functions, the present study increases the number of control points in the generated curves,
thereby enabling observation of irregular curve associations (Tseng and Lai, 2011).

Second, although most studies have identified divestment as a strategy motivated by poor
host country performance (i.e., a remedial measure) and, consequently, reported a negative
or nonsignificant association between divestiture and performance, recent research has
helped clarify the strategic role of divestment (Belderbos, Tong, and Wu, 2020; Dai,

Eden, and Beamish, 2017; Sakhartov, 2018; Tan and Sousa, 2019). If the divestment of
Taiwanese listed electronic information firms from China is strategically significant, a
positive association between divestiture and performance is likely present. The present
study investigates the association between divestiture and performance, identifying the

strategic relevance of investment performance in China to the decisions of Taiwanese
listed electronic information firms to divest from China. Finally, ignoring selection bias
in empirical analysis can result in performance models producing nonrobust estimates
(Leiblein et al., 2002; Shaver, 1998; Tseng and Kao, 2011). To mitigate the effects of

self-selection and sample selection bias, the present study employs two methods, namely
Heckman’s two-stage least squares method and a quasi-experimental method in which
divestiture and non-divestiture cases are separately modeled (Heckman, 1979). Through
empirical testing and through the inclusion of correction factors (i.e., Inverse Mill’s Ratio)

and the probability of decision error variables, the present study provides a solution for
reducing sample selection bias and self-selection bias in empirical strategic research.

28