Page 72 - 臺大管理論叢第33卷第1期

Value Creation and Capture in Developing Countries: The Driver and Mechanism of Offshore Outsourcing
Innovation

p-value < 0.1). The positive coefficient of the interaction term indicates that the design of
project modularity can increase the effect of low-cost talent on the likelihood of choosing
the country with weak IPR protection for innovation outsourcing. As such, Hypothesis 3 is
supported.

4.2 Additional Robustness Checks
A long-standing conundrum in the research on internationalization of innovation

is why firms differ in their decisions and subsequent actions. The issue of endogeneity
arises when such decisions are associated with firms’ abilities. In such situations, ordinary
regression analyses and their results are likely biased (e.g., Chen and Hsiao, 2013; Lo
and Hung, 2015). To check upon the estimation of Hypothesis 1, we conduct an ordinary
regression to show whether bias might exist. In Model 5 of Table 3, specifically, we can

see that the main effect of human capital on the entry of developing countries disappears if
sample selection bias is not corrected by including the inverse Mills ratio. Thus, we believe
that the two-stage modeling approach used in this study is appropriate. Furthermore, we

account for the potential endogeneity of entry choices by using a two-stage least squares
(2SLS) regression. To identify the first stage of our 2SLS regression in Model 6 of Table
3, we use the instrument variable of enhancing capacity for innovation (Lo and Hung,
2015). We expect firms with the motive of enhancing capacity for innovation to pursue
more outsourcing activities but did not expect that enhancing capacity for innovation

affects the focal firm’s preference on location choices. Model 6 reports the results of the
second-stage 2SLS regression (the first-stage regression is not shown but available upon
request), showing that the previously estimated effects remain the same. Furthermore, we

run additional multinomial logistic regressions with the analysis of marginal effects to
inspect the estimation of Hypothesis 1. The estimated coefficients represent the utility of
choosing very strong IPR protection countries (i.e., IPR values greater than the mean plus
two standard deviations), strong IPR protection countries (i.e., IPR values greater than the
mean plus one standard deviation but smaller than the mean plus two standard deviations),

or weak IPR protection countries (i.e., IPR values smaller than the mean plus one standard
deviation). The robustness checks support the previous findings. More specifically,
the availability of low-cost talent decreases the likelihood of outsourcing innovation

activities to very strong IPR protection countries (β = -0.04, p-value < 0.01) and strong

64