128 / 274
以實質選擇權觀點探討探索性與利用性活動對公司績效之影響：中介效果模型
128
Although our results provide support for our arguments, some caution is warranted as
we study just a single industry. Although we consider that the semiconductor industry is the
best choice for revealing exploration and exploitation activities under uncertainty, our results
might not be generalizable to other industries. For example, our binomial lag methodology
identified an optimal exploration lag pattern on a firm’s performance, yet one must be careful
in applying this pattern to other industries.
Another limitation of this study is that we cannot match each case of exploitation with
previous exploration because of the non-additive nature of exploration activities (Vassolo et
al., 2004), that is, the benefits of these activities interact with subsequent exploitation to
affect firm performance. Because the purpose of our study was to highlight the sequential
decisions involving exploration and exploitation, matching specific instances of exploration
and exploitation was not our major concern. Future research might use a different data
structure (e.g., a patent database) to examine the influence of the quantity and quality of
exploration on exploitation.
Furthermore, merely categorizing exploration as real options may not be consistent with
the real options reasoning used in initial investment (Adner, 2007). However, we believe that
conceptualizing exploitation as striking options somewhat resolves the concern for the
“impossibility of proving failure” (Adner and Levinthal, 2004).
Despite these reservations, our study makes several theoretical and empirical
contributions to our understanding of exploration and exploitation. First, we broadened the
use of real options reasoning by integrating ROR with classical exploration-exploitation
arguments (March, 1991). Second, we resolved the debate between the advocates of the
crowd-out and alternative enhancing hypotheses on exploration and exploitation. By
reconceptualizing exploration and exploitation as creating and striking options in two
functional domains, we uncovered the relationship between exploration and exploitation and
also the sequences for these two kinds of activity.
In addition, this study provides new empirical evidence on exploration and exploitation.
Specifically, it provides empirical support for an alternative enhancing relationship in the
context of R&D resources management. Few previous studies have provided direct empirical
evidence for balancing these two activities, although the benefits of doing so are commonly
recognized. Moreover, it should be kept in mind that investment decisions involving resource
allocation drive the direction a firm takes in its development activities, and they influence the
source of the returns. Based on longitudinal data in the semiconductor industry, our findings