Page 81 - 臺大管理論叢第32卷第2期

NTU Management Review Vol. 32 No. 2 Aug. 2022

observe how firms conduct exploitative and exploratory learning. Specifically, I identify
which smartphone manufacturers the commits belonged to by email. For example,
xxxxx@nokia.com is regarded as Nokia’s commit record, and xxxxx@samsung is
Samsung’s commit record.

Second, the period from the old- to the new-generation Android OS is about a
year. Upon releasing the old-generation Android OS to the public, Google will continue
to release the old-generation Android OS as the maintenance version because it solves

software problems or improves product performance. Moreover, six months before the
official release date, the new-generation Android OS is released as the developer version.
Based on above-mentioned release condition, I distinguish the period of exploitative
and exploratory learning carried out by smartphone manufacturers in the generational
transition of the Android OS (See Figure 1). In addition, I calculate firms’ contribution in

the hardware folder under the platform folder in the maintenance version, and in the sub-
folder under the platform folder in the developer version as two independent variables for
the two types of learning.

Third, this study calculates the elapsed time between the dates of a smartphone
manufacturer first releases a new version of an Android smartphone, and the releasing dates
of Google’s Nexus or Pixel after each new version of the Android OS is released. This
elapsed time is regarded as time-to-market, the primary dependent variable. In addition,
this study counts the number of models launched by the 11 smartphone manufacturers in

each Android version from the PhoneDB database as the second dependent variable. This
study also controls for the firms’ heterogeneity in their new product developments, such as
hardware components, nationality, and relationship with Google.

4. Findings

I use the ordinary least squares method for multiple regression analysis. I find that the
origin of the smartphone manufacturer, the CPU chip manufacturer, and display resolution

all do not affect the time-to-market. Meanwhile, cooperation between smartphone
manufacturers and Google has a significant positive relationship with time-to-market.
Moreover, the exploratory learning of the new generation platform technologies and the
exploitative learning of the old-generation ones significantly impact the time-to-market.

73