Innovation and Change in the Process of Alliance Formation in the Japanese Electronics Industry

This paper examines the changing process of strategic alliance formation in the Japanese electronics industry between 1985 and 1998. With data on 128 Japanese electronics/electrical machinery makers, we use a dyad panel regression methodology to address a series of hypotheses drawn from embeddedness and strong/weak tie theory on how keiretsu and prior alliance networks have constrained partner choice in new R&D and nonR&D alliances. We argue and find that the keiretsu effect is smaller on R&D than nonR&D alliances, and that this is truer of the "weaker-tie" horizontal keiretsu than the "stronger-tie" vertical keiretsu. Dividing our time series into four periods (1984-88, 89-90, 91-94, 95-98), however, reveals some important variations in the keiretsu role over time. The horizontal and vertical keiretsu effects on R&D alliances had vanished by 1991-94 (the post-bubble recession era), but they continued in the nonR&D case, in part, we believe, because these provided a means of reducing costs and capacity in a stringent macroeconomic environment. Following previous strategic alliance research, we further examine how the prior alliance network conditioned strategic alliance formation in Japanese electronics and how those patterns varied over time. The data suggest that, as the strategic alliance founding process became "disembedded" from Japan's legacy keiretsu networks, it was driven increasingly by prior direct and indirect alliance ties.