An Evolutionary Model of Changes in Industrial Leadership and Catch-up by Latecomers: Collective Learning-by-doing, Endogenous Innovation, and the "Incumbent Trap"

Abstract

When a radical technological innovation arrives in an industry, incumbents tend to fall behind because they are locked-in to the established technology that they have mastered. Meanwhile, latecomers adopt the new technology. As a result, industrial leadership tends to shift from incumbents to latecomers as new technology dethrones the established one. This tendency is termed as the incumbent trap or incumbents curse in the literature. The incumbent trap phenomenon has been observed repeatedly in the history of leadership changes in various industries.

However, this tendency is not universal. There are many counterexamples in which leadership doesnt shift from incumbent leaders to latecomers or followers after a radical technological change. The literature has insufficiently identified the boundary conditions for the incumbent trap phenomenon, which has resulted in controversy and confusion in the literature. Especially, the incumbent trap research has largely unaddressed collective learning-by-doing effects on the incumbent trap dynamics. In many industries, firms within the same technology camp benefit from externalities of other firms learning-by-doing, which I term collective learning-by-doing. The present study identifies boundary conditions for the incumbent trap dynamics in which collective learning-by-doing plays a role.

For this purpose, the present study focuses on two factors: 1) the basic productivity of new technology and 2) the expected proportion of firms adopting the new technology. I argue that industrial leadership is more likely to shift from incumbent leaders to followers or latecomers if the basic productivity of new technology is neither too high nor too low (Hypothesis 1). In addition, I argue that leadership change is more likely to occur if the two factors have opposite values: the basic productivity of new technology is high (low) and the expected proportion of firms adopting the new technology is low (high) (Hypothesis 2).

To support my theoretical arguments, I build an agent-based simulation model in an evolutionary economics perspective. Two situations are modeled. In the basic model, a radical innovation is exogenously given to firms at a certain period of time, and afterwards in each period, each firm decides whether to adopt this innovation or not in its search for more profit. Their adoption decisions are based on the comparison of the expected productivity of their currently employing technology and that of the new technology. Adding to this setting, an extended model allows for firms to do R&D so that the firms can improve the productivity of the technology they are employing, which I term as endogenous innovation.

Hypotheses 1 and 2 are supported both in the basic and the extend models. In the extended model, the probability of leadership change decreases overall, compared with those of the basic model, in which innovation are only exogenous to firms. This is because incumbent leaders have superior R&D capabilities to followers, which helps incumbents defend their leadership.

The present study makes three contributions. First, it links the incumbent trap phenomenon to externalities caused by firm's learning-by-doing, or collective learning-by-doing. Second, I demonstrate numerically that the initial gap in productivity between old and new technologies affects the probability of changes in industrial leadership from incumbents to latecomers. Third, the present study demonstrates in a formal model R&D effects in the incumbent trap phenomenon.