Strategic Autonomy vs. Race to the Bottom: Self-Defeating Industrial Policy in the Presence of Network Externalities

Nigar Hashimzade and Haoran Sun

Abstract

Industrial policy has returned to the centre of economic governance, particularly in sectors like semiconductors and AI where network externalities make market dominance self-reinforcing. This paper develops a two-country model to examine the strategic trade-off between R&D subsidies and foreign-firm taxation. We show that while network externalities initially strengthen the case for a ``carrot-and-stick'' policy mix, they also create a strategic trap: when domestic and foreign products are close substitutes, non-cooperative industrial strategies become counter-productive. In this situation, aggressive domestic subsidies intensify rent-dissipating competition, while a tax on the foreign rival serves as a more efficient instrument for profit-shifting. We find that the resulting subsidisation race can reduce aggregate welfare, particularly when network effects are moderate and rivalry is intense. The results suggest that for sectors with strong consumption externalities, unilateral industrial strategies may be self-defeating unless carefully calibrated to the degree of product differentiation.

Why this matters 

In industries like semiconductors, AI, and electric vehicles, success breeds more success: the more people use a technology, the more valuable it becomes. Governments are now competing to lead these industries using a combination of support for their own firms (the "carrot," such as R&D subsidies) and penalties on foreign rivals (the "stick," such as taxes or trade restrictions). This paper shows that the carrot-and-stick combination can work when countries' products are sufficiently different. But when products are similar, the same policy backfires: both governments spend more while neither gains a lasting advantage, leaving everyone worse off.

The model opens several directions for future research, including whether falling behind in these markets becomes permanent over time, whether new technology generations (such as next-generation chips or AI models) create windows for trailing countries to catch up, and how international cooperation on subsidy rules should be designed when these self-reinforcing dynamics are present.

On the empirical side, I am developing a programme of work using ONS firm-level microdata to test whether the productivity gains from AI adoption and public procurement are in fact larger in markets with these self-reinforcing properties, connecting the theory directly to evidence for better-targeted UK industrial policy. This is ongoing research.

Manuscript available on request