The Heterogeneous Impact of Market Size on Innovation: Evidence from French Firm-Level Exports

GAICIL Insights

Open a bigger market and every firm sells more.¹ Only some of them start inventing. The Heterogeneous Impact of Market Size on Innovation: Evidence from French Firm-Level Exports, by Philippe Aghion, Antonin Bergeaud, Matthieu Lequien and Marc J. Melitz, tests the long-standing prediction that market size spurs inventive activity, at the firm level and with unusual precision. Using three exhaustive datasets covering production, exports and patenting in French manufacturing, the authors build firm-level export demand shocks responding to conditions in a firm's export destinations, exogenous to its own decisions.

The theory survives, with a sharp twist. On average, firms respond to a larger market by patenting more, with the response arising several years after the shock, reflecting the time innovation takes. The positive impact on innovation is entirely driven by firms with above-median initial labor productivity within their sector. The demand shock raises contemporaneous sales and employment for all firms regardless of productivity. Everyone gets busier; only the leaders get smarter.

The model explains the skew with few additional restrictions. In a framework of endogenous innovation and competition with firm heterogeneity, more productive firms gain more from innovating when markets expand because they can better exploit the larger market. Less productive firms expand production without investing more in innovation. Demand is a fertilizer that works only on certain soil.

The authors are, to their knowledge, the first to identify a causal impact of firm-level market size on innovation independent of sector-level dynamics, controlling for arbitrary sector-level year-on-year changes, and widespread across manufacturing. The detailed timing of demand changes allows precise measurement of the lag before patenting responds.

Now the AI angle. AI markets are global, and market access shapes incentives to invest in AI innovation. The paper implies that expanding access spurs innovation, with the response concentrated among the most productive firms. In AI, where leading firms invest heavily in research, larger markets may amplify their innovative effort. Scale rewards those already able to use it.

That concentration raises a hard question. If market expansion mainly spurs innovation by leaders, it may widen gaps between leaders and laggards, consistent with concerns about rising concentration in technology markets. Policies expanding market access should be paired with policies helping less productive firms build the capabilities to innovate, so benefits spread. Trade openness and capability building are two halves of one strategy.

The lag matters for evaluation. Innovation responds to demand with a delay, reflecting the time to conduct research and secure patents. Evaluators of trade or market access policies should expect innovation effects to emerge over years. AI development cycles can be rapid, so the lag may differ, while the principle that innovation follows demand holds. Judging a trade deal's innovation effect after one year is premature.

Trade policy is directly implicated. Export demand shocks raised innovation among productive firms, suggesting open trade and foreign market access support innovation. Restrictions shrinking effective market size could reduce incentives to innovate, particularly for the most capable firms. In AI and related technologies, where trade restrictions have grown more common, this evidence is a valuable reminder of what market access buys. A tariff on an AI firm's customers is also a tax on its research budget.

The method is a model for innovation research. Building exogenous demand shocks from conditions in export destinations separates the effect of market size from the many factors jointly driving innovation and sales. Similar approaches could study how demand for AI applications affects AI innovation.

Economic nationalists might start here: shrinking the market your best firms can reach may slow their inventing years before it shows up anywhere. And widening the market without lifting the laggards may simply enlarge the leaders' lead.