Where’s the risk in that?

A new study shows that German grants are too wary

Great comic by Tom Fishburne showing the proclivity for risk in business and innovation. Source: https://marketoonist.com/2024/08/safeisrisky.html

Germany funds research generously. The federal government supplies roughly a third of the country’s R&D spending and disburses much of it through a project-funding apparatus with around 40,000 active grants. A new analysis looked at what is funded with the money. Overwhelmingly, it is spent on the firms and the technologies Germany already has.

Companies older than 20 years capture about 64 percent of the funding. Large firms with more than 250 employees take roughly 65 percent of the money while making up only a small share of the recipients. Young firms receive about 10 percent. The funding also clusters in the middle of the technology-readiness scale: 64 percent of supported projects sit at TRL 4 to 6, the validation-and-demonstration band, while early-stage research at TRL 1 to 3, which is for testing the idea with first experiments, is barely represented. At the top of the recipients table sit Bosch, Zeiss SMT, SAP, and BMW. The study found a cute name for the pattern: the middle-technology-trap. It traces a path dependency in which firms that already know how to win grants keep winning them, while newcomers, who the evidence says contribute disproportionately to radical innovation, are screened out.

This is similar to the pattern I wrote about earlier on the applied research apparatus in Germany. Fraunhofer, the applied-research arm closest to industry, functions as a mirror of Germany’s existing industrial composition, because it must earn 70 percent of its budget from the firms it serves, and those firms want a combustion engine two percent more efficient, not a new paradigm. The obvious reply to that argument was always that the state holds the other lever. Fraunhofer follows the market by design, but the federal government funds basic research and steers project money, and the state can be countercyclical. It can buy the frontier that the market neglects.

But here we see that it doesn’t work through the grants. The state’s own project funding is procyclical too which means it mirrors the same industrial base, backs the same incumbents, and rewards the same incremental work.

This means Germany’s innovation system has no circuit breaker. Max Planck selects frontier fields but defines itself by its distance from application and cannot hand off its discoveries. Fraunhofer reaches the market but is pulled toward whatever its industrial clients are willing to pay for. And the federal funding catalogue, which sits between them and could in principle correct for both, instead reproduces the bias.

Every instance of German R&D either fails to reach the frontier or is structurally drawn back toward the position the country already occupies. The same risk aversion in the academic career system, where short-term contracts discourage ambitious research, also governs how the state funds firms.

The trouble is that the frontier science increasingly does not exist on German soil in the fields that matter most. The Max Planck Society fell from fourth to eleventh in the Nature Index between 2021 and 2025, and its chemistry ranking dropped from the top five to fourteenth. From 2021 to 2024, German-affiliated authors appeared on 32 of the 100 most-cited AI papers, compared with 218 for the United States and 142 for China. A funding system designed to back new entrants and early-stage firm research would, within a few years, leave German firms far better equipped to absorb external knowledge. It would not change what there is to absorb. Absorptive capacity presupposes a frontier to draw from, and in AI, chemistry, and physical and materials science, Germany cedes its former excellence.

So the country faces two failures, and they run on two different clocks. The failure the study found is a funding allocation problem, fixable by rule changes within an electoral cycle. The production failure my previous essays have documented is slower and deeper, because you cannot conjure frontier output, or the scientists who generate it, on a budget cycle. No number of grants to frontier science within firms will be enough to counterbalance weak public research at the frontier. The study suggests that a small-firm quota, stronger early-phase funding, and sharper prioritization of strategic technologies are the right instruments. These policy changes are necessary, but not sufficient.

The German innovation system is underambitious at every link, and the policy debate keeps discovering new links to blame: the startups, the stock options, the venture funds, and now the funding catalogue, while leaving the source of the chain untouched. The state, the one actor positioned to fund what the market will not, funds what the market already does.