The System That Reproduces Itself

Germany spends more than 20 billion euros of public money annually on research and innovation. The question of whether this money is well spent is rarely asked — at least not publicly, at least not with numbers. This is not because the question has no answer. It is because the system is structurally built in a way that makes the question difficult to ask.

I. The Architecture

The German research and innovation system consists of at least five overlapping layers that cross-finance each other and present a unified front to policymakers. At the top stand the four large non-university research organisations: the Max Planck Society, the Helmholtz Association, the Fraunhofer Society and the Leibniz Association. Below them operate funding organisations such as the German Research Foundation and a dozen project management agencies. At the base — or rather at the margins — sit innovation agencies at federal and state level: SPRIN-D, Bayern Innovativ, the Innovation Agency Baden-Württemberg and their counterparts in every other federal state.

Together these institutions manage budgets totalling well over 20 billion euros annually. Almost exclusively from public funds. Almost exclusively with guaranteed annual increases: since 2005 the Pact for Research and Innovation has secured the major organisations annual budget growth — first five percent, then three percent. Independent of results. Independent of economic conditions. Independent of whether the work bears fruit.

This is the first structural anomaly: in Germany, research organisations are not funded by performance but by membership. Whoever is once part of the system remains in it — and grows steadily within it.

II. The Numbers That Exist

This table reveals the fundamental problem. The higher the public share, the weaker the measurable output — and the more diffuse the metrics with which institutions describe themselves. Fraunhofer is the most honest construction: 30 percent basic funding from the state, 70 percent from the market. Whoever fails to attract industrial contracts receives less state money. That is a genuine, built-in quality test.

Helmholtz is the opposite: 90 percent state-funded, 46,000 employees, six billion euros — and the primary success indicator is publication counts and the operation of large-scale facilities. This is not a criticism of the science conducted there. It is an observation about the ratio of input to demonstrable social benefit.

III. The Fraunhofer Paradox

Fraunhofer is the flagship of the German system — and on closer inspection also its most instructive symptom. The industry revenues of 867 million euros sound impressive until one sets them in proportion: against a total budget of 3.6 billion euros and 32,000 employees, the industry share amounts to barely 24 percent. Three quarters of the budget comes from public sources.

More revealing: Fraunhofer registered 507 inventions in 2024 and filed 439 patent applications. Licensing revenues from intellectual property outside the industrial sector amounted to two million euros. This means: thousands of patents produced — and the direct economic return from intellectual property is negligible. The transfer does take place, but it takes place invisibly, embedded in contract research that is not invoiced by rights.

This is structurally problematic because it means: the most valuable output — the transferred know-how — is visible in no public number. Fraunhofer cannot demonstrate what it is worth. And whoever cannot demonstrate what they are worth also cannot have explained why they would not have that worth.

IV. The Pact as an Institutional Preservation Mechanism

The Pact for Research and Innovation, in force since 2005 and already extended three times, is the centrepiece of the German system — and its central structural flaw. It guarantees five organisations annual budget increases, independent of results, independent of economic conditions. In times of low inflation this sounds like planning security. In times of six percent inflation it means real purchasing power loss with nominally rising budgets — and simultaneous displacement effects within the ministerial budget: what the major organisations receive as a guarantee is missing elsewhere.

Universities have felt this most directly. While the non-university research organisations received guaranteed annual increases, the core budgets of universities stagnated. Excellence grew at the top, the base eroded. This is not a side effect of the Pact. It is its logical consequence: a system that privileges existing institutions disadvantages all others.

The Pact for Research and Innovation is not optimised for impact. It is optimised for institutional preservation. That is the difference between an innovation system and an institution-protection system — and Germany has the second.

V. The Project Management Agencies: The Invisible Cost Centre

Rarely discussed, but structurally significant: the system of project management agencies. Projektträger Jülich, VDI/VDE Innovation + Technik, VDI Technologiezentrum, DLR as project manager, TÜV organisations as project managers — an entire ecosystem of institutions that administer funding programmes. They assess applications, accompany projects, handle settlement. They conduct no research of their own. They develop no strategy of their own. They bear no risk of their own.

For every euro approved as research funding in Germany, there is at least one — often several — administrative layers that themselves live from public funds in order to decide how public funds are distributed. No one has ever publicly calculated what running this administrative apparatus costs, measured against the money that actually reaches researchers. The figure exists. It is simply not published.

VI. The Innovation Agencies: Subsidised Hope

At the lower end of the system sit the innovation agencies — Bayern Innovativ, NRW.Innovation, the Innovation Agency BW and their counterparts in every federal state. Bayern Innovativ is exemplary: 40 million euros annual revenue, 300 employees, 32,000 "clients". The numbers sound like impact. On closer inspection they describe only input.

Bayern Innovativ publishes not a single output indicator. Not how many of the 32,000 contacts led to market-ready products. Not how much private capital was triggered by Bavarian networking activity. Not how many of the brokered partnerships still exist after one year. This is not an oversight. It is systemic logic: an institution that has no market test cannot know whether it is doing anything useful — and therefore selects input figures that always look good.

The result is what one might structurally describe as the "subsidisation of hope": money flows, activity arises, contacts are made — but whether an innovation ultimately emerges that would not have arisen without the agency remains systematically unclear.

VII. What Is Missing: The Market Test as a Structural Principle

The decisive difference in the system is not the size of the institutions, not their scientific quality, not their headcount. It is the presence or absence of a built-in market test — an instance that, independent of political goodwill, judges whether the output is actually demanded.

Fraunhofer has this test, imperfect but real: companies pay for contract research from their own budgets. The DFG has it, also imperfect: peer review means that researchers judge other researchers, which at least corresponds to academic quality standards. All other major institutions do not have it, or only in a substantially weakened form. Helmholtz measures its standing in publication counts and large-scale facilities. The Leibniz Association counts reports and media appearances. The state innovation agencies count event participants.

The structural problem is not that these institutions do poor work. The problem is that they have no way of distinguishing good from poor work — and therefore the system produces no incentives to improve. An institute that delivers poor results receives three percent more the following year regardless. An institute that delivers excellent results also receives three percent more. The Pact makes both equal.

VIII. The Excluded

The analysis above concerns the inside of the system. There is an outside that appears in no statistics: the innovators who have no access.

The German innovation system is oriented towards institutional actors — universities, research centres, accredited companies, consortia. The individual inventor, the lateral entrant, the medium-sized company without an academic apparatus behind it — they are not provided for in the architecture of the system. Funding applications presuppose academic institutions as partners. Project management agencies approve applications according to criteria tailored to institutionalised research. Innovation agencies broker contacts between actors who are already in the system.

This is the actual paradox of the German innovation system: it is constructed to promote innovations — but structurally prevents the most radical innovations, which are radical precisely because they arose outside existing institutions, from entering the system. A system optimised for institutional preservation also preserves the institutional actors — at the expense of all others.

The question is not whether Germany spends too much or too little on research and innovation. The question is whether what is spent closes the system or opens it. At present it closes it. It locks in those who are already inside — and excludes everyone else.

IX. What This Means

A system that reproduces itself has a characteristic property: it is very stable. Individual institutions can deliver poor results without being sanctioned. New actors can develop superior solutions without being rewarded. Resources flow into established channels — not because these channels are the best, but because they exist.

This does not mean that the system produces no innovations. It produces them — but not because of its structure, rather despite it. The major successes of the German research landscape arise in the spaces that the system does not fully control: in the autonomous institutes of the Max Planck Society, in the few Fraunhofer institutes that have genuine market relevance, in university laboratories that conduct excellent research despite shrinking core budgets.

A better system would not spend more money. It would distribute the existing money differently: according to impact rather than institutional membership, according to market test rather than affiliation, with real consequences for institutions that cannot provide evidence of their impact — and with genuine access for actors who are currently structurally excluded.

That would not be a revolution. It would be the application of the principle that has been valid in the Fraunhofer model since 1973 to the entire system: whoever delivers results receives money. Whoever delivers none receives less. That this thought is considered radical in Germany says more about the system than any number could.