A Conditional Verdict on Corporate Venture Capital

The evidence in this series answers the opening question in a qualified way: corporate venture capital does not automatically make start-ups more innovative because it is not a neutral input that simply adds money to a fixed production process. CVC first selects start-ups that already look technologically promising, then changes their access to resources, markets, and ecosystems in ways that can increase diffusion while reducing independent exploration. The outcome depends on the parent’s position, the start-up’s objective, the terms of the relationship, and the alternatives that remain available.

This conclusion is not anti-CVC. It is a more demanding account of what corporate venture capital is. A minority investment can provide capital, technical knowledge, supply-chain access, customers, distribution, talent, credibility, and a route to market. It can also expose the start-up to the parent’s priorities, information needs, governance influence, and bargaining power. The same relationship can improve the reach of an existing technology while narrowing the range of new technologies the start-up is willing or able to pursue.

The practical task is therefore to judge a specific CVC relationship rather than a label. A founder, corporate investor, or policymaker should ask what was true before the investment, what changed afterward relative to a relevant alternative, which kind of innovation moved, how the ecosystem and governance shaped the movement, and who can capture value at exit. Those questions produce a conditional judgment where a headline about “more innovation” cannot.

The evidence says CVC changes the path, not automatically the amount of innovation

The first part of the answer is selection. In the OECD study’s 2001–2022 sample, 31% of CVC-financed start-ups filed at least one patent, compared with 15% of start-ups financed by other VC. Among start-ups that patented, the CVC group filed an average of 2.43 new patents per year versus 2.02 for the other-VC group. Average yearly VC funding was about USD 22.39 million for CVC-financed start-ups and USD 12.55 million for other VC-financed start-ups. In a deal-level model, having a patent was associated with a 2.01 percentage-point increase in the probability of CVC financing, roughly a 50% premium over the model’s 4% unconditional probability.

Those figures establish that the CVC group starts from a different baseline. Corporate investors select for visible technological signals, financial capacity, and in many cases a plausible technological relationship with the parent. The result is important, but it is not yet a treatment effect. If a company already has a patent portfolio, more capital, and a technology that fits a corporate ecosystem, its later performance cannot be credited entirely to the CVC investment.

The second part of the answer is the post-investment trajectory. The OECD event study compares start-ups receiving CVC in their first funding event with other VC-backed start-ups whose first funding occurs in the same year. The analysis reports no significant pre-treatment differences in the relevant placebo tests, with p-values of 0.81 for patent applications and 0.93 for citations. That makes the post-funding divergence more informative than a raw comparison, although it does not remove every observational assumption.

After funding, the reported average effect is negative for log patent applications, at -0.033, and positive for log patent citations, at 0.203. The result is not “CVC reduces innovation” in an undifferentiated sense. It is a change in the measured composition of output. Relative to other VC-backed start-ups, CVC-backed companies file fewer patents while their patents receive more citations. Among start-ups that had patented in the three years before funding, 31% of the CVC group filed no patent afterward, compared with 39% of the other-VC group. The relative decline therefore appears to involve fewer filings rather than a simple abandonment of patenting.

This is why the two-axis model matters. Patent filings measure the intensity of one kind of inventive activity. Citations measure later references to the start-up’s patent stock and can indicate influence, use, or visibility. A start-up can move downward on filing intensity and upward on citation reach. It may be concentrating effort on a smaller number of projects, commercializing an existing invention, integrating with a platform, or benefiting from a corporate channel that exposes its technology to more users and collaborators. The data support the mixed pattern; they do not determine which internal decision produced it.

Citations to patents filed before the CVC investment make diffusion a plausible interpretation. The technical content of those earlier patents existed before the corporate relationship, so the increase in later citations cannot simply mean that CVC improved the quality of an already-filed invention. A corporate partner may have helped others encounter, adopt, or build on the technology. Yet citation growth is not direct evidence of revenue, product adoption, social value, or the start-up’s total innovation. The report treats diffusion, visibility, and ecosystem use as suggestive mechanisms, not as fully identified causes.

The third part of the answer is context. CVC is concentrated in an environment where a small number of highly active digital corporations can influence the average. About 80% of CVC-financed start-ups in the sample operate in high-digital-intensity sectors. Only 47% of CVC parents are in high-digital-intensity sectors, but those parents account for about 70% of CVC investments. The investment-weighted share matters because the behavior of a relatively concentrated group can shape the pooled result.

For CVC from high-digital-intensity parents, the estimated filing effect is -0.041, compared with -0.017 for parents in low- to medium-intensity sectors. The corresponding citation effects are 0.255 and 0.085. The high-digital group is therefore associated with both a larger relative reduction in patent filings and a larger increase in citations.