The Conditional Verdict: Preserve the Exit, Scrutinize the Deal

The subject of this series has been the acquisition of innovative start-ups by established firms, viewed across the moment before control changes and the years after it. The final question is whether a transaction that rewards risky entry can also weaken the innovation that made the target valuable. The evidence supports a conditional judgment rather than a universal one: preserve the exit option that helps bring founders and capital into the market, but scrutinize deals in which control over a potentially threatening idea can be more valuable than developing it.

That conclusion has two parts because the transaction has two time horizons. Before acquisition, a founder and investors care about the expected value of creating a company, including the possibility of a profitable sale. After acquisition, the new owner decides how much autonomy, research funding, implementation effort, and market access the idea receives. A deal can therefore be good news for entry and a bad environment for the target’s next invention. Treating those statements as contradictory is the result of compressing a sequence into a single outcome.

What the combined evidence can and cannot establish

The first fact to preserve is selection. The OECD study follows a deliberately defined population of young, innovation- and growth-oriented firms across 60 countries from 2001 through 2021. It contains 93,855 start-ups and 17,732 acquired start-ups after linking start-up, acquisition, and patent data. Acquired targets are more likely to have a visible patent and, among firms that patent, tend to have larger and more highly cited portfolios. In one extensive-margin specification, holding a patent raises the acquisition probability by 0.84 percentage points against a 2.4% mean probability, an estimated premium of about 34.9%. The main contrast is often whether a start-up has crossed into visible patenting at all, not whether it has accumulated an enormous portfolio.

The buyers are selected too. Incumbents that acquire start-ups are more innovation-oriented than firms that engage with them through minority investments. They are more likely to hold patents, to hold more of them, and to hold more highly cited portfolios. Real target–acquiror pairs also show greater technological proximity than fictitious alternatives and are less likely to cross industries or countries. These patterns are consistent with a buyer seeking technology it can understand and absorb. They are also consistent with a technically capable incumbent buying a firm whose future growth could threaten its existing position.

Selection changes how the later decline must be interpreted. A target that was more innovative before acquisition cannot be compared with a randomly chosen young firm after acquisition. Its pre-deal patenting may have signalled quality, reduced information asymmetry, attracted financing, or made the firm strategically important. A technically strong parent may also have had a different patenting trajectory from the beginning. Those differences do not disappear simply because the legal transaction is completed.

The second fact is the target-side outcome. In the OECD evidence, acquired start-ups’ standardized patent flows fall substantially after acquisition relative to suitable comparison firms. Before the event, the target and not-yet-acquired groups follow similar trends; afterward, the target’s flow falls more, by roughly 1.5 times the unconditional average yearly flow over the reported horizon. The descriptive table tells the same broad story in a different unit: excluding the acquisition year, mean patents fall from 2.4 before the deal to 1.1 after, while citation-weighted patents fall from 93.5 to 8.5.

The decline has both an extensive and an intensive margin. About half of acquired start-ups filed patents before acquisition, while approximately 10% continue doing so afterward. Among those that remain visible as patenting firms, new work receives fewer citations and the average radicalness of the portfolio falls by about 10% from its pre-acquisition level. The result is therefore not merely that a few firms stopped filing. The firms that continue filing also show a weaker measured trajectory.

The descriptive table and the event study should not be treated as the same estimate. The table aggregates activity over periods of different lengths and has changing observation counts because recent acquisitions do not have a complete post-deal window. The event study measures how the gap between acquired and comparison firms changes relative to the year before acquisition, using a standardized flow. The numbers belong to different estimands, but their agreement gives the target-side pattern a coherent shape: visible new work is concentrated before control changes hands, and the measured quality of surviving work falls afterward.

The third fact is the parent-side result. If the target’s patents disappeared only because the parent took over the research and registered equivalent inventions under its own name, the acquiror’s patenting might rise. The OECD models parent exposure as cumulative acquisition intensity, accounts for other acquisitions and unobserved earlier M&A through a burn-in period, and also uses withdrawn bids as a more demanding comparison in a restricted setting. It finds no significant post-deal increase in acquiror patent counts or patent flows. The citation-weighted patent flow of acquirors declines, while portfolio radicalness does not change.

This makes a simple relocation story less likely, but it does not show that the combined firm created no value. Research can move to an affiliate, become secret, appear in a product rather than a patent, or diffuse through other organisations.