What About The European IPOs?

A data-driven look at 59 Euronext listings from 2021 to 2026, and the one factor that actually separates the winners from the rest.

The US IPO market is having a moment. On May 14, 2026, Cerebras Systems, an AI chipmaker billing itself as Nvidia's most serious challenger, opened on Nasdaq at $350, nearly double its IPO price of $185, and closed the day up 68%. It raised $5.55 billion in the largest U.S. tech listing since Uber's 2019 debut. The roadshow was reportedly 20 times oversubscribed. The excitement was palpable even before SpaceX began circulating its prospectus, with a potential $1.75 trillion valuation that would make it the largest IPO in history. OpenAI and Anthropic are queuing behind it. After years of drought (the rate-hiking cycle of 2022 shrank the U.S. IPO count from over 1,000 to barely 200 in a single year) the window is wide open again, and the biggest names in private markets are rushing through it.

It's hard not to get swept up in the energy. Inspired by the serial blockbuster listing, it's worth asking a question that the fanfare tends to drown out: what's actually happening to the IPO in the old continent? The full distribution, including the ones nobody talks about because they quietly disappeared.

To answer that question, we looked at 59 European companies that went public on Euronext markets between 2021 and 2026. The same period, roughly the same macro backdrop, and a similar mix of growth-stage businesses seeking public capital. What we found is not a story about recovery. It's a story about composition, and about a narrative that investors keep telling themselves that the data doesn't support.

What we did

We started with 103 IPOs across Euronext markets in Paris, Amsterdam, Oslo, Milan, and Dublin. After removing one duplicate (a single firm, KALEON, that appeared under two different tickers on two exchanges), we had 102 unique companies. Of those, 59 had enough retrievable price history to work with.

For each firm, we measured its monthly abnormal return, meaning the difference between what the stock actually returned and what its local benchmark index returned in the same month. A Paris-listed firm is compared to the CAC 40; Amsterdam to the AEX; Oslo to the OBX. This strips out the broad market and focuses on what is specific to the company itself.

We then tracked each firm month by month since its IPO, giving us between 2 and 63 monthly data points per company, and around 2,600 firm-month observations in total.

Getting the statistics right matters here, because European stock returns move together. When the ECB changes rates or a geopolitical shock hits, everything on Euronext tends to move in the same direction. If we ignored that co-movement, the tests will look far more confident than they should. To deal with this, we used Driscoll-Kraay standard errors, a method designed specifically to correct for cross-firm correlation, and we included time fixed effects in every model to absorb macro shocks that hit all firms at once. This approach follows the methodology of Hoechle, Karthaus, and Schmid (2021), a rigorous study of U.S. IPO performance that showed many documented findings were artefacts of using simpler methods on non-random samples.

What we found

The baseline: a persistent penalty. Across the full sample, the average monthly abnormal return was -1.66%. That's a meaningful drag. And there was no statistically significant relationship between how long a firm had been listed and how it performed (p = 0.80). Time, on its own, explained nothing.

A twist: the survivor signal. Things got more interesting, and more misleading, when we narrowed to firms listed for more than three years. In that subset of 42 companies, something appeared: a statistically significant positive trend (p = 0.022), suggesting returns improved over time at roughly +0.17 percentage points per month. We then isolated the 2021-2022 vintage, the 40 firms that went public in those two years, to rule out the effect of mixing firms from different eras. Same signal, still significant at the 5% level (p = 0.025).

At that point, the “time heals wounds” story had real numbers behind it. Then we added two variables.

The Final Model: Size Wins, Time Disappears

To test whether the apparent time trend was real, we added firm-level controls: the log of market capitalisation and the price-to-book ratio. We retrieved these for 38 of the 40 cohort firms, leaving 2,087 firm-month observations in the final model.

The result was unambiguous:

The age effect collapsed from a previously significant +0.0015 to +0.0000057, a reduction of over 99%. A p-value of 0.9938 means we have essentially zero statistical evidence that the passage of time improves IPO returns once we include a firm's size.

Log market capitalisation, by contrast, was the sole significant predictor. Larger firms outperformed smaller ones by a meaningful margin, consistently, across the sample period. The “recovery” we thought we were seeing wasn't firms maturing and finding their footing. It was the size premium dressed up as a time trend.

What This Really Means

The IPO penalty doesn't fade, it gets sorted. The baseline intercept in the final model implies a substantial and persistent underperformance for newly listed companies, independent of macro conditions. That penalty isn't being eroded by the clock. Where it diminishes, it's because the firm has grown large enough to counteract it. Small IPOs that stay small keep paying it.

The non-survivors tell the bigger story. Of the 102 firms we started with, 43 had no retrievable price history: delisted, merged away, or simply gone. That's a 42% attrition rate within five years. This is consistent with one of the most striking findings in the Hoechle et al. (2021) U.S. study: that IPO underperformance is almost entirely driven by non-survivors. When they split U.S. IPOs into firms that remained listed for at least five years and those that didn't, the surviving group showed no significant underperformance at all. The dead firms underperformed by roughly -15% annually. The “IPO discount” is really shorthand for “firms that are going to fail.” We can't make that split cleanly with our European data, but 43 vanished firms in five years makes the pattern hard to ignore.

The apparent discount is a composition effect. The broader finding, that the IPO penalty reflects firm characteristics related to small scale and low survival probability rather than a timing inefficiency that corrects itself, is consistent with the academic literature since the 1990s. Ritter (1991) first documented the long-run underperformance puzzle. Brav and Gompers (1997) showed much of it disappeared when you controlled for the small-growth tilt of IPO firms. Hoechle et al. (2021) confirmed that once you properly account for firm heterogeneity, the puzzle largely dissolves. Our European analysis builds on the same metholodgy at arrives at the same place: the aggregate “IPO underperformance” figure is an average across two very different groups, and the small failures are doing most of the damage.

The lesson from the European data isn't “avoid IPOs.” It's “know which kind you're buying.” The evidence strongly suggests that the size of a firm at listing is the single most important predictor of post-IPO return performance, more than the sector, more than the vintage year, and far more than simply waiting. The mega-caps queuing up in 2026 already have the one characteristic that our data says matters most. The danger is in extrapolating their story onto the long tail of smaller listings that ride the same wave of investor enthusiasm but don't share the same fundamentals.

Times change, markets evolve and the economic landscape reshaping demands for a deeper and up to date understanding of the new phenomena. Quantitative Partners brings the expertise to perform rigurous academic-level analysis that answer real market questions.

This analysis draws on the methodology of Hoechle, Karthaus, and Schmid (2021), “The Long-Term Performance of IPOs Revisited” (University of St. Gallen, SSRN 2929733). Original work on U.S. IPO underperformance: Ritter (1991) and Brav and Gompers (1997). Statistical methods: Driscoll and Kraay (1998).