IPO underperformance (the "new-issues puzzle") is the empirical finding that companies which have recently gone public tend, on average, to deliver lower returns over the following three-to-five years than otherwise-comparable seasoned firms or the broad market. It sits in deliberate tension with its twin anomaly, IPO underpricing — the average ~15–20% first-day pop from offer price to closing price. The paradox: the same security can be a windfall for the allocated short-term flipper and a laggard for the buy-and-hold investor who purchases in the aftermarket. The anomaly is famous, foundational to behavioral finance, and genuinely contested — much of the debate is about whether it is a real abnormal return or an artifact of how returns are measured.

How it's measured

The classic test (Ritter 1991) buys each IPO at its first-day closing price (not the offer price — flippers already captured the pop) and holds for three years, then compares the buy-and-hold return to a firm matched on size and industry. Results are aggregated as a buy-and-hold abnormal return (BHAR) or as a wealth relative (terminal wealth of the IPO portfolio ÷ terminal wealth of the benchmark; below 1.0 = underperformance).

The measurement choices are not innocent — they drive the result:

• Buy-and-hold (BHAR) vs. calendar-time (CTAR): BHAR equal-weights events and compounds; calendar-time portfolios value-weight and avoid the cross-correlation that inflates BHAR t-stats. The anomaly is much weaker, often insignificant, under value-weighted calendar-time methods.
• Benchmark: matching only on size shows large underperformance; additionally matching on book-to-market (small + low-B/M = "small growth") shrinks it dramatically (Brav, Geczy & Gompers 2000).
• Weighting: equal-weighting magnifies the effect because it overweights the many tiny issues; value-weighting attenuates it.

How it's used in practice

Practitioners and academics use the finding three ways. (1) As a caution against aftermarket chasing — the base-rate expectation for a freshly listed, heavily hyped stock held for years is below-average. (2) As an input to event-driven calendars: the most reliable companion effect is the lock-up expiration (commonly 180 days), when insider/VC selling supply hits the float. Field & Hanka (2001), studying 1,948 lockups, documented a small but statistically significant ~−1.5% three-day abnormal return plus a permanent jump in trading volume at unlock, with a much larger drop for venture-backed issues where VCs sell aggressively. (3) As a factor-overlay screen: "net stock issuance" is a recognized cross-sectional return predictor (firms that issue equity, including via IPO, subsequently underperform firms that buy back), and it is often treated as the same phenomenon viewed through a factor lens rather than a standalone IPO trade.

Note what the anomaly does not license: it is a slow, average, portfolio-level tendency, not a short-horizon or single-name signal. It says nothing about the first-day pop and is far too noisy to time an individual name.

Adoption, debate & evidence

The original evidence is strong and oft-cited. Ritter (1991), on 1,526 US IPOs from 1975–84, found a three-year wealth relative of about 0.83 (each $1 in IPOs grew to ~$1.34 vs. ~$1.62 for matched firms) — roughly −27% to −30% cumulative vs. the benchmark. Loughran & Ritter (1995) extended this to seasoned offerings under the "windows of opportunity" hypothesis: firms issue equity when their sector is overvalued, so subsequent returns are poor. Ritter's Warrington (University of Florida) dataset continues to document that equal-weighted three-year buy-and-hold returns of IPOs trail style-matched firms across decades, though the gap varies widely by cohort and is concentrated in small, young, low-profitability, high-volume-period issues.

The skeptical literature is equally serious and is why this is filed as contested:

• Brav, Geczy & Gompers (2000) — underperformance is "concentrated primarily in small issuing firms with low book-to-market ratios"; control for size and B/M and it largely disappears, suggesting it overlaps the small-growth value premium rather than being IPO-specific.
• Eckbo & Norli (2005) — IPO stocks have lower exposure to leverage/liquidity risk factors; on a risk-adjusted basis the abnormal return is small and statistically insignificant.
• Gompers & Lerner (2003), "The Really Long-Run Performance of IPOs" (pre-Nasdaq 1935–72) — underperformance appears in BHAR but vanishes under value-weighted calendar-time measurement, pointing to a methodology effect.
• Schultz (2003) — "pseudo-market-timing": because more firms go public near market peaks, equal-weighted event-time underperformance can arise mechanically even with no ex-ante predictability and no investor irrationality.
• Behavioral side (Miller 1977 divergence-of-opinion; Field & Hanka 2001 and Aggarwal/Krigman/Womack 2002 on lock-ups): short-sale constraints plus heterogeneous, optimistic beliefs let aftermarket prices overshoot fundamental value, then drift down as constraints relax and the lock-up frees supply.

Honest synthesis: the raw effect is robust; the abnormal (risk- and characteristic-adjusted) effect is disputed. Whether it is a genuine free lunch or a relabeling of small-growth risk plus a measurement artifact remains unsettled.

Strengths & limitations

It works best as a prior, not a trade: a base-rate reminder that the average newly public, story-driven small cap is a long-horizon laggard, and that lock-up dates are real supply events. It fails as a precise edge because (a) the result is highly sensitive to method (equal vs. value weight, benchmark, BHAR vs. calendar-time) — change the spec and the anomaly can evaporate; (b) it is regime- and cohort-dependent (hot-issue windows underperform far worse than cold ones; large, profitable, value-priced IPOs often do not underperform); and (c) it is an average over years, useless for short-horizon timing of one name. The single most common misuse is treating "IPOs underperform" as a license to short newly listed stocks — borrow is scarce and expensive early on (the very short-sale constraint that causes the overpricing), early momentum can be violent, and the realized edge is too small and slow to survive shorting costs.

Sources

• Ritter, J. R. (1991), "The Long-Run Performance of Initial Public Offerings," Journal of Finance — original 1975–84 evidence; ~0.83 three-year wealth relative.
• Ritter, J. R., "IPO Long-Run Returns / Updated Statistics," Warrington College of Business, U. Florida — ongoing dataset and decade-by-decade tables. https://site.warrington.ufl.edu/ritter/ipo-data/
• Loughran, T. & Ritter, J. R. (1995), "The New Issues Puzzle," Journal of Finance — windows-of-opportunity / SEO extension.
• Brav, Geczy & Gompers (2000), "Is the Abnormal Return Following Equity Issuances Anomalous?," JFE — disappears under size + B/M matching.
• Eckbo, B. E. & Norli, Ø. (2005), "Liquidity Risk, Leverage and Long-Run IPO Returns" — risk-factor critique.
• Gompers, P. & Lerner, J. (2003), "The Really Long-Run Performance of IPOs," Journal of Finance (NBER w8505) — pre-Nasdaq; value-weighted calendar-time result.
• Schultz, P. (2003), "Pseudo Market Timing and the Long-Run Underperformance of IPOs," Journal of Finance.
• Field, L. C. & Hanka, G. (2001), "The Expiration of IPO Share Lockups," Journal of Finance — ~−1.5% three-day abnormal return at unlock, larger for VC-backed issues.
• Miller, E. (1977) divergence-of-opinion; Aggarwal, Krigman & Womack (2002), "Strategic IPO Underpricing, Information Momentum, and Lockup Expiration Selling," JFE — lock-up / insider selling mechanics.
• Shefrin, Beyond Greed and Fear, ch. on underpricing, underperformance and hot-issue markets.

Dispute flag: the existence of a risk-adjusted IPO abnormal return is genuinely unsettled — the raw underperformance is well documented but may reflect small-growth risk plus pseudo-market-timing rather than a true anomaly.