Beat Earnings, Stock Still Drops 5%: What Happens Next in 12 Markets

We ran a beat-and-dip event study on 12 exchanges. The question: when a company beats earnings estimates but the stock sells off 5% or more anyway, does it recover? In 10 of 12 markets, the answer is no. The sell-off either persists or the stock goes nowhere.

Contents

1. The Setup
2. The Global Picture
3. What's Statistically Significant
4. The Nine Noise Markets
5. Data Quality Notes
6. Why the Results Diverge
7. Takeaway

The Setup

A "beat-and-dip" event is: 1. epsActual > epsEstimated with ABS(epsEstimated) > 0.01 2. Stock drops ≥ 5% from T-1 close to T+1 close (announcement reaction)

For each event, we measure cumulative abnormal return (CAR) at T+21 and T+63 trading days from T+1 close (the dip bottom). CAR = stock return minus a regional benchmark ETF. Market cap floors applied per exchange in local currency.

Total events across 12 markets: 18,624

The Global Picture

p<0.05, p<0.01

†Effective coverage primarily post-2022 due to FMP pipeline timing. ‡China beat rate 32–36% vs 41–60% for other markets; interpret with care.

What's Statistically Significant

Three markets show significant results. The other nine are noise in one direction or the other.

Taiwan: Positive reversion, significant at T+21 and T+63. +1.76% at T+21 (t=2.73), +3.04%* at T+63 (t=2.47). The 10–20% dip category is even stronger: +6.58% at T+21. This is the only market where "buy the beat-and-dip" holds up empirically.

India: Positive at T+63 only. -0.17% at T+21 (not significant), then +2.54%** at T+63 (t=3.55). The reversion is delayed, flat for the first month, then a recovery that takes until 63 days to show up. Data is concentrated 2022–2025 (a strong bull period), so treat this as a recent-regime finding rather than a long-run structural result.

US: Significant negative. -0.22% at T+21 (t=-2.12), -0.84%* at T+63 (t=-4.53). The sell-off continues. This is the largest sample in the study (13,950 events, 25 years) and the most credible negative result.

Sweden: Significant negative at T+63. -2.58%* (t=-2.48). Only 319 events, but the T+63 result is consistent with the US direction and reaches significance.

The Nine Noise Markets

Canada, Japan, Germany, UK, Korea, Brazil, China, and Hong Kong all have T+21 and T+63 t-stats below 1.96. None show a consistent pattern in either direction. This isn't "the strategy doesn't work" in those markets, it's "we can't tell from this data."

Japan (n=776) is notable for a t-stat of essentially zero at T+63. The market reacts to the beat-and-dip event and then, nothing. Complete market efficiency on this specific setup in Japan.

Canada is interesting as a 25-year sample with 737 events. The T+63 result is -1.45% with t=-1.93, which just misses significance. If you're managing a Canada-only book, this is probably a pattern worth tracking.

Data Quality Notes

UK (LSE): FMP earnings data for LSE was sparse before 2022. The 633 events are concentrated in 2022–2025. Results reflect a recent 4-year window, not a structural market property.

India (BSE/NSE): Similar pipeline timing issue. 78% of events are from 2023–2024. The T+63 t-stat of 3.55 is real, but this is recent-period data during a strong Indian bull market.

China (SHZ/SHH): China's beat rate is 32–36% in this sample versus 41–60% for other markets. Lower beats may reflect reporting norms or data coverage gaps rather than fundamental differences in earnings quality. The China result is included with this caveat.

Excluded: Australia (8-year history, adjClose data quality issues), Thailand, Norway, and Switzerland all had fewer than 100 qualifying events, not enough for reliable event study statistics.

Why the Results Diverge

The directional split. Taiwan and India positive, US and Sweden negative, suggests market structure matters more than earnings quality per se.

US: The most efficient market in the sample. Institutional positioning on earnings is sophisticated. A 5%+ sell-off after a beat almost always reflects real information: guidance, revenue, or margin disappointment that the EPS number doesn't capture. In an efficient market, the stock reacts correctly the first time.

Taiwan: Large retail investor base, lower analyst coverage per stock, and a semiconductor-heavy index where earnings quality is genuinely hard to read from EPS alone. The retail overreaction hypothesis is plausible. The institutional re-rating takes 2–3 weeks.

India: Delayed reversion suggests either slower institutional processing or a market structure where the initial sell-off is driven by uninformed price momentum rather than fundamental reassessment. The 2-month lag before recovery is longer than anywhere else.

Takeaway

"Buy the beat-and-dip" is a sound instinct that mostly doesn't work. The market's negative reaction to a genuine earnings beat typically contains real information about forward prospects. Buying against that information in the US, Canada, Japan, Europe, or Korea hasn't produced positive abnormal returns on average over the last 10–25 years.

The exceptions are Taiwan (consistent positive reversion, especially in the 10–20% dip range) and India (delayed positive reversion at T+63, with a data recency caveat). These two markets are worth watching for earnings plays. The rest aren't.

Data note: Results use regional benchmark ETFs (SPY for US, INDA for India, EWJ for Japan, etc.). Market cap floors applied per exchange. All CARs winsorized at 1st/99th percentile.

Part of a series: Post-earnings dip mean reversion tested globally. See individual market studies: US, Taiwan, India.

Run It Yourself

Explore the data behind this analysis on Ceta Research. Query our financial data warehouse with SQL, build custom screens, and run your own backtests across 70,000+ stocks on 20 exchanges.

Data: FMP earnings surprises + adjusted prices, 2000–2025 (varies by exchange). 18,624 total beat-and-dip events across 12 exchanges.