What Killed Post-Earnings-Announcement Drift

When a company reports earnings that beat analyst expectations, the stock jumps immediately, as you'd expect. What Ray Ball and Philip Brown documented in 1968 — and what hundreds of studies since have confirmed — is that the stock kept drifting upward for weeks to months afterward, as if the market's initial reaction to the surprise was systematically too small. Stocks that missed estimates drifted down the same way. This is post-earnings-announcement drift, PEAD, and it's one of the oldest and most robust anomalies in finance: a violation of the idea that all public information gets priced in immediately.

PEAD didn't disappear, exactly, but its economic size shrank substantially from the 1970s and 80s to the 2000s and beyond, and understanding why is a tour through most of the forces that erode any anomaly over time.

The forces that wore it down

Several changes stacked on top of each other across four decades. Decimalization in 2001 cut bid-ask spreads roughly in half overnight, which lowered the cost of trading on the signal and let more capital compete to capture it, narrowing the drift. Algorithmic and quantitative funds built systems specifically to parse earnings releases and analyst estimate revisions within seconds of a release, compressing a drift that used to unfold over weeks into one that mostly happens in the first few days. Analyst coverage broadened and consensus estimates became more widely and instantly disseminated, so the "surprise" itself was measured and repriced faster. And PEAD's economic size was always concentrated in smaller, less liquid stocks with higher transaction costs — as more capital specifically targeted those stocks to harvest the anomaly, the drift there shrank fastest, leaving the effect measurably smaller in the most liquid, most heavily traded names where it's easiest to actually implement.

Worked example

A company reports earnings 8% above the analyst consensus. In an older sample (say, the 1980s), the stock jumps 3% on the announcement day and then drifts up an additional 4% over the following 60 trading days as the market gradually absorbs the full implication of the surprise — a large, slow-moving, tradeable pattern. In a modern sample, the same 8% surprise might produce a 6% jump on the announcement day itself (the market reacting more completely and immediately) and only about 1% of additional drift over the following month, most of it concentrated in the first two or three days. The anomaly is the same shape, but nine-tenths of what used to be a multi-week opportunity has been compressed into the announcement day and the day after.

What this means in practice

PEAD-based strategies today still exist and still generate modest, statistically significant returns in careful studies, but they've shifted toward smaller, less liquid names where competition is thinner, shorter holding periods to capture what's left of the drift before it's arbitraged away, and combining the earnings surprise signal with other information (analyst revisions, guidance language, options-implied moves) rather than trading the surprise alone.