“The Signal Quality of Earnings Announcements: Evidence from an Informed Trading Cartel”, 2020 (; backlinks; similar):
This study examines the revealed preference of informed traders to infer the extent to which earnings announcements are informative of subsequent stock price responses.
2011–42015, a cartel of sophisticated traders illegally obtained early access to firm press releases prior to publication and traded over 1,000 earnings announcements. I study their constrained profit maximization: which earnings announcements they chose to trade [9.25%] vs. which ones they forwent trading.
Consistent with theory, these traders targeted more liquid earnings announcements with larger subsequent stock price movement. Despite earning large profits overall, the informed traders enjoyed only mixed success in identifying the biggest profit opportunities. Controlling for liquidity differences, only 31% of their trades were in the most extreme announcement period return deciles. I model the informed traders’ tradeoff between liquidity and expected returns. From this model, I recover an average signal-to-noise ratio of 0.4.
I further explore 2 potential economic sources of this noise: (1) ambiguous market expectations of earnings announcements and (2) heterogeneous interpretations of earnings information by the marginal investor. Empirically, I document that the informed traders avoided noisier earnings announcements as measured by both sources of noise.
…Empirically, I test whether the informed traders behaved in a manner consistent with market microstructure theory. First, on the extensive margin, the informed traders chose more liquid earnings announcements. Compared to the unconditional mean probability of informed trade, a one standard deviation increase in liquidity increases the probability of trade by 50%. Liquidity is especially important in this setting because of detection risk. Large price impact prior to public disclosures bears the risk of discovery. Second, the informed traders chose earnings announcements with larger ex-post returns. A one standard deviation increase in the magnitude of realized stock returns increases the probability of trade by 19%. This finding confirms the joint hypothesis that informed traders could identify, and preferred to trade on, earnings with larger returns. Furthermore, on the intensive margin, the informed traders more aggressively traded earnings announcements with higher returns. Conditional on a stock that is informed-traded, a one percentage point increase in realized stock returns increases the informed traders’ price impact by 8.5 bps.
…To estimate signal noise from performance, I formulate a model of informed trade. In my model, an investor receives an array of noisy private signals about announcement period returns. The investor seeks to maximize profit by choosing to trade earnings announcements that are liquid and have large returns. The investor’s ability to do so depends on the precision of his return signals (ie. the earnings announcements). I estimate my model using simulated method of moments (SMM), where my moments are average returns, liquidity and their interaction. Using these moments, I recover parameter estimates that imply informed traders were willing to forgo 1% of expected return in exchange for 0.65 standard deviations of liquidity. Their performance implies a low signal-to-noise (SNR) ratio of on average 0.4. Within the context of this natural experiment, this is a causal estimate: signal quality determines performance. For comparison, I consider a simple benchmark trading strategy based on earnings surprise. This benchmark yields a comparable SNR estimate of 0.42. I infer from these low signal-to-noise ratios that earnings announcement press releases are poor signals of subsequent stock price responses.
…This unique natural experiment reveals a general fact that earnings announcements are noisy signals of subsequent market reactions. The informed traders had “perfect foresight” from stolen earnings announcement press releases, but they were only able to enjoy mixed success in predicting next-day stock returns. Their poor performance implies that capital market participants have difficulty mapping earnings information to stock price reactions. The contributions of this paper are to empirically quantify the limited informativeness of quarterly earnings announcements to individual investors, provide evidence on the likely sources of signal noise, and shed light on how this noise affects the behavior of capital market participants.