Institutional Trading around Corporate News: Evidence from Textual Analysis

Alan Guoming Huang is Associate Professor of Finance at the University of Waterloo, Hongping Tan is Associate Professor at McGill University’s Desautels Faculty of Management, and Russ Wermers is the Bank of America Professor of Finance at the University of Maryland. This post is based on their paper, forthcoming in the Review of Financial Studies.

Institutional investors now own over 60% of corporate equities, and account for an even greater proportion of trading volume. Accordingly, institutions play a large role in the incorporation of new information into market prices. However, the mechanism of how institutional investors use information to trade, and how quickly their information-motivated trades are reflected in stock prices is not clear.

This paper studies two high frequency databases to address these issues. First, we construct the most comprehensive corporate news database academically studied, to date, for the period 2000 to 2010 (from the Top Sources of Factiva). This database contains a much broader set of news than prior research based on news releases, and it contains the vast majority of news wire releases, which include their precise time-stamps. Second, we use a high-frequency institutional trading database that holds the precise trades made by institutional investors consisting of about 10% of the market’s overall trading volume over the same time-period. We merge these two datasets to study the reaction to, or prediction of, institutions to the tone of various news events about corporations (i.e., “negative news” or “positive news” by counting the quantity of sentiment words in each news article). We also measure the impact of institutional trading following news events on ensuing stock returns to determine the role institutional high-frequency trading plays in incorporating news into stock prices.

Part of the reason that we carry out this research is that investment communities have a widely held belief that institutional investors are able to “predict” news and trade accordingly, even despite Regulation Fair Disclosure’s prohibition of selective corporate news releases to well-connected investors or analysts. While this may be true in some limited settings, we find that, in our more general setting, institutions trade on the news tone only on the news release day, but not prior to, nor after the news release day. What this means is that institutions react speedily to but do not predictively trade on news. We also find that institutions’ speedy reaction to news pattern is more pronounced for news related to firm fundamentals, and with a less ambiguous content.

A key innovation in this paper is that we use the first news as the starting point to examine the news-trading causal relation. In the case of successive news and trading arrivals, it is hard to disentangle such causal relation (i.e., whether news drives trading or trading elicits subsequent news). In particular, if, for a given firm, news tone is persistent across multiple days in a string of news and traders trade in the direction of news over the period (which the paper verifies), earlier-day trading may simply be reactions to contemporaneous news but could be erroneously categorized as predicting later-day news. Identifying the first news release is thus paramount to knowing whether other follow-up news releases might be anticipated by a sophisticated institutional investor. We ameliorate this potential confounding causality between trading and news in consecutive-day news by “clustering” news across days into a single news event, and examine trading reaction to the entire cluster or to the initial news. We find that institutions speedily react to single-day news (which accounts for 78% of the news articles), but also to consecutive-day news (accounting for the rest 22% of the news articles). They do not predict news in either case.

We find that such news-driven institutional trades result in economically significant abnormal returns. A one standard deviation increase—a standard way of measuring economic significance—in institutional trading results in an annualized abnormal return of 12.5 basis points over five days following the news on top of the return predicted by news tone. This economic significance is meaningful as institutions oftentimes vie for a few basis points in gains. Institutional trading and news tone reinforce each other to predict returns over the next two weeks. While institutional trading predicts returns, the return predictability is mostly present when such trading is consistent with the news direction.

Our results are relevant to practitioners at large for at least a few reasons. First, our results suggest that the trading advantage of institutional investors stems more from their ability to process information in a very timely manner than from their ability to predict (or to obtain) private information about corporate events ahead of their public release. Second, we show that institutions are key to information becoming quickly incorporated into prices, suggesting a market efficiency improving role for institutional investors. And, third, our results suggest that investors should be aware that this high-frequency information is available to the market, and that investors may be able to exploit these data to generate return alpha. Overall, this research sheds light on how public news interacts with institutional trades and hence how institutions resolve information asymmetry in the market place.

The complete paper is available for download here.

Both comments and trackbacks are currently closed.