Print
E-MailKenneth R. Ahern is the Dean’s Associate Professor in Business Administration at the University of Southern California. This post is based on his recent paper.
Trading by investors who have material, non-public information is of first-order importance to liquidity providers, stock market operators, and securities market regulators. Liquidity providers, such as market makers and institutional investors, worry that an informed trader will take advantage of their lack of information. Operators of stock markets worry that the presence of informed investors will drive uninformed investors out of the market. Regulators worry that an unfair advantage by some investors will impair equal access to equity markets.
Though many market participants worry about the presence of informed trading, there is little credible evidence on the validity of existing empirical proxies to identify periods of informed trading, such as bid-ask spreads, Kyle’s lambda, and trade order imbalances. Though these proxies are theoretically grounded, they remain largely untested. This is because validating the proxies requires the rare opportunity to directly observe informed trading.
In a new paper entitled, Do Proxies for Informed Trading Measure Informed Trading? Evidence from Illegal Insider Trades, I exploit detailed data on illegal insider trading to provide new evidence on the validity of a host of proxies for informed trading. Though illegal insider trading does not represent all informed trading, these data help to overcome a number of empirical obstacles. First, the legal documents in insider trading cases provide direct observations of the timing of information flows and trades. Second, the trades documented in illegal insider trading cases are, by definition, based on material non-public information, not speculation or public information. Third, the data include observations of informed trading in a wide range of firms and events. Finally, the longevity of information varies in the data, allowing a comparison of short-lived versus long-lasting information.
The observations of insider trading are hand-collected from all insider trading cases filed by the Securities and Exchange Commission (SEC) and the Department of Justice (DOJ) between 2009 and 2013. The sample includes 312 different firms in 410 different insider trading events over the period 1996 to 2013. Mergers and acquisitions are the most common event in the sample (52%), followed by earnings announcements (28%), news about drug regulation (9%), and other announcements about operations, security issuance, and financial distress. Firms range from recently public firms to the largest firms in the economy, including Microsoft, Procter & Gamble, and Berkshire Hathaway, and the median sample firm is comparable to the median firm on the NYSE.
Using direct observations of insider trading, I compare the power of widely used measures of illiquidity to predict informed trading. Specifically, using intraday data of trades and quotes, I calculate quoted, effective, and realized spreads, price impact, absolute order imbalance, Kyle’s lambda, and parameters from the decomposition of the bid-ask spread. Using daily data, I also calculate the Amihud illiquidity.
First, I show that informed traders strategically time their trades to avoid detection. In particular, without controlling for strategic timing, I find that none of the standard proxies of informed trading are significantly related to illegal insider trading. Instead, sophisticated insiders wait to trade on days when liquidity is high and their trades are harder to identify by market participants and regulators. This causes proxies of informed trading to be low exactly when insiders are trading.
Second, when I control for strategic timing, the results are strikingly different. When information is short-lived, and informed investors do not have the luxury to wait for days with high liquidity, nearly all of the proxies of informed trading are correlated with illegal insider trading, consistent with theoretical predictions.
However, these results are subject to the concern that the data on illegal insider trading is limited to the cases that were detected by the regulators. In particular, if regulators use standard proxies of informed trading to detect illegal insider trading, the results might just reflect bias in the way in which the data were sampled from the population by the regulators
To test for sampling bias, I run tests to separate cases in which regulators are more likely to have used proxies of informed trading to detect illegal insider trading, compared to cases detected by traditional investigations. To separate cases driven by empirical proxies, I identify SEC insider trading cases that were investigated with assistance of FINRA. FINRA is the financial sector’s self-regulatory agency and has the responsibility to monitor markets for suspicious trading behavior. If FINRA detects abnormal behavior using its proprietary algorithms, it refers the case to the SEC. In contrast, the SEC also brings cases that were detected by tips submitted by the public and other non-market investigation techniques.
Once I control for FINRA’s involvement in a case, I find that only two proxies of informed trading still predict illegal insider trading: absolute order flow of stock trades and the autocorrelation of order flow. Traditional measures, such as the bid-ask spread and Kyle’s lambda lose their ability to predict insider trading. Moreover, my results indicate that FINRA’s algorithms rely on unusual changes in bid-ask spreads (quoted by market makers) to detect suspicious activity, but ignore patterns of orders (placed by traders). FINRA’s algorithm might be improved if they rely more heavily on patterns of order flow.
This paper’s results have a number of important implications. First, though standard proxies for informed trading have the power to predict illegal insider trading, they are only effective when information is short-lived and traders cannot strategically time their trades. When information is long-lived, strategic timing nullifies the predictive power of all of the illiquidity measures. Second, the illiquidity measures that are the most reliable predictors are based on order flows, not prices or quotes. This suggests that market makers do not adjust prices and quotes in response to informed trading, contrary to the assumptions of many theoretical models. Finally, the results of this paper help to reveal the conditions under which regulators are more likely to detect insider trading.
The complete paper is available here.
