The Dog That Did Not Bark

Jacques Olivier, Professor of Finance - September 15th, 2008
The Dog That Did Not Bark

How can insider trading¹ patterns be used to anticipate large movements in stock market prices? Jacques Olivier uses his ‘dog that did not bark’ theory to explain the link between an absence of insider trading and stock market crashes. 

Jacques Olivier ©HEC Paris

Jacques Olivier has been professor of finance and economics at HEC Paris since 1996. Currently, he leads the Master in Finance program. He graduated from ESSEC and holds a Ph.D. (...)

See CV

In ‘Silver Blaze’3, Sherlock Holmes tells a Scotland Yard detective that he should focus his attention on a curious incident during the night of the crime: the dog did not bark. Sir Arthur Conan Doyle's short story demonstrates that sometimes there is more to be learned from the absence of a phenomenon than from its presence. Similarly, an absence of insider trading could explain sudden drops in stock prices, when there's no apparent reason for such stock market reactions. So, in an environment where information is often seen as a valuable asset, understanding the behaviour of those who have most of the information (i.e. the insiders), could prove invaluable in predicting, and even anticipating, stock market crashes.


USING INSIDER TRADING TO PREDICT MOVEMENTS IN STOCK MARKET PRICES

Existing literature on insider trading recognizes the informational content of insider acquisitions (an individual who adapts their investment strategy to the behaviour of insiders acquiring stock will earn abnormally high returns on their portfolio), but doesn't recognize that sales of stock contain valuable information. In other words, research undertaken thus far has failed to demonstrate that information released by insiders can help anticipate a drop in stock prices. But, by studying the relationship between the sale of stock by insiders and price movements, Jacques Olivier and JoséM. Marin demonstrate that it is indeed possible to anticipate a crash by observing insider sales. This relationship only becomes more complex when stock prices rise or stock is bought.


EXPLAINING CRASHES BY OBSERVING INSIDER BEHAVIOUR

Insider stock sales often reach a peak several months before a large drop in their price, but sales drop prior to a crash. There are three possible reasons for this phenomenon: • The effect of earning announcements: most of the crashes occur following an announcement by a company that it has failed to meet earnings expectations and most companies prohibit insiders from trading prior to an earning announcement. It is therefore quite logical that insider activity should be low at the time of a crash.

• The desire to escape scrutiny from the SEC4: the agency responsible for regulating the U.S. stock market is particularly vigilant about investigating any insider trading that takes place near the date at which the price movement occurred. Insiders therefore tend to limit their transactions just before a crash to avoid arousing the suspicions of the SEC.

• The ‘dog that did not bark’ theory: investors are informed when insiders sell their stock, and this leads to a partial downward adjustment of prices. But if the insiders have already sold off any stock they could sell, and then stop trading for a long period, investors can only infer that insiders are in possession of bad news, but not how bad the news really is. Market uncertainty increases and expected payoffs fall. This leads to a discontinuity in the price process, i.e. a crash.


HOW THE ABSENCE OF INSIDER TRADING INCREASES THE RISK OF A CRASH

The ‘dog that did not bark’ hypothesis is the only one that withstands the authors' empirical test. The effect of earning announcements isn't a convincing explanation, as insiders are prohibited from trading by their company prior to the earning announcement, which means the link between a drop in share price and an absence of activity is a statistical artefact. The explanation that insiders wish to escape scrutiny from the SEC is equally implausible: the SEC is just as vigilant about investigating insiders who buy stock just before a price rise as it is about those who sell stock just before a crash. And insiders continue to buy stock when prices are rising. However, the study does validate the ‘dog that did not bark’ theory: the absence of insider trading correlates positively with the probability of a crash in the short term.


THE PRACTICAL BENEFIT OF THE ‘DOG THAT DID NOT BARK’ THEORY

The ‘dog that did not bark’ theory demonstrates that uninformed investors react more strongly to the absence of insider transactions than to their presence. The commonly held idea that insider sales contain no valuable information is therefore wrong. On the contrary, insiders send strong signals to anyone who knows how to interpret them: insider sales in the recent past correlate negatively with the probability of a crash, while far away sales correlate positively. This phenomenon will enable anyone to adapt their investment strategy by observing insider behaviour.


1. Insiders or other insiders, as defined under American law: current and former managers, shareholders, their families, and so on.
2. José M. Marin is professor of finance at IMDEA (the Madrid Institute for Advanced Studies, Spain).
3. ‘Silver Blaze’, published in The Memoirs of Sherlock Holmes, by Sir Arthur Conan Doyle, George Newnes Publisher, 1894.
4. SEC: U.S. Securities and Exchange Commission. 


Based on an interview with Jacques Olivier and his article ‘The Dog That Did Not Bark: Insider Trading and Crashes’ (to be published in the Journal of Finance  in October 2008), co-written with José M. Marin².

RESEARCH METHODOLOGY
RESEARCH METHODOLOGY

Jacques Olivier and José M. Marin analyzed data from two different sources:

• The Thomson Financial Insider Filings (TFIF): a database that collects all insider trades reported to the SEC.

• The CRSP database: provides data on returns, total trading volume, and market capitalization. This data was used to identify large movements in stock market prices in three major American markets—the NYSE, NASDAQ, and AMEX markets—between 1985 and 2002. The two researchers used this data to empirically test their model, which explains how an absence of insider trading can cause—in certain specific situations described in the article—significant changes in the value of a company's stock.