The Gateway to Algorithmic and Automated Trading

The Underlying Dog and the Option Tail. Which Wags Which?

Published in Automated Trader Magazine Issue 31 Q4 2013

The direction of information flows in markets has long been the subject of academic debate. Automated Trader talks to Dr. Tyler Brough of Utah State University about recent papers he has co-published with colleague Dr. Ben Blau on this subject in the context of options, their underlying securities and market friction.

Dr. Tyler Brough of Utah State University

Dr. Tyler Brough of Utah State University

Automated Trader: What prompted your research in this area?

Tyler: It builds upon a paper by Chakravarty, Gulen and Mayhew2 which made use of a vector error correction model that was originally designed by Joel Hasbrouck to examine what happens when the same stock trades in different markets. This explored the sharing of information and how the price discovery of smaller markets essentially borrowed from the price discovery process in their larger peers.

However, one of the most intriguing things about Mayhew, Gulen and Chakravati's research is that it used Hasbrouck's econometric tool to explore the price discovery relationship between stocks and their listed options, rather than the same stocks listed on different cash markets. This was methodologically interesting because they obviously could not directly compare price of the underlying and its derivative. Instead they had to use the option price to estimate an implied underlying price, much as one might when calculating implied volatility. That was not a trivial task because it had to accommodate the multiple strikes and expiry dates of the associated options.

Their conclusions were that the option market was essentially leading the underlying market. However, their findings were contradicted by a paper that came out last year3 (albeit based upon a relatively small sample of 39 liquid NYSE stocks) that found that quotes in the options market lagged those in the underlying market.

Automated Trader: I suppose you could argue that the activity of informed traders in option markets would make it plausible for options to lead their associated stocks?

Tyler: Exactly. If you were an informed trader, which market would you prefer to trade in? Although liquidity may be lower in derivative markets, transaction costs are also lower and greater leverage is possible. If information leakage is a concern, an informed trader may also find it easier to mask their intentions in the options market. Finally, if there are short selling constraints in the underlying market, then the options market is the logical alternative venue.

Our view is that where markets are tightly coupled, the activity of informed traders or information flowing through the options market to the underlying market, might drive prior price discovery in the derivatives contract.

Automated Trader: What methodology did you use in your research to test this hypothesis?

Tyler: We used the Hou and Moskowitz methodology of price delay4. This can be thought of as a market model where one can lag the market component out a certain number of periods. Then when one regresses the returns of some security onto that, the significance of the lag coefficients tells you at what rate that security adjusts to market-wide information. So it is essentially a measure of the speed of price discovery.

We used this methodology in conjunction with an event study around option listings. We looked at underlying securities that were eligible for an option contract to list and then compared those that did and didn't list options by calculating their respective price delays (the time taken for market information to affect the price of a security).

Automated Trader: Given that the paper that contradicted Mayhew, Gulen and Chakravati's research was based upon a relatively small selection of stocks, what range of data did you use in your testing?

Tyler: We examined 2,523 stocks from the NYSE and NASDAQ from January 1997 to December 2007 that listed options during that period. We obtained monthly returns, volume, prices, shares outstanding, and market capitalisation data on the stocks from the Center of Research on Security Prices (CRSP) and data on the number of option contracts traded from Bloomberg.

Automated Trader: How did you analyse the price delay data?

Tyler: We used the CRSP data to calculate weekly returns for each stock from Wednesday to Wednesday. We then used the Hou and Moskowitz methodology mentioned earlier to calculate the price delays. We then began our analysis by examining statistics that described our sample both before and after options become available. However, we also controlled for other factors that Hou and Moskowitz found affected the level of price delay, including contemporaneous market capitalisation, book-to-market ratio, institutional ownership, return volatility and turnover.

Automated Trader: What was the outcome?

Tyler: We found that the average delay dropped by nearly 42% after options were listed. This seems to suggest strongly that when options become available to trade, more information flows in and the speed of adjustment for the underlying security increases.

We also examined the delay's return premium in relation to the introduction of options. Again, the results were rather striking. After controlling for other risk factors that influence future returns, we found that stocks with the highest delay outperformed stocks with the lowest delay by 15%. We also found that returns over a one-year forward window prior to options listing were higher than for a one year period post listing, with risk adjusted returns being negative post-listing, versus more than plus 6% pre-listing. This suggests that the profitability of trading strategies that buy and hold stocks with high price delay depends on whether options are available for the underlying stock.

Automated Trader: Did you discover any direct relationship between the delay return premium and option volume?

Tyler: Yes. Our final set of tests examined the delay return premium in relation to the level of option trading activity during the post-listing period. We first sorted stocks into quartiles based on delay, then within each delay quartile we sorted stocks again into portfolios based upon the amount of option volume in each particular stock.

We then examined the returns for each double-sorted portfolio over the next-year. Our hypothesis was that future returns were expected to increase across delay portfolios, thereby suggesting that investors command a return premium in stocks with high delay. In addition, if option trading activity reduced the return premium commanded by delay, then the magnitude of the return premium should also decrease across option volume quartiles.

The results showed that the difference between extreme quartiles was positive and statistically significant, indicating that the delay return premium was present in stocks with the lowest option trading volume. However, the results also showed that the difference between extreme option volume was negative and significant at the 0.01 level, which suggested that the return premium commanded by delay was decreasing as the level of option trading volume rose. These results were robust when sorting stocks in each delay quartile into either call or put volume portfolios.

Automated Trader: So your work effectively combines the econometric measure of delay with an option introduction event study?

Tyler: Yes, and that combination of methods allowed us to draw three robust result-backed conclusions:

1. Once an option is introduced, price delay in the underlying stock decreases.

2. The return premium associated with that price delay also decreases once options are introduced, because traders now have a new channel through which to invest.

3. Price delay (and therefore also its return premium) monotonically decreases as option volume increases, which also further supports conclusion 2.

Automated Trader: Did you also break down the option data by price and expiry?

Tyler: No, we didn't because there is no way of knowing which strikes/expiries (or combination of them) an informed trader would actually be using. Depending upon circumstances they might be trading a variety of ranges and strike prices, plus implementing very complex multi-leg option strategies. In practical terms it wouldn't have made much difference even if we had been able to do this, because it wouldn't have affected the three conclusions we were able to draw.

Automated Trader: So what are the practical implications of your research for professional traders?

Tyler: I think there are a couple of things:

• Firstly, there is a reasonable inference that if the option market moves, then there is a good chance that it will be a leading indicator for movement in the primary market. However, our results were calculated using weekly return data, so there is still scope for exploring this further using higher-frequency data. It is perfectly possible that the same general relationship holds in multiple time frames, which could be actionable as a trading strategy.

• Secondly, any traders currently using strategies that seek to exploit the price delay return premium in stocks without options listed would be well advised to monitor option listings decisions and announcements - or risk finding their returns significantly eroded.