Predicting Option Abuse

Stock options and insider trading are fascinating topics, and when you mix the two as this piece does, you have a very interesting article.

Carr Bettis of Gradient Analytics has completed a project that claims to pin-point when executives exercise their stock options “abnormally early.” Typically these early exercises come about six months before the market decline.

According to Bettis, "What we found was that not only do these early, deep-in-the-money exercises result in share price declines over the period that follows, but they also are associated with earnings misses and future misses,"

Even good theories don’t account for everything: In January 2006, Gradient noted abnormal exercise activity at Smith International Inc. Shares have seesawed since the report, but after a year were flat "We looked at it as a case that didn't work out as expected," Bettis said.

The question is: how much can it do, and where do we have to do more work?

First the basic article, was published in June 2005, see:

http://www.equityincentive.com/whitepapers/EIA%20White%20Paper.pdf

The paper is much broader than the précis and develops the thesis that early executive option exercise is a leading indicator of a market decline.

Being a white paper, it withholds key elements to prevent reverse engineering and general security issues, but there is enough to judge the approach. The company uses a proprietary database, to feed its Gradient’s Equity Incentive Factor model which determines whether the exercise is early or late.

For the Standard Option Expense Model, the companies are first into peer groups by market cap., industry and historical price volatility. With in each group an average of the valuation assumptions is calculated.

These values are then used to calculate Standardized Option Expense. This value is calculated using firm specific information on share price, exercise price and dividends, and peer averages for risk-free rate, expected life, and volatility.

Another measure used is the Objective Option Expense, which is the same as the previous measure, except it uses endogenous expected life and everything else is company specific, except risk-free rate.

The two measures, used in tandem are used identify companies that are not recognizing a correct option expense, issuing options against shareholder interest or when management is using insider information.

Other areas of interest are possible manipulation of option expense using this basic model as a starting place. Gradient has branched into judging governance effectiveness and other sorts of red-flag analysis.

In summary the Gradient has sought to model accounting information to investigate misuse of options and how that might relate to share price. Additionally they have studied the governance issues which might give rise to option abuse.

Critique

As note in the referenced article, Gradient’s techniques can provide false positives, as it were, in this case Smith International. It seems that there are weaknesses in the basic approach which makes the results useful as a signal for further work, and not an end in themselves.

The first issue is peer-group derivation. Volatility as a classification factor is used, but no mention of how it is actually used, which maybe proprietary. Volatility is used, along with size and industry, to define peer groups. Since option prices are extremely sensitive to volatility it is important how it is used for creating peer groups. As sentence or two could have added much clarity, but were not offered.

The next issues are related to the parameters used in the Standardized Option Expense:

-Gradient uses the peer-group average life. I find that comparison odd. Different companies use different standards to award options. An important implication of this relationship is that the computation of a single weighted average life overstates the value of an award if the pool of individual lives widely differed one from the other. Further, companies can vary widely within an industry on who is eligible. There is no indication that this fact is taken into account. This measure is of overwhelming importance in determining if executives are exercising options based on insider information, because groups of grantees can vary radically in their behavior.

-Volatility is probably the most important part of the equation, and it is certainly so for Black-Scholes. I have referenced NERA study which can provide some independent perspective, because Gradient does not offer much on the subject. A company can control volatility in two ways: first, financial leverage can affect volatility of the same reason as it does Beta. Companies can really increase volatility by share repurchase, or radically changing capital structure. The next thing management can do, is generate news, be it projects or acquisitions. A review of utility Betas during the Enron bubble can show that: for some companies like Duke, both Beta and R-square went down, based on news. In short a company can control this factor to some degree.

-As to interest rate, option value is not terrible sensitive to it, but it is unique to the date of issue and the option life. Why not just use what is given?

The Objective Option Expense is more tied to a given company’s actual situation, and is designed to test by recalculation volatility and time to exercise.

-Again historical volatility does not necessarily give future volatility which a company can change as noted above.

-Time to exercise can also change based on the composition of the grantee pool.

The piece then turns to Gradient’s approach to governance, and company ‘low balling’ of option expense. “Mirabile dictu” companies with low option expenses under-perform the market. That is like complaining that crime is down and the prisons are full. What may seem low might be right, for a given company. Also, many analysts just ignore option expense because it is non-cash and is seen as exogenous. What people do look at is potential dilution, which is where good boards of directors concentrate their control efforts. It would seem that Gradient’s work on option valuation is worthwhile only as it can spot anomalies, if then.

I also believe that the criteria for judging management are wrong footed in the extreme. Transaction oriented companies use options a lot and have, to based on the business. Mergers are a big part of the current business scene, but are not mentioned in any detail, in this piece.

In their analysis of governance, they put a lot of stock in fulfilling standard ‘boiler-plate’ requirements. I don’t. I believe that if you want to see is governance is proper you can not rely on any review of simple formalities, but must look deeper:

-How is the Board compensated? Stock options and performance awards make them part of management.

-Is the Board involved in day-to-day business decisions? If they are, then they are not a Board. You have a violation of the separation of duties, which is why you have Boards in the first place.

-How are high-risk areas like energy trading governed? In such circumstances, audit committees without members who have direct experience in those areas are like having no audit committee at all.

In summary, this service has the same foibles as other services that rely on accounting information. The systems used don’t appear to be fine enough to find the dependent variable on a consistent basis. Some of my concerns might be covered by proprietary techniques to which I am not privy, however. In these circumstance it is best practice to mention then drop the issue which improves clarity with no loss of security.