Steve,
Interesting paper in light of your previous posts. I'll agree that the rise in receiveables increases the probability of the existence of earnings manipulation. However, this model's explanatory factor is still just 30%, which might be high from a relative standpoint against most models, but certainly does not make it highly probable that OPMR is manipulating earnings.
Furthermore, the results of the model has not apparently been tested/applied on various 10-year data sets before and after 1992 to test the robustness of the model. It would be interesting to see what percentage of known manipulators this model would find if applied on rolling ten year periods starting in the 1950's through 1998. It would also be interesting to see the number of "non-manipulators" it misclassified during the period.
As to the assumptions made with regard to the cost of errors by the model, I must say that the the authors' supposition that the classification of a non-manipulator as a manipulator is a factor of ten or more lower than the cost of classifying a manipulator as a non-manipulator is incorrect. The author's main justification for this is that there are a relatively infinite alternate investments available which would not reflect this risk. While I agree that the short-term damage of revealing a manipulator is quite quantifiable and often results in a large downward price adjustment, the misclassification of a non-manipulator is no less damaging to the investors of that company.
In all, I think the paper does a good job of identifying variables which can increase the risk that a company may have inflated its earnings. However, a broad brush application of the model is not something that should determine whether a stock is a good investment.
As backup to this statement, I present the follow exerp from the end of the report:
"While the model is cost-effective relative to a strategy of treating all firms as non-manipulators,
its large rate of classification errors makes further investigation of the results an
important element to the model’s implementation. That is, since the model’s variables exploit
distortions in financial statement data that could result from manipulation, one must recognize that
such distortions can have an alternative origin. For example, they could be the result of a material
acquisition during the period examined, a material shift in the firm’s value maximizing strategy, or
a significant change in the firm’s economic environment."
Valuation Guy/Obewon |
