New York – It is of no surprise to any of us that the field of equity research, is actually composed of numerous disciplines and approaches. In assessing the value of research, we generally look at the predictive power of the research recommendations, but there are numerous research companies that typically do not generate recommendations. Even so, the research is very useful to portfolio managers and other investors.
In examining the value added of these research providers, it may be more instructive to separate the research providers into rough methodological buckets and then analyze them according to discipline, rather than in a cross sectional manner. One of the hot areas of research at present is the earnings quality area, where a number of academics have contributed to the development of filtering systems according to the level of accruals that a firm has on its balance sheet. Of course, EQ models have advanced beyond this, but they are still based on this premise.
In its most basic incarnation, the analysis takes firms with different levels of accruals and then sorts them into deciles from the highest level of accruals to the lowest level of accruals. We assert that the distinction between the highest and the lowest accruals certainly does represent different relative value/risk of the sorted companies, but does not necessarily indicate that the companies should be bought or sold. As such, the earnings quality systems do not yield absolute BUY/SELL/HOLD recommendations.
Some of the earnings quality approaches have different expectations for accruals, given the growth/value classification of the particular company. It is common sense that a company while it is growing will have higher accruals than that same company when it is mature. As such, some of the EQ models analyze growth companies separately from value companies.
Others have found that indeed high accruals and future value are not related to whether the companies are growth or value companies. So who is right-quite possibly both approaches are correct.
Suppose you drive every day in the mountains. Most of the time you are better off driving an SUV than a Volkswagon, because the SUV has four wheel drive and is built to handle better in the snow. In terms of the expectations of the two vehicles, therefore, there is a distinct difference between them. However, take both cars and run them onto a sheet of ice at 65 mph and they both behave pretty much the same. So if conditions are bad, but not critical, the SUV is distinguishable from the Volks, but when things are critical, they are not very different.
From an EQ standpoint, it may be that there is a difference between growth and value companies in terms of expected accruals, but that statistically they are the same in when accruals reach a critical level. As a result, firms with the highest level of accruals act pretty much the same regardless of whether they are growth or value companies.
One way to test whether there indeed is a difference is to take the distribution and throw away the tails and the middle. In a decile context, we could envision using the 2nd and 3rd deciles compared to the 8th and 9th deciles.
But we digress, The main point of the article is that the comparing research companies that have similar research methodologies naturally creates a forum on the validity of the models/approach used by the various players and produces meaning dialogue as to the research space in general.
More on this in coming blogs.