tjbresearch.com
Wednesday, July 20th, 2011
decades of work

When you think about it, the profession of securities analysis is not that old (although people have been making judgments about the value of investments since time immemorial).  Similarly, analysis of the work of the analysts themselves only really got going about forty years ago, when academic researchers started studying those who make forecasts about and recommendations on common stocks.

Should you want to get a crash course on the findings of those studies, a new working paper provides a wonderful survey.  It was authored by Mark Bradshaw of Boston College and it’s titled, “Analysts’ Forecasts:  What Do We Know After Decades of Work?”SSRN | The version of the paper I reviewed was dated June 2011.

As Bradshaw notes, “a seemingly disproportionate amount of research has focused on sell-side analysts” versus other investment decision makers, as of five years ago already amounting to over five hundred papers.  His own paper summarizes the main findings of that body of work and its evolution over time, including “what we know (or think we know)” about analyst behavior, via short summaries of key studies, grouped thematically.

Analysts were “elevated to the status of an economic agent in the capital markets” at least in part because information was readily available (about their earnings estimates and recommendations) and “academics very much enjoy analyzing distributions (i.e., means, medians, standard deviations, etc.) and correlations.”  However, Bradshaw sees other reasons for the attention:  Analysts are key intermediaries in the dissemination of market information and they can serve as proxies for other players in the investment ecosystem to study beliefs and actions.

Unfortunately, the research methods used in the academic studies “do not really measure the most interesting part,” the “analysts’ analysis.”  Which is not to say that the studies are unimportant or without merit, just that the “how” of it all ends up being inferred rather than examined.  Bradshaw touches on some promising forays into new avenues of research, but the bulk of the scholarship to date focuses entirely upon a couple of aspects of the job, those estimates and recommendations.

There are too many items of interest in the paper to do justice to them here, many of which I have dealt with previously in one way or another:

Are analysts too optimistic?  The body of work says, “Yes,” but Bradshaw considers that a “sweeping generalization” that doesn’t account for the “meet or beat” phenomenon,the research puzzle | See “taking another look,” which includes a link to Jeff Miller’s review of that issue on A Dash of Insight. the selection biases that come into play in choosing stocks to cover, or the issues related to the use of non-GAAP earnings by data services, which are subsequently used in the academic research.

Are analysts inefficient?  Studies of one sort or another have shown analysts to generally overreact or underreact to certain kinds of information, but isn’t that the nature of decision making in general?  And, as Bradshaw says, “the tyranny of measurement error contaminates our ability to draw strong conclusions.”

Do analysts face potential conflicts of interest?  Yes, indeed — six different kinds of conflicts are reviewed.  As in other areas, though, it’s crucial to look at the economic significance of the apparent conflict making broad statements about the implications of the findings.  (This area of study fits naturally with an examination of incentives, a subject that the author feels has been shortchanged in academic research.)

By virtue of the sheer amount of research that has been written, there is a sense that we understand analysts and how they behave.  But because of the design of that research, Bradshaw says that what we have are studies “of associations, not behavior.”  We end up knowing hardly anything at all about that critical “how.”

Furthermore, what results is “a gross mischaracterization of the analyst’s job function,” with it being thought of as creating earnings estimates and issuing recommendations.the research puzzle | I have written extensively about rethinking the components of an analyst’s job.  See, for example, “those darn analysts.” Those two tasks now rank at the bottom of the heap in terms of the things clients want from the Street, yet they still get virtually all of the attention from academics — and from bloggers and tweeters and the mainstream media.  Rarely has a high-profile job been so misunderstood by so many (partially because the firms themselves perpetuate that focus).

So decades of work by analysts and decades of work by the academics who have studied them, yet there is so much that hasn’t been revealed.tjb research | Although, the “how” which is missing from these analyses is the bread and butter of my consulting business. But if you want to see where we’ve been and get some ideas about where we should go, Mark Bradshaw’s paper is the place to start.