As you might know from this call for survey participants, Regenia Cafini of the FASB is collecting data on several facets of “field test” she is conducting as part of the Financial Statement Presentation project.  As Regenia pointed out to me today, she is an accountant, not a statistician, and she would be “very keen” to share the data with academic researchers who could approach it from a more statistical angle.  Read on for details.

What are the data sets?

There are actually three distinct but related data sets.  Regenia has worked with about 30 companies (preparers) that have restated their 2007 financial statements to comport with the recommendations of the Financial Statement Presentation Discussion Paper.  Thus, the first data set simply includes financial statement information for each company:  the initially-reported version and the restated version.  It is possible that a researcher who wants to work with this data would have to sign a non-disclosure agreement, because public discussion of the full data set might violate Regulation FD.  (Regenia is working with the SEC to get a comfort letter on that point, but that is a story for another day.)  It might be possible to work with a sanitized version of the data, but my own impression is that researchers are going to want access to information that would require a promise of non-disclosure.

After the restatement exercise, Regenia had the preparers answer an extensive survey on their experience with the restatement, how well the restated financials communicate information about the firm, and other matters.  This results in a separate data set, tied to the financials themselves, but quite distinct from it.

Finally, there is the data from the survey of credit analysts, equity analysts and academics (and maybe their students), who are provided with either restated or original financials, and are asked to make a variety of judgments and decisions about how they interepreted the statements, what they thought was important, etc. (Demographic data are also collected.)

The first two data sets are quasi-experiments, in that the variation across industry is likely to influence both the financials and the preparers’ responses.  This last survey is likely to be analyzable as a controlled experiment, because participants will receive some restated and some original financial statements, so you should be able to draw fairly clear inferences from differences in the reactions to different statements for the same companies.

What’s in it for me?

The Financial Statement Presentation project team plans to write a report to the joint Boards about their own analysis, and I am sure they would like to learn what they can from a more statistical/academic analysis.  However, you would be able to use your analyses as part of your own research paper(s).  Would analysis of these data sets be enough for an “A” journal?  I think there is a good chance they are.  However, keep in mind that the surveys were written to inform deliberations, not to provide incremental contributions to the academic literature.  But overall, they look pretty solid to me, and I think the timeliness of the issue and the novelty of the data set will allow creative researchers to pull together a paper that would be of interest to many journals, if not necessarily TAR, JAE and JAR.  Naturally, we understand that these data sets are less valuable if they are given to everyone; also, we want to limit how much time the project team needs to spend answering researchers’ questions.  For these two reasons (along with the nondisclosure issues mentioned above) we will be selecting a single research team to have access to the data.

What tools do I need?

Researchers trying to get a publishable paper out of this data would need to have a good deal of skill manipulating and analyzing data from financial statements, surveys, and experiments.  And of course, some serious interest in Financial Statement Presentation.  This strikes me as an excellent opportunity for a research team that includes at least one archival and one experimental researcher.

I’m interested!  What next?

Send an email to Rob Bloomfield, rjb9 (at) cornell (dot) edu to express your interest.  [If you are wondering why I don't use @ and . in my email address, it's because I am trying to avoid website scrapers that will then add the email to their spam lists.]