Once upon a time, almost all accounting research used armchair reasoning to support claims about what Generally Accepted Accounting Principles should look like.  But in 1968, Ball and Brown published a paper that started a trend toward systematic collection of evidence on how financial information is used.  I always thought that the ultimate goal  was to establish a body of ‘positive’ research that establishes what is, thus preparing us to provide more intelligent normative suggestions on what should be.

However, this week’s JAE conference makes me wonder whether I interpreted this history incorrectly, and in fact the goal is to never, ever make normative statements.  My first inkling of this was Wayne Guay’s presentation on corporate governance (paper with co-authors Armstrong and Weber).  The paper reviewed hundreds of positivist research papers, but in his presentation and Q&A, the authors and discussant shied away from making any claims that some forms corporate governance are good and others are bad, referring to this possibility as ‘a myth.’  The authors and discussant emphasized that context matters, and who could disagree.  That would still suggest that in particular situations, some practices should be more effective and appropriate than others. However, the authors didn’t seem eager to spell those out.

The Kothari, Ramanna and Skinner paper and discussion show an even stronger aversion to admitting normative urges.  I called the paper “abashedly normative” during Q&A because it answers the normative question “What Should GAAP Look Like?”, but denies it is normative:

The analysis is entirely positive, and still we make unambiguous recommendations about what GAAP should look like. As Jensen (1983, p. 320) explains, policy questions are best answered with “knowledge of a wide range of positive theory,” which is what we hope to provide. Watts and Zimmerman (1986, Chapter 1, original emphasis) explain how an exercise such as ours can be positive, not normative:

“Prescriptions require the specification of an objective and an objective function. For example, to argue current cash equivalents should be the method of valuing assets, one might adopt the objective of economic efficiency (i.e., the size of the economic pie available) and specify how certain variables affect efficiency (the objective function). Then one could use a theory to argue that adoption of current cash equivalents will increase efficiency. Theory provides a method for assessing this conditional statement (i.e., do we observe that adoption of current cash equivalents increases efficiency?). But theory does not provide a means for assessing the appropriateness of the objective. The decision on the objective is subjective, and we have no method for resolving differences in individual decisions.”

I agree that judgments over the appropriate objective of accounting standards (or any other policy, whether voluntary or imposed by government) are necessarily normative.  But even if I outsource the statement of an ultimate goal (“the purpose of financial reporting is X”) to someone else, I am still marshaling positive evidence to argue for how GAAP should achieve that goal. Moreover, there is no end to compromise and balance in policy recommendations.  Even if I accept every aspect of the Conceptual Framework’s statement of objectives, stating what GAAP should look like still requires me to judge whether the fair value of a particular asset is sufficiently relevant to trump concerns about reliability.

To see that most people would view this is a normative exercise, we know right where to go:

In philosophy, normative statements affirm how things should or ought to be, how to value them, which things are good or bad, which actions are right or wrong. Normative is usually contrasted with positive (i.e. descriptive, explanatory, or constative) when describing types of theories, beliefs, or propositions. Positive statements are factual statements that attempt to describe reality.

Reluctance to call normative recommendations by their commonly held name makes me uncomfortable for many reasons. First, it suggests that we should be ashamed of making normative statements as part of our jobs.  What could be further from the truth?  We have no compunction in making normative statements when we write reviews (“the authors should …”), advise students (“you should narrow your focus to …”), and hopefully, when we write review papers (“future authors should examine …”).   And given that we academics are paid by philanthropic donations and State governments to conduct research, we should view it as our duty to make normative statements that will help people making the business and policy decisions that shape our world.

Second, and somewhat more worrisome, claiming that some types of normative statements are actually positive is likely to make discussions of policy even more contentious than they already are.  How can positive claims about “what GAAP should look like” not seem more objective and unassailable than normative ones.

Now, I am the first to agree that normative claims are generally more suspect than positive claims, in the commonly understood meaning of the terms.   I am perfectly comfortable with the status quo that peer reviewed articles in our top journals simply make positive statements and stop, or else signal caveats with sirens and flashing lights when in our conclusions we write “Given our positive results, we recommend …. “.

I stand tall and call proudly for more normative claims, which are a central part of our responsibility as researchers.  I’d like to see all of the papers presented at the JAE conference make their normative claims unabashedly, whether they are directing their recommendations toward policy makers or positive researchers.  Of course, I would be particularly pleased if KRS would revise their paper to provide more direct normative answers to the two questions that lie at the heart of FASRI:

  • Given the available positive research, FASB should ….


  • Given FASB’s upcoming agenda, researchers should conduct positive research examining ….

Anyone else interested in these questions should join us.