(See a correction at the end of this post.)
The CFA Institute has recently published a report that summarizes the findings of its member survey on the FASB and IASB’s proposals for recognition and measurement of credit losses. The two most eye-catching statistics are these:
- More respondents (46%) favor a fair value model than any other. A “current expected credit loss” model was favored by 41 percent, with the rest mainly split among “undecided” and the status quo.
- A whopping 92 percent favor a converged solution.
The proponents of current value accounting for loans and debt securities should welcome data that suggests a great deal of opposition to the new models that the boards are proposing, but there are also some obvious weaknesses in the survey methodology, as well as the way the results were reported, that will also give the boards adequate justification to disregard the feedback and continue down their chosen paths.
In order for any survey to produce valid information for decision making, it needs to be constructed so as to bring ‘noise’ in the survey responses down to an acceptable level. In this case, it is not sufficient just to know that the respondents are CFAs. Additional demographic information is necessary to be able to reasonably assume that the respondents are drawn from the population of interest, and that they are likely to constitute a representative sample of that population.
Specifically, the number of respondents was reported to be around 360, but CFAI membership is way higher than that – about 110,000. The report does not disclose how many CFAI members were asked to participate in the survey, how they were selected to be invited to participate, the response rate, and the demographics that would suggest a level of expertise and experience in the specific topics that are being surveyed.*
Since CFAI provided no disclosure regarding any of these questions, I am concerned that invitations to participate in the survey were made without regard to experience and specialization; and moreover, that response rates were very low. If that were the case, a reader of the report should question whether the sample is representative of the population of interest. For me, that population would be the subset of CFAI members who are experienced and independent analysts of banks and similar financial institutions. And by “independent,” I mean that they are not employed by these institutions.
Setting aside those concerns for the moment, however, my biggest disappointment is that CFAI does not separately disclose summaries of the responses of U.S.-based participants. I realize that CFAI brands itself as a global association; but my dog in the hunt, which should be the same as US-based CFAs, is for the FASB to come to the answers that best serve stakeholders in the U.S. capital markets. And by the same token, I expect that IASB members would be primarily interested in the views of non-US CFAs.
As to the convergence question, the result that 92 percent of respondents prefer that the FASB and IASB come up with a converged solution is certainly eye-popping. However, it illustrates how important it is to ask questions that are not burdened by implicit assumptions.
I have little doubt, for example, that if CFAI had asked, “are you in favor of world peace?”, the positive response rate might be only slightly higher than 92% – and for similar reasons. Everybody wants world peace, but not unconditionally. I doubt very much that US respondents to the survey actually intend to convey the impression that they would sacrifice lower quality US GAAP in order to attain convergence.
As it turns out, recent developments show that my concern is much more than mere academic nitpicking. Reuters has reported that the EU is dissatisfied with any approach to loan accounting that does not give a bank’s executives the flexibility to set up cookie jar loan loss reserves – in the name of “prudence.”
That bit of news in and of itself should be more than enough to convince the FASB to remove the term “convergence” from its vocabulary once and for all.
*I incorrectly stated that the the survey didn’t disclose the population sampled or the response rate. In the penultimate paragraph of the report, it is disclosed that 16,001 members were invited to participate, and 361 responded, for a response rate of 2% and a margin of error of +/- 5%.
I apologize for that oversight, although it does not change my conclusions. The response rate is so low that it renders any measures of statistical significance meaningless, unless you can demonstrate that the 2% is a random draw from the population of interest. It is highly uncertain that this is the case; and I don’t see that anything was done to test that critical assumption. However, in private correspondence with CFAI, they wrote, “Our survey was also supplemented by direct outreach to investors. The survey results echo what we heard through such direct outreach.”