Reducing Complexity

The IASB has made its discussion paper Reducing Complexity in Reporting Financial Instruments publicly available now. As complexity is a charge often raised against current accountancy trends by people who are either not aware of or uncomfortable with the more conceptual approach towards accounting used by the IASB, it would be interesting to observe the conclusions drawn in that paper. To make a long story short, I am quite happy with the outcome, namely that the DP accepts that fair value "seems" to be the only measure that is appropriate for all types of financial instruments, then going on to list no less than 24 different measurement methods for financial instruments (table 1). Some random thoughts in that context.

Complexity is bad. That seems to be the tenet of the proponents of reducing it. I tend to disagree, however. In line with Einstein's razor, there is a place for complexity in financial reporting where it faithfully represents economic reality. Complex economic reality necessitates an adequate level of complexity in its description, lest it over-simplifies. Where, however, complexity is merely a consequence of accounting artefacts, it may safely be dispensed with.

Fair value vilified. It is interesting to observe how some people try to make a case against fair value from the current financial market crisis, even though it is quite obvious that this is a clear case of trying to kill the messenger. This position is particularly difficult to understand coming from people with a background in engineering or science. There, it's evident that two variables can only be combined if they have the same unit of measurement: you cannot add metres and kilograms. In accounting, though, that seems to work just fine - it's currency units, after all. Or is it? Notionally, it certainly is, but I'd challenge anyone out there that a $'s worth of an asset measured at cost is very different from the same $ measured at, say, fair value. In short, the unity of measurement is an illusion under the mixed attribute model.