An experiment in fair value accounting: UK investment vehicles

We use the British real estate and investment fund industries as experimental settings where historic cost (HC) and fair value accounting (FVA) can be compared. Both industries have the majority of their assets marked to market and hence the difference between the two accounting systems is profound. However, as the valuation of real estate is arguably more subjective than that of investment funds, we are able to contrast fair value accounting in a near ideal setting with one where it remains important, but where valuation difficulties may permit bias. As this distinction is incorporated in the recently issued SFAS 157, which also formed the basis of the IASB's relevant discussion document, the results of our study may be particularly timely. As expected, we find that fair value income is considerably more value relevant than historic cost income. However, in the presence of changes in FVA balance sheet values, income measures become largely irrelevant. This implies that there is no obvious advantage from adopting FVA income accounting if FVA balance sheet values are available to the user. Furthermore, FVA for our real estate sample is considerably less value relevant than for the investment companies and the evidence for this sample, if not conclusive, is consistent with earnings management. We interpret these results as confirming that fair values are highly relevant and largely unbiased where the values are unambiguous. Where valuation is ambiguous, which will normally be the case, value relevance will be lower and biased accounting may be revealed.