An evaluation of the predictive value of bank fair values
This thesis examines whether the net asset fair values of banks possess predictive ability for the banks’ future cash flows and earnings. This is an important issue considering the arguments for and against the wider use of fair value accounting for banks’ financial instruments and the claim by some that fair values during economic recessions (where markets may be illiquid) are irrelevant and largely unreliable. A number of studies have found that the explanatory power of bank fair values when compared to traditional historical cost are more value-relevant based on capital market reactions. However, there is a very limited literature on how bank fair values are related to the future performance (e.g. earnings and cash flow) of banks. This study fills this gap by providing empirical evidence on the relationship between U.S. bank fair value disclosures and banks’ future performance as measured by operating cash flows and earnings over a three-period future horizon. Furthermore, the thesis provides evidence on the relationship between bank fair values, in terms of the levels classification introduced during the 2008 global financial crisis, and the future performance of banks, thus showing whether market illiquidity affected the underlying relationships. The study examines two distinct periods. The first study period, 1996-2005, was based on annual data of banks with minimum total assets of $US150 million as of year 1996. The second study period from 2008-2010 (this period encompassed the global financial crisis period and also the levels classification of bank fair values according to SFAS 157), was based on quarterly data of banks with minimum total assets of $US150 million as of the first quarter of 2008. The thesis provides strong evidence that there is a predictive relationship between bank fair values and future bank performance. The evidence is strong during the first study period from 1996 to 2005 where the current net asset fair values of on-balance sheet financial instruments of banks were significantly associated with future operating cash flows and operating earnings of such banks over a three-year future time horizon. However, the predictive relationship between net asset bank fair values and operating cash flows is stronger than the predictive relationship between net asset bank fair values and operating earnings. In the second study period, from 2008 until 2010 the empirical results show strong evidence that there is a predictive relationship between level 1 and level 2 bank fair values and future operating cash flows. The findings from the empirical results were that the current quarter’s level 1 and level 2 net asset fair values of banks were significantly associated with the future quarters’ operating cash flows of such banks. The level 3 net asset fair values of such banks in most cases were not significantly associated with the banks’ future quarterly operating cash flows. The corresponding relationships for operating earnings were that the current quarter’s level 1 net asset fair values of banks were positively associated with the future quarters’ operating earnings of such banks. However, the level 2 net asset fair values of banks were negatively associated with the future quarters’ earnings of such banks. This result is in contrast to the results obtained when the predictive relationship between level 2 bank fair values and future operating cash flows was evaluated, where it is found that both level 1 and level 2 net asset bank fair values are positively related to future quarterly bank cash flows. Further empirical analysis showed that a possible reason behind this disparity was that there was a structural change in the relationship between bank operating cash flows and operating earnings over the course of the first and second study periods, where, in particular, for the second study period (which includes the period of the global financial crisis) there was a systematic downward bias in operating earnings relative to the operating cash flows of the sampled banks. This in turn makes operating earnings a poor proxy for operating cash flows during the second study period. The findings from this study provide confirmation that net asset fair values have predictive ability as argued by Ball (2008); Barth (2006b) and Tweedie (2008). The study findings that net asset fair values have predictive ability is consistent with the FASB’s view that the asset values shown in firm financial statements should communicate information about the potential future financial performance of the affected firms (FASB 2010:17). Furthermore, the study also confirms that objectively determined bank fair values based on market prices rather than model based bank fair values provide greater predictive value in relation to future performance as measured by operating cash flows. Lastly, this thesis showed that during the first study period (where there was no financial crisis) that bank size, capital adequacy and growth prospects, had little impact on the results obtained, while for the second study period, there were cases where bank size and bank capital ratios did have a significant impact on the predictive relationship between bank fair values and future cash flows. The study contributes to the fair value accounting and accounting standard-setting literature and highlights that fair values have predictive ability, especially with respect to future operating cash flows of banks, both during and outside of periods of financial crisis.