This study investigates whether fair value accounting contributes to the procyclicality of bank lending. Using banks’ approval/denial decisions on residential mortgage applications to capture banks’ supply of credit, I find no evidence that fair value accounting has procyclical effects on bank lending over the past two business cycles. I further identify two reasons for this result. First, the main accounting item distinguishing fair value accounting from historical cost accounting—unrealized gains and losses on available‐for‐sale securities—does not affect lending decisions. Second, unrealized gains and losses on available‐for‐sale securities are not procyclical, as the risk‐free interest rate rises during some expansionary periods, resulting in unrealized losses, while the risk‐free interest rate (and sometimes the default spread) falls during some recessionary periods, resulting in unrealized gains.

We examine the real effects of FAS 166 and FAS 167 on banks’ loan‐level mortgage approval and sale decisions. Effective in 2010, these standards tightened the accounting for securitizations and consolidation of securitization entities, respectively, causing banks to recognize an estimated $811 billion of securitized assets on balance sheet. We find that banks that recognize more securitized assets exhibit larger decreases in mortgage approval rates and larger increases in mortgage sale rates. These effects significantly exceed those of banks’ off–balance sheet securitized assets, consistent with our results being driven by the consolidation of securitization entities rather than by securitization per se. We conduct tests that help rule out the financial crisis as an alternative explanation for our results. Further analyses suggest that mechanisms underlying the results include consolidating banks’ reduced regulatory capital adequacy, increased market discipline, and consequent desire not to recognize high‐risk mortgages on balance sheet.

This paper studies the ability of long-run risk models to explain out-of-sample asset returns during 1931–2009. The long-run risk models perform relatively well on the momentum effect. A cointegrated version of the model outperforms the classical, stationary version. Both the long-run and the short-run consumption shocks in the models are empirically important for the models' performance. The models' average pricing errors are especially small in the decades from the 1950s to the 1990s. When we restrict the risk premiums to identify structural parameters, this results in larger average pricing errors but often smaller error variances. The mean squared errors are not substantially better than those of the classical CAPM, except for Momentum.