The Sensitivity of Tests of Asset Pricing Models to the IID‐Normal Assumption: Contemporaneous Evidence from the US and UK Stock Markets

Abstract

Standard tests of asset pricing models are based on the iid‐normal assumption. We compare standard test results with those obtained from procedures that do not require iid‐normality. Analysing unconditional and conditional asset pricing models, we find that the use of tests that consider departures from the iid‐normal assumption affect probability values, sometimes by a considerable amount but that test outcomes are not affected. The results also suggest that issues surrounding the testing of joint hypothesis influence probability values and that the use of appropriate tests may be more important when analysing US data than when analysing UK data.