Since the existence of big bath behaviour would confound our analysis, we attempted to identify and exclude those firms engaging in such behaviour. Our first approach was based on the uni-directional definition of big bath behaviour, and excluded the four observations where max Ei 0.7EE (a judgernental cutoff). Second, we also excluded thirty-eight observations where min Ei 1.3EE, leaving a sample of 121. Neither procedure improved the overall significance or explanatory power of the model (although the results remained fairly stable) and therefore the results are not reported here. This suggests that big bath accounting is not present in sufficient strength to impair the explanatory power of the model.