Earnings aggregation and valuation

A fundamental attribute of accounting is the fact that the sum of earnings of shorter intervals represents the earnings of the longer interval of which they are a part. These aggregated earnings are expected to contain fewer measurement errors. Prior literature examines the effect of earnings aggregation focusing mainly on the contemporaneous exp lanatory power of aggregate earnings for returns; higher explanatory power, however, does not imply smaller magnitude of errors in inferring price from earnings. This study addresses the question of whether lengthening the interval for aggregating earnings reduces the magnitude of errors in inferring price. I use estimation samples to obtain estimates of the parameters that relate market prices and accounting data. These estimates are then applied to test samples to infer prices and assess errors. Contrary to expectations, the preliminary results show that, while increasing the contemporaneous explanatory power of earnings for returns, earnings aggregation does not reduce the magnitude of errors in inferring price from earnings in a cross-sectional setting. However, it is very conceivable that the effect of earnings aggregation differs dramatically for different types of firms. Extensions, implementing the tests for different sub-samples, based, for example, on the sign and variance of earnings, are likely to enhance our understanding of the effect of earnings aggregation on valuation.