Is financial statement information about future earnings changes impounded in returns?
The thesis provides empirical evidence on the predictive ability and information content of UK financial statement report numbers. Specifically, I investigate the Ou and Penman (1989a) finding for the US that financial statement numbers convey information about the sign of the one year ahead earnings change, and that this is not reflected in current stock returns. The main motivation, however, of the thesis is the suggestion by Greig (1992) that the Ou and Penman results are driven by variations in accounting ratios across industries. In addition, it is further examined whether the Ou and Penman results are valid only for negative and/or positive values of the earnings changes and whether the Ou and Penman lagged impounding is confined to large and/or companies. The thesis complements the existing U. K. literature by offering this predictive perspective on, and interpretation of, the incremental information content of financial statement accounting numbers. The main results of the thesis provide evidence for a "predictive information link" between some annual report numbers and future earnings changes. However, these annual report numbers capture the temporary and not permanent changes in current earnings, thus for the market to look for the incremental information about future earning changes in accounting numbers is not worthy in terms of money and costs. Only, in the stores industry, the %A in current ratio indeed captures permanent changes in current earnings. The thesis also provides evidence for a "lagged impounding" phenomenon for some of the numbers as well as a "size effect". The incremental information of these accounting numbers is not impounded in current returns. The financial statement information of large companies is earlier reflected in current returns than the information of small companies. The predictive information link is established by fitting binary one-year ahead earnings change prediction models as well as regression models to annual report data for the period 1980-88. The lagged impounding phenomenon and size effect are established by running multivariate regression earnings information models over the period 1980-88.