Autocorrelation in forecasting abnormal profits: evidence for “other information” and tests on the third premise of the Ohlson model

Abstract

The model proposed by Ohlson (1995) is based on the combination of the profits with the book value as the basis for the valuation and provides other perspectives for the use of the Residual Profit Assessment model. In this model, the variable vt defined as “other information”, not captured by Accounting, in addition to the profits, which impact future abnormal profits. It is a variable of difficult specification, for which, in the verification of the model, several proxies have been used by researchers. In this context, assuming that first-order autocorrelation captures all the information about abnormal profits, as described by Ohlson’s third premise (1995), this research aims to identify and analyze the persistence and how the inclusion of terms moving averages as the “other information” variable (vt) helps to answer the questions raised about this variable and improves the use of the Ohlson Model in predicting abnormal profits of companies listed in B3 (Brasil, Bolsa, Balcão). The ARIMA (Autoregressive Integrated Moving Average) model was used, with quarterly time series data, in a sample of 39 companies, from the first quarter of 2002 to the fourth quarter of 2015. The forecasts were made after verification that parameters of the classic model were attended. We conclude that the moving average models were able, on average, to reduce the serial correlation and to provide more accurate forecasts of the abnormal profits.