FORECAST OF RETURNS AND ACCOUNTING GROUNDS IN TIME OF WORLD CRISIS: A STUDY OF PORTFOLIOS USING AUTOREGRESSION VECTORS

Abstract

The objective of this research was to verify whether, in times of global economic crisis, the use of accounting indicators of liquidity, capital structure and profitability possess the ability of generate more profitable investments then benchmarks such as “no risk†asset returns and the return of the market. This analysis was made through the use of forecasts with models of autoregressive vectors (VAR). The accounting indicators and the returns of stocks were used as input data to generate forecasts and consequent formation of investment portfolios - the time series analysis starts on 30/03/1994, ending up at 30/09/2011 with quarterly data. Within a sample of 20 companies, the investment portfolios were composed of the 5 companies with higher expected profitability. Subsequently we compared the actual profitability of the portfolios against the CDI return (riskless asset) and the return of Ibovespa (market return). The results showed that, among the indices studied, the Net Margin and the Return on Equity have greater predictive power, concluding that the indicators that use Net Income in its composition provide more accurate forecasts of stock returns. Moreover, in the comparison of the returns of portfolios formed by VAR models against the return of the market (Ibovespa) and one riskless asset (CDI), in a general way, there was superiority in the profitability on both Ibovespa and CDI, concluding then about the relevance of information in times of crisis.