First, a conditional forecast is obtained using information on one or more
determinants of the autocorrelation coefficient of earnings.For example,
Brooks and Buckmaster (1976) focus on extreme earnings changes, Basu (1997)
examines negative earnings changes, and Lev (1983) identifies economic
determinants like barriers-to-entry in an industry, firm size, product type, and
capital intensity of a firm; also see Freeman et al.(1982) and Freeman and Tse
(1989, 1992).Recent studies estimating conditional forecasts include Fama and
French (2000) and Dechow et al.(1999).