How Naive is the Stock Market's Use of Earnings Information?

Rendleman Jones and Latane (1987) and Bernard and Thomas (1990) report evidence supporting their hypothesis that investors use a quot;naivequot; seasonal random walk model in forming expectations of quarterly earnings. Using the Bernard and Thomas (1990) data we show that the market acts as if it: (1) does not use a seasonal random walk model; (2) does incorporate past earnings changes in forming expectations; (3) does use the correct signs in exploiting serial correlation in seasonally-differenced quarterly earnings; but (4) underestimates the magnitude of the serial correlation. This evidence remains anomalous in the sense that it is consistent with neither the theory of efficient markets nor the quot;naive expectation modelquot; hypothesis nor quot;behaviorial financequot; theories