Previous research suggests an anomaly on the relationship between price and earnings of companies reporting negative earnings (i.e., losses). That is, price and earnings are negatively correlated; the more negative a firm's earnings per share, the higher its stock price. However, Collins et al. (1999) find that the anomaly of negative earnings becomes insignificant after book value of equity is included as an explanatory variable in the price-earnings model. Because the assumptions of Collins' model do not exist in reality, its implications and empirical results are suspicious. This study uses a multi-period empirical model of Ohlson (1995) and compares its explanatory power with that of Collins. The findings show that, with few exceptions, the multi-period empirical model of Ohlson (1995) is superior to Collins' model, that the anomaly of negative earnings does not exist, and that the hook value of equity really plays an important role in explaining price.