This paper aims to identify the effect that accounting bias and economic goodwill have on the variation of the book-to-market ratio (BTM). Using an empirical valuation model to measure the fair value of net assets, I divide the BTM into two components that capture accounting bias and economic goodwill. Accounting bias (BIAS) measures the deviation of a firm's book value (BV) from the fair value of its net assets (FV) under biased accounting. Economic Goodwill (GW) indicates the difference between a firm's market value (MV) and its FV. BIAS reflects the effect of both accounting conservatism and historical cost basis on BV. With GW, the effect of growth can be included in MV, which FV fails to include, or GW may simply reflect how biased accounting numbers lead to market mispricing. Empirical results support my hypothesis that accounting bias and economic goodwill, under the mispricing, can explain most of the BTM anomaly. BIAS ((BV-FV)/MV) has a strong negative relation to future ROE over a nine-year period (the approximate median depreciation period), while GW ((FV-MV)/MV) has a less negative (even positive) relation over shorter terms. When one controls the effect of accounting bias, GW as the growth measure shows a positive relation to future returns for as long as a five-year period. In contrast, BIAS shows a positive relation only temporarily to short-term returns. These results provide two implications. The BTM anomaly for ROE cannot be an equilibrium phenomenon. Rather it is related closely to accounting bias and its effect upon market pricing. Ultimately, the BTM's short-term relation to returns is primarily attributed to accounting bias, whereas its long-term relation to returns is primarily influenced by economic goodwill, which captures growth.