本研究採用Sloan(1996)架構，以Mishkin(1983)測試市場效率性，欲探討備供出售金融資產未實現損益（以下簡稱未實現損益）之價值攸關性，研究結果如下：盈餘預測模式(earning forecasting model)顯示，從未實現損益之變化觀察出金融控股公司似乎有Shefrin and Statman(1985)處分效果（即當期備供出售金融資產未實現利益會於下一期實現，增加下期備供出售金融資產已實現利益），因此當期未實現利益能預測下一期盈餘，而銀行業則無處分效果，其未實現損益不具預測性；合理定價模型(rational pricing model)顯示，只有金控公司未實現損益對異常報酬率達到邊際(10%)顯著，且異常報酬率為隔年六月至後年五月的累積異常報酬率(CAR)與購買並持有報酬率(BHAR)才存在此現象，而不管是銀行業或是整體樣本來看，未實現損益並未被合理定價；結合二個模型測試市場效率性，發現備供出售金融資產未實現損益不具市場效率性，而金融工具評價損益則具有市場效率性。 / This study investigates the value relevance of unrealized gains and losses on available-for-sale securities(hereinafter “UAFS”) by adopting the framework of Sloan(1996) and using Mishkin Test(1983) to examine the market efficiency. The reults are as follows: The earnings forecasting model shows that financial holding companies’ UAFS is predictive of one-year-ahead earnings. It seems that the change of UAFS of financial holdings consists with the theory of disposition effect of Shefrin and Statman(1985) which suggests that investors tend to realize the gains of securities in the next period while hold loss securities. Therefore, it is likely that UFAS could forecast next period earnings of financial holding companies. The rational pricing model shows that only UAFS of financial holdings are rationally priced with marginal significance through abnormal returns (CAR and BHAR) for the period beginning five months after the firm’s fiscal year-end through the fifth month after the next fiscal year-end, whereas the coefficients of banks or all sample are insignificant. Combinig with the two models to test market efficiency, this study finds that unrealized gains and losses on available-for-sale securities are inefficient while the valuation gains and losses on financial instruments are efficient.