當市場存在不完全性及套利行為時,必須以非線性模型評估股價,這是原始Ohlson(1995)股權評價模型所忽略的。本文在保有原Ohlson評價模型所使用的帳面價值與異常盈餘以評估股價的精神下,額外導入帳面價值與異常盈餘的落後期項,且利用McMillan(2001)發展的外生平滑轉換自我迴歸(STARX)模型觀念下,建立一般化線性Ohlson(GLOM)及兩種非線性Ohlson評價模型(NLOM-Logistic及NLOM-Exponential),並比較原始線性Ohlson模型(LOM)與一般化線性Ohlson(GLOM)與兩種非線性Ohlson評價模型(NLOM-Logistic及NLOM-Exponential)預測能力的優劣。 實證結果顯示,導入帳面價值與異常盈餘的落後期項可以增進模型的解釋能力。在非線性Ohlson評價模型中,多數股價存在平滑轉換的現象,以NLOM-Exponential所推估之股價較為精確,而影響股價與股價報酬率變化的因素,在移轉前後區域差異甚大,證明Ohlson模式的線性資訊動態不僅在理論的建構上不合理,而且在實證的預測能力上亦不及非線性的NLOM-Exponential。
Under the incompleteness of stock market and risk-averse investor’s arbitrage behavior, the equity valuation model must be nonlinear. That is ignored by the original linear Ohlson (1995) model. Based on the concept of STARX model derived by Mcmillian (2001), this paper revises the original linear Ohlson model by permitting multi-lagged specification in book value and abnormal earning and adding nonlinear term representing gain in arbitrage. Then I set up the generalized linear Ohlson and two kinds of nonlinear Ohlson model (NLOM-Logistic and NLOM-Exponential) and compare the forecasting performance of these three valuation models with original liner Ohlson model. Empirical study shows that the explanation of the valuation models can be improved by adding multi-lagged specification in book value and abnormal earning. Most stock prices have the phenomenon of smooth transition. NLOM-Exponential has the best forecasting ability among the four valuation models and original Ohlson model is the worst one. The factors influencing price volatility and return between two regimes are different enormously.