股票市場效率與經濟發達程度之關係──遺傳演化神經網路研究

Relation Between Stock Market Efficiency and Degree of Economic Development – A Genetic Neural Network Approach

效率市場假說主張技術分析不能從市場中獲得超額利潤。此一主張長期以來支持與反對各有實際研究支持。本文提出一個假說：市場效率與經濟發達程度成正比，因此在經濟發達程度高的國家市場效率高，無法獲得超額的報酬；但在經濟發達程度低的國家市場效率低，可能可以獲得超額的報酬。由於每個市場適用的技術分析方法可能不同，因此，本研究以具有學習能力的遺傳演化神經網路建構使獲利最大化的交易決策系統。並提出以「市場效率指標」定量衡量市場的效率性、以人均國內生產毛額衡量經濟發達程度，而以包含各種經濟發達程度的22個國家1997~2006共九年的股票市場為樣本，進行實證研究。研究結果顯示，經濟發達程度越低的國家，其股票市場在考慮交易成本下使用技術分析獲得超額利潤的可能性越高。在測試期間，人均國內生產毛額可解釋市場效率指標61%的變異，F統計達0.002%顯著水準，可見市場效率與經濟發展程度有關。

The market efficiency hypothesis advocates that technical analysis cannot obtain the excess profit from the market. There have been many real studies that supported and refuted to the hypothesis, respectively, for a long time. This paper proposes a hypothesis that market efficiency is proportional to the degree of economic development. In order to confirm this hypothesis, this study constructed a trading decision-making system that maximized profit by using genetic neural networks with the learning ability, and proposed “the market efficiency index” to measure the market efficiency and per capita GDP to measure the degree of economic development. Twenty two countries, representing all degrees of economic development and nine years (1997~2006) of stock market data were used as the samples to conduct the study. The findings showed that the lower the economic development, the higher the possibility of obtaining obvious excess profit by using technical analysis while considering transaction cost. In the test period, the per capita GDP may explain 61% of the variation of the market efficiency index, and the F statistics reach a remarkable 0.002% level; therefore, the economic development degree obviously affects the market efficiency.