透過您的圖書館登入
IP:54.86.180.90
  • 學位論文

金融市場的異質代理人建模

Essays on Heterogeneous-Agent Models of Financial Markets

指導教授 : 陳思寬

摘要


傳統經濟學運用代表性個人 (representative agent)、經濟人 (Homo economicus )、 理性預期 (rational expectations) 等建模方法, 在經濟理論的建立上, 有極為豐碩 的成果。這類方法的好處是易於數學上的分析處理, 對真實世界的經濟現象確 實有一定的解釋能力, 也提供建模者一些重要的洞見 (insights)。然而, 在經濟 建模的過程中, 若持續忽略人的有限理性、異質化行為、學習適應行為, 也將會 使得互動、協調、環境反饋等重要元素在經濟模型中付之闕如, 這類模型要拿 來解釋例如金融市場上的異常現象 (anomalies), 就會顯得捉襟見肘。近年來, 經濟學與財務金融學出現典範移轉 (paradigm shift), 具有限理性的異質代理人 模型成為主流, 探討學習適應、互動協調等個體行為如何產生和真實世界相近 的總體現象。 本文採用三種不同類型的異質代理人模型, 對金融市場進行建模分析, 分 別是分析方法 (analytical approach)、實證方法 (empirically based approach) 與 模擬方法 (simulation approach)。第一個議題, 我們在一個簡單的股票市場模 型中, 引入擁有資訊的理性投機者 (informed rational speculators)、動量交易者 (momentum traders) 與反向交易者 (contrarian traders) 三種異質交易者, 以分 析方法分析各種交易者的交易行為如何影響股價, 並分析不同外生參數設定 下, 三種異質交易者的獲利性。 第二個議題, 我們考慮有兩檔股票存在的股票市場, 運用台灣股票市場的實 證資料, 以電子類報酬指數 (electronics sector index TRI) 與未含金融電子股報酬指數 (non-finance non-electronics sub-index TRI) 代表兩種不同類型 (style) 的股票, 運用基於實證的信念適應性系統 (empirically based adaptive belief system) 這種建模方式, 以觀察基本分析者 (fundamental traders)、技術分析者 (technical traders) 和轉換者 (switchers) 這三種交易者在市場中人數的市場占 比 (market fractions) 演變狀況, 並探究轉換者類型投資 (style investing) 型為和 類股輪動 (sector rotation) 的可能關係。 第三個議題, 我們運用 AIE-ASM 軟體模擬一個人工股票市場 (artificial stock market), 市場中具有基於遺傳規劃的眾多異質自主智能交易者 (GP-based heterogeneous autonomous traders), 我們觀察人工股票市場中的價量關係 (stock price-volume relation), 並與真實股票市場進行對比, 最後探究人工股票 市場中個人交易行為與股票市場總體現象的關係 (micro-macro relation)。

並列摘要


Traditional economics achieved fruitful results in constructions of economic theories through modeling methods such as representative agent, Homo economicus and rational expectations. The advantages of such kind of methods are the ease in mathematical deduction and analysis, the existence of a certain degree of explanatory ability of the real-world economic phenomenon, and the provision of critical insights to the modelers. However, continuous ignorance of the bounded rationality and heterogeneity and learning and adapting behaviors of human beings in the modeling process will lead to the lacking of essential elements of economic models including interactions, coordination and environment feedbacks. These types of models run short in explaining, for example, anomalies in financial markets. Recently, paradigm shifts have emerged in economics and finance, where models with heterogeneous-agent models with boundedly rational agents have become mainstream in discussing how behaviors of learning, adapting, interacting and coordinating produce macro phenomena similar to that in the real world. Here we adopt three different types of heterogeneous-agent models to model and analyze financial markets, which are the analytical approach, the empirically based approach and the simulation approach. In the first essay, we introduce informed rational speculators, momentum traders and contrarian traders into a simple stock market model and use analytical approach to analyze how trading behaviors of each kind of traders affect stock prices, and also analyze the profitability of the three types of heterogeneous traders under different sets of parameters. In the second essay we consider a stock market with two stocks existing. We use empirical data from Taiwan Stock Exchange (TWSE) and let electronics sector index TRI and non-finance non-electronics sub-index TRI represent two different styles of stocks, and we use a modeling method of empirically based adaptive be- lief system to observe the evolution of market fractions of fundamental traders, technical traders and switchers, and look into the possible relationship between style investing behaviors and sector rotation. In the third essay we use the AIE-ASM software to simulate an artificial stock market, in which GP-based heterogeneous autonomous traders exist. We observe the stock price-volume relation in the artificial stock market and compare that to the real stock market, and finally look into the micro-macro relation between individual trading behaviors and the macro phenomena in the artificial stock market.

參考文獻


De Long, J.B., A. Shleifer, L.H. Summers, and R.J. Waldmann (1990a), “Noise traders risk in financial markets,” Journal of Political Economy, 98, pp. 703–738.
Alchian, A. (1950), “Uncertainty, evolution and economic theory,” Journal of Political Economy, 58, pp. 211–221.
Amilon, H. (2008), “Estimation of an adaptive stock market model with heterogeneous agents,” Journal of Empirical Finance, 15, pp. 342–362.
Anufriev, M., T. Assenza, C. Hommes, and D. Massaro (2013), “Interest rate rules and macroeconomic stability under heterogeneous expectations,” Macroeconomic Dynamics, 17, pp. 1574–1604.
Arifovic, J. (2000), “Evolutionary algorithms in macroeconomic models," Macroeconomic Dynamics 4, pp. 373–414.

延伸閱讀