- 學位論文
股票與不動產市場泡沫崩盤之偵測機制-噪音交易與能量觀點
Detection of the Bubble Crash in Financial and Real Estate Markets-Noise Trading and Energy Perspectives
摘要
本文以噪音交易(Noise trading)與能量觀點藉由相關文獻提出關於噪音交易者(Noise trader)於崩盤前異常行為的假說，並以HHT時頻分析方法分析股票市場以及不動產市場，觀察高頻波動於接近崩盤時是否與正常時期有顯著差異，更進一步提出兩個偵測崩盤的指標：高頻波動(HFV)以及高頻能量比重(PRH)，並以此二指標對於歷史資料進行分析。 自各市場HFV和PRH指標圖中，可明顯看出高頻分量波動以及能量比重在測試段相對於對照段於接近崩盤時有顯著的增加，並以ROC(Receiver operating characteristic)曲線對指標進行區別力效度檢驗，其AUC(Area under curve)值為0.9436，代表指標於本文所研究的指數對於偵測崩盤與沒崩盤有極好的區別力，藉此可驗證本文所提出的假說。 最後將指標做應用，提供指標在不同國家，不同門檻值(threshold)下的型Ⅰ、型Ⅱ錯誤率，並將不同門檻值對應到各市場，提供訊號超過此門檻值時平均與實際崩盤之距離，以提供不同投資期距與不同風險偏好程度的投資人選擇適當的崩盤偵測機制，以便投資人選擇適宜的退場時機，對於政府而言可藉由此偵測機制及早因應崩盤所帶來的各種經濟、社會、政治問題。
並列摘要
This paper proposes the hypotheses about noise traders’ irrational behavior before collapsing from noise trading and energy perspective and employs the HHT, which is a time frequency method, to observe whether there is a significant difference between regular fluctuations of markets and crashes in high frequency volatility in the stock and real estate markets. Furthermore, the paper presents two indicators, which are HFV and PRH to detect the financial crash and use these two indicators to analyze the historical data. We observe that the value of indicators increasing tremendously before collapsing from two indicators’ graphs and use the ROC (Receiver operating characteristic) curve to verify our indicators’ discrimination and the AUC (Area under curve) value, which equals 0.9436. The result shows that the indicators have an excellent discrimination between detecting collapse and non-collapse. Finally, we apply the indicators in different indexes and offer the typeⅠ、typeⅡ errors under different thresholds to investors who have different risk preferences so that investors can choose a proper detection mechanism and exit the market in time before collapsing. With the detection mechanism, the government can make precautions for the economic, social, and political problems that collapse brings.