本研究以台灣槓桿型ETF及反向型ETF為研究對象，比較其與傳統型ETF在不同市場情緒下，追蹤績效之差異，為客觀定義市場情緒狀態，運用兩區域門檻自我迴歸（Threshold Autoregressive, TAR）方法針對恐慌指數進行估計，將市場情緒區分為悲觀及樂觀等2種狀態，以觀察在不同市場情緒下追蹤績效之差異，另考量中美貿易戰之發生，全球經濟產生結構性的變化及衝擊，故以中美貿易戰之起始日區分，估計前後兩期間之門檻值，捕捉因時而異的門檻效果。鑑於中美貿易戰對全球經濟產生結構性變化及衝擊，本文亦檢測VIX作為市場情緒反應及價格發現指標的可參考性。
其次，考量槓桿型ETF及反向型ETF有別於傳統ETF以直接持有標的商品方式複製損益，係以投資期貨等衍生性商品方式構建槓桿，追蹤持有標的分屬現貨市場及期貨市場，其交易制度、流動性及投資人操作需求均有所不同，將影響ETF之追蹤能力，故本文利用DCC-GARCH模型檢測各類ETF與其追蹤標的市場之是否存在非對稱關係，並將前述門檻差異納入模型，期透過捕捉動態相關係數之變化，了解追蹤績效的變化。另本文並探討不同重大經濟事件內，ETF與其追蹤標的市場相關係數及追蹤誤差之差異，供投資者預期市場下跌時，避險之參考。
實證結果顯示，槓桿型ETF及反向型ETF因追蹤績效不佳造成成交量與報酬呈現不對稱的反向關係，另以追蹤誤差做為投資人買進持有ETF的參考依據，恐會造成投資決策偏誤。中美貿易戰後因經濟結構改變，市場波動性及預期恐慌心理加劇，使傳統型ETF及槓桿型ETF之追蹤績效變差。恐慌指數於中美貿易戰後不具門檻效果無法有效反映市場情緒狀態市場，且在高度波動下，不具價格發現能力。在考量事件因素後，傳統型ETF追蹤績效顯現出大幅度差異，此外，金融事件對追蹤誤差的影響明顯大於貨幣政策、經濟結構變化和非金融事件。

This study takes Taiwan’s leveraged ETF and inverse ETF as object of study to track the difference of performance under different market sentiment compared with traditional ETF.  In order to objectively define the state of market sentiment, the two-section threshold autoregressive (TAR) method is used to estimate panic index, and market sentiment is divided into two states, pessimistic and optimistic, to observe the difference in tracking performance under different market sentiment.  In addition, considering the occurrence of Sino-US trade war and the structural changes and impacts of the global economy, the threshold values of the two periods before and after the trade war are estimated to capture the threshold effects that vary from time to time based on the starting date of the Sino-US trade war. In view of the structural changes and impacts of the Sino-US trade war on the global economy, this paper also tests the referential value of VIX as an indicator of market sentiment response and price discovery. 
Secondly, different from traditional ETF which directly holds the target commodities to copy profits and losses, leverage ETF and inverse ETF build leverage by investing in derivative commodities such as futures to track holdings in the spot market and futures market. The trading system, liquidity and investors’ operational needs are different, which will affect the tracking ability of ETF. Therefore, this paper uses DCC-GARCH model to detect whether there is asymmetric relationship between various ETF and the tracking target market, and incorporates the mentioned threshold differences into the model, hoping to understand the changes in tracking performance by capturing the changes in dynamic correlation coefficients. In addition, this paper also discusses the differences of correlation coefficient and tracking error between ETF and its tracking target in different major economic events, which can be used as a reference for investors to avoid risks when they expect the market to fall.
The empirical results show that the poor tracking performance of leveraged ETF and inverse ETF results in an asymmetric inverse relationship between trading volume and return, and the tracking error is taken as the reference basis for investors to buy and hold ETF, which may lead to investment decision errors. After the Sino-US trade war, due to changes in economic structure, the market volatility and expectation panic has increased, worsening the tracking performance of traditional ETF and leveraged ETF. The panic index has no threshold effect after the Sino-US trade war and cannot effectively reflect the emotional state of the market. Moreover, it has no price discovery ability under high fluctuation. After taking into consideration the event factors, this paper finds that the tracking performance of traditional ETF shows significant differences and the impact of financial events on tracking errors is significantly greater than monetary policy, economic structural changes and non-financial events.

CONTENTS
ABSTRACT IN CHINESE I
ABSTRACT IN ENGLISH II
CONTENTS IV
LIST OF TABLES VI
LIST OF FIGURES VII

CHAPTER 1 INTRODUCTION	1
1.1 Research background	1
1.2 Motivation		6
1.3 Objectives		8
1.4 Process and Structure	11
CHAPTER 2 LITERATURE REVIEW	14
2.1 Introduction of leveraged ETF and inverse ETF	14
2.2Analysis of the tracking error behavior of leveraged ETF	18
2.3 Major financial events and ETF hedging performance	21
CHAPTER 3 RESEARCH METHOD AND EMPIRICAL MODEL	24
3.1 Variables and processing	24
3.2 Threshold autoregressive model (TAR)	27
3.3 Generalized Autoregressive Conditional Heteroskedastic model (GARCH)	28
3.4 DCC-GARCH model	29
3.5 Model setting	31
CHAPTER 4 DATA SOURCES AND PROCESSING	34
CHAPTER 5 EMPIRICAL RESULTS AND ANALYSIS	37
5.1 Basic descriptive statistics of variables	37
5.2 Verification of unit-root	41
5.3 Empirical model estimation results	43
CHAPTER 6 CONCLUSION	55
REFERENCE	59
LIST OF TABLES
【Table 5.1.1】Summary statistics	40
【Table 5.2.1】ADF unit-root test 	42
【Table 5.2.2】PP unit-root test		42
【Table 5.3.1】Empirical Results from Asymmetric DCC-GARCH models	47
【Table 5.3.2】The Empirical Results from Asymmetric DCC-GARCH models before China-US Trade Wars (before Mar. 21, 2018)	51
【Table 5.3.3】The Empirical Results from Asymmetric DCC-GARCH models after China-US Trade Wars (after Mar. 21, 2018)		52
【Table 5.3.4】Mean Value of Correlation Coefficients and Tracking Errors during Various Events Periods	54
LIST OF FIGURES
【Figure 1.1.1】market capitalization of Taiwan’s ETF	5
【Figure 1.4.1】The research structure diagram of the thesis	13

Aitken, M. & Comerton-Forde, C. (2003). How Should Liquidity be Measured? Pacific-Basin Finance Journal, 11(1), 45-59.
Avellaneda, M., & Zhang, S. (2010). Path-dependence of leveraged ETF returns. SIAM Journal on Financial Mathematics, 1(1), 586-603.  
Bansal, V. K. (2015). Impact of volatility on leveraged ETFs. Journal of Business and Behavioral Sciences, 27(1), 86-94
Bansal, V. K., & Marshall, J. F. (2015). Tracking error decomposition and return attribution for leveraged exchange traded funds. Global Finance Journal,28, 84-94.
Bansal, V. K., & Marshall, J. F. (2015). A tracking error approach to leveraged ETFs: Are they really that bad?.Global Finance Journal,26, 47-63.
Bollerslev, T. (1986). Generalized Autoregressive Conditional Heteroscedasticity.  Journal of Econometrics, 31, 307-327.
Chaudhary, G. M., Hussain, I., & Shaheen, I. (2018). Global financial crisis and stock markets volatility: A comparison of diverse financial structure countries. Journal of Contemporary Research in Management, 13(1), 23-40. 
Cheng, M., & Madhavan, A. (2009). The dynamics of leveraged and inverse exchange-traded funds. Journal of Investment Management, 7(4), 43–62
Corrado, C. J., & Miller Jr, T. W. (2005). The forecast quality of CBOE implied volatility indexes. Journal of Futures Markets, 25(4), 339-373.
Deng, G., & McCann, C. (2012). The properties of short term investing in leveraged ETFs. Journal of Financial Transformation, 35, 27-38.
Dumas, B., Fleming, J., & Whaley, R. E. (1998). Implied volatility functions: Empirical tests. The Journal of Finance, 53(6), 2059-2106.

Elton, E. J., Gruber, M. J., Comer, G., & Lai, K. (2002). Spiders: Where are the bugs? Journal of Business, 75, 453-472.
Engle, R. (2002). Dynamic conditional correlation: A simple class of multivariate generalized autoregressive conditional heteroskedasticity models. Journal of Business & Economic Statistics, 20(3), 339-350.
Gabriel, V. (2015). Sensitivity, persistence and asymmetric effects in international stock market volatility during the global financial crisis. Revista De Métodos Cuantitativos Para La Economía y La Empresa, 19, 42-65. 
Giese, G. (2010). On the Risk-Return Profile of Leveraged and Inverse ETFs.  Journal of Asset Management, 11, 219-228
Guo, K., & Leung, T. (2016). Understanding the tracking errors of commodity leveraged ETFs. St. Louis: Federal Reserve Bank of St Louis. 
Guo, K., & Leung, T. (2014). Commodity leveraged ETFs: tracking errors, volatility decay and trading strategies. The Journal of Risk, 8-13.
Hansen, B. (1996). Inference When a Nuisance Parameter is Not Identified Under the Null Hypothesis. Econometrica, 64(2), 413-430.
Hunter, M. H., Lu, X., McLeod, R. W., & Mehran, J. (2013). Bad news bears: Effects of expected market volatility on daily tracking error of leveraged bull and bear ETFs. Managerial Finance, 39(12), 1169-1187.
Lin, A. J., Chang, H. Y., & Hsiao, J. L. (2019). Does the Baltic Dry Index drive volatility spillovers in the commodities, currency, or stock markets? Transportation Research Part E: Logistics and Transportation Review, 127, 265–283.
Lu, L., Wang, J., & Zhang, G. (2009), “Long Term Performance of Leveraged ETFs,” Working Paper, Baruch College.
Luo, Y. (2014). Economic freedom, financial crisis and stock volatilities in emerging markets. International Journal of Financial Management, 4(1), 1-10. 
Murphy, R., & Wright, C. (2010). An empirical investigation of the performance of commodity-based leveraged ETFs. Journal of Index Investing, 1(3), 14–23 
Rompotis, G. (2016). Return and volatility of emerging markets leveraged ETFs. Journal of Asset Management, 17(3), 165-194.
Trainor Jr, W. J., & Carroll, M. G. (2013). Forecasting Holding Periods for Leveraged ETFs Using Decay Thresholds: Theory and Applications. Journal of Financial Studies & Research, 2013(1), 179-190.
Tsalikis, G., & Papadopoulos, S. (2018). Assessing the performance of American and European Leveraged Exchange Traded Funds. Investment Management and Financial Innovations, 15(2), 165–182.