In this paper, we analyze the change in structure which occurred in Taiwan stock index, while finding a better non-linear model. We examine the out-of sample performance of non-linear time series SETAR model by employing Taiwan Stock Exchange Capitalization Weighted Stock Index over the period from January 3, 2005 to December 31, 2009. Furthermore, we do the unit root test before the model setting and then compare the out-of-sample forecasting performances between standard linear ARIMA model and non-linear SETAR model. Empirically, we find that non-linear SETAR model has superior forecasting power than linear ARIMA model does in Taiwan stock market.

In this paper, we analyze the change in structure which occurred in Taiwan stock index, while finding a better non-linear model. We examine the out-of sample performance of non-linear time series SETAR model by employing Taiwan Stock Exchange Capitalization Weighted Stock Index over the period from January 3, 2005 to December 31, 2009. Furthermore, we do the unit root test before the model setting and then compare the out-of-sample forecasting performances between standard linear ARIMA model and non-linear SETAR model. Empirically, we find that non-linear SETAR model has superior forecasting power than linear ARIMA model does in Taiwan stock market.

Abstract i
Table of Contents ii
List of Tables iii
List of Figures iv
1. Introduction 1
1.1 Background and Motivation 1
1.2 Purpose 1
1.3 Thesis Structure 2
2. Literature Review 4
3. Empirical Models 7
3.1 Unit Root Test 11
3.2 Chow Breakpoint Test 17
3.3 SETAR Model 18
3.4 Out-of-sample Forecasting Performance 20
4 Empirical Results 21
4.1 Sample Description 21
4.3 Results of ARIMA Model Selection 24
4.4 Results of Chow Breakpoint Test 25
4.5 Results of SETAR Model 27
5. Conclusions 34
References 36
List of Tables
Table 1 Sample Description ................................................................................ 22
Table 2 Results of Unit Root Test ....................................................................... 23
Table 3 Results of Model Selection of ARIMA .................................................. 24
Table 4 Results of Model Selection of SETAR ................................................... 27
Table 5 Comparison of Forecasting Power ......................................................... 28
Table 6 Forecasted Values and Original Data - November, 2009 ....................... 30
Table 7 Forecasted Values and Original Data - December, 2009........................ 31
List of Figures
Figure 1 Steps of the Box-Jenkins Approach ...................................................... 11
Figure 2 Trend of Taiwan Stock Exchange Capitalization Weighted Stock Index ......... 21
Figure 3 Sample Description ............................................................................... 22
Figure 4 Breakpoint by Using Chow Test ........................................................... 25
Figure 5 Breakpoint in Taiwan Stock Exchange Capitalization Weighted Stock Index . 26
Figure 6 Original Series and Predicted Value from the SETAR and ARIMA
Model using Forecasting Method (2009) ............................................ 32
Figure 7 Forecasting Interval of the ARIMA Model and the Forecasted Value
from Both Models (2009) ...................................................................... 33

1.Bhargava, A. (1986), On the theory of testing for unit roots in observed time series, Review of Economic Studies Vol. 53(3), 369-384.
2.Boero, G. and Marrocu, E. (2002), The performance of non-linear exchange rate models: a forecasting comparison, Journal of Forecasting, Vol. 21(7), 513-542.
3.Boero, G. and Marrocu, E. (2003), The performance of SETAR models: a regime conditional evaluation of point, interval and density forecasts. Working Paper.
4.Box, G. and Jenkins, G. (1970), Time Series Analysis: Forecasting and Control, San Francisco, CA: Holden-Day.
5.Box, G. and Jenkins, G. (1976), Time Series Analysis: Forecasting and Control, San Francisco: Holden-Day.
6.Brooks, C. (2002), Introductory Econometrics for Finance, Cambridge: Cambridge University Press.
7.Cerrato, M., Hyunsok, K. and MacDonald, R. (2008), 3-Regime symmetric STAR modeling and exchange rate reversion, Working Paper.
8.Clements, M. and Smith, J. (2001), Evaluating forecasts from SETAR models of exchange rates, Journal of International Money and Finance, Vol. 20(1), 133-148.
9.De Gooijer, J., Ray, B. and Kräger, H. (1998), Forecasting exchange rates using TSMARS, Journal of International Money and Finance, Vol. 17(3), 513- 534.
10.Dickey, D. and Fuller, W. (1979), Distribution of the estimators for autoregressive time series with a unit root, Journal of the American Statistical Association, Vol. 74(366), 427-431.
11.Dielman, T. (1986), A comparison of forecasts from least absolute value and least squares regression, Journal of Forecasting, Vol. 5(3), 189-195.
12.Granger, J. and Newbold, P. (1974), Spurious regressions in econometrics, Journal of Econometrics, Vol. 2(2), 111-120.
13.Hansen, B. (1996), Inference when a nuisance parameter is not identified under the null hypothesis, Econometrica, Vol. 64(2), 413-430.
14.Hansen, B. (2001), The new econometrics of structural change: dating breaks in U.S. labor productivity, Journal of Economic Perspectives, Vol. 15(4), 117-128.
15.Kapetanios, G. and Shin, Y. (2006), Unit root tests in three-regime, Econometrics Journal, Vol. 9(2), 252-278.
16.Kräger, H. and Kugler, P. (1993), Nonlinearities in foreign exchange markets: a different perspective, Journal of International Money and Finance, Vol. 12(2), 195-208.
17.LeBaron, B. (1992), LeBaron, Some relationships between volatility and serial correlations in stock market returns, Journal of Business Vol. 65(2), 199-219.
18.Maddala, S. and Kim, I-M. (1998), Unit Roots, Cointegration and Structural Change, Cambridge: Cambridge University Press.
19.Makridakis, S. (1993), Accuracy measures: theoretical and practical concerns, International Journal of Forecasting, Vol. 9(4), 527-529.
20.Nelson, C. and Plosser, C. (1982), Trends and random walks in macroeconomic time series: some evidence and implications, Journal of Monetary Economics, Vol. 10(3), 139-162.
21.Priestley, M. (1980), State-dependent models: a general approach to nonlinear time series analysis, International Journal of Forecasting, Vol. 1(1), 47-71.
22.Tong, H. (1977), Some comments on the Canadian lynx data (with discussion). Journal of the Royal Statistical Society. Series A (General), Vol. 140(4), 432-436.
23.Tong, H. (1978), On a threshold model, in Chen, C. H.(ed.), Pattern Recognition and Signal Processing, Amsterdam: Sijhoff and Noordoff.
24.Tong, H. and Lim, K. S. (1980), Threshold autoregression, limit cycles and cyclical data (with discussion), Journal of the Royal Statistical Society, Series B, Vol. 42(3), 245-292.
25.Tong, H. (1990), Non-Linear Time Series. A Dynamical System Approach, Oxford, UK: Oxford University Press.
26.Tong, H. (2007), Birth of the threshold time series model, Statistica Sinica, Vol. 17(1), 8-14.
27.Tsay, R. (1989), Testing and modeling threshold autoregressive processes, Journal of the American Statistical Association, Vol. 84(405), 231-240.