近來，臺灣教育部為招收更多國際學生而實行新政策。本研究分析政策實行的效果、學生來臺原因及在臺學生的滿意度。研究一以時間序列分析1954至2008年的國際學生人數，ARIMA模式的預測效果良好(RER=0.83%)，未來三年內國際學生人數將突破二萬人。研究二以問卷調查分析學生選擇留學國家及學校的選擇因素，同時分析學生來臺動機、所遇困難及在臺滿意度間的關係 (N=210)。結果顯示容易的簽證程序及臺灣的學術聲望影響學生來臺讀書；此外，財務、後勤及生活條件影響學生選擇留學學校。學生因臺學術聲望來臺且無食物適應困難者，願意再次來臺念書；學生因具吸引力課程及方便生活環境來臺且無硬體設備及學習適應方面困難者，願意建議他人來臺留學。根據研究結果，政府應加強國內教學品質，並協助提升學校學術聲望；增加赴海外任教教師，以教學專業提升我國國際形象。有興趣吸引國際學生的學校應可從健全學校行政體制著手，並聚焦於完善學生招生、入學諮詢輔導工作及協助生活適應。 / Recently, the Taiwanese Ministry of Education has implemented a new policy to recruit more international students. This study analyzes the policy’s effects and the reasons motivating students to study in Taiwan. The number of international students in Taiwan from 1954 to 2008 was used to generate a time series model to estimate the number of international students before and after the policy implementation. This ARIMA model (RER=0.83%) demonstrated a dramatic increase in the number of international students over the years. Using a questionnaire designed to examine determinants that affect international students’ decisions for choosing Taiwan and their respective schools (N=210), this study then study the relationship between the students’ motivations, the challenges they faced, and their degree of satisfaction with their experiences. Results demonstrated that an easy visa application process and Taiwan’s acclaimed academic reputation attracted students. Furthermore, financial, logistical, and living reasons affected the students’ choices in schools. Students who found living and educational conditions favorable and who felt intellectually stimulated would choose again to study in Taiwan and would recommend others to do so. Based on these results, the government should strengthen the quality of instruction and further promote Taiwan’s universities abroad. Schools interested in attracting international students should develop structured administrative systems focused on recruiting students and helping them transition into a new environment.

於這篇論文中，我們運White (2000)的Reality Check與Romano and Wolf (2005)的stepwise multiple test檢測交易策略的獲利性以更正資料窺探的偏誤。不同於先前運用資料窺探法則的研究，我們的研究以技術分析及時間序列預測兩者為依歸來建立交易策略，另外我們探討的市場集中在六個主要的亞洲股票市場。大致上，我們發現鮮少證據支持技術交易策略的獲利性；於基礎分析中且考慮交易成本時，只有少數幾個獲利性交易法則出現於兩個興新市場。另外在子樣本期間中，我們發現獲利性策略的表現並不穩定且這幾年來獲利性有逐漸變弱的趨勢。在進階分析中，我們發現沒有任何交易策略表現優越於基本的買進持有策略。 / In this paper, we exam the profitability of trading strategies by using both White’s (2000) Reality Check and Romano and Wolf (2005)s’ stepwise multiple test that correct the data snooping bias. Different from previous studies with the data snooping methodology, our analysis set the universe of forecasts (trading strategies) based on both technical analysis and time series prediction, and the markets which our investigation focuses on are six major Asian stock markets. Overall we find little supportive evidence for the profitability of trading strategies. Our basic analysis shows that there are only few profitable trading strategies detected for two emerging markets while transaction costs are taken into account. Moreover, the performances of the profitable strategies are unstable and the profitability becomes much weaker in the recent years as we find in the sub-periods. In further analysis, we also find that there is no trading strategies in our universe that can outperform the basically buy and hold strategy.

本研究運用Vasecik、ARMA與VEC三種時間序列模型，以蒙地卡羅法，模擬未來五年台幣利率、美元利率與新台幣兌美元匯率的隨機漫步過程，並分析壽險公司的資產、負債與業主權益價值，在利率與匯率的隨機過程中所受到的影響。 藉由蒙地卡羅模擬之隨機漫步過程，本研究發現在利率模型方面，Vasicek利率模型因具有均數回歸的特性，較VEC模型擁有更穩定的隨機漫步過程；在匯率模型方面，VEC模型因同時考量長期影響與短期影響的效果，較ARMA模型擁有較穩定的漫步過程。 在負債面的模擬結果中，當利率下跌時，保單應提列準備金價值的成長速度較利率上升時快，此點反應壽險公司在低利率的環境下，將面臨較嚴峻的資本要求；同時，藉由歷史資料以Vasicek債券評價模型估計之利率期間結構，整體結構呈現負斜率與凹口向上的走勢，在此情形下，短期利率的值較長期利率的值高，保單應提列的準備金價值較原始估計時更高。 在長期的低利率環境中，上述現象反應於長期保單的價值變化尤為明顯。本研究建議在進行保單的精算訂價時，不應僅以預定利率做為保單全期的折現因子，而應將長期的利率風險納入考量。 同時，匯率的變化亦嚴重衝擊壽險公司的業主權益，在模擬結果中，當匯率落於風險值時，壽險公司配置於美元資產的減損將造成業主權益呈現虧損，此點亦反應當壽險公司將資產配置於海外時，必須謹慎地評估外匯避險的相關策略。 整體而言，在本研究中，將資產配置偏重台幣的投資策略擁有較穩定的業主權益價值，並在短期擁有較佳的風險轉換報酬能力；另一方面，將資產配置偏重美元的投資策略在長期擁有較佳的風險轉換報酬能力，然而，也因其擁有較高的風險值，壽險公司可能面臨較嚴重的損失。本研究建議壽險公司在進行海外資產配置時，應謹慎地將利率風險與匯率風險納入考量。 / This article uses Monte Carlo simulation method to forecast the random walk process of Taiwan interest rate, US interest rate, and Taiwan US dollar exchange rate between next five years. The simulation base on three time series model：Vasecik, ARMA and VEC. Through the random walk process, this article aims to analyze the influence in asset, liability and equity by the change of interest rate and exchange rate. In this paper, we find that the Vasicek interest rate model has a more stable stochastic process than the VEC model, which because of the effect by mean reversion. On the other hand, because the VEC exchange rate model takes both long-term and short-term impact in concern, it has a more robust stochastic process than the ARMA model. Through the simulation results of the liabilities, we find that when the interest rate fell, the reserve value of insurance policy will rise faster, which makes life insurance companies face more severe capital requirements in the low interest rate environment. Besides, we also find that the interest rate term structure in the Vasicek Bond Pricing Model displays negative slopes with concave upward, which means the value of short-term interest rate higher than the value of long-term interest rate. In this situation, the reserve value of insurance policy will become much higher than the value original priced. In the long-term low interest rate environment, the impact of interest rate risk has more effect in the long-term insurance policy. This paper suggests that when pricing the costs of insurance policy, we should not only use one interest rate as the full term discount factor. The better way is to discount with the interest rate term structure. Overall, in this paper, the asset allocation strategy, which focus on Taiwan commercial bonds, has both better performances in value at risk and better ability to covert risk into revenue in the short term. On the other hand, the asset allocation strategy, which focus on US commercial bonds, has better ability to covert risk into revenue in the long run. When conducting overseas asset allocation, we suggest that life insurance companies should carefully consider interest rate risk and exchange rate risk.

當我們利用Shewhart管制圖(Shewhart control chart)或累積和管制圖(Cumulative-sum chart. CUSUM chart)來偵測製程時,通常假設製品係獨立取自一個服從均數μ和標準差為σ的獨立常態分配的管制下進行。但是若產品特性值呈現自相關時,這類管制圖就可能發生誤導的結果。本文利用時間序列模式來解決具相關變數的管制圖問題。並考慮利用非線性時間序列模式及特別原因管制圖(special-cause control chart)來檢視台灣經濟景氣指標是否處於控制中的狀態。並討論特別原因管制圖的連串長度分佈(run length distribution)。在最後的實例分析中,介紹自動控制的觀念。 / Traditionally, in the quality control process, such as: Shewhart control chart or CUSUM chart, it is assumed that the observation process follows an i.i.d normal distribution. If the assumption for independence fails, that is when the process exhibits type of autocorrelation, we need to find a more reliable decision method. In this paper, we will apply the time series analysis and structure changed concept to slove the serial correlation problem. The idea of automatic control can be applied in the explanation of this nonlinear process. Finally, a time series about the monitoring indicators of Taiwan is discussed in detail as an example.

觀光事業素有無煙囪工業之稱，自政府於民國四十八年全力推動發展以來 ，來華觀光旅客人數即不斷地成長，此對促進國民外交與增加政府的外匯 收入上有莫大的幫助。觀光旅客人數的多寡，直接影響本地觀光業者與政 府相關單位對觀光業軟硬體設施的投資，像是觀光旅館的興建、導遊人員 的培訓以及整體策略的規劃。不當的評估或不正確的需求預測，都將導致 大量觀光資源的閒置或浪費。本研究計劃主要應用簡算法、時間趨勢模式 、時間序列模式、計量經濟模式，尋找並建立來華觀光旅客長短期需求預 測模式，並針對總體或各主要市場的需求，利用各種模式評估準則提出一 最佳預測模式，以供政府相關單位與觀光業者作為往後政策釐定以及投資 計劃擬定時的參考。

本論文主要研究以類神經網路模式預測季節性時間序列之有效性。利用適 當地建構樣本訓練集,網路經訓練後可作為季節性時間序列之預測工具。 文中亦提出移動學習法以期提高預測之準確度。並以台灣地區每季進口商 品與勞務總值則作為實證之研究。此季節性時間序列因受離群值之影響而 增加其預測困難度。實證結果顯示類神經網路模式之預測表現較傳統之統 計方法優異,即使此序列受到離群值之干擾。 / We investigate the effectiveness of neural networks for predicting the future behavior of seasonal time series. Utilizing the training set constructed properly, we can train the network who can be used to predict the future of seasonal time series. A shifting-learning method is also employed in order to obtained a better forecasting performance. The quarterly imports of goods and services of Taiwan between the first quarter of 1968 and the fourth quarter of 1990 are studied in the research. The series are contaminated with outliers, which will increase the difficulty of forecasting. Empirical results exhibit that neural networks model free approach have better prediction performance than the classical Box-Jenkins approach, even the series are contaminated with outliers.

University of Macau / Faculty of Science and Technology / Department of Mathematics

近年來輻狀基底函數類神經網路 (Radial Basis Function Networks , RBFN) 應用在時間序列相關問題上已有相當的成果。在這篇論文裡，我們嘗試建構一個電腦軟體工具，可以很容易造出 RBFN，應用在時間序列預測相關問題上。更進一步的說，我們的電腦工具可以輕易做出即時修正，完全符合使用者的需求。我們一開始先複習 RBFN 的基本架構， 並說明如何應用到時間序列的問題上。接著我們研究近年來相當受到重視的 T-RBF (Temporal RBF) 架構。最後，我們解釋如何使用 Adobe Flex 去建構我們所需要的電腦軟體工具。這個工具是跨平台的程式，並且不論是雲端計算或是單機應用皆很合適。 / During recent years, applying Radial Basis Function Networks (RBFN) to time series problems yields many important results. In this thesis, we try to implement a cross-platform computer tool that can easily construct a RBFN applied to time series forecasting problems. Moreover, the RBFN created by this computer tool can do real-time modification to fit specific needs. We first review the basic structures of RBFN and explain how it can be applied to time series problems. Then, we survey on so called temporal radial basis function (T-RBF) model, which draws much attention these years. Finally, we explain how we use Adobe Flex to create a computer tool as we mentioned in the beginning. The computer application is cross-platform and is suitable for both cloud computing and desktop applications.

In this paper, we propose two types of sieve bootstrap, univariate and multivariate approach, for the generalized method of moments estimators of time series data. Compared with the nonparametric block bootstrap, the sieve bootstrap is in essence parametric, which helps fitting data better when researchers have prior information about the time series properties of the variables of interested. Our Monte Carlo experiments show that the performances of these two types of sieve bootstrap are comparable to the performance of the block bootstrap. Furthermore, unlike the block bootstrap, which is sensitive to the choice of block length, these two types of sieve bootstrap are less sensitive to the choice of lag length.

在全球經貿體系自由化下，國際資金流通快速，匯率變動也非常頻繁，廠商的產銷決策與營運，面對匯率風險更加難以掌控。如何掌握匯率的變動，並採取有效的避險措施，是廠商從事貿易必須面臨之重要課題。本研究採用自我迴歸整合移動平均模式、倒傳遞類神經網路及混合式自我迴歸整合移動平均模式及倒傳遞類神經網路模型進行未來即期匯率報酬率之預測。試圖找出合適的新台幣兌美元即期匯率之預測模型，並將其應用於外匯避險操作。 研究結果顯示，關於預測誤差的績效表現，整體來說，以自我迴歸整合移動平均及倒傳遞類神經網路混合式模型表現最佳，顯示傳統時間序列模型捕捉匯率報酬率走勢之能力，藉由倒傳遞類神經網路捕捉其線性預測誤差中非線性的部分，可更符合資料的特性，加強匯率報酬率預測的準確性。考慮預測方向的正確性，在兩個不同的準則下(SR、PT)，皆以自我迴歸整合移動平均模型表現最差，代表其在進行匯率報酬率之預測時正確率較為不足。而在PT檢定當中，倒傳遞類神經網路模型及混合式模型皆達到顯著。因此利用人工智慧模型對報酬率之方向進行預測是有效的，又以自我迴歸整合移動平均及倒傳遞類神經網路混合式模型表現最好。總結來說，利用倒傳遞類神經網路模型針對自我迴歸整合移動平均模型做非線性的調整，同時涵蓋未來匯率報酬率線性與非線性的部分，使得自我迴歸整合移動平均模型之預測誤差、方向準確性皆得到改善，藉由倒傳遞類神經網路捕捉其線性預測誤差中非線性的部分，可更符合資料的特性，加強匯率報酬率預測的準確性。