"The fundamental theorem of asset pricing in finance states that the price of any asset is its expected discounted payoff. Ideally, the payoff is discounted by a factor, which depends on parameters present in the market, and it should be unique, in the sense that financial derivatives should be able to be priced using the same discount factor. In theory, risk neutral valuation implies the existence of a positive random variable, which is called the stochastic discount factor and is used to discount the payoffs of any asset. Apart from asset pricing another use of stochastic discount factor is to evaluate the performance of the of hedge fund managers. Among many methods used to evaluate the stochastic discount factor, generalized method of moments has become very popular. In this paper we will see how generalized method of moments is used to evaluate the stochastic discount factor on linear models and the calculation of stochastic discount factor using generalized method of moments for the popular model in finance CAPM. "

1997年7月2日泰國宣布放棄泰銖釘住美元而改採管理浮動匯率制度，乃掀起東南亞國家貨幣貶值的危機﹔同時這些國家股價亦遭滑鐵盧之災。東南亞國家金融風暴（亞洲金融風暴）的面貌是﹕幣值與股價同時大幅下滑。但1980年代末期，台灣股價暴漲與新台幣對美金匯率之升值幾乎是同時間發生的。新台幣升值幅度相當的大，且新台幣升值是漸進的，這予投機者以機會。亞洲金融風暴後，央行力守一美元匯兌新台幣在28.6元的匯率，動用近50億美元，打擊投機客，並同時調降存款準備率，放出1350億元的強力貨幣，予股市一股強力的活水。其目的是守住匯價，亦可使股價止跌。然而，市場的表現不是如此，1997年9月與10月股價卻直直下落。 　　研究期間為1990年1月1日至1999年1月31日止，並分成三子期間，以利比較風暴後兩市場互動關係，資料為發行量加權平均股價指數收盤價及新台幣對美元的銀行間平均收盤即期匯率之日資料；類股指數之研究期間為1997年7月1日至1999年1月31日止。運用GARCH模型進行實證分析，並利用一般化動差估計式(Generalized Method of Moments Estimator; GMM) 來估計所建構的迴歸式，以期達成下列目的﹕ 1. 確認台灣股票市場及外匯市場之互動結構關係。 2. 亞洲金融風暴後，其結構關係變化為何？ 3. 確認亞洲金融風暴後，台灣外匯市場對股票市場之各類股的互動結構 關係。 實証結果為： 1. 亞洲金融風暴前，有匯率對股市的報酬率波動外溢效果，亞洲金融風暴後，此報酬率波動外溢效果較風暴前減少，而其它影響股市的因素（非關匯率因素）反而逐漸增強。 2. 股匯市從單向關係（只有匯市影響股市）演變成雙向互動關係，且股市對匯市影響力增強。 3. 金融保險類、水泥窯製類及造紙類，此三類最不受匯市的影響。 4. 塑膠化工類、營造建材類、食品類及紡織纖維類，此四類受當期匯市報酬率的負影響，亦即新台幣升值，此四類股股價會上漲。而機電類股，則受滯延4期匯市報酬率的負影響。 5. 營造建材類，報酬率波動受到其他因素（非關匯率因素）影響很大。 目錄 第一章 緒論 第一節 研究背景與動機……………………………………… 1 第二節 研究目的……………………………………………… 6 第三節 研究限制……………………………………………… 7 第二章 相關理論探討 第一節 匯率的意義、種類及其影響因素…………………… 8 第二節 匯率變動對股票價格的影響………………………… 11 第三節 效率資本市場理論…………………………………… 14 第三章 相關文獻探討 第一節 國外相關文獻………………………………………… 16 第二節 國內相關文獻………………………………………… 19 第四章 研究方法 第一節 相關模型……………………………………………… 25 第二節 分析程序與方法……………………………………… 27 第五章 資料來源與處理 第一節 資料來源與研究期間………………………………… 34 第二節 資料處理……………………………………………… 34 第三節 基本統計分析………………………………………… 35 第六章 實証結果與分析 第一節 股市與匯市報酬率及報酬率波動外溢效果實証結果 46 第二節 各類股與匯市報酬率及報酬率波動外溢效果實証結果 ………………………………………………………… 72 第七章 結論與建議……………………………………………… 76 參考文獻…………………………………………………………… 108 表次 表5-1 股匯市報酬率之基本檢定統計量……………………… 37 表5-2 1997/7/1至1999/1/31各類股之基本檢定統計量…… 41 表6-1 1990/1/1至1999/1/31外溢效果（使用Pearson 交叉相關檢定）………………………………………… 48 表6-2 1990/1/1至1999/1/31外溢效果（使用GMM估計式）… 53 表6-3 1990/1/1至1994/12/3外溢效果（使用GMM估計式）… 58 表6-4 1995/1/1至1997/6/30外溢效果（使用GMM估計式）… 63 表6-5 1997/7/1至1999/1/31外溢效果（使用GMM估計式）… 68 表6-6 1997/7/1至1999/1/31外溢效果（匯市與金融保險類股 ，使用GMM估計式）……………………………………… 74 表6-7 1997/7/1至1999/1/31外溢效果（匯市與水泥窯製類股 ，使用GMM估計式）……………………………………… 78 表6-8 1997/7/1至1999/1/31外溢效果（匯市與塑膠化工類股 ，使用GMM估計式）……………………………………… 82 表6-9 1997/7/1至1999/1/31外溢效果（匯市與營造建材類股 ，使用GMM估計式）……………………………………… 86 表6-10 1997/7/1至1999/1/31外溢效果（匯市與機電類股 ，使用GMM估計式）……………………………………… 90 表6-11 1997/7/1至1999/1/31外溢效果（匯市與食品類股 ，使用GMM估計式）……………………………………… 94 表6-12 1997/7/1至1999/1/31外溢效果（匯市與造紙類股 ，使用GMM估計式）……………………………………… 98 表6-13 1997/7/1至1999/1/31外溢效果（匯市與紡織纖維類股 ，使用GMM估計式）……………………………………… 102

In this paper, we propose a new method for noise robustness base on Gaussian Mixture Model (GMM), and the method we proposed can estimate the noise feature effectively and reduce noise effect by plain fashion, and we can retain the smoothing and continuity from original feature in this way. Compared to the traditional feature transformation method MMSE(Minimum Mean Square Error) which want to find a clean one, the different is that the method we proposed only need to fine noise feature or the margin of noise effect and subtract the noise to achieve more robustness effect than traditional methods. In the experiment method, the test data pass through the trained noise classifier to judge the noise type and SNR, and according to the result of classifier to choose the corresponding transformation model and generate the noise feature by this model, and then we can use different weight linear combination to generate noise feature, and finally apply simple subtraction to achieve noise reduction. In the experiment, we use AURORA 2.0 corpus to estimate noise robustness performance, and using traditional method can achieve 36:8% relative improvement than default, and the our method can achieve 52:5% relative improvement, and compared to the traditional method our method can attain 24:9% relative improvement.

With the development of multimedia technology, speech recognition technology has increasingly become a hotspot of research in recent years. It has a wide range of applications, which deals with recognizing the identity of the speakers that can be classified into speech identification and speech verification according to decision modes.The main work of this thesis is to study and research the techniques, algorithms of speech recognition, thus to create a feasible system to simulate the speech recognition. The research work and achievements are as following: First: The author has done a lot of investigation in the field of speech recognition with the adequate research and study. There are many algorithms about speech recognition, to sum up, the algorithms can divided into two categories, one of them is the direct speech recognition, which means the method can recognize the words directly, and another prefer the second method that recognition based on the training model. Second: find a useable and reasonable algorithm and make research about this algorithm. Besides, the author has studied algorithms, which are used to extract the word's characteristic parameters based on MFCC(Mel frequency Cepstrum Coefficients) , and training the Characteristic parameters based on the GMM(Gaussian mixture mode) . Third: The author has used the MATLAB software and written a program to implement the speech recognition algorithm and also used the speech process toolbox in this program. Generally speaking, whole system includes the module of the signal process, MFCC characteristic parameter and GMM training. Forth: Simulation and analysis the results. The MATLAB system will read the wav file, play it first, and then calculate the characteristic parameters automatically. All content of the speech signal have been distinguished in the last step. In this paper, the author has recorded speech from different people to test the systems and the simulation results shown that when the testing environment is quiet enough and the speaker is the same person to record for 20 times, the performance of the algorithm is approach to 100% for pair of words in different and same syllable. But the result will be influenced when the testing signal is surrounded with certain noise level. The simulation system won’t work with a good output, when the speaker is not the same one for recording both reference and testing signal.

No description available.

Breast cancer is a devastating disease that affects thousands of women every year influencing their psychological and physical well-being for many years after being diagnosed. The goal of the current study was to determine if there are distinct trajectories of functioning among breast cancer patients in the domains of negative psychological adjustment, positive psychological adjustment, and physical adjustment. This was accomplished using growth mixture modeling. Another goal of this study was to determine whether demographic, medical, and psychosocial variables were able to distinguish among the trajectories. The study combined women from two samples spanning 10 years providing a sample size of 376 women diagnosed and treated for breast cancer. These women were recruited to participate in a 10-week cognitive behavioral stress management intervention and were either randomized to the 10-week experimental condition or a one-day control group. It was hypothesized that distinct trajectories would emerge for each of the domains and that psychosocial variables (i.e., social support, benefit finding, and emotional approach coping) would distinguish among the trajectories. This study was able to statistically identify multiple classes or trajectories of adjustment, consistent with findings reported by Helgeson and colleagues (2004) and Donovan and colleagues (2007). It is difficult to say, however, whether these classes differ in clinically significant ways. The present study also provides a cautionary note to researchers who intend to use growth mixture modeling to identify different trajectories of functioning and the limitations associated with this statistical technique. First, it is important to start this process with strong empirical or theoretical support for the possibility of different classes or trajectories. Without this foundation it becomes difficult to justify why a certain number of classes were chosen. Another limitation of this statistical approach is that there is not a standard method for determining the best number of classes. There are conflicting opinions among researchers in the field about the best fit index to use when the multiple fit indices do not converge. A serious issue related to this is the fact that classes are used for interpreting results and drawing conclusions and inferences. Therefore, clinicians using GMM must be careful when deciding on the number of classes and the clinical inferences drawn from these analyses. Further research needs to be conducted validating these statistical techniques.

Neoclassical models imply convergence of the entire distribution, not just the mean income levels. In this paper, we analyze convergence in income inequality by using the considerably enlarged data bases, from the World Bank (Povcal) and the World Institute for Development Economic Research (WIDER). Convergence in gini indices of inequality is tested across 55 countries. We consider three sample subsets; one for the developing countries, second of the developed countries and third with all countries together. We test for convergence in gini indices over a period of 5, 10, 15, 20 and 25 years. Additionally we use cross-section (OLS),panel (GMM) and a novel OLS estimation methods. Our results uniformly indicate that inequality levels among developing countries converged. Evidence of convergence is weaker among developed countries. Developing countries appear to converge faster than developed countries.

Speech intelligibility represents how comprehensible a speech is. It is more important than speech quality in some applications. Single channel speech intelligibility enhancement is much more difficult than multi-channel intelligibility enhancement. It has recently been reported that training-based single channel speech intelligibility enhancement algorithms perform better than Signal to Noise Ratio (SNR) based algorithm. In this thesis, a training-based Deep Neural Network (DNN) is used to improve single channel speech intelligibility. To increase the performance of the DNN, the Multi-Resolution Cochlea Gram (MRCG) feature set is used as the input of the DNN. MATLAB objective test results show that the MRCG-DNN approach is more robust than a Gaussian Mixture Model (GMM) approach. The MRCG-DNN also works better than other DNN training algorithms. Various conditions such as different speakers, different noise conditions and reverberation were tested in the thesis.

This thesis consists of two related chapters and one unrelated chapter. The first focuses on the health of immigrants in Canada, using the most up-to-date Canadian Community Health Surveys (CCHS). I re-investigate the previously well-established Healthy Immigrant Effect (HIE)--the finding that upon arrival, immigrants are relatively healthier than the native-born population; but that this health advantage declines over the years after migration. Measures of health used in this study include self-assessed health status, the likelihood of being overweight or obese, and the incidence of various chronic conditions. The first part of this chapter replicates the heavily cited work of McDonald and Kennedy (2004) by pooling multiple years of CCHS, and estimating a model controlling for immigrants' cohorts of arrival to disentangle the true effect of years-since-migration (YSM) from the cohort effects. The second part of this chapter takes a closer look at the more recent cohorts of arrival of immigrants. Here I use a matching method to compare various measures of health between immigrants who arrived before and after the implementation of the Immigration and Refugee Protection Act (IRPA). It is important to note that this study does not make any direct link between the implementation of IRPA and the health outcomes of immigrants. It merely observes and compares the health of two different cohorts of immigrants, making no assumptions as to whether these changes are a result of IRPA. My main finding is that the initial health advantage is no longer present for more recent cohorts of immigrants to Canada, and that these recent cohorts of immigrants face higher health risks associated with being overweight or obese. The other two chapter--Chapter 2, Income Processes and Intra-household Risk Sharing, and Chapter 3, Health Shocks and Income Dynamics--deal with different aspects of modelling of labour income risk over the life cycle using the US Medical Expenditure Panel Survey (MEPS). In Chapter 2, I take advantage of MEPS's large sample size (some 40,000 households) to concentrate on estimating income risk-sharing among couples. This refers to an intra-household insurance mechanism that allows couples to diversify labour income risks; for instance, they can and dynamically coordinate labour-supply decisions in response to income shocks. Specifically, this study decomposes income volatility, distinguishing between single and couple household types, and models couples' income risk-sharing as the covariance of the husband and wife's income variance for both transitory and permanent components. I use an innovative identification strategy, assuming the invariability of market price for labour to marital status, to uncover couple-specific risk-sharing parameters by allowing the income profile of singles and non-singles to have partial common structure. I find evidence of risk-sharing between spouses in response to both transitory and permanent income shocks, suggesting that couples' earning capability might be partially insulated from the impact of transitory and permanent income risk. Chapter 3 is co-authored with two of my supervisors, C. Deri Armstrong and G. Dunbar. The work is done primarily by myself, except for the Introduction, where both co-authors contribute to the writing. G. Dunbar also contributes to parts of the sections on Heterogeneous Health Impacts and Endogeneity, and to the Conclusion. This chapter also uses MEPS data, but focuses on understanding the significance of the negative health shocks in decomposing labour income risk. As in Chapter 2, we break down the cross-sectional variance of residual earnings into transitory and permanent components. We then propose a method to decompose the heterogeneity of health shocks impacts by partitioning the cross-sectional variance of residual earnings into a health and non-health component. We use emergency room (ER) visits as a proxy for negative health shocks, and we separately consider the impact of these negative health shocks for several groups, such as single men and single women with no child, single mothers, and couples. We also probe the role of health insurance in attenuating the income effects of health shocks, and put forward a creative method to control for misspecification biases in the income regressions--the usual ability bias. Our results suggest that heterogeneity in health shocks is gender-differentiated. We find that health shocks have heterogeneous impacts for single women with no child, as well as single mothers; but no such evidence is found for single men. For couples, we find that having health insurance coverage reduces the impact of negative health shocks on income volatility by roughly 10 percentage points.

Net interest margin (NIM) is an important indicator of a bank's operational efficiency. Based on the balance sheet data of 189 major listed banks in Europe from 2010 to 2019, this thesis studies the bank's NIM mechanism in a negative interest rate environment. This thesis focuses on the system GMM method and the results show that the policy interest rate is positively related to NIM in the long run and negatively related in the short run, but the relationship between the two is not significant in the short run. Moreover, in a negative interest rate environment, bank NIM's sensitivity to policy interest rates has greatly increased, especially the policy of interest rate cuts. In addition, the sensitivity of NIMs of different banks to policy interest rates also differs significantly. Generally, the NIMs of banks with a high degree of internationalization and larger size are less sensitive to changes in policy interest rates, while the NIMs of banks with a higher share of retail business in their total business are more sensitive to changes in policy interest rates. Finally, through the value-at-risk analysis and stress test, this thesis concludes that the policy interest rate, net loan-to-asset ratio, non-performing loan ratio and inflation rate are sensitive factors of NIM. When NIM is subject to a...