Dissertations / Theses on the topic 'Threshold regression model'

Zero-inflated data abound in ecological studies as well as in other scientific and quantitative fields. Nonparametric regression with zero-inflated response may be studied via the zero-inflated generalized additive model (ZIGAM). ZIGAM assumes that the conditional distribution of the response variable belongs to the zero-inflated 1-parameter exponential family which is a probabilistic mixture of the zero atom and the 1-parameter exponential family, where the zero atom accounts for an excess of zeroes in the data. We propose the constrained zero-inflated generalized additive model (COZIGAM) for analyzing zero-inflated data, with the further assumption that the probability of non-zero-inflation is some monotone function of the (non-zero-inflated) exponential family distribution mean. When the latter assumption obtains, the new approach provides a unified framework for modeling zero-inflated data, which is more parsimonious and efficient than the unconstrained ZIGAM. We develop an iterative algorithm for model estimation based on the penalized likelihood approach, and derive formulas for constructing confidence intervals of the maximum penalized likelihood estimator. Some asymptotic properties including the consistency of the regression function estimator and the limiting distribution of the parametric estimator are derived. We also propose a Bayesian model selection criterion for choosing between the unconstrained and the constrained ZIGAMs. We consider several useful extensions of the COZIGAM, including imposing additive-component-specific proportional and partial constraints, and incorporating threshold effects to account for regime shift phenomena. The new methods are illustrated with both simulated data and real applications. An R package COZIGAM has been developed for model fitting and model selection with zero-inflated data.

Mestrado em Econometria Aplicada e Previsão<br>Neste trabalho, avaliamos a capacidade da curva de Phillips salarial Neo-Keynesiana (CPSNK) proposta por Galí (2011) para descrever a inflação dos salários nos EUA durante o período 1965-2018. De forma a estudar esta relação, empregamos um modelo de regressão de limiar que nos permite examinar a existência de não-linearidades. Os nossos resultados sugerem que a taxa de inflação salarial é bem descrita por um modelo de limiar com 3 regimes em que a variável de limiar é a taxa de desemprego. As estimativas para os parâmetros de limiar dividem a CPSNK em regimes consistentes com períodos de recessão profunda, de flutuações moderadas do ciclo económico e de crescimento prolongado. Encontramos evidência empírica consistente com a relação negativa entre a inflação salarial e a taxa de desemprego prevista pela CPSNK quando a taxa de desemprego está entre os limites de 5.69% e 7.63%. Quando a taxa de desemprego está fora deste intervalo, esta relação parece desaparecer. Para avaliar a robustez das nossas estimativas, incorporamos a possível endogeneidade dos regressores e da variável de limiar ao estimar o modelo de regressão limiar estrutural proposto por Kourtellos et al. (2016). Neste contexto, concluímos que os nossos resultados não são muito diferentes quando permitimos que os regressores sejam endógenos. Por outro lado, as estimativas dos coeficientes de limiar obtidas quando a variável de limiar é considerada como endógena implicam uma redução significativa do número de observações no segundo regime.<br>The main purpose of this work is to evaluate the ability of the New Keynesian wage Phillips curve (NKWPC), proposed by Galí (2011), to describe U.S. wage inflation dynamics over the 1965-2018 period. To study this relationship, a threshold regression model that allows assessing the existence of regime-switching nonlinearity is employed. Our results suggest that wage inflation dynamics are well described by a 3-regime threshold model where the best threshold variable is the current unemployment rate. The estimated thresholds split the NKWPC into regimes consistent with periods of deep recessions, moderate business cycle fluctuations and prolonged expansions. We find evidence that the negative relationship between wage inflation and unemployment implied by the NKWPC holds when unemployment is between the thresholds 5.69% and 7.63%; when unemployment is outside this band the relationship seems to break down. To assess the robustness of our estimates, we account for possible endogeneity of the regressors and the threshold variable by using the structural threshold model proposed by Kourtellos et al. (2016). In this setting, we conclude that our baseline results are not very sensitive to endogeneity affecting the regressors. In contrast, the threshold estimates obtained when the threshold variable is considered as endogenous yield a substantial reduction in the number of observations in the second regime.<br>info:eu-repo/semantics/publishedVersion

The trustworthiness of the blood pressure (BP) readings acquired by oscillometric home-based monitoring systems is a challenging issue that requires patients to see the doctor for trusted measurements, especially those who are obese or have cardiovascular diseases such as hypertension or atrial fibrillation. Even with the most accurate monitors one may get different readings if BP is repeatedly measured. Trusted BP readings are those measured with accurate devices at proper measurement conditions. The accurate monitors need an indicator to assure the trustworthiness of the measured BP. In this work, a novel algorithm called the Dynamic Threshold Algorithm (DTA) is proposed that calculates trusted boundaries of the measured systolic and diastolic pressures from the recorded oscillometric waveforms. The DTA determines a threshold from the heart rate of subjects to locate the oscillometric pulse at the mean arterial pressure (PULSEMAP) and uses the peak, trough, and pressure of the located pulse to calculate the trusted boundaries. In terms of accuracy, a modeling approach is employed to estimate BP from the arterial lumen area oscillations model in the diastolic region (ALA-based). The model requires compliance parameter ‘c’ to estimate BP. To this end, a pre-developed linear regression model between ‘c’ and the corresponding amplitude ratio of the PULSEMAP is employed to evaluate ‘c’. The proposed method uses ‘c’ and estimates BP by minimizing differences between peak and trough amplitudes of the actual and corresponding simulated waveforms. The proposed DTA and ALA-based methods were tested on two datasets of healthy subjects and one dataset of sick subjects with cardiovascular diseases, and results were validated against corresponding references and compared with two popular maximum amplitude and maximum/minimum slope algorithms. Mean absolute error (MAE) and standard deviation of errors (STDE) are used to evaluate and compare the results. For healthy subjects, the MAE of the estimated systolic (SBP) and diastolic (DBP) blood pressures was improved up to 57% and 57% with an STDE of 55% and 62%, respectively. For sick subjects, the MAE was improved up to 40% and 29% with an STDE of 36% and 20% for SBP and DBP, respectively.

With the increasing demand and interest in wind power worldwide, it is interesting to study the effects of running windfarms on the activity of reindeers and estimate the associated Zone of Influence (ZOI) relative to these disturbances. Through simulation, Hierarchical Likelihood (HL) and adaptive Lasso methods are used to estimate the ZOI of windfarms and catching the correct threshold at which the negative effect of the disturbances on the reindeer behaviour disappears. The results found some merit to the explanation that the negative effect may not disappear abruptly and more merit to the fact that a linear model was still a better choice than the smooth polynomial models used. A real-life data related to reindeer faecal pellet counts from an area in northern Sweden were windfarms were running were analyzed. The yearly time series data was divided into three periods : before construction, during construction and during operation of the windfarms. Logistic regression, segmented model, and HL methods were implemented for data analysis by using covariates as distance from wind turbine, vegetation type, the interaction between distance to wind turbine and time period. A significant breakpoint could be estimated using the segmented model at a distance of 2.8 km from running windfarm, after which the negative effects of the windfarm on the reindeer activity disappeared. However, further work is needed for estimation of ZOI using HL method and considering other possible factors causing disturbances to the reindeer habitat and behaviour.

This thesis presents (a) results of a field evaluation of the Traffic Speed Deflectometer (TSD) in the United States (b) deflection thresholds to classify the pavement structural condition obtained from the TSD for a small subset of the Pennsylvania secondary road network. The results of the field evaluation included: (1) repeatability of the TSD, (2) ability of the TSD to identify pavement sections with varying structural conditions, and (3) consistency between the structural number (SNeff) calculated from the TSD and SNeff calculated by the Pennsylvania Department of Transportation (PennDOT). The results showed consistent error standard deviation in the TSD measurements and that the TSD was able to identify pavement sections that varied in structural condition. Comparison of the SNeff calculated with TSD measurements, using an empirically developed equation by Rohde, with the SNeff calculated by PennDOT’s Pavement Management System based on construction history showed similar trends, although the TSD-calculated SNeff was higher. In order to develop deflection thresholds, a model that related the pavement surface condition to pavement surface age and structural condition was developed. Structural condition thresholds were then selected so that the pavement surface condition predicted from the model for a 10-year-old pavement surface fell within one of the three condition categories (Good, Fair, and Poor), to identify pavements in good, fair and poor condition. With Overall Pavement Index(OPI) characterizing the surface condition and Deflection Slope Index(DSI) characterizing the structural condition, the DSI threshold that separates structurally good from structurally fair pavements was determined as follows: (1) the OPI threshold that separates pavements with good surface condition from those with fair surface condition was obtained from the Pennsylvania Pavement Management System (PMS) and (2) the DSI thresholds were calculated using the determined OPI value and the model equation.<br>Master of Science<br>This thesis presents (a) some of the results of a field evaluation of the Traffic Speed Deflectometer (TSD) in the United States (b) deflection thresholds to classify the pavement structural condition obtained from the TSD for a small subset of the Pennsylvania secondary road network. The results of the field evaluation included: (1) repeatability of the TSD: which is the variation in repeated TSD measurements on the same section of the road, (2) ability of the TSD to identify pavement sections with varying structural conditions, and (3) consistency between the structural number (SNeff) calculated from the TSD and SNeff calculated by the Pennsylvania Department of Transportation (PennDOT). The pavement structural number is an abstract number expressing the structural strength of the pavement. The results showed that the TSD measurements were repeatable and that the TSD was able to identify pavement sections that varied in structural condition. Comparison of the SNeff calculated with TSD measurements, using an empirically developed equation by Rohde, with the SNeff calculated by PennDOT Pavement Management System based on construction history showed similar trends, although the TSD-calculated SNeff was higher. In order to develop deflection thresholds to categorize pavements in different condition: good, fair and poor, a model that related the pavement surface condition to pavement surface age and structural condition was developed. Structural condition thresholds were then selected so that the pavement surface condition predicted from the model for a 10-year-old pavement surface fell within one of the three condition categories (Good, Fair, and Poor), to identify pavements in good, fair and poor condition. With Overall Pavement Index(OPI) characterizing the surface condition and Deflection Slope Index(DSI) characterizing the structural condition, the DSI threshold that separates structurally good from structurally fair pavements was determined as follows: (1) the OPI threshold that separates pavements with good surface condition from those with fair surface condition was obtained from the Pennsylvania Pavement Management System (PMS) and (2) the DSI thresholds were calculated using the determined OPI value and the model equation.

O limiar de anaerobiose (LA) é definido como a intensidade de exercício físico em que a produção de energia pelo metabolismo aeróbio é suplementada pelo metabolismo anaeróbio. Este índice constitui-se de um delimitador fisiológico de grande importância para o fornecimento de informações concernentes aos principais sistemas biológicos do organismo, os quais estão envolvidos na realização de um exercício físico. O LA é um importante parâmetro de determinação da capacidade aeróbia funcional de um indivíduo. Diversos métodos são usados para estimar o LA durante exercício. Existem métodos invasivos, como a medida repetida da concentração de lactato sanguíneo; e métodos não-invasivos, por meio de análise de variáveis biológicas como medidas contínuas dos gases respiratórios, através da análise de mudança do padrão de resposta das variáveis ventilatórias e metabólicas, e também pela análise da mudança do padrão de resposta da freqüência cardíaca (FC) frente a um exercício físico incremental. O objetivo deste estudo foi comparar e correlacionar o LA determinado por métodos não-invasivos de análise visual gráfica das variáveis ventilatórias e metabólicas, considerado como padrão-ouro neste estudo, e pelo modelo matemático de regressão linear bi-segmentado utilizando o algoritmo de Hinkley, aplicado a série de dados de FC (Hinkley – FC) e da produção de dióxido de carbono ( CO2) (Hinkley – CO2). Metodologia: Treze mulheres jovens (24 ± 2,63 anos) e dezesseis mulheres na pós-menopausa (57 ± 4,79 anos), saudáveis e sedentárias realizaram teste ergoespirométrico continuo do tipo rampa em cicloergômetro (Quinton Corival 400), com incrementos de 10 a 20 Watts/min até a exaustão física. As variáveis ventilatórias e metabólicas foram captadas respiração a respiração (CPX-D, Medical Graphics), e a FC batimento a batimento (ECAFIX, ACTIVE-E). Os dados foram analisados por testes não paramétricos de Friedman, Mann-Whitney e correlação de Spearman. Nível de significância de ? = 5%. Resultados: Os valores das variáveis potência (W), FC (bpm), consumo de oxigênio relativo ( O2) (mL/kg/min), O2 absoluto (mL/min), CO2 (mL/min) e ventilação pulmonar ( E) (L/min) no LA não apresentaram diferenças significativas entre as metodologias (p > 0,05) nos dois grupos de mulheres estudadas. A análise de correlação dos valores de potência em W, FC em bpm, O2 em mL/kg/min, O2 em mL/min, CO2 em mL/min e E em L/min, entre o método padrão-ouro com o Hinkley – CO2 foram respectivamente: rs=0,75; rs=0,57; rs=0,48; rs=0,66; rs=0,47 e rs=0,46 no grupo jovem, e rs=-0,013; rs=0,77; rs=0,88; rs=0,60; rs=0,76 e rs=0,80 no grupo pós-menopausa. Os valores de correlação do método padrão-ouro com Hinkley – FC para as variáveis potência em W, FC em bpm, O2 em mL/kg/min, O2 em mL/min, CO2 em mL/min e E em L/min, obtidas no LA foram respectivamente: rs=0,58; rs=0,42; rs=0,61; rs=0,57; rs=0,33 e rs=0,39 no grupo de jovens, e rs=0,14; rs=0,87; rs=0,76; rs=0,52; rs=0,33 e rs=0,65 no grupo pós-menopausa. O grupo pós-menopausa apresentou melhores valores de correlação em relação ao grupo de jovens, exceto para as variáveis potência e consumo de oxigênio absoluto (mL/min). Este fato pode estar relacionado a uma maior taxa de variação e magnitude das variáveis analisadas em indivíduos jovens em relação aos de meia-idade, sendo, desta forma, obtida melhor adequação do modelo matemático estudado em mulheres de meia idade. Conclusão: O algoritmo matemático de Hinkley proposto para detectar a mudança no padrão de resposta da CO2 e da FC foi eficiente nos indivíduos de meia-idade, portanto, a metodologia matemática utilizada no presente estudo constitui-se de uma ferramenta promissora para detectar o LA em mulheres saudáveis, por ser um método semi-automatizado, não invasivo e objetivo na determinação do LA.<br>The anaerobic threshold (AT) is defined as the intensity level of physical exercise at which energy production by aerobic metabolism is supplemented by anaerobic metabolism. This index provides a physiologic delimitation of great importance to supply the organism biological systems information involved in physical exercise performance. The AT constitutes a most important determining of an individuals functional aerobic capacity. Several methods are used for estimating the AT during exercise. There are invasive methods that require repeated blood lactate accumulation; and there exist non-invasive methods by biological variables analysis, like continuous respiratory gases determination by analysis of changes in pattern respiratory and metabolic responses, and heart rate (HR) responses too. The aim of the present study was to compare AT obtained by a graphic visual method of ventilatory and metabolic variables, considered by gold standard method in the present study, with the bi-segmental linear regression mathematic model of Hinkley’s algorithm applied in a HR (Hinkley – HR) and carbon dioxide output ( CO2) (Hinkley – CO2) data. Methodology: Thirteen young women, 24 ± 2,63 years old, and sixteen postmenopausal women, 57 ± 4,79 years old, leading healthy and sedentary life style were submitted to an incremental test in a cicloergometer electromagnetic braking (Quinton Corival 400), with 10 to 20 W/min increments up to physical exhaustion. The ventilatory variables were registered breath-to-breath (CPX-D, Medical Graphics) and HR was obtained beat-to-beat (ECAFIX, ACTIVE-E), over real time. The data were analyzed by Friedman’s test and Spearman’s correlation test, with a level of significance set at 5%. Results: The Power output (W), HR (bpm), oxygen uptake ( O2) (mL/kg/min), O2 (mL/min), CO2 (mL/min) and pulmonary ventilation ( E) (L/min) data in AT have showed no significant differences (p > 0,05) between methods to determine AT in both women groups. The correlation analysis of power output in W, HR in bpm, O2 in mL/kg/min, O2 in mL/min, CO2 in mL/min and E in L/min values, determined by gold standard method and by Hinkley – CO2 data were respectively: rs=0,75; rs=0,57; rs=0,48; rs=0,66; rs=0,47 and rs=0,46 in young group, and rs=-0,013; rs=0,77; rs=0,88; rs=0,60; rs=0,76 and rs=0,80 in postmenopausal group. The correlation analysis by gold standard method and Hinkley – FC in AT of power output in W, HR in bpm, O2 in mL/kg/min, O2 in mL/min, CO2 in mL/min and E in L/min data were respectively: rs=0,58; rs=0,42; rs=0,61; rs=0,57; rs=0,33 and rs=0,39 in young group, and rs=0,14; rs=0,87; rs=0,76; rs=0,52; rs=0,33 and rs=0,65 in postmenopausal group. The postmenopausal group presents better correlations values than young group, except in power output and O2 (mL/min) data. This may be related to more variability rate and higher kinetics responses to variables studied in young group in relation to postmenopausal group. Nevertheless, there was obtained better mathematical model adequacy in middle-age women. Conclusion: the Hinkley’s mathematical algorithm proposed to detect the response patterns changes of CO2 and HR variables was efficient to detect AT in health postmenopausal women’s group, therefore, the mathematical methodology used in the present study showed be a promissory tool because this method represent a semi-automatized, non invasive and objective measure of AT determination.

碩士<br>國立嘉義大學<br>應用數學系研究所<br>106<br>Clinically, homocysteine has been shown to be an important risk factor for heart disease and can be effectively managed by monitoring. In real case, if the detected value is normal below a certain value and is abnormal after exceeding a certain value, this specific value is called “threshold”. In this study, four threshold regression models proposed by Fong et al. (2017) are used to explain the change of homocysteine before or after the threshold. Therefore, consider the homocysteine as a threshold variable and introduce its associated explanatory variable construction threshold regression model. Finally, the four threshold regression models can be divided into two modes: "acute" and "chronic". The threshold regression model can be used to find out that the homocysteine value of patients with different patterns at 16 μmol/L.

博士<br>國立政治大學<br>經濟學系<br>90<br>This dissertation includes three financial applications using time series methods. The first article investigates the asymmetric effects of information transmissions in stock markets using threshold vector autoregression model. The second article uses asymmetric cointegration to study the long-run equilibium relationships among Chinese stock markets. The third article uses panal cointegration to test purchasing-power parity (PPP). Firstly, we examines the asymmetric effects of information transmissions of Nasdaq stock market on Taiwan, Japan, and Korea stock markets by using a threshold vector autoregressive model. And also, we check whether Nasdaq stock market have different impacts on organized stock exchanges (including TAIEX, NIKKEI 225 Index, Korea Composite Index) and over-the-counter markets (including Taisdaq Index, Jasdaq Index, and Kosdaq Index) or not. The empirical results indicate that negative innovations in Nasdaq market (bad news regime) have large influence on Asia stock markets. Particularly, the positive innovations in Nasdaq market (good news regime) have small influence on Asia stock market. The large negative innovations in Nasdaq market have great influence than those of the large positive innovations on Jasdaq Index and Kosdaq Index in bad news regime. The second article uses Enders and Siklos''s (2001) asymmetric cointegration model to investigate the long-run asymmetric equilibrium relationships. The empirical results find that there exits an asymmetric cointegrated relationship between Shanghai A share index and Shenzhen A share index for the period from October 1992 to August 2001. The adjustment parameters of error correction term at Shanghai A share market are larger in bad-news regime than those in good-news regime. This result reveals investors at Shanghai possess over-reaction behavior on news of stock market. Moreover, there exists a bi-directional Granger causality between Shanghai A share index and Shenzhen A share index. We find there exists an asymmetric cointegrated relationship between Shenzhen A share index and Shenzhen B share index after 19 February 2001. Furthermore, the Shenzhen B share index leads Shenzhen A share index after 19 February 2001. The third article uses Pedroni''s (2001) panel cointegration test to examine the validity of PPP hypothesis by two different price indces, i.e. Big Mac prices and CPI. Our panel observations include 14 countries from 1992 to 1999. The empirical evidence indicates Big Mac PPP and CPI PPP is supposed if we use nominal exchange rate as the explanatory variable. Nevertheless, the Big Mac PPP is valid but CPI PPP not valid if we use price level as the explanatory variable. Moveover, our concludtion does not influenced by productivity bias.

碩士<br>國立暨南國際大學<br>財務金融學系<br>98<br>This paper examines whether the corporation’s default point is different from KMV model without the default distribution. First, we adopt Probit regression model and Merton option model to estimate the past expected default points and then exam these default points. Second, we use threshold regression model to test whether the debt ratio threshold variable exists between default point and short term and long term debt. Finally, we verify whether the coefficients of short term and long term debt are different from the KMV model. Our empirical results indicate that the past expected default points differ from the KMV model. According to leverage ratio, there is a structural change between historical default points and short term and long term debt. The results shows the coefficients of short term and long term debt are almost significantly greater than the KMV model

碩士<br>國立雲林科技大學<br>財務金融系碩士班<br>95<br>Whether investment could make a profit, is the key point that each investor cares about most. There were a lot of research results on returns of Taiwan stock, but whether the structural change will take place under different control parameters to explain the relationship between the factor of parameters and returns of Taiwan stock relatively rare. In this paper the methods are applied to a 10-year sample of 283 Taiwan firms. We use the threshold regression model with panel data to test the threshold effect and estimate the explanation ability for dependent variable and independent variables. We investigate whether earnings per share (EPS) existing threshold effect on Taiwan stock returns. We also incorporate the return on equity (ROE), growth ratio of return on asset, retained earnings to equity ratio, and gross profit ratio variables to analyze the direction and impact among the different regimes of EPS on Taiwan stock returns. The empirical result shows that EPS, ROE, and growth ratio of return on asset are statistically significant, indicating a positive relationship with stock returns. Taking the EPS as the threshold variable, we conclude that the threshold effect exists; there are three thresholds and four regimes between the ROE and Taiwan stock returns.

碩士<br>義守大學<br>財務金融學系<br>100<br>This study employs a threshold regression model to see if there is a nonlinear relationship between the interest rates and stock index before and after the central banks cut interest rates. This study uses the data from America, Australia, India and Taiwan, and divides the data into two periods: the period when central banks increase the interest rates and the period when central banks cut the interest rates, and investigate whether there are threshold effects of interest rates on stock prices. To avoid the spurious regression problem, this study applies the cointegration test to see if there is long-run relationship among model’s variables. Based on the traditional economic theory, the lower the interest rate is, the lower the cost of capital is and the more investment incentives there are, so interest rates and stock prices should be negatively correlated. However, empirical results indicate that interest rates and stock prices are positively correlated as the interest rates start to increase; the negative relationship appears only after interest rates cross above a threshold. In the period when central banks cut the interest rates, there is a positive relationship between interest rates and stock prices as the interest rates start to fall, the negative relationship appears only after interest rates fall below a threshold. In addition, the cointegration test finds a significant cointegration relationship before and after central banks cut the interest rates. The threshold regression model shows a nonlinear relationship that differs from the traditional economic wisdom; it not only provides investment strategies for stock investors but also policy implications for central banks.

碩士<br>國立中山大學<br>經濟學研究所<br>102<br>We employ the panel threshold regression model provided by Hansen (1999) to examine the relationship between inflation and economic growth for seventeen countries from 1971 to 2010. The key innovation of the model is to emphasize the inference of financial development on the relationship between inflation and economic growth. Empirical results indicate that the relationship between inflation and economic growth exist a single threshold effect. For high financial development countries, positive and significant relationship between inflation and economic growth are observed, but the positive relationship fails to be observed for low financial development countries.

碩士<br>國立臺北科技大學<br>商業自動化與管理研究所<br>94<br>In recent years, corporate governance seems a popular financial issue after Enron and World Com such cases burst out. In these cases, the fraudulent techniques mostly come from the accounting management, and “income-smoothing” is one of the commonest ways. The managers usually make use of transitory earnings（or non-operating earnings）to smooth the income numbers in the financial statement so as to fascinate investors. In this research, the author makes use of the threshold variable “operating earnings”, and wants to find out the threshold point(s) which divides the sample into regimes. In or out of one regime, the power and influencing direction of main explaining variable (non-operating earnings) would be different. The result shows that in Taiwan stock market form 2000 to 2004, the relationship between non-operating income and corporate value actually can be divided into three regimes (two threshold points) with threshold variable “operating earnings”. With the operating earnings of our sample companies perform better and better, it is clear that non-operating earnings do more and more help to the corporate values.

In this dissertation, a new non-global regression model - the partial linear threshold regression model (PLTRM) - is proposed. Various issues related to the PLTRM are discussed. In the first main section of the dissertation (Chapter 2), we define what is meant by the term “non-global regression model”, and we provide a brief review of the current literature associated with such models. In particular, we focus on their advantages and disadvantages in terms of their statistical properties. Because there are some weaknesses in the existing non-global regression models, we propose the PLTRM. The PLTRM combines non-parametric modelling with the traditional threshold regression models (TRMs), and hence can be thought of as an extension of the later models. We verify the performance of the PLTRM through a series of Monte Carlo simulation experiments. These experiments use a simulated data set that exhibits partial linear and partial nonlinear characteristics, and the PLTRM out-performs several competing parametric and non-parametric models in terms of the Mean Squared Error (MSE) of the within-sample fit. In the second main section of this dissertation (Chapter 3), we propose a method of estimation for the PLTRM. This requires estimating the parameters of the parametric part of the model; estimating the threshold; and fitting the non-parametric component of the model. An “unbalanced penalized least squares” approach is used. This involves using restricted penalized regression spline and smoothing spline techniques for the non-parametric component of the model; the least squares method for the linear parametric part of the model; together with a search procedure to estimate the threshold value. This estimation procedure is discussed for three mutually exclusive situations, which are classified according to the way in which the two components of the PLTRM “join” at the threshold. Bootstrap sampling distributions of the estimators are provided using the parametric bootstrap technique. The various estimators appear to have good sampling properties in most of the situations that are considered. Inference issues such as hypothesis testing and confidence interval construction for the PLTRM are also investigated. In the third main section of the dissertation (Chapter 4), we illustrate the usefulness of the PLTRM, and the application of the proposed estimation methods, by modelling various real-world data sets. These examples demonstrate both the good statistical performance, and the great application potential, of the PLTRM.<br>Graduate

碩士<br>國立高雄第一科技大學<br>財務管理所<br>97<br>The previous study finds that the variables of the company characteristics are the main decision of exchange rate exposure, but the dependence of the variables of the company characteristics and exchange rate exposure doesn’t have conclusion in concert. Therefore, this paper applies non-linear threshold regression model to analyze the relationship between changes in exchange rate and firm valuation. The sample is gathered from the quarterly data of the listed multinational corporations in Taiwan for the period 1998-2008. Empirical evidence showed that using export ratio, firm size and book value to market value ratio as the threshold variable. We find statistically significant evidence of a single threshold effect. All of them indicate that exchange rate exposure is effected by company characteristics to perform asymmetric relationship.

碩士<br>國立成功大學<br>統計學系碩博士班<br>95<br>The researcher use Hansen Panel Threshold Regression model to verify Taiwan stock market have phenomenon of unusual return. To verify size effect, E/P effect and BV/MV effect related with unusual return of Taiwan stock market. We would further to explore that unusual return is effected by company size and BV/MV to perform unsymmetrical relationship. Therefore, we need to understand level and fluctuation based on different companies.We found that company size have negative relationship with return and E/P have positive relationship with return from tests. The Taiwan stock market have size effect and E/P effect. BV/MV and unusual return have double threshold effect. Depending on threshold areas to divide three unsymmetrical areas have negative, positive and positive performances. Most companies are belonged to last two areas. The higher BV/MV is, the higher the average rate of unusual return is. Therefore, BV/MV effect have direct proportion with unusual return. With company size, E/P and BV/MV, we found that company size is reacted with BV/MV and others is not related.

碩士<br>國立東華大學<br>自然資源與環境學系<br>103<br>Abstract According to the need of environment conservation, recreation, and local economic development, natural-based tourism has become one of the fastest growing sectors of the global tourism industry. Since the Taiwan government began implementing its policy of a two-day weekend in 2001, the general public has a longer weekend holiday and the quality of the recreational experience has begun to attract the attention of the public. Previous research has emphasized the perceptions of crowding and user satisfaction. Researchers and managers also consistently engage in assessing the most appropriate recreation carrying capacity to maximize user satisfaction. Hansen’s (1999) threshold regression model is adopted to examine whether visitors’ crowding perception varies with the degree of the growth rates of visitor arrivals, which is defined as the threshold variable. By using the quarterly survey reports of the visitors’ satisfaction of the Yangmingshan National Park during the period 2010-2014, the empirical results show that there indeed exists a threshold effect of the growth rate of visitor arrivals in the crowding perception link. Specifically, visitors’ crowding perception increases more at the growth rate of visitor arrivals higher than -6.07% than at the growth rate of visitor arrivals no greater than -6.07%. These findings may provide the officers served in the Yangmingshan National Park to evaluate an appropriate recreation carrying capacity to maximize visitors' perception of satisfaction.

碩士<br>國立臺北大學<br>國際財務金融碩士在職專班<br>107<br>The business model of commercial banks mainly focuses on absorbing the public deposits and lending to the capital demander. Banks play the role of credit intermediary. Therefore, credit risk is the biggest risk for banks. With the quantification of risk assessment standard by Basel and IFRS, credit risk model has become an important tool to control risks at financial institutions. This study is based on Taiwan listed and OTC companies (including emerging stock market) of the manufacturing industry during from 2000 to 2017. Paired samples of having one defaulted company paired with two non-defaulted companies have been adopted. There are four financial reports including the defaulted year (T) and the previous three years (T-1, T-2, T-3) for each company. In total there are 668 financial reports of defaulted companies and 1,312 reports of non-defaulted companies in this study. Furthermore, the paired samples have been divided into two groups - 80% of in-sample and 20% of out-of-sample. In-sample is used to build up the model and out-of-sample is used to validate the model. Logit model was adopted so as to build a stable early warning model of financial distress. Then threshold logistic regression is applied to Logit model, hence this study is called “Threshold Logistic Regression Model”. The debt ratio is be the threshold variable. It is to explore the changes of financial features after considering threshold variable by the model. The final model has seven variables including debt ratio, the ratio of working capital to total asset, total asset turnover, the ratio of retained earnings to asset, basic earning power, shareholding ratio of directors and company scale. Each variable has significant effects for the prediction of company’s financial distress. The results of validation show that the AUC of in-sample is 88.80% and the predicting accuracy is 83.17%; the AUC of out-of-sample is 86.78% and the predicting accuracy is 82.65%. In the defaulted year and the previous three years, the predicting accuracy of in-sample is 87.66%, 87.12%, 81.12% and 76.67% respectively. The more long away from the defaulted time, the more declining the discriminating power the model is and the validation result of out-of-sample is the same as in-sample. However, the declining degree is not serious and it represents that the model has a good and stable prediction. In consideration of the threshold effect, the threshold value of debt ratio is 60.488% and some variables are not significant effects for financial distress prediction in each group, including the “debt ratio”, “the ratio of retained earnings to asset” in high debt group and the“total asset turnover” in low debt group. It means that companies with different debt level have the different causes of financial distress. Besides, the variable “company scale” in high debt group shows that the bigger the company scale is, the higher chance to have financial distress. This conclusion is different from past studies showing that the large companies have lower chance to have financial distress. The results of coefficient test show that the two variables “the ratio of working capital to total asset” and “company scale” in the two groups of high debt and low debt are different. Under the threshold effect of debt ratio, it shows different impact on leading to financial distress.

博士<br>國立中山大學<br>財務管理學系研究所<br>95<br>The number of insurance company has grown rapidly in Taiwan due to insurance deregulation since 1992. The main challenge insurance industry face is the declination of profit due to the increasing of competitors. The operator of insurance company is able to face this question and offer the solution, then a company has better solvency. So we explore two issue, one is to investigate the relationship between asset risk and capital adjustment decision in Taiwan’s life insurance industry from 1993 to 2005, and the other is to provide some empirical evidences of retention limit of excess of loss reinsurance in Taiwan’s property insurance industry. In the first issue, a life insurance company is in less risk and has better solvency when it has more capital or higher ratio of capital; however, this also brings higher opportunity cost which means in long run, the average profit will be lower. There is no conclusion how to balance the relationship between capital adjustment and risk taking decision in life insurance industry though this topic is intensively discussed these days. Therefore, with the methodology of panel data threshold regression, we divide life insurance companies into two categories according to “life insurance and annuity insurance premiums to total premiums ratio”. One is life insurance Company of indemnification, and the other is the one of savings. In conclusion, we identify the negative correlation between capital ratio and risk of life insurance company of indemnification and the positive correlation between capital ratio and risk of life insurance company of savings. In the second issue, because of the increase of natural disaster in Taiwan recently, the property insurance company has to face what the reinsurance companies are not willing to underwriter, so excess of loss reinsurance has become the viable solution in Taiwan’s property insurance industry. We apply extreme value theory to the tail of Taiwan property insurance claim for VaR estimation and calculate retention limit of excess of loss reinsurance. The empirical results show that the distribution of Taiwan property insurance claim is fat-tailed. We suggested using Generalized Pareto Distribution (GPD) to model the data with extreme loss and conclude retention limit of excess of loss reinsurance.

碩士<br>淡江大學<br>財務金融學系碩士在職專班<br>99<br>Since 2008, Taiwan authorities has allowed mainland tourists to visit Taiwan. The number of the tourists for leisure has grew up fast, and that growth brought Taiwan tourism into a ‘boom’ and also the study of international tourist hotels in Taiwan become the contemporary popular doctrine. This study hypothesize that there exist threshold effects of five important explanatory variables on the business performance of Taipei international business hotel. Therefore, this study empirically employs the panel threshold regression model elaborated by Hansen (1999) to fulfill the aims of this research. However, the empirical results failed to reject the null hypothesis of no threshold, so that we apply the traditional multiple regression model for our further evaluation for the business performance of international tourist hotels. The significant findings of our traditional LS provide an easy understanding of strategy implications. In the situation that it is difficult to control the number of the inbound passengers to Taiwan and that the number of hotel rooms (size) is not easy to expand in the short term, the empirical results show significant positive effects of four strategy variables, PRI、POT、GOT and ROT, on housing rates, the proxy for business performance of international hotel. The positive effects of PRI and ROT on housing rates implies that the higher price of the average room fee comes with the better service for entertaining the hotel guests and thus attracting more tourists to international tourist hotels in Taipei. This evidence is in conformity with the “high price / high quality” managerial strategy. The significantly positive effects of POT and GOT on housing rates implies that the marketing strategy of international tourist hotel in Taipei should be oriented by the group tourists. The implication of the above strategies is in accordance with the pllicy nowadays by Taiwan authorities for allowing mainland visitors to Taiwan. The strategy that high quality services come along with the high price of room fee bringing more tour groups seems an urgent matter to managerial staffs for increasing the business performance.of Taipei international tourist hotel.

碩士<br>中原大學<br>國際經營與貿易研究所<br>105<br>Abstract This paper adopts panel smooth transition autoregression (PSTAR) models to evaluate the threshold effects of price level and unemployment on lottery sales. To perform the empirical estimation, we employ the panel data of lottery sales in 42 US states during the period of 2003 to December 2011. Thus, we have 4536 observations. The empirical results are summarized as follows: 1. The threshold effects of price level and unemployment on lottery sales are nonlinear. 2. A rise in unemployment will lead to the decrease of a family’s income. To compensate for the decrease, the members of the family increase the incentive for purchasing lottery tickets, which causes a more volatile in lottery sales and a smaller persistence of lottery sales. Contrarily, a decrease in unemployment will increase a family’s income, which decreases the incentive for purchasing lottery tickets and the volatile in lottery sales, and increases the persistence of lottery sales. 3. A rise in consumer price level will reduce the real income of a family. To compensate for the loss in real income, the members of the family have higher incentive for purchasing lottery tickets. Again, this will cause a more volatile in lottery sales and a smaller persistence of lottery sales. On the contrary, a decrease in consumer price level will increase a family’s real income, which decreases the incentive for purchasing lottery tickets and the volatile in lottery sales, and increases the persistence of lottery sales.

碩士<br>嶺東科技大學<br>高階主管企管碩士在職專班<br>106<br>Nonlinear Analysis of Capital Structure and Operating Performance in Chinese Beverage Listed Companies: Evidence from a Panel Threshold Regression Model Student: Hui-Chuan Lin Advisor: Dr. Hsu-Ling Chang Executive Master of Business Administration (EMBA) Department of Business Administration, Ling Tung University Abstract This thesis mainly focuses on the relationship between capital structure and corporate performance in China beverage listed companies. Both qualitative analysis and quantitative analysis are applied for the investigation. The core content of this thesis is the construction of panel threshold regression model. Specifically, panel threshold regression model is utilized to find out whether Chinese beverage listed company has an optimal capital structure. On this basis, this thesis devotes to reveal the inherent relation between capital structure and corporate performance in China. In this thesis, a sample of Chinese A-share listed beverage companies is selected, and the time period is during 2001 to 2016. In order to ensure the robustness of the empirical results, this thesis uses the asset liability ratio and the ownership concentration as indicators of capital structure. The Tobin's Q value and the rate of return on net assets are utilized as a listed company operating performance indicators, and the scale of the company's assets and company growth are utilized as the control variables. The results show that both Tobin's Q and return on assets have a single threshold effect on asset liability ratio and ownership concentration. Specifically, the capital structure and operational performance show a significant inverted U relationship, which shows that China's A-share beverage listed companies generally have the optimal capital structure for the maximization of the operational performance. However, many listed companies deviate from the optimal capital structure in reality. This study provides empirical evidence for the determination of the optimal capital structure of China's A-share beverage listed companies. Finally, based on the empirical results, this paper puts forward the irrational improvement of the capital structure. Keywords: Beverage Listed Companies; Capital Structure; Operational Performance; Panel Threshold Regression Model

碩士<br>淡江大學<br>財務金融學系碩士班<br>100<br>In this study, we examine the asymmetrical nonlinear relationships and this paper employs the panel smooth transition regression model (PSTR) technique. We assume purchasing managers’ Index for transition variables in this model to investigate the smooth transition threshold effect between stock index returns. The results as follow: Wherther purchasing managers’ Index is before or after the transition threshold value, we can’t find a relationship between exchange rate and stock index returns. When purchasing managers’ Index is before the transition threshold value,we find a negative relationship between inflation index and stock index returns.When purchasing managers’ Index is after the transition threshold value,we find a positive relationship between inflation index and stock index returns.

Tese de doutoramento em Economia apresentada à Faculdade de Economia da Universidade de Coimbra<br>In this dissertation we take an empirical approach to the study of the impact of financial development on consumption. We use four different methodologies, which stress different aspects of consumption behavior and different ways of considering financial development's effect. We start by estimating a Panel Vector Autoregression (PVAR) model to study the relation between a new broad measure of financial development and a set of core macroeconomic variables which includes consumption. The results from these estimations suggest that while financial development has a positive impact on Gross Domestic Product (GDP) and on investment, the evidence of an impact of financial development on consumption is weaker. These results survive a battery of robustness checks and are stronger in a subsample of countries with higher levels of financial development. In the second empirical chapter, using the Panel Smooth Transition Regression (PSTR) method, a Panel Error Correction Model (ECM) is estimated for consumption, in order to study how the relation between consumption and its determinants changes for different values of financial development. The results suggest that higher financial development leads to a smaller reaction of consumption to its determinants in the short run, but also to a faster adjustment towards the long run equilibrium. In the third empirical chapter, we apply Panel Threshold Regression (PTR) techniques and a First Difference Generalized Method of Moments (FD-GMM) methodology to estimate the threshold model. The difference is in the assumption of exogeneity of the explanatory variables. PTR assumes exogeneity while FD-GMM does not. We regress consumption on income, wealth and the interest rate and consider as measures of financial development both a broad measure of financial system development and a measure of credit. In both cases we find evidence of a threshold in the financial sector variable which affects how consumption responds to the other variables in the model. Finally, we use nonparametric regression methods in our fourth empirical chapter. The model, which includes lagged consumption, income, wealth, the interest rate and financial development, aims at describing in a more general way the kind of link one can find between consumption and financial development, as well as providing some evidence of the shape of the relationship between consumption and its typical explanatory variables. Our results point towards distinctly nonlinear links between consumption and the regressors. As for financial development, we find that its effect over consumption is highly dependent on the value of the remaining regressors.<br>Nesta dissertação estudamos o impacto do desenvolvimento do sistema financeiro sobre o consumo através de uma abordagem empírica. Para tal usamos quatro metodologias, as quais salientam aspectos diferentes do comportamento do consumo e formas diferentes de incorporar o efeito do desenvolvimento do sistema financeiro. Começamos por estimar um modelo Panel Vector Autoregression (PVAR) para estudar a relação entre uma nova medida lata do desenvolvimento financeiro e um conjunto de variáveis macroeconómicas cruciais, entre as quais se inclui o consumo. Os resultados das estimações sugerem que o desenvolvimento do sistema financeiro tem um impacto positivo sobre o produto interno bruto (PIB) e sobre o investimento, mas que o efeito sobre o consumo é menos notório. Estes resultados sobrevivem a uma bateria de testes de robustez e são mais fortes numa subamostra de países com níveis de desenvolvimento financeiro mais elevados. No segundo capítulo empírico, usamos o procedimento de Panel Smooth Transition Regression (PSTR) para estimar um modelo Panel Error Correction Model (ECM) para o consumo, de forma a averiguar como varia a relação entre o consumo e os seus determinantes quando o nível de desenvolvimento financeiro toma valores diferentes. Os resultados indicam que o aumento do nível de desenvolvimento financeiro reduz a reacção do consumo aos seus determinantes no curto prazo, mas também conduz a um ajustamento mais rápido em direcção ao equilíbrio de longo prazo. No terceiro capítulo empírico, aplicamos as metodologias de Panel Threshold Regression (PTR) e First Difference Generalized Method of Moments (FD-GMM) para estimar com modelo com efeito threshold. A diferença reside na hipótese de exogeneidade das variáveis explicativas: o procedimento PTR supõe exogeneidade e o procedimento FD-GMM não. Estimamos o modelo para o consumo tendo como variáveis explicativas o rendimento, a riqueza e a taxa de juro, e considerando como medidas do nível de desenvolvimento financeiro ou uma medida lata do desenvolvimento do sistema financeiro ou o crédito. Em ambos os casos encontramos indícios da existência de um efeito threshold do nível de desenvolvimento financeiro sobre a forma como o consumo responde às outras variáveis do modelo. Finalmente, usamos métodos de regressão não paramétricos no quarto capítulo empírico. O modelo, que inclui o consumo desfasado, o rendimento, a riqueza, a taxa de juro e o nível de desenvolvimento financeiro, procura descrever de forma mais geral o tipo de ligação que se pode encontrar entre o consumo e o nível de desenvolvimento financeiro, bem como ser informativo quanto à forma funcional da relação entre o consumo e os seus determinantes típicos. Os nossos resultados apontam no sentido de as ligações entre o consumo e os seus determinantes serem claramente não lineares. Quanto ao nível de desenvolvimento financeiro, concluímos que o seu efeito sobre o consumo depende fortemente do valor das restantes variáveis.

The main purpose of this study is to empirically assess the relationship between inflation and economic growth in Ethiopia using quarterly dataset from 1992Q1 to 2010Q4. In doing so, an interesting policy issue arises. What is the threshold level of inflation for the Ethiopian economy? Based on the Engle-Granger and Johansen co-integration tests it is found out that there is a positive long-run relationship between inflation and economic growth. The error correction models show that in cases of short-run disequilibrium, the inflation model adjusts itself to its long-run path correcting roughly 40% of the imbalance in each quarter. In addition, based on the conditional least square technique, the estimated threshold model suggests 10% as the optimal level of inflation that facilitates growth. An inflation level higher or lower than the threshold level of inflation affects the economic growth negatively and hence fiscal and monetary policy coordination is vital to keep inflation at the threshold.<br>Economics<br>M. Com. (Economics)

Background: This study compares three methodologies appropriate for the analysis of longitudinal time-to-event data. The Cox model is well researched and frequently used. Threshold regression, however, is relatively new and there are few articles describing its application in biomedical statistics. A linear mixed model provides an alternative interpretation of a continuous outcome rather than time to an event. A longitudinal study of the time to onset of diabetic nephropathy, a common complication of Diabetes Mellitus, is used to compare the three models with respect to their explanatory and predictive abilities and utilitarian value to researchers. Methods: The study entails a secondary data analysis of 1160 retrospective patient records, collected at a diabetic clinic at Kalafong Hospital, Pretoria. Model selection was based on current literature, backward elimination of insignificant variables (p>0.2) and the Akaike and Bayesian Information Criterion. Survival and hazard functions and ratios were determined for the survival data. Risk categories in the Cox model evaluated discrimination, while threshold regression predicted survival probabilities for specific patient profiles. The linear mixed model predicted albumin-creatinine ratio values, a marker for the diagnosis of diabetic nephropathy. Results: The Cox model, stratified by glucose control, gender, hypertension, type of diabetes and smoking status, had an AIC of 81 and was the most parsimonious model. Threshold regression, with an AIC of 1428, indicated duration of diabetes as a significant factor in the process of health deterioration. Individual variation in weight and total cholesterol amongst patients was accounted for by the linear mixed model, with an AIC of 3755. Conclusion: All three regression models provided valuable insight into underlying risk factors of diabetic nephropathy and should form part of a multi-faceted approach to analysing longitudinal survival data.<br>Dissertation (MSc)--University of Pretoria, 2014.<br>lk2014<br>School of Health Systems and Public Health (SHSPH)<br>MSc<br>Unrestricted