Esta tese é composta por três ensaios que versam sobre a complexa relação entre desenvolvimento e meio ambiente. No primeiro ensaio é feita uma revisão das principais abordagens teóricas que tratam do assunto, com destaque para a Curva Ambiental de Kuznets (CAK) e para os conceitos de sustentabilidade fraca e forte. Ambos apontam o crescimento econômico como um importante indutor de impacto ambiental. No entanto, a CAK afirma que a geração de riqueza pode ser aliada na proteção ambiental quando o país alcança um patamar de desenvolvimento suficientemente elevado para que o meio ambiente passe a ser incorporado como um bem superior. Já a conceituação de sustentabilidade não reforça esta conclusão ao passo que sugere que o capital natural precisa ser preservado: seja relativamente (conceito fraco) ou absolutamente (conceito forte). Para estes, a relação entre crescimento e impacto é sempre direta. As abordagens multidimensionais vão além e afirmam que a relação entre o homem e os ecossistemas se dá uma forma impossível de ser mensurada através da simples relação entre crescimento econômico e degradação. Para se aproximar minimamente deste modo de pensar, um conjunto maior de indutores precisa ser considerado. O ensaio 2 desta tese apresenta modelos na literatura que buscam estimar impactos ambientais causados pela atividade humana através de um número maior de indutores (produtivos, populacionais, institucionais, entre outros). Estes modelos (IPAT e STIRPAT) têm como característica a possibilidade de estimar o impacto ambiental esperado através da verificação das variáveis consideradas indutoras. Mas, duas fragilidades são especialmente importantes: i) a variável que mede impacto ambiental é, em geral, reportada com poluição apenas; e, ii) os modelos não fornecem cenários não lineares nos indutores, ou seja, o patamar das variáveis não faz diferença no tamanho do impacto gerado. Então, é proposta a aplicação de um modelo logístico ordenado que seja capaz de estimar as probabilidades de diferentes níveis de impacto ambiental, que é medido através do tamanho da Pegada Ecológica dos países. A metodologia proposta, além de contar com uma medida mais ampla e completa de impacto ambiental, também é capaz de fornecer cenários com maior riqueza informativa, pois levam em consideração os níveis das variáveis indutoras. No trabalho, foram avaliados cenários a partir de um modelo que contou com as seguintes variáveis: percentual de população urbana, percentual de população entre 15-64 anos, densidade demográfica, PIB per capita e percentual de área agrícola. Os sinais e magnitudes das variáveis utilizadas são condizentes com o que a literatura vem apresentando. O diferencial da proposta é, de fato, a possiblidade de análise de cenários através de outra perspectiva. O ensaio 3 da tese aplica este modelo estimado para o globo aos municípios gaúchos. A partir da suposição de que os impactos ambientais são homogêneos ao redor do mundo, foi possível estimar as probabilidades de impacto para cada município. Com estes resultados, foi elaborada uma análise regional dos potenciais impactos ambientais dos municípios a partir dos indutores que apresentavam. Trata-se de um exercício interessante quando se considerada a carência de informações ambientais neste nível de desagregação. / This thesis presents three essays on the complex relationship between development and the environment. On the first essay is performed a survey of the main theoretical approaches on this matter with an emphasis on the Environmental Kuznets Curve (EKC) and the weak and strong sustainability concepts. Both approaches points out the economic growth as an important driver of environmental impact. However, the EKC sustains that the pursuit of economic wealth can be responsible for environmental protection as the country reaches a certain level of income which turns the environment into a superior good. On the other hand, the concepts of sustainability do not enforce this conclusion either in relative terms (weak concept) or absolute terms (strong concept); in this scope, the relationship between growth and environment is constantly direct. The multidimensional approaches propose that the relationship between humankind and the ecosystems are established in a way which is impossible to measure through economic growth and environmental degradation. To become closer to this sort of approach it is necessary to consider a wider set of drivers. The essay 2 of this thesis analyses some statistical models that aim to estimate environmental impacts caused by human activity based on a larger number of drivers (production, population, institutions, etc.). These models (IPAT and STIRPAT) present as a major characteristic the ability to estimate the expected environmental impact grounded on the drivers considered. Nevertheless, two important caveats should be considered: i) the impact variable is most of the time just a pollution measure; and, ii) the models do not offer a nonlinear analysis on the drivers, i.e., the level of the drivers does not make any difference in terms of impact. So, we propose an ordered logistic model that is suitable for estimating the probabilities of different levels of environmental impact which is represented by the size of the Environmental Footprint. The proposed methodology, besides counting on a wider and more complete measure of environmental impact, is capable to offer scenarios with a richer degree of information because the level are effectively taken into consideration. The study has evaluated scenarios base on the following drivers: percentage of population in urban areas, percentage of population between 15-64 years old, demographic density, GDP per capita, and the percentage of agricultural land. The signs and magnitudes of the drivers have proven to be the same as those found in the literature. The main difference of this proposal is, in fact, the possibility of analyzing scenarios through a different (wider) perspective. The essay 3 applies the model, which was prior estimated for the globe, to the municipalities of the state of Rio Grande do Sul, Brazil. Based on the assumption that the environmental impact generated by those drivers is homogeneous around the world, it was possible to estimate the probabilities of environmental impact for each municipality and to draw a regional study of these results. It is an interesting exercise if you take into account that there exists a huge lack of information of environmental conditions for municipalities.

Doctor of Philosophy / Department of Civil Engineering / Sunanda Dissanayake / Young drivers are over-represented in crashes when compared to other age group drivers. Distracted driving is one of the major causes of traffic crashes by young drivers. The objective of this study was to assess the hazards of distracted driving among teenage (15–20 year old) and young-adult (21–26 year old) drivers in Kansas. This study used five years of crash data from the Kansas Crash and Analysis Reporting System (KCARS) database from 2011 to 2015. A multinomial logit modeling was used to identify the odds that a driver with a certain type of distraction would be involved in one of the three most common crash types: rear-end, angular, and single-vehicle crashes. Furthermore, ordered logistic modeling was used to analyze the crash data to identify the odds of more severe injuries for teenage and young-adult distracted drivers and their passengers involved in crashes. Survey data was used to develop a structural equation model (SEM) to define the relationship among young drivers’ characteristics (e.g., participants’ socioeconomic and demographic status), attitudes, and behaviors associated with distracted driving and cell phone use while driving. Preliminary analysis showed that more than 12% of the total young drivers’ crashes were distraction-affected crashes. According to the multinomial logit model results, most distraction types for teenage and young-adult drivers are related to rear-end or angular collisions. However, when distracted by cell phones at night, teenage drivers had a greater probability of being involved in single-vehicle crashes. In addition, when teenage drivers drove with their peers as front-seat passengers and were distracted in/on vehicle or by other electronic devices, they were more likely to be involved in single-vehicle crashes. Young-adult drivers distracted in/on vehicle or by cell phones under different conditions such as while driving old or sport utility vehicles, on curved roads, or at intersections, they were more likely to be involved in single-vehicle or angular crashes. Whereas, when they were inattentive during the weekend, rear-end collisions were the most likely collision type. According to the results of the ordered logistic model, teenage and young-adult drivers were more likely to be severely injured in cell phone-related crashes. More specifically, female teenage drivers had a greater probability of being severely injured than male teenage drivers when they were distracted by a cell phone, inside the vehicle, or were inattentive. Young-adult drivers that were distracted on road construction work zones by a cell phone or inside the vehicle, they and their passengers had a greater likelihood of sustaining a severe injury. The SEM results revealed that teenage drivers are more prone than young-adult drivers to drive while distracted and are less likely to support the Kansas laws that ban cell phone use while driving. Also, the model results showed that young drivers who have been involved in crashes or near-crashes during the previous year are more likely to drive while distracted. These results indicate that distractions create threats to the lives of young Kansas drivers, their passengers, and other road users.

Esta tese é composta por três ensaios que versam sobre a complexa relação entre desenvolvimento e meio ambiente. No primeiro ensaio é feita uma revisão das principais abordagens teóricas que tratam do assunto, com destaque para a Curva Ambiental de Kuznets (CAK) e para os conceitos de sustentabilidade fraca e forte. Ambos apontam o crescimento econômico como um importante indutor de impacto ambiental. No entanto, a CAK afirma que a geração de riqueza pode ser aliada na proteção ambiental quando o país alcança um patamar de desenvolvimento suficientemente elevado para que o meio ambiente passe a ser incorporado como um bem superior. Já a conceituação de sustentabilidade não reforça esta conclusão ao passo que sugere que o capital natural precisa ser preservado: seja relativamente (conceito fraco) ou absolutamente (conceito forte). Para estes, a relação entre crescimento e impacto é sempre direta. As abordagens multidimensionais vão além e afirmam que a relação entre o homem e os ecossistemas se dá uma forma impossível de ser mensurada através da simples relação entre crescimento econômico e degradação. Para se aproximar minimamente deste modo de pensar, um conjunto maior de indutores precisa ser considerado. O ensaio 2 desta tese apresenta modelos na literatura que buscam estimar impactos ambientais causados pela atividade humana através de um número maior de indutores (produtivos, populacionais, institucionais, entre outros). Estes modelos (IPAT e STIRPAT) têm como característica a possibilidade de estimar o impacto ambiental esperado através da verificação das variáveis consideradas indutoras. Mas, duas fragilidades são especialmente importantes: i) a variável que mede impacto ambiental é, em geral, reportada com poluição apenas; e, ii) os modelos não fornecem cenários não lineares nos indutores, ou seja, o patamar das variáveis não faz diferença no tamanho do impacto gerado. Então, é proposta a aplicação de um modelo logístico ordenado que seja capaz de estimar as probabilidades de diferentes níveis de impacto ambiental, que é medido através do tamanho da Pegada Ecológica dos países. A metodologia proposta, além de contar com uma medida mais ampla e completa de impacto ambiental, também é capaz de fornecer cenários com maior riqueza informativa, pois levam em consideração os níveis das variáveis indutoras. No trabalho, foram avaliados cenários a partir de um modelo que contou com as seguintes variáveis: percentual de população urbana, percentual de população entre 15-64 anos, densidade demográfica, PIB per capita e percentual de área agrícola. Os sinais e magnitudes das variáveis utilizadas são condizentes com o que a literatura vem apresentando. O diferencial da proposta é, de fato, a possiblidade de análise de cenários através de outra perspectiva. O ensaio 3 da tese aplica este modelo estimado para o globo aos municípios gaúchos. A partir da suposição de que os impactos ambientais são homogêneos ao redor do mundo, foi possível estimar as probabilidades de impacto para cada município. Com estes resultados, foi elaborada uma análise regional dos potenciais impactos ambientais dos municípios a partir dos indutores que apresentavam. Trata-se de um exercício interessante quando se considerada a carência de informações ambientais neste nível de desagregação. / This thesis presents three essays on the complex relationship between development and the environment. On the first essay is performed a survey of the main theoretical approaches on this matter with an emphasis on the Environmental Kuznets Curve (EKC) and the weak and strong sustainability concepts. Both approaches points out the economic growth as an important driver of environmental impact. However, the EKC sustains that the pursuit of economic wealth can be responsible for environmental protection as the country reaches a certain level of income which turns the environment into a superior good. On the other hand, the concepts of sustainability do not enforce this conclusion either in relative terms (weak concept) or absolute terms (strong concept); in this scope, the relationship between growth and environment is constantly direct. The multidimensional approaches propose that the relationship between humankind and the ecosystems are established in a way which is impossible to measure through economic growth and environmental degradation. To become closer to this sort of approach it is necessary to consider a wider set of drivers. The essay 2 of this thesis analyses some statistical models that aim to estimate environmental impacts caused by human activity based on a larger number of drivers (production, population, institutions, etc.). These models (IPAT and STIRPAT) present as a major characteristic the ability to estimate the expected environmental impact grounded on the drivers considered. Nevertheless, two important caveats should be considered: i) the impact variable is most of the time just a pollution measure; and, ii) the models do not offer a nonlinear analysis on the drivers, i.e., the level of the drivers does not make any difference in terms of impact. So, we propose an ordered logistic model that is suitable for estimating the probabilities of different levels of environmental impact which is represented by the size of the Environmental Footprint. The proposed methodology, besides counting on a wider and more complete measure of environmental impact, is capable to offer scenarios with a richer degree of information because the level are effectively taken into consideration. The study has evaluated scenarios base on the following drivers: percentage of population in urban areas, percentage of population between 15-64 years old, demographic density, GDP per capita, and the percentage of agricultural land. The signs and magnitudes of the drivers have proven to be the same as those found in the literature. The main difference of this proposal is, in fact, the possibility of analyzing scenarios through a different (wider) perspective. The essay 3 applies the model, which was prior estimated for the globe, to the municipalities of the state of Rio Grande do Sul, Brazil. Based on the assumption that the environmental impact generated by those drivers is homogeneous around the world, it was possible to estimate the probabilities of environmental impact for each municipality and to draw a regional study of these results. It is an interesting exercise if you take into account that there exists a huge lack of information of environmental conditions for municipalities.

Esta tese é composta por três ensaios que versam sobre a complexa relação entre desenvolvimento e meio ambiente. No primeiro ensaio é feita uma revisão das principais abordagens teóricas que tratam do assunto, com destaque para a Curva Ambiental de Kuznets (CAK) e para os conceitos de sustentabilidade fraca e forte. Ambos apontam o crescimento econômico como um importante indutor de impacto ambiental. No entanto, a CAK afirma que a geração de riqueza pode ser aliada na proteção ambiental quando o país alcança um patamar de desenvolvimento suficientemente elevado para que o meio ambiente passe a ser incorporado como um bem superior. Já a conceituação de sustentabilidade não reforça esta conclusão ao passo que sugere que o capital natural precisa ser preservado: seja relativamente (conceito fraco) ou absolutamente (conceito forte). Para estes, a relação entre crescimento e impacto é sempre direta. As abordagens multidimensionais vão além e afirmam que a relação entre o homem e os ecossistemas se dá uma forma impossível de ser mensurada através da simples relação entre crescimento econômico e degradação. Para se aproximar minimamente deste modo de pensar, um conjunto maior de indutores precisa ser considerado. O ensaio 2 desta tese apresenta modelos na literatura que buscam estimar impactos ambientais causados pela atividade humana através de um número maior de indutores (produtivos, populacionais, institucionais, entre outros). Estes modelos (IPAT e STIRPAT) têm como característica a possibilidade de estimar o impacto ambiental esperado através da verificação das variáveis consideradas indutoras. Mas, duas fragilidades são especialmente importantes: i) a variável que mede impacto ambiental é, em geral, reportada com poluição apenas; e, ii) os modelos não fornecem cenários não lineares nos indutores, ou seja, o patamar das variáveis não faz diferença no tamanho do impacto gerado. Então, é proposta a aplicação de um modelo logístico ordenado que seja capaz de estimar as probabilidades de diferentes níveis de impacto ambiental, que é medido através do tamanho da Pegada Ecológica dos países. A metodologia proposta, além de contar com uma medida mais ampla e completa de impacto ambiental, também é capaz de fornecer cenários com maior riqueza informativa, pois levam em consideração os níveis das variáveis indutoras. No trabalho, foram avaliados cenários a partir de um modelo que contou com as seguintes variáveis: percentual de população urbana, percentual de população entre 15-64 anos, densidade demográfica, PIB per capita e percentual de área agrícola. Os sinais e magnitudes das variáveis utilizadas são condizentes com o que a literatura vem apresentando. O diferencial da proposta é, de fato, a possiblidade de análise de cenários através de outra perspectiva. O ensaio 3 da tese aplica este modelo estimado para o globo aos municípios gaúchos. A partir da suposição de que os impactos ambientais são homogêneos ao redor do mundo, foi possível estimar as probabilidades de impacto para cada município. Com estes resultados, foi elaborada uma análise regional dos potenciais impactos ambientais dos municípios a partir dos indutores que apresentavam. Trata-se de um exercício interessante quando se considerada a carência de informações ambientais neste nível de desagregação. / This thesis presents three essays on the complex relationship between development and the environment. On the first essay is performed a survey of the main theoretical approaches on this matter with an emphasis on the Environmental Kuznets Curve (EKC) and the weak and strong sustainability concepts. Both approaches points out the economic growth as an important driver of environmental impact. However, the EKC sustains that the pursuit of economic wealth can be responsible for environmental protection as the country reaches a certain level of income which turns the environment into a superior good. On the other hand, the concepts of sustainability do not enforce this conclusion either in relative terms (weak concept) or absolute terms (strong concept); in this scope, the relationship between growth and environment is constantly direct. The multidimensional approaches propose that the relationship between humankind and the ecosystems are established in a way which is impossible to measure through economic growth and environmental degradation. To become closer to this sort of approach it is necessary to consider a wider set of drivers. The essay 2 of this thesis analyses some statistical models that aim to estimate environmental impacts caused by human activity based on a larger number of drivers (production, population, institutions, etc.). These models (IPAT and STIRPAT) present as a major characteristic the ability to estimate the expected environmental impact grounded on the drivers considered. Nevertheless, two important caveats should be considered: i) the impact variable is most of the time just a pollution measure; and, ii) the models do not offer a nonlinear analysis on the drivers, i.e., the level of the drivers does not make any difference in terms of impact. So, we propose an ordered logistic model that is suitable for estimating the probabilities of different levels of environmental impact which is represented by the size of the Environmental Footprint. The proposed methodology, besides counting on a wider and more complete measure of environmental impact, is capable to offer scenarios with a richer degree of information because the level are effectively taken into consideration. The study has evaluated scenarios base on the following drivers: percentage of population in urban areas, percentage of population between 15-64 years old, demographic density, GDP per capita, and the percentage of agricultural land. The signs and magnitudes of the drivers have proven to be the same as those found in the literature. The main difference of this proposal is, in fact, the possibility of analyzing scenarios through a different (wider) perspective. The essay 3 applies the model, which was prior estimated for the globe, to the municipalities of the state of Rio Grande do Sul, Brazil. Based on the assumption that the environmental impact generated by those drivers is homogeneous around the world, it was possible to estimate the probabilities of environmental impact for each municipality and to draw a regional study of these results. It is an interesting exercise if you take into account that there exists a huge lack of information of environmental conditions for municipalities.

The Family Forest Research Center recently conducted a mail survey of about 1,400 Massachusetts landowners. Respondents were given questions about themselves and their land and were then asked to rate three carbon sequestration programs in terms of their likelihood to participate. An ordered logit model is used to estimate probabilities that landowners would participate in various improved forest management programs. There are several estimation issues to consider with the ordered logit model. The relative merits of alternative models, including the multinomial and binomial logit, rank-ordered logit, binary logit and mixed ordered logit are discussed. Results of the ordered logit indicate that older males with less education and who own less than 100 acres are less likely to participate in an improved forest management program. All landowners are less likely to participate in a program that requires a management plan and that has a lengthy time commitment, low revenue stream and early withdrawal penalty. Policy implications and direction for future research are discussed.

This project deals with exploring 1) travel-related health risk perception, and 2) actions taken to mitigate that health risk. Ordered logistic regression models were used to identify factors associated with the perceived risk of contracting influenza at work, school, daycare, stores, restaurants, libraries, hospitals, doctor’s offices, public transportation, and family or friends’ homes. Based on the models, factors influencing risk perception of contracting influenza in public places for discretionary activities (stores, restaurants, and libraries) are consistent but differ from models of discretionary social visits to someone’s home. Mandatory activities (work, school, daycare) seem to have a few unique factors (e.g., age, gender, work exposure), as do different types of health-related visits (hospitals, doctors’ offices). Across all of the models, recent experience with the virus, of either an individual or a household member, was the most consistent set of factors increasing risk perception. Using such factors in examining transportation implications will require tracking virus outbreaks for use in conjunction with other factors. Subsequently, social-health risk mitigation strategies were studied with the objective of understanding how risk perception influences an individual’s protective behavior. For this objective, this study analyzes travel-actions associated with two scenarios during an outbreak of influenza: 1) A sick person avoiding spreading the disease and 2) A healthy person avoiding getting in contact with the disease. Ordered logistic regression models were used to identify factors associated with mitigation behavior in the first scenario: visiting a doctor’s office, avoiding public places, avoiding public transit, staying at home; and in the second scenario: avoiding public places, avoiding public transit, staying at home. Based on the models for Scenario 1, the factors affecting the decision of avoiding public places, avoiding public transit, and staying at home were fairly consistent but differ for visiting a doctor’s office. However, Scenario 2 models were consistent with their counterpart mitigation models in Scenario 1 except for two factors: gender and household characteristics. Across all the models from Scenario 1, gender was the most significant factor, and for Scenario 2, the most significant factor was the ratio of household income to the household size. / Master of Science

The prospects for the next Generation - whether young people, regardless of their backgrounds, have equal chances of social success - pose a momentous problem for modern societies. Inequality of opportunity, often reflected by social immobility, is a threat to the egalitarian promise and the stability of your society. This work argues that social capital transmission plays an important role for the chances of social success in Western societies. For the example of Germany, it is reasoned that weak social capital environments deepen existing disadvantages. Even though all levels of education are easily accessible and affordable, Germany has one of the lowest levels of educational mobility among the industrialized countries of the world. Problems appear to be systematic, since the decision regarding entry into higher secondary education is made at early age and is left, in most cases, with the parents, who rely on their own educational trajectory. Outside of the school environment, differences in social capital inheritance explain educational immobility. With the use of the German Socio-Economic Panel survey from 1984 to 2014, various analyses about the relation between social capital and educational success are performed. Social capital, which is helpful for educational and social success, clearly depends on the educational family background. This indirect link has been disregarded in past contributions. Alternative forms of schooling, such as comprehensive and all-day education, as well as a delay of the decision regarding entry into higher education, could help improving unequal social capital inheritance.

Readjusting Historical Credit Ratings using Ordered Logistic Re-gression and Principal Component Analysis The introduction of the Basel II Accord as a regulatory document for creditrisk presented new concepts of credit risk management and credit risk mea-surements, such as enabling international banks to use internal estimates ofprobability of default (PD), exposure at default (EAD) and loss given default(LGD). These three measurements is the foundation of the regulatory capitalcalculations and are all in turn based on the bank’s internal credit ratings. Ithas hence been of increasing importance to build sound credit rating modelsthat possess the capability to provide accurate measurements of the credit riskof borrowers. These statistical models are usually based on empirical data andthe goodness-of-fit of the model is mainly depending on the quality and sta-tistical significance of the data. Therefore, one of the most important aspectsof credit rating modeling is to have a sufficient number of observations to bestatistically reliable, making the success of a rating model heavily dependenton the data collection and development state.The main purpose of this project is to, in a simple but efficient way, createa longer time series of homogeneous data by readjusting the historical creditrating data of one of Svenska Handelsbanken AB’s credit portfolios. Thisreadjustment is done by developing ordered logistic regression models thatare using independent variables consisting of macro economic data in separateways. One model uses macro economic variables compiled into principal com-ponents, generated through a Principal Component Analysis while all othermodels uses the same macro economic variables separately in different com-binations. The models will be tested to evaluate their ability to readjust theportfolio as well as their predictive capabilities. / Justering av historiska kreditbetyg med hjälp av ordinal logistiskregression och principialkomponentsanalys När Basel II implementerades introducerades även nya riktlinjer för finan-siella instituts riskhantering och beräkning av kreditrisk, så som möjlighetenför banker att använda interna beräkningar av Probability of Default (PD),Exposure at Default (EAD) och Loss Given Default (LGD), som tillsammansgrundar sig i varje låntagares sannoliket för fallissemang. Dessa tre mått ut-gör grunden för beräkningen av de kapitaltäckningskrav som banker förväntasuppfylla och baseras i sin tur på bankernas interna kreditratingsystem. Detär därmed av stor vikt för banker att bygga stabila kreditratingmodeller medkapacitet att generera pålitliga beräkningar av motparternas kreditrisk. Dessamodeller är vanligtvis baserade på empirisk data och modellens goodness-of-fit,eller passning till datat, beror till stor del på kvalitén och den statistiska sig-nifikansen hos det data som står till förfogande. Därför är en av de viktigasteaspekterna för kreditratingsmodeller att ha tillräckligt många observationeratt träna modellen på, vilket gör modellens utvecklingsskede samt mängdendata avgörande för modellens framgång.Huvudsyftet med detta projekt är att, på ett enkelt och effektivt sätt, skapaen längre, homogen tidsserie genom att justera historisk kreditratingdata i enportfölj med företagslån tillhandahållen av Svenska Handelsbanken AB. Jus-teringen görs genom att utveckla olika ordinala logistiska regressionsmodellermed beroende variabler bestående av makroekonomiska variabler, på olikasätt. En av modellerna använder makroekonomiska variabler i form av princi-palkomponenter skapade med hjälp av en principialkomponentsanalys, medande andra modelelrna använder de makroekonomiska variablerna enskilt i olikakombinationer. Modellerna testas för att utvärdera både deras förmåga attjustera portföljens historiska kreditratings samt för att göra prediktioner.

信用評等制度在美國已有百年以上歷史，而我國自民國80幾年開始發展評等制度，截至目前，僅有中華信用評等公司與台灣經濟新報社兩家公司提供評等服務，而台灣經濟新報社更將金融保險業排除於評等對象之外。站在穩定市場競爭、保障消費者權益、配合監理需求，以及輔助專案投標等方面來看，市場上的確需要一套能反映產險業行業特性之評等模式。 本文以美國接受A.M.Best評等之產險公司為研究對象，運用三種統計方法：多元區別分析(Multiple Discriminant Analysis，MDA)、羅吉斯迴歸(Unordered Logistic Regression，ULR)、順序性羅吉斯迴歸(Ordered Logistic Regression，OLR)，來建構產險公司之信用評等模式。樣本選擇方面：估計樣本，選取美國1993年到1996年接受A.M.Best評等之產險公司327家；保留樣本，為1997年78筆資料。 而本文預定達成目標如下： 一、建立等級預測模型：參考Ederington(1985)所作債券等級預測模型，以獲利能力、槓桿、流動性、投資風險、準備金適足性五類指標共38個財務比率，透過三種統計模型，建構等級預測模型。 二、藉由等級預測之建立，尋找能有效區別產險公司評等等級之財務指標，並分析其影響程度。 三、力求模型公信力：無論變數選擇或權數決定，皆由統計軟體按照樣本特性選取產生，減少人為主觀判斷。 在決定研究對象之初，因考慮到國內產險公司接受評等之家數不多，且年數又太短，資料數量無法據以建立評等模式，因而決定以美國的產險公司為對象，再以台灣樣本作為保留樣本，預測之等級結果僅供參考之用。 / Three possible models of the P-L Insurers rating process are estimated and compared：1. Muitiple Discriminant Model, 2. Unordered Logistic Model, 3. Ordered Logistic Model. Each model is estimated for a sample of 327 American P-L insurance companies using the same 38 independent variables. The three estimated models are then employed to predict ratings for a holdout sample of 78 companies. The study analyzes 1993 through 1997 data for a sample of P-L insurers that acquired A.M.Best Financial strength ratings between December 31,1993, and December 31, 1997. Empirical evidence suggests that even when models with the same basic structure were compared, differences in estimation procedures resulted in quite different coefficient estimates and classifications. The muitiple discriminant model clearly outperformed the regression model, while the unordered logistic model was clearly superior to the ordered logistic model.