Submitted by Cristiana Freitas (criscgf@hotmail.com) on 2018-05-24T19:38:01Z No. of bitstreams: 1 Dissertação EPGE - Cristiana Freitas.pdf: 1002118 bytes, checksum: 04f0edb6067713b5b5b09ee6fa22d053 (MD5) / Approved for entry into archive by GILSON ROCHA MIRANDA (gilson.miranda@fgv.br) on 2018-05-28T16:29:15Z (GMT) No. of bitstreams: 1 Dissertação EPGE - Cristiana Freitas.pdf: 1002118 bytes, checksum: 04f0edb6067713b5b5b09ee6fa22d053 (MD5) / Made available in DSpace on 2018-06-14T20:15:06Z (GMT). No. of bitstreams: 1 Dissertação EPGE - Cristiana Freitas.pdf: 1002118 bytes, checksum: 04f0edb6067713b5b5b09ee6fa22d053 (MD5) Previous issue date: 2018-04-27 / This thesis aims to estimate the demand for auto insurance in the city of Rio de Janeiro. Due to the availability of consumer level data, a brief review of discrete choice models was made, focusing on Logit with random coefficients. In order to effectively implement the models, we had to deal with a peculiarity of the insurance industry: prices vary according to each individual. As we only have the effective prices of the contracted policy, a random forest model was tested to generate expected premium values for the insurers. / Esta dissertação tem como objetivo estimar a demanda por seguro de automóvel na cidade do Rio de Janeiro. Em razão da disponibilidade de dados a nível do consumidor, foi feita uma breve revisão de modelos de escolha discreta, com destaque para o Logit com coeficientes aleatórios. Para ser possível a efetiva implementação dos modelos, foi preciso lidar com uma peculiaridade do setor de seguros: os preços variam de acordo com o indivíduo. Como só temos os preços efetivos da apólice contratada, foi testado um modelo random forest para gerar valores esperados de prêmio das seguradoras.

The diploma thesis is focused on predictive modeling in credit risk management. Banks and financial institutions are mainly interested in it to estimate the probability of client's default in order to make a decision about which client will be accepted and which client will be rejected. The theoretical part includes an introduction of credit scoring and a description of discrete choice models. The linear probability model, the probit model and the logit model are described in detail. The logit model is afterwards used for the prediction of client's default. The practical part is focused on a statistical description of the dataset and a description of how to work with it before we start with the development of the credit scoring model. After that follows the estimation of the model on testing sample, its testing and the estimation of the model on full sample with a description of individual steps of calculation and outputs of the program SPSS.

Das Ziel der Masterarbeit ist es zu klären, ob private Haushalte variable Strompreise akzeptieren und die Stärke der charakteristischen Merkmale zu ermitteln. Der dazu benötigte Datensatz wurde mit Hilfe einer Stated Choice Befragung im Internet ermittelt. Dafür wurden drei Gerätekategorien und ein Dreitarifmodell formuliert. Für die Auswertung wurde ein additiv verknüpftes Teilwert-Nutzenmodell und ein binäres Logit-Choice-Modell verwendet. Signifikante Variablen wurden über das Top-down-Verfahren bestimmt. Das Modell und die Schätzer der Preise wurden im Anschluss durch Hypothesentest getestet. Als signifikante Schätzer, mit stark negativen Werten, stellten sich in allen Kategorien die Preise heraus. Die Hypothesentests ergaben, dass es sinnvoll ist, die Schätzer der Preise für die ersten beiden Gerätetypen zusammenzufassen. Es stellte sich eine unterproportionale Preiselastizität der Nachfrage heraus und eine hohe Akzeptanz ein Teil der Geräte in den Nachtstunden zu nutzen. Insgesamt sprachen sich mehr Teilnehmer für ein Smart-Meter-Gateway aus als dagegen. / The aim of the master thesis is to clarify whether private households accept variable electricity prices and determine the strength of the characteristic attributes. The data set was determined with a Stated Choice survey on the Internet. For this purpose, three device categories and a three-tariff model were formulated. An additive-linked partial-value model and a binary logit choice model were used for the evaluation. Significant variables were determined by using the top-down method. Afterwards the model and the estimators of the prices were tested with statistical hypothesis testings. The price emerged as a significant estimator, with strongly negative values in all categories. The hypothesis tests showed that it is useful to conclude the estimators of the prices for the first two device types. The results showed an inelastic price elasticity of demand and a high acceptance of the subjects to use a part of the devices in the night hours. Overall, more participants decide to use a smart meter gateway than to refuse it.

An evidence-based policy debate about future fuel demand requires reliable estimates for fuel price elasticities. Such predictions are often based on revealed preference (RP) data. However, this procedure will only yield reliable results in the absence of severe structural discontinuities. In order to overcome this potential limitation we used a situational stated preference (SP) survey to estimate the response to hypothetical fuel price changes beyond the scope of previous observations. We elicit fuel price elasticities for price increases up to four Euros per liter and find that the situational approach predicts the actual responses to previously observed fuel price changes very well. We conclude that applying a situational approach is particularly useful, if behavioral predictions for unprecedented (non-monetary) policy interventions or supply side shocks are of interest that go beyond the reach of standard RP approaches.

Esta tesis es una colección de tres ensayos en el campo de la economía industrial. Los dos primeros ensayos son contribuciones empíricas en el tema de la discriminación de precios. El último ensayo está relacionado con la heterogeneidad en las expectativas futuras de precios. Ese ensayo es una contribución a la literatura que trata el tema de demanda dinámica. El objetivo del primer capítulo es determinar la relevancia de la discriminación de precios en los mercados de alimentos orgánicos (cereales para desayuno). El principal problema al responder a esta pregunta es que los costes no son observables y, por lo tanto, a priori, no puedo saber si el sobreprecio de los alimentos orgánicos se debe a la discriminación precios o simplemente por diferencias de costes. Para evitar este problema, utilizo un modelo de elección discreta para obtener la elasticidad de la demanda y, con un modelo de oferta, obtener el valor de los costes marginales. Lo resultados indican que, aproximadamente, el 6% de la diferencia de precio entre productos orgánicos y productos no orgánicos se debe a la discriminación de precios. Además, yo encuentro que un impuesto en los productos no orgánicos no es suficiente para disminuir la discriminación de precios, y tiene un efecto negativo en el bienestar social: hay reducción en el excedente del consumidor que es superior al aumento de los beneficios de las empresas. Por fin, encuentro que la discriminación de precios resulta de la existencia de consumidores con renta alta en el mercado. En el segundo capítulo yo estudio si hay una relación no monótona entre los descuentos por cantidad y el nivel de competencia en el mercado. Los resultados encontrados sugieren que la relación es sí no-monótona y que tiene formato en U. Eso implica que los descuentos por cantidad son más comunes en los mercados menos concentrados y en los más concentrados. Además, los resultados sugieren que la firma líder es la responsable por esa situación. Por fin, el tercer capítulo analiza la heterogeneidad en la expectativa de precios en el mercado de un bien no duradero (refrescos) y que son comprados con alta frecuencia. Los resultados sugieren que, en media, los consumidores son racionales y forman expectativas con respecto al futuro. Sin embargo, consumidores de baja renta forman expectativas basadas en un modelo de Markov de primer orden (donde solo importa el precio actual y del período anterior para formar sus expectativas), y a los consumidores con renta mediana solo les importa el precio actual al decidir comprar (o sea, son impacientes - miopes).

Hoy en día, las Pequeñas y Medianas Empresas (PYMES) son consideradas como una fuente de crecimiento, empleo e inclusión social; sin embargo, estas sufren de limitaciones recurrentes respecto al acceso al crédito sobre todo si son lideradas por mujeres. Por ello, la presente investigación tiene como objetivo determinar la existencia de discriminación de género en el acceso al crédito en PYMES en el Perú en el año 2017 mediante la Encuesta Nacional de Empresas 2018 (ENE) para lo cual se utilizara un modelo econométrico Probit y se analizaran variables como edad, educación del propietario, sector económico, etc., lo cual demuestra que la probabilidad de que una PYME logre acceder al crédito disminuye cuando este presenta como propietario una mujer y afirmando la hipótesis sobre si en realidad la estructura de la empresa varía dependiendo del sexo del propietario de la PYME teniendo énfasis en la educación del propietario y sector económico al cual la PYME pertenece. / Nowadays, Small and Medium Enterprises (SMEs) are considered a source of growth, employment, and social inclusion; however, they suffer from recurrent limitations in access to credit, especially if they are led by women. For that, reason, the aim of this research is to determine the existence of gender discrimination in access to credit in SMEs in Peru in 2017 through the National Survey of Companies 2018 (ENE). To do so, a Probit econometric model will be used and variables such as age, owner's education, economic sector, etc. will be analyzed, which shows that the probability of an SME gaining access to credit decreases when the owner is a woman and affirms the hypothesis that the structure of the business varies depending on the sex of the SME owner, with emphasis on the education of the owner and the economic sector to which the SME belongs. / Trabajo de investigación

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18255号 / 工博第3847号 / 新制||工||1590(附属図書館) / 31113 / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 小林 潔司, 准教授 宇野 伸宏, 准教授 松島 格也 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

The research utilizes a Discrete Choice Experiment to assess consumer preferences and determine blockchain-based food traceability's relative importance and utility in the food purchasing process. The study specifically tests five selected attributes of olive oil, including price, organic label, olive oil type, country of origin, and blockchain traceability. Additionally, a survey is conducted to evaluate Swedish consumers' knowledge levels concerning blockchain technology and its application in the food system to improve transparency. The study also explores consumer evaluations of safety and sustainability information and their willingness to pay price premiums for food products verified for safety and environmental claims. The findings indicate that while Swedish consumers possess moderate knowledge of blockchain technology, their understanding of its specific application in food traceability is limited. Nevertheless, consumers highly value access to accurate information about the safety and sustainability of their food purchases. A considerable portion of respondents express a willingness to pay a premium for products verified for safety, authenticity, and sustainability. The choice experiment results reveal that blockchain-based food traceability ranks lower in relative importance than the price and organic label but higher than olive oil type and country of origin. The findings of this study contribute to bridging the existing gap in the understanding of blockchain-based food traceability technology among Swedish consumers.

Natural disasters are increasing in frequency and intensity and have devastating impacts on individuals, both humanitarian and economic, particularly in developing countries. Microinsurance is seen as one promising instrument of disaster risk management, however the level of demand for respective projects remains low. Using behavioural games and a discrete choice experiment, this paper analyses the demand for hypothetical microinsurance products in rural Cambodia and contributes significant household level evidence to the current research. A general preference for microinsurance can be found, with demand significantly affected by price, provider, requirements for prevention and combinations with credit. Furthermore, financial literacy, risk aversion, levels of trust and previous disaster experience impact the individual demand for flood insurance in rural Cambodia.

<p> </p> <p>This dissertation consists of three essays that examine the response of small businesses to disruptions in their environment. The first two essays focused on small non-farm businesses in the United States and how they deal with natural disasters. The last essay examined smallholder farm households in Malawi, and how their household labor allocation decisions are affected by land allocation to estates in their communities. The individual essays are summarized as follows:</p> <p>Essay 1: <em>A Conditional Process Approach to Understanding the Role of Adjustment Strategies and Disaster Experience in Racial Disparities in Small Business Performance. </em>Considering that most minority owned businesses have limited access to formal systems, this essay explored how race could indirectly affect business performance (measured as percentage revenue growth) through the adoption of three informal strategies: customer base expansion, supplier base expansion and family adjustment strategies. It also explored whether these indirect effects are moderated by experience with natural disaster. The results showed that being a racial minority was positively associated with revenue growth such that on average, minority business owners experienced 29% higher revenue growth than white-owned businesses (p<0.05) on business performance. It also showed a modest indirect effect of race on revenue growth through each mediating strategy (p<0.5). However, the results did not support a moderating role for disaster experience. </p> <p>Essay 2: <em>Willingness to Pay for Comprehensive Cyclone Insurance Coverage by Small Business Owners: Evidence from the Coastal States of the United States</em>. Small businesses in the coastal United States are usually uninsured or underinsured for cyclone events. The underinsuring of these businesses could be a result of limited insurance coverage as well as individual characteristics of small business owners. Using a discrete choice experiment, this essay used a hypothetical comprehensive cyclone insurance to understand what insurance attributes are important to small business owners. It also examined the role of previous disaster experience, charity hazard as well as temporal orientation on the willingness to pay for the disaster insurance. This study used a discrete choice experiment to elicit insurance preferences from small coastal businesses which employed less than 100 employees. A mixed logit model was used to analyze the data. The results showed that business owners exhibited positive marginal utilities from policies that covered flood, windstorm, and business interruption regardless of the combination. Notably, the mixed logit model showed that on average, business owners were willing to pay up to 450%, 472%, and 482% more than their total monthly business insurance premium payment for insurance that covers flood and business interruption, windstorm and business interruption, and flood, windstorm and business interruption respectively. Ordinary Least Squares (OLS) regression showed that respondents who had previously experienced cyclones were more willing to pay for the presented insurance policies than those who have not. Future orientation was also found to be positively associated with the marginal willingness to pay for the insurance policies.</p> <p>Essay 3: <em>Estates and Small-Holder Agricultural Labor Dynamics in Sub-Saharan Africa:</em> <em>A Case Study of Malawi. </em>Estates are larger than the average farm holdings, which mostly grow one crop, require large capital investment, are centrally managed and rely a lot on hired labor. With such large investments in agricultural land, the labor decisions of smallholder households in Africa will likely be altered. This essay therefore examined the role of estate farms on smallholders’ allocation of labor between on-farm, and off-farm demand and supply of casual labor using the <em>ganyu</em> system of Malawi as a case-study. Using the Malawi Integrated Household Panel survey covering the years 2010, 2013, 2016, and 2019, we estimated the effect of estates on the participation of smallholders on these labor decisions as well as the number of days spent in each activity. We also investigated the effect of these estates on community agricultural labor (<em>ganyu) </em>wage rates and the share of income accruable to ganyu and crop production. Linear probability (LPM), as well as tobit-correlated random effects (CRE) regressions were used to test these effects. Both models showed that the share of estates had a negative correlation with <em>ganyu </em>demand. The Tobit CRE regression showed that on average a 1% increase in the percentage share of agricultural land occupied by estates was associated with a modest 0.04% (p<0.01) decrease in the number of days <em>ganyu </em>labor was demanded, and a 0.02% increase in the number of days household members spent on their own farms. Further results showed that households in communities with higher shares of estates participated in less non-crop farming activities especially wage employment. We also found that the negative relationship between estates and <em>ganyu </em>demand was accentuated among households with higher levels of assets, and farm income. Finally, we found a modest negative relationship between share of estates and community <em>ganyu </em>wage rates </p>