Cette thèse s'intéresse aux fondements du comportement des usagers face à la congestion dans les transports publics. Elle se compose de trois essais distincts. Les deux premiers essais examinent l'adaptation des usagers au manque de ponctualité et à l'inconfort. Le troisième essai offre une analyse empirique de l'effet « foule ». Dans le premier chapitre, je présente un modèle de compétition bimodale entre une ligne de transport public sujette aux retards, et un mode alternatif. A l'équilibre du marché, le niveau de fiabilité choisi par la firme en charge des transports publics augmente avec le prix du mode alternatif via un effet « demande » évoquant l'effet Mohring. L'étude de la qualité de service optimale montre que souvent, la fiabilité des transports publics et donc leur fréquentation à l'équilibre sont inférieurs à ce qu'ils seraient à l'optimum social. Dans le second chapitre, afin d'étudier le coût de l'inconfort et ses implications quant au choix de l'heure de déplacement, je développe un modèle structurel dans lequel les usagers des transports public choisissent entre voyager dans un véhicule bondé et arriver à destination à l'heure désirée ou alors voyager plus tard ou plus tôt pour éviter la foule. J'établis la distribution des usagers parmi les véhicules à l'équilibre et à l'optimum social, et montre que l'optimum social peut être décentralisé par une tarification fine par véhicule. Les propriétés du modèle sont comparées avec celles du modèle du « bottleneck » et des modèles de congestion routière. Dans le troisième chapitre, j'analyse l'influence de la densité d'usagers sur la satisfaction liée au confort durant un déplacement en transport public. De plus, je décris l'anatomie de l'effet « foule » en testant différents aspects (odeur, bruit, position debout...) comme des causes d'inconfort lorsque la densité d'usagers augmente. J'identifie un net effet « foule » : en moyenne, un usager supplémentaire par mètre carré diminue de 1 la satisfaction liée au confort qui est mesurée sur une échelle de 0 à 10. Je ne trouve pas de corrélation entre temps de transport et l'effet « foule ». Cependant, cet effet augmente avec le revenu des usagers. Trois causes de désutilité liée à la foule sont identifiées : une plus grande probabilité de devoir voyager debout, un usage limité du temps, et des conditions de déplacement plus bruyantes. Ces désagréments sont plus importants chez les femmes et les usagers les plus aisés. / This dissertation addresses the foundations of user’s behavior with respect to the congestion in public transport. It is made of three distinct essays. The two first essays investigate how users get used to lack of punctuality and crowding. The third essay presents an empirical analysis of the crowding effect. In the first chapter, I consider the modeling of a bi-modal competitive network involving a public transport mode, which may be unreliable, and an alternative mode. The public transport reliability set by the public transport firm at the competitive equilibrium increases with the alternative mode fare, via a demand effect. This is reminiscent of the Mohring Effect. The study of the optimal service quality shows that often, public transport reliability and thereby patronage are lower at equilibrium compared to first-best social optimum. In the second chapter, to study the behavioral implications and costs of crowding, I develop a structural model in which public transport users face a choice between traveling in a crowded train and arriving when they want, and traveling earlier or later to avoid crowding but arriving at an inconvenient time. I derive the user equilibrium and socially optimal distribution of passengers across trains, show how the optimum can be decentralized using train-specific fares, and characterize the welfare gains from optimal pricing. Properties of the model are compared with those obtained from the bottleneck and flow congestion models of road traffic. In the third chapter, I investigate the influence of in-vehicle crowding on the comfort satisfaction experienced during a public transport journey. Moreover, I describe the anatomy of the crowding effect by testing various nuisance factors (Smell, Noise, Standing...) as channels through which crowding may decrease the comfort satisfaction. I find a clear crowding effect: on average, an extra-user per square meter decreases by one the expected 0 to 10 scale individual well-being. I do not find any empirical evidence of this effect being intensified by the travel time. However, the crowding effect increases with the income of users. I find three causes of crowding disutility: a higher probability to stand for all or part of the journey, a poorer use of the time during the journey, and noisier travel conditions. These features of discomfort matter more for women and wealthy individuals.

Three groups of participants were exposed to different unit price sequences. Unit prices for all groups ranged from unit price 1 to 21. Analyses of demand curves, response rates, session duration, and elasticity coefficients suggest that the sequence of exposure to unit prices can affect the elasticity of demand. In addition, the size of unit price contrast, direction of unit price change, and proximity to experimental milestones also may affect the consumption of monetary reinforcers.

Chapter 2 compares government consultation via an opinion poll and a citizens' jury". In a jury, about fiffteen volunteers spend several days learning about a policy choice before voting. If the public is ill-informed, the government trades-off "information" against participation". Jurors have better information than poll respondents, but constitute a smaller sample. More- over, participation costs may bias the jury sample. Indeed, the literature suggests that costs might induce "neutrality": over-representation of the minority to the extent that the result is uninformative. I show that although the minority will often be over-represented, "neutrality" is a knife-edge result here so juries may be worthwhile. Extensions consider compensating jurors and excluding "special interests". Chapter 3 uses evidence from the allocation of regeneration funding to motivate a model in which central government may ask councils to compile apparently pointless dossiers to ap- ply for money, because the dossiers provide information about councils' competence. I then consider when the government might prefer a simpler but less flexible auction-type process. The UK government's ability to "ring-fence" money, obliging councils to spend it on its priority, is central to chapter 3. Chapter 4 develops the analysis of auctions in this context. With variations in competence, ring-fencing effectively imposes type-specific minimum bids. I characterise equilibrium bidding and show that the ring-fencing constraint may not only increase bids, but actually induce councils to contribute resources. Continuing the themes of participation and competence in the policy process, chapter 5 examines parish councils' use of a general spending power. I find parishes with more well-educated and older citizens (groups with generally higher political participation) are more likely to use their powers. Further investigation suggests that these citizens matter because they are involved in governance, rather than because they exert democratic pressure. This has impli- cations for wider neighbourhood governance policy.

This thesis explores learning and formation of cognitive and non-cognitive skills within formal and non-formal environments, as well as the impact of migration on fertility behaviour in three separate empirical studies. In two of the papers (Chapters 2 and 4) I utilise a 13 year individual-level panel data-set from rural Tanzania, while the third one (Chapter 3) is based on cross-sectional data that I collected in urban Bombay slums in 2007. I consider skill acquisition and learning in a number of spheres. First, I adopt the conventional notion of school-based learning and examine the role of income shocks in evolution of schooling inequalities, in rural Tanzania. I find evidence of shock-induced permanent changes in the schooling of those affected by the shocks in later childhood (age 7-13), 10-13 years later. Further, I find suggestive evidence that the household short-term labour response may to be one of the mechanisms for these long-term effects. Next, I broaden the definition of learning to include acquisition of non-cognitive skills. Although there is growing recognition of the importance of these, there is no evidence, within a developing country context, on effectiveness of interventions targeting them. The second paper is an evaluation of a long-term non-formal schooling intervention in Bombay slums, which works on raising non-cognitive skills, including self-esteem, a sense of agency, and aspirations of children. It shows that, like cognitive skills, non-cognitive skills can be effectively raised through sustained intervention, offering evidence of substantial positive programme effects. The final paper turns to examining the impact of migration among young women on fertility behaviour. Econometric panel data methods are combined with an instrumenting strategy to offer evidence of a causal positive impact of migration on the age at which women start having children, which is shown to be likely to have permanent effects on total fertility. The findings are most consistent with the presence of temporary post migration disruption effects.

This thesis is comprised of three essays. These essays model facets of illicit markets using economic theories of reputation and go on to develop new theories of reputation. In chapter 1, we start by defining illicit markets. We build a model of trade in an illicit market where honesty is enforced through reputation. We then show how criminals' heterogeneity in their ability to avoid detection can have interesting consequences. In particular, such heterogeneity can explain why illicit market participants might sometimes adopt surprisingly blatant trading strategies. Heterogeneity among criminals in terms of how they are affected by enforcement can give rise to counter-intuitive market reactions to changes in the level of enforcement. Chapter 2 considers the market for stolen payment data such as credit card numbers. The central observation is that stealing financial data and using it for personal gain are two very different skills. An online market exists through which those who steal data can sell it to those who can ``cash in''. The low prices in these markets are not evidence of a failure of the reputational system, but of large outside options on one side of the market. Finally chapter 3 considers reputation more generally. We find folk theorem and incomplete information models place too much faith in the knowledge and rationality of short-lived players. We propose instead a new model with more realistic assumptions about the information available to short-lived players and their ability to use it. This model generates natural cycles in the reputation of the rational long-lived player.

This thesis comprises three principal essays, each of which provides a contribution to the literature on the nonparametric approach to demand analysis. In each essay, I develop novel techniques that follow in the revealed preference tradition, and apply them to tackle a series of questions that concern the mechanisms underlying consumer spending decisions. Each technique developed is tightly linked to a particular nonparametric theory of choice behaviour and is explicitly designed for use with a finite set of observations. My work draws heavily upon results from finite mathematics, into which I integrate insights from information theory and integer programming. The output of this endeavor is a set of methodologies that are largely free of auxiliary assumptions over the form of the unobserved structural functions of interest. Providing greater detail on the work to come, my first essay extends and clarifies the nonparametric approach to forecasting demand behaviour at new budget regimes. Using insights from information theory and integer programming, I construct an operational nonparametric definition of global rationality and develop a methodology that facilitates the recovery of globally rational individual demand predictions. This is the first attempt in the literature to develop a systematic methodology to impose global rationality on nonparametric demand predictions. The resulting forecasts allow for unrestricted preference heterogeneity in the population and I demonstrate how these predictions can be used for coherent welfare analysis. In my second and third essays, I prove new revealed preference testability axioms for models that extend the traditional neoclassical choice framework. Specifically, in my second essay, I address the intertemporal allocation of spending by collectives, whilst my final essay integrates taste variation into the utility maximisation framework. In both of these essays, I develop my testable results into practical algorithms that allow one to recover salient features of individual preferences. In my second essay, a methodology is developed to recover the minimal intrahousehold heterogeneity in theory-consistent discount rates, whilst my final essay develops a quadratic programming procedure that facilitates the recovery of the minimal interpersonal and intertemporal heterogeneity in tastes that is required to rationalise observed choice patterns. Applying these techniques to consumption micro-data yields new empirical insights that are of relevance to the applied literatures on time discounting, family economics and the public policy debate on tobacco control.

Este estudo, dividido em três artigos, tem por objetivo analisar as contribuições da demografia, educação e mercado de trabalho no desenvolvimento brasileiro ao longo do século XX. Os três artigos estão encadeados em uma sequência temporal tal que o primeiro artigo concentra-se na transição demográfica brasileira para as coortes nascidas entre 1890 e 1960. O segundo artigo preocupa-se em entender como os efeitos da transição demográfica são absorvidos pelas novas coortes e seu efeito na taxa de matrícula. O terceiro artigo investiga como as coortes mais jovens, com mais capital humano, mudam a oferta relativa da economia e este processo impacta os salários relativos por grupos de qualificação. No primeiro artigo, os microdados dos Censos demográficos foram utilizados na investigação usando variáveis de educação, localidade e coorte de nascimento. Uma vez controlando por coorte e localidade, assumindo a hipótese que na localidade municipal as coortes são mais homogêneas quanto à exposição da oferta escolar, educação explicou aproximadamente 30% da queda da fecundidade. A simulação contrafactual evidenciou que caso não tivesse ocorrido o aumento da escolaridade das coortes, a transição demográfica seria mais lenta e gradual. No segundo artigo, dados da PNAD foram construídos para investigar a probabilidade de matrícula assumindo que a transição demográfica gera uma variação exógena no tamanho absoluto e no tamanho relativo das coortes. Os resultados evidenciaram que para as séries com maiores avanços na taxa de matrícula, diminuição do tamanho da coorte apresentou-se negativamente relacionada com a probabilidade de matrícula. Porém, a magnitude deste efeito não gerou mudanças significativas na simulação contrafactual e, para a primeira série do ensino fundamental, o avanço da escolaridade dos pais mostrou-se significativo. No terceiro artigo, dados da PNAD são utilizados para se investigar mudanças relativas no mercado de trabalho ocasionadas pela entrada de novas coortes com melhores indicadores educacionais. Os resultados apontam que mudanças na oferta relativa de grupos etários mais jovens conduziram a mudanças em seus salários relativos e as mudanças nos salários relativos de grupos etários mais velhos são explicadas por mudanças na oferta relativa agregada da economia. A simulação contrafactual demostrou que, caso não houvesse mudança na oferta relativa da economia, os salários de qualificados em relação a não qualificados apresentaria tendência crescente entre 1981 e 2013. / This study, divided into three articles, aims to analyze the contributions of demography, education and the labor market in Brazilian development throughout the 20th century. The three articles are connected in a temporal sequence such that the first article focuses on the Brazilian demographic transition for the cohorts born between 1890 and 1960. The second article is concerned with understanding how the effects of the demographic transition are absorbed by the new cohorts and affect school attendance. The third article investigates how the younger cohorts, with more human capital, change the relative supply of the economy and this process impacts relative wages by qualification groups. In the first article, the microdata of demographic Census were used in the investigation using variables of education, municipality and birth cohort. Once the cohort and locality control, assuming the hypothesis that in the municipality the cohorts are more homogeneous regarding the exposure of the school supply, education explained approximately 30% of the decrease of the fertility rate. The counterfactual simulation showed that if there had not been an increase in cohort education, the demographic transition would be slower and gradual. In the second article, PNAD data were constructed to investigate the probability of enrollment assuming that the demographic transition generates an exogenous variation in the absolute size and relative size of the cohorts. The results showed that for the series with the greatest advances in enrollment, a decrease in cohort size was negatively related to the probability of enrollment. However, the magnitude of this effect did not generate significant changes in the counterfactual simulation, and for the first grade of elementary school, the progress of parents\' schooling was significant. In the third article, PNAD data are used to investigate relative changes in the labor market caused by the entry of new cohorts with better educational indicators. The results indicate that changes in the relative supply of younger age groups have led to changes in relative wages and changes in the relative wages of older age groups are explained by changes in the aggregate relative supply of the economy. The counterfactual simulation showed that if there were no change in the relative supply of the economy, relative wages from skilled to unskilled workers would show a growing trend between 1981 and 2013.

Esta tese reformula o modelo de teoria de decisão de Savage relaxando a hipótese implícita de que uma conseqüência é uma descrição perfeita de uma determinada situação. Axiomas comportamentais sobre preferências definidas no espaço de atos são introduzidos e uma representação na forma de Utilidade Esperada é derivada. Em particular, como em Savage, há uma única probabilidade subjetiva sobre os estados da natureza. O ganho de flexibilidade da reformulação apresenta uma solução para o paradoxo de Ellsberg que não faz uso de múltiplas probabilidades subjetivas, e uma reinterpretação da aversão ao risco no modelo de Utilidade Esperada convencional. / This thesis reformulates the Savage\'s Decision Theory model relaxing the implicit hypothesis that a consequence must be a perfect description of a situation. We introduce Behavioral axioms on preferences defined over the set of acts and derive a new Expected Utility functional representation. Like on Savage\'s, there is an unique prior over states of world. The flexibility gain of this representation presents a solution for the Ellsberg paradox that does not rely on multiple priors, and allows for a new interpretation of risk aversion on the Expect Utility model.

This dissertation focuses on how changes in SNAP (Supplemental Nutrition Assistance Program, formerly called food stamp) policies affect the consumption choices for recipient households and how macroeconomic environment affects individuals’ behaviors. In the first chapter, I examine the consumer response to SNAP benefit change. In the second chapter, I examine the investor attention under different stock market valuations. In the third chapter, I examine the impact of rising housing prices on individuals’ marriage entry. In the first chapter, we examine how SNAP benefits were spent. The American Recovery and Reinvestment Act of 2009 (commonly known as the Stimulus Act) temporarily increased the benefit level for SNAP recipients and released qualification requirements. We use a consumer panel with detailed transaction records to analyze the impact of SNAP benefits changes on participant household’s consumption choices. We find that the marginal propensity to spend (MPS) on food out of SNAP benefits increase is 0.45. The MPS out of SNAP benefit decrease is -0.85. SNAP participant households are more sensitive toward the benefits decrease than increase. We also study how SNAP benefits spent are in response to benefit changes. We find that with more SNAP benefits recipient households consume much more tobacco and with less benefit they consume much less alcohol. Results are robust to various placebo tests. In the second chapter, I examine the impact of stock market valuation on investor attention. Investor attention affects stock return variance and risk premia. Using Google Trend data, I come up with a new proxy for measuring investor attention. This chapter investigates investor attention effect on the merger announcement and how market valuations and days of the week affect investor attention. With the Google search index, this study finds that investor attention has positive correlations on market reactions following the merger announcement. In the third chapter, I estimate the impact of increasing on first marriage age in China. The first marriage age plays a very important role in the population economy, especially for a demographic dividend. The marriage market is affected by income, education, wealth inequality, consumption, etc. In China, the first marriage age has fallen since the late 1990s. This chapter focuses on how the price of the apartments change people decisions on marriage and ascribes that part of the delay to first marriage age is due to increase in housing price over the same period. In China, social norms require men to own an apartment before they get married. Based on the empirical evidence of population dynamics, the chapter suggests that a 1% increase in housing price will result in 0.016 years delay in the age of first marriage in China. When the housing price increases too much, it will have an even larger effect on marriage decisions because the high housing price becomes unaffordable for young couples. The result is robust with a wide range of model specifications.

We first describe a general class of optimization problems that describe many natu- ral, economic, and statistical phenomena. After noting the existence of a conserved quantity in a transformed coordinate system, we outline several instances of these problems in statistical physics, facility allocation, and machine learning. A dynamic description and statement of a partial inverse problem follow. When attempting to optimize the state of a system governed by the generalized equipartitioning princi- ple, it is vital to understand the nature of the governing probability distribution. We show that optimiziation for the incorrect probability distribution can have catas- trophic results, e.g., infinite expected cost, and describe a method for continuous Bayesian update of the posterior predictive distribution when it is stationary. We also introduce and prove convergence properties of a time-dependent nonparametric kernel density estimate (KDE) for use in predicting distributions over paths. Finally, we extend the theory to the case of networks, in which an event probability density is defined over nodes and edges and a system resource is to be partitioning among the nodes and edges as well. We close by giving an example of the theory’s application by considering a model of risk propagation on a power grid.