Thèses sur le sujet « Endogenous regressor »

The first chapter presents new methods for identifying the structural parameters of linear triangular systems, simultaneous systems, and structural vector autoregressions. The second chapter presents a new method for identifying an endogenous regressor in linear models of time series data
Thesis (PhD) — Boston College, 2006
Submitted to: Boston College. Graduate School of Arts and Sciences
Discipline: Economics

The focus here is on the log-normal version of the spatial interaction model. In this context, we consider spatial econometric specifications that can be used to accommodate two types of dependence scenarios, one involving endogenous interaction and the other exogenous interaction. These model specifications replace the conventional assumption of independence between origin-destination-flows with formal approaches that allow for two different types of spatial dependence in flow magnitudes. Endogenous interaction reflects situations where there is reaction to feedback regarding flow magnitudes from regions neighboring origin and destination regions. This type of interaction can be modeled using specifications proposed by LeSage and Pace (2008) who use spatial lags of the dependent variable to quantify the magnitude and extent of feedback effects, hence the term endogenous interaction. Exogenous interaction represents a situation where spillover arise from nearby (or perhaps even distant) regions, and these need to be taken into account when modeling observed variation in flows across the network of regions. In contrast to endogenous interaction, these contextual effects do not generate reaction to the spillovers, leading to a model specification that can be interpreted without considering changes in the long-run equilibrium state of the system of flows. We discuss issues pertaining to interpretation of estimates from these two types of model specification, and provide an empirical illustration. (authors' abstract)

Submitted by Maike Costa (maiksebas@gmail.com) on 2016-04-12T13:00:18Z No. of bitstreams: 1 arquivo total.pdf: 1701778 bytes, checksum: 4e6e0ca5635cf810784f97567b64095f (MD5)
Made available in DSpace on 2016-04-12T13:00:18Z (GMT). No. of bitstreams: 1 arquivo total.pdf: 1701778 bytes, checksum: 4e6e0ca5635cf810784f97567b64095f (MD5) Previous issue date: 2014-03-10
This paper conducts tests for structural breaks in the reaction function of the Central Bank of Brazil to evaluate possible changes in the conduct of monetary policy in Brazil, taking into account the reaction function of the regressors are potentially endogenous variables. For this, we uses the methodology developed by Hall et al. (2012) who, using an extension of the framework developed by Bai and Perron (1998), develops a method capable of identifying multiple structural breaks at unknown periods. The main results indicate presence of structural breaks in the three reaction functions studied. Furthermore, the actions of monetary policy through interest rate Selic seem to suffer greater influence of deviations of inflation around its target, compared to changes in the output gap and the exchange rate.
Este estudo realiza testes de quebra estrutural na função de reação do Banco Central do Brasil para avaliar possíveis mudanças na condução da política monetária doméstica, levando-se em conta que os regressores da função de reação são potencialmente variáveis endógenas. Para isto, é utilizada a metodologia desenvolvida por Hall et al. (2012) que, utilizando uma extensão da estrutura desenvolvida por Bai e Perron (1998), elaboram um método capaz de identificar múltiplas quebras estruturais em períodos desconhecidos. Os principais resultados apontam para a presença de quebras estruturais nas três funções de reação estudadas. Além disso, as ações da política monetária, por meio da taxa de juros Selic, parecem sofrer maior influência dos desvios da inflação em torno de sua meta, comparativamente a variações no hiato do produto e na taxa de câmbio.

A presente tese é constituída de três ensaios que abordam duas relevantes questões que estão intrinsecamente relacionadas em macroeconomia: política monetária e inflação. No primeiro ensaio, nós procuramos averiguar não linearidades na função de reação do Banco Central do Brasil (BCB) através da estimação de regressões quantílicas inversa, sugerido por Wolters (2012) e proposto por Chernozhukov and Hansen (2005, 2006). Este método nos possibilitou detectar não linearidades na função de reação do BCB sem a necessidade de fazer suposições específicas acerca dos fatores que determinam essas não linearidades. Em específico, nós observamos que: i) a resposta da taxa de juros ao hiato da inflação corrente e esperada foi, em geral, mais forte na parte superior da distribuição condicional da taxa de juros Selic; ii) a resposta ao hiato do produto apresentou uma tendência crescente e significativa na parte inferior da distribuição condicional da taxa Selic; iii) a resposta do BCB à taxa de câmbio real foi positiva e mais elevada na cauda superior da distribuição condicional da taxa Selic. No segundo ensaio, nós investigamos a existência de não linearidades na função de reação do Banco Central do Brasil (BCB) decorrentes de incertezas desse policymaker acerca dos efeitos do hiato do produto sobre a inflação. Teoricamente, nós seguimos Tillmann (2011) para obter uma regra de política monetária ótima não linear que é robusta às incertezas acerca do trade-off produto-inflação na curva de Phillips. Além disso, nós realizamos testes de quebra estrutural para avaliar possíveis mudanças na condução da política monetária brasileira durante o regime de metas de inflação. Os resultados indicaram que: i) as incertezas acerca da inclinação na curva Phillips implicaram em não linearidades na função de reação do BCB; ii) não se pode rejeitar a hipótese de uma quebra estrutural nos parâmetros da regra monetária ocorrendo no terceiro trimestre de 2003; iii) houve um aumento na resposta da taxa Selic ao hiato do produto e uma redução da reação ao hiato da inflação corrente no regime Meirelles- Tombini; e iv) o BCB também tem reagido à taxa de câmbio durante o regime Meirelles- Tombini. No terceiro ensaio, nós procuramos analisar os determinantes da inflação no Brasil através da estimação da Curva de Phillips Novo-Keynesiana (CPNK) proposta por Blanchard e Galí (2007) e a versão padrão proposta por Galí e Gertler (1999). Além disso, realizamos testes de quebras estruturais para avaliar possíveis mudanças na dinâmica da inflação brasileira durante o período de 2002 a 2014. Os resultados indicaram que: i) os testes de quebra estrutural apontam a existência de pelo menos uma mudança estrutural nos coeficientes da CPNK; ii) o componente forward-looking da inflação é dominante, embora sua relevância tenha sido reduzida após 2004; iii) a taxa de desemprego tem afetado negativamente a inflação, embora seja observado uma redução desse impacto nos últimos anos; iv) as mudanças na taxa de câmbio apenas tiveram efeitos sobre a inflação na primeira subamostra e tem perdido relevância no período mais recente; v) o efeito do hiato do produto sobre a inflação corrente diminuiu nos anos recentes; vi) em geral, nós rejeitamos a hipótese nula de uma curva de Phillips vertical no longo prazo a um nível de significância de 5%, mas não a 1%.
This thesis is composed of three essays to address two important issues that are intricately related in macroeconomics: monetary policy and inflation. In the first essay, we seek to investigate nonlinearities in the reaction function of the Central Bank of Brazil (CBB) by estimating inverse quantile regressions (IVQR), suggested by Wolters (2012) and proposed by Chernozhukov and Hansen (2005, 2006). This method enabled us to detect nonlinearities in the CBB’s reaction function without the need to make specific assumptions about the factors that determine these nonlinearities. In particular, we observed that: i) the response of the interest rate to the current and expected inflation was, in general, stronger in the upper tail of the conditional interest rate distribution; ii) the response to the output gap showed a growing and significant trend in the lower tail of the conditional Selic rate distribution; iii) the response of the CBB to the real exchange rate was positive and higher in the upper tail of the conditional Selic rate distribution. In the second essay, we investigate the existence of nonlinearities in the reaction function of the Central Bank of Brazil (CBB) arising from this policymaker’s uncertainties about the effects of the output gap on inflation. Theoretically, we follow Tillmann (2011) to obtain a nonlinear optimal monetary policy rule that is robust to uncertainty about the output-inflation trade-off of the Phillips Curve In addition, we perform structural break tests to assess possible changes in the conduct of the Brazilian monetary policy during the inflation-targeting regime. The results indicate that: i) the uncertainties about the slope in the Phillips curve implied nonlinearities in the CBB’s reaction function; ii) we cannot reject the hypothesis of a structural break in the monetary rule parameters occurring in the third quarter of 2003; iii) there was an increase in the response of the Selic rate to output gap and a weaker response to the current inflation gap in Meirelles Tombini’s administration; and iv) the CBB has also reacted to the exchange rate in Meirelles-Tombini’s administration. In the third essay, we proposed to analyze the determinants of inflation in Brazil through the estimation of the new Keynesian Phillips curve (NKPC) proposed by Blanchard and Galí (2007) and the standard version proposed by Galí and Gertler (1999). In addition, we perform structural break tests to assess possible changes in the dynamics of inflation in Brazil during the period 2002 to 2014. The results indicated that: i) structural break tests indicate the existence of at least one structural change in the coefficients of NKPC ; ii) the forward-looking component of inflation is dominant, though its importance has been reduced after 2004; iii) the unemployment rate has negatively affected inflation, although a reduction of this impact has been observed in recent years; iv) changes in the rate of exchange only had effects on inflation in the first subsample and losing relevance in the most recent period; v) the effect of the output gap on the current inflation has declined in recent years; vi) overall, we reject the null hypothesis of a vertical Phillips curve in the long term at a significance level of 5%, but not 1%.

China's economic development has performed spectacularly during the period of China's economic transition as a result of radical economic reform in the all markets. The country has also gone through extensive fiscal reforms in the last three decades. However, a number of problems have been associated with such rapid economic growth. One of these has been raising inequality. In both Keynesian and neoclassical endogenous growth theories, public spending can play an important role for economic growth and inequality. The majority of previous studies have focused on the relationship between public spending and economic growth, or between public spending and inequality separately. There is no doubt that public spending has an effect on both economic growth and equity simultaneously. In this respect, this thesis attempts to address the problems that have emerged during the period of China's fiscal reforms, and seeks to examine the effects of public spending on economic growth and equality in the same model. This thesis investigates the dynamic relationships among these three variables in China. For aggregate national data, vector error correction model (VECM) has been used. Analysis at the provincial level is based on the panel vector auto-regression (PVAR) model. These methods help to solve the endogeneity in estimations. The national level analysis indicates that total public spending shows a long term Granger causality with GDP per capita, which supports the positive growth effect of public spending in the Keynesian and endogenous growth model. Social public spending has a negative effect on real output per capita in both the short term and long term, but it also has a negative impact on income inequality. Moreover, we find that a higher level of real GDP per capita will increase the level of inequality, but a higher level of inequality has a negative effect on real GDP per capita in the long term. Furthermore, total provincial public spending and provincial social spending have either a non-significant effect on economic growth. On the other hand, the SOEs' investment has a significant, positive growth effect at both the national and provincial level. As for the redistributive role of the public spending, the provincial total public spending and social spending have played an important role on income distribution. Furthermore, the Gini coefficient has a positive effect on the per capita growth rate at the provincial level, but the economic growth has no significant impact on the Gini coefficient.

Submitted by Ana Luiza Neves de Holanda Barbosa (ana.barbosa@ipea.gov.br) on 2012-01-17T17:48:38Z No. of bitstreams: 2 tese_final_analuizaholanda.pdf: 1401652 bytes, checksum: fd6723241862cf776f8dcdc02a400af2 (MD5) tese_final_analuizaholanda.pdf: 1401652 bytes, checksum: fd6723241862cf776f8dcdc02a400af2 (MD5)
Approved for entry into archive by Andrea Virginio Machado (andrea.machado@fgv.br) on 2012-03-05T19:50:53Z (GMT) No. of bitstreams: 2 tese_final_analuizaholanda.pdf: 1401652 bytes, checksum: fd6723241862cf776f8dcdc02a400af2 (MD5) tese_final_analuizaholanda.pdf: 1401652 bytes, checksum: fd6723241862cf776f8dcdc02a400af2 (MD5)
Made available in DSpace on 2012-03-07T19:29:24Z (GMT). No. of bitstreams: 2 tese_final_analuizaholanda.pdf: 1401652 bytes, checksum: fd6723241862cf776f8dcdc02a400af2 (MD5) tese_final_analuizaholanda.pdf: 1401652 bytes, checksum: fd6723241862cf776f8dcdc02a400af2 (MD5) Previous issue date: 2011-09-12
This study investigates the wage gap between the public and private sectors in Brazil. The analysis is carried out with 2009 microdata from the Pesquisa Nacional por Amostra de Domicílios (PNAD) and it takes account its complex sample design. The main contribution of this study is the estimation of a switching endogenous regression model that corrects the selection bias in the choice of employment sector. This model allows for the identification of some factors that determine the entrance of the individual in the public sector labor market. The public-private wage gap is calculated by gender as labor supply varies significantly between women and men. The results show that public sector wages are higher than those in the private sector. In particular, the public-private wage gap for women is higher than for men.
A presente tese engloba três artigos sobre diferencial de salários e estimação de demanda no Brasil. O primeiro artigo investiga o diferencial de salários entre os trabalhadores dos setores público e privado. A principal contribuição deste estudo é a estimação de um modelo de regressão com mudança endógena (endogenous switching regression model), que corrige o viés de seleção no processo de escolha setorial realizada pelos trabalhadores e permite a identificação de fatores determinantes na entrada do trabalhador no mercado de trabalho do setor público. O objetivo do segundo trabalho é calcular a elasticidade-preço e a elasticidade-despesa de 25 produtos alimentares das famílias residentes nas áreas rurais e urbanas do Brasil. Para tanto, foram estimados dois sistemas de equações de demanda por alimentos, um referente às famílias residentes nas áreas rurais do país e o outro sistema associado às famílias residentes nas áreas urbanas. O terceiro artigo busca testar a validade do modelo unitário para solteiros(as) e a validade do modelo de racionalidade coletiva de Browning e Chiappori (1998) para casais no Brasil. Para tanto, foi estimado um sistema de demanda do consumo brasileiro com base no modelo QUAIDS, que apresenta uma estrutura de preferências flexível o suficiente para permitir curvas de Engel quadráticas.

This thesis studies the effects of import competition from China and Eastern Europe on the health and fertility decisions of German individuals working in manufacturing. Individuals are matched with separate measures of exposure to competition from China and Eastern Europe, respectively. To isolate exogenous supply shocks from the origin, instrumental variables for competition from each of China and Eastern Europe are constructed. Results in Chapter 1 suggest that higher import competition worsens individual health via job displacement, wage decline, shortened employment duration, increased reliance on welfare and less future orientation, with Chinese import competition affecting individuals twice as much. Health declines as individuals increase their visits to the doctor, exercise less frequently and have a higher probability of developing chronic illness. Also, there is some evidence that individuals do not tend to become disabled but may be slowly pushed into chronic illness. Findings in Chapter 2 show that import competition negatively affects the individual’s probability of having children via reduced earnings, lower satisfaction with personal income and shortened employment duration. The chapter then investigates effects of import exposure by gender. Results show that male and female fertility choices differ upon rising import competition. Higher import exposure lowers female earnings and job autonomy, which in turn generates a lower opportunity cost of work, to the point where having children would become a more rewarding alternative for female workers. By contrast, increased import exposure negatively affects male workers’ fertility through reduced earnings and employment duration.

碩士
佛光大學
經濟學系
96
We derived the endogenous switching regression model (ESRM) to solve the independent variables which sample selection lead to a biased inference about we estimated the retired household of the leisure expenditure equation and near retired household of the leisure expenditure equation with the OLS, because of the household heads would have been evaluated between the differences of the retired or unretired situation on leisure expenditures. We used the data from the Academia Sinica Research Program Center Databases Resources one of the 2006 Survey of Family Income and Expenditure Consumer Expenditure. The empirical result shows that household head had been transferred to be a retiree which was decided by the criterion function and didn’t relate to education which was to be the symbol the social, but reference with the family income, characteristic family, home ownership, physical situation, the presence of a social security system, the effect of alternative policies on retirement behavior, the expected to lifetime etc. After the household heads retired which impacted on the leisure activities, have been differed from the unretired household, so they had to cut down the leisure expenditures. On the other hand, if the household heads doesn’t retire and still work in the job market, so they will spend more money on leisure expenditures for taking a vacation or going abroad. The household heads compared with the conditional of retired and near-retired who were to be a retiree and choice to engage in leisure activities and consume recreational goods, those will get the utility above the average level of expectation. The household heads transferred to be a retiree and will spend more money on leisure expenditures or be accomplished the planning of travel round the world who retired have above-average expected object from retirement. Alternatively, the near-retired household heads who have held the home ownership, unworried about their living, who could arrange the journey to take a vacation who unretired have above-average expected object from unretirement. This study provides the independent of the self-selection and the differentiate between the retired household and near-retired household, also provides information to researchers, policy makers, financial counselors, and business managers who focus on the leisure expenditures of retired and near-retired households.

碩士
佛光大學
經濟學系
98
This study is to improve inferential biases caused by estimation of each parameter of the family expenditure function of nuclear and non-nuclear families with the Ordinary Least Squares (OLS). Because every family can evaluate its family expenditure in two different situations by itself to see which one has more comparison benefit, then decide if changing family status. We adapt Endogenous Switching Regression Model, and use the data “Taiwan Social Change Survey 2007, Phase 5, Wave 3” from the Institute of Sociology, Academia Sinica. The importance of the endogenous switching regression model is to endogenize family type, which is the extraneous variable of effecting family expenditure. Why choose family type as endogenous variables, rather than general variable such as age, sex and level of education? It is because of that family type not only will directly influence the family expenditure, but more also may effect working situation, and then influence family expenditure. Through the endogenous switching regression model, we get the related coefficient of the nuclear family and non-nuclear family’s working situation. It shows that both family types are different symbols relations. The nuclear family whose occupation is self-runner has higher family expends. The possible reason could be that himself/herself is the owner, has higher incomes and more flexible time to arrange recreational activity. On the contrary, the opposite has no time to do recreational activity and therefore effects family expends. Regarding to the nuclear family which has saving behavior, because both parties of couple are easier to carry out the family financial managing plans and they also have plans on family expends, therefore they have more expend ability and family expends. As a result, if different family types have high income and well family financial managing plans, it is positive help for family fortune. This essay manifests the differences of family expends between nuclear and non-nuclear families. Therefore, while planning family financial management, every family type should consider the relativity and characteristic between nuclear and non-nuclear family types.

The costs of acquiring parcels by condemnation are usually significantly higher than those for property acquired by negotiation, suggesting that Right-of-Way (R/W) acquisition costs may best be described by two different regression equations. This paper develops a switching regression model of acquisition cost to simultaneously predict the probability of whether a parcel will go to condemnation rather than be acquired via negotiation and the corresponding acquisition costs under these two regimes. The error terms of the selection equation and the two cost equations follow a trivariate normal distribution to reflect correlations across unobserved factors (such as a land owner's tenacity or a site's view value). When this model is calibrated using data on properties acquired across the state of Texas for transportation projects between 2008 and 2011, results suggest that R/W appraisers and staff should pay special attention to commercially used parcels in urban areas involving a partial taking with a relatively small remainder. Comparison of cost estimates between the two regimes (condemnation vs. negotiation) suggests that condemned parcels will have, on average, 78% higher acquisition costs across the 1,710 acquired properties and 51% greater price variation. These results suggest that it is much more costly to acquire a property and more difficult to accurately predict its costs if it cannot be acquired via negotiation. The application of model estimates to an example corridor highlights the value of simulation to capture all modeling uncertainties. This two-regime model is further extended to a three-category multinomial endogenous switching, allowing for differential cost estimates across negotiation-deed, administrative settlement, and condemnation contexts. A model of acquisition time -- from the agency's initial-offer date to its final possession date is also developed, to examine the effects of condemnation on acquisition duration. The results suggest that condemnation proceedings add approximately 7 to 8 months, on average, to parcel acquisitions by the Texas Department of Transportation. Taken together, such switching models for condemnation versus negotiated settlement highlight the benefits of avoiding condemnation proceedings in R/W acquisition. Estimation results illuminate the relative importance of various parcel and owner attributes, impacting the nature and cost of acquisition, and enhancing opportunities for R/W staff to identify more contentious properties and establish more reliable budget estimates.
text

碩士
佛光大學
經濟學系
96
This study cited the 2006 Survey of Taiwan Family Income and Expenditures (STFIE) to explore the linear relationship of people’s expenditures among medical health, leisure expense, living area, outcome, gender of house owner, age, education, occupation and social insurance. In this study, the researcher adopted the endogenous switching regression model (ESRM) to find out the influential reasons and relationship between the leisure involvement and health care expenditure on family unit. It is to solve the assumption errors caused by the self-selection problem, when using the approach of OLS to analyze the function variables of health care expenditure in different leisure involvement group. The remarkable difference between this study and the past studies is to take the variable of leisure expenditure as an endogenous variable and obtain the relative possibilities. In addition, the model of probit is used to explore the effects of health care expenditure against different leisure involvements. Furthermore, under the application of ESRM, the researcher wanted to propose better explanations how the leisure involvement effects the health care expenditure The empirical result shows that no matter high leisure involvement group or low leisure involvement group will take part in suitable leisure activities. Consequently, their health care expenditure relatively drops down. The conclusion corresponds with the result of this study when the leisure involvement is endogenouslized. Keyword: leisure involvement, health care expenditure, endogenous switching regression model

碩士
佛光大學
經濟學系
97
This study investigates the effects of dietary quality and exercise of Taiwanese elder on obesity with Nutrition and Health Survey in Taiwan, NAHSIT,1997~2002 of Taiwanese nutrition and health survey (1997~2002) which Department of Health authorize Institute of Biomedical Sciences, Academia Sinica to transact; The total taken sample quantity is 2049. The Endogenous Switching Regression Model is important to affect the Exogeneous Variable(exercise is endogenized in this study) of elder obesity. Exercise can affect elder obesity directly but also changed by dietary quality indirectly and affect elder obesity further. Hence, with Endogenous Switching Regression Model the study takes the exercise level of sample-observing as the deciding variable of regression model-transferring. The effects of dietary quality and sociodemographic variables on bodyweight and evaluations of the improvement of elder dietary quality for obesity effects through the concept of dietary elasticity are focused. Finally evaluate the result of obesity improvement. The differences between this study and other records related with dietary, exercise and obesity are as following: 1. Exercise is endogenized. In a short period, it’s hard to change dietary quality for elder but the exercise level also is another way to affect obesity level. 2. Try to use binary model to investigate the effect level of dietary quality for obesity under the different energy activities level. The study states that: 1. Due correlation coefficient are different sign so takes that if there is regular exercise or not with comparative advantage as economical meaning. The group with exercise hobby has lower Body Mass Index because of exercise; on the other hand, the group without exercise habit has lower Body Mass Index because of no exercise habit as well. 2. Use dietary elasticity to evaluate the improvement of elder obesity. The elder with exercise improve their own obesity by changing dietary quality. If elder with continued exercise increase Healthy Eating Index by 10% then the obesity can be improved by 0.69%; this result is a statistically significance. While the dietary elasticity of elder without continued exercise isn’t a statistically significance and the sign is positive. Hereby it may say that changing dietary quality won’t improve obesity situation due there’s not continued exercise. For elder, there must be enough physical activity to make healthy dietary quality lead to lower obesity level (ex. low BMI.) if there is no proper physical activity but only with good dietary quality then the obesity still doesn’t become low obviously; instead, it may increase their own BMI. This study believes that endogenized exercise values for scholars and strategy executives who care about obesity popular ration very much.

JEL Classification: J31, J41, J68, C21, C24, C33
This thesis aims to study the effects of the promotion of flexibility at the margin in the Portuguese labour market through the facilitation of the use of fixed-term contracts. We present three empirical essays that assess the effects of the 2004 change in legislation that extended the maximum legal duration of fixed-term contracts from three to six years. Given the high labour market segmentation and representativeness of fixed-term contracts in the Portuguese labour market, we consider that these studies may contribute to the design of future public policies. The empirical analysis is conducted using the linked employer-employee database Quadros de Pessoal for the period between 2002 and 2011. The results of the three essays suggest that fixed-term contracts may play different roles in the labour market, namely as screening devices and quantitative adjustment tools, and that this fact should be taken into account when the effects of asymmetric employment protection reforms are analysed. After controlling for several micro and macro variables, we find evidence that the extension of the fixed-term contract for a longer period had negative effects on the probability of conversion of the contract and contributed to increase the wage inequality between workers on permanent or converted fixed-term contracts and those that did not obtain a more stable employment relationship. Besides, we find evidence that this change in legislation did not contribute to increase employment growth and decrease the rate of creation of fixedterm jobs, especially due to the effects of the proportion of non-converted fixed-term contracts. This research suggests that labour market segmentation should be tackled and policy makers should promote measures aiming to stimulate the conversion of fixedterm into open-ended contracts.
Esta tese pretende estudar os efeitos da promoção da flexibilidade à margem, através da facilitação do uso dos contratos a termo, no mercado de trabalho português. Para tal, apresentam-se três ensaios onde se analisam empiricamente os efeitos de uma alteração de legislação que ocorreu em 2004 e que aumentou a duração legal máxima dos contratos a termo certo de três para seis anos. Dada a elevada segmentação do mercado de trabalho e a representatividade dos contratos a termo, considera-se que estes estudos podem contribuir para a definição de políticas públicas no futuro. A análise empírica é realizada utilizando a base de dados Quadros de Pessoal para o período compreendido entre 2002 e 2011. Os resultados dos três ensaios sugerem que os contratos a termo podem desemprenhar papéis diferentes no mercado de trabalho, como processos de seleção e de ajustamentos quantitativos, e que isso deve ser tido em consideração quando se analisam os efeitos de reformas assimétricas da protecção ao emprego. Após controlar por diversas variáveis micro e macro, encontra-se evidência de que a possibilidade de utilizar o contrato a termo por um maior período de tempo teve efeitos negativos na probabilidade de conversão do contrato e contribuiu para aumentar a desigualdade salarial entre trabalhadores com contratos sem termo ou contratos convertidos em sem termo e trabalhadores que não obtiveram uma relação de emprego mais estável. Para além disso, mostra-se que esta alteração de legislação não se traduziu num aumento do crescimento do emprego e que teve inclusivamente um efeito negativo na criação de emprego com contratos a termo, devido principalmente ao efeito da não conversão dos contratos. Esta tese sugere que se deve combater a segmentação do mercado de trabalho português e promover medidas para estimular a conversão dos contratos a termo em contratos sem termo.