We investigate empirically the usefulness of price-cap and quality regulation in terms of allocative efficiency and welfare. An analytical framework allows us to determine sufficient conditions for an increase in welfare. We propose Malmquist productivity indices and their decomposition to check the conditions and to see whether it was a better-solved trade off between quality and costs that caused the welfare increase. The application of this method to a representative sample of Norwegian distribution system operators yields strong evidence for a positive effect of quality regulation on welfare. (author's abstract) / Series: Working Papers / Research Institute for Regulatory Economics

Este trabalho pretende contribuir para a ampliação do debate em torno da trajetória dos ganhos de produtividade no Brasil, examinando o desempenho da produtividade total dos fatores (PTF) agregada no período 1987/2002, nas unidades da federação, através da estimação de intervalos de confiança para o índice de Malmquist e suas componentes variação de eficiência e taxa de progresso tecnológico. Para tanto, aplica-se a metodologia de bootstrap, conforme Simar e Wilson (1999a) aos índices obtidos pela técnica DEAMalmquist, de modo que se os intervalos de confiança – com probabilidade de 90% e 95% - contenham a unidade, o índice em questão não é significativamente diferente de 1 e, portanto, não é possível concluir que existam mudanças na PTF, na eficiência ou na tecnologia. Com a aplicação do bootstrap, a conclusão mais evidente é que se deve ter cautela na análise e comparação entre unidades produtivas através do mero exame dos índices de Malmquist calculados. Em alguns casos, a técnica corrobora os resultados encontrados através do índice, mas em outros se conclui que não se pode afirmar que as variações sejam, de fato, estatisticamente significantes. / This tesis estimates the confidence intervals for output oriented Malmquist indices of productivity and their decompositions – changes in efficiency and changes in technology – in the period 1987/2002 taking into consideration Brazilian States. For this study, a bootstrap algorithm described in Simar and Wilson (1999a) was used, this procedure is a consistent estimation for obtaining confidence intervals for Malmquist indices. So the purpose is to contribute for further debates on total factor productivity performance in Brazil, providing results for both 90% and 95% confidence intervals. They allow assessment of the null hypothesis of no total factor productivity change which indicates that the corresponding measures are not statistically different from unity. The interpretation is straightforward. In the confidence interval case, if it contains the unity, then the corresponding measure is not significantly different from one at the significance level, i.e., it is not possible to conclude that changes occurred in productivity. In contrast, when the interval excludes the unity, one can conclude with confidence that the corresponding measure is significantly different from unity. With the application of bootstrap methodology, it is possible to conclude the necessity of being cautious while analyzing and comparing productive units through the mere exam of the calculated Malmquist indices. In some cases the technique corroborates the results found through the indices, whereas in others it may be concluded that it is not possible to state that variations are, in fact, statistically significant.

Este trabalho pretende contribuir para a ampliação do debate em torno da trajetória dos ganhos de produtividade no Brasil, examinando o desempenho da produtividade total dos fatores (PTF) agregada no período 1987/2002, nas unidades da federação, através da estimação de intervalos de confiança para o índice de Malmquist e suas componentes variação de eficiência e taxa de progresso tecnológico. Para tanto, aplica-se a metodologia de bootstrap, conforme Simar e Wilson (1999a) aos índices obtidos pela técnica DEAMalmquist, de modo que se os intervalos de confiança – com probabilidade de 90% e 95% - contenham a unidade, o índice em questão não é significativamente diferente de 1 e, portanto, não é possível concluir que existam mudanças na PTF, na eficiência ou na tecnologia. Com a aplicação do bootstrap, a conclusão mais evidente é que se deve ter cautela na análise e comparação entre unidades produtivas através do mero exame dos índices de Malmquist calculados. Em alguns casos, a técnica corrobora os resultados encontrados através do índice, mas em outros se conclui que não se pode afirmar que as variações sejam, de fato, estatisticamente significantes. / This tesis estimates the confidence intervals for output oriented Malmquist indices of productivity and their decompositions – changes in efficiency and changes in technology – in the period 1987/2002 taking into consideration Brazilian States. For this study, a bootstrap algorithm described in Simar and Wilson (1999a) was used, this procedure is a consistent estimation for obtaining confidence intervals for Malmquist indices. So the purpose is to contribute for further debates on total factor productivity performance in Brazil, providing results for both 90% and 95% confidence intervals. They allow assessment of the null hypothesis of no total factor productivity change which indicates that the corresponding measures are not statistically different from unity. The interpretation is straightforward. In the confidence interval case, if it contains the unity, then the corresponding measure is not significantly different from one at the significance level, i.e., it is not possible to conclude that changes occurred in productivity. In contrast, when the interval excludes the unity, one can conclude with confidence that the corresponding measure is significantly different from unity. With the application of bootstrap methodology, it is possible to conclude the necessity of being cautious while analyzing and comparing productive units through the mere exam of the calculated Malmquist indices. In some cases the technique corroborates the results found through the indices, whereas in others it may be concluded that it is not possible to state that variations are, in fact, statistically significant.

Este trabalho pretende contribuir para a ampliação do debate em torno da trajetória dos ganhos de produtividade no Brasil, examinando o desempenho da produtividade total dos fatores (PTF) agregada no período 1987/2002, nas unidades da federação, através da estimação de intervalos de confiança para o índice de Malmquist e suas componentes variação de eficiência e taxa de progresso tecnológico. Para tanto, aplica-se a metodologia de bootstrap, conforme Simar e Wilson (1999a) aos índices obtidos pela técnica DEAMalmquist, de modo que se os intervalos de confiança – com probabilidade de 90% e 95% - contenham a unidade, o índice em questão não é significativamente diferente de 1 e, portanto, não é possível concluir que existam mudanças na PTF, na eficiência ou na tecnologia. Com a aplicação do bootstrap, a conclusão mais evidente é que se deve ter cautela na análise e comparação entre unidades produtivas através do mero exame dos índices de Malmquist calculados. Em alguns casos, a técnica corrobora os resultados encontrados através do índice, mas em outros se conclui que não se pode afirmar que as variações sejam, de fato, estatisticamente significantes. / This tesis estimates the confidence intervals for output oriented Malmquist indices of productivity and their decompositions – changes in efficiency and changes in technology – in the period 1987/2002 taking into consideration Brazilian States. For this study, a bootstrap algorithm described in Simar and Wilson (1999a) was used, this procedure is a consistent estimation for obtaining confidence intervals for Malmquist indices. So the purpose is to contribute for further debates on total factor productivity performance in Brazil, providing results for both 90% and 95% confidence intervals. They allow assessment of the null hypothesis of no total factor productivity change which indicates that the corresponding measures are not statistically different from unity. The interpretation is straightforward. In the confidence interval case, if it contains the unity, then the corresponding measure is not significantly different from one at the significance level, i.e., it is not possible to conclude that changes occurred in productivity. In contrast, when the interval excludes the unity, one can conclude with confidence that the corresponding measure is significantly different from unity. With the application of bootstrap methodology, it is possible to conclude the necessity of being cautious while analyzing and comparing productive units through the mere exam of the calculated Malmquist indices. In some cases the technique corroborates the results found through the indices, whereas in others it may be concluded that it is not possible to state that variations are, in fact, statistically significant.

Cette thèse propose une analyse économique de la qualité et de l'efficience des universités et des systèmes universitaires au niveau international, à travers l'utilisation des indicateurs intrants et extrants et la méthode d'enveloppement des données (Data Envelopment Analysis DEA). La thèse est composée de quatre chapitres. Le premier chapitre intitulé "les classements des universités : une mise en perspective critique" présente et évalue la pertinence des indicateurs intrants/extrants les plus utilisés par les classements des universités. Il est l'occasion d'exposer un certain nombre de critiques présentes dans la littérature et de focaliser sur un problème méthodologique répandu dans les classements. Il s'agit de l'utilisation des inputs (intrants) comme des mesures de la qualité des universités. Une telle pratique confond moyen et résultat et fait l'impasse sur les concepts de base des modèles de comptabilité en termes de fonction de production et d'efficience. Le deuxième chapitre intitulé "caractéristiques et classements des universités : autour de quelques facteurs capables d'expliquer les différences de performance entre les universités", compare les résultats de deux classements: QS-Times et Shanghai et propose une liste de facteurs capables d'expliquer pourquoi il existe de telles différences de qualité, selon ces classements, entre les universités. [...] Le troisième chapitre intitulé "la performance et l'efficience des universités et leurs déterminants: une évaluation à partir des classements mondiaux des universités et de la méthode DEA" évalue sur la base d'une méthode DEA l'efficience de 214 universités issues de 13 pays différentes afin de savoir si les universités les mieux classées parmi les classements traditionnels sont aussi les universités qui utilisent le mieux leurs ressources. [...] Enfin, le quatrième chapitre, titré "l'efficience des systèmes universitaires de 35 pays et leurs déterminants: une évaluation par la méthode DEA et le calcul des indices de Malmquist (2006-2012)" évalue l'efficience et les performances des systèmes universitaires de 35 pays. Il propose de nouveaux scores d'efficience globale qui viennent compléter les deux premières études sur ce sujet d’Agasisti (2011) et St. Aubyn et al (2009). Les nouveautés par rapport à l'article d'Agasisti (2011) sont au nombre de cinq: l'échantillon est plus important (35 pays au lieu de 18), la période d'observation est actualisée, l'évolution de l'efficience entre deux périodes est calculée, le nombre des intrants et des extrants incorporé dans chaque modèle est plus élevé et un modèle spécifique d'évaluation de l'efficience de la recherche est proposé. Il conforte la thèse selon laquelle les systèmes universitaires de la Suisse et du Royaume-Uni sont les plus efficients. Il montre aussi sur la base d'un calcul des indices de Malmquist qu'entre 2006 et 2012 l'efficience en matière d'enseignement des 35 systèmes universitaires évalués a eu tendance à diminuer alors que l'efficience de la recherche et de l'attractivité- réputation de ces mêmes systèmes a plutôt eu tendance à augmenter. Ce qui permet de mieux évaluer l'effet des réformes inspirées par l'usage du classement de Shanghai sur les systèmes universitaires. Ces réformes conduisent les enseignants-chercheurs à délaisser l'enseignement pour la recherche. / This thesis aims to economically analyze the quality and efficiency of universities and university systems at an international level of comparison, by using input/output indicators and the Data Envelopment Analysis (DEA) method. The thesis is composed of four chapters. The first chapter entitled "university rankings: a critical perspective" presents and evaluates the relevance of inputs/outputs indicators used by most university rankings. It is the opportunity to present a number of criticisms found in the literature and focus on a common methodological problem in the rankings. It is the use of inputs as measures of university quality. This practice confuses means and results and ignores the basic concepts of accounting models in terms of production functions and efficiency. The second chapter entitled "characteristics and rankings of universities : around some factors that can explain the differences in performance between universities", compares the results of two rankings: QS-Times and Shanghai and offers a list of factors that may explain why there are such differences in quality, according to these rankings between universities. [...] The third chapter entitled "performance and efficiency of universities and their determinants: an evaluation using world university rankings and DEA methodology" evaluates on the basis of a DEA methodology the efficiency of 214 universities from 13 different countries, in order to find if the top ranked universities among traditional rankings are also universities that best utilize their financial and human resources. [...] The fourth chapter titled "efficiency of university systems in 35 countries and its determinants: an assessment by DEA methodology and the calculation of Malmquist indices (2006-2012)" assesses the efficiency and performance of university systems of 35 countries. It offers new scores for overall efficiency that complement the first two studies on this topic in the literature by Agasisti (2011) and St.Aubyn et al (2009). Compared to the article of Agasisti (2011), we identify five new developments in our study : the sample is higher (35 countries instead of 18), the observation period is updated, the evolution of efficiency between two periods is calculated, the number of inputs and outputs incorporated into each model is higher and a specific model for evaluating the efficiency of research is proposed. Our study confirms the thesis that the university systems of Switzerland and the United Kingdom are the most efficient. It also shows based on the calculations of Malmquist indices between 2006 and 2012 that teaching efficiency of 35 reviewed university systems has a tendency of declining while the research efficiency and that of attractivity-reputation is rather increasing. This allows a better assessment of the impact of reforms inspired by the Shanghai ranking on university systems. These reforms led the academic staff of universities to abandon their focus on teaching in favor of research activities.