Generalization of the residualization procedure. Properties and environmental applications

• Autores: Claudia García García
• Directores de la Tesis: Catalina Beatriz García García (dir. tes.), Román Salmerón Gómez (dir. tes.)
• Lectura: En la Universidad de Granada ( España ) en 2020
• Idioma: inglés
• Número de páginas: 197
• Tribunal Calificador de la Tesis: Agustín Hernández Bastida (presid.), Teresa García-Muñoz (secret.), Dimitrios Voliotis (voc.), Juan Gómez García (voc.), María del Mar López Martín (voc.)
• Programa de doctorado: Programa de Doctorado en Ciencias Económicas y Empresariales por la Universidad de Granada
• Materias:
  • Ciencias económicas
    • Econometría
      • Indicadores económicos
      • Modelos econométricos
  • Ciencia política
    • Ciencias políticas
      • Política del medio ambiente
• Enlaces
  • Tesis en acceso abierto en: DIGIBUG
• Resumen

  • The dissertation presented here aims to clarify the role of multicollinearity in an econometric model and proposes residualization as a good methodology not only to deal with the problem but also to achieve another type of interpretation of the explanatory variables from the model under consideration. Chapter 1 gave the reader an initial introduction to the problem and oered a brief explanation of the methodology being presented. Chapter 2 then looked in depth at the multicollinearity problem and the traditional methodologies used in this eld. Chapters 3 and 4 explain the methodology further; Chapter 3 focuses its attention on earlier works in the eld: criticism of the method and methodological preliminaries, while Chapter 4 presents the generalization of the method, together with the justication and properties of the residualization procedure. These two chapters and in particular Chapter 4, are the main contribution of this Thesis. Finally, Chapter 5 presents the empirical part: three specic models on environmental economics, which present strong collinearity problems. In conclusion, even when the goal of the study is to predict (where it has been concluded previously that collinearity is not signicant), it is highly recommended to mitigate the problem because the researcher needs to be very sure of the continuity of the relationships between explanatory variables in the future because, if the relationship changes, the forecast based on the initial model may be unreliable as well.