Econometric modeling of self-exciting process

• Autores: Yutao Li
• Directores de la Tesis: Miguel Angel Delgado González (dir. tes.)
• Lectura: En la Universidad Carlos III de Madrid ( España ) en 2019
• Idioma: español
• Tribunal Calificador de la Tesis: Taisuke Otsu (presid.), Juan Carlos Escanciano (secret.), Juan M. Rodríguez-Poo (voc.)
• Programa de doctorado: Programa de Doctorado en Economía por la Universidad Carlos III de Madrid
• Materias:
  • Ciencias económicas
    • Econometría
      • Modelos econométricos
• Enlaces
  • Tesis en acceso abierto en: e-Archivo
• Resumen

  • This thesis focuses on modeling state dependence, a phenomenon where past experiences do alter the path of future events. The general idea behind this concept is that differences among individuals are not merely explained by their characteristics, but also by their past experiences. Typical examples of state dependence in economics include the incidence of accidents, labor force participation and unemployment, consumer's purchase behaviors, learning etc.

    I use the self-exciting process to incorporate the state dependent structure in economic models. The self-exciting process is a counting process whose filtration includes a $\sigma$-field that is generated by the process itself.

    Chapter I introduces notation and provides an introduction to the self-exciting process. A minimum distance estimation (MDE) method is also introduced. Monte Carlo exercises are performed to investigate the MDE performance. I also provide a comparison among the self-exciting process approach and existing methods such as counting data regression and duration models.

    Chapter II studies how past doctor visiting records could alter the preference individuals' future medical consumption choices, especially through the channel of a cost-sharing health insurance plan. This study contributes to the existing health insurance literature by providing additional evidence that supports the shadow price theory. While most previous studies that endorse the shadow price use specific company or social security designs, I investigate the shadow price using a genuine health insurance contract where medical cost has multiple sources.

    Chapter III contributes to the study of work absence. Early works in this literature usually focus on absence duration and assume independence among the duration. I took another approach in this study where both absence duration and working duration (the length of a working spell until an absence occurred) are analyzed. Special attention is given to studying how the state dependent absence score could affect these duration. I study a particular firm who has installed a experience-rated work absence regulation. I also distinguish between short and long term absences and find that individuals have different state dependent reaction to different types of absences.

    Chapter IV investigates the classical unemployment duration problem. I restrict the attention to Spanish Youth, who are well known for their high job turnover rate. A crucial element in this literature is the separation of the state dependent and the unobserved heterogeneity. I did so by assuming a multiplicative duration structure and perform a first ratio transformation on the individual's unemployment duration to swipe out the unobserved heterogeneity. The new estimator could be regarded as an extension to the existing dynamic panel data model but it avoids using instrumental variables and could allow unit root autoregressive coefficient and non-stationarity process.