This dissertation contains a variety of contributions to econometric theory. Broadly speaking, econometrics may be categorized depending on what type of data is being analyzed, the two main categories being time series data and cross sectional data. A third category is panel data, combining cross sectional data observed over time. This is the area at which this dissertation is aimed. Panel data models may themselves be divided into sub models depending on what the explanatory variable looks like. In the most common type of model, the explanatory variable is a continuous variable. In another class of models, the explanatory variable is time to an event. The analysis of such models is called Duration Analysis. Duration analysis utilizes panel data since the explanatory variables are measured over time as well as over individuals/firms/countries. The first paper, Behind the Diffusion Curve: An Analysis of ATM Adoption, co-authored with Sunil Sharma, is aimed at making contributions to duration analysis. An important aspect of applying duration analysis in economics is the fact that the data is grouped, something which is typically not the case in the other fields. The first paper suggests a new method of evaluating the fit of the model when the data is grouped. The second paper, Evaluating the Proportionality Assumptions for the Duration of Unemployment is also aimed at making contributions to duration analysis. The models used in economics for studying time to an event is almost exclusively based on the Cox’s proportional hazard model. There are many examples where such a specification is unrealistic. This paper suggests a new method of testing the proportionality assumption that has several advantages. The third paper, Stochastic Frontier Production Functions with Errors in Variables is meant to make a contribution to the literature on cross sectional data when you have measurement errors in the explanatory variables (labor and capital). The model I consider is the one studied by Aigner et al (1977) specifying the production function as a Cobb-Douglas. This paper develops a procedure for consistently estimating the parameters in the production function and the technical efficiencies when there is measurement errors but where the reliability of the data is known.
|Tilldelningsdatum||1998 okt. 27|
|Status||Published - 1998|
Bibliografisk informationDefence details
Name: Brännäs, Curt