9701959 Hodgson Economic time series models can often be reduced to a sequence of uncorrelated innovations that may be characterized by complicated forms of dependence, heterogeneity, and non-Gaussianity (i.e., non-normality) that are difficult to model. Despite long-standing awareness of this fact, the range of econometric methodologies currently used by empirical researchers rarely extends much beyond maximum likelihood techniques that assume these innovations to be Gaussian, such as ordinary least squares. The objective of the project is to facilitate the extensions of this range through the development of adaptive maximum likelihood techniques and semiparametric efficient estimators applicable in many modeling contexts of interest to financial and macroeconomics, and through the development of a software package implementing these techniques. The educational component of this CAREER award provides graduate students training in adaptive estimation techniques as part of the syllabi of advanced graduate level courses on either nonparametric econometric techniques, theory of efficient estimation, applied time series econometrics, or applied financial econometrics. Graduate students will also be trained in the use of the software developed under this project. .
