Least Squares Estimation of Conditionally Heteroscedastic Autoregressions.
Abstract
The least squares estimate of the autoregressive parameter of a conditionally heteroscedastic autoregression is consistent and asymptotically normal. Failure to recognize conditional heteroscedasticity results in the underestimation of the variance of the least squares estimate, and in extreme cases, this effect can be substantial. The least squares estimate is not asymptotically distribution free, rather, the asymptotic distribution depends on the form of the conditional heteroscedasticity. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1984
- Accession Number
- ADA145994
Entities
People
- A. F. L. Nemec
Organizations
- University of Washington