An official website of the United States government
Here’s how you know
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
Secure .gov websites use HTTPS
A lock (
) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.
We present a method for estimating regression models with autoregressive integrated moving average (ARIMA) time series errors. The method maximizes the likelihood for different groups of parameters (AR, regression, and MA) separately within each iteration. The idea is to gain numerical efficiency by using generalized least squares (GLS) to maximize the likelihood over the regression and the autoregressive parameters, leaving only the moving average parameter estimates to be obtained by a nonlinear optimization routine. The method uses the "exact likelihood" suggested by Hillmer and Tiao (1979) that is the exact likelihood for pure moving average models. Implementing the method amounts to feeding vectors of the regression and lagged dependent variable to routines that calculate exact likelihood residuals for pure MA models, and then doing regression with these residuals to get the regression and AR parameters. In this way the same software used for exact MA likelihood estimation may be easily modified and used to estimate models with AR and regression effects.
Share
Related Information
Some content on this site is available in several different electronic formats. Some of the files may require a plug-in or additional software to view.
Top