On the stability of recursive least squares in the Gauss-Markov model

Salies, Evens (2004): On the stability of recursive least squares in the Gauss-Markov model.

Abstract

In the Gauss-Markov regression model, one can always update the least square estimate of the slope vector, given new observations at the values of the explanatory variables. The updated estimate is often considered as a time-varying state of an auto-regressive system in Kalman filtering and recursive least squares theory. This note shows that the auto-regressive matrix of this dynamic system once centered has its largest eigenvalues equal to 1 and one eigenvalue that is less than 1.

• Item Type: MPRA Paper
• Language: English
• Keywords: Recursive Least Squares, Gauss-Markov model.
• Subjects: C - Mathematical and Quantitative Methods > C3 - Multiple or Simultaneous Equation Models ; Multiple Variables; C - Mathematical and Quantitative Methods > C4 - Econometric and Statistical Methods: Special Topics

References

Harvey, A. C. (1990). The Econometric analysis of time series. 2nd ed., Cambridge, MA: MIT Press.

Kianifard F., and Swallow, H. (1996). A review of the development and application of recursive residuals in linear models. Journal of the American Statistical Association, 91(433), pp. 391-400.

Magnus, J. R. and Neudecker, H. (1991). Matrix Differential Calculus with Applications in Statistics and Econometrics. John Wiley. Printed in Great Britain.