RUS  ENG
Full version
JOURNALS // Zhurnal Srednevolzhskogo Matematicheskogo Obshchestva // Archive

Zhurnal SVMO, 2011, Volume 13, Number 2, Pages 25–35 (Mi svmo234)

On nonstandard estimation methods in autoregression models in unstable cases

A. N. Startseva, T. S. Mirzaevb

a Plekhanov Russian State Academy of Economics
b Institute for Mathematics and Information Technologies of the National Academy of Sciences of Uzbekistan

Abstract: In the report is suggested some estimates of autoregression parameters that are different in comparison of the least squares estimates (LSE). LSE in the unstable (critical) cases, i.e. when the characteristic equation roots are lain on the unit circle, have, as a rule, compound limit distributions. Suggested by us nonstandard estimates in majority cases have more simple limit distributions.

Keywords: the first order simple autoregression, one-dimensional simultaneous autoregression, spatial autoregression of the first order, simple autoregression of the second order, parameter estimates, nonstandard approach, limit distributions.

UDC: 519.22

Received: 24.06.2011



© Steklov Math. Inst. of RAS, 2024