Linear Time Series Models for Stationary data - Feweb
Linear Time Series Models for Stationary data - Feweb
Linear Time Series Models for Stationary data - Feweb
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Autocovariance, -correlation, correlogram<br />
In ARIMA modelling, autocovariance is the main <strong>data</strong> characterstic to<br />
be modelled. It is not a nuisance, but the object of study. Systematic<br />
autocovariances entail <strong>for</strong>ecastability!<br />
The autocovariance function, γk, is seen as a function of lag k. One<br />
often plots the autocorrelation function which has the same shape<br />
and contains the same in<strong>for</strong>mation on the dynamics, but is<br />
dimensionless:<br />
ρk = γk<br />
γ0<br />
, k = 1, 2, . . ..<br />
Consistent estimates <strong>for</strong> weakly stationary processes are obtained by<br />
ρk = rk = γk<br />
,<br />
γ0<br />
a plot of these against k = 1, 2, . . . is known as the correlogram.<br />
Chapter 7.1 Heij et al, TI Econometrics II 2006/2007 – p. 11/24