Upward sloping time series curve problem

malbec

Why does upward sloping time series curve imply it is not covariance stationary and any problem for this time series analysis??

WarrenB1

It's not bad if it's a linear or log-linear time series, y = b0 +b1(t) +E. That doesn't require mean reversion if I remember correctly. If it's autoregresive, it needs to have a mean reversion (and not a unit root ie b1 must be significantly different from 1)

Kapie

yes, that would be bad

redskins44

Covariance stationary means that the mean of the time series is constant. If it's sloping upwards, it's not constant. It's not so much a problem as something that can be easily remedied by differencing the time series or fitting the trend.

iteracom

You need to convert it to covariance stationary and then test for AR 1 and AR 2