A strong positive partial autocorrelation of lag 1 means that an observation is highly correlated with its previous observation whereas a near zero PACF indicates no correlation.

Does that mean that a near zero PACF suggests no variability within the observations while a positive PACF suggests increasing variability?

Take a person's sleep duration over a period of one month as an example. If he or she sleeps consistently close to 8 hours a day, the day-to-day variability is low and the PACF is near zero.

Are these 2 concepts similar or am I understanding them incorrectly?


1 Answer 1


They are slightly different, by variability I am guessing you refer to the variance. You can have a noise process $\epsilon_t \sim N(0, \sigma^2)$ that is independent across time and that can be highly variable depending on the variance $\sigma^2$.

A real world example would be stock market returns, they are very much volatile and they exhibit near zero PACF at the monthly level.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.