Is this thing necessarily bad?? Doens’t seem like it…says “if time series are cointegrated the error term from regressing one on the other is cov. stationary and t test reliable”. is this right? And is dickey fuller used for ARCH specifically or for any type of AR model? im confused.
Cointegrated time series = good = we can use regresion
dicky fuller is used for any AR time series to check whether theres any unit root… ARCH arch use t-test…
dicky fuller is used for any AR time series to check whether theres any unit root… arch use t-test…