lecture 13 Flashcards
what is a formal test for autocorrelation, not a correlogram?
The Durbin watson test
how do you know id there is positve autocorrelation or negative
The Durbin-Watson tests produces a test statistic that ranges from 0 to 4.
Values close to 2 (the middle of the range) suggest less autocorrelation,
values closer to 0 or 4 indicate greater positive or negative autocorrelation respectively.
when do you use the Box Ljung test
when you have many circumstances of autocorrelated values
what is a lagged dependant variable
agged Dependent Variable. A dependent variable that is lagged in time. For example, if Yt is the dependent variable, then Yt-1 will be a lagged dependent variable with a lag of one period
when you have a lagged variable why can’t you use Durbin Watson
it will always tend to 2, so a Dubin H test should be considered
the coursework likely has lagged variables
- so you must use Durbin H test
what is row ‘p’
the correlation coefficient
how do you estimate ‘p’
covarience of ut, ut-1 over variance of that squared
what is autocorrelation
degree of correlation between the values of the same variables across different observations in the data.
what is the durbin H stat defined as
row_hat sqrt( T/(1-T*var(bhat)))
what is 1st order autocorrelation
first-order autocorrelation, occurs when the consecutive errors are correlated
so if the data at 2017 and 2018 are correlated it is auto correlation.
however if the data at 2003 and 2009 are correlated it is 2nd order
why do you use the breauch godfrey test
to find 2nd order auto-correlation
how do you use the breusch godfrey test
- estimate the model and generate a set of residuals
- estimate the auxiullary regression –> regress orginal regressor on their lagged values
- carry out the F test or the chi squared for the significane of lagged residuals
what are the effect of autocorrelated errors on the OLS estimation
OLS is no longer the best linear unbiased estiamtor, they may be more efficient
OLS is still unbiased
what are the effect of autocorrelated errors on coefficent standard errors
Hypothesis tests based on the OLS standard errors will be
unreliable when there are autocorrelated errors.
how do caculate the autocorrelation efficient, whats the formula
using DW
row_hat = 1 - (DW/2)
