The McGraw-Hill Series Economics essentials of economics brue, McConnell, and Flynn Essentials of Economics

HAC (heteroscedasticity- and autocorrelation-consistent) standard errors 
or
simply 
Newey–West standard errors.
We will not present the mathematics behind the
42
This is especially so if the regressors exhibit a trend, which is quite common in economic data.
43
W. K. Newey and K. West, “A Simple Positive Semi-Definite Heteroscedasticity and Autocorrelation
Consistent Covariance Matrix, 
Econometrica,
vol. 55, 1987, pp. 703–708.
guj75772_ch12.qxd 23/08/2008 05:20 PM Page 447

---

Newey–West procedure, for it is involved.
44
But most modern computer packages now
calculate the Newey–West standard errors. It is important to point out that the
Newey–West procedure is 
strictly speaking valid in large samples
and may not be appro-
priate in small samples. But in large samples we now have a method that produces
autocorrelation-corrected standard errors so that we do not have to worry about the EGLS
transformations discussed in the previous section. Therefore, if a sample is reasonably
large, one should use the Newey–West procedure to correct OLS standard errors not only
in situations of autocorrelation only but also in cases of heteroscedasticity, for the HAC
method can handle both, unlike the White method, which was designed specifically for
heteroscedasticity.
Once again let us return to our wages–productivity regression (12.5.1). We know that
this regression suffers from autocorrelation. Our sample of 46 observations is reasonably
large, so we can use the HAC procedure. Using 
EViews 4,
we obtain the following regres-
sion results:
ˆ
Y
t
=
32.7419
+
0.6704
X
t
se
=
(2.9162)
*
(0.0302)
*
(12.10.1)
r
2
=
0.9765
d
=
0.1719
where
*
denotes HAC standard errors.
Comparing this regression with Eq. (12.5.1), we find that in both the equations the esti-
mated coefficients and the 
r
2
value are the same. But, importantly, note that the HAC stan-
dard errors are much greater than the OLS standard errors and therefore the HAC 
t
ratios
are much smaller than the OLS 
t
ratios. This shows that OLS had in fact underestimated the
true standard errors. Curiously, the 
d
statistics in both Eqs. (12.5.1) and (12.10.1) are the
same. But don’t worry, for the HAC procedure has already taken this into account in cor-
recting the OLS standard errors.