132
Lecture 33  Dummy variables
1) We are going to return to the study of dummy variables.
Recall that we showed that one should always
use one less variable than the number of categories one is looking at.
Recall also that the interpretation of
the dummy variable coefficient is that it measures the change from the category that has been left out to the
included category.
Finally, recall that we said we would very rarely use a dummy variable that would take
on more than 2 values.
If we have more than 2 categories, it is better to use additional dummy variables.
One simple point I would like to make here is that if we are using a multiplicative equation form and want
to convert it into a linear form using logs, and if, in addition, one of the independent variables is a dummy
variable, then we do not want to let the dummy variable take on the value zero and one since the log is not
defined for the value of zero. .
This is very easy to correct.
Simply let the dummy variable take on the
values one and e.
Then the log of the dummy variable will take on the value zero and one.
(Do you see
why?)
2) Deseasonalization
is a term that is used to correct for specific events that occur at certain times during
the year, but not at others.
For instance, consumption expenditure increases dramatically during the month
of December.
This, of course, has to do with the fact that holidays occur during this month.
So, people
will not be following their normal consumption patterns.
If we were to look at consumption during
December and generalize from what occurs during this month to the rest of the year, we would of course be
badly off in our estimates.
When we deseasonalize data, we are removing the effects of these unusual
events from the data.
Suppose that we have quarterly data.
We can deseasonalize the data by including
dummy variables for the quarterly periods.
Since there are four quarters during the year, we will include
three dummy variables.
The coefficients then tell us the change in the dependent variable in one quarter
compared to the omitted quarter.
So, if the fourth quarter is the one we omit (Since this is the quarter in
which consumption expenditure is likely to be unusual), then the coefficient on each of the remaining three
dummy variables tells us the difference in consumption expenditure between that quarter and the fourth
quarter. .
The three dummy variables we would include in the regression equation are:
X
1
=
in quarter 1, 0 otherwise
Y
B
B X
B X
X
t
t
=
+
+
+
.
ε
t
1
1
=
=
X
in quarter 2, 0 otherwise
X
1 in quarter 3, 0 otherwise
2t
3t
b) In our earlier discussion of dummy variables, we gave examples in which dummy variables were used to
change the intercept of the line.
But, dummy variables can also be used to change the slope of the line.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
This is the end of the preview.
Sign up
to
access the rest of the document.
 Spring '11
 YongJinPark
 Econometrics, Regression Analysis, dummy variables, dummy variable, independent variables, Ramsey, structural change

Click to edit the document details