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

FIGURE 10.4
Scatterplot for
Example 10.2 data.
40,000
0
20,000
2000 4000 6000
0
–10
–5
5
0
0
4000
2000
6000
0
C
Yd
W
I
2000 4000 6000
–10
–5
0
5
0
4000
2000
6000
40,000
20,000
0
guj75772_ch10.qxd 12/08/2008 02:45 PM Page 341

---

342
Part Two
Relaxing the Assumptions of the Classical Model
10.8
Remedial Measures
What can be done if multicollinearity is serious? We have two choices: (1) do nothing or
(2) follow some rules of thumb.
Do Nothing
The “do nothing” school of thought is expressed by Blanchard as follows:
28
When students run their first ordinary least squares (OLS) regression, the first problem that they
usually encounter is that of multicollinearity. Many of them conclude that there is something
wrong with OLS; some resort to new and often creative techniques to get around the problem.
But, we tell them, this is wrong. Multicollinearity is God’s will, not a problem with OLS or
statistical technique in general.
What Blanchard is saying is that multicollinearity is essentially a data deficiency prob-
lem (micronumerosity, again) and sometimes we have no choice over the data we have
available for empirical analysis.
Also, it is not that all the coefficients in a regression model are statistically insignificant.
Moreover, even if we cannot estimate one or more regression coefficients with greater pre-
cision, a linear combination of them (i.e., estimable function) can be estimated relatively
efficiently. As we saw in Eq. (10.2.3), we can estimate 
α
uniquely, even if we cannot esti-
mate its two components given there individually. Sometimes this is the best we can do with
a given set of data.
29