Econometrics I

Previous class:

 Regression with time series data

 Normality assumption

 Confidence interval approach and test of significance approach of testing hypothesis

Any Question?

Today’s content:

 2-t rule of thumb (Ref. Gujrati)

 Exact level of Significance : p value (Gujrrati)

 Regression analysis and ANOVA (Gujrati and 2 different sheets provided in the google classroom)

 Reporting the Results of Regression Analysis (Gujrati)

 Restricted and unrestricted regression (Woolridge)

 Testing the linear combination (Woolridge)

 Testing multiple regression (Woolridge)

 F and t test (Woolridge)

 The “Zero” Null Hypothesis and the “2-t” Rule of Thumb

A null hypothesis that is commonly tested in empirical work is H0: β2 = 0, that is, the
slope coefficient is zero. If there is no relationship between Y and X to begin with, then
testing a hypothesis such as β2 = 0.3 or any other value is meaningless. This null
hypothesis can be easily tested by the confidence interval or the t-test approach
discussed in the preceding sections. But very often such formal testing can be shortcut
by adopting the “2-t ” rule of significance, which may be stated as

Reject the null hypothesis if |t| = ˆ β se ( ˆ β ) >

2 2 Reject the null hypothesis if |t| = ˆ β se ( ˆ β ) > tα
2 2

tα/2 for the appropriate degrees of freedom for for the appropriate degrees of freedom for one
two tail test tail test
The Exact Level of Significance: The p Value
It should be clear from the discussion so far that whether we reject or do not reject the null hypothesis
depends critically on α, the level of significance or the probability of committing a Type I error. If we try
to reduce a Type I error, a Type II error increases, and vice versa. That is, given the sample size, if we try to
reduce the probability of rejecting the true hypothesis, we at the same time increase the probability of
accepting the false hypothesis. So there is a trade-off involved between these two types of errors given
the sample size. Now the only way we can decide about the trade-off is to find out the relative costs of
the two types of errors.

Then, If the error of rejecting the null hypothesis which is in fact true (Error Type I) is costly relative to
the error of not rejecting the null hypothesis which is in fact false (Error Type II), it will be rational to set
the probability of the first kind of error low. If, on the other hand, the cost of making Error Type I is low
relative to the cost of making Error Type II, it will pay to make the probability of the first kind of error
high (thus making the probability of the second type of error low).
The Exact Level of Significance: The p Value

Once a test statistic (e.g., the t statistic) is obtained in a given example, why not
simply go to the appropriate statistical table and find out the actual probability
of obtaining a value of the test statistic as much as or greater than that obtained
in the example?

This probability is called the p value (i.e., probability value), also known as the
observed or exact level of significance or the exact probability of committing a
Type I error. More technically, the p value is defined as the lowest significance
level at which a null hypothesis can be rejected.

To illustrate, let us return to our wages-education example. Given the null

hypothesis that the true coefficient of education is 0.5, we obtained a t value of
3.2 in Eq. (5.7.4). What is the p value of obtaining a t value of as much as or
greater than 3.2? Looking up the t table given in Appendix D, we observe that
for 11 df the probability of obtaining such a t value must be smaller than 0.005
(one-tail) or 0.010 (two-tail).
REGRESSION ANALYSIS AND ANALYSIS OF VARIANCE

• TSS = ESS + RSS

• A study of these components of TSS is known as the analysis

of variance (ANOVA) from the regression viewpoint.
REGRESSION ANALYSIS AND ANALYSIS OF VARIANCE
The computed F value is seen to be 108.3026. The p value of this F statistic corresponding to 1
and 11 df cannot be obtained from the F table given in Appendix D, but by using electronic
statistical tables it can be shown that the p value is 0.0000001, an extremely small probability
indeed. If you decide to choose the level-of significance approach to hypothesis testing and fix α at
0.01, or a 1 percent level, you can see that the computed F of 108.3026 is obviously significant at
this level. Therefore, if we reject the null hypothesis that β2 = 0, the probability of committing a
Type I error is very small. For all practical purposes, our sample could not have come from a
population with zero β2 value and we can conclude with great confidence that X, education, does
affect Y, average wages.
 Reporting the Results of Regression Analysis

 The first set of parentheses are the estimated standard errors of the regression coefficients
 The figures in the second set are estimated t values computed from Eq. (5.3.2)
 The figures in the third set are the estimated p values

 Thus, for 11 df the probability of obtaining a t value of 10.3428 or greater is 0.00009,

which is practically zero. By presenting the p values of the estimated t coefficients, we can
see at once the exact level of significance of each estimated t value
 Reporting the Results of Regression Analysis

• under the null hypothesis that the true population slope value is zero, the
exact probability of obtaining a t value of 10.3428 or greater is practically
zero.

• Recall that the smaller the p value, the smaller the probability of making a
mistake if we reject the null hypothesis. Earlier we showed the intimate
connection between the F and t statistics, namely, F1,k = t2 k .

• Under the null hypothesis that the true β2 = 0, Eq. (5.11.1) shows that the
F value is 108.30 (for 1 numerator and 11 denominator df) and the t value
is about 10.34 (11 df); as expected, the former value is the square of the
latter value, except for the round off errors.
Reporting the Results of Regression Analysis
 Restricted and Unrestricted model

• The Unrestricted Model: This is the model without any of the restrictions imposed. It contains all
the variables exactly as in the original regression .
• The Restricted Model: This is the model on which the restrictions have been imposed. For
example all regressors whose coefficients have been set to zero are excluded and any other
restriction has been imposed.
• Suppose for example we estimate a model of the form
Yi  a  b1 X i 1  b2 X i 2  b3 X i 3  b4 X i 4  b5 X i 5  ui
• We may wish to test hypotheses of the form {H0: b1=0 and b2=0 against the alternative that
one or more are wrong} or
{H0: b1=1 and b2-b3=0 against the alternative that one or more are wrong} or {H0: b1+b2=1 and
a=0 against the alternative that one or more are wrong}
• This lecture is inference in this more general set up.
• We will not outline the underlying statistical theory for this. We will just describe the testing
procedure.
 Example 1: Restricted vs unrestricted regression

• Suppose we want to test that :H0: b1=0 and b2=0 against the alternative that
one or more are wrong in:
Yi  a  b1 X i1  b2 X i 2  b3 X i 3  b4 X i 4  b5 X i 5  ui Unrestricted model
Yi  a  b3 X i 3  b4 X i 4  b5 X i 5  ui Restricted Model

Example 2: Restricted vs unrestricted regression

• Suppose we want to test that : H0: b1=1 and b2-b3=0 against the alternative that one or
more are wrong :
Yi  a  b1 X i1  b2 X i 2  b3 X i 3  b4 X i 4  b5 X i 5  ui Unrestricted model

Yi  a  X i 1  b2 X i 2  b2 X i 3  b4 X i 4  b5 X i 5  ui Restricted Model

• Rearranging we get a model that uses new variables as functions of the old ones:

(Yi  X i 1 )  a  b2 ( X i 2  X i 3 )  b4 X i 4  b5 X i 5  ui
 Testing a Linear Combination
• Suppose instead of testing whether b1 is equal to a constant, you want to test if it is equal
to another parameter, that is H0 : b1 = b2
• Use same basic procedure for forming a t statistic

ˆ  ˆ
t 
se   ˆ 
1 2
ˆ 
1 2

Since

se ˆ1  ˆ 2    
Var ˆ1  ˆ 2 , then
Var  ˆ  ˆ   Var  ˆ   Var  ˆ   2Cov ˆ , ˆ 
1 2 1 2 1 2

se  ˆ  ˆ    se  ˆ     se  ˆ   2 s 
1
2 2 2
1 2 1 2 12

where s is an estimate of Cov ˆ , ˆ 

12 1 2
13
 Testing a Linear Combo (cont)
• So, to use formula, need s12, which standard output does not have
• Many packages will have an option to get it, or will just perform the test for you
• In Stata, after reg y x1 x2 … xk you would type test x1 = x2 to get a p-value for the test
• More generally, you can always restate the problem to get the test you want

• Suppose you are interested in the effect of campaign expenditures on outcomes

• Model is voteA = b0 + b1log(expendA) + b2log(expendB) + b3prtystrA + u
• H0: b1 = - b2, or H0: q1 = b1 + b2 = 0
• b1 = q1 – b2, so substitute in and rearrange  voteA = b0 + q1log(expendA) + b2log(expendB - expendA)
+ b3prtystrA + u

• This is the same model as originally, but now you get a standard error for b1 – b2 = q1 directly from the
basic regression
• Any linear combination of parameters could be tested in a similar manner
• Other examples of hypotheses about a single linear combination of parameters:
• b1 = 1 + b2 ; b1 = 5b2 ; b1 = -1/2b2 ; etc 14
 Testing Multiple Linear Restrictions: (Exclusion)

• Everything we’ve done so far has involved testing a single linear restriction, (e.g. b1 = 0
or b1 = b2 )
• However, we may want to jointly test multiple hypotheses about our parameters
• A typical example is testing “exclusion restrictions” – we want to know if a group of
parameters are all equal to zero

Now the null hypothesis might be something like H 0: bk-q+1 = 0, ... , bk = 0

The alternative is just H1: H0 is not true. Can’t just check each t statistic separately, because we
want to know if the q parameters are jointly significant at a given level – it is possible for none to
be individually significant at that level.
A particular t statistic tests a hypothesis that puts no restrictions on the other parameters.
Besides, we would have three outcomes to contend with—one for each t statistic 15
Testing Multiple Linear Restrictions: (Exclusion)
 Testing Multiple Linear Restrictions: (Exclusion)

• To do the test we need to estimate the “restricted model” without xk-q+1,,

…, xk included, as well as the “unrestricted model” with all x’s included
• Intuitively, we want to know if the change in SSR is big enough to
warrant inclusion of xk-q+1,, …, xk

F
 SSRr  SSRur  q
, where
SSRur  n  k  1
r is restricted and ur is unrestrict ed
Economics 20 - Prof. Anderson 17
 The F statistic

• The F statistic is always positive, since the SSR from the restricted model can’t be less than the SSR
from the unrestricted
• Essentially the F statistic is measuring the relative increase in SSR when moving from the unrestricted
to restricted model
• q = number of restrictions, or dfr – dfur
• n – k – 1 = dfur

To decide if the increase in SSR when we move to a restricted model is “big enough” to
reject the exclusions, we need to know about the sampling distribution of our F stat
Not surprisingly, F ~ Fq,n-k-1, where q is referred to as the numerator degrees of freedom
and n – k – 1 as the denominator degrees of freedom.

18
The F statistic (cont)
f(F) Reject H0 at a
fail to reject significance level
if F > c

reject
(1 -a) a
0 c F
Economics 20 - Prof. Anderson 19
 Relationship between F and t Statistics

What happens if we apply the F statistic to the case of testing significance of a single
independent variable
we know that the t statistic on bk can be used to test this hypothesis. The question, then, is: do
we have two separate ways of testing hypotheses about a single coefficient? The answer is no.

It can be shown that the F statistic for testing exclusion of a single variable is equal to the square
of the corresponding t statistic. Since t2 n2k21 has an F1,n2k21 distribution, the two approaches
lead to exactly the same outcome, provided that the alternative is two-sided. The t statistic is
more flexible for testing a single hypothesis because it can be directly used to test against one-
sided alternatives. Since t statistics are also easier to obtain than F statistics, there is really no
reason to use an F statistic to test hypotheses about a single parameter.
We have already seen in the salary regressions for major league baseball players that two (or
more) variables that each have insignificant t statistics can be jointly very significant. It is also
possible that, in a group of several explanatory variables, one variable has a significant t statistic
but the group of variables is jointly insignificant at the usual significance levels.
 Relationship between F and t Statistics

For concreteness, suppose that in a model with many explanatory variables we cannot
reject the null hypothesis that b1, b2, b3, b4, and b5 are all equal to zero at the 5% level,
yet the t statistic for b ˆ 1 is significant at the 5% level. Logically, we cannot have b1 = 0
but also have b1, b2, b3, b4, and b5 all equal to zero! But as a matter of testing, it is
possible that we can group a bunch of insignificant variables with a significant variable
and conclude that the entire set of variables is jointly insignificant. (Such possible
conflicts between a t test and a joint F test give another example of why we should not
“accept” null hypotheses; we should only fail to reject them.)

The F statistic is intended to detect whether a set of coefficients is different from zero,
but it is never the best test for determining whether a single coefficient is different
from zero. The t test is best suited for testing a single hypothesis. (In statistical terms,
an F statistic for joint restrictions including b1 = 0 will have less power for detecting
b1 = 0 than the usual t statistic. See Section C.6 in Appendix C for a discussion of the
power of a test.
The R2 form of the F statistic

• Because the SSR’s may be large and unwieldy, an alternative form of

the formula is useful
• We use the fact that SSR = SST(1 – R2) for any regression, so can
substitute in for SSRu and SSRur

F
R
R q 2
ur
2
r 
, where again
 
1  R  n  k  1
2
ur

r is restricted and ur is unrestrict ed

23