Date post: | 14-Apr-2018 |
Category: |
Documents |
Upload: | muhendis8900 |
View: | 216 times |
Download: | 0 times |
of 51
7/30/2019 Econometrics_ch6
1/51
405ECONOMETRICSChapter # 5: TWO-VARIABLE REGRESSION:
INTERVAL ESTIMATION AND HYPOTHESIS TESTING
Domodar N. Gujarati
Prof. M. El-Sakka
Dept of Economics. Kuwait University
7/30/2019 Econometrics_ch6
2/51
Introduction
The theory of estimation consists of two parts: point estimation and interval
estimation. We have discussed point estimation thoroughly in the previoustwo chapters. In this chapter we first consider interval estimation and then
take up the topic of hypothesis testing, a topic related to interval estimation.
7/30/2019 Econometrics_ch6
3/51
INTERVAL ESTIMATION: SOME BASIC IDEAS
Look at the estimated MPC in Yi= 24.4545 + 0.5091Xi, which is a single
(point) estimateof the unknown population MPC2. How reliable is thisestimate? A single estimate is likely to differ from the true value, although
in repeated sampl ing its mean value is expected to be equal to the true value.
In statistics the reliability of a point estimator is measured by its standard
error. Therefore, we may construct an interval around the point estimator,
say within two or three standard errors on either sideof the point estimator,such that this interval has, say, 95 percent probability of including the true
parameter value.
Assume that we want to find out how close is, say,2to 2. We try to find
out two positive numbers and , the latter lying between 0 and 1, such that
the probability that the random interval (2 , 2+ )contains the true 2is1 . Symbolical ly,
Pr (2 2 2+ ) = 1 (5.2.1)
Such an interval, if it exists, is known as a confidence interval.
7/30/2019 Econometrics_ch6
4/51
1 isknown as the confidence coefficient; and
(0 < < 1) is known as thelevel of significance.
The endpoints of the confidence interval are known as the conf idence limi ts
(also known as critical values), 2 being the lower confidence limit and
2+ the upper confidence limit.
If = 0.05, or 5 percent, (5.2.1) would read: The probability that the
(random) interval shown there includes the true2is0.95, or 95 percent.
The interval estimator thus gives a range of values within which the true2
may lie.
7/30/2019 Econometrics_ch6
5/51
It is very important to know the following aspects of interval estimation:
1. Equation (5.2.1) does not say that the probability of2lying betweenthegiven limits is1 . What (5.2.1) states is that the probability of
constructing an interval that contains2is1 .
2. The interval (5.2.1) is a random interval; that is, it will vary from one
sample to the next because it is based on2, which is random.
3. (5.2.1) means: If in repeated sampling confidence intervals like it areconstructed a great many times on the 1 probability basis, then, in the
long run, on the average, such intervals wil l enclose in 1 of the cases the
true value of the parameter.
4. The interval (5.2.1) is random so long as2is not known. Once2is
known, the interval (5.2.1) is no longer random;it isfixed. I nthis situation2is either in the f ixed interval or outside it. Therefore, theprobability is
either 1 or 0, if the 95% confidence interval were obtained as (0.4268 2
0.5914), we cannot say the probabil i ty is 95% that this interval includes the
true2. That probabil i ty is either 1 or 0.
7/30/2019 Econometrics_ch6
6/51
CONFIDENCE INTERVALS FOR REGRESSION
COEFFICIENTS1AND2
Confidence Interval for2
With the normality assumption for ui, the OLS estimators1and 2arethemselvesnormally distr ibutedwith means and variances given therein.
Therefore, for example, the Variable
is a standardized normal variable. Therefore, we can use the normal
distribution to make probabilistic statements about 2. If 2
is known, animportant property of a normally distributed variable with mean and
variance 2is that the area under the normal curve between is about
68 percent, that between the limits 2is about 95 percent, and that
between 3is about 99.7 percent.
7/30/2019 Econometrics_ch6
7/51
But 2is rarely known, and in practice it is determined by the unbiased
estimator 2. If we replace by , (5.3.1) may be written as:
where the se (2)now refers to the estimated standard error. Therefore,
instead of using the normal distribution, we can use thet distr ibutionto
establish a confidence interval for2as follows:
Pr (t/2 t t/2) = 1 (5.3.3) where t/2 is the value of thetvar iable obtained from thetdistribution for /2
level of signi f icanceand n 2 df; i t is often called thecritical t valueat/2
level of signi f icance.
7/30/2019 Econometrics_ch6
8/51
Substituti on of (5.3.2) into (5.3.3)Yields
Rearranging (5.3.4), we obtain
Pr [2 t/2 se (2) 2 2+ t/2 se (2)] = 1 (5.3.5) Equation (5.3.5) provides a 100(1 ) percent confidence interval for 2,
which can be written more compactly as
100(1 )% confidence interval for 2:
2 t/2 se (2) (5.3.6)
Arguing analogously, and using (4.3.1) and (4.3.2), we can then write:
Pr [1 t/2 se (1) 1 1+ t/2 se (1)] = 1 (5.3.7)
or, more compactly,
100(1 )% confidence interval for 1:
1 t/2 se (1) (5.3.8)
7/30/2019 Econometrics_ch6
9/51
Notice an important feature of the confidence intervals given in (5.3.6) and
(5.3.8): In both cases thewidth of the conf idence interval is proportional to thestandard error of the estimator. That is, the larger the standard error, the
larger is the width of the confidence interval. Put differently, the larger the
standard error of the estimator, the greater is the uncertainty of estimating
the true value of the unknown parameter.
7/30/2019 Econometrics_ch6
10/51
We found that2= 0.5091, se (2)= 0.0357, anddf= 8. I f weassume =
5%, that i s, 95% conf idence coeff icient, then thettable showsthat for 8 df thecritical t/2 =t0.025= 2.306. Substi tuting these values in(5.3.5), the 95%
confidence interval for2is asfollows:
0.4268 2 0.5914 (5.3.9)
Or, using (5.3.6), it is
0.5091 2.306(0.0357)
that is,
0.5091 0.0823 (5.3.10)
The interpretation of this confidence interval is: Given the confidence
coefficient of 95%, in 95 out of 100 cases intervals like (0.4268, 0.5914) will
contain the true2. But, we cannot saythat the probability is 95 percent that
the specific interval (0.4268 to 0.5914) contains the true2because this
intervalis now fixed; therefore,2either l ies in it or does not.
7/30/2019 Econometrics_ch6
11/51
Confidence Interval for1
Following (5.3.7), we can verify that the 95% confidence interval for1ofthe consumptionincome example is
9.6643 1 39.2448 (5.3.11)
Or, using (5.3.8), we find it is
24.4545 2.306(6.4138)
that is,
24.4545 14.7902 (5.3.12)
In the long run, in 95 out of 100 cases intervals like (5.3.11) will contain the
true1; again the probabil ity that this particular f ixed interval includes the
true1is either 1 or 0.
7/30/2019 Econometrics_ch6
12/51
CONFIDENCE INTERVAL FOR2
As pointed before, under the normality assumption, the variable
2= (n 2) 2/2 (5.4.1)
follows the2distribution withn 2 df. Therefore, we can use the2
distributionto establish a confidence interval for 2
Pr (21/2 2 2/2 )= 1 (5.4.2)
Where21/2and2/2 are two cri tical values of
2obtainedfrom the chi-
square table for n 2 df.
Substituting2f rom (5.4.1) into (5.4.2) and rearr anging the terms, weobtain
which gives the 100(1 )% confidence interval for 2.
7/30/2019 Econometrics_ch6
13/51
To illustrate, we obtain 2= 42.1591and df = 8. If is chosen at5 percent,
the chi -square tablefor 8 df gives the following critical values: 20.025= 17.5346, and
20.975= 2.1797.
These values show that the probabil ity of a chi -square value exceeding
17.5346 is 2.5 percent and that of2.1797 is 97.5 percent. Therefore, the
interval between these two values is the 95% confidence interval for2, as
shown diagrammatically in Figure 5.1. Substituting the data of our example into (5.4.3 the 95% confidence interval
for 2is as follows:
19.2347 2154.7336 (5.4.4)
The interpretation of this interval is: If we establish 95% confidence limits
on 2and if we maintain a priori that these limits will include true 2, we shall
be right in the long run 95 percent of the time.
7/30/2019 Econometrics_ch6
14/51
(Note the skewed characteristic of the chi-square distribution.)
7/30/2019 Econometrics_ch6
15/51
HYPOTHESIS TESTING: THE CONFIDENCE-INTERVAL
APPROACH
Two-Sided or Two-Tail Test
To illustrate the confidence-interval approach, look at the consumptionincome example, the estimated (MPC),2, is 0.5091. Suppose we postulate
that:
H0:2= 0.3 and H1:2 0.3
that is, the true MPC is 0.3 under the null hypothesis but i t is less than or
greater than0.3 under the alternative hypothesis. The alternative hypothesisis a two-sided hypothesis. It reflects the fact that we do not have a strong
expectationabout the direction in which the alternative hypothesis should
move from the null hypothesis.
Is the observed2compatible with H0? To answer this question, let us refer
to the confidence interval (5.3.9). We know that in the long run intervalslike (0.4268, 0.5914) will contain the true2with 95 percent probabil i ty.
7/30/2019 Econometrics_ch6
16/51
Consequently, in the long run such intervals provide a range or limits within
which the true2may lie with a conf idence coeff icientof, say, 95%. Therefore, if2underH0falls with in the 100(1 )% confidence interval, we
do not reject the null hypothesis; if it lies outside the interval, we may reject it.
This range is illustrated schematically in Figure 5.2.
Decision Rule: Construct a 100(1 )% confidence interval for 2. If the 2
under H0falls within this conf idence interval, do not reject H0, but if it fallsoutside this interval, reject H0.
Following this rule, H0:2= 0.3clearly l iesoutside the 95% confidence
interval given in (5.3.9). Therefore, we can reject the hypothesis that the
true MPC is 0.3, with 95% confidence.
7/30/2019 Econometrics_ch6
17/51
7/30/2019 Econometrics_ch6
18/51
In statistics, when we reject the null hypothesis, we say that our finding is
statistically signif icant. On the other hand, when we do not reject the nullhypothesis, we say that our finding is not statistically significant.
One-Sided or One-Tail Test
I f we have a strong theoretical expectationthat the alternative hypothesis is
one-sided or unidirectional rather than two-sided. Thus, for our
consumptionincome example, one could postulate that H0: 20.3 and H1: 2> 0.3
Perhaps economic theory or prior empirical work suggests that the
marginal propensity to consume is greater than 0.3. Although the procedure
to test this hypothesis can be easily derived from (5.3.5), the actual
mechanics are better explained in terms of the test-of-significance approachdiscussed next.
HYPOTHESIS TESTING
7/30/2019 Econometrics_ch6
19/51
HYPOTHESIS TESTING:
THE TEST-OF-SIGNIFICANCE APPROACH
Testing the Significance of Regression Coefficients: The t Test
An alternati ve to the conf idence-intervalmethod is the test-of-significanceapproach. It is a procedure by which sample results are used to verify the
truth or falsity of a null hypothesis. The decision to accept or reject H0is
made on the basis of the valueof the test statistic obtained from the data at
hand. Recall
follows the t distribution with n 2 df.
7/30/2019 Econometrics_ch6
20/51
The confidence-interval statementssuch as the following can be made:
Pr [t/2(2 *2)/se (2) t/2]= 1 (5.7.1) where*2 is the value of2under H0. Rearranging (5.7.1), we obtain
Pr [*2 t/2se (2) 2 *2+ t/2se (2)] = 1 (5.7.2)
which gives the confidence interval in with 1 probability.
(5.7.2) is known as the region of acceptanceand the region(s) outside the
conf idence interval is (are)called the region(s) of rejection (ofH0) or the
cri tical region(s).
Using consumptionincome example. We know that2= 0.5091, se (2) =
0.0357, and df = 8. If we assume = 5 percent, t/2= 2.306. I f we let:
H0: 2= *2 = 0.3 and H1: 2 0.3, (5.7.2) becomes Pr (0.2177 20.3823) = 0.95 (5.7.3)
Since the observed2l ies in thecritical region, we reject the null hypothesis
that true2= 0.3.
7/30/2019 Econometrics_ch6
21/51
7/30/2019 Econometrics_ch6
22/51
In practice, there is no need to estimate (5.7.2) explicitly. One can compute
the t value in the middle of the double inequal ity given by (5.7.1)andseewhether it lies between the critical t values or outside them. For ourexample,
t = (0.5091 (0.3)) / (0.0357) = 5.86 (5.7.4)
which clearly lies in the critical region of Figure 5.4. The conclusion remains
the same; namely, we reject H0.
A statistic is said to be statistically signi f icant if the value of the test statistic
lies in the cr itical region. By the same token, a test is said to be statistically
insignif icant if the value of the test statistic lies in the acceptance region.
We can summarize the ttest of significance approach to hypothesis testing
as shown in Table 5.1.
7/30/2019 Econometrics_ch6
23/51
7/30/2019 Econometrics_ch6
24/51
7/30/2019 Econometrics_ch6
25/51
Testing the Significance of2: The 2Test
As another illustration of the test-of-significance methodology, consider thefollowing variable:
2=n 2 (2/2) (5.4.1)
which, as noted previously, follows the2distribution with n 2 df. For the
hypothetical example, 2= 42.1591 and df = 8. I f we postulate that
H0: 2= 85 vs. H1:
2 85, Eq.
(5.4.1) provides the test stati stic for H0. Substituti ngthe appropriate values in
(5.4.1), it can be found that under H0, 2= 3.97. I fwe assume = 5%, the
critical 2values are 2.1797 and 17.5346.
Since thecomputed2l ies between these limi ts, we do not reject H0
. The2
test of signif icance approach tohypothesis testing is summarized in Table
5.2.
7/30/2019 Econometrics_ch6
26/51
7/30/2019 Econometrics_ch6
27/51
HYPOTHESIS TESTING: SOME PRACTICAL ASPECTS
The Meaning of Accepting or Rejecting a Hypothesis
If on the basis of a test of significance in accepting H0, do not say we acceptH0. Why? To answer this, let us revert to our consumptionincome example
and assume that H0: 2(MPC) = 0.50. Nowthe estimated value of the MPC
is2= 0.5091with a se (2) = 0.0357. Thenon the basis of the t test we find
that t = (0.5091 0.50)/0.0357 =0.25,which is insignificant, say, at = 5%.
Therefore, we say accept H0. Butnow let us assume H0: 2= 0.48. Applyingthe t test, we obtain t = (0.5091 0.48)/0.0357 =0.82, which too is statistically
insigni f icant. So now we sayaccept this H0. Which of these two null
hypotheses is the truth? We do not know. It is therefore preferable to say
do not reject rather than accept.
7/30/2019 Econometrics_ch6
28/51
The Zero Null Hypothesis and the 2-t Rule of Thumb
A null hypothesis that is commonly tested in empirical work is H0: 2= 0,that is, the slope coefficient is zero.
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
7/30/2019 Econometrics_ch6
29/51
The rationale for this rule is not too difficult to grasp. From (5.7.1) we know
that we will reject H0: 2= 0 if t =2/ se (2) > t/2when2> 0
or
t =2/ se (2) < t/2when2< 0
or when
|t| =|2/se (2)| > t/2 (5.8.1)
for the appropriate degrees of freedom.
Now if we examine the t, we see that for df ofabout 20 or more a computed t
value in excess of 2 (in absolute terms), say, 2.1, is statistically significant at
the 5 percent level, implying rejection of the null hypothesis.
7/30/2019 Econometrics_ch6
30/51
Forming the Null and Alternative Hypotheses
Very often the phenomenon under study will suggest the nature of the nulland alternative hypotheses. Consider the capital market line (CML) of
portfolio theory,
Ei= 1+ 2i,
where E = expected return onportfolio and = the standard deviation of
return, a measure of r isk. Sincereturn and risk are expected to be positivelyrelated, natural alternative hypothesis to the null hypothesis that2= 0
would be 2> 0. That is, one would not choose to consider valuesof2less
than zero.
Consider the case of the demand for money. Studies have shown that the
income elasticity of demand for money has typically ranged between 0.7 and1.3. Therefore, in a new study of demand for money, if one postulates that
the income-elasticity coefficient2is 1, the alternati ve hypothesis couldbe
that21, a two-sided alternative hypothesis.
Thus, theoretical expectations or prior empirical work or both can be relied
upon to formulate hypotheses.
7/30/2019 Econometrics_ch6
31/51
Choosing , the Level of Significance
To reject or not reject the H0 depends critically on , or the probabil i ty ofcommitting aType I errortheprobabil i ty of rejectingthe true hypothesis.
But even then, why is commonly fixed at the 1, 5, or at the most 10 percent
levels? For a given sample size, if we try to reduce a Type I error, a Type I I
error increases, and viceversa. That is, given the sample size, if we try to
reduce the probability of rejecting the true hypothesis, we at the same timeincrease the probability of accepting the false hypothesis.
Now the only way we can decide about the tradeoff is to find out the relative
costs of the two types of errors.
Applied econometricians generally follow the practice of setting the value of
at a 1 or a 5 or at most a 10 percent level and choose a teststatistic thatwould make the probability of committing a Type II error as small as
possible. Since one minus the probability of committing a Type II error is
known as the power of the test, this procedure amounts to maximizing the
power of the test.
7/30/2019 Econometrics_ch6
32/51
The Exact Level of Significance: The p Value
Once a test stati sticis obtained in a given example, why not simply go to theappropriate statistical table and find out the actual probability of obtaininga value of the test statistic as much as or greater than that obtained in theexample? This probability is called the p value (i.e., probabili ty value), alsoknown as theobserved or exact level of significance or the exact probabilityof committing a Type I error. More technically, the p value is def ined as the
lowest signi f icance level at which a nul l hypothesis can be rejected. To illustrate, given H0 that the true MPC is 0.3, we obtained a t value of 5.86
in(5.7.4). What is the p value of obtaining a t value of as much as or greaterthan 5.86? Looking up the t table, for8 df the probability of obtaining sucha t value must be much smaller than0.001 (one-tail) or 0.002 (two-tail). Thisobserved, or exact, level of significance of the t statistic is much smaller than
the conventionally, and arbitrarily, fixed level of significance, such as 1, 5,or 10 percent. If we were to use the p value just computed, and reject the nullhypothesis, the probability of our committing a Type I error is only about0.02 percent, that is, only about 2 in 10,000!
7/30/2019 Econometrics_ch6
33/51
REGRESSION ANALYSIS AND ANALYSIS OF VARIANCE
In Chapter 3, the following identity is developed:
y2i=y2i+u2i=22x2i+u2i (3.5.2) that is, TSS = ESS + RSS. A study of these components of TSS is known as
the analysis of variance (ANOVA).
Associated with any sum of squares is its df, the number of independent
observations on which it is based. TSS has n 1dfbecause we lose 1 df in
computing the sample mean Y . RSS hasn 2df.
ESS has 1 df which follows from the fact that
ESS =22x2iis a function of
2only, since x2iis known.
Let us arrange the various sums of squares and their associated dfin Table
5.3, which is the standard form of the AOV table, sometimes called the
ANOVA table. Given the entries of Table 5.3, we now consider the following
variable:
7/30/2019 Econometrics_ch6
34/51
F= (MSS of ESS) / (MSS of RSS)
=22x2i/(u2i/(n 2)) (5.9.1) =22x
2i/
2
If we assume that the disturbances uiare normally distributed, and ifH0 is
that2 = 0, then it can be shown that the Fvariable of (5.9.1) follows the F
distribution with 1 dfin the numerator and (n 2) dfin the denominator.
7/30/2019 Econometrics_ch6
35/51
What use can be made of the preceding Fratio? It can be shown that
E22x2i= 2+ 22x2i (5.9.2)
and
Eu2i/ n2= E(2) = 2 (5.9.3)
(Note that2 and 2 appearing on the right sides of these equations are the
true parameters.) Therefore, if2 is in fact zero, Eqs. (5.9.2) and (5.9.3) both
provide us with identical estimates of true 2. In this situation, theexplanatory variable Xhas no linear influence on Ywhatsoever and the
entire variation in Yis explained by the random disturbances ui. If, on the
other hand,2 is not zero, (5.9.2) and (5.9.3) will be different and part of the
variation in Ywill be ascribable to X. Therefore, the Fratio of (5.9.1)
provides a test of the null hypothesis H0:2 = 0.
7/30/2019 Econometrics_ch6
36/51
To illustrate, the ANOVA table for consumptionincome is as shown in
Table 5.4. The computed Fvalue is seen to be 202.87. The pvalue of this Fstatistic using electronic statistical tables is 0.0000001, the computed Fof
202.87 is obviously significant at this level. Therefore, reject the null
hypothesis that2 = 0.
REPORTING THE RESULTS OF REGRESSION ANALYSIS
7/30/2019 Econometrics_ch6
37/51
REPORTING THE RESULTS OF REGRESSION ANALYSIS
Employing the consumptionincome example of as an illustration:
Yi= 24.4545 + 0.5091Xi
se = (6.4138) (0.0357) r2 = 0.9621 (5.11.1)
t = (3.8128) (14.2605) df = 8
p = (0.002571) (0.000000289) F1,8 = 202.87
Thus, for 8 df the probability of obtaining a tvalue of 3.8128 or greater is
0.0026 and the probability of obtaining a tvalue of 14.2605 or larger is
about 0.0000003.
Under the null hypothesis that the true population intercept value is zero,
thep
value of obtaining at
value of 3.8128 or greater is only about 0.0026.Therefore, if we reject this null hypothesis, the probability of our
committing a Type I error is about 26 in 10,000. The true population
intercept is different from zero.
7/30/2019 Econometrics_ch6
38/51
If the true MPC were in fact zero, our chances of obtaining an MPC of
0.5091 would be practically zero. Hence we can reject the null hypothesisthat the true MPC is zero.
Earlier we showed the intimate connection between the Fand tstatistics,
namely, F1, k= t2k. Under the null hypothesis that the true2 = 0, the F
value is 202.87, and the tvalue is about 14.24, as expected, the former value
is the square of the latter value.
EVALUATING THE RESULTS OF REGRESSION ANALYSIS
7/30/2019 Econometrics_ch6
39/51
EVALUATING THE RESULTS OF REGRESSION ANALYSIS
How good is the fitted model? We need some criteria with which to
answer this question. First, are the signs of the estimated coefficients in accordance with
theoretical or prior expectations? e.g., the income consumption model
should be positive.
Second, if theory says that the relationship should also be statistically
significant. The pvalue of the estimated tvalue is extremely small. Third, how well does the regression model explain variation in the
consumption expenditure? One can use r2 to answer this question, which is
a very high.
7/30/2019 Econometrics_ch6
40/51
There is one assumption that we would like to check, the normality of the
disturbance term, ui. Normality Tests
Although several tests of normality are discussed in the literature, we will
consider just three:
(1) histogram of residuals;
(2) normal probability plot (NPP), a graphical device; and
(3) the JarqueBera test.
7/30/2019 Econometrics_ch6
41/51
Histogram of Residuals.
A histogram of residuals is a simple graphic device that is used to learnsomething about the shape of the probability density function PDF of a
random variable.
If you mentally superimpose the bell shaped normal distribution curve on
the histogram, you will get some idea as to whether normal (PDF)
approximation may be appropriate.
7/30/2019 Econometrics_ch6
42/51
Normal Probability Plot.
A comparatively simple graphical device is the normal probability plot(NPP). If the variable is in fact from the normal population, the NPP will be
approximately a straight line. The NPP is shown in Figure 5.7. We see that
residuals from our illustrative example are approximately normally
distributed, because a straight line seems to fit the data reasonably well.
7/30/2019 Econometrics_ch6
43/51
7/30/2019 Econometrics_ch6
44/51
JarqueBera (JB) Test of Normality.
The JB test of normality is an asymptotic, or large-sample, test. It is alsobased on the OLS residuals. This test first computes the skewness and
kurtosis, measures of the OLS residuals and uses the following test statistic:
JB = n[S2/ 6+ (K 3)2/ 24] (5.12.1)
where n= sample size, S= skewness coefficient, and K= kurtosis coefficient.
For a normally distributed variable, S= 0 and K= 3. In that case the valueof the JB statistic is expected to be 0.
7/30/2019 Econometrics_ch6
45/51
The JB statistic follows the chi-square distr ibution with 2 df.
If the computed pvalue of the JB statistic in an application is sufficientlylow, which will happen if the value of the statistic is very different from 0,
one can reject the hypothesis that the residuals are normally distr ibuted. But if
the pvalue is reasonably high, which will happen if the value of the statistic
is close to zero, we do not reject the normality assumption.
The sample size in our consumptionincome example is rather small. If wemechanically apply the JB formula to our example, the JB statistic turns
out to be 0.7769. The pvalue of obtaining such a value from the chi-square
distribution with 2 df is about 0.68, which is quite high. In other words, we
may not reject the normality assumption for our example. Of course, bear
in mind the warning about the sample size.
A CONCLUDING EXAMPLE
7/30/2019 Econometrics_ch6
46/51
A CONCLUDING EXAMPLE
Let us return to Example 3.2 about food expenditure in India. Using the
data given in (3.7.2) and adopting the format of (5.11.1), we obtain thefollowing expenditure equation:
FoodExpi= 94.2087 + 0.4368 TotalExpi
se = (50.8563) (0.0783)
t= (1.8524) (5.5770)
p= (0.0695) (0.0000)*
r2 = 0.3698; df = 53
F1,53 = 31.1034 (pvalue = 0.0000)*
7/30/2019 Econometrics_ch6
47/51
As expected, there is a positive relationship between expenditure on food
and total expenditure. If total expenditure went up by a rupee, on average,expenditure on food increased by about 44 paise.
If total expenditure were zero, the average expenditure on food would be
about 94 rupees.
The r2 value of about 0.37 means that 37 percent of the variation in food
expenditure is explained by total expenditure, a proxy for income. Suppose we want to test the null hypothesis that there is no relationship
between food expenditure and total expenditure, that is, the true slope
coefficient2 = 0.
7/30/2019 Econometrics_ch6
48/51
The estimated value of2 is 0.4368. If the null hypothesis were true, what is
the probability of obtaining a value of 0.4368? Under the null hypothesis, weobserve from (5.12.2) that the tvalue is 5.5770 and the pvalue of obtaining
such a tvalue is practically zero. In other words, we can reject the null
hypothesis. But suppose the null hypothesis were that2 = 0.5. Now what?
Using the ttest we obtain:
t= 0.4368
0.5 / 0.0783
= 0.8071
The probability of obtaining a |t| of 0.8071 is greater than 20 percent. Hence
we do not reject the hypothesis that the true2 is 0.5.
7/30/2019 Econometrics_ch6
49/51
Notice that, under the null hypothesis, the true slope coefficient is zero.
The Fvalue is 31.1034. Under the same null hypothesis, we obtained a tvalue of 5.5770. If we square this value, we obtain 31.1029, which is about
the same as the Fvalue, again showing the close relationship between the t
and the Fstatistic.
Using the estimated residuals from the regression, what can we say about
the probability distribution of the error term? The information is given inFigure 5.8. As the figure shows, the residuals from the food expenditure
regression seem to be symmetrically distributed.
Application of the JarqueBera test shows that the JB statistic is about
0.2576, and the probability of obtaining such a statistic under the normality
assumption is about 88 percent. Therefore, we do not reject the hypothesisthat the error terms are normally distributed. But keep in mind that the
sample size of 55 observations may not be large enough.
7/30/2019 Econometrics_ch6
50/51
7/30/2019 Econometrics_ch6
51/51