UNIT 14 TESTS FOR SPECIFICATION

ERROR

Structure
14.1 Introduction
14.2 Objectives
14.3 Tests for Identifying the Most Efficient Model
14.3.1 The 𝑅 Test and Adjusted 𝑅 Test

14.3.2 Akaike Information Criterion

14.3.3 Schwarz Information Criterion

14.3.4 Mallow’s 𝐶 Criterion

14.4 Caution about Model Selection Criteria

14.5 Let Us Sum Up
14.6 Answers to Check Your Progress Exercises

14.1 INTRODUCTION
In the previous Unit we highlighted the consequences of specification errors.
There could be three types of specification errors; inclusion of an irrelevant
variable, exclusion of a relevant variable, and incorrect functional form. When
the econometric model is not specified correctly, the coefficient estimates, the
confidence intervals, and the hypothesis tests are misleading and inconsistent. In
view of this, econometric models should be correctly specified.

While building a model we face a lot of difficulties in specifying a model

correctly. In some cases economic theory is quite transparent about the dependent
variables and the independent variables. In some other cases still it is in a
hypothesis stage. Researchers are still working in that area to confirm the
hypothesis suggested by others. In such cases, what we have a dependent variable
and a set of explanatory variables. Out of these explanatory variables we have to
select the most appropriate ones.


Dr. Sahba Fatima, Independent Researcher, Lucknow.
Econometric theory suggests certain criteria and test statistics. On the basis of Tests for Specification
Error
these criteria we select the most appropriate econometric model. We describe
some of these criteria below.

14.2 OBJECTIVES
After going through this Unit, you should be in a position to
 identify econometric models that are not specified correctly;
 take remedial measures for correcting the specification error; and
 evaluate the performance of competing models.

14.3 TESTS FOR IDENTIFYING THE MOST

EFFICIENT MODEL
As pointed out above, econometric models should be specified correctly. Any
spurious relationship should be identified and excluded from the model. There
are certain tests for this purpose. These tests can be used under specific
circumstances in conjunction with practical understanding of the variables and an
enlightened study of it through the related literature. Following tests are most
commonly used for model testing and evaluation.
14.3.1 The R2 Test and Adjusted-R2 Test
We have discussed the concept of coefficient of determination (𝑅 ) in Unit 4. As
you know, the coefficient of determination indicates the explanatory power of a
model. If, for example, 𝑅 = 0.76 we can infer that 76 per cent variation in the
dependent variable is explained by the explanatory variable in the model.
We define R2 as follows:
𝑅 = = 1− ... (14.1)

where TSS = Total Sum of Squares

ESS = Explained Sum of squares
RSS = Residual Sum of Squares
As you know,
TSS = RSS + ESS ... (14.2)
Dividing both sides of equation (14.2) by TSS, we find that
+ =1 ... (14.3)

Since 𝑅 = , we observe that 𝑅 lies between 0 and 1 necessarily. Its

closeness to 1 indicates better fit of the model. If 𝑅 is close to one, RSS is much
smaller compared to ESS. Therefore, very little residual will be left. Thus a

191
Econometric Model model with higher R2 is preferred. You should however keep in mind that a very
Specification and high R2 indicates the presence of multicollinearity in the model. If the R2 is high
Diagnostic Testing
but the t-ratio of the coefficients are not statistically significant you should check
for multicollinearity. The R2 is calculated on the basis of the sample data.
Thus the explanatory variables included the model are considered for estimation
of R2. Variables not included in the model do not account for the variation in the
dependent variable.
There is a tendency of the R2 to increase if more explanatory variables are added.
Thus, we are tempted to add more explanatory variables to increase the
explanatory power of the model. If we add irrelevant explanatory variables in a
model, the estimators are unbiased, but there is an increase in the variance of the
estimators. This makes forecast and analysis on the basis of such models
unreliable.
In order to overcome this difficulty, we use the ‘adjusted-R2’. It is denoted by 𝑅
and defined as follows:
⁄( )
𝑅 =1− ⁄( )
= 1 − (1 − 𝑅 ) … (14.4)

where n is the number of observations and k is the number of regressors. As you

know the TSS has a degree of freedom of (𝑛 − 1) while the ESS has a degree of
freedom of (𝑛 − 𝑘 ). Thus, 𝑅 takes into account the degrees of freedom of the
model. The 𝑅 penalises the addition of explanatory variables. It is observed that
there is an increase in 𝑅 only if the t-value (absolute number) of the additional
explanatory variable is greater than 1. Hence, superfluous variables can be
identified and eliminated from the model. The restriction here is to regress all the
independent variable against the same dependent variable.
Remember that we can compare the 𝑅 of two models only if the dependent
variable is the same. For example, we cannot compare two models if in one
model the explanatory variable is Y and in the other model the explanatory
variable in logY.
14.3.2 Akaike Information Criterion (AIC)
Another method for identifying the mis-specification in a model is Akaike
Information Criterion (AIC). This method also penalises the addition of
regressors as we can see from the formula below:
⁄ ⁄
𝐴𝐼𝐶 = 𝑒 ∑ =𝑒 … (14.5)

where k is the number of regressors (explanatory variables) and n is the number

of observations.
We can further simplify equation (14.5) as

ln 𝐴𝐼𝐶 = + ln … (14.6)

192
where ln 𝐴𝐼𝐶 is the natural log of AIC, and is the penalty factor. Tests for Specification
Error
Remember that the model with a lower value of lnAIC is considered to be better.
Thus, when we compare two models by using the AIC criterion, the model with
lower value of AIC has a better specification. The logic is simple. An
econometric model that reduces the residual sum of squares is a better specified
model.
14.3.3 Schwarz Information Criterion
The Schwarz Information Criterion (SIC) also relies on the RSS, like the AIC
criterion mentioned above. This method also is popular for analysing correct
specification of an econometric model. The SIC is defined as follows:

⁄ ∑ ⁄
𝑆𝐼𝐶 = 𝑛 = 𝑛 … (14.7)

If we take in log-form, equation (14.7) is given as

ln 𝑆𝐼𝐶 = ln 𝑛 + ln … (14.8)

where [(𝑘⁄𝑛) ln 𝑛] is the penalty factor. Note that the SIC criterion imposes a
harsher penalty for inclusion of explanatory variable compared to the AIC
criterion.
14.3.4 Mallow’s 𝑪𝒑 Criterion
When we do not include all the relevant variables in a model, the estimators are
biased. The Mallow’s Cp Criterion evaluates such bias to find out whether there
is significant deviation from the unbiased estimators. Thus, the Mallow’s Cp
Criterion helps us in selecting the best among competing econometric models.
If some of the explanatory variables are dropped from a model, there is an
increase in the residual sum of squares (RSS). Let us assume that the true model
has k regressors. For this model, 𝜎 is the estimator of true 𝜎 . Now, suppose we
drop p regressors from the model. The residual sum of squares obtained from the
truncated model is 𝑅𝑆𝑆 . The Mallow’s Cp Criterion is based on the following
formula:
𝐶 = − (𝑛 − 2𝑝) ... (14.9)

where n is the number of observations.

While choosing a model according to the 𝐶 criterion, the model with the lowest
𝐶 value is preferred.

14.4 CAUTION ABOUT MODEL SELECTION

CRITERIA
We have emphasized earlier that econometric models should be based on
economic theory and logic. Therefore, while constricting an econometric model,

193
Econometric Model you should go by the theoretical appropriateness of including or excluding a
Specification and variable. In order to have a correctly specified model, a thorough understanding
Diagnostic Testing
of the theoretical concepts and the related literature is necessary. Also, the model
that we fit will only be as good as the data that we have collected. If the data
collected does not suffer from, say, multicollinearity or autocorrelation, we are
likely to have a more robust model.
As mentioned earlier, the criteria for selecting an appropriate model primarily
rests on the theory behind it and the strength of the collected data. Many a time,
we observe certain relationship between two variables. Such relationship
however may be superficial or spurious. Let us take an example. At a traffic light,
cars stop when the signal is red. It does not mean that cars cannot move when
there is red light in front of them. It also does not mean that traffic light has some
damaging effect on moving cars. The reason is observance of traffic rules. Unless
we look into the traffic rules and go by observation only, our reasoning will be
wrong. The dependent variable and the independent variable both may be
affected by another variable. In such cases the relationship is confounded.
You should note one more issue regarding selection of econometric models.
Different test criteria may suggest different models. For example, economic logi
suggests that there could two possible econometric models (say, model A and
model B) for a particular issue. You may come across a situation such that 𝑅
test suggests model A and AIC criterion suggest model B. In such situations you
should carry out a number of tests and then only chose the best model.
Adjusted R-squared, Mallows 𝐶 , p-values, etc. may point to different regression
equations without much clarity to the econometrician. Thus, we conclude that
none of the methods for model selection listed above are adequate by itself.
There is no substitute to theoretical understanding of the related literature,
accurately collected data, practical understanding of the problem, and common
sense while specifying an econometric model. We will discuss further on the
model selection criteria in the course BECC 142: Applied Econometrics.
Check Your Progress 1
1) Explain why 𝑅 is a better criterion than R2 in model specification.
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................

194
2) Explain how the AIC and BIC criteria are applied in selection of Tests for Specification
econometric models. Error

.......................................................................................................................
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................

3) What precaution you should take while selecting an econometric model?

.......................................................................................................................
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................
.......................................................................................................................

14.5 LET US SUM UP

Selection of an appropriate econometric model is a difficult task. We have to take
into account the economic theory and logic behind the econometric model. There
could be many competing models for a particular issue.
There a certain criteria on the basis of which the best econometric model is
selected. These criteria could be 𝑅 , AIC, BIC, and Mallow’s Cp. We have
described the formulae for these test criteria in the Unit.

14.6 ANSWERS TO CHECK YOUR PROGRESS

EXERCISES
Check Your Progress 1
1) In Sub-Section 14.3.1 we have compared between R2 and 𝑅 . The 𝑅 takes
into account the degrees of freedom.
2) You should describe the test statistics used in AIC and BIC criteria (see
Section 14.3). The model with lowest value of test statistics is preferred.
3) Go through Section 14.4 and answer.

195
 APPENDIX TABLES
Table A1: Normal Area Table
Z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

0.0 0.0000 0.0040 0.0080 0.0120 0.0160 0.0199 0.0239 0.0279 0.0319 0.0359

0.1 0.0398 0.0438 0.0478 0.0517 0.0557 0.0596 0.0636 0.0675 0.0714 0.0753

0.2 0.0793 0.0832 0.0871 0.0910 0.0948 0.0987 0.1026 0.1064 0.1103 0.1141

0.3 0.1179 0.1217 0.1255 0.1293 0.1331 0.1368 0.1406 0.1443 0.1480 0.1517

0.4 0.1554 0.1591 0.1628 0.1664 0.1700 0.1736 0.1772 0.1808 0.1844 0.1879

0.5 0.1915 0.1950 0.1985 0.2019 0.2054 0.2088 0.2123 0.2157 0.2190 0.2224

0.6 0.2257 0.2291 0.2324 0.2357 0.2389 0.2422 0.2454 0.2486 0.2517 0.2549

0.7 0.2580 0.2611 0.2642 0.2673 0.2704 0.2734 0.2764 0.2794 0.2823 0.2852

0.8 0.2881 0.2910 0.2939 0.2967 0.2995 0.3023 0.3051 0.3078 0.3106 0.3133

0.9 0.3159 0.3186 0.3212 0.3238 0.3264 0.3289 0.3315 0.3340 0.3365 0.3389

1.0 0.3413 0.3438 0.3461 0.3485 0.3508 0.3531 0.3554 0.3577 0.3599 0.3621

1.1 0.3643 0.3665 0.3686 0.3708 0.3729 0.3749 0.3770 0.3790 0.3810 0.3830

1.2 0.3849 0.3869 0.3888 0.3907 0.3925 0.3944 0.3962 0.3980 0.3997 0.4015

1.3 0.4032 0.4049 0.4066 0.4082 0.4099 0.4115 0.4131 0.4147 0.4162 0.4177

1.4 0.4192 0.4207 0.4222 0.4236 0.4251 0.4265 0.4279 0.4292 0.4306 0.4319

1.5 0.4332 0.4345 0.4357 0.4370 0.4382 0.4394 0.4406 0.4418 0.4429 0.4441

1.6 0.4452 0.4463 0.4474 0.4484 0.4495 0.4505 0.4515 0.4525 0.4535 0.4545

1.7 0.4554 0.4564 0.4573 0.4582 0.4591 0.4599 0.4608 0.4616 0.4625 0.4633

1.8 0.4641 0.4649 0.4656 0.4664 0.4671 0.4678 0.4686 0.4693 0.4699 0.4706

1.9 0.4713 0.4719 0.4726 0.4732 0.4738 0.4744 0.4750 0.4756 0.4761 0.4767

2.0 0.4772 0.4778 0.4783 0.4788 0.4793 0.4798 0.4803 0.4808 0.4812 0.4817

2.1 0.4821 0.4826 0.4830 0.4834 0.4838 0.4842 0.4846 0.4850 0.4854 0.4857

2.2 0.4861 0.4864 0.4868 0.4871 0.4875 0.4878 0.4881 0.4884 0.4887 0.4890

2.3 0.4893 0.4896 0.4898 0.4901 0.4904 0.4906 0.4909 0.4911 0.4913 0.4916

2.4 0.4918 0.4920 0.4922 0.4925 0.4927 0.4929 0.4931 0.4932 0.4934 0.4936

2.5 0.4938 0.4940 0.4941 0.4943 0.4945 0.4946 0.4948 0.4949 0.4951 0.4952

2.6 0.4953 0.4955 0.4956 0.4957 0.4959 0.4960 0.4961 0.4962 0.4963 0.4964

2.7 0.4965 0.4966 0.4967 0.4968 0.4969 0.4970 0.4971 0.4972 0.4973 0.4974

2.8 0.4974 0.4975 0.4976 0.4977 0.4977 0.4978 0.4979 0.4979 0.4980 0.4981

2.9 0.4981 0.4982 0.4982 0.4983 0.4984 0.4984 0.4985 0.4985 0.4986 0.4986

3.0 0.4987 0.4987 0.4987 0.4988 0.4988 0.4989 0.4989 0.4989 0.4990 0.4990
196
 Table A2: Critical Values of Chi-squared Distribution

df\area 0.1 0.05 0.025 0.01 0.005

1 2.706 3.841 5.024 6.635 7.879
2 4.605 5.991 7.378 9.210 10.597
3 6.251 7.815 9.348 11.345 12.838
4 7.779 9.488 11.143 13.277 14.860
5 9.236 11.071 12.833 15.086 16.750

6 10.645 12.592 14.449 16.812 18.548

7 12.017 14.067 16.013 18.475 20.278
8 13.362 15.507 17.535 20.090 21.955
9 14.684 16.919 19.023 21.666 23.589
10 15.987 18.307 20.483 23.209 25.188

11 17.275 19.675 21.920 24.725 26.757

12 18.549 21.026 23.337 26.217 28.300
13 19.812 22.362 24.736 27.688 29.819
14 21.064 23.685 26.119 29.141 31.319
15 22.307 24.996 27.488 30.578 32.801

16 23.542 26.296 28.845 32.000 34.267

17 24.769 27.587 30.191 33.409 35.718
18 25.989 28.869 31.526 34.805 37.156
19 27.204 30.144 32.852 36.191 38.582
20 28.412 31.410 34.170 37.566 39.997
21 29.615 32.671 35.479 38.932 41.401
22 30.813 33.924 36.781 40.289 42.796
23 32.007 35.172 38.076 41.638 44.181
24 33.196 36.415 39.364 42.980 45.559
25 34.382 37.652 40.646 44.314 46.928
26 35.563 38.885 41.923 45.642 48.290
27 36.741 40.113 43.195 46.963 49.645
28 37.916 41.337 44.461 48.278 50.993
29 39.087 42.557 45.722 49.588 52.336
30 40.256 43.773 46.979 50.892 53.672

197
 Table A3: Critical Values of t Distribution

Df\p 0.25 0.10 0.05 0.025 0.01 0.005

1 1.0000 3.0777 6.3138 12.7062 31.8205 63.6567

2 0.8165 1.8856 2.9200 4.3027 6.9646 9.9248

3 0.7649 1.6377 2.3534 3.1825 4.5407 5.8409

4 0.7407 1.5332 2.1318 2.7765 3.7470 4.6041

5 0.7267 1.4759 2.0150 2.5706 3.3649 4.0321

6 0.7176 1.4398 1.9432 2.4469 3.1427 3.7074

7 0.7111 1.4149 1.8946 2.3646 2.9980 3.4995

8 0.7064 1.3968 1.8595 2.3060 2.8965 3.3554

9 0.7027 1.3830 1.8331 2.2622 2.8214 3.2498

10 0.6998 1.3722 1.8125 2.2281 2.7638 3.1693

11 0.6974 1.3634 1.7959 2.2010 2.7181 3.1058

12 0.6955 1.3562 1.7823 2.1788 2.6810 3.0545

13 0.6938 1.3502 1.7709 2.1604 2.6503 3.0123

14 0.6924 1.3450 1.7613 2.1448 2.6245 2.9768

15 0.6912 1.3406 1.7531 2.1315 2.6025 2.9467

16 0.6901 1.3368 1.7459 2.1199 2.5835 2.9208

17 0.6892 1.3334 1.7396 2.1098 2.5669 2.8982

18 0.6884 1.3304 1.7341 2.1009 2.5524 2.8784

19 0.6876 1.3277 1.7291 2.0930 2.5395 2.8609

20 0.6870 1.3253 1.7247 2.0860 2.5280 2.8453

20 0.6870 1.3253 1.7247 2.0860 2.5280 2.8453

21 0.6864 1.3232 1.7207 2.0796 2.5177 2.8314

22 0.6858 1.3212 1.7171 2.0739 2.5083 2.8188

23 0.6853 1.3195 1.7139 2.0687 2.4999 2.8073

24 0.6849 1.3178 1.7109 2.0639 2.4922 2.7969

25 0.6844 1.3163 1.7081 2.0595 2.4851 2.7874

26 0.6840 1.3150 1.7056 2.0555 2.4786 2.7787

27 0.6837 1.3137 1.7033 2.0518 2.4727 2.7707

28 0.6834 1.3125 1.7011 2.0484 2.4671 2.7633

29 0.6830 1.3114 1.6991 2.0452 2.4620 2.7564

30 0.6828 1.3104 1.6973 2.0423 2.4573 2.7500

inf 0.6745 1.2816 1.6449 1.9600 2.3264 2.5758

198
 Table A4: Critical Values of F Distribution
(5% level of significance)

df2/df1 1 2 3 4 5 6 7 8 9 10
1 161.448 199.500 215.707 224.583 230.162 233.986 236.768 238.883 240.543 241.882
2 18.513 19.000 19.164 19.247 19.296 19.330 19.353 19.371 19.385 19.396
3 10.128 9.552 9.277 9.117 9.014 8.941 8.887 8.845 8.812 8.786
4 7.709 6.944 6.591 6.388 6.256 6.163 6.094 6.041 5.999 5.964
5 6.608 5.786 5.410 5.192 5.050 4.950 4.876 4.818 4.773 4.735
6 5.987 5.143 4.757 4.534 4.387 4.284 4.207 4.147 4.099 4.060
7 5.591 4.737 4.347 4.120 3.972 3.866 3.787 3.726 3.677 3.637
8 5.318 4.459 4.066 3.838 3.688 3.581 3.501 3.438 3.388 3.347
9 5.117 4.257 3.863 3.633 3.482 3.374 3.293 3.230 3.179 3.137
10 4.965 4.103 3.708 3.478 3.326 3.217 3.136 3.072 3.020 2.978
11 4.844 3.982 3.587 3.357 3.204 3.095 3.012 2.948 2.896 2.854
12 4.747 3.885 3.490 3.259 3.106 2.996 2.913 2.849 2.796 2.753
13 4.667 3.806 3.411 3.179 3.025 2.915 2.832 2.767 2.714 2.671
14 4.600 3.739 3.344 3.112 2.958 2.848 2.764 2.699 2.646 2.602
15 4.543 3.682 3.287 3.056 2.901 2.791 2.707 2.641 2.588 2.544
16 4.494 3.634 3.239 3.007 2.852 2.741 2.657 2.591 2.538 2.494
17 4.451 3.592 3.197 2.965 2.810 2.699 2.614 2.548 2.494 2.450
18 4.414 3.555 3.160 2.928 2.773 2.661 2.577 2.510 2.456 2.412
19 4.381 3.522 3.127 2.895 2.740 2.628 2.544 2.477 2.423 2.378
20 4.351 3.493 3.098 2.866 2.711 2.599 2.514 2.447 2.393 2.348
21 4.325 3.467 3.073 2.840 2.685 2.573 2.488 2.421 2.366 2.321
22 4.301 3.443 3.049 2.817 2.661 2.549 2.464 2.397 2.342 2.297
23 4.279 3.422 3.028 2.796 2.640 2.528 2.442 2.375 2.320 2.275
24 4.260 3.403 3.009 2.776 2.621 2.508 2.423 2.355 2.300 2.255
25 4.242 3.385 2.991 2.759 2.603 2.490 2.405 2.337 2.282 2.237
26 4.225 3.369 2.975 2.743 2.587 2.474 2.388 2.321 2.266 2.220
27 4.210 3.354 2.960 2.728 2.572 2.459 2.373 2.305 2.250 2.204
28 4.196 3.340 2.947 2.714 2.558 2.445 2.359 2.291 2.236 2.190
29 4.183 3.328 2.934 2.701 2.545 2.432 2.346 2.278 2.223 2.177
30 4.171 3.316 2.922 2.690 2.534 2.421 2.334 2.266 2.211 2.165
40 4.085 3.232 2.839 2.606 2.450 2.336 2.249 2.180 2.124 2.077
60 4.001 3.150 2.758 2.525 2.368 2.254 2.167 2.097 2.040 1.993
120 3.920 3.072 2.680 2.447 2.290 2.175 2.087 2.016 1.959 1.911
inf 3.842 2.996 2.605 2.372 2.214 2.099 2.010 1.938 1.880 1.831

199
 Table A4: Critical Values of F Distribution (Contd.)
(5% level of significance)

df2/df1 12 15 20 24 30 40 60 120 INF

1 243.906 245.950 248.013 249.052 250.095 251.143 252.196 253.253 254.314
2 19.413 19.429 19.446 19.454 19.462 19.471 19.479 19.487 19.496
3 8.745 8.703 8.660 8.639 8.617 8.594 8.572 8.549 8.526
4 5.912 5.858 5.803 5.774 5.746 5.717 5.688 5.658 5.628
5 4.678 4.619 4.558 4.527 4.496 4.464 4.431 4.399 4.365
6 4.000 3.938 3.874 3.842 3.808 3.774 3.740 3.705 3.669
7 3.575 3.511 3.445 3.411 3.376 3.340 3.304 3.267 3.230
8 3.284 3.218 3.150 3.115 3.079 3.043 3.005 2.967 2.928
9 3.073 3.006 2.937 2.901 2.864 2.826 2.787 2.748 2.707
10 2.913 2.845 2.774 2.737 2.700 2.661 2.621 2.580 2.538
11 2.788 2.719 2.646 2.609 2.571 2.531 2.490 2.448 2.405
12 2.687 2.617 2.544 2.506 2.466 2.426 2.384 2.341 2.296
13 2.604 2.533 2.459 2.420 2.380 2.339 2.297 2.252 2.206
14 2.534 2.463 2.388 2.349 2.308 2.266 2.223 2.178 2.131
15 2.475 2.403 2.328 2.288 2.247 2.204 2.160 2.114 2.066
16 2.425 2.352 2.276 2.235 2.194 2.151 2.106 2.059 2.010
17 2.381 2.308 2.230 2.190 2.148 2.104 2.058 2.011 1.960
18 2.342 2.269 2.191 2.150 2.107 2.063 2.017 1.968 1.917
19 2.308 2.234 2.156 2.114 2.071 2.026 1.980 1.930 1.878
20 2.278 2.203 2.124 2.083 2.039 1.994 1.946 1.896 1.843
21 2.250 2.176 2.096 2.054 2.010 1.965 1.917 1.866 1.812
22 2.226 2.151 2.071 2.028 1.984 1.938 1.889 1.838 1.783
23 2.204 2.128 2.048 2.005 1.961 1.914 1.865 1.813 1.757
24 2.183 2.108 2.027 1.984 1.939 1.892 1.842 1.790 1.733
25 2.165 2.089 2.008 1.964 1.919 1.872 1.822 1.768 1.711
26 2.148 2.072 1.990 1.946 1.901 1.853 1.803 1.749 1.691
27 2.132 2.056 1.974 1.930 1.884 1.836 1.785 1.731 1.672
28 2.118 2.041 1.959 1.915 1.869 1.820 1.769 1.714 1.654
29 2.105 2.028 1.945 1.901 1.854 1.806 1.754 1.698 1.638
30 2.092 2.015 1.932 1.887 1.841 1.792 1.740 1.684 1.622
40 2.004 1.925 1.839 1.793 1.744 1.693 1.637 1.577 1.509
60 1.917 1.836 1.748 1.700 1.649 1.594 1.534 1.467 1.389
120 1.834 1.751 1.659 1.608 1.554 1.495 1.429 1.352 1.254
inf 1.752 1.666 1.571 1.517 1.459 1.394 1.318 1.221 1.000

200
 Table A4: Critical Values of F Distribution (contd.)
(1% level of significance)
df2/df1 1 2 3 4 5 6 7 8 9 10
1 4052.181 4999.500 5403.352 5624.583 5763.650 5858.986 5928.356 5981.070 6022.473 6055.847
2 98.503 99.000 99.166 99.249 99.299 99.333 99.356 99.374 99.388 99.399
3 34.116 30.817 29.457 28.710 28.237 27.911 27.672 27.489 27.345 27.229
4 21.198 18.000 16.694 15.977 15.522 15.207 14.976 14.799 14.659 14.546
5 16.258 13.274 12.060 11.392 10.967 10.672 10.456 10.289 10.158 10.051
6 13.745 10.925 9.780 9.148 8.746 8.466 8.260 8.102 7.976 7.874
7 12.246 9.547 8.451 7.847 7.460 7.191 6.993 6.840 6.719 6.620
8 11.259 8.649 7.591 7.006 6.632 6.371 6.178 6.029 5.911 5.814
9 10.561 8.022 6.992 6.422 6.057 5.802 5.613 5.467 5.351 5.257
10 10.044 7.559 6.552 5.994 5.636 5.386 5.200 5.057 4.942 4.849
11 9.646 7.206 6.217 5.668 5.316 5.069 4.886 4.744 4.632 4.539
12 9.330 6.927 5.953 5.412 5.064 4.821 4.640 4.499 4.388 4.296
13 9.074 6.701 5.739 5.205 4.862 4.620 4.441 4.302 4.191 4.100
14 8.862 6.515 5.564 5.035 4.695 4.456 4.278 4.140 4.030 3.939
15 8.683 6.359 5.417 4.893 4.556 4.318 4.142 4.004 3.895 3.805
16 8.531 6.226 5.292 4.773 4.437 4.202 4.026 3.890 3.780 3.691
17 8.400 6.112 5.185 4.669 4.336 4.102 3.927 3.791 3.682 3.593
18 8.285 6.013 5.092 4.579 4.248 4.015 3.841 3.705 3.597 3.508
19 8.185 5.926 5.010 4.500 4.171 3.939 3.765 3.631 3.523 3.434
20 8.096 5.849 4.938 4.431 4.103 3.871 3.699 3.564 3.457 3.368
21 8.017 5.780 4.874 4.369 4.042 3.812 3.640 3.506 3.398 3.310
22 7.945 5.719 4.817 4.313 3.988 3.758 3.587 3.453 3.346 3.258
23 7.881 5.664 4.765 4.264 3.939 3.710 3.539 3.406 3.299 3.211
24 7.823 5.614 4.718 4.218 3.895 3.667 3.496 3.363 3.256 3.168
25 7.770 5.568 4.675 4.177 3.855 3.627 3.457 3.324 3.217 3.129
26 7.721 5.526 4.637 4.140 3.818 3.591 3.421 3.288 3.182 3.094
27 7.677 5.488 4.601 4.106 3.785 3.558 3.388 3.256 3.149 3.062
28 7.636 5.453 4.568 4.074 3.754 3.528 3.358 3.226 3.120 3.032
29 7.598 5.420 4.538 4.045 3.725 3.499 3.330 3.198 3.092 3.005
30 7.562 5.390 4.510 4.018 3.699 3.473 3.304 3.173 3.067 2.979
40 7.314 5.179 4.313 3.828 3.514 3.291 3.124 2.993 2.888 2.801
60 7.077 4.977 4.126 3.649 3.339 3.119 2.953 2.823 2.718 2.632
120 6.851 4.787 3.949 3.480 3.174 2.956 2.792 2.663 2.559 2.472
inf 6.635 4.605 3.782 3.319 3.017 2.802 2.639 2.511 2.407 2.321

201
 Table A4: Critical Values of F Distribution (contd.)
(1% level of significance)

df2/df1 12 15 20 24 30 40 60 120 INF

1 6106.321 6157.285 6208.730 6234.631 6260.649 6286.782 6313.030 6339.391 6365.864
2 99.416 99.433 99.449 99.458 99.466 99.474 99.482 99.491 99.499
3 27.052 26.872 26.690 26.598 26.505 26.411 26.316 26.221 26.125
4 14.374 14.198 14.020 13.929 13.838 13.745 13.652 13.558 13.463
5 9.888 9.722 9.553 9.466 9.379 9.291 9.202 9.112 9.020
6 7.718 7.559 7.396 7.313 7.229 7.143 7.057 6.969 6.880
7 6.469 6.314 6.155 6.074 5.992 5.908 5.824 5.737 5.650
8 5.667 5.515 5.359 5.279 5.198 5.116 5.032 4.946 4.859
9 5.111 4.962 4.808 4.729 4.649 4.567 4.483 4.398 4.311
10 4.706 4.558 4.405 4.327 4.247 4.165 4.082 3.996 3.909
11 4.397 4.251 4.099 4.021 3.941 3.860 3.776 3.690 3.602
12 4.155 4.010 3.858 3.780 3.701 3.619 3.535 3.449 3.361
13 3.960 3.815 3.665 3.587 3.507 3.425 3.341 3.255 3.165
14 3.800 3.656 3.505 3.427 3.348 3.266 3.181 3.094 3.004
15 3.666 3.522 3.372 3.294 3.214 3.132 3.047 2.959 2.868
16 3.553 3.409 3.259 3.181 3.101 3.018 2.933 2.845 2.753
17 3.455 3.312 3.162 3.084 3.003 2.920 2.835 2.746 2.653
18 3.371 3.227 3.077 2.999 2.919 2.835 2.749 2.660 2.566
19 3.297 3.153 3.003 2.925 2.844 2.761 2.674 2.584 2.489
20 3.231 3.088 2.938 2.859 2.778 2.695 2.608 2.517 2.421
21 3.173 3.030 2.880 2.801 2.720 2.636 2.548 2.457 2.360
22 3.121 2.978 2.827 2.749 2.667 2.583 2.495 2.403 2.305
23 3.074 2.931 2.781 2.702 2.620 2.535 2.447 2.354 2.256
24 3.032 2.889 2.738 2.659 2.577 2.492 2.403 2.310 2.211
25 2.993 2.850 2.699 2.620 2.538 2.453 2.364 2.270 2.169
26 2.958 2.815 2.664 2.585 2.503 2.417 2.327 2.233 2.131
27 2.926 2.783 2.632 2.552 2.470 2.384 2.294 2.198 2.097
28 2.896 2.753 2.602 2.522 2.440 2.354 2.263 2.167 2.064
29 2.868 2.726 2.574 2.495 2.412 2.325 2.234 2.138 2.034
30 2.843 2.700 2.549 2.469 2.386 2.299 2.208 2.111 2.006
40 2.665 2.522 2.369 2.288 2.203 2.114 2.019 1.917 1.805
60 2.496 2.352 2.198 2.115 2.028 1.936 1.836 1.726 1.601
120 2.336 2.192 2.035 1.950 1.860 1.763 1.656 1.533 1.381
inf 2.185 2.039 1.878 1.791 1.696 1.592 1.473 1.325 1.000

202
Table A5: Durbin-Watson d-statistic Level of Significance = 0.05 k= no. of regressors

________________________________________________________________________________________________

203
GLOSSARY
Association : It refers to the connection or relationship between
variables

Alternative : It is the hypothesis contrary to the null hypothesis.

Hypothesis Null hypothesis and alternative hypothesis are
mutually exclusive.

Alternative : In hypothesis testing, alternative hypothesis states a

Hypothesis condition that is opposite to the null hypothesis. It is
expressed as 𝐻 : 𝛽 ≠ 0, i.e., the slope coefficient is
different from zero. It could be positive or negative.

Analysis of Variance : This is a technique that breaks up the total

(ANOVA) variability of data into two parts one statistical and
the other random.

ANCOVA Model : This is a model which involves both a quantitative

and a dummy variable. The form of such a model
will be like: 𝑌 = 𝛽 + 𝛽 𝐷 + 𝛽 𝑋 + 𝑢 .

ANOVA Model : This is a regression model containing only a dummy

explanatory variable. The functional form of this is
like: 𝑌 = 𝛽 + 𝛽 𝐷 + 𝜇 .

Autocorrelation : The Classical Linear Regression Model assumes

that the random error terms are not related to each
other. In other words, there exists no correlation
between the error terms associated with each
observation. This assumption is referred as the
assumption of no autocorrelation.

Base or Benchmark : The dummy variable which takes the value 0 is

Category referred to as the ‘base or benchmark category’.

Continuous Random : It refers to a random variable that can take infinite

Variable number of values in an interval are called
continuous random variables.

Cochrane-Orcutt : This is a transformation procedure suggested by

Procedure Cochrane-Orcutt. It is helpful in estimating the
value of the correlation coefficient between the
error terms. The transformation, enables the
application of the OLS method, and yields estimates
of parameters which enjoy the BLUE property.
Confidence Interval : In order to test the population parameter, a
Approach confidence interval can be constructed about the true
but unknown mean. If the population parameter lies
within the confidence interval, the null hypothesis is
accepted; otherwise it is rejected.

Classical Linear : It refers to a linear regression model that establishes

Regression Model a linear relationship between the variables, based on
certain specified assumptions.

Chow Test : This test visualizes the presence of structural change

that may result in differences in the intercept or the
slope coefficient or both. This in referred to as
parameter instability. For examining this we perform
Chow Test

Causal Relationship : The relationship between the variables where one

can figure out the cause and the effect between
the two variables.

Confidence Interval : It is the range of values that determines the

probability that the value of the parameter lies
within the interval.

Chi-square : Chi-square distribution is the distribution which is

Distribution the sum of squares of k independent standard normal
random variables.

Composite or Two- : In hypothesis testing, a composite hypothesis covers

Sided Hypothesis a set of values that are not equal to the given or
stated null hypothesis.

Confidence Interval : It refers to the probability that a population

parameter falls within the set of critical values taken
from the Table.

Discrete Random : It refers to random variables that can assume only

Variable countable values.

Distribution Function : Distribution function of a real valued random

variable gives a value at any given sample point in
the sample space.

Deterministic : It represents the systematic component of the

Component regression equation. It is the expected value of the
dependent variable for given values of the
explanatory variable.

205
 Econometric Model : These are statistical models specifying relationship
between relationships between various economic
quantities.

Differential Intercept : In the ANOVA model 𝑌 = 𝛽 + 𝛽 𝐷 + 𝜇 , since

Coefficient there is no continuous regression line involved, the
slope coefficient 𝛽 actually measures by how much
the value of the intercept term differs between the
two categories (e.g. male/female) under
consideration. For this reason, 𝛽 is more
appropriately called as the ‘differential intercept
coefficient’.

Dummy Variable : Response to a dummy variable like gender

Trap (male/female), caste (general/SC-ST/OBC), etc. are
called as categories. Depending on the ‘number’ of
such categories, we must consider including the
number of dummy variables in the regression
carefully. Usually, this should be ‘one less than the
number of categories’. Failing to do this will land us
in a situation called as the ‘dummy variable trap’.
This means we will face a situation of
multicollinearity with no unique estimates, or
efficient estimates, of the parameters. The general
rule for introducing the number of dummies is that,
if there are m attributes or categories, the number of
dummy variables introduced should be ‘m – 1’.

Dummy Variables : There are variables which are qualitative in nature.

Also known as dummy variables, these variables are
referred differently like: indicator variables, binary
variables, categorical variables, dichotomous
variables.

Durbin h-statistic : The Durbin- Watson technique fails to operate when

the regression model involves the lagged value of
dependent variable as one of the explanatory
variables. In such models, the h – statistic, also
suggested by Durbin, is useful to identify the
presence of autocorrelation in the regression model.

Durbin-Watson Test : The test helps detect a first order autocorrelation.

(d-statistic) The test statistic employed is:
(𝑒 − 𝑒 )
𝑑=
𝑒
206
Estimator : A method of arriving at an estimate of a parameter.
Estimation of : This process deals with estimating the values of
Parameters parameters based on measured empirical data that
has a random component.
Estimation : The process of estimating any population parameter.
F-Distribution : It is a right-skewed distribution used for analysis of
variance. F-statistic is used for comparing statistical
models and to identify the model that best fits the
population.
Forecasting : Forecasting is a technique that predicts the future
trends by using historical data. The method of
forecasting is generally used to extrapolate the
parameters such as GDP or unemployment.
Goodness of Fit : An overall goodness of fit that tells us how well the
estimated regression line fits the actual Y values.
Such a measure is known as the coefficient of
determination, denoted by R2. It is the ratio of
explained sum of squares (ESS) to total sum of
squares (TSS).
Glejser Test : The Glejser Test is similar to the Park Test.
Obtaining ei from the original model, Glejser
suggests regressing the absolute values of ei, i.e.,
e i on the X variable expected to be closely
associated with the heteroscedastic variance 𝜎 .
Goldfeld-Quandt Test : In this method of testing for heteroscedasticity, we
first arrange the observations in increasing order of
Xi variable. Next we exclude C observations in the
middle of dataset. Thus, (n – C)/2 observations in
the first part and (n – C)/2 observations in the last
part constitute two groups. We then proceed to
obtain the respective residual sum of squares RSS1
and RSS2. The RSS1 represents the RSS for the
regression corresponding to the smaller Xi values
and RSS2 to that of the larger Xi values. We conduct
F-test to check for the presence of
heteroscedasticity.

Gauss Markov : Under the assumptions of classical linear regression

Theorem model, the least squares estimators are Best Linear
Unbiased Estimate (BLUE). This means, in the class
of all unbiased linear estimators, the OLS estimates
have the minimum or least variance.
207
 Hypothesis : It is a tentative statement that we propose to test. It
is based on the limited evidence. Hypothesis is
formulated on the basis of economic theory or some
logic.

Homoscedasticity : A crucial assumption of the Classical Linear

Regression Model (CLRM) in that the error term ui
in the population regression function (PRF) is
homoscedastic, i.e., they have the same variance  2 .
Such an assumption is referred to as the assumption
of homoscedasticity.

Heteroscedasticity : If the variance of ui is 𝜎 , i.e., it varies from one

observation to another, then the situation is referred
to as a case of heteroscedasticity.

Interactive Dummy : This is a variable like DX in which there is one

dummy variable and one quantitative variable. It is
considered in the multiplicative form to enable us to
see whether the slope coefficients of two groups are
same or different. The functional form of this type
of regression is 𝑌 =𝛽 +𝛽 𝐷 +𝛽 𝑋 +
𝛽 (𝐷 𝑋 ) + 𝑢 .

Jarque-Bera (J-B) : This is an asymptotic or large sample test based on

Test OLS residuals in order to test the normality of the
error term. Coefficient of skewness: S, i.e., the
asymmetry of PDF. Measure of tallness or height of
population distribution function: K
For normal distribution S = 0, K = 3
Jarque and Bera constructed J-Statistics given by
( )
𝐽 = 𝑆 +

Linear Regression : In linear regression models the functional form of

the relationship between the variables is linear.

Mathematical Model : A description of system using mathematical

concepts

Multicollinearity : The classical linear regression model assumes that

there is no perfect multicollinearity, implying no
exact linear relationship among the explanatory
variables, included in multiple regression models.

208
MWD test : This is the test for the selection of the appropriate
functional form for regression as proposed by
Mackinnon, White and Davidson. The test is hence
known as the MWD Test.

Null Hypothesis : The null hypothesis (also called Strawman

hypothesis) states that there is no relationship
between the variables. The coefficients are
deliberately chosen as zero to find out whether Y is
related to X at all. If X really belongs in the model,
we would fully expect to reject the zero-null
hypothesis H0 in favour of the alternatives
hypothesis H1 that it is not zero.

Near or imperfect : The case when two or more explanatory variables

multicollinearity are not exactly linear this reinforces the fact that
collinearity can be high but not perfect.
“High collinearity” refers to the case of “near” or
imperfect” or high multicollinearity.

Null Hypothesis : It is the hypothesis that there is no significant

difference between specified population, the
observed difference is mainly due to sampling or
experimental error.

Normal Distribution : It is a very common probability distribution. The

curve is bell-shaped and the area under the normal
curve is 1.

Ordinary Least : Ordinary Least Squares (OLS) is a method for

Squares Method estimation of the unknown parameters in a linear
regression model. The OLS method minimizes the
sum of the squares of the errors.

Parameters : It is a measurement of any variable. A numerical

quantity that characterizes a given population

Prediction : A regression model explains the variation in the

dependent variable on the basis of explanatory
variables. Given the values of the explanatory
variables, we predict the value of the dependent
variable. The predicted value is different from the
actual value.
Parameter : A quantity or statistical measure for a given
population that is fixed. The mean and the variance
of a population are population parameters.
209
 p- value : It is the lowest level of significance when the null
hypothesis can be rejected.

Power of Test : The power of any test of statistical significance is

defined as the probability that it will reject a false
null hypothesis. The value of the power of test is
given by (1  ).

Population : A population regression function hypothesizes a

Regression Function theoretical relationship between a dependent
(PRF) variable and a set of independent or explanatory
variables. It is a linear function. The function
defines how conditional expectation of a variable Y
responds to the changes in independent variable X.

Perfect : The case of perfect multicollinearity mainly reflects

multicollinearity the situation when the explanatory variables and
perfectly correlated with each other implying the
coefficient of correlation between the explanatory
variables is 1.

Park-Test : If there is heteroscedasticity in a dataset, the

heteroscedastic variance i2 may be systematically
related to one or more of the explanatory variables.
In such cases, we can regress i2 on one or more of
such X- variables. Such an approach, adopted in the
Park-test, helps detect the presence of
heteroscedasticity.

Random Variable : A variable which takes on values which are

numerical outcomes of a random phenomenon.

Regression : A regression analysis is concerned with the study of

the relationship the explained or dependent variable
and the independent or explanatory variables.

Residual Term : The actual value of Y is obtained by adding the

residual term to the estimated value of Y. The
residual term is the estimated value of the random
error term of the population regression function.

Ridge Regression : The ridge regressions are the method of resolving

the problem of multicollinearity. In the ridge
regression, the first step is to standardize the
variables both dependent and independent by
subtracting the respective means and dividing by
their standard deviations.
210
Statistical Inference : It refers to the process of deducing properties of
underlying probability distribution of the parameters
by analysing data.

Standard Normal : It refers to a normal distribution with mean 0 and

Distribution standard deviation 1.

Statistical Inference : It refers to the method of drawing inference about

the population parameter on the basis of random
sampling.
: It is an assumption about a population parameter.
Statistical
Hypothesis: This assumption may or may not be true. This
statistical hypothesis is either accepted or rejected
on the basis of hypothesis testing.

Stochastic Error : The error term represents the influence of those

variables that are not included in the regression
model. It is evident that even if we try to include all
the factors that influence the dependent variable,
there exists some intrinsic randomness between the
two variables.

Subsidiary or : When one explanatory variables X is regressed on

Auxiliary Regressions each of the remaining X variable and the
corresponding R 2 is computed. Each of these
regressions is referred as subsidiary or auxiliary
regression.

t- Distribution : It refers to a continuous probability distribution that

is obtained while estimating mean of normally
distributed population where sample size is small
and population standard deviation is unknown.

Test of significance : The method of inference used to either reject or

Approach accept the null hypothesis. This approach makes use
of test statistic to make any statistical inference.

Test Statistic : A test statistic is a standardized value that is

computed from a sample during the hypothesis
testing. On the basis of test statistics one can either
reject or accept the null hypothesis.

Type I Error: : In the statistical hypothesis testing, type I error is the

incorrect rejection of true null hypothesis. The value
is given by alpha level of significance.

211
 Type II Error : The error that occurs when we accept a null
hypothesis that is actually false. It is the probability
of accepting the null hypothesis when it is false.

Variance Inflation : R 2 obtained variables auxiliary regression may not

Factor (VIF) be completely realiable and is not reliable indicator
of collinearity. In this method we modify the
formula of var (𝑏 ) and (𝑏3 ), var (𝑏 ) =
∑

White’s General : This is a method to test the presence of

Heteroscedasticity heteroscedasticity in a regression model. In this, the
Test residuals obtained from original regression are
squared and regressed on the original variables, their
squared values and their cross-products. Additional
powers of original X variables can be added.

SOME USEFUL BOOKS

Dougherty, C. (2011). Introduction to Econometrics, Fourth Edition,
Oxford University Press
Gujarati, D. N. and D.C. Porter (2010). Essentials of Econometrics, Fourth
Edition, McGraw Hill
Kmenta, J. (2008). Elements of Econometrics, Second Edition, Khosla
Publishing House
Maddala, G.S., and Kajal Lahiri (2012). Introduction to Econometrics,
Fourth Edition, Wiley
Wooldridge, J. M. (2014). Introductory Econometrics: A Modern
Approach, Cengage Learning, Fifth Edition

212