Research Journal of Mathematical and Statistical Sciences ________________________________ISSN 23206047

Vol. 1(6), 7-13, July (2013)

Res. J. Mathematical & Statistical Sci.

Generating Critical Values of Unit Root Tests

Akter Rinat1 and Majumder Ajit Kumar2
1

Department of Natural Science, Daffodil International University, Dhaka, BANGLADESH

2
Department of Statistics, Jahangirnagar University, Dhaka, BANGLADESH

Available online at: www.isca.in

Received 31st May 2013, revised 6th June 2013, accepted 5th July 2013

Abstract
In modeling time series econometrics nonstationarity test is very essential part to make some early decision. In order to test
whether the time series is stationary or nonstationary, it is necessary to know the critical values for different sample sizes.
Initially we required different critical values for testing unit root. But these critical values are not available in a wide range
to us. So we need to generate extent tables of critical values. In this paper, we generated critical values for different sample
size using Monte Carlo Simulation for testing unit root.
Keywords: Monte Carlo simulation, DF test, ADF test, critical values.

Introduction
In classical testing approach, critical values are required to test hypothesis. Dickey, D. A., and W. A. Fuller 1, 2, Dickey, D. A.3 ,
Dickey, D., A., Bell, W. R. and Miller, R, B.4 , Gujarati, D. N.5, Hamilton, J. D.6 and Maddala, G. S.7 reported such values for
different tests of testing unit roots, but though reported are very limited. This paper attempts to generate critical values for different
sample sizes. Simulated critical values cannot be calculated by exact mathematical form. In this paper we generated simulated
critical values of DF and ADF tests for different sample size and compare the simulated critical values with the tabulated values.

Critical Values
Critical values for a test of hypothesis depend upon a test statistic, which is specific to the type of test, and the significance level
, which defines the sensitivity of the test. A value of = 0.05 implies that the null hypothesis is rejected 5% of the time when
it is in fact true. The choice of is somewhat arbitrary, although in practice values of 0.1, 0.05, and 0.01 are common. Critical
values are essentially cut-off values that define regions where the test statistic is unlikely to lie; for example, a region where the
critical value is exceeded with probability if the null hypothesis is true. The null hypothesis is rejected if the test statistic lies
within this region which is often referred to as the rejection region(s).

Simulated Critical Values

The distribution form of test statistic of the unit root tests look like t -statistic, but its actual distribution differ from t -statistic. So
there is no exact distribution pattern of these tests. So we need some simulated critical values. For this case we have written the
GAUSS program. By using this program we can generate critical values for any sample sizes. Our simulated critical values are
approximately similar to the tabulated critical values and the critical values only for sample size 25, 50, 100, 250, 500, .

Generated Sequence of Observations from Different Models

Monte Carlo Simulation required from generated observations. In this section we demonstrated how to generate sequence of
observations for the model y t = y t 1 + u t and in the same way we generated observations of the newly proposed distance
based model.
Consider the model,

Model 1 : yt = y t 1 + u t , Model 2 : yt = + y t 1 + u t , Model 3 : yt = + t + yt 1 + u t ,

Where u t ~ N (0 ,

In these models, we test the hypothesis H 0 : = 1 vs

International Science Congress Association

H1 : < 1

Research Journal of Mathematical and Statistical Sciences ___________________________________________ ISSN 23206047

Vol. 1(6), 7-13, July (2013)
Res. J. Mathematical & Statistical Sci.
Table-5
Comparison between simulated critical value of Dickey-Fuller and Augmented Dickey-Fuller table
tnc
tc
tct
Sample size
1%
1%
5%
5%
1%
1%
5%
5%
1%
1%
5%
25
-2.68
-2.63
-1.99
-1.95
-3.88
-3.86
-3.11
-3.09
-4.65
-4.55
-3.64
50
-2.64
-2.62
-1.97
-1.97
-3.63
-3.63
-2.98
-2.97
-4.20
-4.14
-3.42
100
-2.60
-2.60
-1.95
-1.97
-3.51
-3.55
-2.89
-2.92
-3.59
-3.59
-3.20
250
-2.56
-2.60
-1.93
-1.96
-3.45
-3.47
-2.86
-2.88
-3.85
-2.69
-3.06
500
-2.56
-2.57
-1.95
-1.94
-3.19
-3.44
-2.54
-2.86
-3.80
-2.48
-3.01
-2.57
-2.55
-1.92
-1.94
-3.42
-3.45
-2.86
-2.52
-2.44
-2.48
-1.78

5%
-3.79
-3.54
-2.91
-2.00
-1.79
-1.77

From the above tables we have shown that our simulated critical values are approximately same to the critical values given for
Dickey-Fuller test. In the following table we generated the critical values of Dickey-Fuller test and Augmented Dickey-Fuller test
for different sample sizes and are given in table 6. If the random seed changes then the critical values of Dickey-Fuller test and
Augmented Dickey-Fuller test one given in the table.

Sample Size
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

Table-6
Generated critical value of Dickey-Fuller table for sample size n=20, 21,.50
tc
tnc
1%
5%
1%
5%
1%
-2.7912
-2.0226
-4.0478
-3.2178
-4.9185
-2.7665
-2.0098
-4.0097
-3.1894
-4.8272
-2.7297
-1.9791
-4.0058
-3.1775
-4.8232
-2.6966
-1.9916
-3.9049
-3.1309
-4.7512
-2.706
-1.9847
-3.9295
-3.1464
-4.6634
-2.6897
-1.9969
-3.8827
-3.116
-4.6587
-2.7172
-1.9965
-3.8807
-3.1139
-4.6291
-2.7062
-1.9946
-3.8324
-3.1094
-4.5906
-2.7473
-2.0093
-3.8396
-3.0871
-4.5686
-2.6972
-1.9848
-3.8591
-3.0923
-4.5575
-2.6686
-1.9804
-3.8206
-3.0739
-4.5386
-2.649
-1.9858
-3.7788
-3.0644
-4.5209
-2.6653
-1.9681
-3.8078
-3.0674
-4.4785
-2.6852
-1.9725
-3.7743
-3.0568
-4.4334
-2.675
-1.974
-3.7538
-3.0445
-4.4181
-2.643
-1.9802
-3.7424
-3.0409
-4.4289
-2.667
-1.9784
-3.7169
-3.0314
-4.3974
-2.6771
-1.9835
-3.7302
-3.0295
-4.3919
-2.6765
-1.9747
-3.7555
-3.0369
-4.3479
-2.6617
-1.9822
-3.6732
-3.0116
-4.3409
-2.659
-1.9707
-3.725
-3.0205
-4.3336
-2.6347
-1.9634
-3.6809
-3.0105
-4.3077
-2.6679
-1.9741
-3.6974
-3.0096
-4.3046
-2.6503
-1.9751
-3.6763
-2.9995
-4.2747
-2.6871
-1.9878
-3.6781
-2.9939
-4.2793
-2.6663
-1.9591
-3.6595
-2.999
-4.2519
-2.6463
-1.9883
-3.6642
-2.988
-4.2502
-2.6313
-1.9812
-3.6698
-2.9931
-4.2612
-2.6331
-1.9557
-3.6638
-2.9875
-4.2087
-2.6386
-1.9548
-3.6617
-2.9953
-4.2248
-2.6468
-1.9752
-3.6317
-2.9813
-4.2083

International Science Congress Association

tct
5%
-3.9973
-3.9415
-3.9051
-3.8622
-3.861
-3.8197
-3.8218
-3.7762
-3.7781
-3.7526
-3.7457
-3.7165
-3.7226
-3.7095
-3.6853
-3.6766
-3.6628
-3.641
-3.6536
-3.6236
-3.6127
-3.5997
-3.5969
-3.5766
-3.5727
-3.5519
-3.5507
-3.5529
-3.536
-3.5267
-3.5101

Research Journal of Mathematical and Statistical Sciences ___________________________________________ ISSN 23206047

Vol. 1(6), 7-13, July (2013)
Res. J. Mathematical & Statistical Sci.
Table-8
Generated critical value of Dickey-Fuller table for sample size n=20, 21,.50 (changing the random seed)
tnc
tc
tct
Sample Size
1%
5%
1%
5%
1%
5%
20

-2.7641

-2.0099

-3.9205

-3.1107

-4.6271

-3.7599

21

-2.7318

-1.9992

-3.9208

-3.1109

-4.6058

-3.7328

22

-2.7329

-2.0027

-3.8645

-3.0865

-4.5545

-3.7118

23

-2.7158

-1.9863

-3.8685

-3.086

-4.4909

-3.6963

24

-2.7058

-1.9977

-3.8372

-3.0573

-4.4834

-3.6604

25

-2.7106

-1.9811

-3.8238

-3.0635

-4.4378

-3.6526

26

-2.7621

-1.9875

-3.8398

-3.0668

-4.4201

-3.6525

27

-2.7232

-1.9984

-3.8146

-3.065

-4.4138

-3.6529

28

-2.6708

-1.9825

-3.7855

-3.0363

-4.3888

-3.6236

29

-2.6567

-1.9712

-3.7934

-3.0363

-4.4006

-3.6253

30

-2.6777

-1.9873

-3.7384

-3.0147

-4.364

-3.5997

31

-2.7

-1.9723

-3.7272

-3.0068

-4.3663

-3.6113

32

-2.7142

-1.9903

-3.7263

-3.0043

-4.3215

-3.5805

33

-2.6855

-1.9771

-3.7394

-3.0088

-4.3329

-3.5892

34

-2.6424

-1.9739

-3.7024

-3.0046

-4.3263

-3.5788

35

-2.6606

-1.9568

-3.7097

-2.9918

-4.2657

-3.5595

36

-2.6761

-1.9675

-3.6794

-2.9844

-4.2703

-3.5464

37

-2.6483

-1.9734

-3.6593

-2.9725

-4.27

-3.548

38

-2.6522

-1.9779

-3.6907

-2.9786

-4.2221

-3.5344

39

-2.6696

-1.9737

-3.6812

-2.9932

-4.2224

-3.5178

40

-2.6451

-1.9704

-3.6131

-2.9706

-4.2036

-3.4999

41

-2.6629

-1.981

-3.6445

-2.9846

-4.2229

-3.4986

42

-2.6542

-1.9888

-3.6271

-2.9833

-4.1835

-3.4939

43

-2.6633

-1.9774

-3.6383

-2.9645

-4.2152

-3.4987

44

-2.6297

-1.9619

-3.5997

-2.9555

-4.1727

-3.4823

45

-2.6518

-1.9746

-3.6142

-2.9619

-4.1564

-3.4858

46

-2.6519

-1.9915

-3.6212

-2.9501

-4.1405

-3.4816

47

-2.6488

-1.9596

-3.6307

-2.9607

-4.1375

-3.4689

48

-2.6349

-1.9674

-3.6104

-2.9602

-4.187

-3.4906

49

-2.64

-1.9718

-3.61

-2.9572

-4.1062

-3.4497

50

-2.6414

-1.9591

-3.6118

-2.9543

-4.1126

-3.4502

International Science Congress Association

11

Research Journal of Mathematical and Statistical Sciences ___________________________________________ ISSN 23206047

Vol. 1(6), 7-13, July (2013)
Res. J. Mathematical & Statistical Sci.

Conclusion
Unit root test is potentially a serious problem for time series data. Most of the unit root tests have estimation and testing problems.
Hence the test of hypothesis becomes invalid and gives serious misleading conclusions about the statistical significant for the
estimated coefficients as well as the model. There are different types of unit root tests for testing unit root. But these exiting tests
have a number of drawbacks. In this paper we introduce some concept of simulated critical values. Also we discuss how to
generate critical values of Dickey-Fuller type statistics and finally we generated simulated critical values for a wide range sample
sizes. These tables can be used to tests unit roots for different sizes. Considered in this paper, the program is written in GAUSS
code and is available to the author.

References
1.

Dickey D.A. and Fuller W.A., Distribution of the Estimators for Autoregressive Time Series with a Unit Root, Journal of the
American Statistical Association, 74, 427-431 (1976)

2.

Dickey D.A. and Fuller W.A., Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root, Econometrica,
49, 1057-1072 (1981)

3.

Dickey D.A., Estimation and Hypothesis Testing for Nonstationary Time Series, Ph.D. Thesis, lowa Stat University, Ames,
lowa (1976)

4.

Dickey D.A., Bell W.R. and Miller R.B., Unit roots series models: Tests and implications, Am, Statistical 40, 12-26 (1986)

5.

Gujarati D.N., Basic Econometrics, 4th ed., McGraw Hill, New York (2003)

6.

Hamilton J.D., Time Series Analysis, Princeton, N. J,: Princeton University Press (1994)

7.

Maddala G.S., Introduction to Econometrics, 2nd ed., Maxwell Macmillan, New York (1997)

8.

Dickey D.A. and Gonzalez-Farias G., A New Maximum Likelihood Approach to Testing for Unit Roots, unpublished paper
presented at the 1992 Joint Statistical meetings, August 9-13, Boston (1992)

9.

Dickey D.A., Hasza D.P. and Fuller, W.A., Testing for Unit Roots in Seasonal Time Series, Journal of the American
Statistical Association, 79, 355-367 (1984)

10. Enders W., Applied econometric time series, John Wiley and Sons, New York (1995)
11. Fuller, W. A., Introduction to Statistical Time Series, New York John Wiley (1976)
12. Evans G. and Savin N.E., Testing for Unit Roots: I, Econometrica, 49, 753-779 (1981)
13. Evans G. and Savin N.E., Testing for Unit Roots: II, Econometrica, 52, 1241-1269 (1984)
14. Schwert G.W., Tests for Unit Roots: A Monte Carlo Investigation, Journal of Business and Economic Statistics, 7, 147-159,
(1989)
15. Greene W.H., Econometric Analysis, 3rd ed., Prentice Hall, Upper Saddle River, New Jersey (1997)
16. DeJong D.N., Nankervis J.C., Savin N.E. and Whiteman C.H., The Power Problems of Unit Root Tests in Time Series with
Autoregressive Errors, Journal of Econometrics, 53, 323-343 (1992)
17. DeJong D.N. and Whiteman C.H., Reconsidering Trends and Random Walks in Macroeconomic Time Series, Journal of
Monetary Economics, 28, (1991)
18. Judge G.G., Griffiths W.E., Hill R.C., Lutkephol H. and Lee T.C., The Theory and Practice of Econometrics, 2nd ed., John
Willey and Sons, New York (1985)
19. Majumder A.K., one-Sided and Partially One-Sided Hypothesis Testing in Economics, unpublished PhD Thesis, Monash
University, Australia (1999)
20. Kwiatkowski D., Phillips P.C.B., Schmidt P. and Shin Y., Testing the Null Hypothesis of Stationary Against the Alternative
of a Unit Root, Journal of Econometrics, 54, 159-178 (1992)

International Science Congress Association

13