Scribd
Upload
13 views

Δy = α + θy + β Δy β Δy: t t-1 1 t-1 + ..... +. k t-k

This report analyzes two monthly time series - the Consumer Price Index and interest rates in the US - from April 1953 to August 2012. Augmented Dickey-Fuller tests show both series are non-stationary. A VAR model in differences is estimated. Impulse response functions show the effect of a one standard deviation shock dies out after a few periods. Engle-Granger and Johansen cointegration tests indicate the two non-stationary series are cointegrated, meaning they have a long-run equilibrium relationship.

Uploaded by

Diamante Gomez
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
13 views

Δy = α + θy + β Δy β Δy: t t-1 1 t-1 + ..... +. k t-k

This report analyzes two monthly time series - the Consumer Price Index and interest rates in the US - from April 1953 to August 2012. Augmented Dickey-Fuller tests show both series are non-stationary. A VAR model in differences is estimated. Impulse response functions show the effect of a one standard deviation shock dies out after a few periods. Engle-Granger and Johansen cointegration tests indicate the two non-stationary series are cointegrated, meaning they have a long-run equilibrium relationship.

Uploaded by

Diamante Gomez
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Econ 910 Report

HW # 6
Eugenio Paulo
In this homework I use two monthly series. The first is Consumer Price Index for All Urban
Consumers: All Items (CPIAUCNS) and the second series is Interest Rates, Government
Securities, Government Bonds for United States (INTGSBUSM193N). The series start in April
1953 through August 2012.
First I do the Augmented-Dickey Fuller Test to determine if the series are stationary. The test
equation is:

yt = + yt-1 + 1yt-1 +

.....+. kyt-k

H0: = 0
H1: 0
For the CPIAUCNS series, the test statistic is -2.7434 and the critical value at 5% significance
level is -3.41, hence we fail to reject the presence of unit root, which suggests that the series is
non-stationary. For the INTGSBUSM193N series the test statistic is -1.6574, and the critical
value at 5% significance level is -2.86, hence we fail to reject the null hypothesis and conclude
that the series is non-stationary.
Next I estimate a VAR in differences model for the variables.
The VAR in differences for CPIAUCNS is:
Coefficients:
(Intercept)

L(CPIAUCNS, 1)

-0.13220

1.00174

L(INTGSBUSM193N, 1)
0.03864

The VAR in differences for INTGSBUSM193N is:


Coefficients:
(Intercept)
-0.13220

L(CPIAUCNS, 1)
1.00174

L(INTGSBUSM193N, 1)
0.03864

Using a recursive assumption I calculated the impulse response functions 1, 2 and 3 steps ahead.
The IRF formulas are:
INTGSBUSM193Ncont <- dynlm(INTGSBUSM193N ~ L(CPIAUCNS,0:1) +
L(INTGSBUSM193N,1))
This yields the following Impulse response functions
print(irf)
[,1]

[,2]

[1,] 1.000000000

0.1153970000

[2,] 0.154242891

0.0161926681

[3,] 0.022181571

0.0007202025

[4,] 0.001578660

-0.0016759936

[5,] -0.001617907

-0.0020302791

[6,] -0.002096341

-0.0020658994

[7,] -0.002150512

-0.0020519729

[8,] -0.002138655

-0.0020305259

[9,] -0.002116712

-0.0020081192

[10,] -0.002093418

-0.0019857750

[11,] -0.002070135

-0.0019636505

[12,] -0.002047072

-0.0019417680

The 1, 2 and 3 steps ahead led to the following:

irf[2,]
[1] 0.15424289

0.01619267

irf[3,]
[1] 0.0221815711

0.0007202025

irf[4,]
[1] 0.001578660

-0.001675994

Test for cointegration


Imposed cointegration vector
Estimate

Std. Error

t value

Pr(>|t|)
0.00733 **

(Intercept)

-0.1321959

0.0491578

-2.689

L(CPIAUCNS, 1)

1.0017426

0.0002472

4052.483

< 2e-16 ***

L(INTGSBUSM193N, 1) 0.0386448

0.0060046

6.436

2.26e-10 ***

L(z, 1)

NA

NA

NA

NA

Estimate

Std. Error

t value

(Intercept)

0.063014

0.030937

2.037

0.042 *

L(CPIAUCNS, 1)

0.994901

0.003792

262.376

<2e-16 ***

L(z, 1)

-0.995243

-263.365

<2e-16 ***

0.003779

Pr(>|t|)

Engle-Granger estimated cointegration vector


The test statistic is -3.35 and the critical values are:
Critical values for test statistics:
1pct

5pct 10pct

tau1 -2.58 -1.95

-1.62

So, we reject the null hypothesis, and conclude that the residuals are stationary, hence the series
are cointegrated.
Johansen Method
alues of teststatistic and critical values of test:
test

10pct 5pct 1pct

r <= 1 | 3.94 7.52 9.24 12.97


r = 0 | 98.02 13.75 15.67 20.20
So, we reject the null that r = 0, and conclude that there is cointegration.

You might also like