The two variables are clearly related. The shorter term rate is
significantly lower than the longer term rate. This is due to
liquidity and risk premiums. The two variables are not
stationary since the means are not constant. Both variable
appear to be I(1), although we should test for unit roots.
GRETL can do this for us.
First we open GRETL and choose the path
MODEL
TIME SERIES
COINTEGRATION TEST
ENGLE-GRANGER
As shown in the figure below
BOX 1
BOX 2
Augmented Dickey-Fuller test, order 1, for R30
sample size 178
unit-root null hypothesis: a = 1
test with constant
estimated value of (a - 1): -0.0242551
test statistic: tau_c(1) = -1.53538
asymptotic p-value 0.5158
BOX 3
Cointegrating regression OLS estimates using the 180 observations 1959:1-2003:4
Dependent variable: R20
VARIABLE
COEFFICIENT
STDERROR
T STAT P-VALUE
const
-0.390957
0.0971120
R30
0.821951
0.0117602
BOX 4
lag order 1
sample size 178
unit-root null hypothesis: a = 1
estimated value of (a - 1): -0.193805
test statistic: tau_c(2) = -4.00502
asymptotic p-value 0.007024
1
2 t 1 t
This time we run OLS on our ECM to get the following output