Scribd
Upload
132 views

Co-Integration and Error Correction Model

1) The document analyzes the relationship between 3-month and 6-month treasury bill rates over time using regression and cointegration analyses. 2) Augmented Dickey-Fuller tests show that both time series contain a unit root and are non-stationary, but cointegration tests show they follow a similar trend, indicating a long-term equilibrium relationship. 3) An error correction model is used to account for short-term disequilibriums, showing that about 19% of the short-term disequilibrium is corrected each month.

Uploaded by

Saima Munawar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
132 views

Co-Integration and Error Correction Model

1) The document analyzes the relationship between 3-month and 6-month treasury bill rates over time using regression and cointegration analyses. 2) Augmented Dickey-Fuller tests show that both time series contain a unit root and are non-stationary, but cointegration tests show they follow a similar trend, indicating a long-term equilibrium relationship. 3) An error correction model is used to account for short-term disequilibriums, showing that about 19% of the short-term disequilibrium is corrected each month.

Uploaded by

Saima Munawar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Assignment 3

Submitted by: Saima Munawar (25456)


A regression is called spurious regression if the time series of regressors and
regressed variable involved in the analysis have a unit root. However, the regression
analysis would not be spurious if the time series involved in regression have cointegration.
Since treasury bill rates are expressed in terms of interest rates, we don’t take log of IV and DV for
regression analysis. Figure 1 shows graphical representation graphical representation of 3-month
treasury bill rates (tb3) and 6-month treasury bill rates (tb6) and six-month Treasury bill rates. It
seems that both time series are non-stationary since they have trend. However, they both seemed
to follow similar trend this indicates that they might have cointegration.

Figure 1Line graph - 3 Month and 6 Month Treasury Bill Rates


Table 1

Table 1 shows Augmented Dickey-Fuller test for tb3 and tb6 time series with intercept and trend
included at five lagged terms.

The Augmented Dickey-Fuller Test of 3-month Treasury Bill rates time series shows that p > 0.05 and
calculated t-statistic values in absolute terms is less than critical values at 5% level. Hence, we retain
the null hypothesis that 3-month treasury bill rates (tb3) has a unit root at 95% confidence interval.
Similarly, Dickey-Fuller Test of 6-month Treasury Bill rates time series shows that p > 0.05 and
calculated t-statistic (tau) values in absolute terms is less than critical values at 5% level. Hence, we
retain the null hypothesis that 3-month treasury bill rates (tb3) has a unit root at 95% confidence
interval. Conclusively, we can say that both tb3 and tb6 time series are non-stationary at 5% level.
Table 2

The regression analysis between tb3 and t6 is shown in table 2. It shows $1 change in tb3 will
increase the tb6 by $ 0.98. The model captures 99.555 percent variations in dependent variable
significant at 99% confidence interval F(1, 348)= 77680.38, p < 0.001. However, the regression
between these two timeseries can be spurious if they lack cointegration.
Table 3

In order to test the cointegration between tb3 and tb6, Engle-Garner (EG) test is conducted. For this,
error term (ERR1) of regression between tb6 and tb3 is generated. Dickey-Fuller test of an error
term of a regression analysis is in fact EG Test. The Augmented Dickey-Fuller Test of ERR1 without
intercept or trend is AEG Test of cointegration of tb3 and tb6. Table 3 shows AEG test results. It
shows that p < 0.001 and the calculated tau statistics is greater than critical values at 1%, 5% and
10% level. Therefore, there is a very strong evidence to reject the null hypothesis that ERR1 has unit
root. Since ERR1 is stationary time series, it implies that tb3 and tb6 time series are cointegrated.

AEG proves that there is long-term or equilibrium relationship. However, there may be disturbances
or short-term disequilibrium. This short-term equilibrium can be treated by Error Correction Model
(ECM) as expressed below.

d(tb6) = B0 + B1 d(tb3) +B2 ERR1(-1)


Table 4

All the variables in the ECM must be stationary. ERR1 is stationary as per the above discussion. ADF
test of both d(tb3) and d(tb6) in Table 4 shows that p < 0.001 and the calculated tau statistics is
greater than critical values at 1%, 5% and 10% level. Therefore, there is a very strong evidence that
d(tb3) and d(tb6) timeseries are stationary.
Table 5

The regression analysis of ECM in Table 5 shows that all variables are statistically significant at 1%
level t (346) > 2, p < 0.001. The model explains almost 91% variations in the dependent variable –
d(tb6) – significant at 99% confidence interval F (2, 346) = 1767, p < 0.001. The coefficient of d(TB3)
shows that 1% change in 3-month treasury bill rates will result in 0.879% change in the same
direction in 6-month treasury bill rates CP. The coefficient of the error correction term of ECM
ERR1(-1) shows that 18.74% of short-term disequilibrium is corrected every month. Hence, the
short-term disequilibrium is corrected to reach long -term equilibrium in 5.33 months.

You might also like