A bug in the command "makecoint"
Posted: Mon Dec 26, 2022 11:58 pm
The official document said that the command "makecoint" makes a new series containing the cointegration.
For example, establish an ARDL model with two variables x and y. And suppose the cointegration equation gets from the estimation is y = a + bx.
Then, the command should create a series z= a+bx, if it works well.
However, it seems that the command creates a series w= y-(a+bx), which is not the cointegration series.
Of course, sometimes we do want to calculate series "w" too. In fact, the reason I find this bug is that I want to calculate series "w" at first.
Hence, I suggest correcting the bug in the command "makecoint"; meanwhile, design another command to calculate series "w" directly.
For example, establish an ARDL model with two variables x and y. And suppose the cointegration equation gets from the estimation is y = a + bx.
Then, the command should create a series z= a+bx, if it works well.
However, it seems that the command creates a series w= y-(a+bx), which is not the cointegration series.
Of course, sometimes we do want to calculate series "w" too. In fact, the reason I find this bug is that I want to calculate series "w" at first.
Hence, I suggest correcting the bug in the command "makecoint"; meanwhile, design another command to calculate series "w" directly.