CointReg request for programs
Posted: Wed Jul 28, 2010 1:47 pm
I am finding that my long-run and error-correction forecasting models are much improved by using FM-OLS estimates of the static long-run.
A programing version of the interactive COINTREG command that would produce some sort of vector output (estimated level coefficients at a minimum) would be very useful, allowing automation of sub-sample explorations and one-step-ahead or moving window based forecasts.
I have in mind something akin to the vector output produced by the COINT command that conducts Johansen tests on a group.
Or perhaps the equations produced by cointreg could be a separate type of equation object, with their own set of data elements?
.
A programing version of the interactive COINTREG command that would produce some sort of vector output (estimated level coefficients at a minimum) would be very useful, allowing automation of sub-sample explorations and one-step-ahead or moving window based forecasts.
I have in mind something akin to the vector output produced by the COINT command that conducts Johansen tests on a group.
Or perhaps the equations produced by cointreg could be a separate type of equation object, with their own set of data elements?
.