[B,CovRes,R2] = regress(M,Lhs,Rhs,...)
M
[ model ] - Model on whose covariance matrices the popolation regression will be based.
Lhs
[ char | cellstr ] - Lhs variables in the regression; each of the variables must be part of the state-space vector.
Rhs
[ char | cellstr ] - Rhs variables in the regression; each of the variables must be part of the state-space vector, or must refer to a larger lag of a transition variable present in the state-space vector.
B
[ namedmat | numeric ] - Population regression coefficients.
CovRes
[ namedmat | numeric ] - Covariance matrix of residuals from the population regression.
R2
[ numeric ] - Coefficient of determination (R-squared).
'matrixFmt='
[ 'namedmat'
| 'plain'
] - Return matrices B
and CovRes
as either namedmat
object (i.e. matrices with named rows and columns) or plain numeric arrays.Population regressions calculated by this function are always centred. This means the regressions are always calculated as if estimated on observations with their uncondional means (the steady-state levels) removed from them.
The Lhs and Rhs variables that are log variables must include log( )
explicitly in their names. For instance, if X
is declared to be a log variable, then you must refer to log(X)
or log(X{-1})
.
[B,C] = regress('log(R)',{'log(R{-1})','log(dP)'});