@FM estimates a cointegrating relation among the listed variables using fully modified least squares. The original reference is Phillips and Hansen(1990). This does the extension from Hansen(1992b). Fully modified least squares first estimates the cointegrating vector by least squares, then does a non-parametric correction for small-sample endogeneity. The first listed variable is treated as the "dependent" variable for the calculations.

Note that this assumes that there is only one cointegrating vector connecting the variables. If the cointegrating rank is 2 or larger (which is very likely if you have more than four variables), then this will estimate some linear combination of the cointegrating vectors, but you have no control over its structure. @SWDOLS is an alternative which can handle a larger rank in the cointegration space and @JOHMLE is also available for testing and estimating systems with larger cointegration rank.

@FM(options) start end

# list of endogenous variables

Wizards

This is included as one of estimators in the Cointegration Estimation Wizard on the Time Series menu.

Parameters

start, end |
range to use. By default, the common defined range of the endogenous variables. |

Options

DET=NONE/[CONSTANT]/TREND

Chooses which deterministic components are to be included in the cointegrating regression.

LAGS=number of lags in the windows used the modifications [1]

This gives the number of lags in a lag window used for estimating the long-run variance of the residuals.

DEFINE=(output) EQUATION to define describing the cointegrating relation [not used]

This has the first listed endogenous variable as the dependent variable and the remaining endogenous variables plus any deterministic variables on the right.

[PRINT]/NOPRINT

TITLE="title for report" ["Fully Modified LS"]