**[TS] dfactor postestimation** -- Postestimation tools for dfactor

__Postestimation commands__

The following standard postestimation commands are available after
**dfactor**:

Command Description
-------------------------------------------------------------------------
**estat ic** Akaike's and Schwarz's Bayesian information criteria
(AIC and BIC)
**estat summarize** summary statistics for the estimation sample
**estat vce** variance-covariance matrix of the estimators (VCE)
**estimates** cataloging estimation results
**forecast** dynamic forecasts and simulations
**lincom** point estimates, standard errors, testing, and
inference for linear combinations of coefficients
**lrtest** likelihood-ratio test
**nlcom** point estimates, standard errors, testing, and
inference for nonlinear combinations of coefficients
**predict** predictions, residuals, influence statistics, and
other diagnostic measures
**predictnl** point estimates, standard errors, testing, and
inference for generalized predictions
**test** Wald tests of simple and composite linear hypotheses
**testnl** Wald tests of nonlinear hypotheses
-------------------------------------------------------------------------

__Syntax for predict__

**predict** [*type*] {*stub******|*newvarlist*} [*if*] [*in*] [**,** *statistic* *options*]

* statistic* Description
-------------------------------------------------------------------------
Main
**y** dependent variable, which is **xbf** + **residuals**
**xb** linear predictions using the observable independent
variables
**xbf** linear predictions using the observable independent
variables plus the factor contributions
__fac__**tors** unobserved factor variables
__r__**esiduals** autocorrelated disturbances
__in__**novations** innovations, the observed dependent variable minus the
predicted **y**
-------------------------------------------------------------------------
These statistics are available both in and out of sample; type **predict**
*...* **if e(sample)** *...* if wanted only for the estimation sample.

* options* Description
-------------------------------------------------------------------------
Options
__eq__**uation(***eqnames***)** specify name(s) of equation(s) for which
predictions are to be made
**rmse(***stub******|*newvarlist***)** put estimated root mean squared errors of
predicted objects in new variables
__dyn__**amic(***time_constant***)** begin dynamic forecast at specified time

Advanced
__smeth__**od(***method***)** method for predicting unobserved states
-------------------------------------------------------------------------

*method* Description
-------------------------------------------------------------------------
__on__**estep** predict using past information
__sm__**ooth** predict using all sample information
__fi__**lter** predict using past and contemporaneous information
-------------------------------------------------------------------------

__Menu for predict__

**Statistics > Postestimation**

__Description for predict__

**predict** creates a new variable containing predictions such as expected
values, unobserved factors, autocorrelated disturbances, and innovations.
The root mean squared error is available for all predictions. All
predictions are also available as static one-step-ahead predictions or as
dynamic multistep predictions, and you can control when dynamic
predictions begin.

__Options for predict__

The mathematical notation used in this section is defined in *Description*
of **[TS] dfactor**.

+------+
----+ Main +-------------------------------------------------------------

**y**, **xb**, **xbf**, **factors**, **residuals**, and **innovations** specify the statistic to
be predicted.

**y**, the default, predicts the dependent variables. The predictions
include the contributions of the unobserved factors, the linear
predictions by using the observable independent variables, and
any autocorrelation.

**xb** calculates the linear prediction by using the observable
independent variables.

**xbf** calculates the contributions of the unobserved factors plus the
linear prediction by using the observable independent variables.

**factors** estimates the unobserved factors.

**residuals** calculates the autocorrelated residuals.

**innovations** calculates the innovations.

+---------+
----+ Options +----------------------------------------------------------

**equation(***eqnames***)** specifies the equation(s) for which the predictions are
to be calculated.

You specify equation names, such as **equation(income consumption)** or
**equation(factor1 factor2)**, to identify the equations. For the
**factors** statistic, you must specify names of equations for factors;
for all other statistics, you must specify names of equations for
observable variables.

If you do not specify **equation()** and do not specify *stub******, the
results are the same as if you had specified the name of the first
equation for the predicted statistic.

**equation()** may not be specified with *stub******.

**rmse(***stub******|*newvarlist***)** puts the root mean squared errors of the predicted
objects into the specified new variables. The root mean squared
errors measure the variances due to the disturbances but do not
account for estimation error.

**dynamic(***time_constant***)** specifies when **predict** starts producing dynamic
forecasts. The specified *time_constant* must be in the scale of the
time variable specified in **tsset**, and the *time_constant* must be
inside a sample for which observations on the dependent variables are
available. For example, **dynamic(tq(2008q4))** causes dynamic
predictions to begin in the fourth quarter of 2008, assuming that
your time variable is quarterly, see **[D] datetime**. If the model
contains exogenous variables, they must be present for the whole
predicted sample. **dynamic()** may not be specified with **xb**, **xbf**,
**innovations**, **smethod(filter)**, or **smethod(smooth)**.

+----------+
----+ Advanced +---------------------------------------------------------

**smethod(***method***)** specifies the method used to predict the unobserved
states in the model. **smethod()** may not be specified with **xb**.

**smethod(onestep)**, the default, causes **predict** to use previous
information on the dependent variables. The Kalman filter is
performed on previous periods, but only the one-step predictions
are made for the current period.

**smethod(smooth)** causes **predict** to estimate the states at each time
period using all the sample data by the Kalman smoother.

**smethod(filter)** causes **predict** to estimate the states at each time
period using previous and contemporaneous data by the Kalman
filter. The Kalman filter is performed on previous periods and
the current period. **smethod(filter)** may be specified only with
**factors** and **residuals**.

__Examples__

Setup
**. webuse dfex**
**. dfactor (D.(ipman income hours unemp) = , noconstant ar(1))** **(f = ,**
**ar(1/2))**

Forecast changes in **ipman** 6 months into the future, using dynamic
predictions starting in December 2008, and then graph the series
**. tsappend, add(6)**
**. predict Dipman_f, dynamic(tm(2008m12)) equation(D.ipman)**
**. tsline D.ipman Dipman_f if month>=tm(2008m1),** **xtitle("")**
**legend(rows(2))**

Predict and graph the unobserved factor along with changes in **ipman**
**. predict fac if e(sample), factor**
**. tsline D.ipman fac, lcolor(gs10) xtitle("") legend(rows(2))**