Find optimal lag length                                           (STB-7: sts2)
-----------------------

    ^findlag^ yvar [xvars] [weight] [^, n^lags^(^#^) d^etail ^z^ero regress-options ]


Description
-----------

^findlag^ reports four statistics that help determine the optimal lag length of
the dependent variable yvar in a regression.  The four statistics are the root
mean squared error of the regression (RMSE), Akaike's Information Criterion
(AIC), Amemiya's Prediction Criterion (PC), and Schwarz's Information Criterion
(SC).  Each statistic is calculated for regressions containing containing 1 to
^nlags()^ lags of yvar; if ^zero^ is specified, each statistic is calculated for
regressions containint 0 to ^nlags()^ lags of yvar.

For each statistic, the apparent optimal number of lags is reported.  If 
^detail^ is specified, the values of the statistics along with the p-values
of the F test test on excluding all the lags and the t-test on excluding
the last lag of yvar are reported for each regression.


Options
-------

^nlags(^#^)^ specifies the number of lags to attempt and defaults to 4 if not
    specified.

^detail^ requests that detailed output of results be presented.

^zero^ requests that lag 0 be included in the search.

regress_options refers to any of the options of the regress command; see ^help^
    ^regress^.



(Examples follow.)







Examples
--------

 . ^findlag gnp^                 estimates the optimal # of lags in a
                               univariate regression of GNP; 1 to 4 lags
                               considered.

 . ^findlag gnp, nlags(6)^       same as above; 1 to 6 lags considered.

 . ^findlag gnp, nlags(6) zero^  same as above, 0 to 6 lags considered.

 . ^findlag consump gnp^         estimates optimal # of lags in a regression
                               of consumption on GNP.


Author
------
 
Sean Becketti, Federal Reserve Bank of Kansas City.


Also see
--------

    STB:  sts2 (STB-7)
On-line:  ^help^ for ^dicky^, ^lag^