VARMAX Procedure

Functional Summary

The statements and options available in the VARMAX procedure are summarized in Table 1.

Table 1: Functional Summary

Description Statement Option
Data Set Options
Specifies the input data set VARMAX DATA=
Writes parameter estimates to an output data set VARMAX OUTEST=
Includes covariances in the OUTEST= data set VARMAX OUTCOV
Writes the diagnostic checking tests for a model and the cointegration test results to an output data set VARMAX OUTSTAT=
Specifies the input data set for scenarios CONDFORE SDATA=
Writes the statistics of simulated forecasts to an output data set CONDFORE OUT=
Writes the simulated forecasts to an output data set CONDFORE OUTSIM=
Writes the conditional covariance matrix to an output data set GARCH OUTHT=
Writes actuals, predictions, residuals, and confidence limits to an output data set OUTPUT OUT=
BY Groups
Specifies BY-group processing BY
ID Variable
Specifies the identifying variable ID
Specifies the time interval between observations ID INTERVAL=
Controls the alignment of SAS date values ID ALIGN=
Options to Control the Optimization Process
Specifies the optimization options NLOPTIONS
Printing Control Options
Specifies how many lags to print results MODEL LAGMAX=
Suppresses the printed output MODEL NOPRINT
Requests all printing options MODEL PRINTALL
Requests the printing format MODEL PRINTFORM=
Controls plots produced through ODS GRAPHICS VARMAX PLOTS=
PRINT= Option
Prints the correlation matrix of parameter estimates MODEL CORRB
Prints the cross-correlation matrices of independent variables MODEL CORRX
Prints the cross-correlation matrices of dependent variables MODEL CORRY
Prints the covariance matrices of prediction errors MODEL COVPE
Prints the cross-covariance matrices of the independent variables MODEL COVX
Prints the cross-covariance matrices of the dependent variables MODEL COVY
Prints the covariance matrix of parameter estimates MODEL COVB
Prints the decomposition of the prediction error covariance matrix MODEL DECOMPOSE
Prints the residual diagnostics MODEL DIAGNOSE
Prints the contemporaneous relationships among the components of the vector time series MODEL DYNAMIC
Prints the parameter estimates MODEL ESTIMATES
Prints the infinite order AR representation MODEL IARR
Prints the impulse response function MODEL IMPULSE=
Prints the impulse response function in the transfer function MODEL IMPULSX=
Prints the partial autoregressive coefficient matrices MODEL PARCOEF
Prints the partial canonical correlation matrices MODEL PCANCORR
Prints the partial correlation matrices MODEL PCORR
Prints the eigenvalues of the companion matrix MODEL ROOTS
Prints the Yule-Walker estimates MODEL YW
Model Estimation and Order Selection Options
Specifies the initial parameter values for non-linear optimization when the model is estimated through the maximum likelihood method INITIAL
Centers the dependent variables MODEL CENTER
Specifies the degrees of differencing for the specified model variables MODEL DIF=
Specifies the degrees of differencing for all independent variables MODEL DIFX=
Specifies the degrees of differencing for all dependent variables MODEL DIFY=
Specifies the estimation method MODEL METHOD=
Selects the tentative order MODEL MINIC=
Suppresses the current values of independent variables MODEL NOCURRENTX
Suppresses the intercept parameters MODEL NOINT
Specifies the number of seasonal periods MODEL NSEASON=
Specifies the order of autoregressive polynomial MODEL P=
Specifies the Bayesian prior model MODEL PRIOR=
Specifies the order of moving-average polynomial MODEL Q=
Centers the seasonal dummies MODEL SCENTER
Specifies the degree of time trend polynomial MODEL TREND=
Specifies the denominator for error covariance matrix estimates MODEL VARDEF=
Specifies the lag order of independent variables MODEL XLAG=
GARCH-Related Options
Specifies how to calculate the constant (unconditional) correlation matrix in the CCC (DCC) GARCH model GARCH CORRCONSTANT=
Specifies the type of the multivariate GARCH model GARCH FORM=
Specifies the order of the GARCH polynomial GARCH P=
Specifies the order of the ARCH polynomial GARCH Q=
Specifies the type of the univariate GARCH model for each innovation in the CCC or DCC GARCH model GARCH SUBFORM=
Cointegration-Related Options
Specifies the restriction on the drift in the VECM COINTEG ECTREND
Prints the results from the weak exogeneity test of the long-run parameters COINTEG EXOGENEITY
Specifies the restriction on the cointegrated coefficient matrix COINTEG H=
Specifies the restriction on the adjustment coefficient matrix COINTEG J=
Specifies the nonlinear constraints that the adjustment coefficient matrix and the cointegrated coefficient matrix are both full rank COINTEG NLC
Specifies the variable name whose cointegrating vectors are normalized COINTEG NORMALIZE=
Specifies a cointegration rank COINTEG RANK=
Prints the Johansen cointegration rank test MODEL COINTTEST=
(JOHANSEN= )
Prints the Stock-Watson common trends test MODEL COINTTEST=(SW= )
Prints the Dickey-Fuller unit root test MODEL DFTEST=
Specifies the vector error correction model (obsolete)[a] MODEL ECM=
Long Memory Options
Specifies the Vector autoregressive fractionally integrated moving average model MODEL FI
Tests and Restrictions on Parameters
Tests the Granger causality CAUSAL GROUP1=
GROUP2=
Places and tests restrictions on parameter estimates BOUND
Places and tests restrictions on parameter estimates RESTRICT
Tests hypotheses on parameter estimates TEST
Forecasting Control Options
Specifies the size of confidence limits for forecasting OUTPUT ALPHA=
Starts forecasting before end of the input data OUTPUT BACK=
Specifies how many periods to forecast OUTPUT LEAD=
Suppresses the printed forecasts OUTPUT NOPRINT
Conditional Forecasts and Scenario Analysis Options
Specifies the size of the credible interval CONDFORE ALPHA=
Specifies the number of multistep forecast values to compute CONDFORE LEAD=
Specifies the number of burn-in iterations CONDFORE NBI=
Specifies the number of Monte Carlo iterations CONDFORE NMC=
Specifies whether and how to consider the uncertainty of parameters CONDFORE PARM=
Specifies a nonnegative integer to use as the seed for generating random number sequences CONDFORE SEED=
Specifies the numeric variable that identifies each scenario CONDFORE SID=

[a] Starting with SAS/ETS 14.1, it is recommended that you use the COINTEG statement instead.


Last updated: June 19, 2025