DOCUMENTATION
Sign In

Get access to My SAS, trials, communities and more.

Econometrics Procedures
 

DYNAMICLINEAR Procedure

Data Table Output

PROC DYNAMICLINEAR can create multiple output data tables. All output data tables are listed in Table 1. The information in these tables is described in the following sections.

DISCOUNTFACTOR Statement

OUTVERIFICATION= Data Table Generated by the Statement

The OUTVERIFICATION= output data table contains the following variables:

RowID

index of the rows.

Variable_i

value of the discount factors for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement.

VarName

name of the parameter. The values are beta, deltaphi, and deltagamma.

VarID

FactorID value of the parameter.

FILTER Statement

COVARIANCE= Data Table Generated by the PRIOR=/NAIVEPOSTERIOR=/POSTERIOR= Suboption of the OUTFILTERINFO= Option

The COVARIANCE= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

ParmValue_k

parameter value for the covariances of the normal-gamma distributions at the time given by the sum of the ID Variable value and the LeadStep value. The parameters correspond to those in the local-level simultaneous graphical dynamic linear model in the section Simultaneous Graphical Dynamic Linear Models. The dimension p of the variable is the sum of the number of the variables in the ith variable parental set and 1, and k equals 1 comma ellipsis comma p.

RowID

index of the rows.

VarName

name of the variable that corresponds to the variable in the MODEL statement.

VarID

order of the variables as they appear in the MODEL statement.

ParmName

name of the parameter.

ParmID

order of the parameters as they appear in the local-level SGDLM.

GAMMA= Data Table Generated by the PRIOR=/NAIVEPOSTERIOR=/POSTERIOR= Suboption of the OUTFILTERINFO= Option

The GAMMA= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

Variable_i

value of the shape and rate parameters for the normal-gamma distributions at a given time step for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement. These parameters correspond to the time given by the sum of the ID Variable value and the LeadStep value.

RowID

index of the rows.

VarName

name of the parameter. The possible values are shape and rate.

VarID

FactorID value of the parameter. The possible values are 1 and 2.

KLDIVERGENCE= Data Table Generated by the OUTFILTERINFO= Option

The KLDIVERGENCE= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

Effective Sample Size

value of the effective sample size at a given time step. The value is calculated on the basis of the KL divergence during the importance sampling of posterior updating. For more information about the calculation, see Gruber and West (2016).

KL Divergence

value of KL divergence at the time given by the sum of the ID Variable value and the LeadStep value.

LAMBDA= Data Table Generated by the PREDICTIONSAMPLE=/RECOUPLINGSAMPLE= Suboption of the OUTFILTERINFO= Option

The LAMBDA= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

Variable_i

samples of the precision parameter for observation errors at a given time step for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement. The given time step is the sum of the ID Variable value and the LeadStep value.

RowID

index of the rows.

MEAN= Data Table Generated by the PRIOR=/NAIVEPOSTERIOR=/POSTERIOR= Suboption of the OUTFILTERINFO= Option

The MEAN= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

ParmValue

parameter value of the means for the normal-gamma distributions at the time given by the sum of the ID Variable value and the LeadStep value.

RowID

index of the rows.

VarName

name of the variable that corresponds to the variable in the MODEL statement.

VarID

order of the variables as they appear in the MODEL statement.

ParmName

name of the parameter.

ParmID

order of the parameters as they appear in the local-level simultaneous graphical dynamic linear model in the section Simultaneous Graphical Dynamic Linear Models.

NU= Data Table Generated by the RECOUPLINGSAMPLE= Suboption of the OUTFILTERINFO= Option

The NU= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

Variable_i

samples of the observation errors at a given time step for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement. The given time step is the sum of the ID Variable value and the LeadStep value.

RowID

index of the rows.

THETA= Data Table Generated by the PREDICTIONSAMPLE=/RECOUPLINGSAMPLE= Suboption of the OUTFILTERINFO= Option

The THETA= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

ParmValue_k

samples of the state vectors at the time given by the sum of the ID Variable value and the LeadStep value. The dimension p of the variable is the sum of the number of the variables in the ith variable parental set and 1, and k equals 1 comma ellipsis comma p.

RowID

index of the rows.

VarName

name of the variable that corresponds to the variable in the MODEL statement.

VarID

order of the variables as they appear in the MODEL statement.

FILTER and FORECAST Statements

OUT= Data Table Generated by the Statements

Note: The variable names display a maximum of four decimal places. Any percentile values lower than 0.0001 are replaced by 0.0001 for representation.

In the output table, you might encounter instances where variance values of all variables are listed as 0. This situation typically arises when the NSIMULATIONS= option in the PROC DYNAMICLINEAR statement is set too low; the result is that each CPU thread handles only one job or none at all. Such a scenario can lead to an insufficient number of simulations for accurately estimating variance. To solve this problem, consider increasing the NSIMULATIONS= option value, ensuring that each thread processes a sufficient number of jobs to produce more reliable variance estimates.

The OUT= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

Mean_i

sample mean at a given time step for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement. These sample means are derived from the prior normal-gamma distributions that correspond to the time given by the sum of the ID Variable value and the LeadStep value.

Variance_i

sample variance at a given time step for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement. Like the sample mean, this variable is calculated using the prior normal-gamma distributions that correspond to the time given by the sum of the ID Variable value and the LeadStep value.

P_ij

jth (j equals 1 comma ellipsis comma upper J) percentile points at a given time step for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement. The percentile points are specified by the PERCENT=number-list suboption of the OUT= option, and J is the number of elements in the number-list.

OUTCOVARIANCE= Data Table Generated by the Statements

The OUTCOVARIANCE= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

Variable_i

the covariance value at a given time step between the ith and jth variables, , where m is the number of variables in the MODEL statement. The sample covariance is calculated using the samples in the OUTSIM= output data table.

RowID

index of the rows.

VarName

name of the variable that corresponds to the variable in the MODEL statement.

VarID

order of the variables as they appear in the MODEL statement.

OUTSIM= Data Table Generated by the Statements

The OUTSIM= output data table contains the following variables:

ID Variable

value of the variable in the ID statement.

LeadStep

value of h in the h-step-ahead forecast.

Variable_i

samples of forecasting observations at a given time step for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement. These sample means are derived from the prior normal-gamma distributions that correspond to the time given by the sum of the ID Variable value and the LeadStep value.

RowID

index of the rows.

INITIALDISTRIBUTION Statement

COVARIANCE= Data Table Generated by the OUTVERIFICATION= Option

The COVARIANCE= output data table contains the following variables:

RowID

index of the rows.

ParmValue_k

parameter value for the covariances of the initial normal-gamma distributions. The parameters correspond to those in the local-level simultaneous graphical dynamic linear model in the section Simultaneous Graphical Dynamic Linear Models. The dimension p of the variable is the sum of the number of the variables in the ith variable parental set and 1, and k equals 1 comma ellipsis comma p.

VarName

name of the variable that corresponds to the variable in the MODEL statement.

VarID

order of the variables as they appear in the MODEL statement.

ParmName

name of the parameter.

ParmID

order of the parameters as they appear in the local-level SGDLM.

GAMMA= Data Table Generated by the OUTVERIFICATION= Option

The GAMMA= output data table contains the following variables:

RowID

index of the rows.

Variable_i

value of the shape and rate parameters for the initial normal-gamma distributions for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement.

VarName

name of the parameter.

VarID

FactorID value of the parameter.

MEAN= Data Table Generated by the OUTVERIFICATION= Option

The MEAN= output data table contains the following variables:

RowID

index of the rows.

ParmValue

parameter value of the means for the initial normal-gamma distributions.

VarName

name of the variable that corresponds to the variable in the MODEL statement.

VarID

order of the variables as they appear in the MODEL statement.

ParmName

name of the parameter.

ParmID

order of the parameters as they appear in the local-level simultaneous graphical dynamic linear model in the section Simultaneous Graphical Dynamic Linear Models.

PARENTALSET Statement

OUTVERIFICATION= Data Table Generated by the Statement

The OUTVERIFICATION= output data table contains the following variables:

RowID

index of the rows.

Variable_i

value of the adjacency matrices for the ith variable, i equals 1 comma ellipsis comma m, where m is the number of variables in the MODEL statement.

VarName

name of the variable that corresponds to the variable in the MODEL statement.

VarID

order of the variables as they appear in the MODEL statement.

Last updated: July 09, 2026