Enhancements in SAS/ETS® 13.1 Software


SAS/ETS 13.1 introduces many new estimation features, including Bayesian options, limited dependent models with endogenous regressors, compound distribution models, a new data access engine, and many other enhancements. SAS/ETS has also refactored its vector autoregressive procedure, PROC VARMAX, which now estimates parameters at far greater speeds and with more precision than ever before.

Multivariate Bayesian Estimation

The QLIM procedure provides users with Bayesian estimation methods for all models that the procedure supports. PROC QLIM enables users to estimate models of limited dependent variables such as the following:

These models can now be estimated using Bayesian methods. The main features include the following:

Predictive Analysis of Ticket Sales

Limited Dependent Variable Models with Endogenous Regressors

The QLIM procedure now supports full information maximum likelihood (FIML) estimates for structural models that include endogenous regressors. These endogenous regressors can be continuous, discrete, censored, or truncated. Among these estimable models are the following:

PROC QLIM also supports a variety of endogeneity and overidentification tests that use these new FIML estimates.

FRED Interface Engine

Federal Reserve Economic Data (FRED) can now be accessed from a new SAS/ETS access engine utility. This utility conveniently enables you to dynamically query the most up-to-date information from the FRED database and include these data to enrich and enhance your models.

Multivariate Time Series Models

PROC VARMAX now estimates two new variants of multivariate GARCH models, known as GARCH-DCC and GARCH-CCC, and its internal infrastructure has been improved. In many cases, these new improvements enable more complex models to be estimated more quickly and with greater precision than ever before. These improvements enable you to put more complicated models into production.

Conway-Maxwell Poisson Model

The COUNTREG procedure now supports the Conway-Maxwell Poisson (CMP) model, which provides greater flexibility by allowing the scale parameter to be a function of regressors. The CMP allows for over- or underdispersion of outcomes, which ordinary Poisson would constrain. The CMP regression model can also be estimated using zero-inflated models.

Aggregate Loss

Compound Distribution Models

SAS/ETS 13.1 includes a new procedure for simulating compound distribution models (CDM): the HPCDM procedure enables you to simulate aggregate loss distribution models, which are useful in assessing operational risk and defining appropriate capital requirements in situations in which rare but catastrophic events occur. The HPCDM procedure combines results of two other SAS/ETS procedures, the COUNTREG and SEVERITY procedures, giving the modeler extreme flexibility with which to model the number of events and size of losses. This flexibility may take the form of distributional assumptions on the loss and count data models and different regressors in each model. The HPCDM procedure includes the following features:

Panel and Time Series Tests

A variety of new model specification tests have been added to the PANEL and AUTOREG procedures. These tests check the statistical assumptions of the models concerning stationarity, cointegration, and structural change, and they include p-values that are generated by high performance simulation methods. Many software packages report only selected critical values for these tests. The AUTOREG procedure now supports the following test statistics:

For panel data models, the PANEL procedure supports these new test statistics:

Other Enhancements to SAS/ETS

The QLIM procedure now supports the Heckman two-step estimator for sample selection models. This is an alternative to the maximum likelihood estimates that the procedure already supports. The consistent standard errors for this two-step estimator are automatically generated for the second-stage OLS parameter estimates.

The SEVERITY and HPSEVERITY procedures now support a new OUTSCORELIB statement to create scoring functions, which help simplify the computation of distribution functions with fitted parameter estimates. This eliminates the need of writing a complex DATA step that reads the estimates from the OUTEST= data set. Both procedures also now support a new OFFSET= option in the SCALEMODEL statement to model the scale parameter per unit which is a measure of exposure.

The COUNTREG procedure now supports STORE and enhanced CLASS statements. The STORE option enables previously estimated models to be used for out-of-sample prediction and other post-estimation routines. The procedure is also multithreaded for faster computation.

The system GMM estimator proposed by Blundell and Bond for dynamic panel models is now available in the PANEL procedure.

For More Information

For more information, ask your organizationís SAS representative to contact the SAS Customer Interaction Center at 1.800.727.0025.

Download a pdf version of this document.