The MODEL Procedure
The MODEL procedure provides parameter estimation, simulation, and forecasting for a system of one or more nonlinear equations.
Model definition, parameter estimation, simulation and forecasting can be performed interactively in a single SAS session or models can be stored in files and reused and combined in later runs.
The features of the MODEL procedure include
-
tools to analyze the structure of a simultaneous equation system
-
nonlinear regression analysis for systems of simultaneous equations, including weighted nonlinear regression
-
estimation and simulation of systems of ordinary differential equations
-
a full range of nonlinear and iterated parameter estimation methods, including
-
NLS
-
NLSUR
-
NL2SLS
-
NL3SLS
-
iterated SUR
-
iterated 3SLS
-
GMM
-
FIML
-
-
ARIMA, PDL, and other dynamic modeling capabilities
-
Moore-Penrose generalized inverse to generate a parameter covariance matrix for singular estimation problems
-
profile likelihood confidence intervals on parameter estimates
-
dynamic multi-equation nonlinear models of any size or complexity
-
vector autoregressive error processes and polynomial lag distributions for the nonlinear equations
-
goal-seeking solutions of nonlinear systems to find input values needed to produce target outputs
-
dynamic, static, or n-period-ahead forecast simulations
-
Monte Carlo simulation using parameter estimate covariance and across-equation residuals covariance matrices or user specified random variables
-
a variety of diagnostic statistics:
-
model R²
-
Durbin-Watson
-
exact p-values reported for generalized Durbin-Watson
-
asymptotic standard errors and t tests
-
first stage R²
-
covariance estimates
-
collinearity diagnostics
-
simulation goodness-of-fit
-
Theil inequality coefficient decompositions
-
Theil relative change forecast error measures
-
Chow test
-
Hausman's specification tests
-
-
block structure and dependency structure analysis for the nonlinear system
-
automatic calculation of needed derivatives using exact analytic formulas
-
efficient sparse matrix methods for model solution
-
tests for nonlinear functions of the parameter estimates
-
nonlinear restrictions (equality and inequality restrictions) on the parameter estimates
-
bounds on the parameter estimates
-
heteroscedasticity tests, normality tests, and autocorrelation tests
You can now use expressions on the left side of the equal sign to write the model equations, and you can specify the lag length as a variable rather than a constant. In addition, functions are available to compute moving averages from lagged values. For general form equations (unnormalized), the SEIDEL and JACOBI methods are available.
Statistics and Operations Research Home Page | SAS/ETS Software