SIMU_REGRESSION

SIMU_REGRESSION combines two or more existing regression models into a simultaneous regression system. Each equation is estimated separately, but the command reports the system as a unified model with cross-equation residual diagnostics.

Syntax

SIMU_REGRESSION(regression1, regression2, ...)

Example

r1 = REGRESSION(LIST(LOGDIFF(CPIAUCSL), LOGDIFF(PPIACO), LOGDIFF(M2SL)))

r2 = REGRESSION(LIST(LOGDIFF(PCE), LOGDIFF(CPIAUCSL), LOGDIFF(PAYEMS)))

s = SIMU_REGRESSION(r1, r2)

EX(s, slideshow)

Description

A simultaneous regression system is useful when several economic or financial relationships should be viewed together rather than as isolated regressions. The command displays each equation, model coefficients, fitted values, residuals, autocorrelation diagnostics, and system-level residual covariance and correlation matrices.

The residual correlation matrix is the key diagnostic. If residuals are strongly correlated across equations, a system estimator such as SUR or 3SLS may improve efficiency or better represent the structure of the model.

Output

• System overview
• System statistics
• Residual correlation matrix
• Residual covariance matrix
• Individual regression panels for each equation
• Regression slideshow output

Extracts

EX(s, slideshow)
EX(s, residual_correlation)
EX(s, residual_covariance)

Notes

SIMU_REGRESSION is currently a system wrapper around existing regression operators. It does not yet re-estimate the equations jointly. Future versions may support SUR, 2SLS, and 3SLS estimation using the same system structure.