Overview (2SLS command)

2SLS performs two-stage least-squares regression to produce consistent estimates of parameters when one or more predictor variables might be correlated with the disturbance. This situation typically occurs when your model consists of a system of simultaneous equations wherein endogenous variables are specified as predictors in one or more of the equations. The two-stage least-squares technique uses instrumental variables to produce regressors that are not contemporaneously correlated with the disturbance. Parameters of a single equation or a set of simultaneous equations can be estimated.

Options

New Variables. You can change NEWVAR settings on the TSET command prior to 2SLS to evaluate the regression statistics without saving the values of predicted and residual variables, or you can save the new values to replace the values that were saved earlier, or you can save the new values without erasing values that were saved earlier (see the TSET command). You can also use the SAVE subcommand on 2SLS to override the NONE or the default CURRENT settings on NEWVAR.

Covariance Matrix. You can obtain the covariance matrix of the parameter estimates in addition to all of the other output by specifying PRINT=DETAILED on the TSET command prior to 2SLS. You can also use the PRINT subcommand to obtain the covariance matrix, regardless of the setting on PRINT.

Basic Specification

The basic specification is at least one EQUATION subcommand and one INSTRUMENTS subcommand.

  • For each specified equation, 2SLS estimates and displays the regression analysis-of-variance table, regression standard error, mean of the residuals, parameter estimates, standard errors of the parameter estimates, standardized parameter estimates, t statistic significance tests and probability levels for the parameter estimates, tolerance of the variables, and correlation matrix of the parameter estimates.
  • If the setting on NEWVAR is either ALL or the default CURRENT, two new variables containing the predicted and residual values are automatically created for each equation. The variables are labeled and added to the active dataset.

Subcommand Order

  • Subcommands can be specified in any order.

Syntax Rules

  • The INSTRUMENTS subcommand must specify at least as many variables as are specified after WITH on the longest EQUATION subcommand.
  • If a subcommand is specified more than once, the effect is cumulative (except for the APPLY subcommand, which executes only the last occurrence).

Operations

  • 2SLS cannot produce forecasts beyond the length of any regressor series.
  • 2SLS honors the WEIGHT command.
  • 2SLS uses listwise deletion of missing data. Whenever a variable is missing a value for a particular observation, that observation will not be used in any of the computations.