.-
help for ^reswage^ - 1.0 - 1/1/05
.-

Estimates reservation wage model using Breen's (1996) likelihood function -------------------------------------------------------------------------

^reswage^ ^depvar^ ^varlist1^, SELect(^varlist2^)[dummy(^censor_indicator^) > ]

where

^depvar^ is the dependent variable that is censored for a subset of the sam > ple

^varlist1^ is the set of independent variables thought to affect ^depvar^

^varlist2^ is the set of independent variables thought to affect reservatio > n wages

^censor_indicator^ is the dummy variable indicating if an observation is ce > nsored

^aweights^, ^fweights^, and ^pweights^ are not allowed in this version of ^ > reswage^

Description -----------

^reswage^ uses maximum likelihood and Breen's likelihood function (in ^reswage_ > ll.ado^, see "Remarks" below for bibliographic information) to estimate a selection regr > ession model with endogenous selection. ^reswage^ provides an alternative to the stan > dard Heckman two-step or MLE selection models, with the primary distinction being ^r > eswage^ relaxes the assumption of the exogeneity of the sampling mechanism; e.g., wage > offers influence the likelihood of employment and hence the odds of observing wages.

Sample selection models such as Heckman are notoriously difficult to identify, > and ^reswage^ is no exception to this general rule. The likelihood function as spe > cified in ^reswage_ll.ado^ is identified via the assumption of no error correlation be > tween the error terms in the equations predicting observed wages and reservation wage > s. In models where there are no good instruments to adequately differentiate between > these two equations, there will likely be difficulties in getting ^reswage^ to conver > ge.

Options -------

dummy(^censor_indicator^) specifies which cases have censored values of ^depvar > ^. If this option is not utilized, a dummy variable will be created for you on the > basis of whether ^depvar^ is missing.

Remarks -------

For detailed information on the underlying likelihood function, see REGRESSION > MODELS: CENSORED, SAMPLE-SELECTED, OR TRUNCATED DATA. Richard Breen, Sage University P > aper series on Quantitative Applications in the Social Sciences, series no. 7-011, T > housand Oaks, 1996. The likelihood function is provided on pp. 55-57.

Examples --------

. ^use http://www.stata.com/data/jwooldridge/eacsap/mroz^ . ^reswage lwage educ exper, select(nwifeinc kidslt6 kidsge6 age educ)^

or equivalently

. ^use http://www.stata.com/data/jwooldridge/eacsap/mroz^ . ^reswage lwage educ exper, sel(nwifeinc kidslt6 kidsge6 age educ) dummy(inlf > )^

.- Authors: John Reynolds, Department of Sociology, Florida State University, john.reynolds > @fsu.edu Karen Frank, Department of Sociology, Florida State University, kc04d@fsu.edu