{smcl}
{* documented: January 31, 2006}{...}
{* revised: February 26, 2006}{...}
{cmd:help mtreatnb} {right:also see:  {help postestimation commands}}
{hline}

{title:Title}

{cmd :mtreatnb} {hline 2} Multinomial treatment effects of a negative binomial regression model


{title:Syntax}

{p 8 14 2}
{cmd:mtreatnb} {depvar} [{indepvars}] {ifin} {weight} {cmd:,}
	{opt mtreat:ment(depvar_mt indepvars_mt)} {opt sim:ulationdraws(#)}
	[{it:{help mtreatnb##mtreatnb_options:options}}]

{synoptset 20 tabbed}{...}
{marker mtreatnb_options}{...}
{synopthdr :options}
{synoptline}
{synopt :{opt base:category(#)}}value of {it:depvar} that will be the base 
	category.{p_end}
{synopt :{opt pre:fix(string)}}allows you to choose a prefix other than
	{cmd:_I} for the indicator variables{p_end}
{synopt :{opt r:obust}}specifies the robust or sandwich estimator of variance{p_end}
{synopt :{opth cl:uster(varname)}}adjust standard errors for intragroup 
	correlation{p_end}
{synopt :{opt sca:le(#)}}allows you to choose the standard deviation of the
	normally distributed quasirandom variables; default is 1{p_end}
{synopt :{opt start:point(#)}}allows you to choose the starting point in the
	Halton sequence from which the quasirandom variates are generated; 
	default is 20{p_end}
{synopt :{opt altfac:tors(string)}}allows you to choose the starting values 
	for the parameters associated with the latent factors.{p_end}
{synopt :{opt altst:art(string)}}allows you to choose the starting values 
	for all parameters{p_end}
{synopt :{it:{help mtreatnb##mtreatnb_maximize:maximize_options}}}control the 
	maximization process; some options may be useful{p_end}
{synopt :{opt ver:bose}}allows you to display iteration logs and estimates
	tables for the mixed multinomial logit and nb regressions{p_end}
{synoptline}
{p2colreset}{...}
{p 4 6 2}{it:depvar}, {it:indepvars}, {it:depvar_mt}, and {it:indepvars_mt} 
	may contain time-series operators; see {help tsvarlist}.{p_end}
{p 4 6 2}{cmd:fweight}s, {cmd:pweight}s, {cmd:iweight}s, and {cmd:aweight}s 
	are allowed; see {help weight}.{p_end}
{p 4 6 2} See {help postestimation commands} for features available after 
	estimation.


{title:Description}

{pstd}
{cmd:mtreatnb} fits a treatment-effects model that considers the effects of an
endogenously chosen multinomial treatment on another endogenous count outcome,
conditional on two sets of independent variables.  The treatment variable is
modeled via a multinomial logit and the outcome via a negative binomial
regression.  The model is fitted using maximum simulated likelihood.  The
simulator uses Halton sequences.


{title:Options}

{phang}
{cmd:mtreatment(}{it:depvar_mt indepvars_mt}{cmd:)} specifies the variables
for the multinomial treatment equation.  {it:depvar_mt} must have more than
two and less than 10 categories.  This option is an integral part of
specifying the treatment-effects model and is required.

{phang}
{opt simulationdraws(#)} specifies the number of simulation draws per
observation and is required.  These draws are based on Halton sequences.

{phang}
{cmd:basecategory(}{it:#} or {it:string}{cmd:)} is the value or label of
{it:depvar} that will be the base category in the multinomial treatment
equations.

{phang}
{opt prefix(string)} lets you choose a prefix other than {cmd:_I} for the
indicator variables created from the multinomial treatment variable.  The
default is a set of indicator variables starting with {cmd:_I}.  When you use
{cmd:mtreatnb}, it drops all previously created indicator variables starting
with the prefix specified in the {cmd:prefix()} option or with {cmd:_I} by
default.

{phang}
{opt robust} uses the robust or sandwich estimator of variance.  The default
is the traditional calculation based on the information matrix.

{pmore}
The standard errors are robust by construction, i.e, they are of the
Huber-White form.  This form is provides correct standard errors when maximum
simulated likelihood estimation is used.

{phang}
{opt cluster(varname)} adjusts standard errors for intragroup correlation.

{phang}
{opt scale(#)} lets you choose the standard deviation of the normally
distributed quasirandom variables. The default is {cmd:scale(1)}.

{phang}
{opt startpoint(#)} lets you choose the starting point in the Halton sequence
from which the quasirandom variates are generated.  The default is
{cmd:startpoint(20)}.

{phang}
{opt altfactors(string)} lets you choose the starting values for the
parameters associated with the latent factors. Specify these values as
comma-separated numbers.  The default starting values are zero.

{phang}
{opt altstart(string)} allows you to choose the starting values for all
parameters.  Specify these values as comma-separated numbers

{marker mtreatnb_maximize}{...}
{phang}
{it:maximize_options}: {opt dif:ficult}, {opt tech:nique(algorithm_spec)},
{opt iter:ate(#)}, [{cmdab:no:}]{opt lo:g}, {opt tr:ace},
{opt grad:ient}, {opt showstep}, {opt hess:ian},
{opt shownr:tolerance}, {opt tol:erance(#)}, {opt ltol:erance(#)},
{opt gtol:erance(#)}, {opt nrtol:erance(#)}, {opt nonrtol:erance}; see
{help maximize}.  

{pmore}
Because {cmd:mtreatnb} has a complicated likelihood function, {opt difficult}
may be a useful option if the default setup fails.

{pmore}
In addition, {opt altfactors(string)} and {opt altstart(string)} may be useful
to generate alternative starting values if the default setup fails.  Note that
if {opt altstart} is used, the intermediate mixed multinomial logit and nb
regressions are not estimated.


{title:Example}

{phang}{cmd:. mtreatnb docvis age, mtreat(instype1 age firmsize) sim(50)}


{title:References}

{p 4 8 2}Deb, P., and P. K. Trivedi. Forthcoming. Specification and simulated 
	likelihood estimation of a non-normal treatment-outcome model with 
	selection: application to health care utilization. 
	{it:Econometrics Journal}.

{p 4 8 2}Gouriéroux, C., and A. Monfort. 1996. 
	{it:Simulation-Based Econometrics Methods}. Oxford: Oxford University 
	Press.


{title:Author}

	Partha Deb, Hunter College, USA
	partha.deb@hunter.cuny.edu


{title:Also see}

{psee}
Online:  {help postestimation commands}; {helpb treatreg}; {helpb nbreg}; 
	{helpb mlogit}
{p_end}