{smcl}
{* *! version 0.1.0  23jul2007}{...}
{cmd:help bitobit}
{hline}

{title:Title}

{p2colset 5 21 23 2}{...}
{p2col :{hi:[R] bitobit} {hline 2}}Bivariate Tobit regression
{p2colreset}{...}


{title:Syntax}

{p 8 16 2}
{cmd:bitobit estimate}
{ifin}
{cmd:,}
{cmd:y1}({it:{help varname:depvar1}}) {cmd:y2(}{it:depvar2}{cmd:) x1(}{it:{help varlist}}{cmd:) x2(}{it:varlist}{cmd:) censor1(}{it:varname}{cmd:) censor1(}{it:varname}{cmd:)} [{opt l:evel(#)}]

{title:Description}

{pstd}
{cmd:bitobit} fits a two equation seemingly-unrelated model of the {it:y1} variable on the {it:x1} variables and the {it:y2} variable on the {it:x2} variables, where the censoring status is determined by the {it:censor1} and {it:censor2} variables.


{title:Options}

{pstd}{opt estimate} Any arbitrary string of letters or numbers could be substituted for the word "estimate"; an argument before the comma is required so the syntax command recognizes this as an estimation command and not a replay command.

{pstd}{opt y1} designates the dependent variable in the first equation.

{pstd}{opt y2} designates the dependent variable in the second equation.

{pstd}{opt x1} designates the independent variables in the first equation.

{pstd}{opt x2} designates the independent variables in the second equation.

{pstd}{opt censor1} designates the variable that indicates the censoring status of the first equation.  For observations where this variable takes on a value of zero, the actual value is equal to the observed value of y1 (no censoring).  For observations where this variable takes on a value of one, the actual value is above the observed value of y1 (left censoring).  For observations where this variable takes on a value of negative one, the actual value is less than the observed value of y1 (right censoring).

{pstd}{opt censor2} designates the variable that indicates the censoring status of the second equation.  For observations where this variable takes on a value of zero, the actual value is equal to the observed value of y2 (no censoring).  For observations where this variable takes on a value of one, the actual value is above the observed value of y2 (left censoring).  For observations where this variable takes on a value of negative one, the actual value is less than the observed value of y2 (right censoring).

The behavior of this routine is undefined if censor1 or censor2 contain
values other than -1, 0, or 1.

{title:Examples}

{phang}{title:Fixed censoring threshhold}

{phang}{cmd:. generate fridge_censor = 0}

{phang}{cmd:. replace fridge_censor = -1 if refrigerators == 0}

{phang}{cmd:. generate wash_censor = 0}

{phang}{cmd:. replace wash_censor = -1 if washing_machines == 0}

{phang}{cmd:. bitobit estimate, y1(refrigerators) x1(temperature food_price interest_rate GDP) y2(washing_machines) x2(interest_rate GDP clothes_price dry_cleaning_price) censor1(fridge_censor) censor2(wash_censor)}

{phang}{title:Variable censoring threshhold}

{phang}{cmd:. bitobit estimate, y1(age_entering_league) x1(height weight family_income) y2(age_leaving_league) x2(height weight family_income) censor1(hasnt_joined_league_to_date) censor2(hasnt_left_league_to_date)}

{title:Author}

{phang}{browse "http://users.drew.edu/dlawson":Daniel Lawson}, Drew University, dlawson@drew.edu

{title:Version}

{phang}This is version 0.1 released August 3, 2007.