{smcl}
{* 31jan2006}{...}
{hline}
help for {hi:redpace}{right:Mark Stewart (31 Jan 2006)}
{hline}
{title:Maximum Simulated Likelihood Estimation of Random Effects Dynamic Probit Models with Autocorrelated Errors}
{p 4 12}{cmd:redpace} {it:depvar} {it:varlist} ({it:varsinit}) [{cmd:if} {it:exp}] [{cmd:in} {it:range}]
[{cmd:,} {cmdab:i(}{it:varname}{cmd:)} {cmdab:t(}{it:varname}{cmd:)} {cmdab:rep(}{it:#}{cmd:)}
{cmdab:seed(}{it:#}{cmd:)} {cmdab:s:eg(}{it:#}{cmd:)} {cmdab:h:alton} {cmdab:p:rimes(}{it:matname}{cmd:)}
{cmdab:d:rop(}{it:#}{cmd:)} {cmdab:f:rom(}{it:matname}{cmd:)} {cmdab:ma:vg} {cmdab:no:auto} ]
{title:Description}
{p}{cmd:redpace} estimates a Random Effects Dynamic Probit Model with Autocorrelated Errors by
Maximum Simulated Likelihood.
{title:Options}
{p 0 4}{cmd:i(}{it:varname}{cmd:)} specifies the variable name that contains the cross-section
identifier, corresponding to index {it:i}.
{p 0 4}{cmd:t(}{it:varname}{cmd:)} specifies the variable name that contains the time-series
identifier, corresponding to index {it:t}.
{p 0 4}{cmd:rep(}{it:#}{cmd:)} specifies the number of replications, or draws.
{p 0 4}{cmd:seed(}{it:#}{cmd:)} specifies the initial value of the pseudo-random number seed to
be used by the {cmd:uniform()} function. Use the {cmd:seed} option to ensure
reproducibility of results. The number specified must be an integer. The default is 81234567.
{cmd:seed} is ignored if {cmd:halton} is specified.
{p 0 4}{cmd:seg(}{it:#}{cmd:)} specifies symmetric systematic sampling (with antithetics) to be
used and specifies the number of segments of the unit interval that are to be used.
{cmd:seg(2)} corresponds to antithetic sampling. The default value is 1 (i.e. standard sampling).
If {cmd:seg} is specified, the number of replications must be a multiple of the number of segments,
and the number of segments must be a multiple of 2.
{p 0 4}{cmd:halton} specifies that a Halton quasi-random sequence is to be used rather
than pseudo-random numbers, which is the default. This option requires the program
{cmd: mdraws}.
{p 0 4}{cmd:primes(}{it:matname}{cmd:)} specifies a 1x(T-1) matrix
containing the prime numbers to be used for the Halton sequences. Ignored
if {cmd:halton} not specified. The numbers specified must be integers. It is
the user's responsibility to ensure that they are primes.
{p 0 4}{cmd:drop(}{it:#}{cmd:)} specifies the number of initial elements of the Halton sequence
to be dropped for burn in. Ignored if {cmd:halton} not specified. The default is 0.
{p 0 4}{cmd:from(}{it:matname}{cmd:)} specifies a matrix containing starting values for the
parameters of the model. Use this option to check that a global maximum has been found.
Also use to reduce required number of iterations or to restart a previously
halted run. The default uses a pooled probit
for {it:t>=2} and separate probit for the initial period reduced form.
{p 0 4}{cmd:mavg} specifies that the first-order moving average model should be used.
The default is the first-order autoregressive model.
{p 0 4}{cmd:noauto} specifies that the model without autocorrelated errors is to be estimated (by MSL).
This is useful for comparing with the Gaussian-Hermite quadrature estimates
of the same model to see if the number of replications chosen is large enough in
that case.
{title:Remarks}
{p}For fuller details of the estimator and an application, see Stewart (2005).
For more information on the {cmd:redpace} command, sample output and an illustration
and discussion of its usage see Stewart (2006).
{p}The lagged dependent variable must be constructed by the user and must appear
as the first variable in {it:varlist}. It is the user's responsibility to ensure
that both this variable and {it:depvar} are binary 0/1 variables.
{it:varlist} should additionally contain the explanatory variables in the main equation.
{it:varsinit} should contain the variables in the initial period reduced form.
{p}{cmd:redpace} requires a balanced panel, that is the number of "time"
observations for each cross-section unit must be the same. This is checked by
the program.
{title:Examples}
{p 8 12 2}{inp:. redpace y Ly x1 x2 (x1 x2 z1 z2), i(id) t(time) rep(500) seed(945430778) }
{p 8 12 2}{inp:. matrix pp = (3, 7, 11, 13, 17) }
{p 8 12 2}{inp:. redpace y Ly x1 x2 (x1 x2 z1 z2), i(id) t(time) halton primes(pp) drop(20) }
{title:Author}
{p 4 4}Mark Stewart, Economics Department, University of Warwick, U.K.{break}
{title:References}
{p 4 8} Stewart, M.B. (2005), "The inter-related dynamics of unemployment
and low-wage employment", forthcoming {it:Journal of Applied Econometrics.}
{p 4 8} Stewart, M.B. (2006), "Maximum Simulated Likelihood Estimation of Random Effects
Dynamic Probit Models with Autocorrelated Errors", forthcoming {it:Stata Journal.}
{title:Also see}
{p 0 19}On-line: help for {help est}, {help ml}.{p_end}