{smcl}
{* 24feb2004}{...}
{hline}
help for {hi:grouping}
{hline}

{title:Calculate grouping indicator for linked employer-employee data}

{cmdab:grouping} {it:newvarname} {cmd:,} {cmdab:i:var(varname)} {cmdab:j:var(varname)}

{title:Description}

{p} {cmd:grouping} calculates a unique "group" number for each firm
and each individual in a linked employer-employee dataset.  Each firm
is indexed by the {cmd:jvar} variable, and each individual is indexed
by the {cmd:ivar} variable.  The variable {it:newvarname} stores the
resulting group numbers.

{p} A group contains all the individuals who have ever worked for any
of the firms in a group, and all the firms at which any of the workers
were employed.  Thus, in most reasonable cases, the first group will
contain almost all workers and firms. To be in a separate group a firm
must have employed no workers who ever worked for any firm in another
group.  A firm which experiences no turnover will be in a group of its
own.

{p} The algorithm for computing the groups is based on that given by
Abowd, Creecy & Kramarz (2002).


{title:Authors}

Richard Upward
School of Economics
University of Nottingham
richard.upward@nottingham.ac.uk

Thorsten Schank
Volkswirtschaftliches Institut 
Friedrich-Alexander-University Erlangen-Nuremberg
thorsten.schank@wiso.uni-erlangen.de


{title:References}

{p} Abowd, J., Creecy, R.H. and Kramarz, F. (2002) "Computing person
and firm effects using linked longitudinal employer-employee data" US
Census Bureau Technical Paper TP-2002-06.