Date:         Fri, 29 Dec 2006 11:08:59 -0500
Reply-To:     Ken Borowiak <EvilPettingZoo97@AOL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Ken Borowiak <EvilPettingZoo97@AOL.COM>
Subject:      Re: Lag or array question

On Fri, 29 Dec 2006 10:20:50 -0500, Peter Flom <Flom@NDRI.ORG> wrote:

>Thanks!
>
>That worked perfectly
>
>Thanks to Wensui, as well.
>
>Peter
>
>>>> <Nathaniel_Wooding@Dom.com> 12/29/2006 10:14 am >>>
>Oops
>
>Forgot to take into the change between MSAs
>
>Nat
>
>DATA A;
 >INPUT MSA    Year             Unemp;
>CARDS;
>1         1990              5.1
>1         1991              6.
>1         1992              7
>1         1993              9
>2         1990              1
>2         1991              2
>2         1992              55
>Data b;
>   set a;
>   by msa;
>   if first.msa then count=0;
>   count+1;
>
>      lagUnemp=lag2(unemp);
>      if count le 2 then lagUnemp=.;
>
>proc print;
>run;
>
>
>
>             Peter Flom
>             <Flom@NDRI.ORG>
>             Sent by: "SAS(r)                                           To
>             Discussion"               SAS-L@LISTSERV.UGA.EDU
>             <SAS-L@LISTSERV.U                                          cc
>             GA.EDU>
>                                                                   Subject
>                                       Lag or array question
>             12/29/2006 10:01
>             AM
>
>
>             Please respond to
>                Peter Flom
>              <Flom@NDRI.ORG>
>

>
>Hello
>
>I have a data set that has (among other things) the following structure
>
>MSA    Year             Unemp
>1         1990              5.1
>1         1991              6.0
>
>.. (through 2002)
>2         1990               4.1
>2         1991                4.2
>
>.... (for 96 MSAs)
>
>I would like to create a variable of lagged unemployment by 2 years.  So,
>lagunep for 1992 would be unemp for 1990, for each MSA.
>
>I'm sure this is easy for the data step gurus, but I can't see how best to
>do it.
>
>(for context, I will then be using lagunemp in  a multilevel model using
>other variables in the data set.
>
>
>TIA and HNY
>
>Peter
>
>
>
>Peter L. Flom, PhD


In case you are interested in an SQL solution ...

proc sql ;
  select   T1.*,
           T2.Unemp as LagUnemp
  from     A T1
      left join
           A T2
           on T1.MSA=T2.MSA &
       T1.Year-T2.Year=2 ;
 quit ;

HTH,
Ken