calculating lag

[AbuChowdhury]

Thanks. I will try the code and let you know. I did not read the documentation at this moment. Will you please send me the link?

[data_null__ · Accepted solution]

I like this version a bit better.

data difasset2;
   set asset;
   by company_id;
   array _v

  ca   cl   stdebt;

   array _d[3,3] dca1 dcl1 dstdept1 dca2 dcl2 dstdept2 dca3 dcl3 dstdept3;
   do i = 1 to dim(_v);
      _d[1,i] = dif1(_v);
      _d[2,i] = dif2(_v);
      _d[3,i] = dif3(_v);
      end;
   if first.company_id then c=0;
   c + 1;
   do i = c to dim(_v);
      do j = 1 to dim(_v);
         call missing(of _d[i,j]);
         end;
      end;
   drop i c j;
   run;

[AbuChowdhury]

Hi again, I need to calculate ca at time t+1 as well. So, ca values at t-1, t and t+1.

[data_null__]

I should have guessed. :smileyplain:  I should have asked before now. Do you have SAS/ETS software?  Type in PROC EXPAND; RUN; and see what your messages are.  If that works consult the documentation.  It does lags leads and a lot more. 

[art297]

This sure sounds like a FIFO problem. What do you need to do with the lagged data once you get them?

The following keeps the current and up to the last three values for each variable and then calculates the average of the up to 4 values:

data want (drop=counter);

  set have;

  by companyISIN notsorted;

  retain counter .;

  array ca_stack {0:3};

  array cl_stack {0:3};

  array stdebt_stack {0:3};

  retain ca_stack cl_stack stdebt_stack;

  if first.companyISIN then do;

    call missing(of ca_stack{*});

    call missing(of cl_stack{*});

    call missing(of stdebt_stack{*});

    counter=0;

  end;

  counter+1;

  ca_stack{mod(counter,4)} = ca;

  cl_stack{mod(counter,4)} = cl;

  stdebt_stack{mod(counter,4)} = stdebt;

  if year not in (2003,2013) then do;

    ca_avg = mean(of ca_stack{*});

    cl_avg = mean(of cl_stack{*});

    stdebt_avg = mean(of stdebt_stack{*});

    output;

  end;

run;