dividing variable with lagged variable

[Sanaullah Farooq]

I am getting back to STATA after a very long time and is having some difficulties. I want to create a scaled CAPEX variable by dividing CAPEX (t) with Assets (t-1). It is an unbalanced panel data for which means that for every unique firm, the firm-level observations are not equal. My sample data looks like as follows:

* Example generated by -dataex-. To install: ssc install dataex
clear
input str18 firmid int time float(assets capex)
"0876711D LN Equity" 2010   788.6     2.3
"0876711D LN Equity" 2011     727      .5
"1218069D LN Equity" 2005 275.615  67.671
"1218069D LN Equity" 2006 717.403 116.276
"1218069D LN Equity" 2007 790.272 161.478
"1218069D LN Equity" 2008 903.164 192.963
"1218069D LN Equity" 2009 937.341 149.963
"1218069D LN Equity" 2010 910.465 108.437
"1218069D LN Equity" 2011 823.901  82.631
"1334987D LN Equity" 2008   212.8     1.6
"1334987D LN Equity" 2009   245.4     1.8
"1334987D LN Equity" 2010   223.2     4.5
"1561649D LN Equity" 2005 203.671 10.4137
"1561649D LN Equity" 2006 247.038  6.2588
"1561649D LN Equity" 2008 690.709 12.1927
"1561649D LN Equity" 2009  830.02 19.4424
"1561649D LN Equity" 2010  1565.9    30.5
"1561649D LN Equity" 2011  1343.3    41.5
"1561649D LN Equity" 2012  1488.1    22.6
"1561649D LN Equity" 2013  1271.1    21.6
"1561649D LN Equity" 2014  2122.2      26
end

[Tahseen Hasan]

This should do the trick.

*Create unique *numeric* IDs for your firms
egen fid = group(firmid)

*Ensure your data is properly sorted
sort fid time

*Create the lagged (t-1) value of asset based on specific firms
by fid: gen lag_assets = assets[_n-1] if time==time[_n-1]+1

*Create the scaled capex variable
gen scaled_capex = capex / lag_assets

[Clyde Schechter]

Here's a simpler way

xtset firmid time
gen scaled_capex = capex/L1.assets