Computing indices from panel data

[Assogba Hodonou]

I have a panel Data from 1995 to 2021 for 14 products. I want to normalize var1 by its value in 2008 across time. How can I do that?

[Andrew Musau]

* Example generated by -dataex-. For more info, type help dataex
clear
input float year str6 product float x
2006 "invest"  317.6
2006 "kstock"    2.8
2006 "mvalue" 3078.5
2007 "invest"  391.8
2007 "kstock"   52.6
2007 "mvalue" 4661.7
2008 "invest"  410.6
2008 "kstock"  156.9
2008 "mvalue" 5387.1
2009 "invest"  257.7
2009 "kstock"  209.2
2009 "mvalue" 2792.2
2010 "invest"  330.8
2010 "kstock"  203.4
2010 "mvalue" 4313.2
end
format %ty year

isid product year
bys product: egen val2008= max(cond(year==2008, x, .))
gen wanted= x/val2008

Res.:

. sort product year

. l, sepby(product)

     +----------------------------------------------+
     | year   product        x   val2008     wanted |
     |----------------------------------------------|
  1. | 2006    invest    317.6     410.6   .7735022 |
  2. | 2007    invest    391.8     410.6   .9542133 |
  3. | 2008    invest    410.6     410.6          1 |
  4. | 2009    invest    257.7     410.6   .6276181 |
  5. | 2010    invest    330.8     410.6   .8056502 |
     |----------------------------------------------|
  6. | 2006    kstock      2.8     156.9   .0178458 |
  7. | 2007    kstock     52.6     156.9   .3352454 |
  8. | 2008    kstock    156.9     156.9          1 |
  9. | 2009    kstock    209.2     156.9   1.333333 |
 10. | 2010    kstock    203.4     156.9   1.296367 |
     |----------------------------------------------|
 11. | 2006    mvalue   3078.5    5387.1   .5714577 |
 12. | 2007    mvalue   4661.7    5387.1    .865345 |
 13. | 2008    mvalue   5387.1    5387.1          1 |
 14. | 2009    mvalue   2792.2    5387.1   .5183123 |
 15. | 2010    mvalue   4313.2    5387.1   .8006535 |
     +----------------------------------------------+

For a review of this technique, see Section 9 of https://www.stata-journal.com/articl...article=dm0055

[Assogba Hodonou]

Thank you so much, Andrew.