Creating a change variable

[Geert Van]

Hey guys,

I'm trying to add change variables but it doesn't really work. For the variable coc I need the percentage change (year-(year-1)/(year-1)*100), and for the variables irq, at, beta, lev and bm I need the absolute change (year-(year-1)). Note that every firm (firm id) has its own time series.
Does anybody know the easiest way to create these variables? Showed below are the first rows of my data-set.

Any help would be appreciated. Thanks!

firm id	year	coc	irq	at	beta	lev	bm	csr
1	2011	-	0	-	-	-	-	0
1	2012	-	0	3618.793	-	0.405952	-	0
1	2013	-	0	6613.581	-	0.314353	0.216467887	0
1	2014	0.2352727	5.21	11573.5	1.04333	0.376943	0.514705297	0
1	2015	0.097794	26.8	14155.22	1.88716	0.332634	1.206526532	0
3	2010	-	86.67	4946.5	-	0.162438	0.442065995	0
3	2011	0.1431113	85.08	5323.9	-	0.143072	0.436034294	1
3	2012	0.1606975	88.63	5627.1	-	0.13993	0.449940363	1
3	2013	0.1445998	86.61	5736	-	0.15713	0.365684741	1
3	2014	0.100875	89.72	6325.5	-	0.169915	0.441926384	1
3	2015	0.1558223	85.14	5932.2	-	0.187839	0.584596788	1

[Geert Van]

After In posted this, I kept trying to find the right command.

I tried to create a lag variable for all the variables, so I can subsequently create the change variable.

I tried to do this with the following command:



xtset firmid year
gen laggedcoc = L1.coc
gen laggedirq = L1.irq
gen laggedsize = L1.size
gen laggedlev = L1.lev
gen laggedbeta = L1.beta
gen laggedbm = L1.bm

This only worked for the variable irq because this variables has no missing values.
For the other variables the command didn't create any lag variables. Why does it not create the lag variables for the values that are not missing? What changes do I have to make in the command to get all the lag variables?

Thanks in advance.

[William Lisowski]

Welcome to Statalist.

Based on what you've shown us of your data, the variables other than irq seem to have "-" indicating a missing value. Since Stata represents missing values for numeric variables using a single period character, or a period followed by a lower-case letter, this suggests the data were imported into Stata as string variables rather than as numeric variables.

First use

describe

to confirm this, and then use commands like the following to convert your string variables to numeric.

destring coc, force replace

You should review the output of help describe and especially help destring to understand what these commands do.

Once you successfully convert your data to numeric, commands like the following should do what you seek.

xtset firmid year
generate coc_pctchg = 100*(coc-L1.coc)/L1.coc
generate irq_chg = irq-L1.irq

or perhaps

xtset firmid year
generate coc_pctchg = 100*(D1.coc)/L1.coc
generate irq_chg = D1.irq

For this, help tsvarlist provides useful guidance.

[Geert Van]

Hi William,

Thank you so much for your help.

I indeed copied the data from excel into stata, and therefore a wrong indicator for a missing value was present.

After converting my data into numeric, the first command did exactly what I was looking for, thanks again!!