xtset, gives me an error

[Alex Smilgins]

Hi!

I have the following dataset:

unit_number, Year, costs

I would like to calculate the difference in costs for each unit, and I would to do that for different time intervals. In the dataset I have years 2005-2015. Thus, if I define yearly changes, it works:

xtset unit_number Year, delta(1)
gen costs_lead=F.costs
gen diff_costs=costs_lead - costs

But then if I want the 3- year changes it gives me an error:

xtset unit_number Year, delta(3)
gen costs_lead=F.costs
gen diff_costs=costs_lead - costs

The error: "time values with period less than delta() found"

So, it is basically telling me that I can get yearly changes instead of 3-year changes. But I am interested in 3 year changes. I do not understand why this is an error and not a warning.

Please let me know how I should fix this.

Thanks!

[Carlo Lazzaro]

Alex:
what if you create a 3-year -timevar-?

[Alex Smilgins]

Alex:
what if you create a 3-year -timevar-?

I am not sure what you mean. Do you mean deleting some observations such that it does not see smaller time periods?

But I cannot do that, because I need the following:
if 2005 - then value from 2008 should be taken.
if 2006 - then value from 2009 should be taken.
if 2007 - then value from 2010 should be taken.
if 2008 - then value from 2011 should be taken.
if 2009 - then value from 2012 should be taken.
if 2010 - then value from 2013 should be taken. if 2011 - then value from 2014 should be taken.
if 2012 - then value from 2015 should be taken.
if 2013 - then value should be missing (no year 2016)
if 2014 - then value should be missing (no year 2017)
if 2015 - then value should be missing (no year 2018)

So, I need all the observations in the dataset.

[Nick Cox]

xtset unit_number Year, delta(1)
gen costs_lead3 = F3.costs
gen diff_costs = costs_lead3 - costs

[Alex Smilgins]

xtset unit_number Year, delta(1)
gen costs_lead3 = F3.costs
gen diff_costs = costs_lead3 - costs

It works, thank you! :-)