I am looking for some techniques that would help me measure the (over-time) impact of a variable to another.

So let's say we have annual time series data for GDP for 5 countries and I wanted to see how much educational spending influences the GDP over a 20 year period. I understand the impact would not be instant so maybe a distributed lag model is required but what other methods can I use to measure that?

Relatedly, how would I measure the impact to GDP of a material increase of educational spending for one of the countries of interest?

Please assume all data is accurate and available and that I will be using R (in case you'd like to suggest some libraries).


  • $\begingroup$ SARIMAX autobox.com/pdfs/SARMAX.pdf is the general approach when you only have 1 dependent series . This is an ADL or PDL model which allow for purely contemporaneous structure , if adequate. $\endgroup$
    – IrishStat
    Mar 27, 2020 at 12:48


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.