st: Allowing different lag lengths by country in an error correction model with panel data

["Tejaswi Velayudhan (tvelayudhan@cgdev.org)" <TVelayudhan@cgdev.org>]

I have a country level dataset with data from the years in which a demographic health survey (DHS) was run in each country. This means it's sort of an unbalanced panel dataset, with the catch that the length of the gap between years of the series is different for each country. I would like to use this data to run an error-correction or auto-regressive distributed lag model but instead of controlling for y_t-1 and x_t-1 on the right hand side, I would be adjusting for y_t-k and x_t-k, where k varies by country. I have been looking for a way to adjust for the length of the gap (k), but surprisingly cannot find any literature on this or even papers that have dealt with this sort of data. Perhaps I am just not looking for the right keywords but I thought that someone on this listserv may be able to point me in the right direction. 

One possibility I have considered is imputing the t-1 value of the variable based on the t-k value with some discounting factor assumed as a function of k. Is there a more straightforward adjustment of the ECM or ARDL model that allows you to work with t-k instead of t-1? 

Thanks,

Teju

Tejaswi Velayudhan
Research Assistant
Center for Global Development


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/