# Organizing data using time series multivariate regression?

I am trying to understand how I can organize the following data since none of what I learned in my undergrad econometrics course works. I am running out of ideas.

I am trying to measure how the impact of a construction boom affects rural migration into the city. The two equations I have tried estimating so far are:

1. Gross output of construction projects (measured in dollars) would be a function of the number of people employed at year's end, labor productivity (dollar/person) and a dummy for a specific government policy.

2. Year end population would be a function of per capita living consumption (in dollars), per capita residential floor space in welfare facilities, number of beds in welfare facilities, expenditure per capita for food, unemployment at end of the year, and per capita income of rural residents.

The data I have comes in two year increments (i.e. one value for each one of the above mentioned variables for, say, 2011-2012) and I have data for each of those variables above for 22 districts.

So far I have tried multivariable regression to some success but the fact that I have to run so many regressions on Excel doesn't seem like its very efficient and there's no way to see all this progressing on a time scale which makes my analysis quite difficult.

Can any experienced econometricians please give me some advice (besides find an easier topic) for working with this? Specifically I would like to know 1) how to deal with one value given per two years and 2) how to convey 22 districts worth of information over 8 years in a way that does not involve running too many regressions. Thank you in advance!

• Can you clarify your question beyond "give me some advice"? Are you trying to run your analyses in Excel? You included the [time-series] tag, do you actually have data over time, or just data for 22 districts for the 11-12 time period? Oct 20, 2014 at 2:34
• Yes, I am trying to do it in Excel and I have data for 22 districts over the 2004-2005, 2006-2007, 2008-2009 and 2010-2011 periods. I edited my question as well. Oct 20, 2014 at 2:43
• That's sort of time-series, but [panel-data] is a better description (& tag). From a practical standpoint, the appropriate models cannot be fit in Excel. If you aren't familiar w/ more sophisticated software (R, Stata, etc), you will need to try a different topic (sorry) or work w/ a statistical consultant. Oct 20, 2014 at 2:49