I have to build a multivariate regression model as a project for my course in time series analysis. I have decided to model the industrial production index of the USA. I am stuck on the selection of the regressors which can adequately explain the time series of the industrial production. I have already selected some regressors (producer price index, unemployment, interest rates, population, auto and truck sales, inflation, philly fed seasonal adj. index, PMI) but I think I need some more regressors to correctly specify the model.

Can you suggest me some macroeconomic time series I can use as a regressor for the industrial production index? I need data from 1960-70 to 2016.


1 Answer 1


Depending on the goal of your analysis, you may want to take a different approach; but, often, more regressors do not necessarily improve a model. Rather than throwing in as many potentially relate variables into your regression, you may want to start with a more theoretical approach to try an determine the underlying supply side and demand side factors that drive industrial output.

It seems to me that the regressors you have included can be broken down in this way:

  • Supply Side: producer price index and interest rates
  • Demand Side: population, inflation, interest rates, unemployment, and seasonal adjustment
  • Other: auto and truck sales and PMI

While I am sure that the "Other" regressors are correlated with your dependent variable, do you really think they are determinants?

That said, if you were to include a lagged term for auto sales you could make the argument that the more cars/trucks bought in the last period would influence the demand for gas and electric in the current period. Additionally, you may have an argument for the PMI, but my point is that you should try and think of the underlying variables driving the supply and demand for industrial production.

Given the above comments, I would say you could add these potential variables:

  • Supply Side: Government investment in infrastructure, gross fixed capital formation (for the relevant industries), oil prices
  • Demand Side: Per capita income, manufacturing exports, oil prices

I am sure there have also been many studies like this in the past, and the previous literature is always a good place too look for additional ideas in the modeling process.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.