I am currently reading the Paper of Dippel et al. (2015), where the authors estimate the causal effect of imports on voting behavior in Germany. Therefore, they estimate the following first-difference specification, where Yit refers to electoral outcomes, NetExposure refers to import - export exposure, τtr are time-varying fixed effects and Xit is a of control variables:

enter image description here

My question is why should I include "undifferenced" controls (Xit) in the first difference model?

Thank you in advance for any answers!


1 Answer 1


You are allowed to mix differenced and undifferenced variables. Just because dependent variable is differenced does not mean all variables have to be.

Typically you difference variables because:

  • level variables have unit root
  • you actually want to study differenced variables

However, if a variable is stationary and you are interested in an effect of the variable on dependent variable then there is no reason to include it in first differences.


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.