Chapter 7 - Predicting in a Nonlinear World

Click here to read the chapter (link works only for UC affiliates)

Lecture Slides:      Powerpoint     PDF

Learning Objectives

  • Articulate what it means to “get the model right”
  • Learn transformations that can turn a misspecified model into a well- specified model
  • Test for nonlinearities in an econometric model


  • Moving off the Farm

What We Learned

  • An econometric model will almost never be literally right in the sense that it describes precisely how one variable depends on a set of other variables.
    • A good econometric model is as simple as possible and fits the data well.
  • Taking logs of the variables, fitting a polynomial, adding interaction terms, and adding right-hand-side variables are easy ways to improve model specification.