Previous Chapter Chapter 13: Instrumental Variables Regression Next Chapter
Chapter 13: Instrumental Variables Regression
Because of the presence of unobserved variables, social scientists are often unable to use standard methods such as stratification, matching, or regression to estimate the causal effect of a treatment D on an outcome Y. Instrumental variables (IV) regression provides one potential solution to this problem. IV regression involves finding an exogenous variable, the instrument Z, that affects Y only through the causal variable D. The researcher identifies the causal effect of D on Y by estimating the portion of variation in the outcome Y associated with the treatment D that is attributable to the exogenous variable Z.
In this chapter we introduce IV regression. Our discussion is based in part on the presentation in chapters 3 ...