Department of Economics

Recent applied work in economics has displayed renewed interest in the problem of characterizing the causal relations that link economic variables. However, many discussions avoid explicit specification ofwhat has to be true about a formal model to justify an assertion that one variable in it causes another. Such specification is supplied here. Related topics, such as determining whether correlation implies causation, or vice-versa, and when causal coefficients can be estimated using ordinary least squares or instrumental variables regressions, are discussed.