Scanning an old piece by the ever-brilliant Ed Leamer. He says "Our understanding of causal effects in macroeconomics is virtually nil, and will remain so."
I have a lot of sympathy with this after my time on the MPC. It's worth remembering that the model is as he sets out
1. x is the treatment
2. y is the response
3. z are interactive confounders
4. w are additional confounders.
The treatment effect in macro we would like to estimate is .
As he says
"The big problem with randomized experiments is not additive confounders; he big problem with randomized experiments is not additive confounders; it’s the interactive confounders. This is the heterogeneity issue that especially t’s the interactive confounders. This is the heterogeneity issue that especially concerns Heckman (1992) and Deaton (2008) who emphasized the need to study oncerns Heckman (1992) and Deaton (2008) who emphasized the need to study “causal mechanisms,” which I am summarizing in terms of the interactive causal mechanisms,” which I am summarizing in terms of the interactive z variables"
"With interactive confounders explicitly included, the overall treatment effect ith interactive confounders explicitly included, the overall treatment effect β0 + β′ zt is not a number but a variable that depends on the confounding effects. s not a number but a variable that depends on the confounding effects. Absent observation of the interactive compounding effects bsent observation of the interactive compounding effects z, what is estimated is what is estimated is some kind of average treatment effect which is called by Imbens and Angrist (1994) ome kind of average treatment effect which is called by Imbens and Angrist (1994) a “Local Average Treatment Effect,”"
"absent observation of z, the estimated treatment effect should be transferred only into those settings in which the confounding into those settings in which the confounding interactive variables have values close to the mean values in the experiment."
"This is the error made by the bond rating agencies in the recent fi nancial nancial crash—they transferred fi rash—they transferred fi ndings from one historical experience to a domain in ndings from one historical experience to a domain in which they no longer applied because, I will suggest, social confounders were not hich they no longer applied"
