The contrapositive of the fallacy of composition

I wrote something (actually, I just bit a bunch of things Roderick Dewar wrote) over a year ago that kind of glossed over a major point.

We observe [a macroeconomic relationship]. Since this [relationship] is observed across many natural experiments [e.g. countries] where we have no control over the micro degrees of freedom [people] the microfoundations must be largely irrelevant.

There is a consequence of this.

If you think that policy X that focuses on macro observables a, b, c, ... will have the same beneficial effect in different countries (or at different points in history) -- that policy X has reproducible effects -- you must believe that agents don't matter.

This only applies if the set {a, b, c, ...} are macro observables, but that doesn't seem to be a big impasse to this applying to nearly all the various policies practiced or discussed:

Policy: Inflation (price level) targeting

Macro observable: Inflation rate (price level)

Policy: NGDP (level) targeting

Macro observable: NGDP growth rate (NGDP)

Policy: Interest rate targeting

Macro observable: Market interest rate

Policy: Money supply targeting

Macro observable: Money supply (whichever measure: MB, M2, ...)

Policy: Macro stabilization via fiscal policy

Macro observable: Output gap (effectively RGDP or NGDP)

What doesn't this apply to? Most so-called supply side reforms that e.g. impact a particular tax rate or regulation. However, the rationale behind those tends to rely on an assumption that when added up from the individual agent level they turn into a macro-observable result. Some refer to this the fallacy of composition: if you aggregate micro, then there isn't always a reproducible macro observable.

The fallacy of composition is really just the contrapositive of the previous principle. If there is a reproducible macro observable, then it's not the result of aggregating micro.