Economics is not causal. It is epistemic.

Economists are trained to hunt for causes. Identification is the discipline's proudest word: instruments, discontinuities, differences-in-differences — an impressive machinery for isolating the effect of X on Y while the world holds still. I use this machinery, I teach it, and I have no intention of abandoning it.

But after two decades of measuring poverty and advising the institutions that respond to it, I believe the machinery answers a narrower question than we admit. A causal estimate is a statement about a world held constant. Policy happens in a world that is being continuously re-described — by statistical offices, ministries, manifestos, and headlines. Most of what economics does, most of the time, is not discovering causes. It is organizing belief.

Consider something as mundane as a poverty line. In work with Stephan Klasen on Chile, we documented that monetary and multidimensional measures increasingly disagree about who the poor are — and that this divergence is anything but random. Two instruments, both defensible, quietly select different citizens for visibility. Nothing was caused; something was constituted.

The measurement did not photograph poverty. It drew its boundary.

Or take the flypaper effect, that old embarrassment of fiscal federalism: transfer money to a local government and it sticks where it lands, defying the fungibility our models promise. The estimates matter, and in forthcoming work I try to get them right. But what makes the effect consequential is what mayors, councils and finance ministries believe a transfer is for. Institutions are, in a rather precise sense, crystallized beliefs about money.

This is what I mean by epistemic. Models are not mirrors; they are devices for coordinating what a society is able to think. When we say a subsidy “reaches” seventy percent of the energy poor, we have also, silently, defined the energy poor as those the instrument can see. The unseen do not appear in the error term. They do not appear at all.

None of this is an argument against rigor — it is an argument for taking rigor more seriously. If measurement constitutes the visible, then careless measurement is not merely imprecise; it is a quiet act of exclusion. The correct response is not less quantification but more self-aware quantification: methods that state whom they can see, and institutions that remember their categories are revisable.

That capacity for revision — call it epistemic health — is, I have come to think, the scarcest resource in public policy. It is also a proper object of economic research: how beliefs form in complex systems, how they harden into budgets and statutes, and what it costs a society when they can no longer be updated.

So — economics is not causal; it is epistemic. The slogan is deliberately too strong. But between a profession that overclaims causality and one that understands itself as the custodian of a society’s categories, I know which I would rather have advising a ministry.