Do They Work, and How Would You Know

Sanctions are the most used instrument in modern statecraft. They are also the least examined, and the reason is that the headline number everybody quotes is a definition rather than a finding.

Here is the number. The standard reference work, Hufbauer, Schott and Elliott's catalogue of sanctions episodes since 1914, counted roughly a hundred and fifteen cases and judged about forty of them successes. A third. That figure appears in briefing papers, op-eds and ministerial statements, and it has done more to sustain the instrument than any single case.

In 1997 Robert Pape published a paper in International Security called Why Economic Sanctions Do Not Work. He did not gather new data. He went back through the same forty successes and asked, of each one, whether sanctions had actually produced the concession.

His account of what he found: eighteen were settled by the direct or indirect use of force. Eight showed no evidence that the target made the demanded concession at all. Six were not instances of economic sanctions on examination. Three were indeterminate.

Five were left. Five out of a hundred and fifteen is under five per cent.

Kimberly Ann Elliott replied the following year, and her reply is the useful part of the exchange. She argued that Pape had mischaracterised the project: the catalogue was asking whether and how sanctions contributed to desired outcomes, not whether they were the sole cause. Pape responded that a contribution you cannot distinguish from zero is not a finding.

Both are right about their own question. The data are not in dispute; neither is most of the case detail. What is in dispute is the standard of proof, and shifting that standard moves the answer by a factor of seven.

That is worth sitting with, because it is a general warning about policy statistics. A success rate is never just a measurement. It encodes a decision about causation that is usually made once, in a methodological appendix, and then travels for thirty years as a number.

Now the practically useful part, which neither side disputes: the cases that survive scrutiny have a shape.

They involve a modest demand. Not regime change, not withdrawal from claimed territory, but a specific and bounded action the target can take while remaining itself.

They involve a target heavily dependent on the sanctioning party. One dominant trading partner, one source of a critical input, no ready alternative supplier.

And they tend to involve measures imposed quickly and at full strength, rather than ratcheted upward over years. A target given time adapts, finds new partners, builds domestic substitutes and acquires a political stake in not conceding.

The cases that fail cluster just as clearly: regime survival, territorial questions, nuclear programmes. These are matters on which a government will absorb almost unlimited economic cost, because the alternative is not being a government.

Which produces a rule that is close to the inverse of practice. Sanctions are an instrument for small asks of dependent states. They are most often deployed for large asks of determined ones, and then the modest results are read as evidence that more pressure was needed.