Economic sanctions work roughly a third of the time. The largest systematic study, covering 174 cases, judged sanctions at least partially successful in about 34 percent of episodes. Whether that counts as working depends heavily on what the sanctions were trying to achieve, because the success rate varies enormously by goal.

The honest answer is neither the dismissal that sanctions never work nor the assumption that pressure eventually tells. It is that they work under specific conditions that are often absent in the cases that attract the most attention.

The Headline Number Is About One in Three

The standard reference is the Peterson Institute for International Economics database assembled by Hufbauer and colleagues, covering 174 case studies and 204 observations running through 2007.

Episodes are scored from 1 to 16, combining how far the policy goal was achieved with how much the sanctions themselves contributed. A score of 9 or higher counts as at least partially successful, and about 34 percent of cases clear it.

What Success Means in That Statistic

This is where most citations of the number go wrong, and the authors are careful about it.

A score of 9 does not mean sanctions achieved a foreign policy triumph. It means only that sanctions made a modest contribution to a goal that was partly realized. The 34 percent figure is a measure of partial contribution to partial success, not of targets capitulating.

Success Depends Enormously on the Goal

Aggregate rates conceal the most useful finding. Sanctions are far better at small asks than large ones.

• Modest policy changes: 51 percent success.
• Regime change and democratization: 31 percent.
• Military impairment: 31 percent.
• Disrupting minor military adventures: 21 percent.

The pattern is intuitive once stated. Asking a government to release a prisoner or adjust a policy is a cost-benefit calculation. Asking it to dissolve itself is not, because no economic pressure outweighs the survival of those making the decision.

What Makes Sanctions More Likely to Work

Several conditions separate the successes from the failures, and most are visible before sanctions are imposed.

Economic leverage. Trade linkage is decisive. Where bilateral trade is 10 percent or more of the target’s GDP, the success rate is about 50 percent. Where it is under 2 percent, the failure rate is around 80 percent. Successful cases imposed costs averaging 3.3 percent of the target’s GNP, against 1.6 percent in failures.

Regime type. Sanctions against a democracy are roughly twice as likely to succeed as sanctions against an autocracy. Autocratic leaders can insulate themselves from economic pain and pass it downward.

Prior relationship. Sanctions against former allies with cordial prior relations do better. Against antagonistic targets, success falls to about 19 percent.

Speed and breadth. Comprehensive packages combining financial, export and import measures succeed around 40 percent of the time, and imposing them quickly beats escalating in increments, which gives the target time to adapt.

What Makes Them Fail

Duration is the clearest warning sign, and it runs opposite to intuition. Successful episodes lasted 4.4 years on average. Failures lasted 8.4 years. Around two thirds of successful sanctions achieved their goal within three years.

Long-running sanctions are not pressure slowly accumulating. They are usually evidence that the pressure did not work and that the target has adjusted, found new trading partners or restructured around the restrictions.

Offsetting assistance from a rival power undoes the leverage directly. So does a target economy with little exposure to the sender in the first place.

Are They Getting Better?

Somewhat. Comparing periods in the same database, success rates rose from around 32 to 34 percent before 1990 to roughly 39 to 40 percent afterwards, an improvement of about a fifth.

The usual explanation is the post-Cold War shift toward targeted measures aimed at named individuals and entities, such as asset freezes and travel bans, rather than economy-wide embargoes. The UN describes its own regimes as including arms embargoes, travel bans, financial restrictions and commodity restrictions, imposed under Article 41 of the Charter, which covers enforcement measures that do not involve the use of armed force.

The Measurement Problem

One caution belongs on every number above. Judging whether sanctions caused an outcome means comparing the world as it happened with a world that did not occur.

Scholars disagree about how cases in this database are coded, and the 34 percent figure has been contested in both directions. Sanctions also serve purposes that a success rate does not capture, including signaling resolve, constraining a target’s capabilities rather than changing its mind, and satisfying a domestic demand to do something short of war.

The defensible conclusion is narrower than either side of the public argument. Sanctions can work, they work best on limited goals against economically exposed and politically accountable targets, and the longer they run the less likely they are to be the reason anything changes.