Count the Third Year: Sanctions Base Rates Versus the Timelines Politicians Announce

A sanctions package is announced with a stated objective and, usually, an implied clock. The clock is the part worth interrogating. Decades of case data show the economic effect and the political effect running on different scales, and the relationship between elapsed time and the odds of success is not the one the announcement language implies. The intuition behind "give it time to bite" is that pressure accumulates until it crosses a threshold. The record points the other way: episodes that end in a policy change have tended to resolve early, and duration has been a marker of failure rather than of pressure still building.

That is a claim about base rates, checkable against published data. What follows sets out the indicator flow, the interpretation, then the conditions under which the reading breaks down.

The Base Rate Table That Precedes Any Individual Case

The reference dataset is the Peterson Institute for International Economics compilation published as Economic Sanctions Reconsidered (third edition, 2007), coding 174 case studies encompassing 204 observations from the First World War onward. Its headline finding: sanctions were judged at least partially successful in 34 percent of cases. That is a deliberately generous bar, crediting a modest contribution to a partly realised objective rather than a diplomatic victory. The distribution by stated objective is more informative than the aggregate.

Stated objective	Share at least partially successful
Modest policy change	51%
Impairment of military capacity	31%
Regime change or democratisation	31%
Other major policy change	30%
Disruption of a military adventure in progress	21%
All cases	34%

The gradient is consistent: the narrower the demand, the higher the historical hit rate. The objectives that generate the most forceful announcement language — reversing an ongoing military operation, changing who governs — sit at the bottom of the table.

Two further figures bear directly on the timeline question. Successful episodes averaged 4.4 years, failures 8.4 years. And the cost imposed on the target economy averaged 3.3 percent of GNP in successes against 1.6 percent in failures — roughly double. Duration and severity move in opposite directions with respect to outcome: longer episodes were not more successful, they were less.

The Three-Year Mark and What It Actually Measures

A CEPR analysis of the same database sharpens this into a usable heuristic: roughly two-thirds of successful episodes reached their stated goal within three years, while roughly two-thirds of failed episodes ran well beyond it. The same analysis reports a structural break around 1990 — post-1990 episodes show a success rate near 40 percent against roughly 32 to 34 percent for 1946 to 1989, with average concluded duration falling from about 7.8 years to about 4.1 years.

The three-year mark is a selection effect, not a mechanism. Regimes that work tend to work while three conditions still hold together: the coalition is freshly assembled and cohesive, the target has not yet rebuilt its logistics and payment channels, and its buffers have not yet been restructured around the constraint. All three erode with time, in the same direction. Past that point the regime is no longer the instrument that was imposed, though the legal text is unchanged.

Two conditioning variables explain much of the variance. Where bilateral trade with the coalition was 10 percent or more of the target's GDP, the success rate approached 50 percent; under 2 percent, the failure rate was about 80 percent. Success against a democratic target was roughly twice as likely as against an autocratic one. Neither is a lever available after the fact — both are fixed at imposition, so much of the outcome distribution is set before day one.

For scale, the Global Sanctions Data Base codes 729 episodes from 1950 to 2016, extended to 1,045 through 2019. Its Iran work estimates that sanctions cut Iranian trade with the sanctioning countries by roughly 55 percent — a very large economic effect alongside a political outcome that took a further decade to negotiate and was later reversed.

The Maintenance Curve, Read Through One Instrument

The clearest contemporary evidence on how a regime changes shape comes from an instrument with a published parameter and a public revision history: the G7 and EU price cap on Russian seaborne crude. Because it is a number rather than a prohibition, its revisions leave an audit trail.

Date	Parameter	Design change
December 2022	USD 60.00 per barrel	Fixed level, tied to G7 shipping and insurance services
Effective 3 September 2025	USD 47.60 per barrel	EU 18th package; introduces a dynamic rule
Set 15 January 2026, effective 1 February 2026	USD 44.10 per barrel	First application of the dynamic rule; 90-day wind-down for prior contracts

The dynamic rule is specified: the cap sits 15 percent below the average market price for Urals crude over the preceding 22-week reference period, reviewed every six months. The design change is the tell — a fixed cap replaced by an automatic ratchet is an admission that a static parameter loses grip as the priced spread moves.

Whether the ratchet binds is a separate question, and the market data is unambiguous about direction. Tracking by the Centre for Research on Energy and Clean Air puts the July 2026 average Urals price at USD 60.22 per barrel — some 37 percent above the USD 44.10 cap — with a discount to Brent of roughly 26 percent, about USD 21 per barrel. That month, sanctioned shadow-fleet tankers carried 62 percent of Russian crude, against 34 percent on G7-affiliated vessels and 5 percent on unsanctioned shadow tonnage. The cap constrains cargoes that touch coalition shipping and insurance; it does not constrain cargoes that no longer do.

The counterpoint carries equal weight and sits in the same dataset. The persistent 26 percent discount is a real transfer away from the target — at roughly USD 21 per barrel, a standing tax collected by the freight and intermediation layer the constraint created. The picture also differs sharply by product: in July 2026 G7-affiliated tankers still carried 72 percent of Russian refined products, and Russian product loadings fell to 4.7 million tonnes, less than half the 9.6 million tonnes of July 2025. Products are harder to re-route than crude because the buyer set is narrower and specification-bound. The instrument leaks where substitution is cheap and holds where it is not.

Listings as a Coverage Ratio, Not a Score

The second observable is the listing count, routinely reported as a measure of pressure and better read as a measure of the coalition's chase. The European Commission's 20th package, adopted 23 April 2026, is specific enough to decompose.

Component	Figure in the 20th package
Shadow-fleet vessels listed	46 added, 11 delisted, 632 total
Russian banks	20 added, 70 total
Energy-sector entities	36
Military-industrial companies and suppliers	58
Anti-circumvention listings	60, of which 28 in third countries
General additional listings	120 (33 individuals, 83 entities)
New export / import ban values	over EUR 365m / EUR 530m

Two rows carry most of the signal. The shadow-fleet total moved from above 400 in the 18th package (July 2025) to 632 by April 2026. And 28 of the 60 anti-circumvention listings sit in third countries, alongside four third-country financial operators and an oil terminal. Those entries are not the original target; they are intermediaries that appeared because the restriction created a margin worth intermediating.

A rising count is therefore ambiguous — consistent with a coalition closing gaps faster than they open, and equally consistent with one chasing a frontier it cannot get ahead of. The distinguishing evidence is the derivative: whether the priced spread widens while the count rises. On crude over the past year the discount has been broadly stable as the count rose sharply, which favours the second reading.

Enforcement intensity is also not monotonic in time. Tracking by the Center for a New American Security records 3,135 additions to the Specially Designated Nationals list in 2024 against 1,764 in 2025, with Russia-related additions at 74 persons and 38 previously designated persons removed that year. Intensity is a policy choice re-made continuously; it can fall as well as rise while the legal architecture stays in place.

Two Long Arcs Against the Base Rate

The Iran arc is the more instructive because it contains a resolution. UN Security Council Resolution 1696 issued the initial demand on 31 July 2006; Resolution 1737 imposed the first UN sanctions on 23 December 2006, with four further resolutions through June 2010 (1747, 1803, 1835, 1929). Resolution 2231, endorsing the negotiated agreement, was adopted on 20 July 2015 — roughly 8.6 years after the first binding measures, double the 4.4-year average of historical successes. That outcome was then unwound: the E3 notified their snapback assessment on 28 August 2025, the prior measures returned at the end of the 30-day window that September, and the EU Council reimposed its own restrictive measures on 29 September 2025. The full arc runs about 19 years from first demand back to the original architecture.

The Cuba arc has no resolution. Proclamation 3447 was signed on 3 February 1962, took effect on 7 February 1962, and has now run 64 years. The contrast worth holding is that Iran's decade-long arc did eventually produce a negotiated outcome, which a strict three-year heuristic would not have predicted. Base rates describe distributions, not paths, and the tail is populated.

Signals That Would Force a Rethink

Several conditions would make the framework above misleading if applied mechanically.

• Selection bias in the case databases is severe. Episodes get coded when they become visible enough to code. Quiet successes — a threat that changed behaviour before imposition, a narrow financial action resolved without publicity — are underrepresented. The 34 percent figure is a floor on a partially observed sample, not a population parameter.
• Coding "success" requires attributing causation. Assigning a policy change to sanctions rather than to battlefield developments, commodity cycles or leadership turnover is a case-by-case judgement, which is why published ranges are wide.
• Denial objectives run on a different clock from coercion objectives. A measure designed to raise cost and lead time on a procurement channel has no behavioural threshold to cross. Judging an export-control regime by whether policy changed is a category error; the relevant metrics are unit cost and lead time, and those are rarely public.
• Post-2001 financial measures transmit differently. Correspondent-banking and settlement restrictions work through private compliance incentives rather than state enforcement, which is why over-compliance recurs. The historical database is weighted toward trade-era episodes and may understate this channel's speed.
• Asset immobilisation is a distinct case. Roughly EUR 194 billion of immobilised Russian central bank assets sat at Euroclear as of June 2025, about 85 percent of that institution's balance sheet, with an EU windfall contribution of 99.7 percent on the extraordinary revenues from 2024. That is a balance-sheet fact with litigation attached, not coercion on a behavioural timeline.

What Could Shift the Balance

The framing carries specific failure modes, each with an observable that would flag it.

1. A genuine step change in enforcement reach. If secondary measures start removing third-country intermediaries faster than they appear, the substitution frontier stops moving and the erosion behind the three-year effect weakens. The observable is whether the discount to benchmark widens while the count rises.
2. Coalition contraction rather than fragmentation. The failure mode usually discussed is members diverging. The less-discussed one is a coalition narrowing its own target set while keeping the architecture — visible in delistings and a falling annual designation rate, both present in the 2025 data.
3. Buffer exhaustion arriving non-linearly. Reserve drawdown, fiscal deficits and import compression can look stable for years and then move quickly. A slow-moving series is not evidence of a slow-moving process; the base rate does not rule out a late threshold, only says late thresholds are rarer than announcements imply.
4. Commodity price regime shift. A percentage-based cap mechanically tracks the market. If the benchmark falls, the cap falls with it and the nominal constraint tightens with no policy decision behind it — which should not be read as a policy tightening.

What to Watch Next Week

• Scheduled price-cap review dates. Reviews run every six months against a trailing 22-week Urals average. The number to watch is not the new cap level but the realised discount to Brent in the following month — the test of whether the change bound.
• The third-country share of listings in the next package. A rising share means the chase is moving further from the original target; a falling share at a stable count would be the first evidence of the frontier being closed rather than followed.
• Delistings, reported separately from additions. Net figures conceal direction. The 20th package added 46 vessels and delisted 11; whether such removals are enforcement-driven or policy-driven is the informative variable.
• Carrier composition by product. The gap between the crude share on non-coalition tonnage and the product share on coalition tonnage is the cleanest proxy for where substitution is cheap.

Concrete Framework — What to Track, in Order

A checklist for reading a sanctions regime against the historical distribution rather than its announcement.

1. Classify the operative demand before assessing progress. Place it in the goal table above: a regime-change or military-reversal objective starts in a 21 to 31 percent historical band, a narrow policy demand near 51 percent. Announcement language is not the classification.
2. Record the imposition date and mark the three-year point. Past that mark the burden of proof shifts. The question stops being whether pressure is accumulating and becomes what would place this episode in the minority tail.
3. Estimate pre-imposition trade linkage as a share of the target's GDP. Above 10 percent the historical success rate is near 50 percent; below 2 percent the failure rate is near 80 percent. Fixed at day one, and the highest-information variable available before any outcome data exists.
4. Track the priced spread, not the listing count. Discount to benchmark, freight and insurance premia and settlement costs are the transmission channel. Counts measure administrative activity; spreads measure whether the constraint binds.
5. Separate coverage ratio from coverage level. Ask what share of the relevant flow is captured, not how many entities are named. The July 2026 split — 62 percent of crude on sanctioned shadow tonnage, 72 percent of products on coalition tonnage — is a ratio; 632 listed vessels is a level.
6. Hold a symmetric prior on the reverse case. The same data showing duration correlated with failure also shows a cost transfer sustained over years, a halving of re-routable product volumes, and one decade-long arc that ended in a negotiated agreement.

Base rates constrain expectations without determining outcomes. A regime past its third year is not thereby failing, and one inside it is not thereby working. The supported claim is narrower: elapsed time should not be read as accumulated pressure. In the recorded distribution it has more often marked the absorption channels winning.

This article is macroeconomic and geopolitical analysis, not investment or policy advice, and describes historical distributions rather than forecasts of any specific case.