Data as of 10 August 2026   Rates move — verify before acting

Where in the world can you earn dollars?

A map of USD-denominated yield available to individuals — through banks, brokers, insurers, and DeFi. The organising question is not what does it pay but who is allowed to reach it: whether you can open it yourself, need a licensed agent, are capped by a capital-control quota, or are shut out entirely.

Fed funds target
3.50–3.75%
Held at Jan & Mar 2026 meetings
3-month T-bill
~3.60%
The risk-free floor
SGOV 30-day SEC
3.59%
Self-serve benchmark, 6 Aug 2026
Above this line
= risk
Credit, convertibility, duration, funding or smart-contract
Access Rate Class
Route
Yield vs. risk-free
Safety
How to read this

Four things that surprise people

01

The best-paying products are the ones you can't buy yourself

Every Asian participating-savings policy, every US MYGA, every offshore wrapper is agent-gated. The genuinely one-click routes cluster at the bottom of the yield range. That gap is the structural opening for a self-serve product.

02

High local USD rates are compensation for redenomination risk

Egyptian banks pay ~6% on dollar CDs because they need dollars. Argentina pesified deposits in 2001; Lebanon's pre-2019 balances are still frozen six years on. A "USD" deposit can be converted, capped or frozen.

03

Pendle is a market, not a yield source

It splits someone else's yield-bearing token into principal and yield legs. A PT's "fixed APY" is a price traders set. Buy PT-sUSDe and you hold Ethena's solvency risk plus Pendle's contract risk — not Pendle's credit.

04

No DeFi protocol offers a guaranteed floor

Hong Kong par plans guarantee up to 4.40%. US MYGAs guarantee 5–6.3%. Nothing on-chain guarantees anything, because there's no balance sheet behind the promise. Every DeFi rate here is best-efforts.

Access gradient

The five ways a dollar can be out of reach

Colour on this page does one job: it encodes how gated a route is. Rate tells you what you'd earn; the access code tells you whether you can get there at all.

SS — Self-serveOpen an account and buy directly. US, most of Western Europe, the USD-pegged Gulf, most DeFi.
BK — Broker-gatedA licensed agent, broker or private banker is legally required. All Asian par insurance, MYGAs, PPLI, offshore wrappers.
CC — Quota-cappedLegal but capped. India LRS US$250k, China SAFE US$50k, South Africa R11–12m.
PR — RestrictedProhibited or effectively blocked. Vietnam caps USD deposit interest at zero by law.
SC — Stablecoin-onlyCrypto rails are the practical route because the banking system doesn't offer one.
Method

What the yield bar means

Each bar plots the route's yield on a shared scale, with the vertical rule marking the 3.59% risk-free benchmark. Bars are compressed above 10% so outliers stay readable — that tail is not linear. Bars turn orange above 8%, the point past which nearly every figure is leveraged, incentivised, or paid for real convertibility risk.

A solid baris a rate somebody is contractually obliged to pay. A hatched baris not promised — it is an illustration, a projection, or a rate that simply floats. Only 12 of the 50 disclosed yields here are solid.

This is the difference between Egypt's 6.00% certificate of deposit, which is a contract, and AIA GlobalFlexi's 6.50%, which is a projected internal rate of return thirty years out. Plotted as bare numbers they are indistinguishable, and the projection is the higher of the two. The same treatment marks every DeFi rate on the page, because nothing on-chain is promised by anyone.

Where a rate is genuinely unknowable from public sources — offshore wrappers whose return depends entirely on the underlying, Plume vaults that don't publish live APYs — the field reads not disclosed rather than carrying an invented number.

Method

What the safety score means

A yield on its own can't tell you whether you're being paid enough. The score sits beside it as a second axis, out of 12, where higher is safer, so the two can be read together. Open any row to see the full derivation.

It scores the instrument's structure, not our opinion of the provider. Five facts are recorded for every route — what stands behind the dollars, whether any of the return is contractually promised, who is able to change the rate, whether dollars can leave the jurisdiction, and what it costs to get out — and the total is arithmetic over those five. Nobody assigns a row a number. If you think a score is wrong, one of the five lines in the breakdown is the thing to argue with, and it is a fact that can be checked and corrected.

Two rules shape the rest. Convertibility caps everything at 4: a strong balance sheet is no comfort in a jurisdiction that has frozen or redenominated dollars before, so the other axes cannot compensate. And leverage is flagged, not scored, because a leveraged return is not a rate and comparing it to one is a category error.

Three rows carry no score at all. They describe a route to instruments — a Brazilian brokerage, a South African allowance, a Chinese quota — rather than a single instrument, so the inputs cannot be pinned down. They read unscored rather than taking a middling guess, on the same principle that leaves 13 yields at "not disclosed".

A high score is not a recommendation and a low one is not a warning: several low-scoring routes are the only dollar savings their holders can reach, which is the point the access colours make. This is an editorial framework for comparison, not a credit rating, and it is not investment advice.