Ask about the cost of FX hedging for the Brazilian real and the usual answer is the gap between the two central-bank rates: a Selic rate of 13.75% against a Federal Reserve range of 3.75–4.00%, so about 9.5 points a year. On 2 October 2026, B3’s curves priced a three-month hedge at 6.66% a year, mid-market. For a structure that pays dollar holders out of real-denominated assets, three points a year decide whether the product clears.
This article takes that price apart. It explains what the number is and why it is not a fee, who pays it and who collects it, why the market price sits so far below the policy-rate gap, and what the bank actually earns on top. The first article in this series showed how the forward rate is built from two interest rates. Here we put a price on it.
The cost of a hedge is a rate gap, not a fee
The cost of hedging is the forward premium of the dollar, annualized: (F / S)^(365 / days) − 1. On B3’s curves for 2 October 2026 that was (5.3103 / 5.2265)^(365/90) − 1 = 6.66% a year for 90 days.
Nobody writes the investor a bill for 6.66%. What the investor gives up is the gap between the two interest rates: money that would earn about 13% in reais earns about 6% in dollars once the currency is locked. The bank that sells the forward offsets its own position in the money markets—it borrows one currency and lends the other for the same term—so the 6.66% passes straight through it. What the bank keeps is the spread between the price at which it buys and the price at which it sells.
One price, two sides
Every forward has a buyer of dollars and a seller of dollars, and they face the same number with opposite signs. Whoever holds reais today and needs dollars later buys dollars forward and pays the premium: a foreign investor in real-denominated credit, a Brazilian company with a dollar loan, an importer with a dollar invoice. Whoever will receive dollars and needs reais sells dollars forward and collects it: above all, the exporter, who locks in more reais than today’s spot rate would pay.
| Position on 2 Oct 2026 | Hedge | Rate before the hedge | Rate after a 1-year hedge, at mid |
|---|---|---|---|
| Investor in a farm loan at 20% in reais | buys dollars forward | 20% in reais | 12.3% in dollars |
| Farmer with a dollar loan at 8% | buys dollars forward | 8% in dollars | 15.4% in reais |
| Exporter with dollar revenue due in a year | sells dollars forward | spot 5.2265 | forward 5.5801, 6.8% more reais over the year |
The first two rows are the same trade seen from two balance sheets. The conversions are:
- Y_USD — dollar yield of a real-denominated asset after hedging it into dollars (calculated)
- Y_BRL — the asset’s yield in reais, annual
- c — annual cost of hedging reais into dollars for the asset’s term
- R_BRL — cost in reais of a dollar loan after hedging it into reais (calculated)
- r_USD — the loan’s dollar interest rate, annual
- c — annual cost of hedging reais into dollars for the loan’s term
For a one-year hedge c is 6.86%, from B3’s one-year forward of 5.5801: 1.20 / 1.0686 − 1 = 12.3%, and 1.08 × 1.0686 − 1 = 15.4%.
The sign can flip
The rule is simple: whoever sells forward the currency with the higher interest rate pays. Against the dollar the real almost always carries the higher rate, so hedging reais into dollars almost always costs something. Almost: in November 2020, with the Selic at 2%, the three-month hedge into dollars cost only 0.6% a year on B3’s curves. Against currencies whose rates were then above Brazil’s, the sign reversed—hedging reais into South African rand paid the holder up to 4.2% a year (January 2021) and into the Chinese yuan up to 2.0%, as implied by B3’s curves (the cross-currency article covers these pairs). A hedge is a price for exchanging two interest rates, and it can be positive, zero, or negative.
From 9.5% to 6.66%: taking the price apart
The policy-rate gap and the market price differ for four reasons. On 2 October 2026 the steps were as below. These are prices at the close on 2 October 2026. On 5 October, after the first round of Brazil’s presidential election, the real strengthened about 4.6% against the dollar in a single day (PTAX 5.2238 → 4.9859), and the curves moved with it; the arithmetic did not. The decomposition, not the decimals, is the point.
| Step | Effect, points a year | Running total |
|---|---|---|
| Policy-rate gap: Selic 13.75% against the Fed range midpoint of 3.875% | start | 9.51% |
| Use the DI curve for 90 days (13.533%) instead of the Selic: the market expects cuts | −0.21 | 9.30% |
| Count business days: the real accrues on 60 of these 90 calendar days | −0.48 | 8.82% |
| Compound the dollar money-market rate to an effective annual rate | −0.12 | 8.70% |
| Use the onshore dollar rate inside Brazil, 5.77%, instead of the US rate | −2.04 | 6.66% |
| Half-spread plus markup, illustrative 0.0025 reais per dollar | +0.20 | 6.86% |
The first three adjustments are housekeeping, worth 0.8 points together. A hedge for a term must be priced off rates for that term, not off tonight’s policy rate. The day count matters more than it looks. DI accrues only on Brazilian business days, and the window from 2 October to 31 December 2026 contains four weekday holidays (12 October, 2 November, 20 November and 25 December). That leaves 60 accruing days out of 90, fewer than the 252-in-365 average, so the real earns less per calendar day in this quarter than its annual rate suggests.
The fourth adjustment is the big one, and it has a name.
Basis: why the dollar is dearer inside Brazil
Covered interest parity holds between rates that the same arbitrageur can actually earn. Inside Brazil the relevant dollar rate is the cupom cambial, the dollar interest rate implied by B3’s futures. On 2 October it was 5.77%, well above US money-market rates. Because the dollar rate sits in the denominator of the forward formula, a higher onshore dollar rate means a cheaper hedge: 6.66% onshore against about 8.7% at US rates. The gap between the onshore dollar rate and the offshore one is the cross-currency basis.
Over twenty years the onshore three-month dollar rate has sat above the US three-month rate almost all the time:
The median premium from September 2006 to August 2026 was 1.1 points; in four months out of five it stayed between 0.6 and 2.0 points. The spikes line up with moments when dollars inside Brazil became scarce or arbitrage became expensive. The largest, 6.7 points in April 2011, came days after Brazil imposed a 6% tax (IOF) on foreign borrowing with maturities first under one year and then under two (Chamon and Garcia). The only deep negative reading, in October 2008, came from the other side: US interbank rates spiked in the global crisis. The basis is not a curiosity for academics. Since 2008 deviations from parity have been persistent even in the largest currencies (Du, Tepper and Verdelhan, 2018), and in an emerging market they are a second, independent price. An IMF study measures the conventional one-year gap for the real, against LIBOR, at about 1.1 points on average, and finds that almost all of it disappears once Brazilian credit risk is stripped out—it is a price of risk, not arbitrage left on the table (Dao and Gourinchas, 2025).
Two practical consequences follow. A hedged real held inside Brazil earns more than Treasury bills in the United States. That premium is reachable only through the onshore market, which needs a local entity or a registered non-resident account with a Brazilian custodian. A foreign investor without onshore access hedges offshore, with a non-deliverable forward settled in dollars, and the dollar rate implied there can differ. Which instrument a structure can actually use is the subject of the next article in this series.
Spread: what the bank actually earns
A dealer quotes two forward prices: a bid, at which it buys dollars, and an ask, at which it sells them. The hedger pays half the difference relative to the mid-market price, plus whatever markup the bank adds for a corporate client.
- c_cl — hedge cost paid by a client buying dollars forward (calculated)
- F — mid-market forward rate, reais per dollar
- h — half of the bid-ask spread plus any client markup, reais per dollar
- S — spot rate, reais per dollar
- dc — calendar days to maturity
The effect depends on the term. A half-spread of 0.0025 reais per dollar—an illustrative level, not a quote—adds 0.20 points a year to a 90-day hedge, 0.58 points to a one-month hedge and 0.05 points to a one-year hedge. A structure that rolls a one-month hedge pays the spread twelve times a year. That is the real arithmetic of “hedging is expensive”: the rate gap is similar at every tenor, but the spread compounds with every roll.
How wide are spreads in practice? Brazil’s central bank has measured them. In quoted market prices (Bloomberg data, January 2017 to January 2019) the average bid-ask on the real was about 6 basis points, close to the euro’s 4 (BCB, Estudo Especial 41/2019). Companies pay more. On onshore forwards with non-financial firms, measured against B3 futures of the same maturity, the median buy-plus-sell spread at the end of 2018 was about 0.2%, and it widened for smaller tickets and longer tenors. On spot conversions the median buy-plus-sell spread was about 0.9%, so roughly 0.45% each way, and the smallest contracts paid around four times what the largest did (BCB, Estudo Especial 48/2019). The illustrative 0.0025 sits between the two. At the corporate forward level—a half-spread of about 0.005 reais per dollar, half of that 0.2%—the drag becomes 0.41 points a year on a three-month hedge, 1.2 points on a one-month hedge and 0.10 points on a one-year hedge.
Collateral, margin and tax
The price is not the only cost. A bank sells a forward only inside a credit line, under an ISDA Master Agreement or a local equivalent, and often against collateral posted under a credit support annex. On B3 there is no credit line; instead positions are settled against daily margin, and margin calls are paid in cash before the hedged asset pays anything. For an on-chain vault or a small SPV that liquidity requirement can be the binding constraint, more than the 6.66% itself.
Tax is the other line item. Brazil’s tax on financial operations, the IOF, is set by decree and changes often; the rates below are those in force in October 2026 under Decree 6,306/2007 as amended in 2025 (Planalto, consolidated text). FX derivatives—futures, forwards, swaps—pay 0%: the 1% levy introduced in 2011 was cut to zero in June 2013. Foreign investment into Brazilian financial and capital markets pays 0% on the currency conversion. A foreign loan with an average term of up to 364 days pays 3.5% on the conversion, which on a six-month loan is about 7 points a year. The same money can therefore cost nothing or several points a year depending on whether it enters as an investment or as a short loan, which makes the legal route part of the hedge cost. The 2025 amendments were suspended by Congress and partly reinstated by a Supreme Court injunction, so the current rates need checking deal by deal.
Price a hedge: calculator
The calculator uses B3’s curves for 2 October 2026 at four tenors. Pick a tenor, then move the rates, the spread, or the yields you want to convert.
Two experiments are worth a minute. First, lower the onshore dollar rate toward US levels and watch the hedge get more expensive: that is the basis at work. Second, switch between the one-month and one-year presets with the same spread: the amber add-on shrinks as the tenor grows.
Where hedge prices get misread
The policy-rate gap instead of the market price
On 2 October the policy-rate gap overstated the cost of a three-month hedge by 2.85 points: 9.51% against 6.66%. Any model that sets a hedge cost as “local policy rate minus US policy rate” builds in that error. The right input is the forward itself, or the two market curves for the same term.
Mixing conventions
DI compounds over 252 business days; the onshore dollar rate is simple on a 360-day year; most investors think in effective annual rates on 365 days. The same trade on 2 October produces four different “hedge costs” depending on the convention:
| How the same 90-day trade is annualized | Result |
|---|---|
| Forward over spot, compounded on 365 calendar days | 6.66% |
| Forward over spot, compounded on 252 business days | 6.91% |
| DI annual rate against the effective onshore dollar rate | 7.13% |
| DI rate minus quoted dollar rate, no conversion | 7.76% |
Only the first is directly comparable with a dollar yield. Comparing a hedge cost from one row with a yield from another is a 0.3–1.1 point error.
A three-month price for a twelve-month risk
A one-year loan hedged with three-month forwards pays today’s 6.66% for the first quarter and an unknown price for each of the next three rolls. The one-year forward on 2 October was 6.86%. The difference is small on that day; it is not small in a year when rates move, and the roll also re-pays the spread each time.
What this means for an RWA structure
A structure that promises dollar returns from real-denominated assets runs the conversion in reverse: required yield in reais = (1 + target dollar yield) × (1 + c) − 1. To pay 10% in dollars with a one-year hedge at 6.86%, the assets must earn about 17.5% in reais—before credit losses, servicing and the vault’s own fees. That is why hedged structures end up in the riskier parts of the local credit market, and why a “safe 10% in dollars” built on Brazilian assets deserves the question of which loans earn the missing 7.5 points. The alternative is to lend in dollars and leave the currency with the borrower, which moves the problem rather than removing it; our agri-debt adjudication prices that route and its yield calculator breaks it down layer by layer.
The hedge cost is a transfer, not a leak: what the investor gives up, the exporter on the other side collects. Getting it right is about pricing that transfer at the right tenor and in the right place, which is also where the four funding levers of an RWA deal meet the currency.
Python: reproduce the cascade and the convention table
# B3 curves, 2 October 2026, 90 calendar days = 60 business days
selic, fed_mid = 0.1375, 0.03875 # policy rates
di, cup = 0.13533, 0.0577 # DI "pre" curve (252, compound); onshore dollar rate (ACT/360, simple)
S, dc, du = 5.2265, 90, 60
brl_cal = (1 + di) ** (du / 252) # 1.030681 over the 90 days
brl_eff = brl_cal ** (365 / dc) - 1 # 13.04% effective annual
usd_eff = lambda r: (1 + r * dc / 360) ** (365 / dc) - 1
steps = [
("policy gap", (1 + selic) / (1 + fed_mid) - 1),
("DI curve", (1 + di) / (1 + fed_mid) - 1),
("business days", (1 + brl_eff) / (1 + fed_mid) - 1),
("USD compounding", (1 + brl_eff) / (1 + usd_eff(fed_mid)) - 1),
("onshore dollar (basis)", (1 + brl_eff) / (1 + usd_eff(cup)) - 1),
]
for name, v in steps:
print(f"{name:24s} {v:.2%}") # 9.51%, 9.30%, 8.82%, 8.70%, 6.66%
F = S * brl_cal / (1 + cup * dc / 360)
print(f"{(F / S) ** (365 / dc) - 1:.2%} {(F / S) ** (252 / du) - 1:.2%} "
f"{(1 + di) / (1 + usd_eff(cup)) - 1:.2%} {di - cup:.2%}") # 6.66% 6.91% 7.13% 7.76%
Pricing the currency layer of an RWA product?
We price hedges at the tenor and access point a structure actually has, model rolls and margin, and show what the hedged yield really is—before it goes into a term sheet or a vault's documentation.
Get in touchThe takeaway
The cost of hedging reais into dollars on 2 October 2026 was 6.66% a year for three months at mid-market on B3’s curves: the gap between Brazilian and dollar interest rates, measured on market curves, for the right number of days, against the dollar rate that actually exists inside Brazil. It is not a fee; it is what the hedger gives up and what the exporter on the other side collects. The policy-rate gap overstated it by almost three points; the onshore dollar premium cut it by two; the bank’s spread added a fraction that grows with every short roll. A hedged dollar yield is the local yield divided by one plus that number—so the honest way to read any hedged RWA product is to ask for the number, the tenor and the access point, and do the division.