Covered Interest Parity: The Forward Is Just Arbitrage

Last lesson you learned to read forward points off a screen — the tiny adjustment that turns today’s spot into the rate you can lock in for delivery in three months or a year. We treated those points as a given, a number the market hands you. This lesson rips the lid off. It turns out the forward rate isn’t a forecast, a vibe, or a dealer’s hunch about where the euro is headed. It’s pure arithmetic — pinned to the spot rate and the gap between two interest rates by exactly the same no-arbitrage logic that fixed the cost-of-carry forward in the futures topic. Buy a currency forward and you’re really just doing cash-and-carry in two money markets at once. By the end you’ll derive the forward yourself, build the riskless trade that enforces it, and understand why — since 2008 — the textbook formula stopped holding perfectly.

Throughout, we quote the exchange rate as domestic currency per 1 unit of foreign currency (e.g. USD per EUR), call the home interest rate $r_d$ and the foreign rate $r_f$.

What pins the forward rate?

Analogy. Imagine two savings accounts in two countries paying different interest, and a perfectly liquid currency window between them. If you could borrow cheaply in the low-rate country, swap into the high-rate one, earn the fat yield, and swap back with no exchange-rate risk, you’d have invented free money. The universe abhors free money. So the price of swapping back — the forward rate — must be set to exactly cancel the interest advantage. That cancellation is the forward rate. It isn’t a guess about the future spot; it’s the number that makes the two accounts tie.

Definition. The forward exchange rate $F$ is the price agreed today for exchanging currencies at a fixed future date. Covered interest parity (CIP) is the no-arbitrage condition stating that $F$ is determined entirely by the spot rate $S$ and the two interest rates $r_d$ and $r_f$ — “covered” because the forward contract covers (hedges) all the currency risk, leaving a fully locked-in outcome. There is nothing probabilistic about it.

This is the cost-of-carry argument wearing a passport. In the futures topic the fair forward was spot grown by the net cost of carry. Here the “carry” of holding a foreign currency is simply the interest it earns ($r_f$) versus the interest you forgo at home ($r_d$). Same skeleton, new costume.

When it holds

CIP is the tightest no-arbitrage relation in all of finance precisely because both legs — the two deposits and the FX trades — are observable, liquid, and (classically) executable with negligible friction. When you can borrow and lend at $r_d$ and $r_f$ and trade spot and forward freely, CIP holds essentially to the last basis point. The interesting question, saved for the end, is what happens when you can’t.

The two paths

Analogy. You have 1 unit of domestic money today and you want domestic money one year from now, with zero exchange-rate risk. Two roads lead there, and an arbitrageur insists they cost the same.

• Path A — stay home. Deposit your 1 unit at the domestic rate. One year later you have $1 + r_d$. Done.
• Path B — go abroad, covered. Convert your 1 unit into foreign currency at spot. Since $S$ is domestic-per-foreign, 1 unit of domestic buys $1/S$ units of foreign. Deposit that abroad at $r_f$, growing it to $\frac{1}{S}(1 + r_f)$ units of foreign. Today, lock a forward to sell that foreign money back at rate $F$. One year later you convert at $F$ and end with $\frac{F}{S}(1 + r_f)$ units of domestic.

Both paths start with the same 1 unit of domestic money and end in domestic money, with no currency risk in either (Path B’s forward nailed down the conversion the moment you opened it). If two riskless strategies deliver guaranteed amounts, those amounts must be equal — otherwise borrow via the cheap path, invest via the rich one, and harvest the difference forever. So:

That single equation is covered interest parity. Everything else is algebra.

The CIP formula

Take the two-path equality and solve for $F$. Starting from $1 + r_d = \frac{F}{S}(1 + r_f)$, multiply both sides by $S$ and divide by $(1 + r_f)$:

There it is — the covered interest parity forward rate. The forward is the spot, scaled by the ratio of the two gross returns. The low-rate currency’s return sits in the denominator, the high-rate currency’s in the numerator. (This is the per-period form, with $r_d$ and $r_f$ measured over the same horizon as the forward.)

For a forward of length $t$ years with annual rates and simple (money-market) accrual — the market convention for short tenors — scale each rate by the day-count fraction $t$:

A six-month forward uses $t = 0.5$; a 90-day one uses roughly $t = 90/360$ under the ACT/360 convention most money markets quote. The structure never changes — only how much of each annual rate you accrue.

The slider below is this formula made physical. Drag the domestic rate, the foreign rate, or the spot, and watch two things happen: the no-arbitrage forward recomputes from $F = S(1+r_d)/(1+r_f)$, and the two investment paths — stay-home versus go-abroad — always land on the identical ending wealth. They have to. That equality is the whole point.

Worked example

Take the pretest numbers and grind the arithmetic, using the simple per-period form for a one-year forward ($t = 1$):

Let’s sanity-check both paths on $1 of domestic money. Path A: $1 deposited at 5% becomes $1.0500. Path B: $1 buys $1/1.2000 = 0.8333$ EUR, which grows at 3% to $0.8333 \times 1.03 = 0.8583$ EUR, sold forward at 1.2233 for 0.8583 × 1.2233 = $1.0500. Identical to the cent. The forward of 1.2233 is the only rate that makes them tie — quote anything else and one path beats the other, and we’ll go pick that pocket in two sections.

Premium or discount — which currency?

Analogy. Two bank accounts, one paying more interest than the other. If they’re to be equally attractive, the higher-yielding one must come with a built-in penalty on the way out, and the lower-yielding one with a built-in bonus — otherwise everyone piles into the fat yield. In FX that penalty/bonus is the forward premium or discount. The low-interest-rate currency always trades at a forward premium; the high-rate currency at a forward discount. The forward giveth to the low yielder exactly what the interest rate tooketh away.

Definition. A foreign currency is at a forward premium when $F > S$ (it costs more domestic currency to buy forward than spot) and at a forward discount when $F < S$. From $F = S(1+r_d)/(1+r_f)$:

So $F > S$ — the foreign currency at a premium — exactly when the domestic rate is the higher one. The currency with the lower interest rate (here the foreign one, when $r_d > r_f$) is the one that appreciates forward. Put bluntly: the low-yield currency is dear in the forward market, and that premium is precisely the yield you give up by holding it.

Worked example

Re-use $S = 1.2000$ and flip the rate gap two ways to see both signs:

In the first row the EUR (foreign) earns less interest than the USD, so the forward bids it up (1.2233) to compensate — a premium. In the third row the EUR now out-yields the USD, so the forward marks it down (1.1660) to claw that extra yield back — a discount. The forward is a perfectly calibrated equalizer.

Covered interest arbitrage

Now the enforcement. The formula isn’t a courtesy; it’s policed by covered interest arbitrage — the FX twin of cash-and-carry. The recipe whenever the quoted forward strays from the parity forward: borrow the currency the market is implicitly overpaying you to hold, convert at spot, invest at the better rate, and lock the forward to seal a riskless gain.

Setup. Keep $S = 1.2000$, $r_d = 5\%$, $r_f = 3\%$, so the parity forward is $F^* \approx 1.2233$. Suppose a dealer instead quotes a forward of 1.2350 — richer than parity. That means the market is paying too much to convert EUR back into USD a year out. The fix: end up long forward EUR-selling… let’s just build it on $1,000,000 and watch the cash flows.

The over-priced forward overpays you to sell EUR forward, so you want to be holding EUR to sell. Route money the foreign way and dump the proceeds into that rich forward:

Walk the year-end column: the EUR deposit matures at 833,333 × 1.03 = €858,333; selling it forward at 1.2350 yields 858,333 × 1.2350 = $1,060,041; repaying the loan costs $1,050,000; the difference is $10,041, locked in, with no capital down and no exchange-rate risk (the forward fixed the exit). At the parity forward of 1.2233 this profit is exactly zero — selling at 1.2233 yields 858,333 × 1.2233 = $1,050,000, precisely the loan repayment. The $10,041 exists only because the quote of 1.2350 was off parity.

How the trade self-destructs. Doing this, every arbitrage desk on Earth (a) sells EUR forward in size, pushing the forward quote down toward 1.2233, (b) borrows USD, nudging $r_d$ up, (c) buys EUR spot, nudging $S$ up, and (d) lends EUR, nudging $r_f$ down. Every one of those moves shrinks the gap. The profit evaporates the instant the quote snaps back to parity — which, in liquid markets, is essentially instantly. Mirror the whole thing (quoted forward below parity ⇒ borrow EUR, convert, lend USD, buy EUR forward) to enforce the floor. Squeezed from both sides, the forward has nowhere to live but on the CIP line.

When CIP breaks: the cross-currency basis

Here’s the twist that separates the textbook from the trading floor. For decades — right up to 2008 — CIP held so tightly that any deviation was a rounding error; desks treated it as a law of nature. Since the global financial crisis, it doesn’t quite hold anymore. A persistent wedge has opened between the forward you can actually trade and the textbook parity forward, and it has a name: the cross-currency basis.

Definition. The cross-currency basis is the extra spread (often quoted in basis points on the foreign leg of an FX swap) needed to make the two sides of a covered trade actually balance in the real market. Algebraically, reality now reads $1 + r_d = \frac{F}{S}(1 + r_f + b)$, where $b$ is the basis — the fudge factor that shouldn’t exist if CIP were exact. For many currencies against the US dollar, $b$ has been reliably negative (a “dollar premium”): non-US institutions pay a markup to obtain dollars synthetically through FX swaps.

Why the free lunch goes uneaten — limits to arbitrage. The classic arbitrage assumed you could borrow and lend freely and cheaply in both currencies. Post-crisis that assumption broke:

• Balance-sheet is no longer free. Basel III leverage ratios and capital charges mean a bank’s balance sheet is a scarce, costly resource. Putting on a covered-arbitrage trade consumes balance-sheet capacity, and the dealer demands compensation for it. The “risk-free profit” isn’t free — it costs regulatory capital.
• Dollar funding is scarce. Foreign banks have huge dollar liabilities but no natural dollar deposit base, so when dollars get tight (quarter-ends, stress episodes) they’ll pay up via FX swaps to get them. That demand pushes the basis wider exactly when balance-sheet is most constrained.
• Counterparty and liquidity frictions. Credit limits, collateral terms, and bid-ask spreads all chip away at a gap that’s now only a handful of basis points to begin with.

Put together, these are limits to arbitrage: the trade still looks riskless on paper, but the capital it ties up and the funding it requires cost more than the gap pays — so rational arbitrageurs leave it on the table, and the deviation persists. CIP didn’t stop being a good first-order model; it stopped being exact. For this course, the headline is the right one: the forward is overwhelmingly the interest differential, with a small, real-world basis layered on top that you’d measure in basis points, not big figures.

Putting it together

Strip away the FX jargon and covered interest parity is the cost-of-carry argument you already know, applied across two money markets. The forward isn’t a forecast; it’s whatever rate makes “invest at home” tie with “convert, invest abroad, and lock the forward.” That equality hands you $F = S(1+r_d)/(1+r_f)$, and any quote off that line is a riskless profit waiting to be harvested by covered interest arbitrage — which is exactly why, in liquid markets, the line holds. The low-rate currency rides at a forward premium; the high-rate one at a discount; the forward is a yield equalizer, never a crystal ball. And the one real-world asterisk — the post-2008 cross-currency basis — is itself a no-arbitrage story: when the trade costs scarce balance-sheet and pricey dollars, a small persistent wedge survives. Here’s the whole argument on one card:

One mixed recap before we leave the forward behind:

Key Takeaways