Friends, every currency trade you have ever placed has an invisible floor underneath it. That floor is the interest rate differential between two central banks. You may not have been thinking about it when you clicked buy or sell. But the institutional desk on the other side of your trade was. Rate differentials are not the only thing that drives currencies. Volatility regimes, risk appetite, geopolitics, energy shocks, and capital flows all matter, as we have covered in the earlier parts of this series. But rate differentials are the structural driver. They set the baseline around which everything else oscillates. When the noise clears, when the headlines fade, when the positioning squeezes unwind, what remains is the rate differential. And the rate differential is remarkably patient.
The 3.5% Rule: What the Fed Measured
We introduced this finding in Part 3, but it is worth expanding here because it is the quantitative anchor for everything that follows. The Federal Reserve Board’s May 2024 FEDS Note, “Monetary Policy and Exchange Rates
during the Global Tightening,” measured the relationship between 2-year OIS rate differentials and bilateral exchange rates across seven advanced economy currency pairs during the 2021-2024 tightening cycle. Their finding: a 1 percentage point widening of the US-versus-foreign 2-year OIS differential generates approximately 3. 5% dollar appreciation on average. www.federalreserve.gov The sensitivities varied by pair. JPY showed the highest sensitivity at approximately 5. 0%. EUR, GBP, AUD, CAD, NOK, NZD, and SEK all showed significant but lower betas. The combined model (OIS differential plus VIX plus high-yield credit spread) achieved adjusted R-squared values of 0. 30 to 0. 42 for most currencies. About half of the 13% broad dollar appreciation during this period was directly attributable to relatively higher US interest rates. This is the structural floor. When the Fed is at 3. 50-3. 75% and the ECB is at 2. 00%, there is a 150-175 basis point differential supporting USD against EUR. That differential does not disappear because of a bad NFP print or a geopolitical headline. It persists until one of the two central banks changes course. And central banks move slowly.
How Each Bilateral Pair Maps to Its Rate Spread
The structural floor is not the same thickness for every pair. Each currency responds to a different rate differential, and each has a different sensitivity. Here is how the major pairs map.
EURUSD and the US-Germany 2-Year Spread
The European Commission’s Economic Brief 055 (McCoy, 2020) documents the dominant role of the 2-year interest rate differential in driving EURUSD. The brief cites Goldman Sachs research concluding that “the two-year interest rate differential turns out to be the strongest driver of the US dollar.” The full brief is available from the European Commission: economy-finance.ec.europa.eu The St. Louis Fed published a direct visualization of this relationship in February 2026, showing co-movement between the US-foreign interest rate differential and the dollar over time. The visual makes the transmission unmistakable: fredblog.stlouisfed.org For EURUSD specifically, the relevant spread is the US 2-year yield minus the German 2-
year Bund yield (or, more precisely, the US 2-year OIS minus the EUR 2-year OIS). When this spread widens, EURUSD falls (dollar strengthens versus euro). When it narrows, EURUSD rises. The correlation is not perfect, but over any meaningful timeframe, it is the single strongest driver.
USDJPY and the US-Japan Yield Differential
USDJPY has the highest rate sensitivity of any G7 pair, and the reason is structural. Japan has maintained ultra-low or negative interest rates for decades. Even after the Bank of Japan began tentative normalization, the US-Japan rate gap remains enormous compared to any other G7 pair. CME Group research identifies four key drivers of USDJPY: interest rate differentials, quantitative easing balance sheet ratios, economic growth rates, and trade balances. During the current cycle, the US-Japan rate differential widened to its broadest since before the 2008 financial crisis, making rate differentials the overwhelmingly dominant driver. The CME analysis is available here: www.cmegroup.com The ASEAN+3 Macroeconomic Research Office (AMRO) identifies the 5-year differential as the most relevant tenor for USDJPY, reflecting the institutional nature of Japan’s cross- border investment flows (insurance companies, pension funds, and banks managing duration-matched portfolios): www.amro-asia.org Here is the pearl for USDJPY traders: the sensitivity of approximately 5. 0% per 1 percentage point OIS widening means USDJPY is the pair most vulnerable to rate-channel reversals. When the Fed eventually cuts or the BoJ eventually hikes meaningfully, the rate differential narrowing will hit USDJPY harder than any other G7 pair. This is also why USDJPY is the pair most likely to overshoot on safe-haven flows during VIX spikes, because the rate-channel argument is so strong that reversals require extreme force.
USDCAD, Oil, and the Dual-Channel Problem
USDCAD is the pair that most frustrates traders who rely on rate differentials alone. The reason is that USDCAD responds to two competing channels simultaneously: the US- Canada rate spread and the price of crude oil. Canada exports approximately 4 million barrels of crude oil per day. When oil prices rise, Canada’s terms of trade improve, capital flows into Canadian assets, and CAD strengthens
(USDCAD falls). But if oil prices rise because of an inflationary shock, US rates may rise simultaneously, widening the rate differential and supporting USD (USDCAD rises). The two channels can pull in opposite directions. The historical correlation between USDCAD and crude oil is approximately 0. 75 to 0. 80 (inverse for CAD value), but this correlation has weakened in recent years as rate differentials have become more important. During the 2022-2024 hiking cycle, the US- Canada rate differential widened as the Fed tightened faster than the Bank of Canada, and this rate channel dominated the oil channel for extended periods. The practical implication: USDCAD requires both inputs. A rate-only model misses the oil channel. An oil-only model misses the rate channel. You need both, and you need to know which one is dominant in the current regime. When volatility is low and rate differentials are stable, oil tends to dominate USDCAD. When rate differentials are shifting rapidly (around FOMC or BoC meetings), the rate channel takes over.
AUDUSD, China, and Commodity Beta
AUDUSD is often described as a “risk-on” currency, but that description obscures the actual transmission mechanism. Australia’s currency responds primarily to three factors: the US- Australia rate differential, Chinese economic activity (which drives demand for Australian commodity exports), and global risk appetite. FOREX. com analysis shows AUDUSD correlations with Chinese bond yields of 0. 83 to 0. 93 for 5-year and 10-year yields on a 20-day rolling basis, making AUD effectively “a bet on China’s growth and inflation outlook”: www.forex.com The seminal academic paper on commodity currencies is Chen and Rogoff (2003), “Commodity Currencies,” which estimated commodity price elasticities of 0. 5 to 1. 0 for AUD and NZD real exchange rates. The paper is available from Harvard: scholar.harvard.edu A Dallas Fed working paper finds that interest rate factors can account for up to half of the variation in one-year currency returns when nonlinear risk premia are included: www.dallasfed.org For AUDUSD traders, the practical framework is: rate differential sets the structural bias, China determines the cyclical overlay, and global risk appetite determines the timing. When all three align (US rates rising, China slowing, risk-off), AUDUSD declines are powerful and sustained. When they conflict, the pair chops.
Why Taylor Rule Models Beat Everything Else
Here is a finding from the academic literature that should permanently change how you think about FX forecasting. Molodtsova and Papell (2009/2013) tested whether Taylor rule fundamentals could forecast exchange rates out of sample. Taylor rule models use the central bank’s reaction function, essentially estimating what the policy rate “should” be given current inflation and output gaps, and then using the implied rate differential to forecast the exchange rate. Their finding: Taylor rule models demonstrate significant out-of-sample predictability for 12 OECD exchange rates, outperforming the random walk, interest rate parity, monetary models, and purchasing power parity models. The Fed published a related working paper: www.federalreserve.gov And the extended NBER version: www.nber.org The predictability is strongest when central banks credibly follow their rules. During periods when central banks deviate from their reaction functions (such as during QE or forward guidance experiments), the models lose power. This makes intuitive sense: if the rate differential drives currencies, and the Taylor rule predicts the rate differential, then the Taylor rule should predict currencies. It does. The retail trader translation: you do not need to run Taylor rule regressions yourself. But you should understand the implication. The market prices currencies based on where it expects central bank rates to go. If you can read the Fed’s reaction function (what combination of inflation and employment data would trigger a cut or a hold), you can anticipate the rate differential, and therefore the currency direction. Every data release is an input to the reaction function. Every FOMC dot plot is a snapshot of the committee’s own Taylor rule estimate.
The Breakthrough: Exchange Rate Models Now Work
For decades, the dirty secret of academic FX research was that exchange rate models did not work. The random walk, a model that says tomorrow’s exchange rate is today’s exchange rate plus noise, consistently outperformed fundamental models. This was the Meese-Rogoff puzzle, and it made serious economists question whether currencies could be modeled at all. That era is over.
Engel and Wu published “Exchange Rate Models Are Better Than You Think, and Why They Didn’t Work in the Old Days” as NBER Working Paper 32808 in 2024. Their model links monthly bilateral exchange rate changes to real interest rates, expected inflation, trade balances, global risk measures, and liquidity demand. It fits “very well” for the US dollar against all G10 currencies, with fit increasing “almost monotonically” since the 1990s. The full paper is available from the University of Wisconsin: users.ssc.wisc.edu Their explanation for why models work now and did not work before is elegant: credible inflation targeting. Before the 1990s, central bank reaction functions were uncertain. Markets could not reliably predict how the Fed would respond to inflation. This uncertainty injected noise into the rate-to-currency transmission channel that overwhelmed the signal. After the adoption of inflation targeting and transparent communication, the reaction function became predictable. The Taylor Principle (raising rates more than one-for-one with inflation) was satisfied. And the models started working. Here is the counterintuitive finding embedded in this paper: higher US inflation leads to dollar appreciation when monetary policy is credible. This is the opposite of the textbook prediction. The mechanism is that markets price in the Fed’s reaction: higher inflation means higher rates, which means a wider differential, which means a stronger dollar. The inflation weakens purchasing power, but the rate response more than compensates. This creates a feedback loop that every FX trader should understand. Hot CPI or PCE print. Market reprices Fed path. 2-year yield rises. Rate differential widens. Dollar strengthens. All within hours.
The Structural Floor in Practice
Rate differentials do not make you money every day. They make you money over time by telling you which direction the structural wind is blowing. The volatility regime (Parts 1 and 2 of this series) tells you whether that wind is reliable right now or being overwhelmed by a storm. The yield curve spreads (Parts 3 and 4) tell you whether the wind is strengthening, weakening, or shifting. But the rate differential is the wind. Everything else is weather. When the Fed is at 3. 50-3. 75% and the ECB is at 2. 00%, the structural floor under the dollar is real and measurable. It takes approximately 3. 5% of euro appreciation per 1 percentage point of differential narrowing to erode that floor. Until the ECB hikes or the Fed cuts enough to compress the gap, the floor holds. This is what the US Rate Structure panel on the 4xForecaster dashboard (www.federalreserve.gov 4xforecaster. com/) is designed to communicate. The 3-month, 2-year, and 10-
year yields with their daily changes. The front curve spread and term structure spread with directional indicators. Not because the numbers are intrinsically interesting, but because they tell you whether the structural floor, the most important variable in FX, is intact, strengthening, or cracking. When the floor is intact, trade with it. When it is cracking, respect the crack. And when the volatility regime tells you the floor is being overwhelmed by a storm, step aside until the storm passes. Then check the floor again. It will still be there. This is Part 5 of the Macro-to-FX Transmission Series from 4xForecaster. Next: The Dollar Index Decoded: Why DXY Is 57. 6% Euro (And What That Means for You).