Uncovered Interest Parity: The Fama Puzzle and Carry Trade

Key Takeaways

• The Fama puzzle: the Fama regression coefficient on interest differentials is systematically negative (~−0.88 in Froot-Thaler's survey of 75 studies), meaning high-rate currencies tend to appreciate, not depreciate.
• This systematic UIP failure is the foundation of the carry trade, borrowing low-rate currencies (JPY, CHF) and investing in high-rate ones earns the spread until a violent risk-off unwind.
• UIP fails badly at 1-month to 1-year horizons but performs better over 5–10 year horizons (Chinn-Meredith: "It Works, But Not For Long").
• Carry trade excess returns are compensation for crash risk, not a puzzle-free arbitrage; high-rate currencies can lose 10–20% in days during risk-off episodes.

What It Is

UIP is the theoretical sibling of covered interest parity. CIP uses forward contracts to eliminate FX risk; UIP replaces that forward with investors' expectations of the future spot rate. If markets are efficient and risk-neutral, the two should give the same answer. They do not.

The empirical failure is known as the Fama puzzle or forward premium puzzle, after Eugene Fama's influential 1984 paper. Fama showed that not only does UIP fail, the sign is systematically wrong: currencies with higher interest rates tend to appreciate on average, not depreciate. That violation is the foundation of the carry trade.

The Intuition

The logic behind UIP is simple. If dollar rates are 5 percent and yen rates are 0.5 percent, and FX markets set no risk premium, investors should demand a compensating expected yen appreciation of 4.5 percent. Otherwise everyone would pile into dollars.

In practice, speculators do pile into dollars, and the yen frequently does not appreciate. Over many years, high-rate currencies tend to keep delivering positive excess returns above the forward rate. That is what makes the carry trade a persistently profitable strategy, and it is also what makes the Fama puzzle so stubborn.

How It Works

The standard UIP equation is:

E[S_{t+1}] / S_t = (1 + r_domestic) / (1 + r_foreign)

Where:

S_t          = spot exchange rate today (foreign per domestic)
E[S_{t+1}]   = expected spot rate next period
r_domestic   = domestic interest rate
r_foreign    = foreign interest rate

Rearranged, the expected percentage change in the exchange rate should equal the interest differential:

E[delta S] / S = r_domestic - r_foreign

The Fama regression tests this by running realized exchange rate changes on interest differentials. Under UIP, the slope coefficient should be exactly 1.0. In Fama's data and countless replications, the slope is typically negative, often close to minus one. The Froot-Thaler (1990) survey across 75 studies found an average slope of about minus 0.88.

More recent work, including Bussiere and Chinn's "New Fama Puzzle" (2018), finds the sign has flipped closer to positive in the post-2008 period for some currency pairs, especially at short horizons. UIP still fails, just differently. At long horizons of five to ten years, something closer to UIP does re-emerge in the data, which is why the Fed paper by Chinn and Meredith is titled "It Works, But Not For Long."

Worked Example

Consider a one-year carry trade between the dollar and the yen. Spot USD-JPY is 150.00. Dollar rates are 5.0 percent and yen rates are 0.5 percent for one year.

UIP predicts the yen should appreciate by 4.5 percent, so expected USD-JPY one year forward is:

E[S_{t+1}] = 150.00 * (1 + 0.005) / (1 + 0.050)
E[S_{t+1}] = 150.00 * 1.005 / 1.050
E[S_{t+1}] = 143.57

Under UIP, a carry trader borrowing yen at 0.5 percent and lending dollars at 5.0 percent would earn the 4.5 percent interest spread, then give back exactly 4.5 percent when converting back to yen, for zero expected profit.

In reality, USD-JPY has often stayed flat or even appreciated over one-year windows when US rates were above Japanese rates. That means the carry trader keeps the 4.5 percent interest spread plus any FX appreciation, a return that can persist for years until it unwinds violently in risk-off episodes (August 2024, for example).

Common Mistakes

1. Treating UIP as if it holds. It does not, and betting on it is how many introductory currency models get embarrassed. Carry trades that assume UIP holds would never exist, yet they are among the oldest strategies in FX.

2. Confusing the carry trade with free money. Carry works until it does not. When UIP "snaps back" in a risk-off event, high-rate currencies can lose 10 to 20 percent in days, wiping out years of carry income. The excess return is compensation for crash risk, not a puzzle-free arbitrage.

3. Ignoring horizon effects. UIP fails badly at one-month and one-year horizons but performs better over 5 to 10 years. Traders look at short-term failure; long-term investors see more of the UIP pull.

4. Mixing nominal and real rate differentials. UIP is a nominal condition. Real-rate versions exist but require matching expected inflation across countries, which is a separate estimation problem.

5. Not adjusting for time-varying risk premia. Modern currency pricing models (Fama-French-style factors, affine term structure models, habit models) all treat the UIP deviation as a risk premium rather than an irrational pricing error. Treating it as free alpha ignores the tail risk embedded in the trade.

Frequently Asked Questions

What is uncovered interest parity? UIP says that if dollar rates are 5% and yen rates are 0.5%, the yen should appreciate by 4.5% over the period to equalize returns for an unhedged investor. This is a theoretical condition requiring market efficiency and risk-neutrality. In practice it fails systematically, high-rate currencies tend to appreciate, not depreciate.

What is the Fama puzzle? The Fama puzzle is the empirical finding, documented by Eugene Fama in 1984, that the regression of realized exchange rate changes on interest differentials produces a negative slope coefficient, roughly −0.88 on average across 75 studies. This means not only does UIP fail; the sign is backwards. High-rate currencies appreciate rather than depreciate, which is why carry trades are profitable.

What is the carry trade and why does it work if UIP fails? The carry trade borrows in a low-interest-rate currency (JPY, CHF) and invests in a high-rate currency (MXN, BRL, AUD), earning the interest differential. It works because UIP fails: the high-rate currency does not depreciate as UIP predicts. The excess return is compensation for crash risk, when risk appetite reverses, high-rate currencies can fall 10–20% in days, wiping out years of carry income.

Does UIP ever hold? At short horizons (one month, one year), UIP fails consistently. At long horizons of five to ten years, something closer to UIP re-emerges in the data, which is why Chinn and Meredith titled their Federal Reserve paper "It Works, But Not For Long." Long-term investors and central banks operating over multi-year horizons see more of the UIP pull; short-term traders exploit the failure.

What explains the UIP failure, irrationality or risk premia? Modern currency pricing models treat UIP deviations as time-varying risk premia, not as irrational pricing. High-rate currencies pay a carry premium because they tend to depreciate sharply during global risk-off episodes when marginal utility is highest. From this view, carry trade returns are rational compensation for tail risk, not free alpha from a market anomaly.

Sources

1. Chinn, M. and Meredith, G. "Uncovered Interest Parity: It Works, But Not For Long." Federal Reserve IFDP 752. https://www.federalreserve.gov/pubs/ifdp/2003/752/revision/ifdp752r.pdf
2. Bussiere, M. and Chinn, M.D. "The New Fama Puzzle." NBER Working Paper 24342. https://www.nber.org/system/files/working_papers/w24342/w24342.pdf
3. Bussiere, M. and Chinn, M.D. "The New Fama Puzzle." IMF Economic Review. https://link.springer.com/article/10.1057/s41308-022-00161-z
4. CEPR VoxEU. "The New Fama Puzzle." https://cepr.org/voxeu/columns/new-fama-puzzle

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.