Oleg Itskhoki, Dmitry Mukhin
9 min
The Mussa puzzle, named after Michael Mussa’s 1986 observations, highlights a striking historical shift: when the Bretton Woods system of pegged exchange rates collapsed in 1973, both nominal and real exchange rate volatility increased dramatically. For decades, economists have interpreted this as evidence of nominal price rigidities; the logic follows that if nominal exchange rates become volatile, but prices (which are sticky) cannot adjust, then real exchange rates must necessarily fluctuate to compensate. This has long served as a key argument for why monetary policy is non-neutral.
Itskhoki and Mukhin re-evaluate this long-standing assumption. By examining an expanded set of macroeconomic facts, the authors argue that the data simply does not support the standard "nominal rigidity" explanation. If price stickiness were the primary driver, one would expect to see corresponding shifts in other macroeconomic variables during the regime transition. The authors find that no such simultaneous changes occurred, which falsifies both traditional flexible-price Real Business Cycle (RBC) models and standard sticky-price New Keynesian models.
The paper offers a new resolution to the Mussa puzzle by shifting the focus from product markets to financial markets. The authors propose a model of segmented financial markets where exchange rate risk is not distributed evenly across the economy. Instead, this risk is disproportionately held by specialized financial intermediaries. In this framework, the volatility observed after 1973 is driven by the risk premium channel rather than a failure of prices to adjust quickly in response to monetary shocks. This suggests that the impact of monetary policy on exchange rates is fundamentally linked to how financial intermediaries manage and transmit risk.
The Mussa (1986) puzzle is the observation of a sharp and simultaneous increase in the volatility of both nominal and real exchange rates following the end of the Bretton Woods System of pegged exchange rates in 1973. It is commonly viewed as a central piece of evidence in favor of monetary non‐neutrality because it is an instance in which a change in the monetary regime caused a dramatic change in the equilibrium behavior of a real variable—the real exchange rate—and is often further interpreted as direct evidence in favor of models with nominal rigidities in price setting. This paper shows that the data do not support this latter conclusion because there was no simultaneous change in the properties of the other macro variables, nominal or real; an extended set of Mussa facts falsifies both conventional flexible‐price RBC models and sticky‐price New Keynesian models. We present a resolution to the broader Mussa puzzle based on a model of segmented financial market, in which the bulk of the nominal exchange rate risk is held by financial intermediaries and is not shared smoothly throughout the economy, emphasizing the importance of monetary transmission via the risk premium channel.
Alex: That distinction matters. An observation might support monetary effects without identifying the particular mechanism producing them.
Sam: That is the distinction to keep in mind. Their argument does not make the exchange-rate jump disappear. It challenges treating that jump as direct evidence for slow price adjustment.
Alex: What do they add to the original comparison that changes the interpretation?
Sam: They widen the set of facts a model must explain. Alongside the changes in exchange rates, they examine the properties of other macroeconomic variables. The abstract reports no simultaneous change in those other variables, whether nominal or real.
Alex: Then the absence of a broader change is part of the evidence. A model cannot succeed just by reproducing more volatile exchange rates.
Sam: It must also account for the contrast with the rest of the economy. That is the key idea behind their extended Mussa facts. The combination matters, rather than the exchange-rate observation considered alone.
Alex: Can you make that combination concrete using the historical case, without adding details we do not have?
Sam: Before the regime change, exchange rates were pegged, meaning held to a target. After that system ended, nominal and real exchange-rate volatility rose sharply together. But the authors report no simultaneous change in the properties of the other macroeconomic variables.
Alex: Which existing models fail that broader test? Is the criticism limited to models with slow price adjustment?
Sam: No. The authors say the extended facts falsify conventional flexible-price real business cycle models. Those are models where prices can adjust freely. They also say the facts falsify sticky-price New Keynesian models, which incorporate slow price adjustment.
Alex: So simply switching from slow-adjusting prices to freely adjusting prices does not resolve their objection.
Sam: The abstract challenges both conventional approaches. But its wording matters: this is a claim about those model classes facing the extended facts. It is not a demonstration that every conceivable model involving either pricing assumption must fail.
Alex: What alternative do they propose, and what makes it different from changing the price-setting assumption?
Sam: They propose a model of segmented financial markets, where risk is not shared smoothly throughout the economy. In their model, financial intermediaries hold the bulk of nominal exchange-rate risk. The location of that risk becomes central to the explanation.
Alex: That directs attention to who bears currency fluctuations, rather than only how quickly prices respond. How does monetary transmission enter?
Sam: The authors emphasize a risk premium channel. A risk premium is compensation investors require for bearing risk. Their proposed resolution therefore puts the transmission of monetary effects through risk compensation at the center.
Alex: Can we say how that premium changes, or how much exchange-rate volatility the mechanism explains?
Sam: Not from this page. It does not give the relationship’s direction, magnitude, or detailed model structure. We can report that the authors present this mechanism as a resolution, but we cannot reconstruct its quantitative performance.
Alex: Let’s turn to the evidence. When they say exchange-rate volatility rose sharply, what number should a listener carry away?
Sam: There is no numerical effect size in the abstract. It gives the historical break and the qualitative comparison. It does not state how many countries were studied, the observation window, or the uncertainty around the results.
Alex: And “no simultaneous change” could cover several different statistical properties. Do we know exactly what stayed unchanged?
Sam: We do not. The abstract uses “properties” without listing the variables or statistical measures. We should not translate that into a claim that every other economic outcome stayed constant.
Alex: That limits our assessment of a strong word like “falsifies.” We know the authors’ conclusion, but not the tests supporting it.
Sam: Yes. The full paper is needed to assess how they establish the contrast and test the models. Our judgment here is narrower: the abstract states a clear empirical challenge, but does not supply enough detail to independently evaluate it.
Alex: The page also mentions a replication package. Does that help with confidence?
Sam: It says the journal checked the package’s data and code for their ability to reproduce the paper and approved online appendices. That is useful information about reproducibility. It does not, by itself, establish that the economic interpretation is correct.
Alex: So the contribution has two parts: a wider empirical target and a different proposed mechanism. What remains open for the reader is how convincingly those parts fit together.
Sam: That is what to investigate in the manuscript. Does the extended comparison support rejecting the conventional models? And does concentrating exchange-rate risk in intermediaries explain the broader pattern under the authors’ assumptions?
Alex: Who should make time for that full read, and where should they start?
Sam: Researchers in international macroeconomics, exchange rates, and monetary transmission should read it closely. Start with the extended Mussa facts and the comparison against conventional models. Then examine the segmented-market mechanism, especially how intermediaries bearing currency risk connects to the risk premium channel.
Alex: And for researchers outside those areas, what is the one line to remember?
Sam: A jump in exchange-rate volatility is not, by itself, a verdict on how prices adjust. The authors’ challenge is to explain both what changed and what did not.
Alex: Keep that distinction in mind when you read the evidence.