The Exchange Rate Disconnect International Finance Puzzle
The exchange rate disconnect puzzle refers to the weak shortrun relationship between the exchange rate and its macroeconomic fundamentals. In other words, underlying fundamentals such as interest rates, inflation rates, and output don’t explain the shortterm volatility in exchange rates.
For example, shortterm forecasts of macroeconomic exchange rate models are little better than forecasts of random walk models. A shortterm exchange rate determination according to the random walk model implies that tomorrow’s exchange rate equals yesterday’s exchange rate.
Therefore, you need to put into perspective the models or concepts about exchange rate determination and modify your knowledge. Models of exchange rate determination are good approximations for the longrun relationship between exchange rates and macroeconomic fundamentals. The exchange rate disconnect puzzle implies that macroeconomic fundamentals don’t explain shortterm variations in exchange rates.
The exchange rate disconnect puzzle and the purchasing power puzzle are considered pricing puzzles. Therefore, explanations of the exchange rate–related puzzles make use of comparing exchange rates to asset prices. As in the case of exchange rates, asset prices such as stock market indices are highly volatile, and the changes in them don’t strongly correspond to changes in fundamentals.
Similar to exchange rates, random walk models are reasonably consistent with changes in stock prices. In the case of asset prices in general, the assumption is that news affects asset prices. Still, given the relevance of exchange rates as relative prices and their effect on a large number of transactions, it’s surprising that disconnect exists between shortterm changes in exchange rates and underlying macroeconomic fundamentals.
The exchange rate disconnect puzzle has various manifestations. Predictable excess returns are one of them. According to interest rate parity (IRP), the differences in home and foreign interest rates should be equalized by changes in the exchange rate. But studies find two characteristics of shortrun exchange rates that aren’t consistent with the IRP:

Not only do shortterm exchange rates deviate from the IRP, but these deviations are predictable.

The variance of these predictable returns is greater than the variance of the expected change in the exchange rate.
Predictable excess returns have two explanations:

Market forecasts are irrational: This irrationality may arise from the presence of heterogeneous traders in the market.
Heterogeneity among traders means that there are different types of traders such as fundamentalists (they keep an eye on macroeconomic conditions that affect exchange rates), chartists (they use charts or graphs about past changes in exchange rates to forecasts future changes), and noise traders (they follow trends in exchange rates and overreact to good or bad news; also called irrational).
(Some studies consider all traders who don’t make investment decisions based on fundamentals (chartists and noise traders) as noise traders or irrational.) Studies show that irrational traders can affect prices; because they face a higher risk, they can earn higher returns than rational traders.
Differences may arise in the distribution of perceived and measured economic disturbances (perceived by traders and measured by researchers): In this explanation, market participants are rational. But systematic forecast errors reflect the difference between what traders observe as they experience the situation and how researchers make sense of it ex post (or after the fact). Still, this approach doesn’t explain the high variation in excess returns.

Peso problem effect: A peso problem refers to the situation in which market participants’ expectations about a future policy change don’t materialize within the sample period. In the early 1970s, interest rates on Mexican peso were substantially higher than those in the U.S. This was happening despite the fact that the Mexican peso had been pegged for almost a decade.
Nominal interest rates on the Peso were higher than in the United States and reflected higher inflation rates in Mexico along with the probability of a large devaluation of the peso. A couple years passed before, in the late 1970s, the Mexican peso was devalued.
This situation isn’t consistent with standard assumptions about forecast errors. Forecast errors should have a mean of zero and shouldn’t correlate with current information. But if market participants expect a future discrete change in policy or fundamentals, then rational forecast errors can be correlated with current information and may have a nonzero mean in finite samples.