Carry trades are one of those strategies that sounds too simplistic to be true, actually works most of the time, and sometimes blows up for reasons that are often more confusing than the fact that it worked in the first place. The basic idea of a carry trade is to make a list of currencies, ranked by yield. Borrow one of the lower-yielding ones, use it to invest in the higher-yielding one, and... that's the trade.

If you look at interest rate differentials by currency, your first instinct might be that you earn 5.5% per year if you borrow yen at 0.25%, convert that money into the Polish zloty, and lend it out in Poland where the risk-free rate is 5.75%. Your next thought might be that that 5.5% probably represents the market's assessment of the long-run relative depreciation of the zloty in yen terms, and that is indeed a good working assumption.

It just turns out not to be true, historically speaking: over time, the "funding currencies," i.e. low-rate currencies issued by a handful of very stable countries, did outperform flightier ones, but typically by less than the magnitude of that interest rate differential. You can tell a story where this makes sense, and it goes like this: investors either want to save in their domestic currency or in some stable currency, and the richer they get the more their savings will be in that currency. (This is both because there are more desirable dollar-, euro-, and yen-denominated assets and because the consumption basket of a richer person will often have more imports—if they're buying nice cars, constantly upgrading their smartphones, and taking vacations in Miami or Paris or Zurich, their purchases will be denominated in non-local currencies and they'll preserve their purchasing power if their assets are, too.) In that model, carry traders are providing a classic financial service, of taking the less desirable side of a trade and getting paid a bit for it.

This turns out to carry risks. Suppose you run this strategy, and you do the prudent thing of diversifying the currencies you buy—a bit of money in Mexico, some in Brazil, some in Poland, some in Turkey. From time to time, you take a hit, because one of these countries will go through a financial crisis, coup attempt, spate of free-spending populist governance, etc. But there will also be times when all of these currencies are depreciating at once: when there's a global financial crisis and every investor is liquidating risky investments in order to pay debts, those investors are disproportionate sellers of emerging market assets, and disproportionate buyers of dollars, euros, and yen.

So, in moments like that, your carry trade portfolio's return streams are all highly correlated, and it's the bad kind of highly correlated where every side of the trade is losing money—your pesos buy fewer goods and services in Mexico, the dollars you borrowed have appreciated relative to goods and services (and risky financial assets) in the US. And the correlation between this strategy and other strategies has also gone up: if you're running a diversified portfolio with different investment strategies, the time when carry trades lose money is the time when everything else loses, too.

That helps explain why the excess returns from carry trades never quite dropped to zero: the excess return they earn is really a kind of insurance premium, because another reason most investors prefer to own stable currencies rather than speculate in unstable ones is that the unstable ones are more likely to collapse when other assets are losing value, too.

The mechanics of a carry trade can also make it painful. A currency achieves funding-currency status when investors have a reasonable expectation that it's going to have low rates for a while, and if the carry trade expands, they start to make money on both sides: they're earning money every day from the interest differential, but as more people borrow yen and then sell it to buy something else, the yen keeps depreciating. This makes these traders feel incredibly smart—it's just a wonderful feeling when both sides of a pair trade work out exactly as you're expecting. But this means that analyzing the funding currency means tracking a moving target: a few years ago, you were looking at Japan as a slow-growth, rapidly-aging country whose households invest a wild share of their assets in cash and cash equivalents. You were looking at the first country to experiment with effectively zero interest for long-term bonds, and a country that stuck with this experiment for a long time. But, over time, what you were looking at changed—at the peak of the yen carry trade, you were looking at a momentum asset that was continuously depreciating as traders sized up their bet against it, but that also meant that they were increasingly making the same bet as a large number of levered investors who may not have fully understood who the marginal yen seller was and how they'd behave.

The most recent iteration of the yen trade unwound brutally over the last few days, after the Bank of Japan increased interest rates by 25 basis points. Sometimes, that's what it takes: carry trades become possible because of monetary and fiscal policy, but they lead to currency movements that justify a different set of policies. Looking forward, they're mostly a systematic bet on long-term averages, but at the time when they get really exciting, it's because they turned out to be a bet on a specific macro environment that, in part because of that bet, couldn't last forever.

Read More in The Diff

Carry trades and other such situations come up from time to time in The Diff. For example:

• We’ve looked at how 1998 was an important year in financial history because of the unwind of other carry trades.

• In Capital Gains itself we’ve looked at dividend yield, another kind of carry situation.

• And we covered why Japanese stocks were so cheap ($) a few years back.

Join the discussion!