Macroeconomic theory suggests that currencies of countries in a strong cyclical position should appreciate against those in a weak position. One metric for cyclical strength is the output gap, i.e. the production level relative to output at a sustainable operating rate. In the past, even a simple proxy of this gap, based on the manufacturing sector, seems to have provided an information advantage in FX markets. Empirical analysis suggests that [1] following the output gap in simple strategies would have turned a trading profit in the long-term, and [2] the return profile would have been quite different from classical FX trading factors.

Colacito, Riccardo, Steven Riddiough, and Lucio Sarno (2019), “Business Cycles and Currency Returns”

The below are excerpts from the paper. Headings, bracketed/cursive text and emphasis have been added.
The post ties in with the SRSV summary on macro trends.

A most simple “output gap”

“We define the relative strength of the economy based on its position within the business cycle, i.e. whether it is nearer the trough or peak in the cycle. The business cycle…constitutes a key building block in theoretical models of exchange rates…Business cycles are a key driver and powerful predictor of both currency excess returns and spot exchange rate  fluctuations.”

“The output gap is defined as the logarithm of the difference between actual and `potential’ output…Since it is not directly observable, we measure the output gap using industrial production data…for a broad sample of 27 developed- and emerging market economies…and apply several commonly adopted [statistical] methods [for estimating the potential level] in the literature.”

“The use of two-sided filters and revised data [may bias empirical findings]… Hamilton (2018) provides a quantitative analysis of the main drawbacks of the Hodrick-Prescott filter and suggests an alternative procedure for de-trending output and measuring the output gap…We use the Hamilton procedure in our real-time analysis, implementing the procedure recursively conditioning only on data available at the time of sorting.”
see also: Hamilton, J. D. (2018), “Why you should never use the Hodrick-Prescott filter”

“To measure economic activity we collect industrial production data from the OECD’s Original Release Data and Revisions Database. The database provides monthly `vintages’, which reflect the precise time-series available to market participants each month, and is thus free of any subsequent revisions or forward-looking information.”

The predictive power of output gaps for FX returns

We find a strong link between currency excess returns and the relative strength of the business cycle. Buying currencies of strong economies and selling currencies of weak economies generates high returns both in the cross-section and time series of countries.”

“We collect daily bid, mid, and ask spot and 1-month forward exchange rates vis-a-vis the US dollar from Barclays and Reuters via Datastream. The empirical analysis uses monthly data obtained by sampling end-of-month rates from October 1983 to January 2016. Our sample comprises 27 countries…The sample period for each currency differs and thus the number of countries in our sample fluctuates over time. “

“The behaviour of exchange rates becomes easier to explain once exchange rates are studied relative to one another in the cross section, rather than in isolation…We take this empirical step by investigating the cross-sectional properties of currency returns to provide novel evidence on the relationship between currency returns and country-level macroeconomic conditions.”

“Using monthly data from 1983 to 2016, we find that sorting currencies into portfolios on the basis of the differential in output gaps relative to the US generates a monotonic increase in excess returns as we move from portfolios of `weak’ to `strong’ economy currencies. Thus our results imply that currency excess returns are higher for strong economies, a finding that we document to be robust to various ways of constructing currency portfolios.”

“The use of two-sided filters and revised data in the long sample also raises questions as to whether the relationship is exploitable in real-time. We explore this question using a shorter sample of `vintage’ data beginning in 1999 and find that the results are qualitatively identical. The `vintage’ data mimics the information set available to investors and thus sorting is conditional only on information available at the time…A high-minus-low cross-sectional strategy that sorts on relative output gaps across countries, generates a Sharpe ratio of 0.72 before transaction costs, and 0.50 after costs…Moreover, a time series strategy, which goes long (short) currencies issued by countries with output gaps above (below) the US…generates a Sharpe ratio of 0.65 before costs and 0.50 after costs.”

[The figure below] plots the cumulative returns to both currency strategies. Specifically, the figure shows the out-of-sample cumulative returns (left-hand plot), the equivalent in-sample cumulative returns obtained using the Hamilton (2018) linear projection applied to revised data (middle plot), and combination strategies.

“In related work, Dahlquist and Hasseltoft (2019) propose a currency strategy based on economic momentum, defined on the basis of eight economic variables that capture interest rate, price, industrial production, and unemployment information…Their results suggest that the strategy generates high risk-adjusted returns and subsumes the carry trade.”

No strong correlation with classic FX trading factors

“The excess returns from a trading strategy that sorts currencies on relative output gaps generates high risk-adjusted returns that are uncorrelated with the excess returns from popular currency investment strategies, thereby providing tangible diversification gains to global investors.…Returns…are uncorrelated with common currency investment strategies, and cannot be understood using traditional currency risk factors.”

“The predictability stemming from business cycles is quite different from other sources of cross-sectional predictability observed in the literature… We test the pricing power of conventional risk factors using a battery of linear asset pricing models, and do not find evidence that these pricing kernels can price the cross-section of currency returns sorted on output gaps.”

Sorting currencies by output gaps is not equivalent, for example, to the currency carry trade that requires sorting currencies by their differentials in nominal interest rates. We highlight this point in [the figure below] using two common carry trade currencies, the Australian dollar and Japanese yen. The interest rate differential is highly persistent and consistently positive between the two countries in recent decades. A carry trade investor would have thus always been long the Australian dollar and short the Japanese yen. In contrast, the output gap differential varies substantially over time, and an output-gap investor would have thus taken both long and short positions in the Australian dollar and Japanese yen as their relative business cycles fluctuated.”