Understanding macroeconomic transmission channels in emerging market (EM) foreign exchange (FX) markets is key to uncovering valuation investment opportunities. Here we present ways investors can combine macroeconomic assumptions with data science techniques and economic estimates to design a framework that measures EM FX over- and under-valuation.
Our EM FX valuation framework started from the premise that the premium built into the EM FX forward market is the compensation that investors require for assuming the risk that the respective EM currency may experience a decline in its value against the numeraire currency (i.e., the U.S. dollar) over the term of the forward contract. That is, the carry that investors harvest from long EM FX forward positions against the U.S. dollar is their required buffer against perceived EM currency depreciation risks.
A sensible framework for identifying relative richness or cheapness in the EM FX market therefore necessitates the identification of the main factors that drive EM currency depreciation risk. Here we turned to macroeconomic fundamentals to provide us with the necessary building blocks, basing our EM FX valuation framework on the following assumptions:
External balance:all else being equal, markets perceive countries with weaker trade fundamentals as posing greater domestic currency depreciation risks
Inflation:all else being equal, markets perceive countries with higher price inflation as posing greater domestic currency depreciation risks
Real economic growth:all else being equal, markets perceive countries with weaker real economic growth as posing greater domestic currency depreciation risks
Our study is based on a universe of EM FX pairs against the U.S. dollar, as described in the table below.
Emerging Market FX Universe
India, Indonesia, Malaysia, Philippines, Singapore, South Korea, Taiwan, Thailand
Hungary, Israel, Poland, Russia, South Africa
Brazil, Chile, Columbia, Mexico, Peru
*EMEA = Europe, Middle East, and Africa
Valuation Framework: Design and Currency Valuations
We designed a framework that combines information about observable risk factors with observable compensation that the market demands exposure to these risk factors. Our objective was to use this framework to systematically track EM currencies that offer risk premia that are not aligned with macroeconomic fundamentals, i.e., cheap currencies that reward too much in risk compensation given risk factors and expensive currencies that offer too little in risk compensation given underlying macroeconomic dynamics.
In designing our EM FX valuation framework, we asked the following questions:
What methodology should underlie our EM FX valuation framework?
We opted for a model framework that delivers on two important and interconnected aspects:
It does not force the researcher to impose a priori structures on the (unknown) functional form of the modeled relations
It is a priori well suited to capture non-linear dynamics
Based on these two conditions, we believe our choice of random forest regression as our preferred econometric framework is justified.
What is our model’s dependent variable, and what factors can we use to explain its behavior?
We used the carry implied by 1-month forward long EM-short U.S. dollar contracts in our respective country universe as our model’s dependent variable. We then used forward-looking measures of country current account as a percent of GDP, CPI inflation, and real GDP growth obtained from FactSet Economic Estimates as inputs into our random forest regression framework.
In addition, we hypothesized that domestic EM and foreign (i.e., the Federal Reserve in the case of currency pairs against the U.S. dollar) central banks do not react to each incremental macroeconomic news release but to the accumulation of data over time. This introduces a degree of inertia in domestic and foreign interest rate environments and ultimately in the FX risk premium that we aimed to model. We proxied this with the one-period lag of our dependent variable.
How do we obtain country valuation scores?
We fit a monthly random forest regression that modeled our dependent variable on our proxies for central bank policy inertia and forward-looking macro risk factors
We used the fitted random forest regression model and the set of observable explanatory factor data to predict a model-implied “fair value” level for carry (risk premium) for each EM FX pair in our universe
The difference between the actual market level of the risk premium and the model-implied fair value level provided a model-dependent proxy for the degree of richness or cheapness of the respective EM FX market
We converted country value gaps into comparable value proxies by time series scoring the raw richness or cheapness measures of each country
We fit our random forest regression valuation model framework on a monthly frequency over the period September 2013 - October 2021. Daily FX market data and continuous updates to the macroeconomic forecasts from FactSet Economic Estimates provided frequently updated and real-time measures of cross-sectional over- and under-valuation in the investable EM FX universe.
The chart below provides a snapshot of our model’s valuation output at the end of October 2021. As of that date, Brazil and Russia present the most attractive relative value opportunities while the potential for capital enhancement via price appreciation is the most unfavorable for Poland and South Africa.
Emerging Market FX Value as an Investable Factor
We then sought answers to these questions:
Does the concept of value in EM FX markets exist as a distinct phenomenon that is conceptually and empirically different from other traditional factor approaches in this asset class?
Even if value factor orthogonality can be shown, has this factor generated positive premia to investors in a historical context?
To answer these questions, we ran regressions on three investment factors to ascertain to what extent each factor could generate a performance premium beyond that implied by the broad market and other competing traditional investment styles. The factor regression results shown in the table below are based on the monthly performance of our proxy for the EM FX broad market and three EM FX investment factors—carry, momentum, and value—over the period October 2014 - October 2021.
We constructed our EM FX market portfolio as a long-only long-EM equal-weighted portfolio in the EM countries outlined above. The three-factor portfolios are long-only long-EM and equal-weighted and are rebalanced monthly to cover only the subset of our full EM FX universe that falls beyond the median of the respective factor measure.
Empirical evidence supports the view that an explicit valuation factor in the EM FX market captures a distinct phenomenon in this asset class. More specifically, our designed EM FX valuation factor exhibits a statistically significant “alpha” once we control for the factor’s exposure to the broad EM FX market and the influence of the other two traditional factors often employed in systematic EM FX investment strategies. In addition, the value factor was shown to generate positive “alpha” performance over the same historical time span.
Here we present a prototype of an EM FX valuation framework based on assumptions on macroeconomic transmission channels, allowing investors to identify and exploit value opportunities in this asset class in a disciplined and systematic way. This is just one possible prism through which investors can think about uncovering valuation investment opportunities in EM FX markets; this analysis is by no means conclusive. An important question of how investors can time different systematic investment factors in EM FX, including the value factor, to reap maximum factor diversification and performance potential remains open for future research.
This blog post is for informational purposes only. The information contained in this blog post is not legal, tax, or investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.
Associate Director, Research, Alpha Analytics and Optimization
Mr. Emil Margaritov is Associate Director, Research, Alpha Analytics and Optimization at FactSet. In this role, he leads the research and development of systematic smart-beta investment products across multiple asset classes. Prior, he was a quantitative researcher at the Macro team of a multi-strategy hedge fund within the Systematic Fixed Income group at BlackRock, where he was co-leading the research and development of investment strategies across fixed income, foreign exchange, and asset allocation. Mr. Margaritov earned a doctoral degree in economics from Goethe University Frankfurt.
The information contained in this article is not investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.