Emerging-market equities have been one of the investment darlings of 2017, handily outpacing their developed-market peers on a narrative of accelerating global growth. The idea that a booming world economy is consistent with EM outperformance seems so obvious that it isn’t even worth checking—but is that actually the case? I'm a macro strategist who writes Bloomberg's Macro Man column; I’ve also recently started a Mythbusters-style series in Bloomberg Markets magazine where I challenge some of the most basic assumptions in finance. For my latest endeavor, I decided to take a look at what really drives the relative returns of emerging markets.
Given the size and liquidity of the U.S. stock market, I used that as the performance benchmark for emerging-market equities. It turns out that global growth considerations do provide a decent contemporaneous explanation for relative EM performance.
Using monthly data since 1995, I found a correlation of 0.3 between a global purchasing managers’ index (a proxy for world growth) and trailing one-year relative EM out- or underperformance vs. the U.S. market. In a sense, we can distill the essence of relative emerging-market performance even more: The ratio of the MSCI EM Index to the MSCI USA Index seems to track the trends in the spread between Chinese and U.S. gross domestic product growth pretty closely.
This is all well and good, but simply explaining prior performance doesn’t necessarily tell us much about the future relative returns of EM and U.S. stocks. What does information available today say about relative returns in the future?
I ran a regression using a few simple factors to explain one-year forward relative performance of EM and U.S. stocks. The results were fascinating.
It turns out that current global PMI readings are negatively correlated with one-year forward relative EM returns. The better the PMI numbers today, the worse that EM stocks do relative to the U.S. over the next 12 months on average. The correlation since 1995 is -0.3. Unsurprisingly, the global PMI gets a negative coefficient in the regression formula, with a t-statistic that is statistically significant.
I specifically excluded the dollar from the regression—it’s almost tautological that U.S. stocks will outperform (in dollar terms) when the USD appreciates and underperform when it falls.
Which factors do the best job of forecasting future EM performance? Valuation and momentum. I defined valuation as the spread in the earnings yield—that is, the inverse of the price-earnings ratios—between EM and U.S. stocks. The regression shows a positive, statistically significant relationship: The higher the EM earnings yield relative to the U.S., the better the subsequent performance. On this basis, it sure looks like EM stocks are a value play in a growth narrative’s clothing!
Interestingly, though, the factor with the highest statistical significance turns out to be momentum. If EM stocks have outperformed over the previous year, it turns out they do well over the next 12 months, too.
While we also tend to think of EM stocks as a reflation play, generally speaking inflation isn’t so good for them—at least U.S. inflation. The negative relationship is fairly modest, however, and I’d put it down to the monetary policy response from the Federal Reserve.
I should emphasize, however, that the relationship between several of the factors and EM performance doesn’t seem to be stable. The r-squared, a measure of the predictive power of the model, is 0.4. That’s decent but certainly doesn’t suggest that the modeled factors can explain the majority of future EM returns
As a sense check, I broke each of the input factors into deciles and calculated the subsequent one-year relative returns from each. In effect, this helped separate the wheat from the statistical chaff.
For example, there’s a strong linear relationship between the spread in earnings yields and subsequent performance. It turns out that if you buy the market that’s cheap, it tends to do better. Chalk one up for Graham and Dodd there. For what it’s worth, the current EM earnings yield premium to the U.S.—1.7 percent—is in the sixth decile: That’s broadly consistent with EM stocks matching the performance of their U.S. equivalents.
What about that counterintuitive finding that growth, as represented by global PMI, is bad for subsequent EM relative performance? Well, there does seem to be a compelling case that buying EM when the growth outlook seems especially dire can pay handsome dividends. (It’s always darkest before the dawn.) However, while the negative relationship between PMI readings and subsequent EM outperformance seems broadly linear, there are some exceptions. While it may be a stretch to say that strong PMI readings are an active detriment to EM returns, at the very least we can conclude that they are far from an unalloyed positive—contrary to popular belief.
What about momentum, the strongest signal in the regression? Again, there looks to be a strong linear relationship … with a twist. At extreme levels of EM outperformance, it seems as though some sort of mean reversion kicks in and emerging-market stocks cease to outperform.
A similar analysis comparing the U.S. consumer price index with subsequent EM returns shows no evidence of a stable relationship—small wonder it had the lowest t-statistic in the regression.
So if we add everything up, what does our framework suggest about EM performance moving forward? The prognosis is a little less than optimistic: The model is forecasting that EM stocks will slightly lag their U.S. counterparts over the next 12 months.
While that, of course, is no guarantee that the EM narrative is about to end, the results of this study seem sufficiently compelling that we can probably conclude that “strong PMI data leads EM relative performance” is false.
Crise is a macro strategist who writes the Macro Man column for Bloomberg and blogs for Markets Live.