EM Portfolio Construction: Smart Indexes, Frontiers, and ESG

So you’ve learned how to trade EM credit, local rates, and FX. Now what? How do you actually put it all together into a portfolio? Chapter 10 is about the nuts and bolts of portfolio construction, and it has some genuinely surprising findings about indexes, risk parity, and why ESG is unfair to poor countries.

The Problem with Market-Cap-Weighted Bond Indexes

Here’s a weird thing about fixed-income indexes that most people don’t think about. When a country issues more debt, its weight in the index goes up. Index-tracking investors then have to buy more of that country’s bonds. So the more a government borrows, the more money flows to it automatically.

This is the opposite of what bond vigilantes are supposed to do. In theory, markets should punish countries that borrow too much. But market-cap-weighted indexes actually reward them with more capital. The reckoning only gets postponed, and when it finally arrives, investors are holding way more of that country’s bonds than they should be.

For equities, market-cap weighting makes more intuitive sense. You’re allocating more money to successful companies whose stock prices went up. For bonds, you’re allocating more money to countries that borrowed the most. Not the same thing at all.

Equal Weights: Great Until They’re Not

One fix is to weight every country equally. And it works, sort of. An equal-weighted version of the EMBI outperformed the regular EMBI pretty consistently after the 2008 crisis. The information ratio was 0.49 vs 0.38 for the regular index.

But there’s a catch. Equal weighting is basically a bull market strategy. Going into 2008 and during the crisis, the equal-weighted index underperformed. Makes sense: you’re overweighting smaller, riskier credits, and those are exactly the ones that blow up when things go south.

There’s also a practical problem. Even a medium-sized fund would run into liquidity issues trying to buy enough bonds in the less liquid credits. And equal weighting doesn’t protect you from overleveraged small countries. It can actually increase your exposure to them.

GDP Weighting Sounds Smart but Doesn’t Work

The next logical idea: weight countries by GDP. Bigger economy, bigger weight. This would be closer to how equity indexes work, where successful companies get higher weights. It would also solve the liquidity problem since larger countries tend to have more bonds outstanding.

Sadly, it doesn’t generate better performance. GDP weights are almost the opposite of equal weights. Since equal weighting outperforms (by tilting toward smaller countries), GDP weighting underperforms (by tilting toward bigger ones). Smaller countries tend to carry a risk premium, similar to why small-cap stocks have historically outperformed large-caps. You get compensated for taking on that extra risk.

If anything, the authors found that inverse GDP weighting works better. They built an index that gives more weight to smaller economies and less to bigger ones. After briefly lagging during the 2008 crisis, their inverse GDP index mostly outperformed from there.

More sophisticated rules like weighting by debt-to-GDP don’t seem promising either. Safety only pays during short sharp selloffs, but not over the medium term. The risk premium that riskier countries carry more than compensates for the occasional blowup. The authors’ suggestion? Use the trading rules from earlier chapters as the basis for truly smart indexes.

Frontier Markets: Fun While the Music Plays

Frontier markets are the less developed corners of EM. Think countries that don’t meet the liquidity or market structure requirements to be in the main benchmarks. They’re mostly B-rated (the EMBI is mostly BBB), about 40% African, and they almost exclusively issue external (dollar) debt because local currency bonds are harder to sell when you’re a weaker credit.

The JPMorgan NEXGEM index tracks frontier credits, and it has massively outperformed the EMBI since 2003. But here’s the thing: it also got crushed harder during 2008. The information ratio is only slightly better (1.42 vs 1.38 for EMBI). Frontier markets are basically a leveraged bet on the same trade.

Two things make frontiers interesting anyway. First, there’s an inclusion rally whenever a frontier market graduates to the main EMBI index. You get paid for being early. Second, the high correlation between NEXGEM and EMBI means you can hedge your frontier exposure using the more liquid EMBI when things get ugly. So even though individual frontier markets are hard to hedge, a basket of them is manageable.

Portfolio Allocation: Markowitz Is Dead, Long Live Simplicity

The classic Markowitz mean-variance framework is basically useless in practice. It produces extreme corner solutions. Return forecasts are terrible. Even Black-Litterman, which tries to fix the forecasting problem, still suffers from the same issues.

So what actually works? The authors tested several approaches for combining local debt (GBI-EM) and external debt (EMBI):

Risk parity allocates equal risk to each position, usually by leveraging up the safer asset. Applied to GBI-EM and EMBI using three-month trailing volatility, it produced decent results. But a simple 50/50 blend of the two indexes outperformed it. No fancy math needed.

For FX specifically (where liquidity is better), the picture changes. The authors tested three approaches against the GBI-EM currency benchmark:

• Equal volatility (risk parity): IR of 1.31
• Hierarchical risk parity (HRP): IR of 1.23
• Equal weight: IR of 1.14
• GBI-EM currency benchmark: IR of 0.61

All three alternatives crushed the benchmark. Risk parity actually wins in FX, which is interesting since it didn’t win for bonds.

What Is Hierarchical Risk Parity?

HRP is worth explaining because it’s clever. The idea, from Marcos Lopez de Prado (2016), is that the usual variance-covariance matrix has a lot of estimation error in it. You can reduce that error by imposing a hierarchy on the matrix.

Step one: cluster currencies by how correlated they are. Poland and Hungary end up together. South Korea and Singapore get grouped. Thailand is nearby but one node removed. The clusters are intuitive and match how you’d think about these currencies fundamentally.

Step two: reorganize the covariance matrix so similar currencies sit next to each other.

Step three: allocate based on this reorganized cluster structure.

The results are strong. HRP comes in second to simple risk parity in IR terms, but all structured approaches blow past the benchmark. The takeaway is that portfolio construction itself generates alpha. You don’t need perfect forecasts if you build the portfolio intelligently.

Derivatives: Weapons of Mass Alpha

The authors feel strongly that efficient EM portfolios need derivatives. Even in long-only portfolios, here’s why:

Curve trades. Earlier chapters showed that rate cutting cycles are best played at the short end of the curve. In credit, the biggest opportunities can be at the short end when a sovereign avoids default against expectations. Without derivatives, you can’t isolate these trades.

Swap spread trades. These are mostly uncorrelated with the rest of your portfolio. Swap spreads are highly mean-reverting and have a weak directional component. Adding an uncorrelated return stream is always beneficial.

Tax efficiency. In countries like Brazil, Colombia, and Indonesia, capital gains or withholding taxes eat into returns. Total return swaps can sometimes get around these.

Liquidity. In many markets, derivatives are more liquid than bonds. You can replicate a whole country index (all bonds with maturity above one year) with just a few interest rate swaps.

Separating FX and rates. This is maybe the most important point. Without derivatives, your default position is always long bonds, and you hedge with FX whenever you turn bearish. That works when shocks are big (like 2008) or in less developed EM where rates follow FX closely. But in more mature EM countries, rates can sell off while FX rallies, and vice versa. As more EM countries develop, separating FX and rates bets becomes essential. Interest rate swaps make this possible.

ESG in EM: Not Fair

ESG investing has exploded. By end of 2018, assets managed with at least some ESG component hit $12 trillion in the US alone, roughly 25% of total US assets under management. An entire ecosystem of ESG funds, benchmarks, research firms, and academic studies has sprung up.

In EM, the JPMorgan ESG index performs almost identically to its non-ESG version. The authors see this as good news: ESG isn’t costing investors money. You get a free lunch, you just don’t get paid to eat it.

But here’s where it gets complicated. ESG criteria were mostly designed with developed-market equities in mind. There are over 100,000 publicly listed stocks to choose from, so excluding some doesn’t hurt your opportunity set much. In EM sovereign debt, you have fewer than 100 investible countries. Excluding even a few can seriously limit diversification.

The CFA Institute and the UN-backed PRI list dozens of ESG criteria across governance, social, and environmental categories. Everything from corruption and rule of law to CO2 emissions and biodiversity. The question is: how do you weight all of this? Equal weighting is lazy. Letting data availability drive the focus is worse. And combining everything into one score lets bad behaviors hide behind the average.

EM Is Actually Better at ESG When You Adjust for Poverty

This is the most interesting argument in the chapter. Raw ESG scores obviously favor rich countries. Earlier stages of development come with more pollution, worse labor protections, and weaker governance. How much growth a country is willing to sacrifice for cleaner air depends a lot on how wealthy it already is.

But applying rich-country standards to poor countries isn’t really fair. EM countries point out, with some justification, that stricter environmental laws in developed countries are part of what pushed dirty industries to set up shop in EM in the first place. Part of the loss of US manufacturing jobs to China was a reaction to increased worker protection and environmental standards at home.

The authors ran a simple but revealing exercise. They took CO2 emissions per unit of GDP and regressed it against GDP per capita (and GDP per capita squared, since the relationship isn’t linear). Then they ranked countries by the residuals.

The results flip the narrative. In raw terms, the median developed market ranks at 51 (cleaner) while the median EM ranks at 36. DM countries pollute less, no surprise. But after adjusting for development stage, the median DM country moves to rank 19, making it a heavier emitter than the median EM country. The US makes it into the worst 10 and scores worse than China.

When you look at changes (who’s improving fastest), some of the worst polluters have improved the most, including Ukraine, South Africa, and China. ESG-focused funds should be rewarding that improvement.

A Proposal: Single-Factor ESG Funds

The authors propose something that makes a lot of sense. Instead of cramming every ESG factor into one aggregate score, offer single-factor ESG funds. One fund focused purely on CO2 emissions. Another on governance. Another on labor standards.

The weighting of ESG factors would then be determined by how much money flows into each strategy. This is more transparent because investors can see exactly how much each ESG factor costs in terms of performance. And it creates better incentives for governments, since each specific behavior has a clear reward attached to it.

They also suggest focusing on externalities. An externality means the country doesn’t fully bear the cost of its behavior. CO2 emissions are the classic example: cheap dirty energy boosts the local economy, but the climate cost is shared globally. If there are no externalities, countries will probably clean up on their own as they get richer. But with externalities, the incentives are never right without outside pressure.

One more problem: ESG data is painfully slow. CO2 emissions data is yearly, with another year of publication lag. Countries that improve don’t get rewarded quickly. The authors suggest using natural language processing on local newspapers and internet data to generate more timely ESG scores. Faster feedback loops mean investors can focus on changes rather than just levels.

The Bottom Line

Portfolio construction in EM isn’t just about picking the right countries. How you weight them matters enormously. The existing benchmarks aren’t great, and simple alternatives like equal weighting or GDP weighting each have their own problems.

The practical takeaways:

• Market-cap-weighted bond indexes reward profligate borrowers. Not ideal.
• Equal weighting works in bull markets but blows up during crises and has liquidity constraints.
• GDP weighting underperforms because it’s the opposite of equal weighting.
• Frontier markets are higher-beta EM, useful for alpha but not truly diversifying.
• A simple 50/50 blend of local and external debt beats risk parity for bonds.
• For FX, risk parity and HRP both crush the benchmark. Portfolio construction alone generates significant alpha.
• Derivatives aren’t optional. You need them to separate FX and rates, play the curve, and manage taxes and liquidity.
• ESG scores are biased against poor countries. Adjusting for development stage and focusing on improvements rather than levels would make ESG more fair and more useful.

The thread running through all of this: don’t be lazy with your portfolio construction. The benchmark is not your friend, and a little bit of structure goes a long way.

Previous: EM Credit Part 2

Next: Big Data, Machine Learning, and the Future of EM

Book: Trading Fixed Income and FX in Emerging Markets Authors: Dirk Willer, Ram Bala Chandran, Kenneth Lam Publisher: Wiley Year: 2020 ISBN: 978-1-119-59905-0