GMO March 2017 white paper titled, “The What-Why-When-How Guide to Owning Emerging Country Debt”
As GMO enters its 23rd year managing emerging debt portfolios, we offer our perspectives on the frequently asked questions that have come up over the years, including:
- What are the characteristics of external, local, and corporate emerging debt?
- Why and when should one own emerging debt, considering diversification, alpha, and value?
- How should one own emerging debt? Dedicated external, local, corporate, or blended? Active or passive?
What are the characteristics of external, local, and corporate emerging debt?
Emerging debt has evolved as a concept over the last 30 years. In the late 1980s, the asset class got its start when western banks securitized defaulted sovereign bank loans into tradable “Brady bonds.” Over the years, the countries have refinanced these issues into external debt (debt issued in foreign currency) and/or local currency debt (debt issued in their own currencies), while corporates domiciled in these countries have issued corporate debt, mostly in foreign currency.
The graphs in Exhibit 1 depict the yields and the transactions costs (indicative bid-offer spreads on the bonds) associated with the three main benchmarks for the sub-asset classes. The inset exposures table shows the principal macro exposures associated with each type (e.g., “what exposures am I taking to get those yields?”). Appendix 1 discusses language for the risks of each type with more detail, particularly on liquidity risk. The transactions cost graph partially answers the question: “what is the indicative cost to replicate the exposures in the benchmark” assuming “normal” transaction sizes and no market impact? The main conclusions: the bundle of macro exposures varies across type, and the replication costs are high normally and sometimes, around crises, very high.
External debt (USD Sovereign and Quasi-Sovereign)
External bonds are represented by the J.P. Morgan Emerging Markets Bond Index Global (EMBIG), which includes sovereign issues as well as those sub-sovereign and corporate issues that are either 100% government guaranteed or where the issuer is 100% federal government owned. The main EMBIG index is restricted to U.S. dollar issues settled in Euroclear, although J.P. Morgan also produces a euro-denominated benchmark separately. For EMBIG, the exposures table in Exhibit 1 highlights the two principal exposures of the class: sovereign default risk and U.S. dollar interest-rate risk. The weighted average rating is Ba1/BB+/BB+ from Moody’s/S&P/Fitch, and the U.S. interest-rate duration is 6.7. The weighted average price bid-offer is around 80 bps, or just shy of 40x that of an on-the- run U.S. Treasury issue. Note that J.P. Morgan also publishes a “diversified” version of the index (EMBIG-D), in which larger issuers are capped, although the beta of the two is close to one. Each has 65 countries representing 142 issuers. Lately, EMBIG-D has overtaken EMBIG in terms of assets tracking it, according to J.P. Morgan.
Local currency debt (Sovereign)
Nominal1 local currency bonds are represented by the J.P. Morgan GBI-EM family of indices (broad, global, narrow), of which the “global diversified” (GBI-EMGD) is the most widely used. As the benchmark currently only captures 15 countries, some of which are large issuers, investors prefer the capped issuer diversified version. Local currency bonds are bonds issued in the country’s own currency, regardless of whether such a bond is issued and settled locally under local law or in “global” format under foreign law settling in Euroclear. In the first case, the investor first sells his home currency, buys the local currency of the country in question, and then buys the bonds, reversing the process on exit. In the latter case, all transactions (purchase, sale, coupon, and principal payments) settle in U.S. dollars at a “fixing” exchange rate (generally near the current spot rate). In the first case the investor takes cross-border exposure (the risk that upon exit capital outflow restrictions may exist); in the latter case the investor takes sovereign default exposure (the risk that the country may not be able to deliver U.S. dollars). As both are types of sovereign exposure, we view local debt instruments (from a foreigner’s perspective) as having sovereign default risk (GBI-EMGD rating is Baa2/BBB/ BBB). Furthermore, local debt instruments carry local interest-rate exposure (duration is 4.9) and currency exposure. The weighted average price bid-offer including the FX is around 50 bps. Barclays produces a broader index that includes 19 countries (capturing some more developed countries like Korea, Singapore, Israel, and Czech Republic). Given the country composition difference, the beta of the J.P. Morgan index has been 1.14 with respect to the Barclays index, and since 2013 the J.P. Morgan index has underperformed the Barclays index by a cumulative 7.4%.
Emerging corporate bonds are represented by the J.P. Morgan CEMBI family of indices (with broad/narrow and diversified/undiversified variants). The CEMBI Broad Diversified (CEMBIB-D) is the most widely followed. It is a USD-denominated index of bonds issued by corporates domiciled in emerging countries. Interestingly, J.P. Morgan departs from its EMBIG/GBI-EM method for defining “emerging countries” in this context, adopting a regional-based approach instead. In CEMBIB-D’s case, companies headquartered in Latin America, Eastern Europe, Middle East, Africa, or Asia ex-Japan are considered eligible, as are those with 100% of their operations there (as long as the bonds are guaranteed by the local operation). 100% government-owned quasi-sovereigns eligible for EMBIG are specifically excluded, but partially government-owned quasi-sovereigns make up about 30% of CEMBIB-D. A company can migrate from one index to the other based on nationalizations/privatizations. CEMBIB-D contains issues from 51 countries spanning A1/A-rated Qatar and Taiwan to unrated Iraq. The average rating is currently Baa3/BBB-/BBB. Given that the corporates are domiciled in emerging countries that have sovereign default risk themselves to greater or lesser extents, we expect such issuers to pay a spread premium over and above the sovereign’s own credit spread to account for their idiosyncratic risks. As the issues are in U.S. dollars, they also carry U.S. interest-rate risk. The issues contained in CEMBI are generally also included in the widely followed global corporate indices provided by Barclays/Bloomberg, with investment-grade issues included in the Global Aggregate and sub-investment-grade issues in the Global High Yield.
Exhibit 2 details some other attributes of the benchmarks as well as the MSCI Emerging Markets index (MSCI-EM), the emerging equities benchmark, for reference. After all, emerging debt often competes with emerging equities for risk assets.
Based on the table, we observe:
- For all classes of emerging debt, the fraction covered by the relevant benchmarks is much smaller than for equities. This suggests that there are more opportunities for alpha by taking off-benchmark exposure in debt than in equity.
- The local currency sovereign debt universe is 5.8x the size of external sovereign debt, but the index-captured fraction is much smaller in local debt. This is due to the relatively low fraction of the universe captured by the local debt index.2 The most widely followed local debt benchmark has a market capitalization that is 1.6x that of the most widely followed external debt benchmark.
- In external debt, there are two benchmarks with significant assets tracking them, while among the local debt and corporate debt varieties, one benchmark has become dominant. Exhibit 3 details the differences between the two external debt benchmarks, which we don’t view as material in most states of the world (correlation among the two is 1). The key difference is the down-weighting of some of the larger countries (Mexico, Indonesia, China, Russia) and the consequential up-weighting of some of the smaller countries (Peru, Panama, Uruguay, Poland). The overall ratings and spread durations are the same. In comparison, the diversified version has an investment-grade/non-investment- grade split tilted toward the latter; more Africa (due to the up-weighting of smaller countries), less Latin America (mostly less Mexico); and fewer quasi-sovereigns.
?? Appendix 2 shows country-level details across the debt and equity indices. Exhibit 4 shows the countries with at least a 4% weight in one of the major benchmarks. Although China has a big presence in equity markets, corporate bond markets, and quasi-sovereign corporate markets (in EMBIG), it has no investable local currency debt at present. Mexico, Brazil, and Russia have material presences in all benchmarks, making them (with China) the most systemically important to investors with multiasset emerging portfolios.
Why and when should one own emerging debt, considering diversification, value, and alpha?
As with any risk asset, the time to own it in risk-seeking portfolios is when its prospective returns adequately compensate you for its risks, taking into account the diversification potential with other risky assets you already own or may want to buy. We distinguish historical statistical diversification, which affects rebalancing potential, from fundamental diversification, which becomes more important in event scenarios that deviate from statistically-implied ones. In such events one asks: “how much do we own of what?”
From a statistical diversification perspective, using data from 2005 to 2016, J.P. Morgan calculates that:
- All three debt benchmarks have similar correlations with the S&P (~60%), and EMBIG and CEMBIB-D have ~70% correlations with emerging equities.
- Local debt is even more correlated with emerging equities, likely due to the common currency effect.
- EMBIG and CEMBIB-D, dollar-denominated spread products, have higher correlations with other dollar spread products, such as U.S. investment-grade corporates or U.S. high yield.
From a fundamental diversification perspective, we note that the most likely set of uncomfortable exposure questions would likely come from common issuers in CEMBIB-D and the MSCI-EM index family (MSCI-EM plus MSCI-Frontier) – the overlap between CEMBIB-D and MSCI-EM is noticeable. Appendix 3 details our thoughts on this.
The opportunity to do better than the already fairly high yields associated with emerging debt is another key reason investors take interest in emerging debt. Exhibit 6 uses manager data from eVestment, which compiles universes for external and local debt, among other asset classes, to show the returns of the top quartile and median managers relative to the respective benchmarks. We observe:
- In external debt, the median manager beats EMBIG by 1.3-2.3% over the periods shown, with top (5th percentile) managers 50+ bps more than that.
- In local debt (a newer class with less history), the median manager is in line with the benchmark generally, while top managers have added significant alpha, even over the years when local debt struggled in absolute return terms due to the strength in the U.S. dollar.
We conclude that (a) active management is useful; and (b) picking a top manager in local is more important. This is why GMO’s Asset Allocation team includes the expected alpha in its assessment of the attractiveness of emerging debt, since emerging debt alpha has been higher and more persistent.
We find that the reason the median manager struggles relatively in local debt is (a) the benchmark is gross, while the manager pays high custody fees and taxes (J.P. Morgan has estimated this to detract up to 1.5% annually at times from net returns; in external debt the bonds are all Euroclear, with no tax and very low custody fees); (b) a fairly narrow benchmark (only 15 countries) is a limited opportunity set for very benchmark-aware managers; and (c) there is a temptation among many managers to take macro views, given the nature of the exposures involved, and macro views are (in our opinion) low information ratio ideas.
If we recall the exposures table in Exhibit 1, we refer to “top-down” styles as those that involve any of the dots in the table. For example, in local debt this might mean taking a view on the direction of local interest rates or the currencies, a temptation given the low breadth (15 countries) of the index. In external debt, with 142 issuers (64 sovereign and 78 quasi-sovereign), the menu is wider and the terminology a bit confusing. We refer to “top-down” as choosing one macro aspect of a country over another (whether credit, currency, or interest-rate), and we refer to “bottom-up” as choosing one particular bond over another (“security selection”) on a particular credit curve. The security selected bundles any one or more of the credit/currency/rate exposures seen in the table, with some very entertaining examples along the way.3 For our complete thoughts on this topic, please see our 2010 CFA paper, “Deconstructing and Reconstructing Emerging Market Debt: A Bottom-Up Approach to EMD Investing” (available on www.gmo.com).
A gray area with confusing terminology, therefore, includes in-benchmark quasi-sovereigns (applicable only to EMBIG-relative). Were this an equity portfolio, one would say company selection was “bottomup.” However, for a debt portfolio, where there can be many issues (PEMEX, for example, has 26 bonds representing 6.3% of EMBIG), the company choice is a kind of “market selection” distinct from the issue selection. While in equities there is generally only one instrument per company, in debt there may be many bonds of differing levels of seniority or with different creditor rights in the offering documents.
GMO distinguishes itself in the industry as being the value manager that emphasizes bottom-up issue selection as its main alpha driver. For example, of the 32 Mexican sovereign individual bonds in the combined EMBIG and GBI-EMGD benchmarks, which is the best one/ones to own? As all carry sovereign default risk (per the Exhibit 1 exposures table) from a foreigner’s perspective, there’s no need to own them all. In fact, sometimes one outside both benchmarks (excluded merely for its currency denomination – British pound sterling, say) can be even better. After all, hedging GBP currency and interest-rate risk is a trivial exercise for bond managers. We believe that taking this approach is what has allowed us to be at or near the top of the alpha rankings over long periods of time.
To determine absolute value, we begin with the prevailing yields on the three sub-classes, divide those up into payments for known risk factors (from our exposures table), and then ask ourselves: is it enough? Below is a summary of our valuation methods for these principal risks. Once a quarter we update these in our Emerging Debt Report (available on www.gmo.com).
External debt: sovereign default and U.S. interest rates
For sovereign default valuation, we first calculate a “fair” spread of the EMBIG, accounting for the credit rating profile of the EMBIG (used to estimate default probabilities), an estimate of future historical credit transition (based on tabulations of historical sovereign transition), and a recovery value assumption. We then take the ratio of the actual EMBIG spread to the fair value spread and compare it to its history. Assuming a long-term investment horizon, Exhibit 7 suggests that the market shows signs of being attractive when the fair value multiple is above the long-run average and median lines, and unattractive when it lies below. We encourage credit investors to do the same exercise for other credit asset classes to determine relative value.
We haven’t made a similar computation for CEMBI, although given the extra risks outlined in Appendix 3, as conservative credit managers, we’d be inclined to say that the multiple should be higher in all cases due to the further illiquidity of corporates as seen in the bid-offer chart in Exhibit 1. We’d also add an extra hurdle when adding them to portfolios that already have emerging equities, given some of the fundamental overlap.We haven’t made a similar computation for CEMBI, although given the extra risks outlined in Appendix 3, as conservative credit managers, we’d be inclined to say that the multiple should be higher in all cases due to the further illiquidity of corporates as seen in the bid-offer chart in Exhibit 1. We’d also add an extra hurdle when adding them to portfolios that already have emerging equities, given some of the fundamental overlap.
For U.S. interest rates, to which external debt is also sensitive, we count ourselves among the fixed income managers who say “no clue!” Instead, we, along with most, simply look at the level and the curve-implied forward interest rates and wonder when these low rates will end. We share our “hairy graph,” which depicts these ideas, with our clients when discussing (lamenting!) the issue (Exhibit 8).
Local debt markets
To determine local debt market value, we attempt to value the currencies and the local interest rates. Caveat: we believe that currency valuation is the hardest of the macro variables to value, harder even than interest rates. Attempts at currency and interest-rate valuation can be hamstrung by policymaker objectives, so we are careful to make only broad statements on these topics. For currencies, our model analyzes trends in real effective exchange rates and the balance of payments, and measures how far away current values are from their medium-term averages. These are combined into a single value score, as shown in Exhibit 9, where we compare the weighted average of currencies in the GBIEMGD with values for the USD and EUR. As the Exhibit indicates, scores below the zero line indicate potentially “cheap” currencies, while positive scores indicate potentially “rich” currencies, with three-sigma extreme zones highlighted. It matters the funding currency, too, which is why we plot two of the majors: USD and EUR. Please ask us to see this graph relative to other funding currencies.
For local interest rates, we compare real yields offered on local debt (bonds less inflation expectations) to those of the majors (USD and a simple average of USD/EUR/JPY). We use this to determine the “cushion” of emerging local debt relative to the very QE-depressed real yields in G3. We find currently that EM local real yields are in line with their long-run averages, although the real yield differential to G3 is very large presently given their financial repression policies. Indeed, policymaker objectives in the G3 are very at odds with any investor’s measures of value!
See the full PDF below.