In the name of simplicity, shortcuts are frequent when macro investors talk about implied default rates in the high yield market. Numerous assumptions are swept under the rug in order to make tidy statements like “the market is pricing 4% defaults,” so today we reexamine the issue in some detail. What’s truly in the price depends on the timing of defaults, the coupons, current cash prices, and recovery values of the issuers that default, and importantly, investor risk tolerance. We developed a framework to analyze the implied defaults in the marketplace from a top-down macro perspective, still taking account of available combinations of many of these various assumptions. We sought to determine a range of expected defaults reflected by current prices. Then, specific investor views on the assumptions can further narrow the output within the range. The results are useful! We show that the assumptions can influence the output significantly, but even a conservative perspective still results in implied defaults which are quite high.
We start with a typical high yield index and divide it into subsectors (see sector table below), each with its own weighted average coupon, yield, and price. This allows us to isolate Metals & Mining and Energy (“MME”) from everything else. Clearly with regards to who might default, the commodity-linked issuers are top-of-mind, and we build the flexibility to model defaults concentrated there versus pro-rata throughout the high yield market. We also divide the index into buckets of bond prices (see bond price table below), so we can choose to force the defaults into issuers who are already distressed, versus not. This exercise is similar to stressing MME names, but not exactly the same since not all distressed high yield is directly commodity-linked.
The beneficial effect of sustained carry over time is also significant, such that both timing and persistence of defaults influence the results. Naturally, a high yield portfolio pays a spread over Treasurys to compensate investors for default losses, and that spread is contractual for the life of the bonds, not just the next one year. So even though “implied defaults” are typically quoted offhand as a 1-year number, what should be relevant are cumulative defaults over the average life of the portfolio. The typical high yield index has an average life of about 5 years, and in our model we allow for several patterns of defaults over a 5-year horizon.
Lastly, investor risk preferences are a key determinant of what’s “in the price.” The textbook says that investors will always expect to lose less than the high yield spread over Treasurys via default losses. How much less depends on those risk preferences. The more risk averse investors become, the larger the expected return over Treasurys they will demand after accounting for expected default losses. That expected return is the credit risk premium (“CRP”) and it’s a moving target. Academics fix this number at zero, and do all their calculations in the risk-neutral world. For our initial calculations we fix the CRP at 250bp, which is an accepted historical average. We’ll reintroduce flexibility in the CRP at the end.
We use the composition of the Barclays HY 2% Issuer-Constrained Index as of Jan 20, 2016, which on that day had an aggregate yield-to-worst of 9.8%. Metals & Mining and Energy comprised 5.7% and 14.8% of the par amount respectively (though less as a percent of market value). We extract implied default rates over the 5-year approximate life of the portfolio, allowing defaulting bonds to accrue coupon-only until the year of default, wherein accrual drops to zero and cash price minus recovery crystalizes as the default loss, with no reinvestment of recovery. We model composition of defaults in three ways: 1) defaults occur pro-rata across industry sectors according to their par weightings, 2) defaults are concentrated in the Metals & Mining and Energy sectors until those sectors are completely extinguished, with pro-rata defaults thereafter, and 3) defaults are concentrated in the bonds which are currently trading below $60 until those are extinguished, with pro-rata thereafter.
We model the timing of defaults in two ways: 1) evenly across the 5 years – results reflected in Table 1, and 2) a spike in year 1, which recedes by half in year 2, followed by 3 years of 2% default rate – results reflected in Table 2.
These tables allow a narrowing of implied default rates to the set of scenarios which seem most reasonable – to us that means a default spike in the next year concentrated in MME names, followed by a slow bleed thereafter. We take the conservative view that recovery will be on the low end of the spectrum, which leads us to conclude that index is priced for around 24% to 29% of the market to default over its lifetime, highlighted in orange. Thematically though, it’s possible to conclude that much of the market is priced for default regardless of the combination of assumptions. Looking across both tables with wider ranges of assumptions suggests between 23% and 35% of the market is implied to default, and that lower bound is still quite a large number. According to Bank of America Merrill Lynch data, in the 2008-2009 recession, the rolling 5-year cumulative default rate never exceeded 20%.
All of the calculations above depend on the high yield investor earning a total return equal to Treasurys plus the CRP of 2.5%. In practice, in so far as the CRP reflects extra expected return in exchange for say, reduced liquidity and systemic risk aversion, that risk premium should clearly be widening now. Goldman Sachs Research among others, have refined a model for extracting an implied CRP from market prices over time, and their calculations do in fact show a significant widening of the CRP recently, to north of 4% for the high yield market. The calculation is complex – perhaps a topic for a future blog – but the bottom line still stands. Expected returns are rising, and defaults in-the-price are quite high no matter how you slice it.