Investment Grade vs High Yield: Two Different Risk Machines
Investment grade is a rates instrument with a credit accent; high yield is an equity instrument with a coupon and a ceiling. Treating them as one quality dial is the allocation error.
The conventional mental model of corporate credit is a single dial running from safe to risky: investment grade at one end, high yield at the other, with the yield you collect rising as you slide along it. Sophisticated investors mostly know the dial is a simplification. What is less widely internalised is that it is not merely simplified but structurally wrong — the two markets are not the same machine at different intensity settings. They are different machines. The risk factor that dominates an investment grade bond's return variance is not the one that dominates a high yield bond's, and nearly every allocation mistake made with credit traces back to ignoring that flip.
What each machine is actually built from
Start with investment grade. For a high-quality issuer, default over a typical holding horizon is a remote event, so the compensation for expected default loss is a modest slice of the yield. What the holder actually owns, in variance terms, is duration: investment grade issuance tends toward long maturities because strong issuers can sell them, so the instrument's price is dominated by moves in the underlying government curve, with a second factor — spread beta, the market-wide repricing of credit risk premium — layered on top. Call it what it is: a rates instrument with a credit accent. The credit accent matters at the margin and matters enormously at the ratings boundary, but on an ordinary day the government curve is driving the P&L.
High yield inverts the weighting. Maturities are shorter, coupons are larger, and spreads are a far bigger share of total yield — so the government curve recedes and the credit factor takes over. What the holder owns is exposure to the default-and-recovery process: the probability that individual issuers fail, the severity when they do, and the clustering of those failures in recessions. Add the structural feature that most high yield bonds are callable — the issuer can repurchase the debt at a fixed price once conditions improve — and the return profile becomes recognisably equity-like but with a truncated top: full participation in bad outcomes, capped participation in good ones, and a large coupon as payment for accepting that asymmetry. Call this machine an equity instrument with a coupon and a ceiling.
Why the machines differ: three layers of why
The third layer deserves emphasis because it is the least visible from a price chart. Correlation between assets is usually discussed as if it were a property of the assets. It is substantially a property of the holders. Investment grade behaves like a rates instrument partly because it is held by institutions that buy duration for liability reasons and are insensitive to equity drawdowns; high yield behaves like equity partly because it is held by capital that flees at the same moments equity capital flees. When you buy the instrument, you are also buying membership in its holder base's behaviour — a fact that matters most at exactly the moments diversification is supposed to pay.
The illustration: two crises, two different machines exposed
The 2008 crisis displayed the machines at maximum separation. High yield behaved as its construction dictates: spreads widened to distressed levels, prices fell alongside equities, and the asset provided no shelter whatsoever — the default machine was being repriced in real time. Government bonds rallied hard as policy rates were cut. Investment grade sat between its two factors: spreads widened sharply, but the collapsing government curve offset a meaningful part of the damage for longer-duration holders. The episode is the cleanest qualitative demonstration that the two credit markets were answering different questions — high yield was asked who survives, investment grade was asked what is duration worth in a depression, and the answers moved their prices in different directions relative to equities.
The 2022 episode inverted the lesson, which is precisely why the pair of illustrations teaches more than either alone. That year the shock was the rates curve itself: inflation forced policy tightening, government yields rose sharply, and equities fell at the same time. Investment grade — the rates machine — was hit through its dominant factor and delivered unusually deep losses for a high-quality asset, while high yield, structurally shorter in duration, fell by less than its usual crisis behaviour would suggest even as its spreads moved wider. Both machines drew down with equities, and the diversification that investment grade had provided in prior recessions failed — not because the framework broke, but because it worked: the dominant factor of each machine determines when it protects you, and a shock arriving through rates is the one configuration where the rates machine cannot help. Knowing which machine you own means knowing which shocks it is built to absorb and which it is built to transmit.
The strongest case against the two-machine view
The serious objection is continuity. The ratings scale is granular, BBB and BB issuers are frequently near-identical businesses separated by an agency committee's judgment, crossover investors arbitrage the boundary deliberately, and spread levels across the whole quality spectrum move with substantial common variation. On this view, the two-machine framing reifies an administrative line: risk in credit is one continuous surface, and the boundary is a labelling convention, not an economic object. A second objection compounds it: in severe stress, correlations converge anyway — everything credit-shaped falls together — so the machine distinction evaporates exactly when it is claimed to matter.
The continuity objection is correct about issuers and wrong about instruments and owners. The underlying businesses do form a continuum; the securities do not, because structure (maturity, callability, covenants) and holder rules (mandates, index membership, regulatory capital treatment) change discontinuously at the line. An administrative line that forces selling is not merely administrative — it is a market mechanism, and the fallen-angel repricing pattern is its regular, observable signature. As for stress convergence: correlations of direction converge, but magnitudes and recovery paths do not, and the framework's claims are about which factor dominates variance, not about whether signs occasionally agree. The honest concession is narrower: for crossover-rated issuers specifically, machine identity is genuinely ambiguous, and the framework's boundary claims apply weakly there. That is a boundary condition, not a refutation.
What this changes for a portfolio owner
The operational shift is to stop asking how much credit do I own and start asking which machine each holding is, and what job it was hired for. Investment grade is hired to do a rates job — carry with duration, an offset in demand-shock recessions — and should be judged against that job, including its known failure mode when the shock arrives through the curve itself. High yield is hired to do an equity-adjacent job — harvesting the credit risk premium and the asymmetry payment for accepting capped upside — and belongs, honestly accounted, in the same risk bucket as the equity book it will draw down beside. Sterling's Embedded Intelligence enforces exactly this classification when reading a portfolio: exposures are grouped by dominant risk factor rather than by asset-class label, because labels are how the same risk gets counted twice — or hedged zero times.
How to apply this framework
- Classify by dominant factor, not by label. For each credit holding, ask what drives the variance: the government curve and spread beta, or default and recovery. That answer — not the word bonds — determines which portfolio bucket it belongs in and what it can hedge.
- Interrogate the ratings boundary as a mechanism. The BBB/BB line changes owners, indices, and permitted holders. Watch concentration just above the line as a source of forced-selling risk, and recognise fallen-angel repricing as flow, not only fundamentals.
- Match the machine to the shock you are hedging. Investment grade duration is built to absorb demand-shock recessions and to transmit rate shocks; high yield absorbs neither. If the risk that worries you arrives through the curve, the rates machine is exposure, not protection.
- Price the ceiling in high yield. Callability truncates the upside that would otherwise reward you for surviving the bad states. Judge high yield on the asymmetry actually on offer — full downside participation, capped recovery — rather than on yield alone.