Why does credit move in cycles?

Credit conditions do not wander randomly. They move in a recognisable loop that has repeated across markets and decades, driven by a feedback mechanism between lender behaviour and borrower behaviour. Understanding the loop is not the same as timing it — nobody in this course is going to tell you where in the cycle anything currently sits.

The loop

1. Losses are low and funding is cheap. Defaults are rare, recent experience is benign, and capital wants to be lent.
2. Lenders compete. Spreads compress. Competition moves from price to terms: bigger leverage multiples, weaker covenants, more generous EBITDA definitions, longer maturities.
3. Weak credit accumulates quietly. The stock of fragile borrowers grows precisely while nothing appears to be going wrong. Nothing looks wrong, because nothing has been tested.
4. A shock or a repricing arrives. Rates rise, demand falls, a sector turns. Refinancing gets harder for the marginal borrower.
5. Defaults rise, recoveries fall. As Recovery and loss given default showed, these happen together.
6. Lenders retrench. Spreads widen, terms tighten, credit availability shrinks — which itself causes defaults that would not otherwise have happened.
7. Losses are taken, weak structures are cleared, standards are strict. New lending done here is done on excellent terms — and the loop restarts.

The uncomfortable symmetry

The structural point worth carrying: the worst loans are made when spreads are tightest, and the best loans are made when spreads are widest. Underwriting quality moves inversely to compensation. Practitioners talk about credit vintage for exactly this reason — the year a loan was written encodes the standards prevailing when it was written.

Maturity walls

Debt does not come due evenly. The aggregate calendar of when outstanding bonds and loans mature is called the maturity wall, and it concentrates refinancing risk into specific years. A borrower with no maturity for four years can survive a bad market that would kill an identical borrower facing a refinancing in six months. Same credit quality, different timing, different outcome.

What the cycle does to expected loss

Combine the two components with the correlation from unit 3:

• Benign year: PD 2%, recovery 45% → LGD 0.55 → EL = 110bp
• Stress year: PD 10%, recovery 25% → LGD 0.75 → EL = 750bp

Default frequency multiplied by 5; expected loss multiplied by nearly 7. This is why credit returns are not symmetric through time: a long run of quiet years accumulating spread income, punctuated by concentrated periods where a large fraction of it is handed back.

Why this is a mechanism, not a forecast

The loop is well documented, but its duration is not. Phases have run from a couple of years to well over a decade, cycles differ by market and by sector, and policy intervention has repeatedly changed the shape of the descent. Anyone who tells you the cycle is a clock is selling something.

Use it as a lens: it tells you which questions to ask about the terms on which credit is currently being extended, and what would have to be true for conditions to change. It does not tell you when.

Try it now

Look at a full cycle:

1. Below is the full available history of a high-yield bond fund. Each sharp dip in it is a spread-widening episode showing up as a price — mark every one you can find.

2. Now a distress reading rather than a price. The New York Fed's distress index for high-yield bonds is published weekly and runs from 0 (calm) towards 1 (distress); it is built from spreads and market conditions, and is not itself a spread. Across its full history, 7 January 2005 to 21 August 2026, its highest reading was 0.78 on 17 April 2009 and its lowest 0.06, first reached on 26 July 2024 (measured 28 September 2026). Eight weeks of early 2020 are below. Check that 2009 and March 2020 line up with the two deepest dips on the chart above.

3. For the 2020 episode, measure the fund's fall from its early-2020 high to its March low on the chart. Treat the whole fall as spread widening and turn it into basis points using price change ≈ − spread duration × spread change with a duration of 5. Then ask how many years of spread income at an illustrative pre-widening spread of 400bp it would take to earn that back. That ratio is the cycle in one number.