The Credit and Liquidity tab is quiet right now. The Credit Stress model reads 10 out of 100, tagged MINIMAL, and the written summary on the page says credit conditions show minimal stress while growth and the labour market are described as growing. That is not the interesting state. The interesting state is the one where the same tile has climbed, the financial press has found a phrase for it, and you are trying to decide whether to sell something.
Start from a piece of arithmetic that does most of the work. Economic downturns are rare. Episodes of credit spreads widening noticeably are not rare at all. Two things happen at very different frequencies, so the majority of the widenings must resolve without the downturn. That is not optimism, it is counting, and it means the base case for any given credit scare is that it fizzles.
Why I am not handing you a table of dates
The natural shape for this article would be a list of episodes with years attached and spread levels next to them. I am not going to write that, and the reason matters more than the table would.
Numbers recalled from memory and dressed up as a dataset are how bad decisions get made confidently. If I tell you spreads widened by a specific amount in a specific year and then tightened over a specific number of months, you would reasonably treat those figures as measured, and they would not be. The scorecard is a live product reading current data, not a historical archive of past episodes with dates attached, and nothing on the page in front of me establishes what happened in any particular past year.
What I can give you is the structure, which is more durable than a table anyway. Widenings that resolved into nothing have recognisable shapes, and once you have seen the shapes you can classify the next one in real time using series you can look at yourself.
The four shapes a false alarm takes
Almost every widening that goes nowhere fits one of four descriptions, and each has a tell.
- Supply indigestion. A heavy run of new borrowing hits the market and the index widens because buyers have to be paid to absorb it. The tell is that the widening is broad and shallow and that new deals keep getting done. Nothing has gone wrong with the borrowers, there are simply too many of them at once.
- One sector doing the damage. A single industry with a concentrated presence in the index reprices hard, usually because of a commodity price or a specific policy change, and it drags the average with it. The tell is that the index moved while most of its constituents did not. This is the one people most often mistake for a systemic signal.
- A liquidity event with no credit deterioration. Somebody large has to sell, or dealers pull back their willingness to hold inventory, and the price of transacting rises sharply while the ability of borrowers to pay is unchanged. The tell is that it is violent, it is fast, and it affects the most liquid instruments first, which is backwards for a genuine credit problem.
- A policy scare that reverses. The market prices in a path for rates or for a specific intervention, credit widens on that path, and then the path changes. The tell is that rates and credit move together in lockstep, which is unusual when the driver is actually credit quality.
Notice what the four have in common. In none of them is the ability of companies to service their debt actually changing. That is the distinction the tile cannot draw for you, because a spread is a single price and all four of these show up in it identically.
Building the catalogue yourself, properly
If you want the table, build it. It is an afternoon of work and it is worth more than a borrowed list because you will remember doing it.
The spread series and the recession dating are both published by public sources and both are free. Pull a long history of a high yield spread series, mark every episode where it widened materially from a local low, and mark separately the periods the economy was formally in contraction. Then classify each widening as one that overlapped a downturn and one that did not.
Two rules keep this from becoming an exercise in fooling yourself. Define "materially" before you look at the chart, not after, or you will unconsciously pick a threshold that makes your story work. And write down the classification of each episode using only what was visible at the time, since with hindsight every episode looks obviously benign or obviously ominous, and hindsight is exactly the faculty you will not have available when it matters.
When you are done you will have something the scorecard does not provide and does not claim to, which is your own dated history with your own definitions attached.
The features the benign episodes share
Three checks separate a scare you can discount from one you should respect, and all three are available to a retail investor without a data terminal.
The first is breadth. Did the whole credit complex widen, or did one part of it move enough to drag the average? A broad, even widening across quality tiers and sectors is the one that deserves attention. A widening concentrated in one place is usually a story about that place.
The second is whether anything else agrees. This is where the model row on the scorecard earns its keep. The seven models are built on deliberately different inputs, and today they read 41 from Blockcircle Labs, 30 on the GDP two quarter rule, 30 on the yield curve, 20 on the Sahm rule, 10 on credit stress, 39 on leading indicators and a combined score of 29. If the credit model climbs and the labour and output models sit still, you are looking at one witness. The counter labelled models at or above 60 currently reads 0 out of 7, and that counter moving is a different event from one tile moving.
The third is persistence. Supply effects and forced-seller effects clear once the flow is absorbed, usually inside a few weeks. Deterioration in the ability to pay does not clear, it grinds. Waiting three or four weeks before acting costs you the first part of any real move and saves you from most of the fake ones, and on a retail account with transaction costs and taxes that is a trade worth making.
What to do with a scare you have decided to discount
Discounting a warning is not the same as ignoring it, and the difference should show up in your account.
The useful response to a credit scare you believe is one of the four benign shapes is to change the pace of what you were going to do anyway rather than to reverse it. Stop adding to the most speculative position. Hold new cash rather than deploying it on the usual schedule. If you had planned to increase risk this month, do it next month instead. Each of those is close to free if you are wrong about the scare, which is the property you want in a decision made under this much uncertainty.
Write the classification down at the time, with the date and which of the four shapes you think it is and what would change your mind. Then look at the note three months later. Most people discover that their real-time classification was worse than they remember and that the thing they said would change their mind never got checked. Fixing that is a bigger improvement to your results than any refinement to which spread series you watch.