Resolves YES if the NBER Business Cycle Dating Committee declares, by the end of 2025, US Eastern Time, that the US entered a recession at any point in 2023 or 2024. Resolves NO on 2026-01-01 otherwise.
Related questions
I constructed the Sahm Rule indicator based on the monthly U3 unemployment rate posted on the FRED website from 1970 onward and recorded every event where the indicator reached a level of 1/3 (0.33) or higher (except Covid which is not a natural business cycle anyway). For each of these events, I plotted a line of the indicator with the time of the event in the center of the chart (t=0). Each differently colored line in the chart corresponds to a different time period and shows how the Sahm Rule indicator behaved 6 months prior to and after the event. I labeled each line with the date of the event. Dashed lines indicate recession.
As you can see from the chart, every event happened after a recession had already begun with the exception of June 2003 which was a false signal. However, keep in mind that the economy had recently come out of a recession that ended in November 2001, less than 2 years prior. The Sahm Rule indicator reached the 1/3 threshold in April 2024 and stayed there in May.
We know there is a lag of up to a year between the NBER-defined recession start date and the date of the announcement. Hence, considering this empirical data alone (which may or may not be prudent) it suggests that the US economy is more than likely already in a recession. Also, keep in mind that the historical data used to construct this chart are final revisions, and that U3 unemployment figures are often upwardly revised during periods of recession, meaning that the actual final revisions for the U3 unemployment rate for the first half of 2024 could reveal an even sharper rise in the Sahm Rule indicator, possibly already crossing the "original" 0.5 threshold.
When will we get the Sahm Rule indicator for June 2024? https://fred.stlouisfed.org/series/SAHMREALTIME
@Peregrine you might like this Axios article: https://www.axios.com/newsletters/axios-macro-aa394170-45de-11ef-976b-55be70cd83e6.html
They explain that all indicators are flashing red (or about to) but that this time is different :)
By the way, regarding the yield curve inversion, I thought the signal wasn't when it went negative but when it went back above 0 (or -0.15?).
Yup. Based on a long list of empirically tested indicators you should predict recession by EoY 2024 with at least 80 % probability unless you think something has fundamentally changed about business cycles which is precisely what the prevailing sentiment is right now. The truth is that prior to almost every past recession people have been saying the exact same thing that they are saying now: "Yes the indicators point to recession, but this time is different because XYZ". Newsflash: People have always attempted to find some XYZ reasons to justify why this time is different. It almost always ends badly. That's why recession odds are typically the most reliably mispriced bets you can find, both in financial and betting markets, because people keep falling prey to the same delusional thinking.
I agree. Trends on prediction markets:
Here: min of 13% in May, now 21%
Kalshi: min of 7% in April, now 11%
Metaculus: min of 16% in May, now 20%
As we gather more economic data about 2024 and get closer to the end of the year, the probability of a recession this year should go down if the situation was good, not up...
Also interesting: https://www.businessinsider.com/us-recession-outlook-inflation-costs-consumer-spending-affirm-soft-landing-2024-7
Kalshi traders have this at 8% - what am I missing? They are using a different statistic but both are trying to measure negative GDP growth. https://kalshi.com/markets/recssnber/recession
@KevinBurke I think Manifold has an unusually high risk-free interest rate at the moment—even for Manifold—because of the pivot to real money (e.g., smart traders losing liquidity or even leaving the platform). Keep in mind this "end of 2024" market actually might not resolve until 2026-01-01. Still, I'm not sure if that's a full explanation for a ~21% price.
Personally I basically just bet on EMH for basically of these economic and financial markets that aren't long-term enough for transformative AI to apply, and I'm mostly holding positions to rack up as many loans as I can before those are no longer available.
@Jacy I also just think kalshi is wrong here
8% is an absurdly low probability given the yield curve inversion, second spike in inflation, geopolitical risk, reliance of US fiscal deficits, inconsistencies of BLS employment measures, low PMI reads and banking risk from higher rates for longer
@IsarBhattacharjee Well, Kalshi's market is about negative GDP growth this year specifically. If I read into your suggestion about a second spike in inflation, the way that could cause a recession is if it forces the Fed to hike rates further, which would slow economic activity.
There are only a bit less than three quarters left in this year, and rate hikes affect output with a considerable delay. To get two negative prints this year, rate hikes (or the lack of cuts that might otherwise have happened) would have to be affecting output by Q3 at the latest. That's not far away - Q2 will be almost over by the time the FOMC next meets. So it's doubtful that an increase in inflation in the coming months and therefore rate hikes in the months following could affect the probability of having two negative GDP prints this year basically at all. Maybe next year.
That said, markets aren't pricing rate hikes at all, and Powell even made comments in the last presser saying they're not on the agenda given how things currently look. So the above is just to say that even if a second spike in inflation caused more hikes, it still wouldn't affect Kalshi's market much.
I think the same basic argument applies to any other definition of a recession too - that inflation → hikes → slowing whatever metric is slow and unlikely to affect things this year, but pointing to the specific requirement of two negative GDP prints in one year I think should clarify just how short the needed timeframe for that mechanism is in order for it to cause a recession.
As for your other points:
Yield curve inversion is nothing but a market forecast for lower interest rates over the next ten years than two years, which you expect whether there's a recession or not. A soft landing would look like that too.
Geopolitical risk is a claim you could make nearly every year that doesn't move us much away from base rates. I know it doesn't feel like it, it never does, but it's true.
Inconsistencies in data ⇒ recession doesn't follow, sorry. In fact, jobs data is looking great, the Fed would have to be pretty happy about it.
Low PMI, meh, the Fed wants some slowing, slowing is good unless it's too fast, merely pointing to slowing isn't enough. Some slowing is the point, and if we weren't seeing it you'd be pointing to the lack of slowing as a recession risk too (because it might necessitate more rate hikes).
@chrisjbillington totally fair view. I think we'll just disagree on this one.
Labour market revisions drive a lot of NBER decision making, and rates higher for even a bit longer can really stress banks and mortgage holders.
Jobs data - we've just had the biggest tick up in the sahm rule indicator (we're at roughly 0.4 and 0.5 normally indicates recession)
Guess we'll just have to wait and see.
Ultimately this markets pricing is much close to the true probability than the 8% on kalshi which still seems crazily confident to me
we've just had the biggest tick up in the sahm rule indicator
Biggest since when? Oct last year was a bigger increase, then it declined until January, then increased to now.
I rate it as a good indicator. But the way it works, you can expect a slowdown in its increases soon, or even a decline.
Its baseline (the 12-month-minimum of the 3-month-moving-average of unemployment) will start increasing this month, such that increases in the indicator will not be as large for the same increase in the unemployment rate. The three-month-moving-average of unemployment can get as high as 4.2% as of the September release without the Sahm rule triggering.
Most recessions in the past involved a decently faster increase in the unemployment rate than we're seeing now, they don't usually get to the point where the Sahm indicator's baseline starts increasing before the rule triggers.
Guess we'll just have to wait and see.
Indeed!
@chrisjbillington to be fair a lot of the uptick in unemployment data we see now from historic recessions are a result of revisions much later on. They normally don't look anywhere near that bad at the time and then they get revised worse.
Good discussion about this around halfway through this podcast: https://blockworks.co/podcast/forwardguidance/2e9b9bc8-080b-11ef-b1df-632d6d93c38c
A while back (back when I did my masters) I looked into the revision record and it's crazy how much u employment gets revised up to 12 months after
@KevinBurke A recession basically has to start within the next month for Kalshi to resolve yes, whlie it could start December this year for this market.