I’ve seen enough. A recession has started. Let’s discuss starting with a very good indicator that has few false positives and no false negatives.
What is the McKelvey Recession Indicator?
Take the current value of the 3-month unemployment rate average, subtract the 12-month low, and if the difference is 0.30 percentage point or more, then a recession has started.
Claudia Sahm, a former Federal Reserve and White House Economist, changed the indicator from 0.3 to 0.5.
Please consider The Sahm Rule: Step by Step written December 7, 2023 by Claudia Sahm.
I created the Sahm rule, and it’s on me to communicate it well. I try. If you have any questions, please add them to the comments.
Sahm claims to have invented the rule. However, credit should go to Edward McKelvey, a senior economist at Goldman Sachs.
False Positives
To eliminate false positives, Sahm modified the original McKelvey rule from 0.3 to 0.5 but the result is a much larger lag time negating her claim of “real time”.
At 0.5 Sahm eliminates all but one false positive (none if you discount the one and only time the indicator was early, and then by six months).
A trigger of 0.3 produces five false positives, albeit only one since 1964. If October of 2023 is false, that makes six false positives.
Using a trigger of 0.4 results in only two false positives with an average lag of ~1 month.
McKelvey June 2023-Present
At 0.3 we had a weak McKelvey trigger in October but the signal quickly faded below 0.3 for the next five months not resurfacing until April.
Using a compromise 0.4 trigger we have as signal for June, although barely. It’s 0.399 to three decimal places.
Using a fatter crayon of 0.35 we had a weak signal in May, and a stronger one in June.
Calculation Methodology
In her step-by step-example, Sahm takes a one-decimal place input and derives a two-decimal place output, a mathematical non-no.
I calculate the unemployment rate directly to three places, then round in my charts.
For example, for June of 2024, Sahm uses one-decimal unemployment rates of (3.9, 4.0, and 4.1), then calculates a two-decimal output of 0.43. Sigh. I rather doubt this is what McKelvey intended, but that’s a guess.
I calculate unemployment rates of 3.865, 3.964, and 4.055 yielding a slightly lower McKelvey number of 0.399 best thought of as 0.40 if not 0.4 (one decimal place rounded).
This entire exercise is a bit of silliness because the unemployment rate is not accurate to the degree that dedication to any specific number depends on.
Rather than a “rule”, McKelvey is best thought of as an early warning indicator. Using 0.3 as a warning signal (or better yet ~0.35 calculated properly, seems about right).
Using 0.50 is so purposely tight to eliminate false positive that it provides no useful warning in most cases.
October Looks Doubtful
I don’t believe a recession started in October.
Nonetheless, I salute Danielle DiMartino Booth for discussing the indicator at a time no one else was remotely interested in the recession idea (and most still aren’t).
It was this Tweet that piqued my interest in the indicator.
Recession Lead Times in Months, McKelvey vs Sahm
Using a trigger of 0.4 instead of 0.3 produces similar results to 0.3 but with only two false positives instead of five or six (depending on how October is treated and also assuming a recession is underway or soon at hand).
Also using 0.4 as the trigger, the sum of the lags is 12 months in 11 recessions, with two leads of one month and two right on time. That makes the average lag ~1 month.
At 0.40 a recession just triggered for June. At 0.35 a recession triggered in May. Importantly, for five straight months the signal has strengthened. This is unlike October of 2023.
Very Weak Economic Data
On July 1, I commented A String of Very Weak Economic Data Sinks the GDPNow Forecast emphasis added.
I am increasingly confident a recession has started or soon will. Yet despite weak data, Treasury yields rose. What’s happening?
GDPNow Chart Notes
- The base forecast is in blue. It’s the one most observers watch but shouldn’t.
- Real final sales, in red, is the bottom line estimate of the economy.
- The difference between the two line is Change In Private Inventories (CIPI) which nets to zero over time.
The NBER, the official arbiter of recessions, uses real final sales as part of its recession decision process, not the base GDP report.
On July 3, I commented Weakness in ISM Services and Manufactured Goods Hits GDPNow Forecast
The GDPNow forecast took another dip today with the key item, Real Final Sales, now at 0.9 percent [and falling fast].
Jobs Much Weaker than Expected, the Unemployment Rate Ticks Up
On July 5, I commented Jobs Much Weaker than Expected, the Unemployment Rate Ticks Up
Counting negative revisions, there was unexpected weakness across the board in June, especially private and manufacturing payrolls.
On the assumption that weakening data will continue to weaken further and there will not be a miracle save by the Fed or White House, this economy is toast.
Mish, Aren’t You Always Early?
That’s a fair question. I even called a recession somewhat recently that did not happen.
So did most economists, but I was much more vocal and insistent about it.
Admit mistakes and move on.
I should have paid attention to Bob Farrell’s Rule #9: When all the experts and forecasts agree, something else is going to happen.
Let’s discuss the “something else” that did happen to kill the expected recession.
Tax Cuts Explain Surge in Consumer Spending in 2023
On January 29, 2024, I commented Tax Cuts, Not Bidenomics Explains Surge in Consumer Spending in 2023
Also, on January 1, 2023, 38 states had noteworthy tax changes. 37 of those changes put extra money in people’s pockets.
The combination of big wage increases, plus tax cuts, plus Inflation Reduction Act spending murdered the then-pending recession.
What about now?
Weakness Everywhere
There is weakness in housing (new home sales, existing home sales, and starts at the lowest in 4 years), consumer spending, manufacturing (both durable and nondurable good), jobs data (constant negative revisions, QCEW, major survey discrepancies, quits, and a rising unemployment rate), and finally we have major unexpected ISM Services in Contraction.
All of the above items are hard data other than the services ISM.
There is no savior on the horizon this time. The Fed rates to be inactive until it panics in September and that will be much too late to stop a recession that started in May or June.
In contrast to what happened in 2023, only two of us (at least that I am aware of) are vocal about a recession being underway or even likely this time.
Thanks Danielle and Bob Farrell with an emphasis on Note #9.
When Did Recession Start?
I suggest May based on a McKelvey 0.4 trigger and an average lag of about a month.
This also corresponds to more pronounced weakness in many major economic reports.
Click on numerous links in the Weakness Everywhere paragraph above for specific examples.
I give credit to people when I use their idea.
Sahm took someone else’s idea then claimed “I invented it”. She is a fraud.
It’s bad enough to post something like this without credit. It’s another several orders of magnitude worse to claim you invented something that you didn’t.
A cynic might say: What would you expect from a former White House economist?
Mish wrote “that will be much too late to stop a recession that started in May or June.”
— Tell me, ¿What does the Fed do to stop a recession?
Austrian economists tell us that recessions occur when too much malinvestment has happened and needs to be cleaned up. The symptoms of a recession really are symptoms of that cleaning up. Wrong businesses go out of business and their employees need to find new jobs. That we have a recession is a positive sign that the economy is engaged in cleaning up the mess.
You have completely ignored much more reliable employment indicators in continuing and initial claims and failed to address the conflicting indicators.
Unemployment Claims, as a set, is a highly reliable group of indicators, and revisions are very minor. They are not suggesting a recession. Here’s what going on.
Realize all economic rules (e.g. Sahm Rule and McKelvey) have underlying assumptions. The Sahm rule effectively assumes the workforce is appropriately balanced first and then suffers a deterioration resulting in the unemployment rate spiraling up. For the 70 years prior to 2021, this was more or less true. And… this unemployment rate deterioration was always supported by CLAIMS.
This time… claims (which is a better leading indicator) is not reinforcing the recession conclusion. So, instead of ignoring it… let’s try to explain it.
Under recent circumstances, the workforce was NOT BALANCED. Mass early retirements and the interruption of legal immigration (from 2020 to 2022) led to an undersupplied labor force. Due to the undersupply of labor (which was temporary) the unemployment rate reflected an extremely tight (3.4%) but very temporary state.
When labor force participation vastly increased (in 2022 & 23, the labor force grew at the fastest two-year period since 1979), some of these new participants went into the “unemployed” pool. Therefore, they increased the count for the Unemployment Rate without causing the claims data to show a significant increase.
As of now, the 4.1% unemployment rate is a better reflection of a balanced workforce, and the climb is not a recession signal!
Remember this post when no recession develops.
Dear Know-It-All Tex
We will all bow down to your superior Know-It-All-Ism at the appropriate time
Thanks Mish. Before reading this article, I have always thought that Claudia came up the whole thing herself. Now I know it is actually Edward Mckelvey who started it and Claudia just raised the threshold then call it her own.
I give credit to people when I use their idea.
She took someone else’s idea.
Then claimed “I invented it”. She is a fraud.
It’s bad enough to post something like this without credit. It’s another several orders of magnitude worse to claim you invented something that you didn’t.
A cynic might say: What would you expect from a former White House economist?
Question; we always talk about rates and real rates, are any of these numbers adjusted by inflation? can they be? does it matter?
German industrial output declines unexpectedly in May
https://www.reuters.com/markets/europe/german-industrial-output-declines-unexpectedly-may-2024-07-05/
July 5 (Reuters) – German industrial production unexpectedly fell in May, adding to signs that manufacturing in Europe’s largest economy will not recover in the coming months.
Industrial production fell in May by 2.5% compared to the previous month, the federal statistics office said on Friday. Analysts polled by Reuters had predicted a 0.2% rise.
When Trump gets in he’ll drop interest rates to zero to get out of recession. Hopefully, that will fix things.
Interesting September, in the equity markets, are coming.
2/1/2024. The recession has been with us for a while. Inflation has wreaked havoc on most purchases for the majority of people. Biden’s dementia gaslighting have made it worse.
There is weakness in housing (new home sales, existing home sales, and starts at the lowest in 4 years), consumer spending, manufacturing (both durable and nondurable good), jobs data (constant negative revisions, QCEW, major survey discrepancies, quits, and a rising unemployment rate), and finally we have major unexpected ISM Services in Contraction.
There is also a massive head fake in the jobs reports that are not acknowledging that a significant number of jobs ‘created’ were not in fact created rather they were a result of workers being severely injured or killed off by the Covid Vaccines. Disabilities have risen dramatically as have excess deaths since the vaccines were rolled out.
The job market is far far worse than any of the analysts (who are jabbed and who jabbed their children) can imagine – or acknowledge. Nobody wants to admit they have poisoned themselves
My wife told me the recession had started a couple months ago. She’s not an economist, doesn’t read blogs like this and hates it when I talk about it. She goes to the store and sees what’s there.
Sometimes common sense trumps the economic models.
If the recession is severe, which I think it will be (is), then it started a couple years ago, which I also believe.
If the recession is weak or mild, then it started much more recently, and will either be over quick, or break into a severe one, if the damage that has been done, is unable to be papered over, which I believe will be the case in that scenario.
Severe Recession on its way (hidden and here already?)!!!
“$1.8 Trillion in annual Federal deficit spending, and all I got was this lousy recession.”
You mistyped $2,377,787,965,394.50
That’s the change in Debt to the Penny (Total Public Debt) from 7/6/2023 to 7/5/2024 (most recent year’s data).
P.S. Even the deficit is having inflation problems!
I receive automated emails for sales from online retailers and a couple of local restaurants. I am receiving more & more of these lately, which I can only assume that things are slowing down. The emails hope to drive sales.
I am receiving endless emails from the sales people at a couple of digital services providers that I have interacted with in the past but did not engage… They appear to be desperately trying to generate new business using old leads
There is a definite weakness in Florida real estate now. We sold my parents’ place in Palm Beach Country at a 30% discount to what we could have sold it a year and a half ago. The ones who bought it are individuals and not a company which is interesting. Also they had no problem getting a bank loan and could have paid more but didn’t see the need with the market being weak. It’s an anecdote but does show the real state of the market in a previously red-hot emplacement. Strangely enough; this weakness in real estate doesn’t seem to be cooling the Florida economy by very much.
Ouch, 30% is a huge discount for what you think you could have gotten a year ago.
Not sure where your parents place was in Palm Beach County (I live here), but from what I can see when I look at Zillow etc, places are still listing for very high prices and it seems people are getting that. A house we looked at in early 2020 that sold for 800K around the time we bought ours sold again in late 2021 for 1.2 million that astounded us because they did nothing to the home (it was immaculate to begin with so didn’t need anything). That same home went back on the market 2 months ago for 1.46 and since we have a realtor friend we arranged for a viewing (wanted to see what it looked like 4 years later). Nothing had changed (ie no updates) and if anything it was 4 years more run down (because nothing had been updated) and we thought no way are they getting that. Yet a week ago it went contingent for 1.42 (40K off asking on 1.5 million home). So clearly lots of money sloshing around.
https://www.realtor.com/realestateandhomes-detail/8623-Thousand-Pines-Ct_West-Palm-Beach_FL_33411_M57958-13138
Was it not Gundlach at Doubleline that said it is the rise in rates after inversion that corresponds to the start of recessions? My “pub” indicator trigger went off last week as I met a recently laid off person at a bar (the infrequency of me at bars and talking to strangers must mean they are numerous).
Whatever metrics you use, now is a good time to at least be recession-ready.
The graph suggests a threshold value of 0.36 would eliminate all the meaningful false positives. Limiting one’s choices to 0.3, 0.4 or 0.5 is insufficiently granular.
Are the UE rate data revised after-the fact? (I’m thinking that since the data come from the Household Survey and a slowly-varying population estimate, probably not much, whereas the Establishment Survey do get revised a lot…) Does the graph show only the real-time data or the final revised data?
Not at all clear how the upcoming recession will play out. Further fiscal stimulus appears to be constrained by rising interest costs / inflation / devaluation concerns, but Congress might do it anyway.
BTW, there’s a documented tendency of the Establishment Survey data (and associated birth/death model) to overstate employment growth at times when the economy is weakening. This results in stronger negative data revisions as an economy slips into recession. Such revisions are very rare during expansions (link below). So the observation of those negative revisions can itself be a signal of a recession having recently started.
https://research.stlouisfed.org/publications/employment-research/are-unemployment-data-revisions-biased
All that federal spending in an election year was supposed to forestall the recession. It’s typical political idiocy: eating your seed corn and thinking you’ve discovered a new path to prosperity.
“There is weakness in…consumer spending…”
I have my own indicator of economic weakness. Assuming there is a correlation between vacation travel and economic spending, I can say that vacation travel is way down — in spite of what the media is reporting.
I live in an area that is a tourist destination, summer (lakes, forests, rivers, etc.) and winter (skiing, snowmobiling, etc.) and there are a lot of businesses here that depend on tourism. It is markedly down this summer. Everyone is waiting for the big surge to arrive.
Still waiting.
What part of MI are you in? I am from the metro Det area. Love to go up to Harbor Springs in the winter for skiing.
I’m in the northern Rockies.
Supposedly the Great Smokies are dead in prime tourist season also. Judging from the prices I’m seeing that appears to be true. I first noticed slowing travel around Labor Day 2023 when lake country was much slower than usual, whereas spring through summer 2023 tourism was red hot. I agree, it’s not being reported on, but it is a trend nonetheless.
The bloaters greatly prefer the steamrolling inflationary depression that is gutting the country over any kind of recession that could affect their dubious assets.
We have not heard much about the Inverted Yield indicator lately.
Inflation is receding but the economy has not yet dipped into a recession. Next month maybe.
There is still so much Federal money going out, much to the illegal immigrants and the people hired to care for them, much also for the enormous debt service, some for energy transition projects that are not actually being built, or for EV charging stations that are not being built, and some to start up new weapons and ordnance to replace what has been given to the Ukraine. Another trillion dollars over receipts every 100 days. More oil and gas are being used to support these efforts. That is why this time seems different because the government does not seem to mind higher interest rates, they still spend like there is no tomorrow.
Also the creditors are receiving now a trillion dollars a year, that is new……
Plus a new labor force to keep wages down.
Wall street says everything is okay. Fed about to start cutting rates adding more fuel to already burning fire. What could possibly go wrong?
The Bank of Canada cut its rates by 25 basis points earlier this month, on the basis that inflation was falling. Here’s what happened:
Core CPI – goods and services less food and energy – spiked by 4.9% month-to-month annualized in May from April, the hottest since September 2022 (blue), according to Statistic Canada today.
https://fasteddynz.substack.com/p/central-banks-are-in-a-huge-bind
I wonder if the people who could not see a weakening Biden also cannot see a weakening economy. Funny how a weak economy and a weak president go hand in hand.
The end of the economic cycle is always confusing because the economy weakens while the markets rise rapidly into a top. I expect both to continue over the next 3-4 months. The smart money needs a blow off top to suck in retail and sell to the bagholders. They will push the goldilocks narrative and “things will get better when the Fed cuts”. Dollar will fall with rates, which will drive up risk assets. Recession probably officially starts Q4 2024 or even Q1 2025. Current administration will continue to fudge the numbers to make sure no recession starts before the election. If recessions are proportional to the bubble that proceeded them, this one will be worse than 2008. The popping of the greatest asset bubble in history and the end of a 40+ year credit cycle is going to be more brutal than anyone realizes.
Just a mathematical point, significant figures matter more than where the decimal points are located.
As an engineer, that bugged me, too. Significant figure error propagation needs to be reviewed. Mish would do himself a favor to delete his criticism of Sahm surrounding this point.
Nice analysis Mish. It’s “one”’of the reasons I keep reading.
Now; how deep?; how long?; and how have you positioned your investments? Not easy questions; I understand that.
Assuming it’s a recession at all, my bet is it’s shallow, no more than a year in length, and since I am sitting at just under 20% cash, I may selectively sell small portions of some big winners to move back to 25% cash. If things start to look worse than that I will sell more and raise even more cash.
N-gDp is falling. Stocks up as the FED cuts, then stocks fall.
So what is your present position and how do you expect to change it? When do you expect the first rate cut?
No triggers yet. Waiting for FED mistake. You don’t have to make many moves, just the right one. It all depends on how fast the FED acts.
First rate cut in September. CPI will be important.
I’ll probably buy TLT, just because I’m lazy.
Thank you for your replies.!
I love the smell of hyper inflation in the morning … smells like….
Gentlemen … get your wheelbarrows ready!!!
“I even called a recession somewhat recently that did not happen”.
Actually, you don’t yet know that you were not correct. The NBER is generally a year late in admitting to a recession.
Given the blatant manipulation of every other number put out by the Obama, I mean Bidan I guess, administration over the past 3.5 years, one would have to live in a fantasy world to believe that the NBER is t also compromised.
You might eventually be proven correct on your earlier call.
Agree with the call, Mish. My industry depends on advertising and you can see local businesses “pulling in horns” for several months prior to the “official” announcement.
David Rosenberg is there with you
And the question is going to be how DEEP is the recession going to be
Which I suppose depends on what the government does, though everything the government does is actually negative in the long run (OK, well some of it probably has a positive multiplier but not much)
normally … with recessions… the central banks are able to stop them by lowering interest rates.. and easy $$$,,,,
Not this time… this time .. the medicine… will poison the patient.
The Bank of Canada cut its rates by 25 basis points earlier this month, on the basis that inflation was falling. Here’s what happened:
Core CPI – goods and services less food and energy – spiked by 4.9% month-to-month annualized in May from April, the hottest since September 2022 (blue), according to Statistic Canada today.
https://fasteddynz.substack.com/p/central-banks-are-in-a-huge-bind
Fed manages to keep pulling rabbits from their credit hat; frankly, we’re puzzled a full blown recession hasn’t happened yet give how Biden has been like to proverbial drunken sailor with our fiat money.