Claudia Sahm claims to have invented a recession indicator created by Ed McKelvey. Now she says the indicator is wrong. Let’s investigate. 
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.
Edward McKelvey, a senior economist at Goldman Sachs, created the indicator.
Claudia Sahm, a former Federal Reserve and White House Economist, modified the indicator from 0.3 to 0.5.
The rule triggered in August but Sahm is in denial.
Sahm Denial
Marketplace discusses Sahm’s Recession Denial.
When the monthly jobs report from the Labor Department was released in August at 8:30 a.m., it packed a punch. Something called the Sahm Rule had been triggered — it was like an economic fire alarm was going off.
“I was live on the radio, and they read the numbers out loud. I said, ‘OK, so the Sahm Rule says we would be in a recession, but Sahm says we’re not,’” said Claudia Sahm.
Sahm discovered the rule when she was studying previous recessions as part of a project to help policymakers prepare for the next one.
So why is Sahm, the economist, discounting Sahm, the rule, now?
The unemployment rate, Sahm explained, has an Achilles’ heel: It doesn’t only go up when people lose their jobs, it can also go up when the number of people looking for jobs goes up.
“When you have people enter the workforce, it can take longer to find a job, even in the best of times,” she said. “That will push up the unemployment rate.”
Sahm certainly did not discover the rule. She modified Ed McKelvey’s rule with no credit given to McKelvey.
And it might behoove Sahm to actually investigate her explanation.
In 7 of 10 recessions, the labor force was higher in the third month of recession than the start of it.
In isolation, that would tend to raise the unemployment rate as Sahm says. But it is also normal behavior.
The Covid recession only lasted 2 months and was so unusual in many other ways that it’s best to remove it for comparison purposes. But if you insist, then call the score 7 of 11.
Recessions Tend to Start Slowly
In two recessions, 1970 and 1973, employment was higher in the third month of recession than the first, by 83,000 and 353,000 respectively.
Nonfarm payrolls were up by 275,000 and 223,000 respectively.
And that is after revisions!
So, don’t claim the labor market is too strong for a recession to have started.
The McKelvey Recession Indicator Triggered, But What Are the Odds?
On September 10, I commented The McKelvey Recession Indicator Triggered, But What Are the Odds?
Many eyes are on the McKelvey recession indicator. Too many?
That would probably be the case if everyone believed it. [But heck, not even Sahm believes it!]
I calculate the odds based on past history of recessions at well over 50 percent. Click on the link for how I arrived at the percentage.
Recession Supporting Factors
September 3: Construction Spending Growth Slows in May, Stops in June, Negative in July
September 6: Payroll Report: Manufacturing Sheds 24,000 Jobs, Government Adds 24,000, Big Negative Revisions
September 7: BLS Negative Job Revisions 15 of Last 21 Months
September 9: Fed Beige Book Conditions Are Worse Now Than the Start of the Great Recession
“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.”
https://stayathomemacro.substack.com/p/the-sahm-rule-step-by-step
It’s not even clear she implemented the rule properly. There should not be negative numbers. But I don’t know if McKelvey had them or not.
Adam Taggart interviewed her and she blames the media for giving her all the credit.
All she did was change the trigger from 0.3 to 0.5 and I believe using a wrong lookback period that creates bizarre negative numbers.
She popularized the rule, I will give her credit for that. But stole is a better word IMO.
Well stolen is half done.
It is impossible to have a negative number.
Pull up Sahm’s data – negatives all over the place.
I fixed the issue.
https://fred.stlouisfed.org/series/SAHMREALTIME
I do not believe she translated McKelvey Accurately
So, don’t claim the labor market is too strong for a recession to have started.
—
It was never a strong labor market going as far back as 2022 when layoffs started in earnest post covid. What we are seeing now is the cutting of jobs due to AI replacing jobs imo. It will continue for many years ahead. The biggest challenge governments will have is AI replacing human workers in many sectors. Profits can go up in this environment and the stock market can go up. It is possible for some companies to have increasing profits while the economy has rising unemployment and rising GDP. We are entering a new era imo.
My eyes and ears tell me there is no recession when I go out. Traffic. Check. Restaurants packed on weekends. Check. People eating out for lunch during the workday. Check. I can believe some poorer parts of the country might be in recession. But they probably never recovered from the last two recessions so they don’t really count. I will start believing there is a recession when I see low foot traffic in malls, restaurants and nearly everywhere else. People are paying higher ticket prices for these NFL games and not blinking an eye.
I don’t know when McKelvey did his work, but a guy named Elliot Middleton did some pretty good analysis back in the ’90s.
Hi Mike, is the current monthly PMES-McKelvey recession indicator data published anywhere that we can access? That was a fantastic article.
You can follow Sahm here – albeit with negatives
https://fred.stlouisfed.org/series/SAHMREALTIME
I do not believe she translated McKelvey Accurately
You can also follow PMES – I give links here
https://mishtalk.com/economics/improving-the-mckelvey-recession-indicator-no-false-negative-or-positive-signals/
I also explain the negatives
Nobody in Government, working for the Government, taking any money or gifts of any kind from Government, is subsidized by the Government, has Family that works for the Government, relies on the Government for there Businesses etc. will Never call a recession, as it’s a detriment for the gravy train there riding.
It would be like a homeless person complaining the free bed they had to sleep in for the night, was too hard, and the breakfast that was served in the morning was too cold! Not happening if they want the goodness to continue…
This rule misses the underlaying condition of every recession. That is recessions require a fragile financial system as a precursor. Consumer credit card and loan delinquencies are that precondition. Current levels for credit cards is 3.2% when 4.7% is the level of prior recession conditions.
Recessions are triggered when an event causes a temporary pullback in credit extension that proves to be the 1st domino in the line.
The rule is freaking 11 of 11 with no false positives or negatives!
It is missing ….. NOTHING
Hard to apply economic models when there are major differences between when they worked and now.
The biggest factor is the demographics of baby boomers exiting the work force over last 10 years and now have a major impact on the labor force.
Go back even 7 years ago and it was hard to get a public school teacher job, now it is not.
Same goes for a wide range of vocations ranging from nursing to HVAC repair; as they all are in dire need of new workers.
So how do you apply this economic model when there was not this dire need for new workers say in 1981, 1993, 2002, and 2009 ? And there are a lot of unaccounted for gig and work-for-cash workers compared to these previous years.
Another matter is the Federal Government and Federal Reserve could essentially stop a recession (and deflationary spiral) by just Zero Interest Rate Policy and Quantitative Easing (ZIRP + QE).
these “invisible workers” (gig workers and 1099s) are not “actively looking for work” which would be accounted for in the unemployment numbers
and I think we are more in an “invisible workers” economy or more of a two-part-time-job-and-Uber-driver economy
part time workers likely do not collect unemployment and they just get another part time job if they get laid off or work more hours for Uber and Door Dash
I see that a lot in Panama City Beach, Mister Mish
A lot of the “service workers” here fit in this category as they’ll get let go at Holiday Inn Resort, Broken Egg restaurant or Hammerhead Freds, and then get hired at Walmart to collect shopping carts
Its a bad rule when you have unemployment well below 5.5% for so long. I think being below 5.5% is indicative of a healthy labor market.
So with the unemployment rate still well below 5.5% and vocations (from nursing to HVAC to public school teaching) begging for workers due to demographic changes (boomers exiting workforce, etc) then if the 3 month average for unemployment is 4% and the 12 month low is 3%, then so what the difference (3 month avg minus 12 month low) is 1% and a lot greater than the 0.5% threshold as far as SIR (Sahm’s irreleant rule).
ZIRP and QE did not end the Great Recession. What ended that was suspension of key accounting rules, allowing banks that were broke to pretend to be solvent and resume lending.
Also, back in the 1950s and 60s, and even as recently as 2001, there were plenty of recessions with initial unemployment rates below 5.5%. Roughly half of all postwar recessions began with unemployment at or below current levels!
Typical BS Economics
I prefer astrology as a reliable information source.
The Sticky CPI, ex food and energy, is 4.1%. It force people to look for a job. When more people enter the job market it takes longer to find a job, even in the best times. When that happened ignore the Sahm recession signal. If she is correct, McK is junk.
Need to examine the number of job annoucements or openings compared to number of unemployed (or number of job seekers)
Granted some of the job announcements are ruses and fake
But what I posted before on this thread, I think compared to pass recession periods like 1993, 2002, and 2009, I think its a lot different economic conditions such as its more of a two-part-time-plus-Uber-driver economy.
Let alone I never seen this much demand for various vocations from public school teacher to nursing to HVAC maintenance to car repair, because of a major demographics change (i.e., baby boomers continue to leave workforce).
With the 15 trillion or so US government allowed and promoted corporate stock buy-backs in the last 25 years, the current 56.25 trillion capped US stock market will grow to its maximum time-base limit. If buy-backs (rather than corporate tax advantages for R and D and global competitive product development)are continued to be allowed, maybe the time-based upper valuation limit will be used as a predictor of US market turns in the future.
Yeah, look at GE buying backing $10 billion of shares in 1990 and at that time, it was the largest buyback in history.
Through financial engineering gimmicks, Jack Welch created a paper tiger not a technology and innovative leader.
And then that moron Immhelt took over after self-aggrandizing Welch to drive GE nearly in the bankruptcy grave.
As long as mega caps like Google and Amazon innovate and promote productivity tools or resources to the masses, then I’m okay with them buying back shares to benefit their insiders.
Roll Tide.
Every quarter in 2023-2024 has been positive, each except one has been over 2%. How many consecutive quarters of growth do we need to disprove an active recession exists, 10, 20, 150 quarters?
Have you ever heard of revisions?
Are you aware of the discrepancy between GDP and GDI?
The answer would appear to be something like 4.
We will find out perhaps 14 months from now as you should damn well know. And it might behoove you to look at the stock market peak in November of 2007 along with rising employment right before the great recession.
If you have something to add, OK. But all you are offering is bullshit, then I would advise you to stop.
Fair enough, Mish. But many of us put more faith (even if not as coincident as a McKelvey/Sahm rule) in the NBER using GDP numbers for the majority of a recession call. So I think it’s fair to ask about those figures. Do you believe the past GDP numbers will be revised so much lower in the future, that those 2-3+% growth rates will disappear?
If you waited until Q4 of 2008 to notice via GDP the recession that had started in 2007, you lost a lot of money:
– in the stock market
– in real estate markets
– in the employment market unless you kept your job and pay wasn’t market-based
That same recession was anticipated by the market peaking in Oct 2007, and even the laggy Fed had made multiple interest rate reductions…. before NBER finally made a post-election statement in December 2008.
Fair enough, NAI. If there are fools out there investing their money based primarily upon quarterly GDP reports (or late NBER calls), they will probably lose money within the market on a regular basis LOL
But Mish has previously called for a start of this recession in May of this year, or maybe even the first quarter. Then and now, GDP has been growing above a historical rate in real terms (and the unemployment rate has been below historical averages). So of course, the recession may be called for then, or later in August. But I’m still asking whether he or others really believe negative revisions to 2024 GDP and the unemployment rate will be so high as to rise to the ‘recession level’ normally associated with NBER calls (not for some future recession which of course will come eventually happen, but for right now)
I’d say it’s too soon to tell by GDP means. GDP is not measured in real-time with sufficient accuracy. When people say “GDP has been growing…” they are potentially confusing estimates with reality. The preliminary, initial (etc.) GDP reports are heavily estimated and/or based on approximate, incomplete data. Only when the final data comes in about a year later is it reasonably reliable.
Also, in the current case, GDI, which should match GDP, has not been growing to match GDP. That makes the GDP numbers even more suspect than usual.
Regarding whether employment metrics will “rise to the recession level”, the answer is that they already have, for a variety of employment metrics based on the population survey. These include annual % change in full time employment, annual % change in part-time employment, annual % chance in total number of employed. These have been previously discussed in the comments here. (One cannot use the Establishment Survey data for this, it lags in much the same way as the GDP data, in this case because of the estimates used in the business Birth/Death model – estimates which are blissfully unaware of the onset of recessions.)
The NBER called one recession after it was over.
In this case you might wait 14 months for them to decide when it started
Would not surprise me a bit
From Jan 1 to Oct 9 2007 the market overall was up 16% (the peak for Dow and S&P, NAS was the 30th Oct). The recession was identified in December 2007 ( it was announced in Dec 2008)!
The market bottomed on March 9, 2009, down 56% from the peak. It would’ve (should’ve) been much lower but for the Fed bail out. I remember seeing my buy orders not executing as the market headed back up.:)
The recession end was June 2009 (I don’t know when it was announced). It looks like a 3-month +/- lag against the market overall, but don’t wait for the announcement. Those betting on the market might need to consider that.
Democrats and establishment types like this Sahm are not going to suggest the economy is faring poorly or is in a recession right before this election.
If Harris wins, the establishment economists will declare a recession within 9 months of the election.
Then the mainstream media will spoon feed the public that the economy has fully recovered 3 months before early voting starts for the November 2026 election cycle.
Some of the “dismal science” discussions might just go beyond my grasp but there’s one thing that doesn’t.
Taking credit for someone else’s work is an immediate disqualifier if you expect me to take your work seriously. At the least indicate that you’ve modified someone else’s calculations based on some research you’ve done or some other indicator that you think is more important to the work.
I’ve about had enough of the flat out lying that goes on these days.
I’ve gotten used to the Lying. What would blow my mind is Honesty.
Plagiarism seems to be a liberal thing. (kidding)