Since mid-1970s, all national recessions involve California. Or put it another way, the conditional probability of the national unemployment rate spiking given what California looks like right now is quite high.
Our explanation for why soft landing confidence is misplaced: 1) the process governing unemployment (u) is nonlinear, its distribution is multimodal with long right tail. Linear forecasting models on past labor values would tend not to predict sharp upswings. (1/2)
@asokol_econ
Almost every hard landing looks at first like a soft landing
What's standing in the way of a soft landing now:
-The Fed staying too high for too long
-A too-hot economy
-A rise in oil prices
-A financial market rupture
"Planes land. Economies don't."
The ISM employment plummet was so sharp that I emailed the ISM folks to ensure they check if there is an error. This is their reply. So this really begs the question of why nonfarm payroll increased by that much.
Underwhelming holiday retail sales was one of the first piece of data to alert certain Fed officials of a recession back in 2007/2008…the holiday season’s not over yet, but so far not looking good. Mastercard forecasts 3.7% y/y for the season, so far 2.5% y/y for Black Friday.
Jobless claims are very low, whatever seasonal factors one uses. Our latest on
@theterminal
assesses what the low level means. Spoiler: less unemployed people are applying for claims, meaning that the low level doesn't mean it what it used to mean. (1/4)
ISM employment plunge vs NFP blowout — which one will the Fed believe? Here is an employment intentions diffusion index we constructed from regional Fed surveys: it shows a similar steep drop as ISM.
So 42% of the 3.1 million payroll gains in 2023 is due to BLS birth-death model. Here is Fed’s staff explanation back in Jan 2001 in the Greenbook on why the adjustment may overestimate job gains and “may not incorporate shifts in labor market trends in a timely fashion.”
Don’t fight the soft landing narrative? This discussion at the Oct 2007 FOMC meeting (2 months before the start of NBER dated recession) calls into question just how much confidence one can derive from recent soft landing calls. Hint: little. The only true signal then was SLOO.
Our own growth surprises index (which uses 66 indicators) and inflation surprises index (33 indicators) are trying to tell us something. Typically they move in synch, but so far this year they have moved in opposite directions.
Inflation surprises also saw a U-turn in early 23’
A 🧵 to add to the debate on whether this time is different when it comes to recession. In a piece with Bill Dudley today, we looked into reasons why or why not recession rules will hold in this cycle. First, U-1, U-2, U-3, Uflows say recession trigger is either near or here. 1/5
BLS has a very clear exposition of the household vs establishment survey now:
There is a simple chart there that is very striking. No need to say more. It is amazing that markets are pricing in a no landing given this.
What FOMC omits in the minutes says a ton. The 3 things that were absent but which you know they discussed:
1) cuts will happen before 12m change in pce inflation hits 2% (Powell, Daly, Goolsbee all mentioned this)
2) neutral rate may be higher (5 FOMC officials used almost
A weak data point Fed has in their pocket: bank credit contracting for 5-consecutive weeks from early Dec through early January (last seen in 2020, and before that, 2012 before QE3). Hence Waller's point that financial conditions not easing. If credit goes, so goes economy.
While tech layoffs got all the attention, several state/local schools are laying off teachers and admin workers. That’s odd, given how JOLTS and payrolls say education sector still short of workers.
Turns out there are 3 reasons: more states are reporting budget issues, and
Strangely Powell mentioned several times “unexpected” intermeeting events or deterioration in labor market that could prompt earlier rate cuts. But given the low baseline for unemployment this year (4.0%), it is quite easily for it to meet that “unexpected” condition.
We are changing our Fed call to a cut in May on the basis of this jobs report. We have doubts about the January blowout, but the revisions for 23’ shows a hotter labor market (though even further boosted by birth death model). 3 charts:
This is your infrequently-posted reminder that jobless claims is overstating the tightness of labor market. The number of unemployed that is on UI is now less than 1 in 3.
Two reasons: low eligibility and low take up.
On take up, California a good example—max weekly benefit of
Why the unemployment rate can go up even if jobless claims remain low, in BLS own words: “The unemployment data derived from the household survey in no way depend upon the eligibility for or receipt of unemployment insurance benefits.”
If you also believe that immigration drive
We don’t think the CPI revisions are a nothingburger. Disinflation momentum stronger for core services (ie rents). Powell’s pce supercore, and core pce will be lower in end-2023. More room for core goods disinflation. This has to give fed greater confidence.
Turns out that 6 quarters of lags in nominal wage growth historically can predict the directional change of unemployment quite well. Based on that, unemployment rate should be rising much more rapidly according to this simple model. Is job openings doing all the work? Can the u
Highligh from FOMC minutes: “Several participants noted the risk that, if labor demand were to weaken substantially further, the labor market could transition quickly from a gradual easing to a more abrupt downshift in conditions.” They believe in the mechanics behind Sahm Rule.
We estimate that Dec’s core PCE inflation was 0.14% m/m. 1- 3- 6-month annualized measures all below 2%. Solid case for March rate cut based on inflation alone. (Wild card: cpi seasonal factors revisions in Feb).
Just launched our NLP Fedspeak Index on
@TheTerminal
today. Our Fedspeak Index combines Bloomberg's timely Fed reporting+labels trained on my read of Fedspeak. It flagged for us Powell's Dec Dovish pivot. Right now its readings is similar to March 2019-4 months before rate cuts.
Car prices may continue to contribute to disinflation in the rest of the year, but the residual impact of two years of high cars inflation is still working its way to other auto-related services categories. Looks like the order of spillovers go like this: car prices -> car parts
A 🧵 on weather and inflation. While we’re more optimistic about core inflation lately, progress on headline is worrisome. One of the key risk I see: el nino. A little known fact: Arthur Burns indirectly blamed El Niño on contributing to out of control good prices. 1/4
I admire the ernestness of Powell’s answer at the presser on how the Fed won’t be swayed by politics in its decisions.
But let’s face it, there is no such thing as pure data dependency, as so much of policymaking requires a normative view of the economy, which in term is often
Common theme in beige book: Turnover within firms slower, but more job applicants = an unemployed worker will take longer to find a job. That is the typical beginnings of persistent rise in unemployment. Next step: as workers aren’t voluntarily leaving, firms will go to layoffs.
A new FRBNY paper finds that FOMC tends to be more dovish in internal discussions than in public speeches during easing cycles (no such bias for tightening cycles). Tells you just how much to discount Powell’s performance today.
A weedy, potential explanation about the hotter than expect ed core PCE inflation implied by the gdp print today: pce deflator weights uses a Fisher Ideal index formula, vs cpi uses Laspeyres weighting.
Why does that matter? So the PCE inflation uses moving weights, so if the
Car prices may continue to contribute to disinflation in the rest of the year, but the residual impact of two years of high cars inflation is still working its way to other auto-related services categories. Looks like the order of spillovers go like this: car prices -> car parts
Our take on commercial real estate in comparison to the 2023 bank failures - in 1 chart. Need a 40% drop in cre valuations to equate to same amount of total bank asset affected as in 2023. But cre is more spread out risk in time, and across banks, though smaller banks.
Another🧵 on the layoffs coming for state/local education sector. The minicipal fiscal cliff will be at its worst in 24-25, but 23-24 is no good either. Key drivers are:
1. Expiration of federal covid funds (ESSER). This hits 2024-25.
2. Worsening state budgets (hits 23-24)
3.
Why I think dot plot next week will still show 75 (not 50) bps cuts for 2024: I don’t think any of the 10 dots that saw >75bp cut will defect. My guesses of those 10 dots are:
Goolsbee
Cook
Harker
Jefferson
Collins
Williams
Barr
Kugler
Daly
Powell
Umich sentiment survey did a very interesting survey on how people consume economic news. For one thing, it is clear that mainstream media reporters don’t need to worry too much that they are not depicting the economy positively enough.
Another reason why it just isn’t clear the median would shift to 50 bp cuts for 24’: growth surprised on downside (based on 66 indicators) even as inflation surprised up.
A 🧵on what Fed minutes won't tell you but FRB/US model baseline (also published by Fed, but quietly, after Feb 1 meeting) does. 1) the staff model revised down output gap. Now it falls to negative in 3Q 2023 (recession, anyone?)
There are potential rhymes of 70s inflation today: favorable base effect of inflation decline after two years about to exhaust, Fed pivot just when that happens, just when a second oil shock may strike, officials possibly overestimating productivity, Fed keen on achieving a soft
We question the emerging consensus that the nfp neutral pace is 200k+ because it is picking up the surge in immigration. There’s a more straightforward explanation: establishment is more procyclical than household survey. Has been true since 1940s. (H/t to former BLS
We've been suspecting that labor market is weaker than it looks. The decline in JOLTS recently seemed too small to be consistent with the ECI drop in Q2 (lately, price data has seen less drastic revisions than quantity data). A thread on wage Phillips Curve (1/n)
ZipRecruiter withdrew its annual earnings guidance due to “atypical hiring patterns”
“The number of job openings and employers’ willingness to pay for those job openings has been declining significantly from the peaks of prior years," said CEO Ian Siegel
Based on CPI and PPI, we are at 0.42% for headline PCE and 0.34% for core PCE inflation in Feb. Still, we see 12-month change in core touching 2.5% before June.
Powell will testify before the Congress tomorrow. Our NLP algo tags him as the median FOMC participant. But even the middle one is on the hawkish side. Goolsbee is on his own.
Another reason why FOMC not ready to cut: members not yet of broad agreement of that need. Here’s visualizing the dispersion of FOMC views with the help of our new weekly NLP Fed spectrometer. (Interactive version at
@TheTerminal
BECO models —> Fedspeak —> spectrometer)
This 👇. I don’t think housing market has bottomed. As an airbnb host myself, I experienced this and am already seeing houses bought during pandemic being listed in the market again. When the labor market softens more and ppl have to return to office, more chills will be felt.
Why I’m skeptical that the nonfarm payroll neutral pace is now well in the 200ks: that assumes the nonfarm payroll is able to well capture the migrants. Looking at the composition of the 3mil nonfarm payrolls added last year, almost 70% are in sectors that are high skills. (1/3)
Credit ratings are procyclical…history of financials crises in US, Europe, emerging markets are replete with examples of how credit downgrades (in corporate bonds and sovereign alike) act as a financial accelerator that amplify a downturn. Some of that dynamics beginning now.
Credit cards interest now @ 20%, the highest ever in the adulthood of many of gen X and millennials. One reason economy was resilient so far? Pandemic policies artificially surpressed the credit channel. Credit scores are about to drop, delinquencies a straight line up.
My thoughts on future of AI: it will separate the population into two classes — those who control it, and those who will be controlled by it. It will probably increase economic inequality though expanding the economic pie collectively.
One implication of this: when UI claims do rise to a high level, then it is truly painful for people -- more so than the same level of UI in past recession.....because this time around UI claims can no longer pay the bills, and one must have exhausted other options.
Thoughts on Fed minutes: Fed has totally bought in the idea that immigration is already disinflationary, and that the economy is in the midst of a productivity boom. For all the acknowledge of uncertain outlook, there are, in between lines, a lot of confidence about the
Today’s payroll report shows just how hard it is to conduct policy in real time, recalling how Arthur Burns had blamed the deceptive real-time data in missing on inflation. The payrolls data for 2023 that will most approximate the truth will only come out in Feb 2025. (1/2)
It is hard to predict revisions, but we’re going to stick our heads out on this CPI one - with an eye on PCE implications. Two things we predict: 1) more disinflation in end-2023 in OER, and 2) less disinflation in used cars. (With intrepid
@ou_estelle
)
Pretty stunning 2024 outlook from CBO here. 10yr UST to average 4.8% in H2 2024 even with an unemployment rate that would imply a recession. If 10yr UST is still that high in a recession then it would probably make for a deep one. Unless they think R* is much higher.
Am reading the CBO’s “the accuracy of cbo’s budget projections for fiscal year 2023” report right now. On their $1 trillion deficit miss— lower income, corporate, import receipts. Doesn’t read like the rosy picture that GDP gives. More similar to the GDI vibes.
How would the Fed view today’s CPI surprise? First, consider how they view the Jan and Feb surprise:
According to FOMC minutes, “Some” participants didn’t discount the strong inflation readings in January and February, though “a few” said residual seasonality contributed to
(2) Fan chart (based on quantile regressions) for u forecast has risk bands around mean forecast of 3.9% for q423 (matching consensus) highly skewed to the upside. 4.6% and 9.9% u both falls in the 95% confidence interval for 2023 and 24 upside side band.
But the borrowing binge of the superprime really jumps out in the data. Turns out they are the ones buying cars in 2022 and 23…. While suprime and near prime who bought earlier in the pandemic are the ones driving up delinquencies.
Whatever today's dot show, just remember that the new St Louis Fed President Alberto Musalem will assume his position in April. Given his background in international stuff, I would guess that his dot would be a notch lower than the interim president (who likely was one of those
According to this new Fed working paper, "How did households use the stimulus checks? According to the CE survey, up to 60 percent of households mostly used this sizable income supplement to increase spending, while smaller shares reported paying down debt or adding to savings."
New Fed working paper estimates that pandemic fiscal support (the EIP and, interestingly, PPP) boosted demand for cars by about 1.75 million units in 2020.
"..Fiscal programs explain about 70% of the drop in inventory relative to the pre-pandemic period"
Takeaway from Baltimore bridge collapse- one way that Port of Baltimore is small but important is its relative closeness to the Midwest, directly linked through rail. Which is why autos and farm equipment disproportionately goes through that port.
@NickTimiraos
Nick, it is not the weights change that caused the revisions, it is purely the seasonal factors. The new weights won’t be applied retrospectively, and is only relevant starting jan 2023.
On retail - to
@BobEUnlimited
point, wage growth has driven consumption last year (as it does usually), but in 2H 2023, dissaving is playing an increasing role. To the extent that households are optimistic about the future, dissavings can keep supporting spending. But that can
Actually, the Fed not recognizing a recession is key part of the anatomy of a recession in the making..if they recognize it, they could have preempted it by slashing rates sharply, have no recessions! (5/5)
The implied recession probability based on the NLP sentiment for Beige Book (from Cleveland Fed) also spiked just around the BED data shows the negative private payrolls. But now the Beige book probability is back down. Thanks, Powell or Fed's dual mandate?
We estimate that the average wage coverage gap is bigger than ever, abut $1400/week. Why collect benefits when you can earn way more as gig worker or freelancer and actually be able to pay your bills? The lower the coverage of UI, the lower the recipiency rate, at state level.
Bits of puzzle explained. About 30% of cardboard demand came from food manufacturing, timing of cardbox shipments plunge started around when SNAP expired. A loss of $200/month on food is substantial. No wonder beige book also mentioned SNAP. Lower income Americans not doing well.
@FreightAlley
SNAP benefits were cut drastically in March 2023 bc they ended the extra money given out bc of COVID
The people who lost those benefits didn’t magically find other money for food, so they buy less. Result is “demand weakness starting in early April.”
On subprime delinquencies in credit card/auto -- Student loans forbearance inflated/distorted credit scores, so subprime behavior is worse than the usual subprime, and near-prime is more like subprime. Nothing like a bit of miscalculated credit risk to jumpstart credit tightening
A 🙀🧵 about student loans forebearance and credit scores. Fresh Start, started Apr 2022, helped 7.5 million student loan borrowers who had defaulted in payments avoid the negative impact of default. The immediate impact is about +50 jump in their ang credit scores. Avg 550–>600
Two most interesting things to me from FOMC minutes: 1) More dissent than meets the eyes -- "A couple of participants indicated that they favored leaving the target range for the federal funds rate unchanged or that they could have supported such a proposal." (1/2)
“Spending between Thanksgiving and Christmas rose just 3.6 percent over last year, the weakest performance in at least four years, according to MasterCard Advisors.” - (2007, nytimes). In 2023, holiday sales also could be weakest in 5 years…but as they say, just normalizing.
BLS reports and jobless claims are not realtime indicators of labor market stress in the gig economy.
Better to follow Uber/LegalZoom earnings calls, and Obamacare marketplace signups. The BLS reports partially impute job creation from the formation of new entities. But an
Saw a lot on this. Here is BLS’s answer on this in their Q&A. The claim that nonfarm payroll is better at capturing undocumented workers than household survey is just a conjecture at this point.
Many people note a gap between the household survey and establishment survey in the payrolls data.
Interesting work from Brookings suggest the household survey may be incorrect as it is not taking accounting the millions of illegal immigrants.
We highlight two reasons why unemployed aren't apply for claims: 1) eligibility is at all time low pre-recession. and 2) UI benefits not catching up with inflation, opportunity cost of collecting benefits higher than ever before.
This time is different because the increase is due to labor supply? Flaw of this argument is that even in past recessions, entrants into labor force also accounts for at least half if not more of the unemployment inflows. Slowdown in job finding rate is key, not layoffs ( 3/5)
San Diego school district announced major layoffs, as “part of the district’s efforts to close a $94 million gap after COVID relief funds end this year.” It joins many other similar layoffs at the local district level. But what I have not seen before is fractional layoffs.
Real GDI growth accelerated to a robust 4.8% in q4. There are imprints of easier financial conditions in this. Chart here highlights the 1ppt reversal of “net interest payments” from drag to adding. Also a massive increase in “net dividends.”
Over the past year, the monthly household survey of workers has tracked actual job growth more closely than the preliminary business survey. In the coming months, policymakers may want to turn to the household survey as a better real-time barometer of CA's labor market. [4/4]
Why chances are small for a Fed soft landing, and markets are still underpriced in hikes, in charts: 1.) All but five of the 15 tightening cycles we identified since the 1950s ended in recession. (1/4) Read more:
@economics
Powell said March rate cut is unlikely based on today’s data. That frames the tension between our judgemental call (March cut) vs NLP fedspeak call (May cut) well: the former hinged on our forecasts of softer inflation and jobs, while the latter on rhetoric based on current data.
The borrower cohorts from 2021, 2022, and 2023 notably sees higher delinquent rates in unsecured loans than older cohorts.. that is true across credit tiers, including superprime (folks with 780+ credit score), which saw yoy unsecured loans jumped 38.6% in q3 …
This explains the past. Powell looked like a maestro from his pivot. Looking forward—we think the pivot has staunched some of these dynamics. It probably will play out longer. We still think unemployment will go to 4.5% end of this year. And no, the nfp is “strong” last year not
Credit cards interest now @ 20%, the highest ever in the adulthood of many of gen X and millennials. One reason economy was resilient so far? Pandemic policies artificially surpressed the credit channel. Credit scores are about to drop, delinquencies a straight line up.
The employment-cost index is seen inside the Fed as the highest-quality measure of compensation growth.
Q4 data shows the labor market continued to cool.
Wages and salaries for private sector workers excluding incentive paid occupations, slowed in Q4
+0.7% QoQ
+4.3% YoY
Seasonal factors update in household survey tomorrow. A thought excercise: how robust is the current labor market narratives to SA issues? Calendar days and weather matter for appropriate SA. 2017, 2006 will match calendar days, but 1997 el nino signal closer to 23’. (1/2)
Those who said that the tight labor didn’t add to inflation the past few years would be inconsistent to their own views if they now say that the incease in immigration has been disinflationary.
Well, the Business Dynamics revision is up and turns out that Birth/Death models weren't all that inaccurate (/s). As a reminder, they projected 264K jobs created NSA (roughly 100K jobs per month seasonally adjusted).
The actual data was a LOSS of 198K jobs. So the net 500K jobs
Some thoughts on the idea that a new Trump admin will try to “devalue” the dollar. (I put quotes around because that is a wrong usage of the word)
Dollar “devaluation” was a recurrent policy proposal in the first Trump administration, but the idea never took flight because, as