LinkUp Forecasting a Net Loss of 25,000 Jobs In December and a Gloomy New Year

Labor demand in the U.S. has dropped 16% since April and the U.S. economy, as a job creating machine, has been bludgeoned with a sledgehammer

As is always the case this time of year, appropriately so to be sure, the media is flooded with recaps of the prior year and forecasts about the year ahead. It’s one of the great aspects of the New Year and we’re certainly no exception in both consuming and generating them.

We’ll hold off on our annual predictions until after we get December’s jobs report later this week, but the following post provides our final NFP forecast for 2025 as well as a broad overview of our take on where the job market has been, where we think it’s headed, and what that’s likely to mean for the broader economy.

We’ll start with a perfect synopsis of the prevailing consensus view found in one of the countless New Year commentaries - in this case a recap of the most recent episode of Bloomberg’s Everybody’s Business podcast. Bloomberg summarizes the hosts’ forecast for 2026 as follows:

The chances of a recession in the new year are significantly higher than normal, according to most forecasting agencies, but the chances of it actually happening are lower than was predicted for 2022 and 2023. If the economy continues to grow, if the job market doesn’t crater, if food prices don’t spiral out of reach, and if the promise of AI can deliver sufficiently for investors, 2026 could shape up to be good year for the US economy.

Factually accurate for certain, but replete with plenty of conditionals and equivocations. But in addition to adding dozens more ‘ifs’ to the list (as I have ad nauseam for the past year) which most everyone underestimates, ignores, misses altogether, or worse, welcomes, I’d note that the fundamental premise is built around the flawed assessments of 2022 and 2023 (and we’d add 2024 as well).

We’ll refrain (somewhat) from yet another exhaustive revisit of how badly everyone was misreading the fundamentals of the U.S. job market coming out of the pandemic and how that played out across the economy and the markets, but suffice it to say that the job market was far stronger than everyone believed it to be as the underlying mechanics of the labor market operated precisely as expected throughout the period.

From our standpoint (and we’re clearly biased), the collective misread of the post-covid economy was due, in large measure, to fundamentally flawed JOLTS data. Our view, on the other hand, was shaped by our accurate job openings data indexed daily directly from company and employer websites globally that list organic (i.e., non-sponsored or paid) job vacancies that are accurate, diligently maintained, complete, real, and timely. Of course, we also applied 20+ years of experience in and around the human capital management, talent acquisition, and recruitment advertising spaces, but the point is that accurate data provided (and continues to provide) an incredible window into what was/is actually happening in the job market, arguably the most significant driver of the broader economy.

In looking at LinkUp’s labor demand data relative to JOLTS data (see chart below), it’s clear that the two generally correlate with one another - at least they did until early 2022 - and they generally get to the same place over a long enough period of time (sort of).

But looking more closely at the two time series, it’s also very clear that in each of the past 4 years (as numbered below), there were material deviations between LinkUp and JOLTS data.

Again, we wrote endless blog posts in real time during these years, but the quick synopsis is as follows (from our vantage point and focusing on LinkUp data):

In 2022 (1), as companies raised wages and attracted people back to work, jobs were filled, new openings steadily decreased, and equilibrium in the job market was achieved as labor supply and demand gradually came into balance.

In 2023 (2), with the massive labor supply shock subsiding and the economy opening up fully, demand for everything (especially services) surged which led to a whole new wave of job openings and sustained hiring at levels higher than most expected.

The same phenomena occurred again in 2024 (3), aided by (or perhaps driven largely by) stimulative policy coming out of Washington and the early stages of AI investment. Again, sustained demand for everything drove companies to keep hiring, posting more and more openings to fill the jobs required to meet that demand, and again and again, the resilient job market surprised almost everyone as the dark recession storm clouds on the horizon proved, again and again, to be a mirage.

But then came 2025 (4).

Hiring kicked off in Q1 2025 very much in line with typical seasonality patterns, but after the clown show in the Rose Garden in April and the acceleration of insanity spewing out of Washington (see dotted lines above), businesses reduced hiring dramatically not only due to the unprecedented levels of chaos and uncertainty, but also as AI truly started to materially impact business investment and hiring decisions. Since April, labor demand in the U.S. has dropped 16%.

So that’s the basic narrative around the post-covid job market and, to a large degree, the economy in general. If someone can walk me through the narrative behind JOLTS data, the perfect inverse of LinkUp’s data, through Q3 2025 (or any of the past 4 years, for that matter), I’d love to hear it. As noted above, JOLTS data eventually gets to our same endpoint, but the meandering path along the way defies logic and as an ‘official’ data source, it’s been delivering a really wonky, highly questionable signal for a long time.

JOLTS wonkiness this year has been particularly stark. Since the beginning of the year, according to ‘official’ data, labor demand in the U.S. has risen 2% and since April, JOLTS data would indicate that labor demand has risen 7%. In the past 7 months, however, the U.S. economy has added a total net gain of 119,000 jobs - an average of just 17,000 jobs per month. Last year, the economy added an average of 195,000 jobs per month - 11x the rate of monthly job creation since April 2025.

The fact is that the U.S. economy, as a job creating machine, has literally been shut off.

This is not a job market that is softening or cooling or decelerating; this is a job market that has smashed into a concrete wall.

The number of job openings at year-end is precisely where it stood 6 years ago in December, 2019. And even if December’s jobs numbers are positive (which we highly doubt), total job gains in 2025 will be under 650,000. In 2010, in the depths of the Jobless Recovery immediately following the financial crisis that completely froze the U.S. economy, the U.S. generated a net gain of 1.1 million jobs - almost double the number of jobs created this past year.

We can (and must) debate endlessly about what mix of factors has shut the U.S. jobs machine off (e.g., tariffs, policy, D.C. chaos & insanity, AI, immigration & deportations, waning democracy, geopolitics, or whatever else anyone wants to throw into the blender), but the machine has clearly been shut off (or bludgeoned with a sledgehammer) and it seems as if almost no one is noticing.

And by the way, because of the administration’s nativist policies, labor supply has shrunk significantly which has kept the job market as a whole significantly more balanced than it otherwise would have been (see dotted line in chart below).

As a result, and leaving aside the accelerating deterioration of the household survey in any event, the unemployment rate is not only seriously muted these days, it’s become, at least for the time being, a woefully inadequate measure of the overall health of the job market. It will spike up at some point, to be certain, but by then it will far too late and irreparable damage will have already been done.

All that matters at the moment is what’s actually happening with labor demand and what that means for job growth, and by our estimate, the year will end on a sour note with a net loss of 25,000 jobs in December.

So that’s where 2025’s story ends which then, of course, begs the question as to what’s in store for 2026. Not surprisingly given all of the above, our view is that this thing does not end well with the only questions being around timing and severity of the downturn. We’ll obviously be writing extensively on those two questions in the months to come, but similar to our view on the general state of the country relative to consensus, we’ll take the under on timing and the over on severity.

As a preview of what’s to come in future posts, we’ll end by echoing Joe Weisenthal’s recent comments on AI disruption, the market bubble, and the likelihood of a recession in the near future.

On the most recent Ezra Klein podcast with guests Tracy Alloway and Joe Wesienthal from Bloomberg’s Odd Lots podcast, Joe notes in response to whether or not the AI phenomena is a bubble or not:

If the AI bet fails, then we’re going to have a recession and a bunch of people are going to lose their jobs and if the AI bet succeeds, then a bunch of people are going to lose their jobs because AI will be able to replace labor. So either way, fail or succeed, it feels like this ends with a bunch of people losing their jobs.

2026 is going to be one for the books.

Happy New Year.