The AI shrinkflation of white-collar jobs: More work, same pay, fewer perks
Benefits are shrinking, flexible spending accounts are vanishing, and wellness programs are disappearing. But none of it shows up in wage data

Getty Images
The conventional way to measure whether workers are doing better or worse is to look at their wages. By that yardstick, white-collar workers appear to be holding steady: Salaries have been flat to slightly up, and the headline unemployment rate, while ticking higher, remains below historical alarm levels.
But wages are only one piece of what a job is worth. And as the AI era accelerates, and its effects on white-collar work proliferate, the rest of the package has been getting smaller. Here's what to know.
The bag of chips is getting lighter
Aaron Terrazas, former chief economist at Glassdoor, described the dynamic to Quartz as "kind of like shrinkflation," borrowing the consumer-pricing term for a product that shrinks while its price stays the same. In employment, the analogy works like this: The salary number on the pay stub doesn't change, but the benefits around it erode, the non-salary compensation thins out, and the job itself expands to absorb duties that once belonged to two or three people.
That framing suggests the standard metrics used to gauge worker compensation are missing something. Wage data, the figure economists and policy makers most commonly cite, captures only one dimension of a multi-dimensional bargain. If benefits shrink, perks vanish, and workloads grow, a worker's real compensation can fall without any corresponding signal in the data.
The evidence that this is happening sits across multiple surveys and federal datasets.
Benefits are thinning at every level
Start with health insurance, the benefit white-collar workers have long treated as foundational. Annual premiums for employer-sponsored family health coverage reached $26,993 in 2025, 6% higher than in 2024, according to KFF's 2025 Employer Health Benefits Survey. That premium growth outpaced workers' wages, which rose 4%, and inflation, at 2.7%.
When premiums rise faster than pay, someone absorbs the gap. Increasingly, it is the employee. A third of mid-sized companies increased employee contributions to health plans in 2025, according to Sequoia's 2025 Benefits Benchmarking Report. Meanwhile, 53% of companies are taking direct action to reduce health care costs, and 30% are sunsetting what they deem low-value programs. According to Sequoia data, the share of companies offering health plans that fully cover employee-only premiums has fallen for three straight years. Most mid-sized and enterprise employers are now implementing strategies that include introducing employee contributions to plans that were previously zero-cost.
The shrinkage extends beyond premiums. Sixty percent of employers offered a medical flexible spending account in 2025, down from 63% in 2024 and 68% in 2021, according to SHRM's 2025 Employee Benefits Survey. Dependent care FSAs followed the same trajectory: 54% of employers offered them in 2025, down from 58% in 2024 and 65% in 2021. Wellness programs with resources dropped to 39% of employers in 2025, from 53% in 2021, according to SHRM.
Each line item looks modest on its own. Together, they describe a pattern: The non-salary infrastructure that made white-collar work distinctly attractive is eroding, benefit by benefit.
Flexibility itself is becoming a cost
For years, remote and hybrid work functioned as a form of compensation. That implicit benefit is now being clawed back. Amazon $AMZN has called hundreds of thousands employees back to the office full-time. JPMorgan $JPM Chase has largely ended remote work. Microsoft $MSFT now requires hybrid workers to be in the office three or more days each week.
Fully remote job postings fell from 15% of new listings in the fourth quarter of 2024 to 11% in the fourth quarter of 2025, according to Robert Half. And 20% of companies with in-office policies said they would pay hybrid employees less than their in-office counterparts.
Return-to-office mandates do not reduce salary. But when a worker who spent three years saving two hours of commuting time and hundreds of dollars a month in transportation and child care costs is called back five days a week, the effective value of the job declines. That decline does not register in any Bureau of Labor Statistics release.
The job itself is doing more
The third bucket Terrazas identified is the job expanding without a corresponding pay increase. This is the hardest to quantify and the most common complaint among white-collar workers navigating AI-era workplaces.
When headcount shrinks but output expectations do not, remaining workers absorb the difference. Duties multiply. Scope widens. Hours lengthen. The salary stays the same.
According to the Bureau of Labor Statistics, total employer compensation costs for private industry workers averaged $46.15 per hour worked in December 2025, with wages and salaries accounting for 70.1% and benefits accounting for 29.9%. That benefits share has been stable, but it masks what's happening inside the benefit categories: Plans that once covered more now cover less. Programs that once existed no longer do. Workers who once had flexibility now commute. The total cost per hour may hold, but the value delivered to the worker does not.
This is the core measurement problem. Wage data treats compensation as if it were a single number. It is not. A job is a bundle: salary, benefits, flexibility, scope, workload, career trajectory. When employers need to cut costs without triggering the backlash that comes from a visible pay reduction, they cut the parts of the bundle that don't show up in headline figures. The result is that the standard debate over white-collar workers misses the point.
The question is not just whether wages are falling. It is whether the job itself is worth less than it was. By the evidence available, the answer, increasingly, is yes.