Job-Based Health Insurance: A Quarter-Century of Wage Theft
What Would You Do With a Spare Two Hundred Grand?
If you found $200,000 - or even just $125,000 - in the back of your closet, what would you do with it? Ok, not “find the rightful owner” — it’s yours. Save for retirement? Buy or fix up a house? Pay off student loans or credit cards? Bet it on the Super Bowl? That’s how much money our health care system has stolen from San Francisco hotel and food service workers over the last 30 years.
A recent analysis in JAMA Open Network found that if premiums for employer-sponsored family health insurance had increased at the same rate as wages from 1988 to 2019 and employers had paid workers with the savings, the average worker with family coverage would have pocketed an extra $125,340 “or nearly 5% of total earnings over the 32-year period.”
$125,000 is a lot of money. It’s also an underestimate. The study’s authors only examined the growth of health insurance premiums. Yet today’s job-based insurance requires far more out-of-pocket spending - deductibles, copays, “coinsurance” - than in 1988. The authors failed to account for this additional cost increase (or decline in value of insurance), which may have cost workers as much as premiums themselves.
Data provided to Healing and Stealing by UNITE HERE! Local 2 offer a better example of how health care costs rob workers of wages. If San Francisco hospitality workers’ health care costs had gone up at the rate of general inflation between 1996 and 2019 - 8 years shorter than the time-frame of the JAMA Open Network study - each worker would have earned $206,877 in additional cash wages over the years. Current wages for housekeepers - the largest single group of workers in a hotel - would have been $9.68 per hour higher in 2019, $36.12 instead of $26.44.
Unionized San Francisco hospitality workers have fought ferociously for decades to win and preserve comprehensive coverage for themselves and their families - rallying, striking, taking arrests. Despite minor copay adjustments, their benefits haven’t eroded and cover virtually all the costs of illness and injury. Thus the growth in the cost of their health care includes both insurance premiums and the deductibles and coinsurance most of us have come to accept as normal.
The lost wages are measured in what economists call real terms - inflation adjusted 2019 dollars. The nominal dollar figures dating back 23 years are staggering. In 1996, San Francisco employers contributed $267 per month per eligible employee for health benefits, or $1.73 per hour. In 2019, the monthly health benefit contribution was $1,932.15, or $12.55 per hour, seven times more than 1996. At that point, employers were paying more just for health benefits than the state’s minimum wage.
For background, these are monthly contributions by employers for each eligible worker, made to a separate fund jointly governed by the union and employers who participate. The fund negotiates with several insurers from whose plans workers choose coverage. They are a “blended rate” including both single and family coverage (see technical details at “A Little Extra for the Data Curious” below).
Glenn Mercado understands the importance of health benefits and what it takes to keep them. Mercado has worked in the banquet operations department at the San Francisco Marriott Marquis for 18 years, setting up and putting away furniture, equipment and supplies for events. A month shy of 40 and single, he’s pretty healthy, but supporting his father through struggles with poor health made him acutely aware of the need for health care.
In a video interview Mercado told Healing and Stealing he had to learn to drive about ten years ago when his father had a stroke that landed him in the hospital. The stroke was the first of a series of conditions requiring frequent treatment and Mercado needed to make sure his father got to his appointments.
The stroke and long recovery were “really dire for me at the time. It was hard for me to see him ever functioning as a normal human being again. I really felt he was going to be bedridden for the rest of his life.” His father eventually recovered from the stroke, but follow up care, heart surgery and other health care needs keep Mercado behind the wheel, getting his father to his appointments.
One thing makes all his father’s conditions worse - money.
“My dad is better now, but he still makes frequent visits to the hospital, which he absolutely hates. The main reason he hates it is the cost. When he thinks about costs and insurance, it gives him a lot of anxiety. It’s really intense, it’s actually debilitating for him. That’s the thing that really helped me appreciate our health care.”
His health care coverage comes at extreme cost in forfeited wages.
For many people, the bite that insurance takes out of their paycheck isn’t always visible. If a small premium increase is paired with a much higher deductible, the full impact of the change may not be felt until the first bills arrive from the doctor’s office.
By contrast, San Francisco hospitality workers see the money go directly from their pocket into the coffers of hospitals, doctors and insurers every year. The union typically negotiates a set amount of money for wage and benefit increases. So each year, local leaders have to decide how much goes to wages, how much to health care and how much to pensions.
It’s an agonizing choice, and workers can see health care consume raises needed to feed their families, educate their children and pay their mortgages or the constantly rising rents in the Bay Area. From 1995 to 2019, just 56% of the annual increase in San Francisco’s unionized hospitality workers’ compensation went to wages, 36% to health care and another 8% to pensions.
The numbers fluctuate significantly. In 2014, 59% of the raise went to health care and just 27% to wages. Two years later, two-thirds of workers’ raises came in the form of a dollar-an-hour wage increase and just 23% to cover the growth in health care costs.
The data demonstrate how much stronger the real impact of health care cost increases have been than just the premiums shown in the JAMA Open Network study. According to the article, in 1988, health care accounted for 7.9% of the average compensation for households with family coverage. By 2019, it had grown to 17.7%.
In 1996, UNITE HERE! Local 2 workers saw 13.3% of their compensation going to health care. By 2019 the share had risen to 32.2%, nearly twice the national rate found by the the researchers who only looked at premiums.
There are limitations to comparing a single city to national data, but San Francisco hospitality workers’ much higher healthcare costs give some indication of how severely health coverage has eroded even as premiums have skyrocketed. According to data from the Kaiser Family Foundation’s annual employer surveys, the average deductible for single coverage grew by more than a thousand percent from 1996 to 2019 ($133 to $1396), and deductibles are only one form of out of pocket cost-sharing. Most employers and workers are buying much less coverage for much more money than they were three decades ago.
At the risk of making a shameless plug for my former employer it’s worth noting the scale of the economic gains San Francisco hospitality workers have won in recent decades.
Despite the sacrifices required to maintain health coverage that actually allows workers and their families to get treatment, from 1996 to 2019 real wages for San Francisco’s unionized hotel housekeepers increased three times faster than the earnings of the national median worker. Local 2 housekeepers’ wages, adjusted for the growth in the cost of living, grew by 44%. Real earnings for the U.S. median worker grew by just 15%.
Asked what he would do with an extra $200,000 if workers’ health care costs had increased at the same rate as the cost of groceries and transportation, Glenn Mercado smiles, hesitates, but then turns serious.
Hotel banquet business fluctuates, and in the past year Mercado has struggled to get enough hours at work. “This past year is the worst year I’ve experienced at Marriott. The first thing would be use it to invest in retirement or something that would help me accumulate more income. I know this is already a reality that other people need two jobs, but it’s getting to be a reality for me. I need another stream of income.”
California recently created a new agency, the Office of Health Care Affordability, charged with setting a cap on on how fast health care costs can grow. Mercado told the commission that oversees the agency that decades of extreme health care cost growth is also stealing something else, “a healthy social life. I don’t mean a big vacation. Hanging out with friends at a restaurant, going to a community event, festival, or pretty basic things, I can’t really afford to go to them anymore. The reality is that it requires money. The money is for bills or groceries.”
U.S. employer-sponsored health insurance has been transferring money from workers to investors, consultants, both for-profit and “charitable” executives and an avalanche of meaningless administrative work for decades.
It doesn’t have to be that way. What would you do with a spare 200 grand?
A LITTLE EXTRA FOR THE DATA CURIOUS
Ok, a lot extra. Two questions:
1. What are the sources for Healing and Stealing’s data and how were they analyzed?
2. What are the differences between the San Francisco hospitality worker data and the January 16, 2024 article in JAMA Open Network,“Employer-Sponsored Health Insurance Premium Cost Growth and Its Association with Earnings Inequality Among US Families,” by Kurt Hager, PhD, MS, Ezekiel Emanuel, MD PhD, and Dariush Mozzaffar
Start with sources and analysis:
SF Hospitality Workers’ Healthcare Cost and Wage Data: Unionized San Francisco hotel and food service workers get healthcare benefits through a fund that is jointly controlled by management and the union, known as a multi-employer welfare fund, or a “Taft-Hartley” Fund, named for the 1947 law that created the structure. The fund receives a monthly contribution for each worker eligible for healthcare under the terms of the union’s contract with participating employers. The fund, in turn, negotiates premiums with several health plans that members may choose for coverage. Healing and Stealing obtained the amounts of contractually agreed wages, health fund contributions and pension contributions from UNITE HERE! Loal 2 staff dating to 1996.
Wage Loss: To calculate the lost wages, Healing and Stealing began by obtaining the average annual value and annual percentage change in the Consumer Price Index – All Urban Areas (“CPI-U”). dating to 1996 from the Bureau of Labor Statistics website.
The first step is to create an annual “health care spending at the rate of inflation” to compare with the actual healthcare contributions. 1996 is the base year, with both actual and “at inflation” the same, $267. To get an “at inflation” figure for 1997, the 1996 health care contribution rate was multiplied by (1+ 1996 annual CPI-U %change) to establish the first year’s “health care growing at the rate of inflation” value. For 1998, the 1997 “at inflation” number was multiplied by (1+1997 CPI-U %change), and then that process was repeated to yield the expected monthly contributions for each year if spending had increased at the rate of inflation.
To calculate the total of $206,877 in foregone wages, each year’s monthly “at inflation” contribution was subtracted from the actual contribution rate, then multiplied by 12 to yield an annual loss, then added to the cumulative losses from the prior year.
The annual wage losses were then converted to constant 2019 dollars to yield a “real” wage loss figure. For those not familiar with the process, BLS publishes a helpful “how-to” fact sheet with formulas and clear examples here. Table 4 explains the methodology used for these calculations.
To calculate the $9.86 per hour difference in housekeepers’ wages, an hourly “at inflation” contribution rate was created by first dividing the actual monthly health contributions by 154, (average monthly hours for full-time hotel workers in San Francisco), then applying the same methodology used for calculating the raw “at inflation” monthly contributions. The actual 2019 hourly health contributions were $12.55, while the hourly “at inflation” contributions were $2.87, where $12.55 - $2.87 = $9.86.
“More than the State’s Minimum Wage”: Data downloaded from the St. Louis Federal Reserve’s “Federal Reserve Economic Data (FRED) website. See State Minimum Wage Rate for California”. 2019 = $12.00.
“Three times faster than the Median National Worker’s Earnings”: Housekeepers’ real wages were calculated by applying the BLS methodology described above to the hourly wages provided by UNITE HERE! Local 2. National median workers’ earnings are from Current Population Survey Data downloaded from the FRED website. See “Employed full time: Median usual weekly real earnings: Wage and salary workers: 16 years and over”. The CPS data are quarterly. Healing and Stealing uses July 1 for each year. Both metrics were then indexed with 1996=100, and the percentage changes calculated from 1996 to 2019 for Figure 2. For figure 3, the real health care contributions were similarly indexed to 100.
There may be technical arguments why median earnings are not precisely comparable to housekeepers’ negotiated wages. Since the comparison metric is percentage growth and the scale of the difference is large (298%), it is unlikely that any definitional issues would alter the conclusion that, despite the pressure of holding onto real health insurance, unionized San Francisco hospitality workers have still made much stronger real wage gains than the national median worker.
Health Costs as Percentage of Compensation: Hager et al, use family earnings (salaries/wages) plus health insurance costs to define compensation. This is obviously incomplete, as the authors acknowledge. Healing and Stealing applied the same definition to compensation for the UNITE HERE! Local 2 data. Pension and other smaller negotiated benefit contributions were excluded from the calculation to allow for comparability.
Growth in Deductibles: The Kaiser Family Foundation’s annual surveys have recorded the scale of certain forms of cost sharing for decades. Unfortunately, the data have not always been reported in consistent form. This is not necessarily KFF’s fault since employers and insurers have demonstrated significant creativity in extracting out-of-pocket payments from patients over the years, and it’s not always easy to keep up.
Patients share costs in a number of different ways, including copays, coinsurance and deductibles. Hager et al focus on family health insurance premiums. Family deductibles come in several forms (one big deductible or a series of individual deductibles) that can make it difficult to track a single measure over time. And, of course, many plans have separate deductibles for drugs or hospital care. Healing and Stealing offers the growth in the average general annual deductible for single coverage only as one example of out-of-pocket cost growth.
The archive of KFF surveys is here.
The 2019 average annual single deductible ($1,396) is on page 114 of the 2019 report.
The 1996 average is derived from historical data in the 1999 report. The average annual single deductible by plan type is on page 72 and market share by plan type is found on page 57. The deductibles for each plan type were multiplied by their market share percentage and summed. The weighted average uses only the in-network deductibles for PPOs and point of service plans, consistent with the 2019 definitions.
It is frustrating and time consuming to have to aggregate a total average in these older surveys. Healing and Stealing urges KFF to assign staff to go back through its reports and develop consistent metrics to make available to the public. The historic inconsistency in reporting measures on the survey makes an extremely valuable data resource less useful.
Differences between the JAMA Open Network study and San Francisco hospitality worker data
Limitations of the JAMA Open Network Study: Healing and Stealing’s analysis of UNITE HERE! Local 2 health benefit spending and wages differs in several significant ways from the Hager et al article. The authors make several assumptions that tend to underestimate the impact that 30 years of health care cost increases at multiples of underlying inflation have had on workers’ wages and standard of living. Both analyses report their data in constant 2019 dollars. Here are the key issues and differences.
Out of Pocket Costs: As noted in the text, since the benefits and out-of-pocket cost-sharing for San Francisco’s unionized hospitality workers have remained roughly constant, the growth of health care contributions accounts not just for premiums, but also for the value of the dramatic increases in out-of-pocket costs that most other workers have suffered. It’s unfortunate that the authors failed to develop a data set that accounts for out-of-pocket costs, since the erosion in the quality of benefits offered to workers has been nearly as destructive as the drastic increase in premiums themselves. Workers really are paying much more for much less coverage.
Healing and Stealing found a proxy for combined premiums and cost-sharing with a single phone call. I’m an independent journalist earning almost no money beyond my pension. Co-author Ezekiel Emanuel, former President Obama’s top health care advisor, is one of the most powerful physicians in the world; the authors work for universities that had a combined $23 billion endowment last year; two of the authors used money from an NIH grant that, unless there was a typo in the article, they were supposed to use to study health system and state-level policy strategies to improve diet and reduce Cardiometabolic Diseases in the US, and; the article includes a 413-word conflict of interest disclosure larded with corporate board appointments and foundation grants. Given the resources at their disposal, the authors’ failure to even try to generate a number that includes the extreme growth in out-of-pocket costs over the last few decades is disappointing, especially since one of their sources - the Kaiser Family Foundation’s annual employer surveys - has been recording deductible costs for years, albeit in inconsistent form.
Wage and Inflation Comparisons: Hager et al correlate family earnings with the cost of employer-sponsored family health insurance coverage from 1988 to 2019.
Healing and Stealing compared San Francisco hospitality employer contributions for health benefits to the rate of inflation (CPI-U) because, as described in the article text, the relationship between wages and health benefits for San Francisco workers is known down to the penny.
Comparing premium growth to family earnings creates some unnecessary complexities and other academic researchers have used inflation instead, but it’s a defensible approach since the national data sets lack the specificity of the San Francisco hospitality worker data. The relevant question for workers is “how will extreme health cost growth affect my ability to meet all of my other financial responsibilities?”
Taxes: Beyond out-of-pocket costs, the authors may also underestimate the impact of health benefits on wages by subtracting the amount that benefits would be taxed if they were paid as wages. The authors decided that in a hypothetical world where benefit costs were shifted to wages, employers would only have increased workers’ pay by the “value” of the benefits because the fact that health benefits are tax exempt inflates the value. So they subtract the average withholding rate from the lost wages.
This is problematic in several ways. First, the notion that employers generally have any idea of the value of health plans is dubious. They’ve certainly failed to control the cost of them, and federal and state policies have failed for decades by assuming that employers are competent or powerful enough to negotiate benefits effectively. They’re in charge of buying health care and we spend twice as much but get much less than people in other countries.
Second, in years where the need for increases in health spending are low, San Francisco hospitality employers shift the entire value of compensation increases from benefits to wages, and vice versa in years of extreme costs. Employers don’t get to keep the taxes – they deduct them and send them to the government. So it’s money spent either way. This may be a function of the protections that workers receive from having a legally binding union contract. The authors cite research showing that total compensation declines by 52% for each dollar of premium increase. As with most such studies, it doesn’t account for out-of-pocket cost growth.
Presumably, as is the case in San Francisco, if employers are committed to spending a bucket of money on labor, they’ll spend it regardless of the way its spent. What the authors are really saying is that workers lack the power to extract wealth from their employers whenever there’s a significant shift in labor market conditions or regulations. If they are correct that given the opportunity, employers would just casually pocket roughly 28% of money they current spend on health care when shifting it to wages, Healing and Stealing strongly advises people to form unions before the government institutes a universal healthcare financing scheme.
Poor Data Presentation and Inconsistencies Between Text and Exhibits: Hager et al present their results in a way that makes comparing them to other data sources inordinately time consuming. The authors don’t include a data table showing the year-by-year results, only images of graphs. Even for the 1988 and 2019 endpoints, they report the topline $125,340 wage loss number, but nowhere in the text do they report two of the half-dozen most critical metrics from which the number is derived - the 2019 average cost of family premiums and average “net” premium costs. The former is relatively easy to find in the source data. The latter is the authors’ own calculation of premium costs minus taxes. Their source material document consists of a bunch of links to large data sets, so the only way to actually compare their results to other research or analyze points that the authors haven’t fully fleshed out is to recreate their research from scratch.
This kind of behavior is endemic to health care financing research and flouts the supposed norms of the academy. It reduces the pace of reproducibility and distorts future research. It may be explained in this instance by the authors’ eagerness to get to questions of inequality based on race and ethnicity. Those are important goals, but doing so by failing to present the basic data for the entire population is a serious flaw.
The poor presentation is compounded by at least one outright error. On a core metric, the text of the article is inconsistent with the exhibits. As noted above, Hager et al claim that the percentage of compensation associated with family health insurance premiums grew from 7.9% in 1988 to 17.7%. Their Figure 1, reproduced below, allegedly shows “Percentage of Compensation Associated With Health Care Premiums Among Families With Employer-Sponsored Health Insurance From 1988-2019 by Race and Ethnicity”. It’s only an image, but every single race or ethnicity on the chart starts below the supposed 1988 average of 7.9%, except possibly for the line presenting the results for Black Americans. Given the relatively small portion of the population that is Black, it is impossible for this graph to be consistent with the numbers reported in the text. Healing and Stealing assumes the text is accurate, but this kind of sloppiness renders the paper unreliable for other researchers, and the error would be obvious to the authors had they included a line for the total population.
Limits of Local 2 Contract Data: First, a single city offers limited applicability to the country as a whole. San Francisco is a high-cost market, both for health care and the general cost of living. Health care markets are geographic by definition, so the costs underlying employers’ monthly contributions may vary significantly from a national measure.
Second, the analysis compares growth to the national rate of inflation rather than a local figure, and inflation can vary significantly by geography.
Third, the data are not direct premium prices. As noted, the monthly contributions go to a separate Fund, which in turn negotiates premiums for various plans offered to Local 2 members.
As noted in the text, compensation increases are “allocated” annually from a single negotiated amount to wages, health and pension. The allocations vary widely. Health allocations can depend on the prior year’s health spending, bids from insurance subcontractors, expected spending and the level of federally mandated reserves held by the Fund. However, over a 23-year period, the contributions accurately reflect the cost of offering family health care for the covered population and thus are reasonably, if roughly, comparable to the Hager et al data.