The Effects of Rising Health Costs on Salaries and Income Inequality

Can the widely noted recent stagnant salary growth for the average U.S. worker and increased income inequality be explained, in significant part, by the rapidly rising costs of health care benefits? This article argues that they can. First, a simple economic model of the growth and distribution of total compensation, salaries and health benefits is posited. We then report recent statistics relevant to the model’s key concepts, supporting the logical structure and empirical specification of the model and its assertion of the centrality of the cost of health care benefits. Finally, we discuss the economic and political implications of this assertion for plan sponsors and public policymakers.

An Economic Model
Imagine a society where there are five groups of workers, sorted by their levels of total compensation. To simplify and focus the analysis, we assume that total compensation includes only salary and employer-paid health insurance benefits, and we consider only full-time workers.

In the first year of the analysis (representing, quite roughly, 1998), the first group — comprising 60 percent of the workforce — is assumed to have earned $25,000 annually in total compensation. The second group, representing 32 percent of workers, earned $50,000 in total compensation; the third group, 5 percent of workers, earned $80,000; the fourth group, 2 percent of workers, earned $120,000; and the fifth group, 1 percent of workers, earned $350,000 in total compensation.

The average total compensation, across society and all groups of workers, is $40,900, while the total compensation of the median worker is $25,000. The ratio of average to median total compensation, one indicator of inequality, is 1.636.¹

In this model, employers pay the cost of their workers’ health insurance and then deduct the cost from the workers’ total compensation. The cost is initially $3,300 per worker, which roughly represents the cost to employers averaged across single and family coverage in 1998.

Because health insurance benefits cost essentially the same for all workers, these costs deliver a proportionately larger hit to those who earn less. The first group pays roughly 13 percent of its total compensation for health insurance ($25,000 to $21,700), while the highest earning group pays only about 1 percent of its total compensation for health insurance ($350,000 to $346,700).² The average salary in this economy is $37,600 and the median salary is $21,700, so the ratio of average to median salary is 1.733.

Inequality measured by salary is somewhat greater than inequality measured by total compensation, ignoring the fact that each group gets equal expected value from the health care system. Nonetheless, most studies measure inequality by salary or income, excluding employer-paid health insurance benefits, rather than by total compensation, given the generally wider availability of worker- and household-level data on wages and income from surveys and administrative records. The percentage of total salaries in the economy earned by workers whose salaries are significantly affected by the cost of health care benefits — say workers earning less than $100,000 — is 84.6 percent. Stated another way, in this society, 3 percent of the workers earn 15.4 percent of total salaries.

Another relevant social indicator is the ratio of the cost of health insurance benefits to compensation. In our simplified model, this is the same as the ratio of health care spending to income (GDP) calculated and projected every year by the Centers for Medicare & Medicaid Services (CMS) and widely reported.

Under the assumptions described above, the cost of health benefits-to-compensation ratio for society as a whole, in the first year, is 8.06 percent. (The costs of public health programs and the taxes paid to finance the programs, especially rapidly growing and underfunded Medicare and Medicaid, are ignored.) As noted above, for the lowest compensation group, the ratio of the cost of health benefits to compensation is 13.2 percent, declining to 6.6 percent for the second compensation group, and so forth, finally dropping to 0.9 percent for the highest-earning — but smallest in number — compensation group.

Now posit that compensation increases 1 percent annually, in inflation-adjusted terms, and that the cost of health insurance benefits, eliminating the impact of general price inflation, increases 4.5 percent annually for 10 years — the same rates of increase being experienced by all compensation groups. These rates of increase reasonably describe recent experience. Furthermore, assume that the distribution of workers among the compensation groups remains constant. That is, the same 3 percent of workers get more than $100,000 in salary in the first year of the analysis and more than $106,527 (as increased by the rate of growth in average real wages) in the 10th year of the analysis.

What are the measures and indicators of general welfare and inequality 10 years later? Average compensation, at $45,179, has increased a respectable 1 percent annually in real terms (as has a common measure of social welfare, per capita real national income or real GDP). However, the average inflation-adjusted, net-of-cost-of-health-insurance, salary — $40,054 — has increased 0.63 percent annually. Because health insurance benefits now cost $5,125 (in real terms), the lowest-earning, largest-in-number compensation group receives $22,491 in salary — an annual pay increase of only 0.36 percent in real terms, over the 10 years. Many members of this group will perceive that their welfare has lagged behind the general welfare, unless they feel that they are gaining more value from the health care system over time.

Over the 10-year period, the inequality indicator, measured by salary, has worsened for this society, to 1.781 — an increase of about 0.05 (although it remains at 1.636 when measured by total compensation). The 97 percent of all workers who make less than $106,527 now get 84.1 percent of total salaries in the economy, about 0.5 percent less than 10 years earlier. The ratio of health care spending to compensation has increased by more than 3 percentage points to 11.3 percent; for the lowest-compensation group, this ratio is now a more painful 18.6 percent.

Obviously this analysis/model is highly simplified (ignoring business cycles, other trends and temporary conditions, and so on). But while parsimonious and perhaps even insightful, is it a robust description of compensation, salaries, the cost of health benefits and inequality in the United States today? We next compare, for congruence with the model results, official statistics that correspond with these key concepts and measures.

Total Compensation, Salary and Health Benefits Statistics
Figure 1 shows the annual increase in total compensation, salary and the employers’ cost of health benefits over the 10-year period from 1998 to 2007. The cost of health benefits consistently outpaced total compensation, which, in turn, increased faster than salary. The very large jumps in the cost of health benefits in 2002 and 2003 were particularly painful, coming on the heels of a recession and an initially weak economic recovery with sluggish job and wage growth.

On average over the period, the cost of health benefits increased by 6.4 percent annually, compared with 3.6 percent for total compensation and 3.2 percent for salary/wages. Subtracting out a general inflation rate of 2.5 percent produces rates of increase similar to those we used above in the economic model.

Figure 2 suggests that the weakness in salary growth can be largely attributed to the higher cost of health care rather than other benefits, such as retirement.³ From 1998 to 2007, all benefits, as a share of total compensation, increased by about 2.5 percentage points, from roughly 27.5 percent to more than 30 percent. That was the same as the percentage point increase in the compensation share cost of health benefits — about 2.5 percentage points, from just above 5.5 percent to almost 8 percent — although the timing of the increases differed slightly.

The data in Figure 2 include employees, such as part-time workers, who do not receive health insurance or other benefits from their employers. The Kaiser Family Foundation, using nonpublished data from the Bureau of Labor Statistics and limiting the focus to employees with access to employer-provided health insurance, calculated that the ratio of employer-provided health costs to salary increased from 8.2 percent in 1999 to 11 percent in 2005.

Using this same data, the Kaiser Foundation researchers also looked across different occupation groups, across time, as reported in Figure 3 below. The percentage increases in the health insurance cost share for this period were larger for lower-compensation occupations, like laborers and transportation workers, than for higher-compensation occupations, such as executive/managerial, professional and craft/precision production workers.

Figure 4 below shows median and average wages, according to data from the Social Security Administration. Although the SSA data include seasonal and part-time workers, the overall trends would likely be similar for full-time workers only. There are clearly business-cycle effects, so a point-to-point comparison over a longer time period is more instructive than focusing on particular years.

The empirical ratio of average to median wages — the main measure of inequality in society in our model — increased from 1.45 in 1998 to almost 1.50 in 2006, an increase of 0.05. Another measure of inequality, as mentioned above, is the percentage of total earnings in the economy going to workers whose salaries and wages fall below a certain threshold, say the taxable maximum wage for Social Security purposes, which increases with the average wage (in 2008, it is $102,000). In 1998, this percentage was 84.5 percent; in 2005, it was 84.0 percent — a decrease of 0.5 percent, again exactly the change that our model indicated would result exclusively from the accelerating cost of health insurance.

The distribution of workers among earnings groups may be measured by the percentage of workers earning more than a sizable fixed amount in salary, indexed for average economywide wages, in any year. The SSA reports detailed statistics on the distribution of workers across earnings categories year by year. In 1998, 2.45 percent of workers earned more than $100,000 in salary and wages. In 2006, 2.6 percent of workers earned more than $134,000, the equivalent wage in 2006, as increased by the rate of increase of average nominal wages (that is, with no adjustment for price inflation) since 1998. In other words, there was essentially no movement in the distribution of workers across earnings categories, even as other measures of inequality increased. This is all consistent with our economic model. Although other factors might explain longer-term trends in inequality, they are not evident here, over this period.

Finally, we report the well-known ratio of national health expenditures to GDP, estimated by CMS. From 1998 to 2006, the ratio increased from 13.6 to 16 percent — an increase consistent with our other measures of the cost of health benefits.

Taken together, these statistics are quite congruent with our economic model over the last 10 years. Our simple model/analysis appears to accurately reflect recent labor-market conditions (including trends in compensation, wages and inequality) and the cost of health care. The analysis suggests that growing wage inequality and the average worker’s stagnating wages can be explained by the rising cost of health care and its disproportionate effects on average workers. What are the economic and political implications of this supposition?

Implications
An important assumption of our interpretation of the analysis here is that workers do not associate rising health care costs with commensurate improvements in health care value. This assumption may be somewhat exaggerated because surely today’s health care system offers new and more effective medical treatments than before, and workers and the public are aware of and appreciate these advances. But, at the same time, there seems to be considerable skepticism about the extent of the improvement.

In other areas of economic life, we expect new technologies to improve amenities and lower costs. For example, today’s computers offer vastly expanded capabilities and much faster processing than the computers we used 10 years ago — and today’s models are considerably cheaper. Similarly, automobiles have become much more reliable, long-lasting and safer since the 1980s and 1990s, but prices have not surged in real terms. Health care has not experienced technology-induced reductions in costs.

Moreover, many careful studies have found large geographic differences both within the United States and between the United States and other countries in the intensity and cost of health care, with no discernable differences in health outcomes. Some have attributed these inefficiencies, including the overuse of procedures and drugs, in the U.S. health care system to counterproductive incentives inherent in the third-party payer mechanism of low-deductible private or public insurance (other than strict managed care). They believe it would be more effective to load the incentives toward healthier behaviors and prevention rather than unnecessary or avoidable procedures. Other key issues are whether low-wage workers, left to their own devices, would spend as much on health care benefits as high-wage workers and, if so, the effects of differential spending on actual health outcomes.

Because the highest health care bills are concentrated among those with severe and chronic ailments, insurance coverage may also be less appreciated by more moderate consumers of health care, who seem to get less for their money. Furthermore, when the overall cost of insurance coverage increases, this psychological dissonance between perceived value and average cost may intensify, possibly prompting some employers and workers to drop their insurance coverage altogether.⁴

In the United States today, real salaries have not increased significantly for a large share of the workforce because of the rapidly rising cost of health insurance benefits, while the perceived and actual increase in the value of health benefits has fallen behind the rapid increase in costs. Moreover, common indicators generally show more inequality in society, which has fueled demands for more public funding of health insurance for low-wage workers and their families, such as the recently vetoed legislation to expand the State Children’s Health Insurance Program. Similarly, some of the presidential candidates propose to either mandate or provide incentives to extend health insurance coverage to everyone, again supported by general federal government revenue funding, which would effectively shift some of the burden of health insurance coverage from low-wage to higher-wage workers.

Employer-sponsors of health insurance plans have responded to rapidly rising costs, in part, by making these costs more transparent to workers, thereby increasing workers’ appreciation of their health benefits. For example, some employers give their workers an annual statement detailing the total cost of employer-provided benefits, including health insurance benefits. Also, through their more open and explicit design, consumer-driven and high-deductible health plans provide more transparency. These plans are intended to encourage workers to be more cost-sensitive consumers of health care, particularly for ordinary care, and to emphasize preventive care, which is exempt from the high deductibles. They are also expected to reduce the rate of increase in costs and improve value.

Whether to increase public financing of workers’ health care is an important issue with far-reaching economic and political consequences. Even more important, however, is addressing the deeply rooted and widespread perception, and reality, that the increase in the cost of health care has outpaced its increase in marginal value. This will likely require more than redirecting the flows of finances and distributional burden. It will take fundamental and widespread changes to alter incentives, in both private- and public-sector health plans, and enhance the awareness of costs, as well as other direct initiatives in health care, such as study of the relative efficacy and cost of alternative treatments.

¹ A ratio of average to median greater than 1 indicates a concentration of compensation or salaries in the higher percentiles, and, moreover, this ratio is the way the Social Security Administration presents its data for this purpose.
² It is a simplification to assume that the cost of employer-provided health benefits is the same across all compensation groups; nevertheless, studies show that the range is fairly tight from lowest to highest income groups — no more than 1-to-2 — and likely has not changed much over time, partly due to the nondiscrimination rules governing employer benefit plans.
³ Over this period, there may also have been some substitution of other “distant” benefits, like retirement plan benefits, for “immediate” health insurance benefits, especially among lower-wage workers.
⁴ Working through the ebb and flow of health care coverage owing to business-cycle effects, the percentage of wage and salary workers offered coverage from their employer and the percentage of workers taking coverage when offered have remained fairly steady in the last 10 years. According to Employee Benefits Research Institute (EBRI) Issue Brief Number 303, March 2007, the eligibility rate for nonelderly adult wage and salary workers was 74.0 percent in 1997, 76.2 percent in 2001 and 74.0 percent in 2005. Eligibility declined during and after the 2001 recession because of a relative increase in the number of part-time employees. The actual rate of uninsured status among those eligible for coverage increased from 3.8 percent in 1997 to 4.8 percent in 2005; the cost of coverage was the main explanation for this increase. Note that the stability in employer-provided health coverage found in the EBRI study is different from the finding in other studies that health insurance coverage of workers has declined in the last few years, particularly for the lower-paid. Those studies, however, include self-employed workers who typically do not have employer coverage, and, in the 2001 recession and afterwards, the number of self-employed workers increased significantly. More recently, however, the number of self-employed workers has likely stabilized or even declined, as occurs during and after the recovery phase of all business cycles.

May 2008