You are on page 1of 80

Executive Summary Treating Our Ills and Killing Our Prospects

By Steven A. Nyce Towers Watson and Sylvester J. Schieber June 29, 2011

Sylvester Schiebers contributions to this paper were supported by the Coalition for Affordable Health Coverage. The authors are indebted to Michael Slover of Towers Watson for his contributions to the statistical analyses included in parts of the paper and to Susan Farris of Towers Watson for her editorial help in preparing the final draft. All of the conclusions and opinions stated herein are those of the authors and do not necessarily represent those of the Coalition for Affordable Health Coverage, Towers Watson or any of their associates.

Executive Summary
Most of us are aware that health costs have been rising more rapidly than those in most other aspects of our lives and that they seem to have accelerated in recent years. There is also widespread sentiment that the rewards for increasing worker productivity in the United States in recent times have not raised all boats and that middle-class workers are increasingly being left behind with their rewards growing more slowly than productivity levels. Both of these perceptions are widely reported but they are seldom linked. For most workers, the majority of the health expenses they incur each yearand the health cost inflation that goes with themare hidden from the naked eye because they are financed through compensation packages, seemingly paid for by employers who actually consider them part of the cost of hiring workers. Of course, the costs of hiring workers have to be covered by their productivity or the employer cannot continue to offer them jobs. These health benefits costs have gotten so large in recent years and have been growing so fast that they now are contributing to the noticeable slowdown in workers pay and income growth. Robert Gordon has been studying the issue of growing inequality in income and pay in the United States and has raised a cautionary note regarding the conclusions that have been drawn in this regard. Gordon argues that productivity and income growth rates are computed on an inconsistent basis and when income is measured on a per capita basis as productivity is and both are converted into constant dollars using the GDP price deflator rather than separate deflators, the difference in the rates of growth in income and productivity are minimal. Despite this conclusion, Gordons analysis shows that from 2000 to 2007, output per hour was growing at a rate of 1.39 percent per year but mean income per person of working age grew hardly at all over that period. He suggested that further research was required to ascertain what was happening to this disappearing income.1 A substantial part of the answer to Gordons puzzle is that tracking of aggregate compensation is based on the National Income and Product Accounts (NIPA) which includes cash pay for employed workers plus employer contributions for benefits provided to them as part of their total remuneration. The benefit costs include the employers share of payroll taxes for Social Security, Medicare and unemployment insurance plus employer contributions for retirement health benefit plans that they sponsor for their workers. In 2000 under the NIPA annual data series, benefits costs absorbed by employers were 10.6 percent their total compensation costs, rose to 12.5 percent by 2007 and to 13.7 percent by 2009.2 For the most part, the economic literature and the Census data series on the distribution of earnings and income ignore this element of compensation paid to workers and its distribution across the earnings spectrum. The benefits are sizeable enough that they have the potential to skew any analytical results that ignore them. The reason that compensation grew more rapidly in the latter period while pay fell behind was because of higher growth rates in the parts of compensation that do not show up in the regular paycheck. These are benefits costs that employers incur in behalf of workers. The benefit costs include the employers share of payroll taxes for Social Security, Medicare and unemployment insurance plus employer contributions for retirement health benefit plans that they sponsor for their workers. During the 1990s, hourly pay grew at around 1 percent per year i

across most of the earnings spectrum. Pay growth in the period from 2000 through 2009 has been closer to 0.5 percent per year across the middle parts of the earnings distribution but has lagged behind the 1990s at all earnings levels (See Figure 5). The story with total compensation is somewhat different. During the 1990s, total compensation was also growing at a rate of around 1 percent per year over nearly 90 percent of the earnings spectrum. From the 40th to the 90th percentiles of the distribution, however, compensation growth during the 2000 to 2009 period actually exceeded what had been achieved during the 1990s. When benefit costs grow more rapidly than the compensation budget, wage growth is reduced. The growing share of compensation diverted to benefits, shown in Executive Summary Table 1, explains some of the public consternation about what has been happening to disposable earnings. While the major elements of this current discussion focus on health care issues, the results in the table are driven by more than just health care costs over the time spans reflected. During the 1980s, increases in the payroll tax acted as a drag on workers cash rewards as we implemented the changes to the Social Security Act that were adopted in 1977 and 1983 in response to the financing crisis that faced us in the late 1970s and early 1980s. During the most recent decade, employer contributions to employer-sponsored retirement plans have increased dramatically. The largest share of these latter increases has been directed toward defined benefit plans. While these plans have been curtailed in recent years, there is a tail of liabilities that still must be met from these plans that were much more prominent during the 1980s and early 1990s. These added contributions have arisen, in part, because contribution levels were reduced in the 1980s in response to regulatory changes that reduced employer contributions to the plans and to the booming financial markets in the 1990s. The funding problems that have plagued these plans since the beginning of the current century have required the dramatic increase in contributions because the aging of the baby boomers is making the liabilities more immediate, because falling interest rates have accentuated them and because the turmoil in the financial markets had reduced the value of asset trusts in many cases. The sluggish growth in workers disposable income in recent years has been attributed to a variety of things including changing reward structures in the corporate world and tax policy as the focus of many commentaries. Those factors may have played some role in developments but growing benefit costs were likely a much larger reason for the unsatisfactory results many people have had at the pay window in recent years. The underlying factors that have affected the nonwage components of compensation over the past three decades have not played themselves out so these forces will play a continuing role in the rewards picture for the future. In addition to the effects that growing health benefits costs have been having on workers pay and income, there is also evidence that they have been affecting labor markets as well. It is through a job that most workers acquire their health insurance in modern America. Health benefits have become totally engrained in the decisions of workers about all aspects of their workforce choices from what job to take, when or if to leave a position and whether or not to work at all. Single mothers are more likely to take full-time employment to qualify for health benefits than married mothers who can qualify under their husbands insurance. People receiving disability benefits are reluctant to go back to work because they will lose valuable health insurance coverage once they demonstrate they can provide for themselves. Workers unhappy in their current jobs and older workers wanting to find a bridge job as entry into retirement are trapped often in existing jobs with insurance coverage. These benefits cause ii

distortions in labor supply decisions that must ultimately affect the overall efficiency of our economy. Executive Summary Table 1: Share of Compensation Gains Provided in the Form of More Expensive Benefits Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods*
Earnings decile 1 2 3 4 5 6 7 8 9 10 1980-1990* 100.0% 100.0% 90.8% 54.1% 63.9% 43.0% 48.6% 36.8% 29.7% 21.4% 1990-2000 30.4% 23.1% 25.0% 21.3% 17.8% 18.8% 12.4% 9.6% 7.8% 6.8% 2000-2009 35.2% 47.7% 52.3% 60.8% 55.7% 55.3% 54.8% 50.3% 45.0% 37.7%

Source: Updated results of projections presented in Steven A. Nyce and Sylvester J. Schieber, Health Care Inflation, Must Workers Bear the Brunt, Milken Institute Review (Second Quarter 2010), pp. 46-57.
*

Total benefit cost increases in the 1980s for the first and second earnings decile exceeded 100 percent of compensation growth. In both cases, benefit costs increased significantly but total compensation growth was in the negligible first decile and negative in the second.

In Executive Summary Table 2, the isolated effects of health benefits costs increases on compensation are shown for the analysis period. Empirical analyses, by their nature, consider the implications of things like health cost across broad groups of the population. For example, the heart of the analysis developed here is looking at the effects of health benefit cost inflation on the levels of pay realized by the full-time workers by earnings levels. The earnings splits that we have chosen are deciles. When we show in Executive Summary Table 2 that health benefits cost increases have absorbed 23.6 percent of the added compensation reward paid to workers in the bottom earnings decile in the period 2000 through 2009, our result includes all workers in that decile even though many of them do not actually participate in a plan. Because of declining coverage of workers, which has tended to be concentrated at lower earnings levels, the rates at which increasing health costs have crowded dollars out of the paycheck has been dampened. For many of these people, however, the loss of employerprovided health insurance has meant larger out-of-pocket expenditures for health consumption out from their disposable income. It is a classic case of damned if you do or damned if you dont, in that either way, other consumption is adversely affected. It is important to remember that the primary purpose of the analysis is to understand how health inflation is affecting the general rewards for broad groups of the workforce. When an iii

employer is considering the addition or retention of a worker, however, the focus is more narrowly focused on what that worker will cost in comparison to the expected value that he or she will bring to the organization. The marginal costs of workers in the various earnings deciles who actually take health insurance are quite different than the average of all workers in the deciles. Executive Summary Table 2: Share of Compensation Gains Provided in the Form of More Expensive Health Benefits Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods*
Earnings decile 1 2 3 4 5 6 7 8 9 10 1980-1990* 285.8% 100.0% 106.9% 57.2% 74.4% 45.2% 55.5% 38.7% 21.4% 12.1% 1990-2000 26.8% 20.8% 23.6% 21.0% 19.8% 22.5% 15.5% 12.1% 9.1% 2.9% 2000-2009 23.6% 30.4% 30.1% 36.5% 28.9% 26.7% 25.8% 20.1% 15.0% 9.1%

Source: Updated results of projections presented in Steven A. Nyce and Sylvester J. Schieber, Health Care Inflation, Must Workers Bear the Brunt, Milken Institute Review (Second Quarter 2010), pp. 46-57.
*

Health benefit cost increases in the 1980s for the second earnings decile exceeded 100 percent of compensation growth. Benefit costs increased significantly but total compensation growth was negative in the second decile.

Executive Summary Table 3 shows how health benefits care costs have risen relative to wages between 1980 and 2009 for workers who actually enrolled in the health benefit plans offered by their own employers. In 1980, employers costs for those who enrolled in their programs cost single-digits relative to wages for all decile groups except the lowest with the median enrolled employee costing about 6 percent of pay. Over the next three decades, those costs have grown more than threefold relative to wages and reaching more than a third of individuals wages among the lowest decile groups. In fact, for the lowest decile group they have nearly eclipsed half of an employees take home pay in 2009. Likewise, these costs have been growing at much faster pace for the lowest paid highlighting the greater impact that compounding has on the lower pay groups. For example, health benefit costs relative to wages for the second decile were twice those of workers in the ninth decile in 1980, which by 2009 has risen to more than three times. Because of that, health benefits have the potential to make certain workers uneconomical in some cases.

iv

Executive Summary Table 3: Health Benefit Costs as a Share of Wages for Full-Time, FullYear Workers Receiving Health Care Benefits through Their Own Employer
1980 1 2 3 4 5 6 7 8 9 10 15.4% 9.5% 8.0% 7.2% 6.3% 5.8% 5.4% 4.9% 4.3% 3.2% 1990 30.9% 18.7% 15.3% 13.3% 11.6% 9.9% 9.2% 8.2% 6.9% 4.9% 2000 38.1% 22.9% 18.6% 16.0% 14.0% 12.1% 10.8% 9.2% 7.8% 4.7% 2009 49.5% 30.9% 25.5% 22.3% 19.4% 16.8% 14.8% 12.5% 10.2% 6.3%

Source: Developed by the authors. On the labor demand side, the idea that employers treat health benefits as part of the compensation bundle suggests that increasing health premiums for their benefit plans would merely be adjusted by reducing other aspects of the total package. But workers cash pay tends to be sticky downwardit is difficult to reduce pay without causing a variety of disruptions among workers. If employers are forced to absorb health cost increases that exceed the added productivity that workers bring to the table, it is likely they will curtail employment in some cases. Katherine Baicker and Amitabh Chandra have also developed an empirical analysis of effects of rising health benefits costs on labor demand and estimate that a 10 percent increase in health insurance premiums reduces the aggregate probability of employment by 1.6 percent and total hours worked by 1 percent.3 To put these results into context, taking the excessive rates of growth in employers health benefit costs over the past decade compared to productivity growth, health costs have increased an excess of 20 percent. Extrapolating Baicker and Chandras results, this suggests that growing health benefit costs have added 2 percent to the unemployment rate over the period. No one knows for certain what the implications of the 2010 health reform law will be for U.S. workers in terms of their future health costsor even how they will acquire their health insurance coverage in the future. The analysis of what has occurred over the past three decades suggests that a considerable share of the disappointment with the rewards that many workers have received in recent years is due to the voracious appetite that health benefits have brought to bear on their productivity rewards. The extension of the trends on health benefit cost growth that have persisted for decades now suggests that if we cannot bring excessive health care inflation under control, workers prosperity is going to be increasingly threatened. A full-time worker in the second earnings decile in 2009 earned around $25,000 in total compensation on average. If his or her productivity goes up by the rate of growth that the Social Security actuaries estimate, by 2019 this worker will be earning around $36,600 in total compensation but nearly 75 percent of the difference from 2009 will have been consumed by rising health benefit costs. If the worker is being provided family coverage, the cost of health benefits will grow to consume all of v

the added productivity contribution. Do we really think that health cost reform is going to restrain health inflation? Even Peter Orszag, who was the Director of the Office of Management and Budget and a major architect of the health reform package, and Ezekiel J. Emanuel who was a special advisor to the White House and OMB during the reform development suggest that under the new reform model, total health expenditures in the United States in 2030 will only be 0.50 percent less as a share of GDP than under prior law.4 One of the goals of health reform is to broadly expand the coverage of the population under health insurance programs. In a scenario where we assume that future health costs grow at the rate they have been growing since the turn of the century and where we assume that all workers take up health insurance in the context of employer-sponsored plans, our results are sobering. In this case, not only is recent health inflation eating away at cash wages in the compensation package, but expanded coverage will eat away further at aggregate rewards other than health benefits paid to workers. Because the expansion in coverage would be so much more dramatic in the lower ends of the earnings distribution, the growth in the rate of compensation absorption is more dramatic for lower earners. Once again, the results portend that we are at risk of health inflation stealing from our productivity rewards at rates that will leave little for other aspects of the compensation package. The stark result of the sort of scenario presented in Executive Summary Table 4 is that employers simply cannot offer many workers the benefits implied and pay them a wage that might grow over time beyond that and remain competitive in a global economy. The potential subsidization of health insurance under the health care reform law for workers who acquire it outside their employers suggests the economics of offering coverage will change significantly. Many employers, particularly in low-wage industries, will likely eliminate their plans and let workers fend for themselves in the new health exchange marketplace. At the margin, this sort of outcome may work and may even be desirable but we cannot avoid the fact that we are facing a national marketplace here. The mere shifting of health insurance costs from employers compensation packages to a mix of public subsidy and workers contributions out of their remaining disposable wages will not reduce national costs unless we bring excessive inflation under control. If health reform does not reduce the rate of growth in our costs, we must ask who is going to pay the bill? It is important to keep in mind that our projections have not included the bill for additional public subsidies under the new health reform law for workers not receiving employersponsored health care benefits. Nor do they include the potential response that policymakers will ultimately bring to bear on the underfunding of Social Security and Medicare that now persists which we have concluded in a separate analysis will be substantial.5 The Congressional Budget Office projects that the cost of federal entitlement programs will rise from 10 percent of GDP in 2010 to 16 percent of GDP by 2030. Ultimately these costs will have to be absorbed by the productivity of our economysuggesting that at least two-thirds of the bill will come out of workers pockets one way of the other if costs are distributed in accordance with the way our national output is distributed. The workers costs to cover these federal expenditures may not come out of their compensation but they will still be extracted from their paychecks in the form of higher taxes. The potential of these costs hanging over our heads with seemingly little inclination on the part of policymakers to address them suggests that we have extremely little margin for any added subtractions from future workers compensation if we wish to pass on to vi

them the American Dream that each generation is given a chance to better itself compared to those that have come before. Executive Summary Table 4: Share of Compensation Gains Provided in the Form of More Expensive Health Benefits Assuming Expanded Coverage Under PPACA is Paid by Employers Where Health Cost Inflation Rates Persist at Current Rates
Projection Periods Earnings decile All 1 2 3 4 5 6 7 8 9 10 --------------------------------------------------------2009 to 2015 to 2009 to 2015 2030 2030 ---------------------------------------------45.2% 41.9% 42.8% 272.7% 136.8% 94.5% 71.2% 58.5% 47.1% 40.6% 32.2% 26.3% 16.4% 134.4% 87.5% 73.2% 65.0% 57.3% 51.4% 45.6% 38.9% 32.2% 20.3% 169.5% 100.0% 78.6% 66.6% 57.6% 50.3% 44.3% 37.2% 30.7% 19.3%

Source: Developed by the authors as described in the text.

vii

Treating Our Ills and Killing Our Prospects

By Steven A. Nyce Towers Watson and Sylvester J. Schieber June 22, 2011

Sylvester Schiebers contributions to this paper were supported by the Coalition for Affordable Health Coverage. The authors are indebted to Michael Slover of Towers Watson for his contributions to the statistical analyses included in parts of the paper and to Susan Farris of Towers Watson for her editorial help in preparing the final draft. All of the conclusions and opinions stated herein are those of the authors and do not necessarily represent those of the Coalition for Affordable Health Coverage, Towers Watson or any of their associates.

Table of Contents
Introduction .................................................................................................. 2 Background ................................................................................................... 4 Hidden Compensation Masking Workers Rewards ................................ 8 Factoring Benefits into the Rewards Picture ........................................... 13 Pay and Compensation across the Earnings Spectrum .......................... 24 Health Benefit Costs and Labor Markets ................................................ 32 Labor Supply Effects of Health Care Benefits.......................................... 33 Labor Demand Effect of Health Care Benefits ......................................... 40 The Past as Prologue .................................................................................. 51 The Prologue Is almost Past ...................................................................... 55 Health Care Dominates Our Prospects .................................................... 58 Conclusion ................................................................................................... 65 Endnotes ...................................................................................................... 68

Introduction
When most of us were young, we played with learning books that required we draw a series of lines along a sequence of numbers that resulted in the outline of a picture. In some regards, those learning books can serve as an analogy to a number of evolving economic developments that we sometimes observe as independent phenomena which are hard to understand until we link them together to reveal a portrait of our modern lives. In recent years, we have all read, heard about or experienced the pain of health cost inflation through higher premiums, larger copayments and deductibles. In some cases, the situation has been made even worse for those people who have either lost their health insurance or had it curtailed and have subsequently faced economic ruin as they have dealt with serious illness in their families. On a somewhat independent track, another stream of news has documented the evolving pattern of economic advancement in recent years suggesting that the better off in our society seem to be making steady gains in growing their incomes and gaining wealth while the middle class and those further down the economic ladder seem to struggle harder and harder but have been losing ground. The dual recessions that we have experienced since the beginning of the new millennium have stolen many jobs and the new ones being generated in the sluggish recovery now underway do not seem to hold out the promise of a good paycheck, a generous health benefits plan and the secure retirement package that we recall the earlier generations of our forebears enjoying. For many of us, the growing premiums, copayments and deductibles in our health benefits plans have been an aggravation if not painful. What many people do not realize is that, despite some shifting of costs to workers for their health benefits, employers costs for these benefits have consistently continued to grow more rapidly than total compensation budgets. For 2

employers, health benefits costs along with their contributions for pension and retirement savings plans and their contributions for their share of payroll taxes are all part of their compensation costs; what it costs them to hire and maintain workers. When health benefits costs rise, they represent an increase in compensation costs which ultimately have to be covered by the overall productivity of the workers who receive them. Benefit costs are an element of compensation that is generally not visible to us as workers but rising costs for these ultimately have ramifications for workerssometimes in the form of slower growth in wages and sometimes in the form of fewer available jobs for some who might otherwise be employed. In the case of health care benefits, this can also manifest in the form of higher cost sharing through increased contributions out of pay or higher out-of-pocket costs both of which intrude into individuals consumption budgets. When we pull back the veil that hides benefits costs, it helps to explain some of what has been happening to workers pay levels and to their jobs. In the following discussion, we develop in some detail the linkages between benefits that employers provide to workers, their compensation and pay packages as well as the implications for jobs. In some research that brings benefits into the compensation picture and explores their ramifications, the view is restricted to health benefits. As you will see in the discussion that follows, employers contributions for their retirement plans and payroll taxes are also important. Although we broaden the consideration of employer-sponsored benefits beyond health care, once we explore the general implications they have for workers rewards, we step back and take a more detailed look at the implications that excessive health benefits cost inflation has played in recent years and explore the potential future effects they might have if we do not get health inflation under control.

Background
An article in the September 13, 2010 issue of the Bismarck Tribune6 pointed to a report recently issued by the Census Bureau on Income, Poverty and Health Insurance Coverage in the United States: 2008.7 Edward Lotterman, the author of the Bismarck Tribune article, tracked the median earnings of men ages 15 and older who had year-round, full-time jobs over the past several decades. In 1960, these workers had median earnings of $34,152 in inflation-adjusted to 2008 dollars. In 2008, nearly a half century later, the median earnings of this group were $46,367. They had gained a little over $12,000 in 48 years, just a 0.6 percent growth rate in real wages over the period. This track record over nearly 50 years was not particularly impressive but the story was even worse than first implied. Lotterman reported that the median earnings in the annual series had reached $48,452 in 1973 and hadnt gotten back to that level since. Mens median earnings had been over $47,000 three years during the Jimmy Carter Administration and twice during George W. Bushs first term. Lotterman found this result particularly remarkable given that productivity as measured by output per hour of workers in the business sector had doubled over the period from 1973 to 2009. Pursuing the issue further, Lotterman looked at mean income instead of medians and found a different result. Average disposable income had nearly doubled from 1973 to 2009. Mean incomes growing rapidly when median earnings were stagnant or falling meant that the rewards for increasing productivity were being paid to workers at the upper end of the earnings distribution. According to Lotterman this is what happened over the last three decades: From 1979 to 2005, the mean after-tax income for the highest fifth of the population increased by 69 percent. For the poorest fifth, the change over 28 years was 6 percent. For the middle threefifths, the change was about 20 percent.8 Similar stories have become the centerpiece of many 4

political discussions and economic analyses in recent years. There is a widespread sentiment that the rewards for increasing worker productivity have not raised all boats. Robert Gordon has studied the issue of growing inequality in income or pay in the United States and raises a cautionary note regarding the measures often cited to document the phenomenon that Lotterman and others have noted. Gordon recently dissected the reported differences in income and productivity growth that are frequently cited.9 Since productivity growth rates are based on individual worker output, he suggests that comparing such growth with income growth should be done on the same basis. He shows that when income is distributed across the whole working-age population the difference in the mean and median growth in income over the 1979 to 2007 period is about one-fifth the level that Lotterman noted. Gordon also notes another methodological problem with the conventional comparisons of income and productivity, namely the common practice of using inconsistent deflators to put nominal earnings and product on a consistent dollar basis across time. To put them on a consistent basis, Gordon uses the GDP deflator. Finally Gordon notes that the hourly productivity measures generally used in these sorts of analyses are based on worker productivity growth rates for the nonfarm, private business sector. Of course, this leaves out the farm sector as well as government and domestic workers who, between them, account for a significant share of American workers. In considering economy-wide comparisons of growing pay compared to growing productivity levels, there is no justifiable reason to include them in computing pay growth over time but then leave them out of the comparative improvement in productivity rates. The various adjustment factors that Gordon develops in his analysis are presented in Table 1. Under the conventional measure of the gap between income and productivity growth, based on the rate of growth of median household income minus the rate of growth in productivity

in the private, nonfarm business sector, the income-productivity gap for the period 1979-2007 was 1.46 percent per year as shown in line 13 in Table 1. Under Gordons adjusted alternative, the difference in the rates of growth in income and productivity measured only 0.16 percent per year for the period as shown in line 14.10 Table 1: Annual Growth Rates, Median and Mean Income and Productivity for Selected Periods
19792007 1. Census median household income 2. Census mean household income 3. Mean minus median (line 2 minus line 1) 4. Census median income per person (15+) 5. Census mean income per person (15+) 6. Mean minus median (line 5 minus line 4) 7. Deflator used by census (CPI-RS) 8. PCE deflator 9. GDP deflator 10. Median income per person (15+) with GDP deflator 11. Total economy output per hour 12. Nonfarm private business sector output per hour 13. Conventional income-productivity gap (line 12 minus line 1) 14. Alternative income-productivity gap (line 11 minus line 10) 15. Alternative gap as percent of conventional gap 16. Mean income per person with GDP deflator 17. Hours per person 18. Output per person (line 11 plus line 16) 11.0 1.60 0.20 1.86 -8.0 1.58 0.41 1.69 26.9 1.60 -0.08 2.09 -123.8 3.56 0.76 3.06 76.0 0.02 -0.69 1.39 0.16 -0.09 0.52 -1.04 1.62 1.46 1.12 1.91 0.84 2.52 1.95 1.43 2.64 2.71 2.59 1.50 1.66 1.37 1.28 1.65 2.17 3.34 2.30 0.46 2.08 0.49 0.93 0.44 1.15 1.25 0.09 3.50 3.27 3.15 19791995 0.31 0.96 0.65 1.00 1.21 0.21 4.24 4.16 3.87 19952007 0.73 0.90 0.17 1.34 1.29 -0.05 2.51 2.09 2.19 19952000 1.87 2.50 0.63 2.68 2.90 0.22 2.30 1.76 1.64 20002007 -0.09 -0.25 -0.16 0.39 -0.05 -0.44 2.65 2.32 2.58

Source: Robert J. Gordon, Misperceptions about the Magnitude and Timing of Changes in American Income Inequality, (Cambridge, MA: National Bureau of Economic Research, 2009), NBER Working Paper 15331, Table 1.

Despite Gordons evidence that the skewing of income toward higher earners since the 1970s has been exaggerated, there is significant economic literature documenting the dispersion of earnings toward the end of the twentieth century. There had been considerable wage compression during the 1940s, in part due to the structure of wage and labor market controls during World War II and the effects of war requirements on the relative demand for various types of workers. As labor market controls were relaxed after the war, the narrower earnings distribution persisted for some time.11 The earnings and income distributions remained stable into the early 1970s to the extent that Henry Aaron remarked that tracking the changes in these measures was like watching the grass grow.12 From the early 1970s through the 1980s, something changed and male wage rates and family income inequality increased sharply.13 Gordon points to evidence that the decline in unionization, the shift in our balance of payments from being a net exporter to becoming a net importer and the dramatic increase in immigration rates and levels since the end of World War II have each contributed somewhat to wage and earnings inequality in recent decades.14 Whether measured by total income or earnings, the evidence that those in the upper end of the income distribution gained relative to those lower down the distribution, beyond the mid1970s, has resulted in a fairly widely embraced conventional wisdom that the haves in our society have been gaining steadily compared to the have nots and that the process continues unabated. Richard Burkhauser and three colleagues have recently published a number of studies attempting to sort out some of the conflicting evidence from various studies of the earnings distribution. In one of them, they focused on the period 1975 through 2004 and concluded that for the poorest 99 percent of the income distribution, the increase in inequality since 1993 has

been significantly slower in the USA than in the previous two decades.15 Still, there is a strong public perception that lower and middle class workers have realized little or no gain from improving productivity levels in recent years. For example, Robert Gordons summary of the results presented in Table 1 notes that output per person from 2000 to 2007 was growing at a rate of 1.39 percent per year as shown in line 18 but that the mean income per person of working age adjusted by the GDP deflator, shown in line 16, had hardly grown at all.

Hidden Compensation Masking Workers Rewards


Even though the statistical evidence suggests that workers rewards have not tracked their productivity gains, there is something troubling about these results on further scrutiny. Gordon has tracked compensation over the period that analysts have been concerned about the growing dispersion in income and has shown that total compensation growth rates have tracked worker productivity growth rates closely. While he did not show it in the fashion that it is presented in Figure 1, the picture there summarizes Gordons observations regarding the situation. Over the period shown there, productivity and compensation growth rates tracked each other closely and had actually done so over virtually the whole post-World War II era. There was some evidence that total compensation growth might have been trailing off slightly from productivity growth rates after 2000 but Gordon noted that by the second quarter of 2009, it seemed to be back on track and that the share of total GDP paid to workers in compensation at that time was almost identical to what it had been in 1959. With aggregate compensation growing at the same rate as productivity rates but income per working age person growing much more slowly than productivity, the result suggested that somewhere a portion of workers rewards were getting lost in the income measurements being published by the Census Bureau based on the periodic household surveys that are used to track 8

income across the population over time. Gordon suggested further research was warranted to find out what was happening to this income. Figure 1: Growth of Productivity as Measured by Output per Hour of Labor Input and Growth of Total Compensation Paid to Workers as a Multiple of Base Values in 1980
1.8 1.7 1.6 1.5 1.4 1.3 1.2 1.1 1.0 0.9 Produc3vity Compensa3on

0.8 1980 1983 1986 1989 1992 1995 1998 2001 2004 2007

Source: Derived from unpublished data from the Office of the Actuary, Social Security Administration which was derived by the actuaries using Department of Labor information on employment and hours of work by U.S. workers and information from the Bureau of Economic Analysis, National Income and Product Accounts. Wages and benefit costs were converted into constant dollars using the GDP deflator.

A substantial part of the answer to Gordons puzzle is that tracking of aggregate compensation is based on the National Income and Product Accounts (NIPA) which includes cash pay for employed workers plus employer contributions for benefits provided to them as part of their total remuneration. The benefit costs include the employers share of payroll taxes for Social Security, Medicare and unemployment insurance plus employer contributions for retirement health benefit plans that they sponsor for their workers. In 2000 under the NIPA annual data series, benefits costs absorbed by employers were 10.6 percent their total compensation costs, rose to 12.5 percent by 2007 and to 13.7 percent by 2009. For the most part, the economic literature and the Census data series on the distribution of earnings and income 9

ignore this element of compensation paid to workers and its distribution across the earnings spectrum. The benefits are sizeable enough that they have the potential to skew any analytical results that ignore them. Most wage and income inequality research has used either the Current Population Survey (CPS), data developed on individuals from a representative sample of households in the United States, or annual federal income tax filing data. Virtually all of the early economic literature that documented the accelerated growth of earnings and income levels for those at the upper end of the spectrum relative to those at lower levels considered only cash wages and salaries as the rewards paid for work. The press articles that describe the growing income disparity phenomenon also focus solely on cash income when considering earnings. In recent years, there has been an evolving series of studies that have incorporated estimates of employers benefits costs in analyzing pay dispersion among workers. Brooks Pierce at the Bureau of Labor Statistics (BLS) has developed two analyses evaluating the earnings inequality phenomenon using survey data collected to produce the BLS Employment Cost Index (ECI). The survey source of the ECI data is a sample of jobs within employing establishments that are surveyed quarterly. The survey solicits information on the pay and benefits costs associated with incumbents in the sampled jobs. In the first study Pierce

developed, covering the period from 1982 through 1996, he found that dispersion measures including employer-financed benefits were larger than those using wages alone.16 In a subsequent effort, he concluded that wage inequality grew over the two decade period from 1987 to 2007 The ECI data also indicate that benefits costs to employers rose more in high than low wage jobs. The differential growth was great enough so that inequality in compensation more broadly defined grew at least as much as did inequality in wages.17

10

Wankyo Chung used survey data from the U.S. Chamber of Commerce estimates of benefits costs as a percentage of pay across private employers to impute benefit costs to workers surveyed in the CPS who reported being covered by a health and/or retirement benefit plan and concluded that ignoring benefits results in exaggerations of inequality among skilled workers but underestimates of the phenomenon among the unskilled.18 This latter conclusion would arise if coverage rates were lower for low earners even if they received proportional benefits to those higher up the earnings distribution. Helen Levy studied the prevalence of employer-sponsored health insurance on various segments of the workforce broken out by gender, race and immigrant status and how it affected the distribution of rewards across the earnings spectrum. She allocated health benefits costs on a constant basis across workers covered by employer-sponsored plans who were in the CPS. She concluded that considering the provision of such insurance narrow[s] the compensation gap (compared to the wage gap) for women, blacks and Hispanics.19 In a recent analysis, Richard Burkhauser and Kosali Simon have imputed the value of health benefits to CPS respondents indicating they were covered by employer-sponsored plans based on estimates derived from the Medical Expenditure Panel Survey Insurance Component which has been conducted each year since 1996. They concluded that, Adding the value of employer health insurance payments to income reduces measured income inequality and reduces increases in measured income inequality over time.20 In a paper that the current authors published in 2010, we distributed reported aggregate health costs incurred by employers as estimated from the annual NIPA data differentiating health costs by whether single, couple or family coverage was being provided based on actuarial estimates of such differentials for typical employer-sponsored plans. In addition, we also distributed the costs associated with Social Security and Medicare payroll taxes paid by employers and estimates of retirement plan costs

11

incurred by employers that skewed the distribution of benefits among those covered by such plans at increasing rates as workers moved up the earnings distribution. We found that employers benefits costs as a share of compensation rose more rapidly for lower earners in the 1980s and 1990s than for those at higher earnings levels with a reversal in that trend between 2000 and 2007.21 One likely explanation for the differences in findings is that the units being analyzed vary from one study to the next. The ECI data that Pierce uses is based on a sampling of one to eight jobs within private employers establishments so the unit of observation is a job. The data is collected on wages and other compensation costs for incumbents in the sampled jobs. Often the respondents cannot break down the various costs at the specific job level and report them on a broader group of workers. In the case of health benefits, there is no controlling for variations in costs associated with single, couple or family coverage. There is considerable variance from one establishment to the next and one job to the next in terms of part- versus full-time workers, take up rates on benefits, and so forth. The remaining studies mentioned here have been based on micro data, primarily from the CPS and augmented with information to allow estimation of costs associated with employerfinanced benefits to specific workers. Chung restricts the analysis to white men ages 25 to 54 who work full-time and full-year in the private sector. Levy includes women and minorities but restricts her analysis to workers employed 35-hours per week or more in the private sector. Burkhauser and Simon adjust reported household income to include the imputed value of health benefits provided by employers, Medicare or Medicaid but then adjust the derived variable based on the number of individuals in the household. Other than Pierce, none of these other analyses include benefits beyond health care. Pierce includes paid time off which has the potential of

12

being double counted, especially for cases where earnings levels are reported on some sort of annualized rate rather than a specific hourly rate for an actual hour of work. In our earlier analysis, we focused on full-time, full-year workers but included those in both private and public sector and allocated employer costs for benefits as described earlier.

Factoring Benefits into the Rewards Picture


The disparate results in the earlier analyses discussed here raise a number of issues. First of all, the size and relative rates of growth in non-wage compensation suggest that employers benefit costs cannot be ignored in considering workers rewards. Second, the role that employersponsored health benefits plays in the compensation story is important, but the studies that have added it alone to cash pay have generally admitted that this component of non-cash rewards is one-third or less of the benefits package. Third, the segmentation of the workforce may be a significant factor in explaining the differences in the results that have been noted. In order to explore the underlying dynamics of what is happening to workers rewards, the following analysis uses the CPS data from surveys done between 1981 and 2010 to distribute workers by their earnings level each year into 10 groups (deciles) ranked by earnings level. The income supplement questionnaires used to gather the income data in the CPS are administered each March and look retrospectively at income over the prior year so the analytical period includes annual income from 1980 through 2009. The top 1 percent of the earnings distribution was excluded in developing the analysis. The Census Bureau has a policy not to disclose any information that has the potential to allow data users to identify specific respondents to its surveys and thus limits the reporting of extraordinarily high income levels reported on the CPS because doing so could make the few people who received the incomes identifiable. In practical terms, the way this is accomplished is 13

by simply coding all earnings responses above a certain level at their break off point in the earnings portion of their survey, a practice referred to as top coding. We eliminated all instances of top-coded earnings by excluding the top 1 percent of reported earnings in each year because we could not determine their actual pay. Given the concerns that some income segments have been getting ahead of others and that higher-income workers have been gaining the most, we segmented the population into 10 income categories. The first nine groups include 10 percent each of the population studied each year, and the tenth group includes the next 9 percent. The top 1 percent of the earnings distribution is of particular interest from a cash income perspective but we do not believe it has such important implications from the perspective of the distribution of the benefits elements of compensation. Thomas Piketty and Emmanuel Saez have analyzed federal income tax filing data from 1913 forward to document the shares of total reported income in that context that has accumulated to the top 10 percent of tax filers and subcomponents of the group including the top 1 percent.22 They found that the share of income reported by the top 1 percent of the tax filers dropped from 15 to 17 percent of the total in the decade prior to World War II to around 11 percent by the end of the war and then declined somewhat lower to around 8 or 9 percent after the war and remained around that level until it started to rise again in the early 1980s climbing back to 16 percent by the late 1990s. Robert Gordon concludes that the sources of rising inequality in the bottom 99 percent of the income distribution and the top 1 percent should be treated as separate topics with separate explanations.23 The top 1 percent of earners clearly receive more than 1 percent of cash salaries but their share of compensation paid in the form of payroll taxes, health benefits or contributions to tax-

14

qualified retirement plans cannot be nearly so skewed. Employers Social Security payroll taxes are capped well below the earnings level of the top 1 percent. The limits on income considered for tax-qualified retirement plans are also well below the top-coded amounts. In most cases, highly compensated individuals are included in the same health benefit plan as other workers in establishments where they are employed. Because of the limitations on the various benefits considered, their role in compensation and employment decisions relative to the top 1 percent of the earnings distribution becomes increasingly trivial as cash earnings levels increase. The employers cost of a group health benefit for a 50-year old executive with a wife and two children covered by a plan will be exactly the same as that for a 50-year old rank-and-file worker with similar characteristics. A health benefit plan costing an employer $10,000 per year for a worker earning a salary of $40,000 per year is a totally different consideration than the same health benefit being provided to an executive who is being paid $2 million per year. Excluding the top 1 percent of the earnings distribution should not distort an analysis of the implications of employer-sponsored benefits on compensation and hiring practices to any substantial degree. Estimating the effects of employer-sponsored benefits on compensation costs required that we impute the costs of these benefits to individual workers in the CPS annual samples. In allocating employers health benefits costs, we varied the imputed compensation to workers based on the size of their employer, as suggested by existing survey evidence to that effect,24 whether individuals were actually participating in a health insurance plan offered by their employer and whether they were provided individual, couple or family coverage.25 In addition, we estimated employers payroll tax contributions for Social Security and Medicare based on workers reported earnings and the contribution rates stipulated in law and added imputed employers contributions for retirement plans as in our earlier work. We ignored the employer

15

costs for unemployment and workers compensation and life insurance in our analysis which are trivial compared to the costs that we have incorporated and would have little practical effect on the results if included. In order to explore the differences between Pierces results suggesting that the addition of employer-sponsored benefits accentuated the disparity in cash pay and its growth over time, we began with an allocation of benefit costs across the total workforce. Figure 2 shows the estimated share of compensation that paid in the form of benefits, on average, within each of the earnings deciles for 1980 and 2009. While the package of benefits that we included in our analysis is somewhat more restricted than Pierces, the pattern of benefits paid to workers as a share of total compensation in the figure has a similar profile to his results. In 1980, benefits were an increasingly larger component of compensation at each higher earnings decile up until the eighth. If pay rates were skewed toward higher earners, this pattern of benefit rewards would have accentuated pay disparities. Between 1980 and 2009, benefits increased more steeply as a share of compensation from the bottom up through the sixth earnings decile suggesting that increasing benefits costs would have also contributed to any increased skewing of rewards toward earners in the middle part of the distribution relative to those at the bottom. Beyond the sixth decile, the growth in benefits costs relative to total compensation gradually trailed off toward the top end of the distribution. While we could not link our results directly to Pierces, his results showed that health benefit costs as a share of compensation reached a plateau around the 30th percentile of the compensation distribution, held constant until around the 60th percentile or so and then declined over the remainder of the distribution. The earnings cap on wages subject to the Social Security payroll tax suggest that the employer costs associated with this benefit also declines at higher

16

pay rates. Pierce found that retirement plan costs increased as a share of compensation fairly steadily across the compensation spectrum from nothing at the bottom of the distribution to about 6 percent of compensation around the 95th percentile of the distribution. At the median of the compensation distribution, health benefit costs were about 2.5 times retirement plan costs and even at the 80th percentile employers health benefit costs were 50 percent larger than retirement benefits costs.26 Given the relative magnitude of employers health benefits and Social Security payroll tax costs, total benefits costs as reflected in Figure 2 are consistent with the results that Pierce presented. He also found that the differences in the compensation growth rates compared to wage growth between 1987 and 2007 were larger between roughly the 20th and 70th percentiles of the compensation distribution than at either end of that distribution27 which is consistent with our results. Figure 2: Share of Total Compensation Paid in the Form of Benefits by Earnings Decile in 1980 and 2009
.
25% 20% 15% 10% 5% 0% 1 2 3 4 5 6 7 8 9 10 1980 2009

Earnings decile
Source: Derived from tabulations of the Current Population Survey augmented by data from the National Income and Product Accounts for various years.

. 17

The results in Figure 2 occur to a large extent because the prevalence of being in an employer-sponsored health or retirement benefit plan is much lower at lower earnings levels than for workers at higher points on the distribution. To a certain degree, this result arises because the offer of benefits at the bottom of the earnings distribution is less common but also because many workers who are offered benefits do not take advantage of them. One consideration in explaining the pattern in Figure 2 is that workers at the bottom end of the earnings distribution tend to work part time to a greater extent than those further up the spectrum. In our summary estimates of hours worked in 1980, 2.6 percent of total hours were worked by those in the bottom earnings decile and 5.1 percent by those in the second. This climbed a bit to 3.2 percent and 6.8 percent of total hours worked in 2009 in the bottom two earnings deciles respectively. In many cases, employers do not extend benefit offers until workers are working some significant threshold of hours, often at least 20 hours per week and then may require a worker contribution that is proportionately larger as a share of hourly wages than for full time workers.28 Table 2 provides considerable detail in regard to the extent to which full-time, full-year workers were covered by health insurance and the sources of that insurance in 1990, 2000 and 2009. The various categories of workers considered as shown in the column headings are not mutually exclusive. Many of the students and pensioners are included in other columns and each of the right-hand four columns of the table is a subset of the all workers column. The latter includes single and married men who are not included in the other columns. The table indicates that even among workers who are fully engaged in full-time work, many are either covered by health insurance from sources other than their own employers or have no health insurance coverage at all.

18

Table 2: Percentage of Full-Time, Full-Year Workers and Health Insurance Status by Source and Year
All workers Married women Single women Students Pensioners Percentage of all full-time, full-year workers with health insurance 1990 2000 2009 88.8% 86.8% 84.8% 94.1% 91.9% 91.9% 85.6% 83.4% 80.4% 79.8% 82.5% 77.2% 94.2% 92.8% 93.5%

Percentage of workers with insurance from their own employer 1990 2000 2009 70.2% 69.0% 65.5% 62.9% 61.3% 61.7% 76.3% 74.7% 70.2% 44.1% 43.0% 35.5% 69.4% 71.4% 70.0%

Percentage of workers with employer insurance as a dependent 1990 2000 2009 9.9% 11.9% 11.8% 24.1% 26.2% 24.2% 0.9% 2.0% 2.1% 18.5% 20.8% 20.2% 8.1% 9.0% 8.5%

Percentage of workers with insurance from non-employer source 1990 2000 2009 8.7% 5.9% 7.5% 7.0% 4.4% 5.9% 8.5% 6.8% 8.1% 17.2% 18.7% 21.5% 16.7% 12.5% 15.0%

Percentage of workers uninsured 1990 2000 2009 11.2% 13.2% 15.2% 5.9% 8.1% 8.1% 14.4% 16.6% 19.6% 20.2% 17.5% 22.8% 5.8% 7.2% 6.5%

Source: Tabulations of the Current Population Survey.

A significant majority of all workers were covered by some sort of insurance over the years shown in Table 2 although the percentages dropped moderately in each column between 1990 and 2009. Other than the students, the majority of each of the classes of workers acquired their health insurance through their own employer although substantial numbers of married women and students acquired their coverage as a dependent of someone else claiming them as a dependent under another employers health benefits program. Substantial numbers of the 19

students and pensioners acquired health insurance from some non-employer sourcemany of the former likely through insurance programs sponsored for students at institutions of higher learning and the latter were likely covered under some form of retiree health benefit program or through insurance that they acquired in the private markets themselves. Substantial numbers of students and single women were uninsured and the percentages of those without generally increased over the 1990 to 2009 period. Table 3 is similar to Table 2 except that it summarizes the results for the same components of the workforce which worked less than full time for the full year referred to in the various rows. Still the vast majority of workers indicated that they were covered by health insurance in the various years but the sources of coverage were dramatically different than for the full-time, full-year workers. For these workers, one-quarter to one-third reported being covered by health insurance through their own employers. Few of the students received health insurance through their own employers but about half of the individuals reporting that they were receiving a pension did so. Nearly half of the students and married women in Table 3 acquired their health insurance in 1990 as a dependent of another worker in their families although the percentage of married women acquiring their health insurance this way dropped off by 2009. One-quarter to one-third of the single women, students and pensioners had health insurance through some non-employer source. For the single women, many likely qualified for public benefits through state-sponsored insurance programs for low-income families. Many of the students again were likely acquiring their health insurance through group programs run by their academic institutions and the pensioners were likely covered by retiree health plans from prior employers, individual insurance or Medicare.

20

Table 3: Percentage of Workers Who Worked Less Than Full-Time and Full-Year and Health Insurance Status by Source and Year
All workers Married women Single women Students Pensioners Percentage of workers who worked less than full-time, full-year with health insurance 1990 2000 2009 76.0% 76.8% 70.3% 88.3% 86.5% 84.4% 72.7% 73.7% 67.7% 85.1% 85.6% 82.2% 86.3% 88.7% 88.9%

Percentage of workers with insurance from their own employer 1990 2000 2009 27.7% 29.8% 26.0% 24.8% 27.9% 26.5% 24.6% 27.8% 22.9% 5.9% 6.8% 6.4% 47.1% 48.4% 45.8%

Percentage of workers with employer insurance as a dependent 1990 2000 2009 27.6% 30.3% 24.6% 48.7% 48.3% 43.4% 16.8% 19.9% 16.7% 47.4% 54.1% 49.2% 14.3% 16.8% 16.7%

Percentage of workers with insurance from non-employer source 1990 2000 2009 20.8% 16.7% 19.7% 14.8% 10.2% 14.5% 31.3% 26.0% 28.2% 31.8% 24.7% 26.7% 24.9% 23.4% 26.4%

Percentage of workers uninsured 1990 2000 2009 24.0% 23.2% 29.7% 11.7% 13.5% 15.6% 27.3% 26.3% 32.3% 14.9% 14.4% 17.8% 13.7% 11.3% 11.1%

Source: Tabulations of the Current Population Survey.

Figure 3 clearly shows that, in the aggregate, the workers employed on a full-time, fullyear basis were consistently more likely to be covered by their own employers health benefit plans across the earnings spectrum than those working less during the year. Even for workers who were working full time, health benefits coverage at lower earnings levels were quite low in the bottom part of the earnings distribution.

21

Figure 3: Percentage of Workers with Health Insurance Provided by Their Own Employers Based on whether They Worked Full or Part Time and Full or Part Year in 2009 by Earnings Decile
Growth rate

90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 FTFY PTFY FTPY PTPY

Earnings decile

Source: Derived from tabulations of the Current Population Survey augmented by data from the National Income and Product Accounts for various years.

Figure 4 shows the prevalence of workers covered by health insurance provided by someone other than their own employer. For full-time, full-year workers at the bottom of the earnings distribution, coverage rates being garnered from sources other than their own employers is about twice those from their own employer. But for the part-time workers in the bottom two earnings deciles, coverage from other sources is five or six times the rates at which the workers are getting coverage through their own employers. There is no dispute that employers contributions for benefits represents a labor cost and that economic theory suggests such costs would have some moderating effect on the cash wage payable to workers. It makes sense then, that workers in households where health insurance is already available might seek out employment where the rewards for work did not include redundant coverage of little added

22

marginal value especially if such redundant coverage has the potential to reduce the wages that such workers might earn. Figure 4: Percentage of Workers with Health Insurance Provided by Someone Other than Their Own Employers Based on Whether They Worked Full or Part Time and Full or Part Year in 2009 by Earnings Decile
Growth rate

80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 FTFY PTFY FTPY PTPY

Earnings decile

Source: Derived from tabulations of the Current Population Survey augmented by data from the National Income and Product Accounts for various years.

Undoubtedly, some of the people who are employed less than full time and full year are underemployed, especially during periods with sluggish economic growth. However, many parttime workers may prefer to work less than full time. For example, students might work full time during their summer vacations and part time during the school year but do not desire or are not able to work full time while carrying a full academic load. Many parents with younger children also coordinate their work outside the home with childcare and homemaking obligations and prefer something less than full-time employment. Many pensioners maintain their attachment to the workforce for economic or social reasons but prefer the flexibility of part-time or seasonal 23

employment. Given the concentrations of part-time workers toward the bottom of the earnings distribution and the fact that many more of them are gaining most of their health insurance through other means than their own employment, we have chosen to focus the remainder of the analysis on the full-time, full-year workers where the tradeoffs between wages and benefits in the compensation package are more straightforward.

Pay and Compensation across the Earnings Spectrum


Figure 5 shows the compound annual growth rates in inflation-adjusted average hourly pay rates across ten pay deciles from 1980 to 1990, 1990 to 2000, and 2000 to 2009 for full-time, full-year workers. In this case, our pay deciles have different income breaks than those reflected in Figures 3 and 4. In the earlier cases, we distributed all workers across the ten categories whereas the analysis in Figure 5 and subsequent discussion redistributes them across nine categories each made up of one-tenth of the of the full-time, full-year workforce with the tenth category having the remaining 9 percent of all such workers with reported earnings below the CPS top coded income amounts. The hourly wage growth patterns vary considerably over time as might be expected given the evidence on pay dispersion over the past few decades. During the 1980s, there was negative wage growth at the bottom of the earnings spectrum, modest but flat growth across the middle-income segments and progressively higher growth across the top 30 percent of the distribution. This was a decade that started with a hard recession and the elimination of many manufacturing jobs. During the decade, there was a realignment of economic activity as imports and global completion increased. This is the period when virtually all analyses of the pay and income dispersion phenomenon agree that the growth in rewards was disproportionately concentrated toward the upper end of the earnings spectrum. In the 1990s there was significant wage growth across all earnings categories but wages grew 24

considerably more at the top than at lower levels. The most significant growth was concentrated in the top earnings decile. During this period, many middle and upper level managers in private firms were included in pay-for-performance plans and, with rapid economic growth during the mid to late-1990s, earners at the top of the distribution did disproportionately well. The rate of growth in pay clearly fell back during the 2000s and was not as flat across the earnings distribution as it had been during the 1990s. Figure 5: Compound Annual Growth Rates of Inflation-Adjusted Hourly Pay for FullTime, Full-Year Workers by Earnings Decile and for Selected Periods
Growth rate

3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% -0.5% 1 2 3 4 5 6 7 Earnings decile 8 9 10 1980-1990 1990-2000 2000-2009

Source: Derived from tabulations of the Current Population Survey, various years.

Considering the net combined effects of slower wage growth after the turn of the century and the higher unemployment, David Leonhardt summarized the effects on median household income in the New York Times. He found that it had fallen from $51,295 in 1998 to $50,303 in 2008 in inflation adjusted dollars. His assessment of what was happening is that: Its a combination of two trends. One, economic growth in the current decade has been slower than in any decade since before World War II. Two, inequality has risen sharply, so much of the bounty from our growth has gone to a relatively small slice of the population.29 25

In order to expand Leonhardts cash income perspective, we estimated employers contributions for payroll taxes and employer-sponsored benefit plans adding them to workers wages. We did this by relying on CPS respondents indications that they were covered by an employer-sponsored health benefit plan and/or a retirement plan. We used the reported aggregate totals for the cost of these programs from the BEAs NIPA files. Employers payroll tax contributions were estimated based on prevailing law governing contribution rates in each year. Employer contributions to retirement plans were based on reported coverage and a progressive structure of benefits across the earnings distribution. We allocated health benefit costs using reported coverage using differential rates for single and family coverage within those earnings levels. In Figure 6, employer costs for retirement plans including employer-sponsored plans and social insurance benefits, as well as those for employer-sponsored health plans, are added to cash pay. While wage growth in the 2000s falls short of that achieved during the 1990s, there was more compensation growth across almost all the earnings spectrum in the early 2000s than in either of the prior two decades. Figure 6 doesnt support the conclusion that it is only the top end benefitting from added productivity in recent years. Those in the eighth earnings decile and above did somewhat better than those at lower levels, but generally workers did much better than the cash only perspective suggests. The results here do not support the conclusion that the 2000s were much worse for middle-America than the prior decade. If people are generally dissatisfied with their cash rewards from work during the 2000s but compensation growth was more robust across most of the earnings spectrum than it had been in the prior 20 years, it is important to understand what the other elements of compensation are and how they are affecting the actual and perceived welfare of workers.

26

Figure 6: Compound Annual Growth Rates of Inflation-Adjusted Hourly Compensation for Full-Time, Full-Year Workers by Earnings Decile and for Selected Periods
Growth rate

3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% -0.5% 1 2 3 4 5 6 7 8 9 10 1980-1990 1990-2000 2000-2009

Earnings decile

Source: Derived from tabulations of the Current Population Survey augmented by data from the National Income and Product Accounts for various years.

Figure 7 shows the compound annual growth rates of employer contributions for Social Security and Medicare. In the early 1970s, the automatic indexation of Social Security benefits made benefit growth particularly susceptible to high inflation at a time when inflation was abnormally high. This resulted in a financing crisis that led to further legislative changes in 1977 fixing the indexing problem and increasing tax rates and the levels of earnings covered under the payroll tax. This legislation proved insufficient to deal with the short-term financing crisis and additional legislation was adopted in 1983 that further increased the payroll tax rates. The legislative remedies to the problem affected high earners somewhat more than middle and lower earners. These increases were largely absorbed by the early-1990s but the extension of the Medicare payroll tax rate to all earnings enacted by the Omnibus Budget Reconciliation Act of 1993 resulted in a higher growth rate for higher-wage workers during the decade. The increases in contribution rates in the 2000s are largely attributable to increases in real earnings levels. 27

Figure 7: Compound Annual Growth Rates of Inflation-Adjusted Hourly Payroll Taxes Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods
Growth rate 5.0%

4.0% 3.0% 2.0% 1.0% 0.0% 1 2 3 4 5 6 7 8 9 10 1980-1990 1990-2000 2000-2009

Earnings decile

Source: Derived from tabulations of the Current Population Survey augmented by data from the National Income and Product Accounts for various years.

The growth rates in employer contributions to their retirement plans are shown in Figure 8. Regulatory policies reduced contributions to retirement during the 1980s and 1990s and the booming financial markets during the latter 1990s also resulted in further reductions of employer contributions to their retirement plans. The saying that the sons have to pay for the sins of their fathers could not ring truer, however, than it does in this case. The long spell of low contributions to the retirement system changed dramatically at the beginning of the new millennium as the maturing of obligations with the aging of the baby boom and the downturn in financial markets necessitated much larger employer contributions. The even steeper financial market declines in 2008 and 2009 will require continuing higher contribution requirements for retirement plan sponsors for several more years.

28

Figure 8: Compound Annual Growth Rates of Inflation-Adjusted Hourly EmployerSponsored Retirement Benefit Costs Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods
Growth rate

8% 6% 4% 2% 0% -2% -4% -6% 1 2 3 4 5 6 7 8 9 10


Earnings decile

1980-1990 1990-2000 2000-2009

Source: Derived from tabulations of the Current Population Survey augmented by data from the National Income and Product Accounts for various years.

The last major benefit cost, employer-provided health insurance for workers and dependents, has grown at higher rates for the duration of the period being analyzed than any other compensation element. As Figure 9 shows, these costs have outpaced inflation over virtually the entire study period. There was a brief respite during the mid-1990s, when the combination of managed care practices and health providers reaction to the Clinton Administrations attempt at health care reform slowed the growth of health benefit costs. They still grew by an average of 2.5 percent more per year than inflation for the bottom two-thirds of the earnings distribution, while wages grew by only about 1 percent per year. By the end of the 1990s, pent-up anger at managed care programs among the provider community, consumers, the media and policymakers forced insurers to curtail various aspects of managed care. In the new millennium, costs escalated again, although not at the furious pace of the 1980s.

29

Figure 9: Compound Annual Growth Rates of Inflation-Adjusted Hourly Health Benefit Costs Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods
Growth rate

6.0% 5.0% 4.0% 1980-1990 3.0% 2.0% 1.0% 0.0% 1 2 3 4 5 6 7 8 9 10


Earnings decile

1990-2000 2000-2009

Source: Derived from tabulations of the Current Population Survey augmented by data from the National Income and Product Accounts for various years.

An aspect of the story not depicted in these figures is that many employers reacted to escalating costs by curtailing or, in some cases, eliminating their plans. Reductions in employerprovided coverage shift more costs to workers. The analysis here simply distributes employer costs for workers benefit plans, as reported in the National Income and Product Accounts, across the covered workforce for each earnings decile based on those workers in the CPS who reported they were covered by benefits programs. For workers who lost employer-sponsored health insurance coverage, such an outcome appears as a cost reduction in Figure 9. The lower growth rate of hourly health benefit costs shown in the figure for workers in the lowest earnings decile for the 2000 to 2009 period compared to the 1990s was driven by a smaller share of the workers at this earnings level having employer-sponsored coverage. The slower growth in hourly health benefit costs for workers in the lowest earnings decile is misleading; it is a result of the transfer of these costs from some workers compensation budgets to their consumption 30

budgets. Workers who lost employer coverage had to acquire and pay for health insurance and retirement protection on their own or do without. When benefit costs grow more rapidly than the compensation budget, wage growth is reduced. The growing share of compensation diverted to benefits, shown in Table 4, explains some of the public consternation about what has been happening to disposable earnings. The sluggish growth in disposable income has been attributed to a variety of things including changing reward structures in the corporate world and tax policy as the focus of many commentaries. Those factors may have played some role in developments but growing benefit costs were likely a much larger reason for the unsatisfactory results many people have had at the pay window in recent years. The underlying factors that have affected the non-wage components of compensation over the past three decades have not played themselves out so these forces will play a continuing role in the rewards picture for the future. Table 4: Share of Compensation Gains Provided in the Form of More Expensive Benefits Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods*
Earnings decile 1 2 3 4 5 6 7 8 9 10 1980-1990* 100.0% 100.0% 90.8% 54.1% 63.9% 43.0% 48.6% 36.8% 29.7% 21.4% 1990-2000 30.4% 23.1% 25.0% 21.3% 17.8% 18.8% 12.4% 9.6% 7.8% 6.8% 2000-2009 35.2% 47.7% 52.3% 60.8% 55.7% 55.3% 54.8% 50.3% 45.0% 37.7%

Source: Updated results of projections presented in Steven A. Nyce and Sylvester J. Schieber, Health Care Inflation, Must Workers Bear the Brunt, Milken Institute Review (Second Quarter 2010), pp. 46-57.
*

Total benefit cost increases in the 1980s for the first and second earnings decile exceeded 100 percent of compensation growth. In both cases, benefit costs increased significantly but total compensation growth was in the negligible first decile and negative in the second.

31

Health Benefit Costs and Labor Markets


Anyone who has ever taken a course on the principles of economics has learned that market prices affect both the supply of goods or services that will be available in the marketplace as well as the demand for them. The equilibrium that is established in a market is determined where the supply and demand curves intersect establishing the point at which suppliers will offer the same quantity of the market item as the demanders are willing to buy. Generally, when economists study how the supply and demand for labor interact, the price they tend to consider as affecting how much labor is offered or hired is the cash wage. By failing to take into account benefits costs associated with hiring labor, such analyses miss some of the important determinants of labor market activities. In recent years, there have been a number of studies that have analyzed the implications of employers provision of benefits on both labor supply and demand decisions. In the current context, we are going to focus solely on the provision of health insurance and how it has potentially been affecting labor market outcomes. On the labor supply side, the availability of employer-sponsored health benefits provides a set of incentives, rewards and costs that vary considerably from one segment of the potential workforce to the next and are affecting who even enters the workforce in some cases, where they work in many others, often influencing decisions on whether to seek and take alternative employment opportunities or to enter retirement late in the working career. On the demand side, the excess inflation that has persisted in health benefits costs in combination with stickiness limiting offsetting reductions in other components of compensation may be affecting how jobs are structured or even offered at some points in the earnings distribution.

32

Labor Supply Effects of Health Care Benefits


The vast majority of working age adults and their dependents in the United States cover a substantial amount of their health needs through health insurance. For most of them, the insurance has not been available through any sort of public health insurance programs. The primary way that most of these people have acquired their health insurance, in recent decades, has been through employer-sponsored health benefit programs. The reasons for this include the relative efficiency of joining a group benefits plan versus buying individual insurance in private markets, the tax incentives favoring these plans and the perceived role employers play in helping to finance the benefits. As a result, these benefits have become engrained in the decisions of workers about all aspects of their workforce choices from what job to take, when or if to leave a position and whether or not to work at all. In fact, when employees are asked about the most

important factors that influence their decision to join an organization, employees consistently respond that health care benefits is one of the most important factors in making that decision behind only base pay and vacation time. For many older workers, health benefits are second only to base pay in ranking of considerations regarding their job choices.30 However, in order to qualify for coverage under an employer-sponsored health plan the individuals covered must have an employment link to the firm sponsoring the plan. In many cases the employment linkage is not direct in that most employer-sponsored plans cover not only workers employed under them but their dependents as well. For single workers, the way to be covered by an employer-sponsored plan is to take a job with the employer. For couples, the potential of being covered under a spouses employer-sponsored insurance creates added flexibility in acquiring health insurance.

33

Jonathan Gruber and Brigitte Madrian have summarized the results of a number of empirical studies completed up to around 2000 that compared the labor force participation rates and hours worked by married women whose husbands had employer-sponsored health insurance to those of married women whose husbands did not have such coverage.31 They cite several studies which estimated that married women who had health insurance under their husbands employer plans had labor force participation rates 6 to 20 percent lower than married women without such coverage. These same studies also generally found that women who were covered by their husbands health insurance were more likely to work part time than their counterparts without health insurance.32 Some more recent studies have raised questions about the results of the earlier analyses because there may be some sorting characteristics between spouses that are unrelated to husbands employer-provided health insurance status that is exaggerating some of the earlier results. Merve Cebi poses the prospect that it is possible that women with strong preferences to be homemakers may tend to marry men who have labor market traitshigh skills, inclination to work long hours, and the likethat lead them to jobs with health insurance coverage. If that is the case, then it is these other factors that account for some of the measured relationship between husbands with health insurance coverage and the lower labor force participation of their wives. When Cebi controls for the sorting and selection among married couples, the estimated reduction in wives labor force participation rates is 7.7 percent compared to 14 percent under the methods used in the earlier studies. Among married working women, Cebis estimate when controlling for the sorting between spouses is that when husbands are covered by health insurance the probability of their wives being employed fulltime is reduced by 15 percent as opposed to 28 percent without such controls. Finally, when controlling for the sorting phenomenon, working wives estimated working hours are reduced by 6.5 percent

34

compared to 13 percent without the controls. While the estimated effects of reduced work outside the home are lower in every case under Cebis approach, the effects are still negative and significant. But the implication of employer-sponsored health benefits on the labor supply decisions of married spouses is just one facet of the issue to consider. Cathy Bradley and several colleagues studied the work responses of employed women who were diagnosed with breast cancer differentiating by whether they were covered by a spouses health insurance or insurance secured through their own employment.33 Here one would have to presume that the sort of selection characteristics that Cebi was concerned about should not arise. The results of Bradley and her colleagues analysis found that six months after diagnosis, women who had their health insurance through a spouse were twice as likely to have quit work as those who acquired their insurance through their own employer. Elements of the labor response persisted through 12 and 18 months. In addition, the results indicated that among women who continued to work after their breast cancer diagnosis, those who had health insurance through their spouse decreased weekly hours worked by considerably greater amounts than did women with health insurance through their own employer.34 It is not just married couples, however, where alternative coverage of health insurance than through ones own employer can be obtained. There are two major federally sponsored programs that may also discourage potential workers from seeking substantial employment or even entering the workforce. Medicaid is a joint federal-state sponsored health insurance program for low-income individuals who qualify for income assistance under either the Temporary Assistance for Needy Families (TANF) or the federally sponsored Supplemental Security Income (SSI) program for needy aged, blind and disabled individuals. The people who qualify for these benefits tend to have relatively limited skills that would command moderate pay

35

and the prospects of finding jobs with expensive health benefits are low. But TANF or SSI eligibility are subject to strict income limits so those qualifying for them face the prospect of losing their Medicaid coverage if they take jobs that pay meaningful income. Likewise, individuals qualifying for Social Security Disability Insurance (DI) benefits also qualify for Medicare if their disability persists for two years or more. Once an individual qualifies for these benefits, substantial earnings lead to the suspension of the Medicare coverage. The TANF program is aimed at low-income single mothers. Brigitte Madrian cites a number of empirical studies that have found little or no effects on the labor supply decisions of potential workers who qualify for these benefits.35 Partly this may be the result of the timelimited nature of TANF benefits meaning that recipients know they ultimately must find alternative income and take the opportunities when they arise. Another consideration is that the dependent children under the program are likely to continue to receive publicly financed health insurance even if their mothers find employment. In the case of the individuals qualifying for Social Security disability benefits under both SSI and SSDI, the Social Security Advisory Board has studied the operations of these programs and found that the prospects of losing health insurance has resulted in a strong reluctance to take jobs once people get on these programs because of the potential loss of the insurance.36 In the case of the cash benefits those qualified are receiving, they drop gradually as earnings rise, but at a certain threshold the health insurance benefits are lost and most people on the programs are not willing to take that risk. The recovery rate for those who qualify for SSDI benefits is quite low historically. Among beneficiaries who entered the program from July 1980 through June 1981, only 2.8 percent of them had their benefits terminated because of work within 10 years of program entry.37 Among those qualifying for SSDI in 1996, 6.5 percent eventually had their benefits suspended due to work

36

over the subsequent 10 years but many returned to the rolls. The average time off the rolls among this group was 42 months. Only 3.7 percent of this group of beneficiaries had their benefits terminated due to work.38 There is no doubt that public health insurance plans covering low-income individuals are expensive and stressing government budgets but most of the workers or potential workers affected by them are not as significant in terms of diminishing the productive capacity of the economy as the loss of comparable numbers of workers who can command higher earnings. One of the more important groups of individuals that includes many potentially higher earners are individuals who are nearing retirement age and qualify for retiree health benefits of one sort or the other. Here the factors potentially affecting the labor supply decisions of older workers may cut one direction at certain ages and then the other subsequently. For individuals who are not offered retiree health insurance by their employers, the high cost of securing coverage in the individual insurance markets may act as a substantial disincentive to retirement until Medicare eligibility is attained at age 65. Some relief from having to go to the individual insurance market for terminating workers is provided under COBRA benefits mandated under the Consolidated Omnibus Budget Reconciliation Act of 1986 which required that employers offer terminating workers the option to buy extended coverage under their plans for up to 18 months at the cost of the insurance to the employer plus a 2 percent surcharge. This means that anyone reaching age 63 has been able to acquire health insurance at group rates as a bridge to Medicare eligibility in recent years. There have been a variety of empirical studies assessing the implications of employersponsored health benefits on retirement behavior of workers. Alan Gustman and Thomas Steinmeier estimated that workers covered by these benefits compared to those not covered

37

tended to delay retirement until they reached eligibility age and then accelerated their retirement patterns once they became eligible for the benefits but that the effects were quite moderate. They concluded the offer of such benefits only reduced male workers retirement age by an average of 1.3 months.39 Lynn Karoly and Jeannette Rogowski focused on workers ages 55 to 62 over a two-year period, in the mid-1980s, and found that those covered by employersponsored retiree health benefits had a predicted probability of retiring of 24 percent during the study period compared to 12 percent by those who did not have such coverage.40 In a subsequent paper using a different data set, the Health and Retirement Study (HRS), and focusing on the early 1990s period, these latter authors estimated that employer-sponsored retiree health benefits increased the probability of early retirement by 4.3 percentage points relative to that of persons with health insurance while working but no employer-sponsored coverage once they left their jobs.41 David Blau and Donna Gilleskie, using the HRS, estimated that among men who were working at age 51 and covered by health insurance, 74.5 percent of them were still employed at ages 55 to 59 if they were not covered by employer-sponsored retiree health benefits compared to 65.8 percent if they had such coverage.42 Jonathan Gruber and Brigitte Madrian have analyzed the effects of continuation of coverage mandates of the sort that are now part of COBRA based on state level mandates during the 1980s. They concluded that the 18-month continuation of coverage requirements under COBRA reduced the labor force participation rates of men between the ages of 55 and 64 by 3.3 percentage points.43 When we look at what has been happening with labor force participation rates in the United States over the last decade for people near retirement age, there is clearly a confluence of factors at play. Over the last decade, labor force participation of workers, ages 55 to 64, increased from 59.2 percent to 64.9 percent and among those 65 and older rates rose from 12.8

38

percent to 17.4 percent. There are numerous factors that explain this pattern including but not limited to public policy changes in our entitlement programs, changes in the private pension system and the volatile performance of the economy, especially the financial markets which have played havoc with many workers retirement savings. Improving life expectancies and the lower incidence of disabling conditions among older workers have also played roles A combination of forces have clearly pushed the workforce in the direction of extending working careers. The implementation of the reform provisions under the Patient Protection and Accountable Care Act (PPACA) of 2010 could lead to a slowing or even a reversal of these workforce patterns. This law calls for the creation of health exchanges to be set up by the states where individuals can buy health insurance without preexisting condition limitations at rates that will tremendously favor older individuals relative to what has been available in individual insurance markets in the past. In addition, for individuals with incomes well into middleincome ranges, the purchase of insurance through these exchanges will be further subsidized. This suggests that many early retirees will likely face health insurance options that are even more economically favorable than what they have been offered under COBRA and may have a substantial effect on labor force participation rates of individuals in their mid-50s and above when fully implemented. The phenomena that older workers might be reluctant to leave a job prior to qualifying for retiree health benefits or the case of workers with a medical condition unwilling to leave a job with health benefits are closely related to a larger phenomenon that has come to be known as job lock. This concept is that workers cling to jobs when they have health insurance because of the risks associated with losing their coverage if they move to alternative employment or into retirement. This is an area that has resulted in many empirical studies on the response of workers

39

to employer-sponsored health insurance. Gruber and Madrian review 18 studies in their 2002 summary of the literature. They conclude that about one-third of the studies find that employersponsored health insurance significantly affect job choice and mobility, one-third find there is no relationship, and the last third find results that vary depending on how the analysis is done. Sorting through all the measurements and other methodological issues, Gruber and Madrian conclude that there is some job-lock, around 10 percent on the lower bound and maybe up to 25 to 35 percent at the upper end.44

Labor Demand Effect of Health Care Benefits


One question that might naturally arise when thinking about how benefit costs impact employers demand for labor is why they should have any effect at all if employers simply pass along the cost of the benefits in the form of lower wages. In part, the answer is that in the short run, wages tend to be sticky, especially in a downward direction, in that actual reductions to workers nominal pay seems to be relatively rare. If an employer cuts workers pay, the company runs the risk of discouraging workers commitment resulting in shirking and turnover. Some workers earning the minimum wage are protected against wage cuts by regulatory fiat. But even at higher levels, the sticky wage phenomenon has been found to persist in a number of empirical analyses.45 If the cash wage that an employer pays a worker tends to be resistant against reductions, then increasing benefits costs have the potential to raise that workers compensation costs above the productivity level. In the earlier discussion about the role of employer-sponsored benefits costs in the compensation package, the focus was on employer-paid payroll taxes for social insurance benefits, their contributions for their retirement pension and savings plans and their costs associated with health benefit plans. In the case of the payroll taxes and retirement plan

40

expenditures, they tend to be variable costs linked to cash wages over time. There may be short time-period exceptions like the recent dramatic increased contributions required to meet unfunded defined benefit obligations, but these should level back out once the underfunding is resolved. Health insurance provided by employers tends to have a different cost structure across the earnings spectrum than other significant benefits. Health insurance benefits taken by the clerical person earning $8.00 per hour in a company creates a cost roughly equivalent to the benefit provided to a worker earning as much as 10 or 20 times that amount or more. In this regard, employers health benefits costs tend to be much more in the nature of fixed costs for the workers who participate in such plans. Because of that, health benefits have the potential to make certain workers uneconomical in some cases. Table 5 shows how employers health benefits costs have risen relative to wages between 1980 and 2009 for workers who are receiving these benefitsthose who were actually contract holders of the policies offeredby their own employers. In 1980, employers costs for those who enrolled in their programs cost single-digits relative to wages for all decile groups except the lowest with the median enrolled employee costing about 6 percent of pay. Over the next

three decades, those costs have grown more than threefold relative to wages and reaching more than a third of individuals wages among the lowest decile groups. In fact, for the lowest decile group they have nearly eclipsed half of an employees take home pay in 2009. Likewise, these costs have been growing at much faster pace for the lowest paid highlighting the greater impact that compounding has on the lower pay groups. For example, health benefit costs relative to wages for the second decile were twice those of workers in the ninth decile in 1980, which by 2009 has risen to more than 3 times.

41

Table 5: Health Benefit Costs Relative to Wages for Full-Time, Full-Year Workers Receiving Health Care Benefits Through Their Own Employer
1980 1 2 3 4 5 6 7 8 9 10 15.4% 9.5% 8.0% 7.2% 6.3% 5.8% 5.4% 4.9% 4.3% 3.2% 1990 30.9% 18.7% 15.3% 13.3% 11.6% 9.9% 9.2% 8.2% 6.9% 4.9% 2000 38.1% 22.9% 18.6% 16.0% 14.0% 12.1% 10.8% 9.2% 7.8% 4.7% 2009 49.5% 30.9% 25.5% 22.3% 19.4% 16.8% 14.8% 12.5% 10.2% 6.3%

Source: Developed by the authors. Consider the case of two workers, one earning $16,000 per year and the other $160,000 where both receive health benefits that cost their employer $6,000. For simplicity, also assume that no other benefits are provided at the employers expense to either of the workers. Assume that both of these workers productivity rises over the year since their pay was last set that would warrant each of their compensation being increased by 1.0 percent and that the employers health benefit premiums increase by 10 percent at the time annual pay reviews are undertaken. Adding health costs to pay, the low earner receives total compensation initially of $22,000 and the high earner $166,000. If their productivity warrants a 1 percent increase in each of their compensation rates, this means the employer can justify an added $220 for the low earner and an added $1,660 for the highly paid worker. But the health benefit plan creates an incremental cost for the employer of $600 for each worker. For the low earner, the added productivity will not cover the added cost whereas for the high earner it will and leave some incremental amount to be passed on as a pay increase.

42

Table 6 shows the decade-to-decade growth rates in worker productivity levels, measured as output per hour, and total hourly compensation, hourly wages and employer contributions for health benefits per hour of labor input. It is clear from the table that employers health benefit costs have been growing more rapidly for a longer period of time than any of the underlying economic factors used to finance them. Even if we assume that low earners productivity rates have been growing at the average rate of the increase in worker productivity as measured by output per hour the significant difference between productivity improvements and the growth in employers health benefit costs per hour suggests the phenomenon just outlined in the example of the two workers has likely been in play for quite some time. The rapid growth in health benefit costs clearly has the potential to make certain workers uneconomical and is likely affecting their employment prospects. Table 6: Compound Annual Growth Rates in Output per Hour, and Hourly Total Compensation, Wages and Health Benefit Costs in Inflation Adjusted Dollars
Output per hour of work 2.5% 2.6 1.5 1.5 1.8 2.0


Compensation 2.6% 2.8 1.3 1.2 2.2 1.4 Wages 2.3% 2.5 0.6 1.1 2.3 1.1

Decade 1950s 1960s 1970s 1980s 1990s 2000-2008

Employers health benefit costs 12.5% 8.9 8.1 4.9 2.8 3.4

Source: Derived from unpublished data from the Office of the Actuary, Social Security Administration which was derived by the actuaries using Department of Labor information on employment and hours of work by U.S. workers and information from the Bureau of Economic Analysis, National Income and Product Accounts. Wages and benefit costs were converted into constant dollars using the GDP deflator.

One group of workers that might be particularly affected by the fixed-cost characteristics of health benefits is the part-timers. If they are being offered the same coverage

43

at equivalent employee premiums as full-time workers, the health benefit costs are generally going to be a much larger share of their total compensation package simply because the fixed costs have to be covered across a shorter work week. In a somewhat dated study of the provision of health benefits to part- and full-time workers based on sample firms used to construct the U.S. Department of Labors Employment Cost Index (ECI), Mark Lettau and Thomas Buchmueller divided firms as integrated if they offered essentially the same health benefit plans to part-time workers as they did to those working full time, or segregated if they had different plans or conditions for joining such plans. In their sample, between two and three times as many of their sample employers were integrated as opposed to segregated. Across the full sample, there was no differential in employers costs for their health benefit plans for full- and part-time workers at the 25th and median levels of the distribution of cost differences but more than $1,200 per year at the 75th percentile. For most firms offering health benefit plans, the cost of providing the benefit does not vary by whether workers are full or part time.46 The Lettau and Buchmueller study is now more than 10 years old and some things have clearly changed since they analyzed the employer provision of health insurance for part-time workers. The offer of health benefits to part-time workers, however, has held relatively steady over the past decade according to the Kaiser-HRET annual survey of employers as reflected in Table 7. This is a survey of a sample of firms drawn from a list of private employers provided by Dun and Bradstreet and from the Census Bureaus Census of Governments list of public employers with three or more workers. To increase precision, the sample is stratified by industry and the number of workers in the firm. The simple fact that employers have continued to offer health benefits and that their costs for these plans has continued to grow more rapidly than

44

worker productivity or compensation suggests that there has been some potential for health inflation to affect part-time workers employment opportunities. Table 7: Among Firms Offering Health Benefits, the Percentage That Offered Them to Part-Time Workers by Firm Size for Selected Years
1999 3-24 workers 25-199 workers 200-999 workers 1,000- 4,999 workers 5,000 workers or more 19% 26 36 53 61 2001 17% 31 42 55 60 2003 24% 29 38 57 57 2005 27% 29 41 50 59 2007 23% 26 37 54 63 2009 31% 27 44 55 60

Source: The Kaiser Family Foundation and Health Research and Education Trust, Employer Health Benefits 2010, found at: http://ehbs.kff.org/pdf/2010/8085.pdf.

Firms offering health benefits to part-time workers or those working full time, for that matter, does not mean that all workers take the opportunity to join their employers plans. In recent years, we know that employers have increased contribution rates for workers who join their plans making even heavily subsidized coverage less affordable for some employees. Many employers have also raised point-of-care cost sharing through higher deductibles and coinsurance rates and increasingly sought to have workers with working spouses avail themselves of their own employer coverage. Some employers now require that their workers complete a health risk assessment in order to qualify for enrollment in their preferred health plan and some are even requiring that employees achieve a certain biometric score to qualify for these plans. In general, there has been a trend toward adding requirements that either discourage or limit participation in health benefit plans, especially the more generous ones. Figure 10 tracks workers participation in their own employers health insurance programsas opposed to being covered by a spouses plan or some other source of coveragebroken out by work intensity. In

45

the figure, full-time (FT) workers are differentiated from those working part time (PT) and there is a further differentiation by whether workers were employed for the full year (FY) or only part of it (PY). Those individuals working full time but only part year (FTPY) were likely in most cases to be in some sort of seasonal job. Generally, participation in employer-sponsored health benefit plans has clearly declined over the period shown in Figure 10 and has continued to do so over the past decade. Figure 10: Percentage of Workers Who Participated in Health Insurance Plans Sponsored by Their Own Employers by Full- or Part-Time Employment Status by Specified Year
90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Percent FTFY Percent PTFY Percent FTPY Percent PTPY

Source: Authors tabulations of the Current Population Survey.

We saw earlier that health benefits costs rose particularly rapidly during the 1980s, the period leading up to the development and implementation of managed care arrangements and the early implementation of plan features shifting costs to workers. Picking 1994, when Lettau and Buchmueller concluded that health benefits were largely a fixed cost for employers, as the

46

benchmark for evaluating further developments, the actual provision of health benefits continued to fall in each of the work categorizations in Figure 10. By 2007, before the effects of the most recent recession took full effect, the share of workers who were working fulltime on a full-year basis who were actually in a health benefit plan sponsored by their own employers was down 6.8 percent relative to 1994. For those working full time for part of the year, the decline was only 4.5 percent. Those in part-time, full-year status were down 9.6 percent and for those who were part-time, part-year the drop had been 12.1 percent. Across most of the workforce, part-time workers are paid at lower hourly rates than their full-time counterparts when controlling for other characteristics of the work that is being done. So part-time workers compared to full-time workers had both fewer hours of work, on average, and lower average earnings rates per hour resulting in much lower total wages against which increasing health benefits costs were being assessed. On top of that, many employers were more aggressive in shifting plan costs directly to part timers than to their full-time workers. In fact, today, 28 percent of employers with at least 1000 employees dont offer health care benefits to part-time employees and 29 percent offer more limited coverage or subsidy than full-time employees.47 The larger share of total wages that part-time workers had to give up to directly secure their health benefits resulted in them reducing their take-up rates of the offer of benefits to a greater extent than those working full time. Despite the evidence that employers were shifting the fixed costs associated with their health benefit plans to their workers, there likely remained a fixed cost element to the benefits that were being provided and especially for workers in the firms that were slowest to change. To the extent there are fixed costs associated with health benefits, there is a strong economic incentive for employers to concentrate work to minimize the incidence of these costs. For

47

example, if an employer had 120 hours of the sort of work that needs to be done each week, it might be done by five workers working 24 hours each or by three workers working 40 hours each. If the hourly cash wage was the same regardless of whether workers were employed full or part time and the costs of health benefits for each worker were equivalent regardless of how many hours were worked each week, it would be significantly cheaper to hire three full-time workers to do the necessary work in this case than by hiring five part-timers. One way to test whether health costs might be affecting employment levels and patterns is to compare time at work for workers with health insurance compared to those who do not have such insurance. If health benefits are a fixed cost associated with hiring a worker, the economics of the situation suggest that the work hours of those covered by the benefits should rise over time compared to those without such benefits in response to rising health costs. David Cutler and Brigitte Madrian analyzed work patterns of individuals with and without health insurance coverage over the period from 1979 to 1992. They estimated that the work week of workers covered by health benefits increased by between 40 and 45 minutes per week relative to those without health insurance.48 Our estimates using the Current Population Survey are that the share of total hours worked by U.S. full-time, full-year workers climbed from 76 percent in 1980 to 84 percent in 2007. There may be other factors than health insurance at play, but both the conceptual effects of health costs and the empirical evidence suggest that they are leading to less demand for part-time workers where benefits are being offered as well as to shifting plan costs toward part-time workers seeking coverage through their employers. Some might not consider the prospect of growing health costs shifting the demand for labor from part-time to full-time workers as a significant problem as long as it is not having any aggregate effects on total employment levels but there is also evidence that the costlier benefits

48

are reducing overall labor demand. The example we used at the beginning of this section, where the effects of increasing health benefits costs on two workers were compared, demonstrated that the underlying arithmetic of health inflation is a bigger threat to lower paid workers than those higher up the earnings distribution because it is more likely to overwhelm the added compensation warranted by growing productivity. Benjamin Sommers developed a test of the potential effects of the combination of sticky cash wages and excessive health inflation on employment levels of workers at lower earnings levels for the period from 2000 to 2001. Sommers noted that between March 2000 and 2001, health insurance premiums grew by 11 percent but that general inflation in the South was 2.3 percent while it was 3.7 percent in the West so inflation-adjusted health costs were growing more rapidly in the former section of the country than the latter. Because of differential general inflation rates in the Southern and Western regions of the United States during that period Sommers surmised that health benefits likely were creating a larger disincentive for employers to keep low-wage workers in the South than the West. To test his hypothesis, he compared the unemployment rates in 2001 of workers with low education levels who were employed in 2000. For workers who were not covered by employer-sponsored health insurance, the probability of being unemployed in 2001were statistically equivalent. For workers covered by employer-sponsored health insurance in 2000, the probability of being unemployed in 2001 were 2.2 percent higher; the result was statistically significant. Sommers also found that the higher growth rates in real health-benefit costs in the South were associated with slower growth in nominal wages and greater shifting of benefits plan costs to workers.49 The results suggest that the labor market is dynamically responding to excessive growth in health benefits costs in a range of ways that many workers undoubtedly consider adverse to their interests.

49

Katherine Baicker and Amitabh Chandra have also developed an analysis of effects of rising health benefits costs on labor markets. They note that the sorts of data available in surveys like the Current Population Survey do not provide health benefits premium data with which to directly compare wages, employment levels and the like and that estimating these for survey respondents based on industry, firm size, and worker characteristics causes certain statistical problems for studying health premium effects on labor market outcomes. To overcome these problems, they use estimates of state-level medical malpractice premiums as a proxy for health premiums in their analysis. The results of this analysis suggest that a 10 percent increase in health insurance premiums reduces the aggregate probability of employment by 1.6 percent and total hours worked by 1 percent. They also find that growing health benefits costs are now shifting work toward part-time workers with higher rates of not being in health benefit plans than among general workers. Finally, they estimate that a 10 percent increase in health benefits premiums leads to a 2.3 percent offsetting decrease in cash wages.50 To put these results into context, taking the excessive rates of growth in employers health benefit costs over the past decade compared to productivity growth, as shown in Table 5, health costs have increased an excess of 20 percent. Extrapolating Baicker and Chandras results, this suggests that growing health benefit costs have added 2 percent to the unemployment rate over the period. None of the empirical evidence regarding the effects of health benefits cost inflation on labor demand or other labor market effects is as clean and straightforward as we might like. Still, all of the evidence supports a logical conclusion that we cannot continue to have excessive inflation in this component of the compensation package without it having significant effects on virtually every aspect of labor market activity. On the labor demand side of the market, it should not be surprising that employers are shifting costs of health benefits to escape the

50

prospect of having to pay workers more than their productivity warrants and restructuring jobs or eliminating them when that is not possible. Rationalizing how employers might respond to this situation suggests that workers with the lowest skillsessentially the ones concentrated toward the lower deciles of the earnings distributionsare the ones most vulnerable to labor market responses to health inflation. They are the ones who have been losing employer-sponsored health coverage, taking pay cuts and losing jobs because of this situation.

The Past as Prologue


The future growth of compensation components that siphon rewards out of workers paychecks will depend on a variety of factors. If the employer-based pension and retirement savings programs continue to operate at current levels, contributions to the systems should moderate considerably in the next four or five years once remaining unfunded pension liabilities are covered by added contributions that are required under law. What happens to the payroll tax will depend on how policymakers address Social Security and Medicare financing. If most of the under financing in these programs is addressed through higher tax rates, the effects will be similar to those recorded in the 1980s. If much of the financing shortfall is addressed by increasing earnings subject to taxation, the effects will be concentrated at higher earnings levels. What happens to payroll tax in coming years will be extremely important, but the real wild card in this deck is what happens to health costs and the implications will likely be quite significant. Table 8 isolates the effects of rising health benefits costs in the context of compensation increases in recent decades. Health inflation has been making a substantial claim on the share of compensation gains workers have realized for many years now. Even in the 1990s, when health inflation seemed to be under some modicum of control, growing health benefits costs claimed more than one-fifth of the added compensation rewards from the bottom half of the earnings 51

distribution. The picture for the most recent decade is uglier than the 1990s but the 1980s show what can happen when health cost inflation really explodes relative to productivity growth. Then, workers at the bottom of the earnings distribution were taking cuts in their paychecks to cover the growing cost of their health benefits. The median worker was giving up three-fourths of the added rewards for increasing productivity to cover health insurance. Table 8: Share of Compensation Gains Provided in the Form of More Expensive Health Benefits Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods*
Earnings decile 1 2 3 4 5 6 7 8 9 10 1980-1990* 285.8% 100.0% 106.9% 57.2% 74.4% 45.2% 55.5% 38.7% 21.4% 12.1% 1990-2000 26.8% 20.8% 23.6% 21.0% 19.8% 22.5% 15.5% 12.1% 9.1% 2.9% 2000-2009 23.6% 30.4% 30.1% 36.5% 28.9% 26.7% 25.8% 20.1% 15.0% 9.1%

Source: Updated results of projections presented in Steven A. Nyce and Sylvester J. Schieber, Health Care Inflation, Must Workers Bear the Brunt, Milken Institute Review (Second Quarter 2010), pp. 46-57.
*

Health benefit cost increases in the 1980s for the second earnings decile exceeded 100 percent of compensation growth. Benefit costs increased significantly but total compensation growth was negative in the second decile.

The reason that health care is such a wild card in the compensation and employment outlook is because no one really knows what the implications of health reform will be on health costs over the next decade or two. The last time the federal government intruded on the health financing system by introducing a major new national program was in the mid-1960s when Medicare was implemented. There was not much to go on then either in terms of estimating what costs would be under the program. In the decade prior to the adoption of the HI program, 52

hospital costs had been rising about 3 percentage points faster per year than covered wages.51 An Advisory Council on Social Security financing met during 1963 and 1964 to consider these trends and determine what assumption to use in projecting the new HI program costs. The Advisory Council proposed assumptions for the initial projections: hospital costs would rise by 2.7 percent more than wages over the first five years of the programs operations, then trend down to the wage growth rate over the next five years and for all subsequent years. 52 As it turned out, over the first quarter century of the Medicare HI programs operations, the average covered wage subject to the payroll tax grew at an average compound rate of 6.2 percent per year while average daily hospital costs rose at a compound rate of 11.9 percent per year. Over the last 10 years of that period when it was conservatively assumed that hospital costs would grow at the same rate as wages the growth rate in daily hospital costs was outpacing wage growth by 4.5 percentage points per year.53 A second major variable in determining actual HI cost rates was the hospital utilization rate. Estimated utilization rates were based on the 1957 Survey of Beneficiaries conducted by Social Security.54 In making the projections, Robert Myers argued that utilization rates were most likely to conform to a low-cost estimate, at least during the early years of the program, to give recognition to the possibility of success of current efforts for progressive patient care, for reductions in hospitalization costs resulting from development of outpatient hospital diagnostic facilities, and for progressive cost-reducing trends in medical practice. 55 In the final cost projections, the Ways and Means Committee had Myers use higher utilization rates than those used for his original estimates. The increase in the early-year utilization assumption was about 20 percent under the more conservative assumptions. The use of the high-cost utilization rates in later years was considered a safety factor. As the program was implemented, over the

53

first 17 years or so, utilization levels consistently ran some 20 percent higher than even the more conservative Ways and Means Committee assumptions. The estimates that had been repeatedly characterized as conservative turned out to be excessively optimistic. In addition to higher than expected inflation and greater than expected utilization of services, the expanding protections offered through HI became a third contributing factor in cost inflation. In 1972, all those who had received disability benefits for 24 consecutive months under the Social Security Disability Insurance (DI) program or the Railroad Retirement program became eligible for coverage. At the same time, HI benefits were also made available to those younger than 65 with end-stage renal disease who were insured under the Social Security or receiving a DI benefit. By 1975, the number of days of HI-covered care provided to this new group was approaching 10 percent of the elderly caseload. By 1983, the disabled and end-stage renal covered days of care under the HI program were 16 percent of the elderly caseload. The underestimated costs for Medicares HI program were not simply additive they compounded each other. If reality had lived down to expectations, the cost of the HI program in 1990 would have been less than half of what it was. Health cost inflation stretched well beyond the financing of Medicare. It also affected the cost of health insurance benefits that employers were providing to workers. Major policy changes that expand insurance coverage of large segments of the population, change financing incentives, payment mechanisms and the like will almost certainly affect health care pricing and utilization patterns. There is certainly the possibility that the new health reform law will turn out to have a similar set of unanticipated costs as Medicare did, especially when we consider that its major proponents and architects admit that we have to reengineer the delivery system for it to deliver the sort of cost savings they anticipate. Since health care insurance financing is such an important element of the

54

compensation package that the majority of workers receive, health reform will likely continue to play a central role in determining workers employment and wage outcomes.

The Prologue Is almost Past


It is likely that policymakers will continue to debate the efficacy of health reform but the larger threat to implementation may be in the courts. Regardless of whether health reform as enacted in 2010 is sustained, neither our government nor our workers can prosper with the rates of health cost growth we have experienced over the past 40 years. Prior to the passage of the ACA, the CBO had estimated that in 2030, health care spending in the United States would reach 29 percent of GDP;56 in 2009, it was 17.6 percent.57 Peter Orszag and Ezekiel Emmanuel, two of the architects of the reform package within the Obama Administration, wrote of the measure in the New England Journal of Medicine, Yet we would argue that even from a purely green eyeshade viewpoint, the bill will significantly reduce costs. Projections suggest that with reform, total health care expenditures as a percentage of the gross domestic product will be 0.5% lower in 2030 than they would otherwise have been. Undoubtedly, a half percentage point of GDP in 2030 will be a large number but this estimate suggests that health reform will slow the rate of conversion of GDP from spending on other things to health care from an annual rate of around 2.6 percent per year to 2.5 percent. This doesnt look like relief and suggests workers paychecks will continue to suffer from excessive inflation in their health benefits costs. John Cogan, Glenn Hubbard and Daniel Kessler evaluated insurance premium rates in Massachusetts around the implementation of their health reform law in 2007 because that law is considered to be similar to the national reforms embedded in the PPACA. The analysis covered data over the period 2004 to 2008 and compared premium increases in Massachusetts to rising premiums around the country. The 2004 through 2006 period served as a control to determine if 55

Massachusetts had different trends prior to the implementation of health reform compared to other areas. For provision of individual coverage, the authors concluded the Massachusetts reform increased employer premiums by about 6 percent compared to elsewhere. In the family coverage case, the differential was only 1.5 percent across the state but much higher in the Boston area compared to other metropolitan areas around the country and for small employers.58 If the Massachusetts findings play out nationally for an extended period of time, it is likely the conjecturing by Peter Orszag and Ezekiel Emmanuel that health reform will cut costs will come as close to reality as the Ways and Means projections on Medicare costs turned out. A problem with the extension of recent health benefit cost trends is that they are now indexing against a much larger base of current costs than 30 years ago. Table 9 shows the estimated size of employer contributions to health benefits at each earnings decile compared to the wages paid to workers in selected past years. For example, an average worker in the third earnings decile in 1990 received health benefits costing 7.9 percent of compensation but 20 years later this had grown to 11.5 percent of compensation. If employers health benefits inflation persists at 2 percent more than compensation growth over the next 20 years, an additional 5.6 percent of compensation will be displaced. If the excess health benefits inflation jumps to 4 percent, an additional 13.7 percent of compensation will be displaced by health benefits over the next 20 years starting in 2009. For a worker at the seventh decile, the effects are not quite as pronounced but with 2 percent excess health inflation, the extra displacement for health care would be 4.4 percent of compensation over the next 20 years and 10.7 percent at 4.0 percent excess health inflation. The ultimate problem, however, is not exactly how much of compensation is being utilized to finance health benefits, it is whether the growth in these costs per year leaves any additional productivity reward for workers to take home in their paycheck.

56

Table 9: Health Benefits Paid by Employers as a Percentage of Compensation Paid to FullYear Workers by Earnings Decile and for Selected Years
Decile 1 2 3 4 5 6 7 8 9 10 1980 5.8% 5.3% 5.0% 4.8% 4.4% 4.2% 3.9% 3.8% 3.2% 2.4% 1990 7.9% 8.1% 7.9% 7.6% 7.2% 6.3% 6.1% 5.6% 4.8% 3.4% 2000 10.0% 9.4% 9.6% 8.8% 8.3% 7.7% 6.9% 6.2% 5.3% 3.3% 2009 11.6% 11.4% 11.5% 11.4% 10.4% 9.9% 8.9% 7.8% 6.5% 4.1%

Source: Estimated by the authors as described in the text. Notes: Compensation is based on cash pay and employers payments for retirement, health care and payroll taxes (i.e., OASDHI benefits).

The results in Table 9 are remarkably different than those shown in Table 5 and may lead to the conclusion that they are inconsistent. There are two important differences in the results in the two tables. In deriving Table 5, health benefit costs were compared to wages and Table 9 considers health benefit costs relative to compensation. More importantly, Table 5 was derived by considering the relative costs of health benefits compared to wages of workers who actually were enrolled in their own employers health benefit program. In the lower income deciles, the majority of workers do not participate in employer-sponsored health insurance so the average cost of a benefit plan that ranges from $7,000 to $13,000 per year, depending on whether it is single or family coverage, is colossal relative to pay of $7.50 per hour. In deriving Table 9, the total cost of all people in each earnings decile who take health benefits is averaged across all workers in the particular decile, whether they joined their employers health benefit plan or not. As such, the costs are distributed across a much broader population in each decile. The net effect

57

is much greater in the lower deciles than further up the earnings distribution because the actual enrollment rates are so much lower at the bottom of the earnings distribution. As health benefits become an ever larger share of compensation, more and more workers are going to find themselves in the situation that we described earlier for low-wage workers whose health benefits costs were outstripping productivity improvements. This is especially true if employer-financed health insurance persists and continues to experience excessive inflation relative to productivity growth rates.

Health Care Dominates Our Prospects


In order to understand the implications of current and proposed policies for the distribution of productivity rewards in the future, we projected earnings and benefit cost levels to 2030 at 10 earnings levels corresponding to the decile levels of earnings analyzed earlier. In the following discussion, we report the share of productivity gains that will be absorbed by health care benefits in a range of four alternative scenarios. The initial scenario projection assumes that employer-sponsored health benefit costs grow at the same rate as compensation plus 1.5 percentage points per year less than half of recent trend rates and that current employer-sponsored health insurance coverage rates persist. In the second scenario, we assume a more realistic prospect that recent health benefit cost inflation trend rates persist throughout the entire period with coverage rates the same as the baseline. In the third scenario, we assume that health care inflation rates accelerate in line with the notion that the potential added insurance coverage under PPACA may lead to an escalation in health costs somewhat in line with what was experienced over the first couple of decades after Medicare was implemented. Finally, in scenario four, we assume that all the uninsured are covered in the future, and the cost of their insurance is equivalent to the cost of existing coverage 58

provided to workers, adjusted for family composition. This scenario considers the maximum coverage effects that health reform can possibly have as it is implemented. In this scenario, health costs increase at recent historical rates and employer-sponsored health insurance coverage rates expands to cover all workers. In all the scenarios, we assume real compensation grows at 1.52 percent annually from 2009 to 2030, consistent with the 1990 to 2009 period. The first scenario might be characterized as a we wish perspective on future health inflation. As a benchmark, we did not assume any expansions in insurance coverage as a result of health care reform. Under this scenario, we project that compensation will grow over the projection period at a rate of 1.5 percent per year and that health benefit costs will grow at 3.0 percent per year. The health cost differential compared to compensation growth in this projection is about half the rate that has persisted over the past decade in our analysis of fulltime, full-year workers presented earlier. This health benefits differential compared to compensation growth is 0.5 percentage points higher than suggested by Table 5, but that result is for the whole workforce including part-time workers who have completely different coverage and participation characteristics than their full-time counterparts. There are some optimists who believe that health care reform might lead to substantial amelioration in the excessive inflation rates that have persisted in health costs for several decades and this scenario would reflect that potential outcome. There are probably far more skeptics that this scenario might come to pass than optimists, but it is still an important scenario to consider because it shows how much productivity growth would be consumed by health care benefit costs over the next couple of decades in what is likely to be the best possible circumstances. The results of the projections in this scenario are shown in Table 9, broken into three period summaries. In this particular scenario, there is not a great deal of variation from one

59

period summary to the next. In the bottom half of the earnings distribution, something slightly less than one quarter of the projected compensation gains would be consumed by health care benefits during the 2009 to 2015 period and something slightly more than one-quarter over the 2015 to 2030 period. The effects taper off on up the earnings distribution. The important aspect of the results presented in Table 10 is that if we can bring health care inflation down to rates that have really not persisted for any substantial duration over the past half century, we will still be diverting up to a quarter of workers added rewards for improving productivity to higher health benefit costs over the next couple of decades. Our appetites for consuming health care have been profound for quite a long time and we have hidden many of the ramifications by financing much of it in ways where workers have not directly seen the costs. Now it is threatening their very prosperityand this is a scenario less plausible than the others we considered. Many people will judge this best we can do scenario as unsatisfactory. Table 10: Scenario 1, Share of Compensation Gains Provided in the Form of More Expensive Health Benefits Paid by Employers Where Health Cost Inflation Rates Grow at Productivity Growth Plus 1.5 Percentage Points per Year
Projection Periods Earnings decile 1 2 3 4 5 6 7 8 9 10 --------------------------------------------------------2009 to 2015 to 2009 to 2015 2030 2030 ---------------------------------------------24.0% 28.1% 27.0% 23.6% 27.6% 26.6% 23.6% 27.7% 26.7% 23.5% 27.5% 26.5% 21.5% 25.2% 24.3% 20.3% 23.8% 22.9% 18.3% 21.5% 20.7% 16.1% 18.8% 18.1% 13.4% 15.7% 15.1% 8.5% 10.0% 9.6%

Source: Developed by the authors as described in the text.

60

The second scenario was developed assuming that employers health benefit costs growth will continue at rates that have persisted over the past decade. Thus, in this projection, health benefits costs rise at a rate that is 3.2 percentage points more per year than the rate of growth in total compensation; roughly equivalent to a 7 percent increase per year in employers health costs per employee. Once again, this projection assumes that pre-health reform insurance coverage rates will persist. In this case, the projections suggest that health costs will progressively consume increasing amounts of the added compensation rewards workers receive for their improving productivity in future years. Given the clamor in some circles to repeal health reform, this projection might be characterized as the business as usual scenario because we are assuming that recent coverage rates and benefits inflation rates persist into the future. There is nothing about the world prior to health reform that suggests we will realize much lower health cost inflation if we revert to prior policy. In the 2009 to 2015 period, health costs are projected to consume between one-third and 40 percent of workers added compensation over the bottom half of the earnings distribution as shown in Table 11. In the 2015 to 2030 period, the excessive health inflation on an ever growing base is projected to consume more than half of growing compensation for workers in the lower half of the distribution. Under this scenario, workers in the sixth through eighth deciles are projected to be losing compensation growth to more expensive health benefits in the 2015 to 2030 period somewhat faster than the lower wage workers during the 2009 to 2015 period. The health benefits inflation problem has the potential, at current rates, to have significant dampening effects on workers disposable income well up in the earnings distribution within the next few years if it cannot be brought under control. The pre-health care reform status quo is potentially a real problem for many workers who have health benefits.

61

Table 11: Scenario 2, Share of Compensation Gains Provided in the Form of More Expensive Health Benefits Paid by Employers for Full-Year Workers by Earnings Decile and for Selected Periods Where Health Cost Inflation Rates Persist at Current Rates
Projection Periods Earnings decile All 1 2 3 4 5 6 7 8 9 10 --------------------------------------------------------2009 to 2015 to 2009 to 2015 2030 2030 ---------------------------------------------24.9% 35.0% 32.4% 39.1% 38.4% 38.5% 38.3% 35.1% 33.1% 29.9% 26.2% 21.8% 13.9% 54.9% 54.0% 54.2% 53.9% 49.3% 46.6% 42.0% 36.9% 30.7% 19.5% 50.9% 50.1% 50.2% 49.9% 45.7% 43.2% 38.9% 34.2% 28.5% 18.0%

Source: Developed by the authors as described in the text.

The third scenario is based on the assumption that recent health insurance benefit coverage rates persist but that health inflation jumps to 6 percent per year in excess of the compensation growth rate. This projection might be labeled the we fear scenario because it reflects what might happen to health benefit costs if health reform exacerbates health inflation rates. While this scenario does not include the potential coverage expansionary effects of health reform, it isolates the potential health inflation effects on workers compensation. In this case, Table 12 suggests that health inflation could consume up to two-thirds of the added rewards paid to workers in the bottom four earnings deciles between 2009 and 2015 but would take well over half the potential reward for those in the fifth through seventh deciles during the period. In the 2015 to 2030 period, health benefits inflation would be consuming more than 100 percent of 62

compensation growth for the bottom six deciles of the earnings distribution and nearly that for the seventh decile. Given the earlier summary of the research on the effects of health benefits costs growth on the demand for workers, this scenario suggests that a significant increase in recent health benefits inflation rates will either be driving many more workers out of employersponsored health coverage or resulting in the elimination of their jobs. Table 12: Scenario 3, Share of Compensation Gains Provided in the Form of More Expensive Health Benefits Paid by Employers Where Health Cost Inflation Rates Increase to Productivity Growth Plus Six Percentage Points per Year
Projection Periods Earnings decile All 1 2 3 4 5 6 7 8 9 10 --------------------------------------------------------2009 to 2015 to 2009 to 2015 2030 2030 ---------------------------------------------42.6% 80.9% 71.2% 66.8% 65.7% 65.8% 65.5% 60.0% 56.6% 51.1% 44.8% 37.3% 23.7% 127.0% 124.8% 125.1% 124.5% 114.0% 107.7% 97.0% 85.2% 70.9% 45.0% 111.7% 109.8% 110.1% 109.5% 100.2% 94.7% 85.4% 74.9% 62.4% 39.6%

Source: Developed by the authors as described in the text.

Herb Stein, a venerable Washington economist who passed away in 1999, used to observe of health inflation that what cannot go on forever has a tendency to stop. Several years after he initially made the observation, someone pointed out to him that the excessive health inflation rates had persisted long enough that it was putting his observation in question. Stein responded that the observation was still true that excessive health inflation would 63

ultimately have to subside but that economists simply did not know when that might be. If the third scenario developed here were to play out, we might find out fairly soon whether the declining prosperity of a broad swath of American workers will finally bring excessive health inflation to its knees. Even if health inflation persists at recent levels, it will simply take longer to get to the results reflected in Table 12. But the ultimate result of extra compounding of an element of compensation at rates faster than compensation is growing itself has to be crowding out other elements of the compensation package. In the fourth scenario, the projections assume health benefits cost inflation persists at the same rates as we have experienced over the past decade. In this case, however, we assume that employer-provided health insurance coverage expands to the whole workforce. In this case, not only would recent health inflation eat away at cash wages in the compensation package, but expanded coverage would eat away further at aggregate rewards other than health benefits paid to workers. In this scenario, the average costs of benefits for those who have been participating in their employers health benefit plans would extend to all the workers who had been outside employer coverage prior to health reform. Because the expansion in coverage would be so much more dramatic in the lower ends of the earnings distribution, the differences in Table 13 and Table 11 are more dramatic for lower earners. Undoubtedly many workers who wanted coverage in the past would welcome the prospect of the protections an employer would provide but it is not clear that they would be any more likely to pay the price for obtaining the newly available coverage than they have often been in the past if the price is as high as this scenarios results suggest. Once again, the results portend that we are at risk of health inflation stealing from our productivity rewards at rates that will leave little for other aspects of the compensation package for many workers.

64

Table 13: Scenario 4, Share of Compensation Gains Provided in the Form of More Expensive Health Benefits Assuming Expanded Coverage Under PPACA is Paid by Employers Where Health Cost Inflation Rates Persist at Current Rates
Projection Periods Earnings decile All 1 2 3 4 5 6 7 8 9 10 --------------------------------------------------------2009 to 2015 to 2009 to 2015 2030 2030 ---------------------------------------------45.2% 41.9% 42.8% 272.7% 136.8% 94.5% 71.2% 58.5% 47.1% 40.6% 32.2% 26.3% 16.4% 134.4% 87.5% 73.2% 65.0% 57.3% 51.4% 45.6% 38.9% 32.2% 20.3% 169.5% 100.0% 78.6% 66.6% 57.6% 50.3% 44.3% 37.2% 30.7% 19.3%

Source: Developed by the authors as described in the text.

Conclusion
No one knows for certain what the implications of health reform will be for U.S. workers in terms of their future health costsor even how they will acquire their health insurance coverage in the future. The analysis of what has occurred over the past three decades suggests that a considerable share of the disappointment with the rewards that many workers have received in recent years is due to the voracious appetite that health benefits have brought to bear on their productivity rewards. The extension of the trends on health benefit cost growth that have persisted for decades now suggests that if we cannot bring excessive health care inflation under control, workers prosperity is going to be increasingly threatened. A full-time worker in the second earnings decile in 2009 earned around $25,000 in total compensation on average. If 65

his or her productivity goes up by the rate of growth that the Social Security actuaries estimate, by 2019 this worker will be earning around $36,600 in total compensation but nearly 75 percent of the difference from 2009 will have been consumed by rising health benefit costs. If the worker is being provided family coverage, the cost of health benefits will grow to consume all of the added productivity contribution. Do we really think that health cost reform is going to restrain health inflation? Even Peter Orszag, who was the Director of the Office of Management and Budget and a major architect of the health reform package, and Ezekiel J. Emanuel who was a special advisor to the White House and OMB during the reform development suggest that under the new reform model, total health expenditures in the United States in 2030 will only be 0.50 percent less as a share of GDP than under prior law.59 One of the goals of health reform is to broadly expand the coverage of the population under health insurance programs. In a scenario where we assume that future health costs grow at the rate they have been growing since the turn of the century and where we assume that all workers take up health insurance in the context of employer-sponsored plans, our results are sobering. In this case, not only is recent health inflation eating away at cash wages in the compensation package, but expanded coverage will eat away further at aggregate rewards other than health benefits paid to workers. Because the expansion in coverage would be so much more dramatic in the lower ends of the earnings distribution, the growth in the rate of compensation absorption is more dramatic for lower earners. Once again, the results portend that we are at risk of health inflation stealing from our productivity rewards at rates that will leave little for other aspects of the compensation package. The stark result of the sort of scenario presented in Table 13 is that employers simply cannot offer many workers the benefits implied and pay them a wage that might grow over time

66

beyond that and remain competitive in a global economy. The potential subsidization of health insurance under the health care reform law for workers who acquire it outside their employers suggests the economics of offering coverage will change significantly. Many employers, particularly in low-wage industries, will likely eliminate their plans and let workers fend for themselves in the new health exchange marketplace. At the margin, this sort of outcome may work and may even be desirable but we cannot avoid the fact that we are facing a national marketplace here. The mere shifting of health insurance costs from employers compensation packages to a mix of public subsidy and workers contributions out of their remaining disposable wages will not reduce national costs unless we bring excessive inflation under control. If health reform does not reduce the rate of growth in our costs, we must ask who is going to pay the bill? Our projections do not include the potential response that policymakers will ultimately bring to bear on the underfunding of Social Security and Medicare that now persists although we have considered the effects of these in a separate analysis.60 The Congressional Budget Office projects that the cost of federal entitlement programs will rise from 10 percent of GDP in 2010 to 16 percent of GDP by 2030. Ultimately these costs will have to be absorbed by the productivity of our economysuggesting that at least two-thirds of the bill will come out of workers pockets if costs are distributed in accordance with the way our national output is distributed. The workers costs to cover these federal expenditures may not come out of their compensation but they will still be extracted from their paychecks in the form of higher taxes. The potential of these costs hanging over our heads with seemingly little inclination on the part of policymakers to address them suggests that we have extremely little margin for any added subtractions from future workers compensation if we wish to pass on to them the American Dream that each generation is given a chance to better itself compared to those that have come before.

67

Endnotes
Robert J. Gordon, Misperceptions about the Magnitude and Timing of Changes in American Income Inequality, (Cambridge, MA: National Bureau of Economic Research, 2009), NBER Working Paper 15331, Table 1. Computed from U.S. Department of Commerce, Bureau of Economic Analysis, National Income and Product Accounts found at: http://www.bea.gov/national/nipaweb/index.asp. Katherine Baicker and Amitabh Chandra, The Labor Market Effects of rising Health Insurance Premiums, NBER Working Paper No. 11160(Cambridge, MA: National Bureau of Economic Research, 2005). Peter R. Orszag and Ezekiel J. Emanuel, Health Care Reform and Cost Control, New England Journal of Medicine (June 16, 2010), found at: http://healthpolicyandreform.nejm.org/?p=3564. Steven A. Nyce and Sylvester J. Schieber, Health Care Inflation, Must Workers Bear the Brunt? The Milken Institute Review (Second quarter 2010), pp. 46-57. Edward Lotterman, Stagnant Incomes, Restless Voters, (September 13, 2010), found at: www.bismarktribune.com/news/columnists/ed-;ptter,am/article_99eb3092-bc2e-1. U.S. Census Bureau, Income, Poverty and Health Insurance Coverage in the United States: 2008, found at: www.census.gov/prod/2009pubs/p60-236.pdf.
8 7 6 5 4 3 2 1

Edward Lotterman, Stagnant Incomes, Restless Voters.

Robert J. Gordon, Misperceptions about the Magnitude and Timing of Changes in American Income Inequality, (Cambridge, MA: National Bureau of Economic Research, 2009), NBER Working Paper 15331.
10

Ibid, Table 1.

Claudia Goldin and Robert A. Margo, The Great Compression: The Wage Structure in the United States at Mid-Century, Quarterly Journal of Economics (February 1992), vol. 107, pp. 1-34. Henry J. Aaron, Politics and the Professors: The Great Society in Perspective (Washington, DC: Brookings Institution, 1978). Peter Gottschalk and Sheldon Danziger, Inequality of Wage Rates, Earnings and Family Income in the United States, 1975-2002, Review of Income and Wealth (2005), series 51, no. 2, pp. 231-254. Robert J. Gordon and Ian Dew-Becker, Controversies about the Rise of American Inequality, NBER Working Paper No. 13982 (Cambridge, Ma: National Bureau of Economic Research, 2008). Richard V. Burkhauser, Suaizhang Feng, Stephen P. Jenkins and Jeff Larrimore, Estimating Trends in US Income Inequality using the Current Population Survey: The Importance of Controlling for Censoring, NBER Working Paper No. 14247 (Cambridge, Ma: National Bureau of Economic Research, 2008), p.29.
16 15 14 13 12

11

Brooks Pierce, Compensation Inequality, The Quarterly Journal of Economics (2001), pp. 1493-1525.

Brooks Pierce, Recent Trends in Compensation Inequality, U.S. Department of Labor, Bureau of Labor Statistics (July 2008), p. 2.

17

68

Wankyo Chung, Fringe Benefits and Inequality in the Labor Market, Economic Inquiry (July 2003), vol. 41, no. 3, pp. 517-529. Helen Levy, Health Insurance and the Wage Gap, NBER Working Paper No. 11975 (Cambridge, MA: National Bureau of Economic Research, 2006), p. 17. Richard V. Burkhauser and Kosali I. Simon, Measuring the Impact of Health Insurance on Levels and Trends in Inequality, NBER Working Paper No. 15811 (Cambridge, MA: National Bureau of Economic Research, 2010). Steven A. Nyce and Sylvester J. Schieber, Health Care Inflation, Must Workers Bear the Brunt? The Milken Institute Review (Second quarter 2010), pp. 46-57. Thomas Piketty and Emmanuel Saez, Income Inequality in the United States, 1913-1998, Quarterly Journal of Economics (February 2003), vol. 118, no. 1, pp. 1-39. Robert J. Gordon, Misperceptions about the Magnitude and Timing of Changes in American Income Inequality, p. 21. Kaiser Family Foundation and Health Research & Education Trust, Employer Health Benefits 2010 (The Henry J. Kaiser Family Foundation and HRET, 2010), p. 80, found at: http://ehbs.kff.org/pdf/2010/8085.pdf. Aggregate employer contributions to health care coverage are allocated on the basis of multiples of single only coverage. Contact holders with employee plus spouse or child coverage are allocated at two times single only coverage and those with employee plus children or family coverage are allocated at three times single only coverage. Brooks Pierce, Recent Trends in Compensation Inequality, U.S. Department of Labor, Bureau of Labor Statistics (July 2008), p. 42.
27 26 25 24 23 22 21 20 19

18

Ibid, p. 43.

See Shaping Health Care Strategy in a Post-Reform Environment, 16th Annual Towers Watson/National Business Group on Health Employer Survey on Purchasing Value in Health Care, 2011. David Leonhardt, A Decade with No Income Gains, New York Times (September 10, 2009), found at: http://economix.blogs.nytimes.com/2009/09/10/a-decade-with-no-income-gain/. Towers Watsons Retirement Attitudes Survey, Part III: Attraction and Retention. http://www.towerswatson.com/united-states/research/2717. Jonathan Gruber and Brigitte C. Madrian, Health Insurance, Labor Supply, and Job Mobility: A Critical Review of the Literature, NBER Working Paper No. 8817 (Cambridge, MA: National Bureau of Economic Research, 2002). Thomas C. Buchmueller and Robert G. Valletta, The Effect of Health Insurance on Married Female Labor Supply, Journal of Human Resources (199), vol. 34, no. 11, pp 42-70; Craig A. Olson, A Comparison of Parametric and Semiparametric Estimates of the Effect of Spousal Health Insurance Coverage on Weekly Hours Worked by Wives, Journal of Applied Econometrics (1998), vol. 13, no. 5, pp. 543-565; Barbara Steinberg Schone and Jessica Primoff Vistnes, The Relationship between Health Insurance and Labor Force Decisions: An Analysis of Married Women, (2000), unpublished working paper, U.S. Department of Health and Human Services, Agency for Healthcare Research and Quality; and Allison J. Wellington and Deborah A. Cobb-Clark, The Labor-Supply Effects of Universal Health Coverage: What Can We Learn From Individuals with Spousal Coverage?, in Simon
32 31 30 29

28

69

W. Polachek, ed., Worker Well-Being: Research in Labor Economics, Volume 19 (Elsevier Science: Amsterdam, 2000). Cathy J. Bradley, David Neumark, Zhehui Luo and Healther L. Bednarek, Employment-Contingent Health Insurance, Illness, and Labor Supply of Women: Evidence from Married Women with Breast Cancer, NBER Working Paper No. 11304 (Cambridge, MA: National Bureau of Economic Research, 2005).
34 33

Ibid, p. 20.

Brigitte C. Madrian, The U.S. Health Care System and Labor Markets, NBER Working Paper No. 19801 (Cambridge, Ma: National Bureau of Economic Research, 2006), p 13. Social Security Advisory Board, A Disability Program for the 21st Century (2006), p. 24, found at: http://www.ssab.gov/documents/disability-system-21st.pdf. Scott, L. Muller, Disability Beneficiaries Who Work and their Experience under Program Work Incentives, Social Security Bulletin (1992), vol. 55, no. 2, pp. 2-19. Su Liu and David Stapleton, How Many SSDI Beneficiaries Leave the Rolls for Work? More than You Might Think, Disability Policy Research Brief (Mathematica Policy Research, Center for Studying Disability Policy, 2010) vol. 10, no. 1. Alan L. Gustman and Thomas L. Steinmeier, Employer-Provided Health Insurance and Retirement Behavior, Industrial and labor Relations Review (October 1994), vol. 48, no. 1, pp. 124-140. Lynn A. Karoly and Jeannette A. Rogowski, The Effect of Access to Post-Retirement Health Insurance on the Decision to Retire Early, Industrial and labor Relations Review (October 1994), vol. 48, no. 1, pp. 103-123. Jeannette A. Rogowski and Lynn A. Karoly, Health Insuance and Retirement Behavior; Evidence from the Health and Retirement Survey, (Rand, 1997), found at: http://www.rand.org/content/dam/rand/pubs/drafts/2008/DRU1798.pdf . David M. Blau and Donna B. Gilleski, Retiree Health Insurance and the Labor Force Behavior of Older Men in the 1990s, The Review of Economics and Statistics (February 2001), vol. 83, no. 1, pp. 64-80. Jonathan Gruber and Brigitte C. Madrian, Health-Insurance Availability and the Retirement Decision, American Economic Review (1997), vo. 85, no. 4, pp. 938-948. Jonathan Gruber and Brigitte C. Madrian, Health Insurance, Labor Supply, and Job Mobility: A Critical Review of the Literature, NBER Working Paper No. 8817 (Cambridge, MA: National Bureau of Economic Research, 2002). Shulamit Kahn, Evidence of Nominal Wage Stickiness from Microdata, American Economic Review (1997), vol. 87, no. 5, pp. 993-1008; Peter Gottschalk, Downward Nominal Wage Flexibility: Real or Measurement Error? Review of Economics and Statistics (2005), vol. 87, no. 3, pp. 556-568; Alessandro Barattieri, Susanto Basu and Peter Gottschalk, Some Evidence on the Importance of Sticky Wages, NBER Working Paper No. 16130 (Cambridge, Ma: National Bureau of Economic Research, 2010). Michael K. Lettau and Thomas C. Buchmueller, Comparing Benefit Costs for Full- and Part-Time Workers, Monthly Labor Review (March 1999), pp. 30-35. 16th Annual Towers Watson/National Business Group on Health, Employer Survey on Purchasing Value in Health Care, Shaping Health Care Strategy in a Post-Reform Environment, 2011.
47 46 45 44 43 42 41 40 39 38 37 36

35

70

David M. Cutler and Brigitte C. Madrian, Labor-Market Responses to Rising Health Insurance Costs: Evidence on Hours Worked, RAND Journal of Economics (Autumn 1998), vol. 29, no. 3, pp. 509-530. Benjamin D. Sommers, Who Really Pays for Health Insurance? The Incidence of Employer-Provided Health Insurance with Sticky Nominal Wages, International Journal of Health Care Finance and Economics (2005), vol. 5, pp. 89-118. Katherine Baicker and Amitabh Chandra, The Labor Market Effects of rising Health Insurance Premiums, NBER Working Paper No. 11160(Cambridge, MA: National Bureau of Economic Research, 2005). Robert J. Myers, Actuary to the Committee on Ways and Means, Actuarial cost estimates and summary of provisions of the Old-Age, Survivors, and Disability Insurance Systems as modified by the Social Security Amendments of 1965 and actuarial cost estimates and summary of provisions of the Hospital Insurance and Supplementary Medical Insurance Systems as established by such act (Committee on Ways and Means, House of Representatives, 89th Congress, First Session, July 1965).
52 51 50 49

48

Ibid. p. 28.

Average wages were calculated from the Average Wage Index series developed by the Office of the Actuary, Social Security Administration; average daily hospital charges and reimbursement rates were taken from the Social Security Bulletin Annual Statistical Supplement, 1976, p. 178, Social Security Bulletin Annual Statistical Supplement, 1981, p. 209 and Social Security Bulletin Annual Statistical Supplement, 1993, p. 311. Robert J. Myers, Actuarial Cost Estimates for Hospital Insurance Act of 1965 and Social Security Amendments of 1965, Actuarial Study No. 59 (U.S. Department of Health, Education, and Welfare, Social Security Administration, Division of the Actuary, January 1965), p. 7.
55 54

53

Ibid, p. 8.

Congressional Budget Office, The Long-Term Outlook for Health Care Spending (2007), p. 13, found at: http://www.cbo.gov/ftpdocs/87xx/doc8758/11-13-LT-Health.pdf Centers for Medicare and Medicaid Services, National Health Expenditures Fact Sheet, found at: https://www.cms.gov/NationalHealthExpendData/25_NHE_Fact_Sheet.asp. John F. Cogan, R. Glenn Hubbard and Daniel Kessler, The Effect of Massachusetts Health Reform on Employer-Sponsored Insurance Premiums, Forum for Health Economics and Policy (April 2010), vol. 13, no. 2, http://www.bepress.com/cgi/viewcontent.cgi?context=fhep&article=1204&date=&mt=MTI5MTIzNzQ2MA==&acc ess_ok_form=Continue. Peter R. Orszag and Ezekiel J. Emanuel, Health Care Reform and Cost Control, New England Journal of Medicine (June 16, 2010), found at: http://healthpolicyandreform.nejm.org/?p=3564. Steven A. Nyce and Sylvester J. Schieber, Health Care Inflation, Must Workers Bear the Brunt? The Milken Institute Review (Second quarter 2010), pp. 46-57.
60 59 58 57

56

71

You might also like