Bargaining Health Benefits in the Workplace: An Inside View

In early 1999, as a professor of health economics and a union representative at a major university, I served on a health benefits bargaining committee, preparing for formal contract negotiations to begin that summer. The committee had an “issue-based” bargaining format and consisted of faculty union representatives and members of the university administration. They were asked to submit a report that the bargainers could use in the formal negotiations. Both the union representatives and the members of the university administration hoped that the committee could prepare the bargainers well enough so that they would not have to settle health care issues using last-minute compromises and ad hoc formulas. This article examines the process of writing this report and its results.

Negotiating health care coverage in the workplace is a fundamental issue in health care reform in the United States. Largely because of historical factors, most Americans receive health benefits through their (or their spouse's) employer, benefits that are provided and bargained as part of a larger wage and salary package. Pauly (1997, chaps. 1 and 4) described the “confusion” between two different conceptions of what payment for health care and health insurance means to U.S. business. One view, which is held by many business managers, is that like all other costs, health care costs are simply part of doing business, and so reducing them increases profits. A second view, which is shared by most economists, maintains that health costs or health insurance premiums ultimately come out of what would otherwise be money wages for workers. Thus a decrease in health costs (implying a reduction also in benefits) must be offset by an increase in money wages.

The economists’ textbook model, using an ideal labor market, shows how differences in benefit packages may be balanced by changes in money wages. Because wage contracts for unionized employers are typically negotiated for two or three years, negotiators must forecast, albeit with some uncertainty, wage trends, health care demand, and health insurance premiums. Moreover, the health benefit portion of the contract may be a relatively small part of the wage bill. Finally, in the “real” world, wages are negotiated at the same time as are other, nonwage issues such as organizing rules, work rules, or grievance procedures.

In this article I examine a particular labor negotiation in the context of the market model. Although it cannot replicate all the conditions required by an ideal market, it does provide a framework for describing and analyzing labor market processes and results in a collective bargaining environment. I then describe a wage-and-benefit model, the employment setting, the benefits bargained, and the “issue-based” bargaining process used in the negotiations. After analyzing the results, I conclude with observations for both employers and the larger issue of health care reform.

Who Pays for Health Insurance: The Textbook Model

Analysts model the relationship of health insurance and labor costs by starting with labor demand and supply without health insurance and then examining its impact. Following Folland, Goodman, and Stano (2001, chap. 11), consider an employer trying to decide how many workers to hire. Analysts generally assume that a lower market money wage rate prompts an employer to hire more workers, for two reasons: (1) the employer can substitute labor for more expensive equipment, and (2) the employer can sell more products at lower prices and thus needs more workers. For a higher wage rate, the opposite occurs. Assume at the outset that there is no health insurance benefit and that the market wage is $20 per hour. Employers will hire workers as long as the incremental revenue from the goods those workers produce exceeds the $20 per hour wage. To begin, assume that the employer hires 1,000 workers, at an equilibrium money wage of $20 per hour.

Suppose that workers negotiate a health insurance benefit that is worth $1 per hour and costs the employer exactly $1 per hour (henceforth all wage rates are in hourly terms) to provide. The employer, who was previously willing to pay $20, will now pay $20 less the $1 cost to provide the benefit. Other points on the employer's demand schedule—which indicates the number of workers it would hire at different wages—will also change by the $1 cost of the benefit.

Workers who were previously willing to accept a wage of $20 are now willing to supply their labor for $1 less, since they value the benefit at $1. As a result, the net wage (the money wage + the value of the benefit) remains unchanged at $20, but the equilibrium money wage falls to $19, or by exactly the amount of the benefit. The workers accept the lower money wages, and the same 1,000 workers are employed at the same net wage, $19 in money wages plus the $1 benefit. The workers are no worse off at a wage of $19 with the health insurance than they were at $20 without the health insurance, because the insurance is worth the $1 that they lose in the reduced wage.

Assuming that the insurance benefit was worth what it cost ignores the behavior, often termed moral hazard, induced by insurance. Suppose instead that employees negotiate the option to buy any “brand-name” drug at the price of its cheaper generic equivalent. This benefit will likely increase health care expenditures because the more expensive brand-name drugs will displace cheaper generics (hence the moral hazard). The employer must again spend a dollar (per hour) to provide the benefit and so, to compensate, will lower the money wage offer.

If the generic drugs provide the same results as the brand-name drugs do, then at least some workers may not be willing to accept a lower money wage in return for the opportunity to buy the brand-name drugs at lower prices. Moreover, many employees may not use brand-name drugs, or even any prescription drugs, at all, and so they may not greatly value this benefit. Accordingly, if the hourly benefit is worth less than the $1 that it costs the employer to provide, say 50 cents, then the employees will be willing to accept only a 50-cent reduction in their money wage.

In the first example, employees accepted a $1 decrease in their money wage because the benefits provided were worth a dollar. But in the second example, the employees accepted only a 50-cent decrease in their money wage because they do not value the benefit commensurate with the cost to their employer. The new equilibrium money wage of $19.50 ($20 less the $0.50 decrease that the employees are willing to accept) reflects the 50-cent valuation that workers place on the benefit. Adding the dollar that the employer must spend to provide the benefit leads the employer to face a net wage of $20.50. If the employer sells its product in a competitive national market and cannot raise the product's price to reflect the higher (50-cent) labor cost, it will compensate for the higher cost by reducing the labor force to fewer than the 1,000 previously hired. This reduction reflects the economic cost in unemployment of providing a benefit that is worth less to the employees than it costs.

If the employer has some market power and can raise its product prices without losing all its consumers, it may be able to pass on some of the higher labor costs to its consumers. Conversely, using health insurance coverage costing $1 to give employees benefits that they value more than 50 cents, such as a greater choice of providers, causes the money wage they will accept to fall from $19.50 toward $19.00. The employer's net wage will likewise fall from $20.50 toward $20.00, and employment will rise toward 1,000 workers.

This model becomes more complicated in settings with a diverse workforce, that is, with different workers placing different valuations on health care benefits and thus making it difficult to identify marginal preferences. In addition, when the workers belong to unions, voting and political processes give voice to a wider set of workers than just those at the margin.

Nonetheless, the model suggests that it is advantageous to both employers and their workers to offer health care benefits that are valuable to the workers. Otherwise, above-market labor costs reduce employment as well as profits, or (for nonprofit organizations) residuals to spend on alternative uses. With this simple but useful model in hand, we now turn to a particular example.

The Setting

The contractual negotiations described in this article took place in a state university with 13 colleges, including schools of law, medicine, and engineering, and more than 30,000 students.1 The university has had collective bargaining for 30 years. A single union represents the teaching faculty and the academic support staff (library staff, financial aid counselors, and others), and several other unions represent other worker groups. The unions, all locals of nationally organized unions, conduct separate negotiations, but both labor and management understand that all the unions’ wage and benefit packages generally move together, as do the packages for university employees not represented by a union.

In 1999, with the previous three-year contract about to expire, the university (henceforth referred to as the “administration”) and the faculty union prepared to negotiate a new contract. The expiring contract had provided 3.75 percent annual increases in wages and benefits, and the union hoped to raise this figure. The union also wished to establish an agency fee for nonmembers (not all faculty and staff are union members, and nonmembers pay no dues) and to adopt measures that would make it easier to recruit members among current and newly hired faculty and staff.

The university's health benefit package was quite generous well into the 1980s. But because of the employees’ higher payments for fee-for-service (FFS) coverage beginning in the mid-1970s, by 1985 many employees had chosen one of several HMO or PPO alternatives. These plans were available at lower costs than FFS coverage, which was typically chosen by more highly compensated employees and retirees (in 1999 the mean age of those covered by FFS plans was 63 years).

Although the details of the various HMO and PPO plans differed, all offered generous pharmaceutical coverage (copayments of $2) and zero-cost office visits, with little incentive to economize on visits to costly settings such as emergency rooms. The annual family dental benefit had been capped at $1,000 for a number of years. The disability benefit was 50 percent of the worker's salary up to a maximum of $30,000 per year. Because the disability benefit was figured along with Social Security disability, the $30,000 cap was applied to the university's plan plus Social Security (which was capped at $17,000). The university also offered a flexible medical spending account (FSA) option, in which the beneficiary could deduct up to $3,000 of pretax salary each year to pay for uncovered expenses, subject to $1 per week ($51 per year) in administrative fees.

The Process

In early 1999 after the administration and the union agreed to an “issue-based bargaining format” to examine health benefits, they formed a six-member “fact-finding” committee. Supposedly there were no “sides” in this process, although the members obviously represented their group's interests. The administration's team consisted of its chief contract negotiator (also a professional economist), the benefits manager, and an outside attorney acting as a consultant. The union's team included its chief contract negotiator (a law professor specializing in tax policy), myself, and a third member who had headed union negotiations in the past. The meetings began on March 15, about five months before the contract expired. The committee was to submit a report to the bargaining teams by May 1, 1999.

In the March 15 meeting, the administration characterized the current plan as a “rich” one. It asserted that because of the low copayments, services and drugs were overused and that therefore the copayments should be raised. At the second meeting a few days later, the two sides agreed that long-term disability coverage was a health benefit and hence eligible for discussion by the committee. Both sides proposed substantial increases. The benefits manager viewed the existing plan as inadequate, undermining the university's competitive recruiting position.

After initial sparring both among themselves and with the administration, the union members agreed that the committee should investigate the possibility of raising the prescription drug copayments, either a “5–10” ($5 for generic and $10 for brand-name drugs) or a “10–20” (similarly defined) plan. The union was also willing to consider higher outpatient copayments, particularly for emergency room care.

The union proposed that the annual cap of $1,000 for dental care be raised to $1,500. Union negotiators also sought coverage for hearing aids and removal of the flexible spending account (FSA) administrative fee. Although small, this fee discouraged the use of FSAs, particularly by those with relatively low uncovered expenses. As an example, a worker in the 15 percent income tax bracket who had $300 of uncovered expenses would have to pay the $51 per year fee in order to save $45 (15% of $300) in taxes and obviously would not do so. The break-even expenditure level in this case was $340, so that a 15 percent tax rate would yield tax savings equal to the $51 fee.

The committee agreed to create alternative scenarios that the benefits manager would convey to representatives of the managed care organizations with which the university contracted. The union representatives asked if they, too, could talk to metropolitan or statewide labor organizations offering “carve-out” contracting arrangements, particularly for prescription drug services. The administration's chief negotiator did not object, saying that they should use “whatever works.” The union representatives did talk to these organizations, but time constraints precluded substantive input from them. The union's chief negotiator believed, in retrospect, that representatives of the outside (metropolitan/state) labor groups were reluctant to put a quote on the table, fearing that it would be used for bargaining without their getting the contract.

The negotiators expected reductions in the monthly premium quotes, since the health services and the insurance literature suggest some demand responsiveness to increased coinsurance rates or payments. The Rand Health Insurance Experiment (Newhouse and the Insurance Experiment Group 1993) indicates a health care visit coinsurance elasticity of about −0.2.Motheral and Henderson (1999) found in their analysis of pharmaceutical benefits a price elasticity of about −0.35. Similarly, Hillman and colleagues (1999), using data from 1990 to 1992, found a pharmacy copayment price elasticity of about −0.31 for independent practice associations (IPAs), but a smaller elasticity of about −0.08 for network-model HMOs.

We would expect competitive insurers to offer cost reductions reflecting, at the very least, the insurer's smaller share of the health care expenses for the baseline level of services (including drugs) and presumably some additional lower costs related to the reduced moral hazard. These cost reductions would reflect the share of discretionary services in the bundle offered by the HMOs, since the discretionary services presumably would be more responsive than essential services to out-of-pocket payments. For example, Harris, Stergachis, and Ried (1990) distinguished essential medications, such as antihypertensive drugs, cardiac agents, antidiabetic drugs, and thyroid medications, from discretionary medications, such as cough and cold remedies, skeletal muscle relaxants, and nonsteroidal anti-inflammatory drugs.

If insurance follows the textbook definition of moral hazard, leading the employees to purchase more than they would have done without insurance, then the marginal valuation of the health benefits is unlikely to be as high as the marginal cost of providing them. As an extreme example, my local pharmacist told me that he sold prescription shampoo, prescription toothpaste, and prescription shaving cream. At $2 per prescription, these items are considerably cheaper than over-the-counter store- or name-brand substitutes.

If workers find these benefits less desirable and prefer higher money wages, the higher money wages (plus the cost of the benefits) will lead to higher net wages. The burden of the higher net wages is shared by those workers who must be terminated (because their wages exceed the value of their marginal products) and by the consumers, who may have to pay higher product prices if the employer is operating in a less than perfectly competitive market.

When unions and management negotiate new multiyear contracts expecting at least nominal, if not real (in terms of inflation), money wage increases, the same logic holds. All else being equal, if employees’ contributions and copayments do not change even when medical costs rise, then real net wages will, in effect, increase. Higher employee contributions or insurance copayments for health benefits are likely to be offset by larger negotiated money wage packages.

The committee members understood the fungibility of wages and benefits. Recognizing the role of health insurance in reducing risk, the union's chief negotiator nonetheless characterized much of the health package as a “prepayment” for services rather than “true” insurance against unexpected occurrences. Both sides recognized that the size of the overall increase in compensation would be determined in the negotiations. The imposition of higher insurance copayments or the enrichment of the long-term disability benefit would presumably be offset by changes in other parts of the benefit package.

Results of the Committee Process and the Negotiations

The committee recognized that the most important responses would come from the HMO and PPO groups, since the FFS option already had in place a substantial (10%) coinsurance rate. Table 1 lists the HMOs’ responses to proposals that the current plan ($2 prescriptions and free office and emergency room visits) be replaced with a less generous plan. These responses came from three major managed care providers that the university already was using (MC1, MC2, and MC3), and a fourth managed care provider (MC4) that was being considered to broaden the university's geographic coverage based on the workers’ residences. Because dental coverage was self-insured for the employer (although administered by an outside plan), no outside estimates were provided.

TABLE 1.

Proposed Changes in Current Plan: Monthly Rates

	a	b	c	d	e	f	g	h	i
	Projected Current	Projected Current w/Hearing Aid	% diff. from a to b	Projected Current w/$ 10 Rx	% diff. from a to d	Projected Current w/$ 10 Rx $ 10 Outpat. $25 ER	% diff. from a to f	Projected Current w/$ 10 Rx $ 10 Outpat. $50 ER	% diff. from a to h
MC1
Single Person	183a	183	0.2	173	−5.6	168	−8.4	166	−9.2
Two Persons	418	419	0.2	395	−5.6	383	−8.4	380	−9.2
Family	467	468	0.2	441	−5.6	428	−8.4	425	−9.2
MC2
Single Person	176	177	0.5	165	−6.3	157	−11.1	na	—
Two Persons	427	429	0.5	400	−6.3	379	−11.1	na	—
Family	490	492	0.5	459	−6.3	436	−11.1	na	—
MC3
Single Person	172	na	—	164	−5.0	154	−10.7	na	—
Two Persons	403	na	—	383	−5.0	360	−10.7	na	—
Family	453	na	—	430	−5.0	404	−10.7	na	—
MC4b
Single Person	na	196	—	182	−7.2b	172	−12.1b	170	−13.2b
Two Persons	na	459	—	426	−7.2	403	−12.1	398	−13.2
Family	na	515	—	479	−7.2	453	−12.1	448	−13.2

Notes:

^aAll figures are rounded to the nearest dollar.

^bDid not provide health care for university at that time. Percentages use column b as a base and hence overstate decreases in columns e, g, and i, compared with MC1, MC2, and MC3.

Column a of Table 1 shows the projected monthly costs for the year, beginning on September 1, 1999, for single-person, two-person, and family coverage. Column b shows the current projections plus coverage for hearing aids. MC1 quoted increases of $1 per month or less, about 0.2 percent. MC2 quoted increases of $2 per month, or slightly less than 0.5 percent.

Column d shows the impact on premiums (relative to column a) of moving from $2 to $10 drug (Rx) copayments, holding all other features constant. The four plans projected premium decreases of between 5.0 and 7.2 percent (recognizing that the 7.2% decline for MC4 is slightly overstated, since it uses column b rather than column a as its comparison base).

Column f shows the impact of $10 office visit copayments and $25 emergency room copayments, in addition to the $10 Rx copayment. Column g shows that this change yielded premium decreases of between 8.4 and 12.1 percent (again with the 12.1% figure slightly overstated).

Column h provides the incremental impact of increasing the emergency room copayment (relative to column f) from $25 to $50. For MC1, the percentage decrease from column a rose from 8.4 to 9.2 percent. For MC4, the percentage decrease rose from 12.1 to 13.2 percent.

When the managed care proposals arrived, both sides were surprised at the seemingly small impacts of the proposed changes on the monthly premiums. In discussions around the table, the committee surmised that the providers were reducing the premiums by the difference in the consumer payment, with little adjustment for demand induced by moral hazard.

With benefit of hindsight, it appears that the adjustments were consistent with general actuarial practices. Melek and Pyenson (1995, 38–39) demonstrated the use of actuarially determined capitation rates for mental health benefits. Consider, for example, a benefit plan offering 15 inpatient days and 45 outpatient visits in a moderately managed plan and at a national average cost. With copayments of $0 per admission and $10 per outpatient visit, the “per person per month” (PMPM) rate would be $5.62. But with copayments of $200 per admission and $25 per outpatient visit, the PMPM rate would be $4.70, or 16.4 percent lower. This percentage is similar to those of columns f and h when compared with column a in Table 1.

We can also compare the implicit quantity impact of raising the pharmaceutical copayment from $2 to $10 with the demand elasticities from the literature. Although the specific HMO figures are not available, outside data suggest that for MC1, MC2, and MC3, HMO pharmaceutical expenditures were between $22.15 and $25.63 per single HMO member. Drug Benefit Trends (1998) reported an average of $14.26 per member per month for prescription drugs in 1996, an increase of $2.63 from 1995. Raising this figure by $2.63 per year for the three years up to 1999 yields $22.15 per member (table 2, column a) per month, or about 12.1 percent. Horn (1997) reported that the HMOs were spending 14 percent of their monthly charges on drugs. Applying this percentage to table 1's figures yields estimates between $24.09 and $25.63 per member per month (table 2, column b).

TABLE 2.

Implied Pharmaceutical Price Elasticities

	a	b	c	d	e	f	g	h	i
	Estimated Pharmaceutical Expenditures				Pct. Δ in Pharm. Expend.		Implied Elasticities
	Low	High	Pct. Δ in Price Change	Δ in Pharm. Expend. (Table 1) Change	Low	High	Low	High	Mean
MC1	22.15	25.63	133.33	−10.20	−46.05	−39.80	−0.345	−0.298	−0.322
MC2	22.15	24.68	133.33	−11.07	−49.98	−44.86	−0.375	−0.336	−0.356
MC3	22.15	24.09	133.33	−8.58	−38.74	−35.61	−0.291	−0.267	−0.279

Column d shows the change in pharmaceutical expenditures expected from a 133.33 percent price increase (using standard midpoint elasticity calculations of the $8 increase divided by the $6 midpoint) from $2 to $10. Holding market prices constant, the changes in expenditures reflect the changes in pharmaceutical quantities. Columns e and f put the changes into percentage terms, and columns g, h, and i provide the upper and lower bounds and mean estimates. The implied price elasticities center on −0.3, which are consistent with the estimates by Motheral and Henderson (1999), Newhouse and colleagues (1993), and Hillman and colleagues (1999).

If the proposed premium decreases were “small,” what would be considered “large”? Benefit packages cover bundles of inpatient and outpatient services and drug therapies for chronic and acute conditions. If higher costs or acute conditions account for the larger share of the health benefit package, the quotes provided may address cost containment on those items for which the demand is more responsive to higher out-of-pocket prices. If, however, the more expensive items do not respond to changed prices, their component of the premium is not likely to fall.

In early May 1999, the committee issued its report, listing potential options and their estimated costs. The report included two distinct alternatives for increasing the long-term disability benefit, increased dental coverage, and hearing aid coverage and eliminating the FSA administrative charges, and it also considered options regarding higher HMO copayments. Rate quotes from the managed care organizations (table 1) were included as appendices.

The formal negotiations began in June 1999 and continued through the summer. There was no renegotiation of drug, outpatient, or emergency room copayments; that is, no copayment rates were changed in any of the plans. The contract was settled in early September after a brief job action, whose major points of contention were the percentage wage increase and the agency fee.

The resulting faculty wage-plus-benefit package increase was 4 percent per year over three years, slightly more than in the previous contract. The agency fee for nonmembers was not negotiated, although the administration did agree to allow union-recruiting materials to be placed in offer packages sent to prospective faculty and staff members. There were no changes in copayments.

In regard to other health-related benefits, annual dental coverage rose from $1,000 to $1,500. The long-term disability cap (including Social Security) increased from $30,000 to $60,000, and the replacement percentage climbed from 50 to 66 2/3 percent of the base salary. The administrative fee for flexible medical spending accounts was removed. Coverage for hearing aids was not provided. When I asked why this seemingly inexpensive (and not subject to moral hazard) benefit was rejected, the administration's chief negotiator responded, “We couldn't give you everything you wanted.”

Observations

This section of the article analyzes the results of the bargaining on four important issues: (1) the application of the “textbook” model, (2) the impacts of health care costs, (3) the effects attributable to unionized bargaining, and (4) the differences between public and private universities.

The Textbook Model

Although both sides expressed an interest in changing the health insurance benefits, terms remained unchanged. Indeed, the projected cost savings from increasing the copayments for drugs and outpatient and emergency room visits were not even addressed in the formal negotiations. As explained earlier, the expected changes in use related to the demand elasticities of the individual components (and their shares of the total expenditures) and to current actuarial practices did not permit larger adjustments by the insurers.

Compared with the packages at other institutions, our university's prescription drug benefit plan remained “rich.” For the 1999–2000 period, Gabel and colleagues found that the plans of 80 percent of all insured workers in the United States had a tiered structure, with more generous coverage for generic drugs and less generous coverage for brand-name drugs (Gabel et al. 2000). Copayments averaged $7 to $8 for generic drugs, $12 to $14 for brand-name drugs with no generic substitute, and $15 to $20 for brand-name drugs with generic substitutes (HRET 2002).

Experts now attribute the continuation of existing benefit packages to the perceived tight labor market, given the strength of the national and regional economies in 1999. Gabel and colleagues (2000, 150) presented the results of the 2000 Kaiser Family Foundation and Health Research and Educational Trust (KFF/HRET), arguing that “more employers are offering coverage to their workers and companies continued to absorb most of the increases in premiums in 2000—average employee contributions did not rise at all. This is presumably in large part because of the need to attract workers in an economy marked by low unemployment” (table 3).

TABLE 3.

Out-of-Pocket Health Care Spending by Workers with Employer-Based Coverage, and the National Unemployment Rate, Selected Years

Year	Total Out-of Pocket Expendituresa ($1990)	% Increase Expend.	National Unemployment Rateb	% Increase Unemp.
1990	1040		5.6
1993	1208	16.2%	6.9	23.2%
1995	1152	−4.6%	5.6	−18.8%
1997	1084	−5.9%	4.2	−25.0%

Sources:

^aGabel, et al. 2001.

^bU.S. Bureau of the Census 2000, Table No. 643.

Elaborating on this argument, table 3 relates out-of-pocket health care spending (total medical and drug expenses plus health insurance premiums are from Gabel et al. 2001) in real dollars by workers with employer-based coverage for 1990, 1993, 1995, and 1997 to the aggregate U.S. unemployment rate. The two measures were directly related, with higher (lower) unemployment rates going hand in hand with higher (lower) employee spending.

Although the textbook model is deterministic and predicts current wages, it may not predict these wages as precisely for a multiyear contract. In tight labor markets, employers may prefer to provide more generous health care contributions, which may be less visible and more adjustable (through payroll deductions) year by year than stipulated percentage wage increases in a three-year contract. Moreover, while some of our university professors might not agree that the 1999 academic labor market was “tight,” the fact that the benefits for all university (faculty and nonfaculty) employees move together and that local and regional labor markets reflected 30-year lows in unemployment rates may have influenced the administration's bargaining positions and actions. The administration's chief negotiator supported this conclusion, recalling, “It was 1999. State coffers were full. Other universities were offering good packages to faculty unions. There wasn't a dire need to push copay.”

Health Care Costs

A second insight relates to mandated health care costs. The employer chose not to address them, even when changes appeared possible. Employers and analysts often argue that mandated coverages are likely to raise health care costs and the accompanying insurance rates, leading some employers to stop offering insurance or some employees to stop accepting it.

Gabel and colleagues (2000) explained that employers have sought to control drug costs, in part by shifting them to employees. Two- or three-tiered copayment structures now predominate, and “most covered workers have benefits that provide incentive to choose less expensive (generic) drugs.” Although the committee considered such initiatives, no tiered plans survived in the formal negotiations.

Would raising the copayment increase the workers’ taxable income, thereby making a change less desirable? Even though employer-provided insurance is subsidized by taxes and may lead to more insurance than is necessary, consumers still are sensitive to out-of-pocket costs. Furthermore, in the negotiations, both the administration and the union viewed a reduction in, for example, prescription drug costs as an opportunity to free up funds for other parts of the tax-subsidized package that might be valued more at the margin. These other benefits included better dental, hearing aid, and disability benefits.

In addition, although the health benefit package was a major part of the negotiations, it may not have been the most important part for either party. The university's composite fringe benefit rate on wages and salaries was 21.3 percent, with the actual rates ranging from 0 percent (part-time student assistants) to 54 percent (skilled trades). Approximately 6.3 of the 21.3 percentage points were the employer's share of Social Security and Medicare taxes.2 Another 7.1 points (about one-third of the fringe rate) were for medical and dental benefits. Retirement benefits accounted for 6.3 points, and life insurance, workers’ compensation, unemployment insurance, and long-term disability made up the remaining 1.3 points. Even a 10 percent reduction in medical or dental costs, which accounted for 5.85 percent of the wage bill (7.1 points divided by pretax wages of 100% plus the 21.3% fringe rate), would have changed the wage bill by only 0.585 percent. The union was certainly not going to recommend higher copayments, and the administration, which had described the health care plan as “rich,” evidently did not feel that such cost controls were worth negotiating, at least not at this time.

Why? First, aside from monetary benefits, the union's other major goal was to establish an agency fee for nonmembers, thereby increasing the union's revenues and membership. The administration opposed the fee, terming it (according to the union's chief negotiator) “coercive.” Indeed, it may have viewed the higher health costs as an appropriate trade-off.

A second issue relates to the very nature of the negotiations. The administration's chief negotiator recalls that the negotiators “needed more time to work on the copay data.” Citing “lingering doubts” about some of the quotes, he asserted that there was “no time for problems like that when you are dealing with a bargaining list of 70+ items and you have only a few bargaining sessions scheduled. Knowing the vulnerability of the data, either side could kill the proposal by poking at its statistical base.”

A third reason for accepting higher costs suggests that this university and others, consistent with the textbook model, may exert some market power and pass along at least some health care costs in the form of higher student tuition and higher charges to outside research-funding agencies. Our “young” state university has never had a capital campaign or a large endowment to cushion it against revenue losses. Furthermore, the university's ability to pass along costs to students is limited by a state funding formula penalizing institutions that raise undergraduate tuition at a rate higher than the overall inflation rate. Nonetheless, just as neighboring institutions are not perfect substitutes for our university, outside research funders do not “pull their grants” in response to small changes in fringe benefit rates. These factors suggest that as a member of the higher education sector, our university may be subject to somewhat less pressure to limit health care costs than are members of other, more competitive sectors.

Effects of the Unions

Although much of the discussion about employee benefits ignores the dynamics of collective bargaining and negotiation, the literature evaluating the impacts of unions is substantial. Goldstein and Pauly (1976) found that unions do cause compensation packages to shift toward higher health benefits. Shore-Sheppard, Buchmueller, and Jensen (2000) synthesized Goldstein and Pauly's work, noting that for a mix of skilled workers (typical of most large employers and certainly of this university), employers’ benefit decisions weigh the preferences of different worker constituencies, taking account of their value to the firm, their outside employment opportunities, and their willingness to forgo wages in return for health benefits.

Buchmueller, DiNardo, and Valletta (1999) provide an up-to-date summary of how unions have affected health insurance provision and coverage. Citing Freeman and Medoff (1984), they found that in nonunion workplaces, where entry and exit are the primary adjustment mechanisms, employment and compensation outcomes are determined by the preferences of “marginal” workers, who tend to be young and mobile and have little invested in the firm. In contrast, in a unionized environment, the less mobile workers, with more firm-specific investments, have a greater voice. Buchmueller and colleagues characterized these bargained outcomes as more representative of the complete bargaining unit than of the marginal workers.

This characterization fits our university's negotiations in two ways. First, the union's members include both teaching faculty and academic staff. The faculty members are better paid, are generally more mobile, and have more opportunities for outside consulting; and the academic staff are less well paid and generally have less job mobility. Because the bargaining unit contains a larger fraction of the academic staff than faculty members, the bargainers must make sure to address the academic staff's concerns. Second, as noted earlier, any bargaining outcome is generally applied to all the university's unions, as well as to the nonunionized personnel. In effect, then, the union was bargaining for a very diverse group, among whom the professors were generally the highest paid and most mobile.

This diversity was apparently important to the negotiations regarding copayments. According to the chief administration negotiator,

During bargaining sessions, there was some nibbling, but no biting of the copay proposal. There was concern about the “incidence” of copay. A copay is a regressive thing. An academic staff member with a salary of 25K would find any kind of copay a greater burden than you and I. This is an important consideration in the context of collective bargaining, which strives to provide equal benefits to members.

Although “25K” represents the low end of the academic staff's pay scale, this observation is salient. What seemed to be an appropriate and desirable outcome for the more highly paid faculty may have seemed more risky and less attractive to the less highly paid academic staff.

Public versus Private Universities

A fourth issue concerns labor negotiations in universities. Both public and private universities have unionized faculty. Responding to my queries, an attorney from the parent national union explained that about 34 percent of faculty members at two- and four-year public universities are unionized, compared with about 10 percent of the faculty at two- and four-year private universities. Most public universities operate under state laws that make work stoppages illegal in one way or another, although the laws are rarely, if ever, enforced.

The attorney considered the key difference between public and private universities to be the ability of the union negotiators at public institutions to go to the National Labor Relations Board (NLRB). In 1980, however, the U.S. Supreme Court held that because the faculty at Yeshiva University (a private institution) participated in cooperative relationships with their administration through the mechanism of shared governance, they were considered managerial employees and therefore were excluded from the protection of the National Labor Relations Act. This decision effectively allows private universities to “walk away” from negotiations with faculty unions. The attorney advised me that when faculty at a private university are opposed by management, “it is very dangerous to go to the NLRB. Your lawyer would tell you that you'd litigate for three years, at a cost of $100,000, and you would lose.”

How the distinction between public and private institutions affects wage and benefit negotiations is a matter of speculation. Because of the precedent at Yeshiva, there is probably less incentive for private universities to negotiate the wide range of health benefit issues with their unions and also less fear of unionization. Yet private universities must compete for faculty in a national market and for administrative and support staff in local markets—hence they are almost certainly subject to the same market pressures as the public universities are.

Conclusions

The private discussions among the university negotiators produced a general understanding that health benefits and money wages constituted alternative uses for the overall benefit package. Yet when given a chance to address potential costs related to moral hazard, there were no changes, with the result that at least some higher (business) costs (in this imperfectly competitive sector) would be passed on to consumers.

Would the proposed changes have represented an improvement for many workers? The May 1999 report by the union and the administration estimated a coinsurance-based cost savings of $820,000 if applied to all 5,000 (unionized and nonunionized) university employees. The subsequent 1999 contract mandated a new committee of union and administration members to gather information for the next (2002) negotiations. That committee (which I chaired) reported in December 2001 an estimated savings of $1.58 million for all university employees. The two reports thus estimated savings between $164 and $316 per worker, although how they can be packaged and presented to the unionized and nonunionized workers still has not been resolved.

We should examine carefully the demand for health care in regard to coinsurance rates and the definition of benefits. The insurers’ responses (for an admittedly healthy population in a very safe occupation) to proposed increases in insurance copayments, although consistent with estimates from the literature, seemed quite modest.

The insurers’ responses may have reflected strategic first responses in their continuing negotiations with the university, and a second round of negotiations may have resulted in yet lower (i.e., better) quotes. The managed care plans possibly viewed the demand for their products as less elastic than economists have assumed, owing to the potential disruption and loss of productivity when an employer switches plans or insists that the employees share more of the costs.

In addition, the 1999 KFF/HRET study (exhibit 1.2) indicates that 56.9 percent of those covered in the economy work in large (1,000 or more insured workers), predominantly self-insured firms. With imperfect competition in the product market and some insulation from competitive pressures in the insurance market, employers may be able to pass on health care costs to their consumers (in the university's case, both students and research funders).3

The committee's experience in 1999 suggests that when the labor market is tight and when other issues subject to collective bargaining are on the table, raising coinsurance rates and addressing health care costs are not critical to either the administration or the insurers of this university's faculty. Health benefit reformers should thus consider this possibility when evaluating proposals regarding mandated coverage, health care financing, or cost containment.