EVENTS

Date:	Wednesday, June 29, 2005
Time:	9:15 AM -- 11:00 AM
Location:	Wohlstetter Conference Center, Twelfth Floor, AEI 1150 Seventeenth Street, N.W., Washington, D.C. 20036

June 2005

Do Health Accounts Promote Better Spending?

Tax-favored health accounts, including flexible spending accounts (FSAs) and health savings accounts (HSAs), are an increasingly popular way for employees to pay for some of their out-of-pocket health expenses using pre-tax dollars. Do such accounts improve the incentives for workers to purchase health care more carefully? Do employers providing FSAs offer health insurance plans with higher cost-sharing? If so, do the accounts simply shift costs from the employer to the employee, or are employees better off? On June 29, William Jack, a professor of economics at Georgetown University, addressed these and related questions as he presented his new study on FSAs, coauthored with Arik Levinson, also a professor of economics at Georgetown, and Sjamsu Rahardja of the World Bank.

William Jack
Georgetown University

When employers sponsor group health insurance, employees share costs in two ways: (1) by sharing premium payments, and (2) by paying out-of-pocket through co-payments and coinsurance when using medical services. Considerable research has investigated the former case, but there is little research on the effect of cost-sharing. This paper focuses on the second component. 

The study utilizes flexible spending accounts (FSAs) as a proxy for out-of-pocket cost measurements. Few studies have investigated burgeoning out-of-pocket costs, and even fewer studies consider whether the increase can be attributed to the rise of FSAs. The purpose of the paper was to empirically validate whether FSAs can explain the shift in health-care costs to employees, and discern the welfare impact of this shift.

The paper explores three major questions. Do employers offer FSAs in conjunction with higher coinsurance rates? If so, are they in part responsible for the rise in employee contributions made to employer-sponsored health insurance? And more importantly, does this added subsidy mitigate or exacerbate the welfare loss associated with the existing premium subsidy?
 
To empirically address these issues, they analyzed data from the 1993 Employer Health Insurance Survey (EHIS). Compiled by the Robert Wood Johnson Foundation, the EHIS is a cross-sectional index of firm-level data on health insurance plans offered by ten U.S. states. Within these analyses, they controlled for endogenous and exogenous firm characteristics that could bias the impact of FSAs on coinsurance rates.

Among the surveyed firms, FSAs increase coinsurance rates by 7.3 percentage points. Given that the average coinsurance rate for these plans was 17 percent, the increase is statistically and economically significant. As posited by William Jack and Louise Sheiner in an earlier work (“Welfare-improving health expenditure subsidies,” American Economic Review, 1997, 87 (1), 206-21), an FSA’s ability to increase coinsurance rates can reduce the distortion that emerges from inefficient first-dollar health coverage. Still, the increase in out-of-pocket spending was determined to be welfare-neutral, as the net-of-subsidy coinsurance rates were unaffected by the presence of FSAs.

Phil Ellis
Congressional Budget Office

The paper makes a valuable empirical contribution to a debate that has been largely occurring in a vacuum of data. The methodology makes use of observable factors or “instruments” that influence the firm’s likelihood of offering an FSA, but not the coinsurance rate. It was surprising that the instruments effectively controlled for the selection bias and immeasurable attributes of firms. In general, firms that offer health insurance also offer higher wages. Therefore, it is difficult to judge to what extent employees are paying for health insurance through their wages. 

The findings are difficult to explain on theoretical grounds, and the results are ambiguous about causation. Clearly, with the introduction of an FSA, firms increase their gross coinsurance rates. However, this may be offset by the tax subsidy on out-of-pocket costs. The models rely on an imputed average tax rate to determine the value of the FSA subsidy. This is problematic, since it may overestimate the actual tax subsidy on spending through FSAs. Employees do not pay health-care expenses exclusively through “use it or lose it” FSAs; they cannot predict their health-care needs in advance. On the other hand, the data indicate that people with higher tax rates are more likely to use FSAs and to put more money into them. This may have countered the erroneously large estimate of the tax subsidy, but it is difficult to judge.

In addition, the paper unrealistically assumes that everyone spends the same amount on health-care. Among the non-elderly insured population, about 60 percent of people account for 10 percent of health-care spending, while only 10 percent of people account for over 60 percent of health-care spending. There is a tendency to use these data as proof that health savings accounts will not benefit consumers since the majority of spending occurs beyond the deductible offered in these plans. However, a RAND study has found that up-front cost-sharing does discourage hospitalizations and reduces spending, even when consumers have exceeded the deductible.

This paper is the first in what will be a continuing series of publications as more data emerge on the effects of savings accounts. As these papers appear in print, one must use caution and be on the lookout for selection bias. Moreover, when faced with evidence of cost-shifting, it is necessary to determine whether costs were rearranged or benefits reduced.

Louise Sheiner
Federal Reserve Board

The paper takes a novel approach in empirically investigating the effects of tax subsidies on out-of-pocket health expenditures. Effectively, one of two paths emerges when such a tax-subsidy is offered: (1) the nominal coinsurance rates remain constant, ultimately lowering out-of-pocket payments; or (2) participants choose insurance policies with higher cost-sharing provisions.

The theory is primarily one of incentive: given the option of a tax-subsidy, employees are drawn to enroll in more comprehensive and expensive health plans than they would otherwise choose. The paper establishes a “responsiveness” on why some firms institute FSAs whereas others do not, and finds a large effect of FSAs on coinsurance rates. Yet, the paper concludes that FSAs have no measurable effect on net after-tax coinsurance, and thus do not reduce health-care spending. 

The study falls short in some aspects. For example, there is no clear indication of how the instruments under investigation behave. The data suggest that older firms are more inclined to utilize FSAs, yet the authors do not venture a guess as to why this is so. Moreover, the paper excludes managed care and PPOs, even though their usage is quite prevalent.  

The paper focuses on a very narrow aspect of “out-of-pocket” payments, and seemingly dismisses out-of-pocket payments via deductibles, out-of-pocket maximums, and co-payments, among others. Although it is clear that firms with high coinsurance rates are more likely to offer FSAs, it is not so clear whether out-of-pocket spending increases as a result of firms adopting FSAs. In fact, out-of-pocket health expenses have steadily decreased over the past quarter-century. 

Notwithstanding, the study provides a crucial investigation on a topic devoid of careful research. The paper is a necessary steppingstone in examining the workings and worth of FSAs.

Tom Miller
Joint Economic Committee
 
The evidence in William Jack’s paper is preliminary. The 1993 sample underrepresents multiple-offer firms and fails to take into account the growth of FSAs in the past twelve years. Furthermore, the options that firms provide do not reflect employees’ decisions or FSA participation levels.

When discussing the rise in out-of-pocket costs, the paper fails to adjust for overall costs. Deductible levels and employees’ share of premiums, as a percentage, have not charted significant growth in the past few years. In addition, FSA subsidies are assumed to be linear and open-ended like HSA tax subsidies. This fails to take into account that employees behave differently when they must target earnings for a single year. A full tax subsidy will not account for the entire out-of-pocket expenditure created by coinsurance and the individual will bear the cost. Mr. Jack’s paper lacks a sufficient time horizon of data analysis; the paper focuses on data from 1993. One must also examine the integration of FSAs with debit cards to allow consumers access to price information at the point of sale.

The bigger picture to take away from this paper is that the growth of consumer-driven health plans challenges the conventional third-party paradigm. According to a McKinsey study, these accounts increase consumer engagement in overall health and wellness. When people are ready to take the plunge into health savings accounts, we can expect a change in their health-care spending patterns.
 
AEI research assistant Saad Ahmad and AEI intern Clara Magram prepared this summary.