Medical savings accounts (MSAs) have caught the attention of health-care reformers in Canada. Both the interim report of the Senate’s Kirby Committee and the report of the Mazankowski Commission in Alberta highlight MSAs as a possible financing reform, and the upcoming Romanow Commission report will undoubtedly offer its views on the potential place of MSAs in Canada’s health-care system.

Publicly funded MSAs can be likened to vouchers for health care. An MSA-based health-care financing scheme includes two essential features: an individual- or household- specific account whose balances, which can accumulate over time, are earmarked for health-care expenses; and a high-deductible, catastrophic insurance plan that covers expenses above the deductible. Individuals use MSA funds (and personal resources if the MSA funds are not adequate) to pay for health-care expenses below the deductible and, if required, cost-sharing above the deductible. The catastroph- ic insurance policy insures extraordinary, high-cost care. Providers bill patients directly.

By forcing individuals to purchase routine health care at full price up to the deductible, MSAs are designed to counter the incentive under free care for over-utilization, encourage price competition among health-care providers who compete for cost-conscious patients and reduce the inequities of standard user-charge policies by providing resources to individuals to make such purchases.

MSA advocates argue further that MSAs can be integrat- ed into Canada’s system of public health-care financing without compromising universal access to medically neces- sary services while generating substantial benefits, includ- ing increased consumer choice, better access to many serv- ices, greater cost-control, improved system efficiency, and increased personal responsibility and accountability. MSA detractors, in contrast, argue that Canadian-style MSAs will increase public health-care expenditures, decrease system efficiency and compromise health system equity.

What does the evidence say? Three factors make it dif- ficult to draw inferences about the likely effects of publicly financed MSAs in Canada: there is very limited evidence of the performance of MSAs, much of the evidence is of low quality and it comes from settings very different than Canada’s publicly financed health-care system. Singapore, for example, is the only country that has integrated MSAs into its system of health-care finance, and interpretation of their effects differ sharply. MSA advocates tend to view Singapore, with its modern health-care facilities and low per-capita spending on health care, as a shining example of the potential of MSAs. Critics emphasize that MSAs constitute a relatively small portion of Singapore’s financing mix and that Singapore’s rate of increase in health-care expenditures actu- ally increased following the introduction of MSAs in 1984, leading the Singapore government to conclude in 1993 that: ”œMarket forces alone will not suffice to hold down medical costs … [T]he government has to intervene directly to structure and regulate the health system.”

The other country with perhaps the greatest experience with MSAs is the United States. But the MSA literature from the U.S. consists mainly of ana- lytic commentary and policy proposals, reports of the experiences of private firms that have limited scientific validity and generalizability, or predic- tions based on simulation models within America’s multi-payer insurance system for selected popula- tion groups. Two recent pilot projects authorized by the U.S. Congress give one pause regarding the real potential of MSAs, particularly in publicly financed health-care systems. One project was designed to provide information on the effect of MSAs on health expenditures and risk selection among the self-employed or those working in small firms; the second was targeted at beneficiaries in the publicly financed Medicare program, which provides health care to senior citizens. Both pilot projects failed due to lack of beneficiary demand for MSA policies and an unwillingness of private insurers to market MSA- based catastrophic insurance policies, which they judged to be complicated and difficult to sell to ben- eficiaries. In a review of the experience with the Medicare pilot, the U.S. Medicare Payment Advisory Committee concluded that there was no way to adjust the design of the MSA schemes so as to serve the interests of the public Medicare program and elicit sufficient response from private insurers.

Although the MSA literature per se is quite limited, because MSAs rely on demand-side incentives, one can draw on the extensive ana- lytic and empirical research on cost-sharing and insurance markets to identify the likely effects of MSAs when integrated into a system of public health-care finance. In what follows I draw both on other jurisdictions’ experiences with MSAs and on the broader research literature on insur- ance markets to assess claims made about the likely impact of MSA-based financing on health- care expenditures, health system efficiency, con- sumer choice and access and equity.

Will MSAs control expenditures and increase the health-care system’s efficiency? The patient cost-sharing required by MSA financing likely would lead to reductions in the utilization of health care. However, in Canada such reductions would likely be smaller than is commonly pre- dicted in the MSA literature. There are two rea- sons for this. First, elements of the Canadian sys- tem are currently supply-constrained (e.g., because of the shortage of general practitioners and family physicians), which implies that there are unmet demands among the users of the sys- tem. MSA-induced reductions in utilization among current users will therefore be partly off- set by increased utilization by those whose demand is currently unrealized. If the services forgone by those who reduce utilization are less needed than those purchased by people who are able to gain greater access, this would constitute a social gain even if overall health-care costs are not reduced. But while this is possible it is by no means assured: evidence from the user-charge lit- erature clearly demonstrates that charging people deters both necessary and unnecessary utiliza- tion. Second, where the system is not supply-con- strained, reduced patient demand would trans- late under fee-for-service payment into a large decrease in provider incomes. Providers may par- tially offset this decrease by inducing demand for their services, a phenomenon that was observed in the most comprehensive study of user fees for physician services in Canada, which was con- ducted in 1980 by Glenn Beck and John Horne, who analyzed responses to user charges imposed during the 1970s in Saskatchewan.

More importantly for expenditure control, MSA-induced reductions in utilization will not necessarily translate into expenditures savings to the public sector. Many simulations of the effects of MSAs even predict increased overall spending. To see why this might happen, consider the fol- lowing simplified example, illustrated in Table 1, of a universal, publicly financed MSA scheme with an annual public MSA contribution of $1000 per person, a deductible of $1000 and gov- ernment-provided catastrophic insurance. In the example, there are three people whose utilization of physician and hospital services under the cur- rent financing system ranges from $0 to $5600 (panel a). Under MSA financing, because individ- uals pay the full price for services below the $1000 deductible, utilization falls"in the exam- ple, by 20 per cent for physician services and 10 per cent for hospital services (panel b). Expenditures for persons 1 and 2 fall below the deductible, so they pay for their services using MSA funds while still leaving positive MSA bal- ances at the end of the year. The utilization for person 3 ($4960) exceeds the deductible, so she uses all of her MSA contribution of $1000 (leav- ing a zero balance) and then draws on the public catastrophic insurance for the remainder ($3960).

What happens to public health-care spend- ing as a result? Even though total utilization falls under MSAs (from $6000 to $5280), total public expenditures actually increase (by $960 from $6000 to $6960). This is because the decrease in utilization ($720) is more than offset by the extent to which the public MSA contributions exceed the expenditures of those low-users (per- sons 1 and 2) who end the year with positive MSA balances ($1680).

This example illustrates a key principle for MSAs: in order to reduce overall public expendi- tures, the ”œovercontributions” to the MSAs of low- users must be more than offset by the reductions in utilization of high-users. This is unlikely, how- ever. The vast majority of health-care expenditures are incurred by a small proportion of very ill high- users. This pattern of expenditures creates two effects that mitigate against net cost savings. First, the majority of the population would receive an MSA contribution that exceeded its need for health care, so even small per capita overcontribu- tions can be important in aggregate. Second, very ill high-users are not likely to reduce expenditures substantially because their care is covered by the catastrophic insurance (i.e., it is free at the margin) and, in any event, hospital-based care of this sort tends to be the least responsive to price.

Public savings are even less likely if enroll- ment in the MSA scheme is voluntary, as is pro- posed in many Canadian MSA discussions. The imprecision of even the best methods of risk- adjusting MSA contributions makes voluntary MSAs most appealing to those who have low expected health-care expenditures and least appealing to those with high expected expendi- tures, who tend to remain in the traditional insur- ance system. This risk selection can increase gov- ernment expenditures even further under MSAs, as illustrated in Table 2, which modifies the first example to reflect risk selection under MSA-based financing. Only persons 1 and 2 (who are low risk) enroll in the MSA scheme; person 3 remains in the traditional public insurance plan and her utiliza- tion is the same as under the current system of finance. As a result, public expenditures increase from $6000 under the current system to $7600 under a system of voluntary MSAs.

MSA advocates might counter that the above examples fail to capture the dynamic changes to our health-care system that would be induced by the injection of demand-side compe- tition under MSAs, and they would be right. But the historical record, both with respect to MSAs and health-care financing more generally, is that patient-based demand-side competition often fails to generate cost savings or improved effi- ciency. As noted above, the most comprehensive assessment of Singapore’s experience concluded that MSA-related competition was not only inef- fective in controlling costs, but that price increas- es and wasteful over-adoption of new technolo- gies had actually exacerbated expenditure pressures. Singapore’s ability to control costs has resulted not from MSA financing, but rather from a series of supply-side regulatory initiatives that control, for instance, physician supply, physician fees, the number of hospital beds and the introduction of technologies.

Singapore’s experience with demand-side competition is neither anomalous nor surprising. For well-understood reasons, health-care markets do not operate as do most markets for ordinary consumer goods. It has long been observed, for example, that contrary to predictions based on standard market theory, unregulated physician fees often are positively rather than negatively cor- related with physician supply and that consumer- based competition among hospitals is associated with wasteful duplication and increased spending on new technologies.

Finally, if we take the primary objective of health spending to be the improvement of peo- ple’s health, MSA-induced utilization reductions also detract from system efficiency. To be efficiency enhancing, such reductions would have to target unnecessary utilization, thus eliminating only low-value services. However, it is now well established, based on studies from numerous set- tings for a variety of services, that because con- sumers frequently do not possess the information required to reduce only unnecessary health care, cost-sharing deters both necessary and unneces- sary use. Recent Canadian evidence also docu- ments that cost-sharing-induced reductions in necessary care can generate important adverse health effects in vulnerable populations, includ- ing increased use of emergency rooms, increased admissions to hospitals, and even death.

If MSAs won’t reduce demand or increase the system’s efficiency, will they not at least expand consumer choice and increase access to health- care services? MSA advocates emphasize the potential for MSA financing to increase the range of health-care services that individuals can pur- chase at public subsidy. This could occur, for example, if either current-year MSA funds or accu- mulated MSA balances from previous years can be used to purchase a wide range of health-related services (e.g., drugs, dental care, home care, phys- iotherapy, health club memberships). This is cer- tainly an attractive feature of MSAs, given the patchwork system of public coverage in Canada outside the physician and hospital sectors.

But on reflection, basic actuarial principles make it less clear whether, on average, MSAs truly expand choice for everyone. Table 3 expands the simplified example by including expenditures on uninsured services, which are assumed to be $50, $200 and $300, respectively, for persons 1, 2 and 3. If current-year MSA funds can be used to pur- chase uninsured services, but such expenses do not count toward meeting the $1000 deductible in the catastrophic policy, persons 1 and 2 are made better off. Because their expenditures on insured services are less than the deductible, MSA funds can be used to purchase non-insured services, sav- ing $50 and $200 in personal out-of-pocket expen- ditures. Person 3, however, is not better off: he needs available MSA funds to finance currently insured services, leaving no MSA funds to pur- chase non-insured services. MSAs make him no better off than under the current system. In gener- al, only the subset of relatively healthy individuals whose expenditures on currently insured services are less than the deductible have expanded choice under MSAs. This conclusion is unchanged if, as is sometimes recommended, only accumulated MSA balances from previous years can be used to finance the purchase of non-insured services (e.g., only persons 1 and 2 have accumulated balances at the end of the year). Increased choice for every- one is not possible unless new resources are inject- ed or efficiencies arise.

Will MSAs preserve the equity principles that underlie the Canadian system? A number of intrinsic features of MSA financing inevitably generate adverse effects on equity and access. When individuals’ economic resources differ, prices equal to zero are a necessary (though not sufficient) condition for equal access. MSAs’ elimination of free care affects the poor particularly adversely. MSA funds, of course, partially mitigate the effects of requiring individuals to pay the full cost of care up to the deductible. But if accumulated MSA bal- ances can be used to purchase non-health care goods and services, the poor would face a trade-off between health care and necessities, such as hous- ing and food, while upper-income individuals trade off health care against ”œluxuries.”

MSAs redistribute (and re-privatize) revenue raised through the general tax system, financial- ly benefitting those who have less-than-expected health needs. Risk-adjusting the MSA can only partially offset the effects of the well-established correlation between socio-economic status and health, so MSAs benefit healthy, high-income members of society. Simulation models of the effects of introducing voluntary MSAs alongside traditional comprehensive insurance consistently predict that low-risk, high-income individuals benefit the most financially and that high-risk, low-income individuals suffer the most. It is not surprising, therefore, that in some instances MSAs have even been marketed to high-income individuals as tax-preferred savings vehicles.

Finally, it is not at all obvious that MSAs can be easily integrated into the Canadian health- care financing system. On the face of it, a transi- tion to MSA financing appears simple: the govern- ment simply stops paying providers directly and instead deposits the funds into individual MSA accounts. But the recent U.S. Medicare MSA pilot program suggests that integration of MSA financ- ing would not be easy, especially given the historical methods for financing health care in Canada.

In the pilot program, beneficiaries showed little interest in MSAs and private insurers were cor- respondingly reluctant to invest the resources to develop the required insurance plans, which they judged to be complicated and difficult to sell to beneficiaries. As noted above, the Medicare Advisory Payment Commission concluded that there was no way to design the MSA schemes to serve the interests of the publicly financed Medicare program and elicit sufficient response from private insurers. This failure in the U.S. has two major implications for Canada. First, in any system of voluntary MSAs, it may be very difficult to sign up subscribers without investment in sig- nificant marketing and educational programs. Unless people are unhappy or see tangible benefit they may exhibit considerable inertia in moving to a new, more complicated financing arrange- ment. Second, given that the U.S. has a stronger, better developed private insurance market than Canada does, Canadian MSA designs that rely on the private sector for the provision of catastroph-ic insurance are even less likely to succeed.

Providing the catastrophic insurance through the public sector eliminates problems related to private insurance markets and repre- sents a smaller departure from the current sys- tem. But such a scheme faces a number of signif- icant practical hurdles. If the MSAs are voluntary, risk selection will likely be a serious problem, crippling the ability of MSA financing to deliver on the promised benefits. A full-scale switchover from the current system to mandatory MSAs is likely to be a bigger step than many policy-mak- ers would be willing to contemplate, however. Finally, the current role for private health insur- ance in Canada is to provide coverage for servic- es that are not publicly insured"drugs, dental care, physiotherapy, room upgrades for inpatient stays, out-of-country coverage and so forth. If individuals are allowed to use public MSA funds to purchase such services, the existing role for private insurance in Canada evaporates. This cre- ates a conundrum for MSA proponents, who gen- erally advocate greater private-sector financing in Canada. Allowing funds to be used to purchase currently uninsured services is the only way to create the financial incentive for people to use MSA funds prudently, and yet any design that allows this (with publicly provided catastrophic insurance) would be vigorously opposed by the private insurance industry and have the ironic effect of eliminating one of the most important current sources of private finance.

Under plausible scenarios, the integration of MSAs into Canada’s system of health-care financing could lead to higher public costs, reduced efficiency, reduced equity and compro- mised access for many Canadians. This is particu- larly the case if MSAs are voluntary and cata- strophic coverage is provided by private insurers. MSAs embody an inherent tension between fea- tures necessary to generate the promised benefits and features necessary to avoid unwanted adverse effects on costs, efficiency and equity. Those fea- tures that enhance incentives to reduce utilization and use MSA funds prudently"increased cost-shar- ing, greater flexibility in how accumulated MSA funds can be spent, favourable tax treatment of accumulated MSA funds"generate some of the greatest inequities in the distribution of the bene- fits and burdens of health-care financing. MSAs bring with them all the well-known (and inescapable) limitations of demand-side approach- es to expenditure control and utilization manage- ment. In doing so, they conflict directly with the fundamental tenet on which Canada’s current sys- tem of hospital and medical financing is based" payment according to ability to pay and service according to ability to benefit. Careful design may temper some of these deleterious effects, but MSAs have little to offer within publicly financed health- care systems and they put much at risk.


The author would like to acknowledge the financial support of the Canadian Institutes of Health Research and the Canadian Health Services Research Foundation, as well as support from the Ontario Ministry of Health to the Centre for Health Economics and Policy Analysis. Morris Barer, Jonathan Lomas and Steve Morgan provided helpful comments but, like the above-named institutions, do not necessarily share the views expressed.