Leadership Forum - March 2005

Prospects for Real Voluntary Health Insurance in Developing Countries

• Related Event
• Feature Reading

Author: Mark Pauly, Bendheim Professor, Professor of Health Care Systems, Business and Public Policy, and Risk and Insurance, Wharton School, and Professor of Economics, University of Pennsylvania.

Introduction: the Rationale for Private Insurance with Incomplete Public Financing

In most developing countries the system of public financing of health services fails to provide enough funding or resources to satisfy citizen demand for higher quality care. So, despite the usually zero or nominal user price for public services, many households, across the income distribution, pay substantial amounts out of pocket for medical goods and services. Prescription drugs is usually one of the major items of spending, but payments for private physician services or hospital accommodations, or even “tips” within the public system, can amount to significant expenditures, which, when needed, bite deeply into the income and wealth households use for other purposes. Table 1 shows the estimated out-of-pocket percentages for a sample of developing countries with a spectrum of political systems; all are high relative to developed countries, even to the United States, where the percentage paid out of pocket (2003) was 14 percent.

The existence of high and unpredictable out-of-pocket expenses makes a prima facie case for insurance: risk-averse people should prefer to pay a premium close to the average out-of-pocket payment to protect themselves against the relatively rare but possible outcome of a serious illness to a family member associated with an above-average financial drain. While insurance markets can be complex and problems can arise, there would seem to be an overwhelming case, as long as public spending is constrained, for exploring the development of voluntary health insurance in such countries. This is an important point: private insurance will always appear inferior to the hypothetical situation of well-funded public insurance and care managed by a wise and well-informed government, but that is not usually a relevant comparator. Instead, the alternative of constrained public funding is more common in the countries in the table, and the role of voluntary private insurance should be compared to that benchmark. Such voluntary private insurance is not common (although it is not nonexistent either); what are the challenges it faces and what are the prospects that those challenges can be met—with appropriate changes in institutional structure?

Two Misconceptions: Privatization and Affordability

Let us first clear away two fundamentally irrelevant issues that often impede the discussion of voluntary insurance. The first has to do with the role of the public or private sectors in supplying or producing insurance. In essence, the production of insurance is simple: it is only a financial transaction or agreement, in which a set of people all at risk for some unusual loss agree the share those losses when they happen. Those who do not suffer a loss compensate those who do. To make sure any agreement to do this actually is executed, insurers usually collect a sufficient premium to cover the average loss before the loss-producing event occurs, and then use those funds to compensate those who afterwards do suffer a loss. The organizational entity that implements this agreement could be public or private or something in between. The key point is that there is no conceptual necessity that it be of any particular type; in different settings different arrangements will work out better as ways of minimizing the transactions costs (which are the only real resource-consuming activities here) that are part of making and implementing this agreement. It will generally be desirable to give households options to obtain insurance from government sources, from private profit-seeking entities, and from mutual (policyholder-owned) groups organized at the community or occupational level. The pros and cons of “privatization” here should be a nonissue.

The second irrelevant issue is the “affordability” of insurance. It is sometimes said that voluntary insurance is not affordable by the average household in a developing country and so (by implication) only poorly funded and low-quality public services are an option. It is probably true that very low-income households, those at or close to subsistence, will have near-zero out-of-pocket expenses and therefore have no willingness to pay for insurance. For this reason, as will be discussed below, voluntary insurance is not generally a viable solution for the poor. But somewhat further up the income distribution (and not very far up in many countries) we find out-of-pocket payment levels that are large in absolute terms and are quite large relative to household income or wealth. These payments, though rare, can be financially devastating when they do occur, eating up family resources needed to continue farming, to pay children’s school fees, or to maintain adequate housing. It is this group that can feasibly be most helped by voluntary insurance. And it is this group that can afford such insurance, by virtue of the fact that the insurance mitigates the formerly high out-of-pocket payment—if the household with a given set of characteristics is observed coming up with the resources to pay the out-of-pocket payment just at the limit of its financial capacity, it surely can afford the much lower premium that would be charged for health insurance. Indeed, in an intuitive sense the social value of insurance comes from the greater affordability of a small fixed periodic payment (the premium) relative to the rare but devastating high payment when illness strikes. (Estimating both the value of and the demand for voluntary insurance requires a measure of this financial vulnerability, a topic to be considered below).

The Determinants of the Feasibility of Private Insurance: Risk Aversion, Administrative Costs, and Moral Hazard

What then is relevant to the emergence and sustainability of voluntary health insurance markets? The two main factors are common to all insurances: the administrative cost or insurance “loading” relative to households’ risk aversion, and the size of moral hazard. Without a subsidy, real-world insurers (whether private, public, or mutual) need to incur administrative costs to sell insurance, to collect insurance premiums, and to pay medical claims. This expense depends on the institutional structure for engaging in these activities, and it depends as well on the frequency of claims. Usually the most expensive way to produce insurance is to sell it “individually,” one-on-one, and so various customer-grouping arrangements have arisen, often based on the workplace or the community. If a set of households can agree beforehand on the kind of insurance they want, buying it or producing it as a group may lower the administrative cost per insured exposure. If this cost is low, a household, if risk averse, will buy insurance even at a premium somewhat above its expected or average expense. How much higher than the average expense the premium can rise and still be acceptable to buyers depends (tautologically) on their degree of risk aversion, about which little can be said.

The other factor that can influence the depth of coverage (but not, in theory, whether the household finds it worthwhile to purchase at least some insurance) is the existence and strength of moral hazard. Moral hazard arises when the losses an insurance covers increase (in expected value) when coverage is present or set at higher levels. Moral hazard is the mirror image of the improved access to care that insurance provides. Recent research (Nyman, 2004) suggests that it need not always deter the purchase of coverage against medically catastrophic illnesses that can be effectively if expensively treated, nor need it deter coverage against effective preventive care, but it will always deter the purchase of very generous coverage against relatively high-probability, low-cost form of treatment. The household, just like society, needs to trade off the benefit of insurance in providing access to valuable medical care in a way that is not financially devastating against higher premiums. The paradox is that, the more spending is stimulated by coverage, the more premiums will have to increase and so (at an earlier point than if spending were less responsive) the voluntary demand for more generous coverage is reined in. This is sometimes a difficult message for health policymakers to understand, since they often have the goal of increasing the consumption of medical care. That will happen to some extent with voluntary insurance, but, to push it further than what insurance markets would offer, it would be necessary to subsidize the insurance premium.

Designing insurance that will appeal to demanders is thus a tricky balancing act, involving consideration of administrative expense, moral hazard, and risk aversion. Nevertheless, even for households at modest income levels in developing countries, there is good reason to believe that an insurance could be offered at premiums people would be willing to pay. It would not cover everything, and it would require efficiency (and the absence of corruption) in setting up administrative structures and in channeling moral hazard—but it could be done.

Risk Variation, Risk Rating, and Private Insurance: a Controversial Question

The problem of risk variation gets extensive discussion in the theory and the practice of insurance, but it is controversial and often misinterpreted. The only sensible definition here of risk is expected or predicted expense (under a given insurance contract). Of course, even if all buyers of an insurance contract were of equal initial risk, after the fact a few will have high expenses and many will have low or zero expenses. The observation of an uneven distribution of medical spending thus does not prove (or tell us anything much) about the distribution of risk and whether it is equal or unequal. However, both common sense and empirical data do suggest that there are some observable variables that will predict different medical expenses for different households—their age and gender, the household size, the costliness of medical care in the area, and whether or not someone has a chronic condition. If all of the predictors of future medical expense known to buyers of insurance are also known to sellers of insurance, premiums charged by competitive insurers will vary directly with risk. Such risk rating is necessary to get maximum uptake and efficient functioning in unsubsidized markets; if premiums are set much lower than expected expenses for some group (say, older people), and insurers are to cover their costs, they will have to charge premiums much higher than the benefits the low-risk, mostly younger insureds expect to collect, and the latter group will decline coverage.

It is true that risk-rated markets will disadvantage high risks, compared to uniform or community-rated premiums, but they will advantage (and make insurance purchase likely by) the more numerous low risks. This tradeoff is one of the things that makes insurance rating controversial. It is even possible for uniform premiums to deter so many of the lower risks from purchasing that those left in the market are almost all the higher risks, so there is little or no redistribution or advantage for them. Markets will work even less well if buyers have information about risk that insurers do not; “adverse selection” will then ensure with the lower risks either pushed out of the market or purchasing less coverage than they should. Note that, even with risk rating, insurance performs a powerful role of risk pooling; just as described above, the few within a given risk category who are unlucky enough to become seriously ill will get back much more than they paid in premiums, while the majority will have paid in more than they got back. One could label this a “transfer” but, in contrast to governmentally engineered tax and transfer programs, this is one in which (before the fact) the winners and losers are equally eager to participate.

Governments often try to bring about transfers across risk categories in health insurance. For example, they may charge a premium independent of age, thus making transfers from the young (at least those still willing to participate) to the old. This is obviously not risk pooling, since there is almost no risk associated with getting older; it is virtually a sure thing. It is not risk pooling, it is just premium averaging. It is a transfer similar in form to government programs that make transfers from taxpayers to small businesses, lower income people, farmers, and other favored groups. In some cases, for example, people who already have a chronic condition, it is surely appropriate to make such transfers. In theory, doing so through charging the same premiums to high and low risks is a very inefficient way to produce this desirable transfer; it would be much better to raise the funds through general taxation (which, in contrast to premium averaging, would not drive the healthy out of insurance markets through higher premiums) and use the proceeds to lower premiums and/or raise net incomes for those with chronic conditions. Bringing about such transfers in any form will generally (and appropriately, if the transfers are appropriate) require government compulsion. Even in communities with high levels of social coherence, while a program of transfers through premiums may get general approval, it is generally difficult to persuade even community-spirited lower risks to continue to purchase insurance that offers them much less in benefits than they pay in premiums. Some modest amount of “community rating” may be appropriate and sustainable in insurance markets that are voluntary, but use of it on a large scale is generally an inefficient and often unfair way to achieve a good social objective.

Impediments to Private Insurance

There are some potential impediments to the emergence of voluntary health insurance markets. Some have already been hinted at; many governments impose heavy regulation on private or mutual insurers in order to produce across–group transfers that they are unwilling to make explicit through taxation. Regulation is also often used to require reserves to be held by insurers. For health insurers of any reasonable size, reserves generally do not need to be very large relative to premiums, as long as premiums themselves are properly set. This is because illness, like death, can be predicted fairly accurately for a population—“we don’t know who will get sick, but we know how many.” Unexpected epidemics of contagious disease obviously increase insurer risk, but even in developing countries most expenses do not go for contagious disease, and most contagious disease (including AIDS) can now be predicted. It is important for government to put in place the contract law that requires insurers who collect premiums to pay the benefits they promise; the fear of corruption and/or insufficient resources to pay claims makes people unused to insurance wary of it at first. Sometimes regulators use reserve regulation to limit insurer competition, but good regulation can definitely help the market.

The other major obstacle still requires more research to establish its importance, but it could represent serious market and political problems. As noted, the case for insurance is based on out-of-pocket payments in the absence of insurance. The higher such payments, the greater is the willingness of people to pay premiums as an alternative. It seems likely that, although households at many income levels make out-of-pocket payments, the average level of payments will rise with income. This occurs because good health and medical services are both normal goods and because people with growing income desire to spend some of that income on improving the quantity and quality of their survival. An implication, however, is that policies to be purchased voluntarily by lower income households or community need to be lower priced and less lavish than the policies that will appeal to higher income people. If people at all income levels are faced with the same premiums and the same policies, a kind of adverse selection will ensue in which the higher expected expense (though not necessarily sicker) higher income people buy coverage and then engage in behavior that makes the premium unattractive to those with more limited budgets. The upshot is that there needs to be some product differentiation. Doing so may present challenges to insurers, and also to politicians who may find unattractive the prospect of more frugal coverage for lower income people than for higher income people, even though such coverage is a vast improvement over the situation lower income people faced when their only choice was to pay out of pocket. The perfect may become the enemy of the good, and insurance may be inhibited.

The best solution to this problem, as well as to a number of the other problems mentioned above, is to have modest subsidies to voluntary insurance markets. If the subsidy exceeds the insurer administrative cost (which will probably be in the range of 10 to 30 percent of the premium), insurance will look like a bargain (compared to risking out-of-pocket payment) to even the least risk-averse household. Subsidies also mitigate the danger of adverse selection and overly aggressive risk rating—since insurers should be less willing to suspect people who are brought into the market by subsidies as being secret high risks. Some reallocation of even limited public funds to subsidizing real insurance seems to be a policy worth considering.

Data Needs

While much still has to be learned about the prospects for voluntary insurance, one key piece of information for both public and private planning is the level and distribution of out-of-pocket payments relative to income. People demand insurance voluntarily to protect against “financial vulnerability,” and public policymakers appropriately seek to encourage this protection and, as noted, appropriately subsidize it. But to understand which groups should have the highest priority for coverage and at the same time be willing to pay various amounts for it, we need measures of the impact of insurance. One impact measure is obvious to define though difficult to collect in practice: the “health impact” of coverage. This measure asks how much the health of a given subset of the population would improve because of insurance coverage, primarily (one assumes) because of a somewhat higher level of medical care consumption, but also perhaps of the large reduction in periods of financial distress. But it is also useful to obtain a measure of financial distress per se. One such measure that may be interesting to collect is the coefficient of variation of nonhealth consumption spending in the absence of insurance (and the effect that various types of coverage have on it). This measure calculates the standard deviation of nonhealth spending or wealth (with spending high for an uninsured household when no one gets sick but potentially quite low when someone does), and then divides or scales it by the mean level of nonhealth spending for that group. This measure automatically raises the level of financial distress for a given risk of out-of-pocket payments as income (and other consumption spending) falls, and it provides an indicator of the risk that risk-averse people are presumed to be willing to pay to avoid.

Other empirical work would also be helpful, especially that connected to the behavior of competing insurers and mutual insurers. The main point is that, at this stage of our knowledge, it does seem that there are positive prospects for markets in voluntary insurance. And, as developing countries do experience the across-the-board growth in real income they seek, their populations may be better satisfied with such insurance than with the old public system.

References

• Cutler, D., and Zeckhauser R., The anatomy of health insurance, in Culyer and Newhouse, eds., Handbook of Health Economics (North Holland, 2001)
• Nyman, J “Is Moral Hazard ‘Inefficient’? The Policy Implications,” Health Affairs 23(5), 194-199
• Pauly, M., “The Economics of Moral Hazard,” American Economic Review 58 , 1968.
• Pauly, M., and Herring B., Pooling Health Insurance Risk, American Enterprise Institute, 1999.
• Phelps, C., Health Economics, 3 rd Edition (Addison Wesley, 2002)
• Zweifel, P., and Manning W, “ Moral Hazard and Consumer Incentives in Health Insurance,” in Culyer and Newhouse, eds., Handbook of Health Economics (North Holland, 2001)

TABLE 1. OUT-OF-POCKET PAYMENTS AS PERCENTAGE OF NATIONAL HEALTH EXPENDITURES, SELECTED DEVELOPING COUNTRIES, 2001

Source: Neelam Sekhri and William Savedoff, “ Private Health Insurance: Implications for Developing Countries,” WHO Bulletin, February, 2005.

- top -

Related Event: Impact Conference on The Role of Private Health Insurance in Developing Countries , sponsored by The Wharton School of the University of Pennsylvania in collaboration with the International Finance Corporation and The World Bank -  March 15-16, 2005 .

Feature Reading:   Private Health Insurance in OECD countries - the OECD Health Project.