CAN WE DESIGN A MARKET FOR COMPETITIVE HEALTH INSURANCE?
CONCLUSION
Modern microeconomic theory has been used to design new markets of which the spectrum license
auctions are the outstanding success (McAfee, McMillan 1996; McMillan 2003). However, the more
complex the market and the broader the range of products and set of transactions it encompasses,
the more difficult it is to apply theories of competition. The sources of market failure in health care
and health insurance include information asymmetries, widespread uncertainties, systematic
differences in health status which affect both the demand and the ability to benefit form health care,
and broad equity concerns. On top of these, health care is expensive, typically accounting for over 9%
of GDP in developed nations.
The design of a competitive market for health insurance requires four major aspects to be
addressed. First, the appropriate risk adjustment mechanisms must exist, so that there is
competition on process and quality, not just for selection of enrollees. Second, there should be some
advantages from pooling the various sources of health care funds in terms of efficiency and cost
containment. Third, consumers must behave rationally and so they must be able to understand how
to judge differing insurance performance, use this information in forming their preferences, and act
to change insurers as a result. Fourth, insurers must also change their behaviour to become active
purchasers and negotiators on behalf of individuals as patients and the insured population as a
group.
On risk adjustment, although there has been substantial development in approaches particularly
over the last ten years, much health care expenditure remains unpredictable. Unpredictability is an
inherent problem in forecasting the demand for health care, if it were not there would be no need
for insurance. However, it is clear that variations in health care expenditure are not all driven by
health status, and so determining what variation is appropriate remains a crucial issue unless the
status quo is accepted and prediction is validated by observed expenditure. In this case, one might
well (and with much less effort) revert to full cost reimbursement. The central issue here, though,
is not whether perfect prediction is possible but whether information asymmetries are sufficient to
ensure that insurers will not be able to cream skim by selecting good risks. This condition has to be
met for competitive insurance to be efficient and to maintain equity. It seems that with the current
techniques concluding that information asymmetries do not matter is brave.
Funds pooling is being selectively applied far more widely than managed competition. Although this
still requires the use of risk adjustment, where this no choice of insurer by consumers, or enrollees
by insurance funds, there is limited opportunity for cream skimming. The evidence on funds pooling
is mixed. While it seems to hold some possibility for improving efficiency, it is clear, particularly from
the Australian experience, that achieving improved care and lower costs is not a simple matter of
rolling the various funding sources together.
Consumer behaviour is also a crucial issue as it is consumer decision making that will drive
efficiency in this competitive market. On the evidence to date, it can be said that consumers find
it difficult to understand health care insurance, difficult to assess insurers’ performance, time
consuming to have to make decisions on this issue, and more likely to stay with the status quo.
Similarly, the behaviour of insurers is also crucial in driving improved efficiency. There is no evidence
yet that Australian insurance funds have the capacity to become better purchasers of health care. Yet
multiple insurers are associated with higher administrative costs. At the same time, the monopoly
power of a single payer is diluted, with a consequent flow on into higher provider incomes.
So the answer the question posed in the title of this paper is negative. However, the issues involved
present a complex and varied research agenda.
ACKNOWLEDGEMENTS
My thanks to Glenn Jones, Elizabeth Savage and Randall Ellis for helpful comments on this paper.
Any errors remain the responsibility of the author.
14