Moral hazard and adverse selection in Australian private hospitals: 1989–1990

https://doi.org/10.1016/S0167-6296(02)00104-2Get rights and content

Abstract

The Australian hospital system is characterized by the co-existence of private hospitals, where individuals pay for services and public hospitals, where services are free to all but delivered after a waiting time. The decision to purchase insurance for private hospital treatment depends on the trade-off between the price of treatment, waiting time, and the insurance premium. Clearly, the potential for adverse selection and moral hazard exists. When the endogeneity of the insurance decision is accounted for, the extent of moral hazard can substantially increase the expected length of a hospital stay by a factor of up to 3.

Introduction

Different countries have different health insurance and health care systems. The US has a mainly private system with a small public (free) sector that acts as a safety net for the disadvantaged. On the other hand, the UK has a mainly public (free) system with a small private system for those prepared to pay for their health care. Australia is distinctive in having a mixed system with large private and public (free) sectors. In 1989–1990, around 44% of income units had private hospital insurance and 35% of hospital users used a private hospital. In addition, private hospital insurance is chosen at the individual or family level unlike in the US where health insurance is commonly a compulsory part of the employment contract.

There is an on going debate in the US concerning whether the public system should be extended and in the UK concerning whether the private system should be extended. A similar debate is going on in Australia and as in the US and UK is concerned with the appropriate sizes of the private and public sectors. Research that analyses the relationship between health insurance and the use of health care in the private and public sectors is needed to inform this debate.

There are two well-known difficulties associated with providing health insurance to individuals. These arise because the insurer does not know (a) the risk class to which a particular individual belongs and (b) the extent of the loss in well being an individual experiences. In a world with purely private health insurance, ignorance of an individual’s risk class leads to adverse selection as only high-risk individuals purchase insurance or no insurance market exists. Ignorance of the extent of the illness or the actual loss in well being leads to moral hazard as individuals, who have some control over the extent of their treatment and receive insurance payouts on the basis of their health care expenditure, over-utilize health services.

Private hospital insurance takes the form of a schedule of allowances for particular private hospital services that result in individuals with different insurance policies facing a different set of net prices. One reason insurance policies take this form is because expenditure on hospital services is observable by insurance companies while the individual’s health state vector is not. As previously mentioned, this introduces the possibility of moral hazard. Moral hazard can occur because the insurance policy alters the individual’s behavior in a way that decreases the expected profit of the insurance company. For example, (i) the existence of insurance might induce the individual to devote less resources to preventive care and so increase the probability of an insurance claim and decrease the expected profit of the insurance company, or (ii) the existence of insurance might induce the individual to purchase more private hospital services than are strictly needed to return the individual to a healthy state. This paper will be concerned with the latter case though both are manifestations of the same phenomena, namely, that private hospital insurance induces individuals to over-utilize private hospital services. In order to determine the appropriate mix of private and public health service and health insurance provision, studies need to be done to ascertain the extent of moral hazard and adverse selection under various insurance and health care regimes.

The aim of the empirical sections of this paper is to ascertain the extent to which the existence of insurance induces individuals to purchase more private hospital services than they would if they faced the true price of those services rather than the net price under insurance. As such, an indication of the extent of welfare loss that results from the price distortion is given.1 Moral hazard is present if the use of private hospital service, k is decreasing in the ‘net’ price of service, k.

A number of empirical papers have examined the determinants of an individual or family’s insurance choice, Ngui et al., 1989, Propper, 1989, Propper, 1993, Cameron et al., 1988; and Hurd and McGarry (1997). The general findings are that individuals or families are more likely to have private health insurance the greater is their income, the older they are, and if they are employed. Health status variables do not seem to impact on health insurance choice. In the UK, Besley et al. (1999), found that individuals were more likely to have private health insurance the longer were long-term waiting lists for treatment in public hospitals. A second group of papers take an individual’s or families’ insurance status as given and examine the determinants of health service use, Manning et al., 1987, Manning and Marquis, 1996; and Hurd and McGarry (1997). The general findings are that individuals and families consume more health services the more health insurance cover they have, the greater is their income, and to some extent the lower is their health status.

Health insurance choice depends, amongst other things, on expected future consumption of health services and so both the insurance and use decisions are interdependent. Of the papers previously mentioned, only the paper by Cameron et al. models the interaction between health insurance choice and health services use and so is the only paper that can address issues of adverse selection and moral hazard. In a more recent contribution, Lee (1993) also models this interaction. However, neither paper distinguishes between private or public hospital service use and given that much of the current debate concerns this issue, work needs to be done with this emphasis. This is particularly so in the Australian context because private hospital services are provided at different ‘net’ prices depending on the insurance cover held, whereas public hospital services are available free to all. As a result, moral hazard can only be identified amongst private hospital users as this is the only group that faces different ‘net’ prices.2

This paper is an attempt to model and empirically test the interaction between private hospital insurance choice and private hospital service use. A three-period model of health insurance and health care utilization is developed. In the first period, an income unit makes an insurance decision not knowing what its health state will be in the second period. In the second period, the income unit’s health state is realized. Given their first period insurance decision, the income unit chooses the quantity of private hospital services to consume. In the third period, after a waiting time has elapsed, the income unit can consume free public hospital services. The income unit’s insurance decision depends on its probability distribution over health states in period 2, the ‘net’ prices of private hospital services (given insurance), the waiting time for the various services to be available free in a public hospital, insurance premiums, and other socio-economic variables. The consumption of private hospital services in period 2 depends on the realization of the health state, the ‘net’ prices, the waiting time, and other socio-economic variables. The consumption of public hospital services in period 3 depends on the realized health state in period 2, whether private hospital services were consumed in period 2, and other socio-economic variables. The theoretical model produces reduced form equations for insurance choice and private hospital use that provide a rationale for the variables included in the empirical sections of the paper.

The empirical implementation of the model uses data from the 1989–1990 National Health Survey in Australia, but is hampered by a lack of data on insurance premiums, ‘net prices’, and waiting times. Chronic conditions and reason for hospital use are used as proxies for these variables. Insurance choice is estimated with a probit model. In general, it is found that an income unit is more likely to have private hospital insurance the greater is income, age of head of the income unit, health status as measured by chronic conditions, and other socio-economic variables. The significance of some of the health status variables contrasts with previous studies and suggests that adverse selection may be present.

The only quantity variable related to hospital use that appears in the data is the length of the hospital stay. This is the dependent variable in the duration model that is estimated. As insurance choice is endogenous in the model, the estimated probability of insurance is used as a regressor in these duration equations to provide consistent estimates of the moral hazard effect. In addition, reason for hospital service use and other socio-economic variables are also included as regressors. Separate equations are estimated for income units with different structures and at different stages of the life cycle and tests are performed to ascertain whether insurance choice is endogenous, that is, in the context of the theoretical model of this paper, whether adverse selection might be present.

In the private hospital duration equations, evidence of moral hazard was found to be significant for some, but not all income unit types. Where moral hazard was present, the effect of it on duration was quite large, increasing expected durations by a factor of up to 3. In contrast to previous studies, income was generally found to have a negative effect on duration, perhaps reflecting the opportunity cost of time. Finally, insurance choice was found to be endogenous for all but one income unit type suggesting adverse selection might be present.

Section snippets

The model

In this section, a theoretical model is developed which provides reduced form equations for insurance choice, hospital choice, and hospital use. These equations provide the rationale for the inclusion of particular variables in the estimating equations of Section 4. Variables, or proxies for them, are only included in the estimation if the theory suggests they should be included.

The health insurance system

In 1984, a universal system of health care subsidies, known as Medicare, was introduced into Australia. These subsidies are based on a schedule of fees known as the Medical Benefits Schedule (MBS). The government adjusts the MBS annually. Under Medicare, individuals who choose to be treated in a public hospital as a public (Medicare) patient are treated by doctors and specialists nominated by the hospital. These services are free of charge whether or not the individual has private hospital

The empirical modeling strategy

The primary aim of this paper is to ascertain the extent of moral hazard in the provision of private hospital services in Australia.11 In the process, the existence of adverse selection is ascertained as well. Reduced form hospital service use Eq. (6), and its many person income

Private hospital insurance choice

The single and couple income units were separated into young (head aged less than 50 years) and old partitions and separate probit models were estimated for each, the rationale being that young and old, singles and couples are very distinct groups with different behavior. This was tested explicitly by pooling the data and performing a likelihood ratio test, the results of which appear in Table 3. The χ2-statistic of 952.8 exceeded the critical value of 380.5, indicating the model with income

Conclusion

The appropriate size of private and public health sectors and the design of insurance policies depend on the extent of adverse selection and moral hazard. There was some evidence of adverse selection for all but one income unit type. Except for the very old, income units act strategically in purchasing insurance in the sense that, insurance choice is an endogenous variable in hospital use equations. If income units expect to be heavy users of private hospital services, they purchase private

Acknowledgements

We wish to thank Kevin Werksman for invaluable research assistance, Denzil Fiebig, Philip Haywood, participants in the 21st Annual Conference of the Australian Health Economics Society, and participants in seminars at a number of institutions where this work has been presented.

References (21)

There are more references available in the full text version of this article.

Cited by (57)

  • Study of the ex-post moral hazard of laboratory services in Iran

    2021, Health Policy OPEN
    Citation Excerpt :

    On average, the utilization rate of the laboratory services for people with insurance coverage was two times higher than uninsured people (AOR: 2.03; CI: 1.25–3.29; P-Value: 0.001), which a significant positive relationship was found in almost all socio-economic subgroups. This result agrees with the findings of other studies [6,9,11,13,30] which showed that insurance coverage motivates people to consume more than others. Given that the Iranian health system does not have managed care, such a significant moral hazard was expected.

  • Hips and hearts: The variation in incentive effects of insurance across hospital procedures

    2014, Journal of Health Economics
    Citation Excerpt :

    This is perhaps not surprising given the variety of identification strategies used. Savage and Wright (2003) and Lu and Savage (2006) consider selection on observables only. Several studies have used instrumental variables to separate causal effects from selection.

View all citing articles on Scopus
View full text