Self-selection and moral hazard in Chilean health insurance

https://doi.org/10.1016/S0167-6296(02)00121-2Get rights and content

Abstract

We study the existence of self-selection and moral hazard in the Chilean health insurance industry. Dependent workers must purchase health insurance either from one public or several private insurance providers. For them, we analyze the relationship between health care services utilization and the choice of either private or public insurance. In the case of independent workers, where there is no mandate, we analyze the relationship between utilization and the decision to voluntarily purchase health insurance. The results show self-selection against insurance companies for independent workers, and against public insurance for dependent workers. Moral hazard is negligible in the case of hospitalization, but for medical visits, it is quantitatively important.

Introduction

We study the interaction between the purchase (and design) of health insurance and health care service utilization in the Chilean health system. The relationship arises from the effect of the purchase/design of insurance policies on utilization through moral hazard and consumer self-selection. Significant effects of insurance on the demand for health services (moral hazard) have been found in the literature (see Newhouse (1993) for the US; Cameron et al. (1988) for Australia; and Bertranou (1998) for Argentina).1 Regarding adverse selection, Cutler and Zeckhauser (1998), Cutler and Reber (1998), Cameron et al. (1988), and Bertranou (1998), find significant adverse selection among consumers of the various health insurance plans they analyze.

Our study of the behavior of independent workers, with and without insurance, replicates what is traditionally done in the literature. The usual finding is that those who voluntarily purchase health insurance have a higher health risk than an average individual in the population, and consume more health care services than if they were not insured.

Dependent workers2 must spend 7% of their taxable income on health insurance, and must choose between purchasing public insurance (from the National Health Fund or Fondo Nacional de Salud (FONASA)); or private insurance (from one of several Instituciones de Salud Previsional (ISAPREs)).3 Among this group of workers, we look for self-selection and over-utilization associated with the purchase of one kind of insurance, as opposed to the other. To measure self-selection, we analyze the relation between the probability of affiliation to one kind of insurance, and the observable and non-observable characteristics of the beneficiaries. To estimate over-utilization, we compare the utilization of health care services with public and private insurance.

Two characteristics of the Chilean health insurance system allow us to presume the existence of self-selection. In both cases, self-selection is expected to operate against public insurance. The first consists of the difference in premiums in public and private insurance policies (see Aedo and Sapelli, 1999). Public insurance sets its premiums as a percentage of income, while private insurance uses risk rating. Private insurance institutions are allowed, by law, to adjust premiums according to age, sex, and number of dependents. Thus, public insurance is relatively more attractive for those with higher risk factors, since the effective price of public insurance is lower. Thus, those who purchase public insurance, on average, are expected to exhibit higher observable risk than those who purchase private insurance.

The second characteristic that leads to self-selection against public insurance is the free implicit insurance offered by public insurance. First, until recently, public insurance was not able to determine whether or not a patient was affiliated to a private insurance health plan. This provided an incentive for private insurance beneficiaries to make (unauthorized) sequential use of both types of insurance, according to their relative price and quality. Second, private insurance beneficiaries are permitted to switch back to public insurance; thus, public insurance operates explicitly as a last-resort insurer.4 This implies that public insurance beneficiaries, on average, should exhibit greater non-observable risk than private insurance beneficiaries.

Section snippets

The data

The source of our data is the 1996 CASEN Survey.

The model

Cameron and Trivedi (1991) and Cameron et al. (1988) model the individual demand for health insurance and health care services, taking the interdependence between these decisions into account, within a framework of intertemporal (two-period) utility maximization under uncertainty. In the initial period, individuals (or family groups) choose the health insurance plan, without knowledge of the health status which will determine their demand for services during the period to follow. Individuals

Empirical methodology

Our dependent variable is the number of services (physician visits and days of hospitalization) consumed by the household head in the 3 months prior to the survey. We correct for possible selection bias and estimate the probability of purchasing each type of insurance. The choice of health insurance is made by the household head, and is assumed to have been made prior to the 3-month period in question.

Results

We analyze self-selection and over-utilization in six instances: the utilization of physician visits and hospital days within three population groups, holders of a private insurance policy and of public insurance in the case of dependent workers, and members of a health insurance plan (be it private insurance or public insurance) in the case of independent workers. The signs of the estimated coefficients are consistent with those predicted, except for the fact that some variables are not

Conclusion

For independent workers, we find that insurance plans in Chile receive an adverse selection of the population, and their affiliates consume more than they would have if they had not purchased insurance (moral hazard).

When analyzing the choice between public insurance and private insurance for dependent workers, we find that self-selection is present against public insurance based on variables which are observable (by private insurance and researchers). This is to be expected, given the

References (13)

  • A.C Cameron et al.

    The role of income and health risk in the choice of health insurance, evidence from Australia

    Journal of Public Economics

    (1991)
  • C Aedo et al.

    El sistema de Salud en Chile: readecuar el mandato y reformar el sistema de seguros

    Estudios Públicos

    (1999)
  • F Bertranou

    Health care services utilization and health insurance coverage: evidence from Argentina

    Revista de Análisis Económico

    (1998)
  • A.C Cameron et al.

    A microeconomic model of the demand for health care and health insurance in Australia

    Review of Economic Studies

    (1988)
  • Cutler, D., 1994. Market Failure in Small Group Health Insurance. NBER Working Paper no....
  • D Cutler et al.

    Paying for health insurance: the trade-off between competition and adverse selection

    The Quarterly Journal of Economics

    (1998)
There are more references available in the full text version of this article.

Cited by (51)

  • Private health insurance, healthcare spending and utilization among older adults: Results from the Brazilian Longitudinal Study of Aging

    2022, Journal of the Economics of Ageing
    Citation Excerpt :

    For models of healthcare utilization, we compare probit (for any utilization) and OLS regression (for the intensity of utilization among those with any) with results obtained from two-stage-least squares (2SLS) regression models that account for the endogenous nature of having a private health plan (Cameron and Trivedi, 2005). Candidate instrumental variables to identify selection into private health plan coverage are based on previous literature (Galárraga et al., 2010; Hidayat et al., 2004; Jung and Streeter, 2015; Sapelli and Vial, 2003; Waters, 1999) and, after testing a range of candidates, the final instrumental variables include 1) the mean value of the respondent’s vehicle and 2) neighborhood social capital, measured by responses to the question of whether people in one’s neighborhood can be trusted (always, almost always, almost never, never). The argument for the first instrumental variable is that it represents a preference for spending on luxury goods (since the public system is readily available for use without additional expenditures).

  • Internet use and insurance growth: evidence from a panel of OECD countries

    2020, Technology in Society
    Citation Excerpt :

    Individuals with higher education generally have higher incomes and tend to purchase life insurance. For non-life insurance, Sapelli and Vial [71] found a positive link between the probability of purchasing private insurance and education. Consequently, the level of education potentially increases the demand for life insurance and for comprehensive non-life coverage.

  • Health Insurance: Economic and Risk Aspects

    2015, International Encyclopedia of the Social & Behavioral Sciences: Second Edition
View all citing articles on Scopus
View full text