Contenu du sommaire

From Adverse Selection to Selection in Health Insurance: for a Change in Perspective by Agnès Couffinhal One of the commonly raised arguments to justify government intervention in health insurance is the presence of adverse selection. However, an economic analysis based on the Rothschild and Stiglitz model (1976) shows that insurers can manage to overcome information asymmetry by differentiating contracts. A summary of the main findings of this research highlights the conditions required to empirically check for adverse selection and presents the tests to establish whether it is found on the health insurance markets. At the end of this work, we cannot reject the hypothesis that adverse selection is present on the health insurance markets. However, the phenomenon appears to be fairly minimal. Adverse selection would also seem to be just one of the hindrances to the efficient running of the insurance markets and contract differentiation one of the possible solutions open to the insurer. Risk selection by the insurers themselves is another phenomenon, often confused with adverse selection, which can also undermine market operations. A clarification of the link between adverse selection and selection highlights the need to better incorporate this second phenomenon into an analysis of market functioning.

Health Expenditures : the Moral Hazard Assumption by Pierre- Yves Geoffard The analysis of many proposals to reform the French health insurance system, and in particular those focusing on co-payment schemes, is heavily dependent on delicate empirical measures of the elasticity of demand for health care. This paper offers a critical survey of the empirical literature on the moral hazard assumption, according to which more extensive health insurance coverage would lead to more spending. The paper begins by recalling that the Rand Health Insurance Study, a major study conducted in the United States in the 1970s, unambiguously concluded that the price elasticity of demand for medical care was negative, especially for ambulatory care. The paper then stresses the limitations of more recent studies undertaken on French data - limitations which are either inherent to the data or due to the econometric methods employed. In addition, evaluating the redistributive aspects of co-payments and their impact on access to care would require more accurate measures of cross price-income elasticity. Despite all these uncertainties, the lessons drawn from the existing empirical studies suggest that the current French co-payment scheme (the ticket modérateur system) is very far from being fair and efficient.

Health Care Demand Induced by French Self-Employed Doctors. A Microeconometric Study Based on Panel Data by Eric Delattre et Brigitte Dormont A much-debated hypothesis in health economics is the supply-induced demand hypothesis (SID), which posits that suppliers can manipulate demand for their services. This article tests for the existence of SID behaviour using unbalanced panel data covering 7,925 French self-employed Sector 1 and Sector 2 GPs and specialised doctors from 1979 to 1993. These data are representative of the population in question. Estimates provide strong evidence of SID behaviour in Sector 1. Such behaviour entails self-employed doctors restricting the rationing in the number of doctor-patient consultations to which they are subject when doctor density increases and offsetting this phenomenon by raising the volume of care provided per consultation. In Sector 2 (where doctors freely set their fees), doctors' reactions to density variations are compatible with the absence of SID behaviour. An increase in doctor density in Sector 2 brings about a drop in fees and an increase in activity. These reactions are consistent with the predictions of a monopolistic competition model of health care supply. To conclude, we find that a rise in doctor density brings about a rise in the volume of health care supplied in both Sector 1 and Sector 2. In addition, the fee reductions observed in Sector 2 do not help balance the social security system accounts, since these reductions only concern surcharges. Lastly, we establish that our estimated microeconomic elasticities create a massive macroeconomic drift in health care consumption, related to the increase in the number of doctors.

Opportunistic Behaviour of Patients and Doctors: the Contribution of Analyses by Health Care Episode by Pascale Breuil-Genier and Frédéric Rupprecht The "strategic" behaviours attributed to patients (adverse selection, moral hazard) and health care professionals (supply-induced demand) are not easily tested: although it is easy to find a link between health insurance and health care consumption, it is more difficult to single out the behavioural consequences of moral hazard and adverse selection. Similarly, a simultaneous increase in doctor density and in health care consumption may indeed be a consequence of induced demand (whereby doctors create the demand for their own services), but it may also result from an exogenous rise in health care needs combined with a simultaneous rise in the resources required to meet them. In an attempt to identify the determinants of patients' and doctors' behaviour more precisely, we analyse the series of consumption decisions taken for a given illness and a given patient by reconstructing health care episodes from the 1 99 1 -92 health survey. Our analysis shows that a patient' s decision to seek health care in case of sickness is strongly influenced by whether the patient possesses supplementary insurance coverage, but that this factor has no impact on the choice of the type of first consultation (general practitioner or specialist), nor on whether the patient subsequently seeks other consultations. The doctor density explains part of the choice between general practitioners and specialists for the initial consultation, but this variable has no impact on the number of first consultations, nor on whether the patient subsequently seeks other consultations.

Bootstrap Methods in Regression Models by Emmanuel Flachaire In practice, we rarely know the true probability distribution of a test statistic and we generally base tests on its asymptotic distribution. If the sample size is not large enough, the asymptotic distribution could be a poor approximation of the true distribution. Consequently, tests based on it could be largely biased. Bootstrap methods yield a more accurate approximation of the distribution of a test statistic than the approximation obtained from the first-order asymptotic theory. Moreover, they provide a way of substituting computation for mathematical analysis when it proves hard to calculate the asymptotic distribution of an estimator or statistic. In this paper, we present a general methodology of the bootstrap in regression models.