treatment)is supplied. In the case of medical care, however, an additional differential exists in that the doctor will usually also have a substantially greater idea of the usefulness to the patient of alternative'products whereas with other commodities the consumer is presumed to know this better. As the purveyor of a service, this puts the doctor in a special position ince he presumes to tell the patient what he needs as well as supply ing those needs. In an ideal world, as Arrow has pointed out [1], one can conceive of devices which would enable such risks to be insured optimally One possible schemel would be to pay doctors by results so that they had an appropriate incentive for using their superior information as efficientl as possible, with doctors transferring the risk of failure to insurance agencies. The reasons why such mechanisms do not operate are, however easy to identify. The major one must certainly be the enormous costs of discovering whether treatment had been 'successful. How successful is eatment that saves a person's life but renders him permanently disabled How does one measure a treatment that relieves but does not eliminate particular set of symptoms? How successful is a treatment that prolongs life for two months or three, or four? Instead of this kind of mechanism for reducing the costs of uncertainty for the patient, most societies have evolved what is usually called 'the doctor-patient relationship,, the special trust relation between doctors and patients which gives the medical profession a high social status in the unity as trustworthy and partial. It is the same ethic, one may argue, that calls for an absence of any obvious commercialism in the physician's dealings with his patients. It is an ethic which is as old as Hippocrates, and one which appears to be commonly shared across different societies and across different institutional frameworks, in both the market-type health systems and the NHS Whether a periodic bill from one's family physician or the periodic spectacle of an entire profession threatening disruptive action in support of a pay claim is more conducive to this special relationship is difficult to say, and there seems to be no obvious grounds for a priori choice between alternative institutions here 6. a common complaint against health care that is organized in a market is that actuarially fair insurance is not available, apart from the problems due to pooling elements, because charges are loaded by administrative costs. Clearly, if marginal social costs are incurred in administering which social welfare could be increased: assuming a negative sloped demand curve for risk avoidance, too many people would be insured. The absence of an actuarially fair'price cannot therefore be held to be an inefficiency of the market save in comparison with the hypothetical ideal world where 1 Suggested by Arrow, op cit, P. 964. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.Reproduced with permission of the copyright owner. Further reproduction prohibited without permission