The starting point of this paper is the analysis of the economic peculiarities of medical care. Medical practice is full of lessons on the special character of medical care. Any physician or health care manager is able to describe a set of properties distinguishing the activity of a health professional from those of other workers in more conventional activities.
Perhaps one of the most often perceived differences in the sector is the lack of consumers' ability to choose their own basket of goods and services due to their lack of information for decision-making. It is useless to ask a patient whether he/she would prefer chemotherapy instead of radiotherapy in case he/she is able to afford only one of the alternatives. By the same token, it would also be useless to ask the patient whether he/she prefers an immunological examination or a magnetic resonance scan.
Such a situation is aggravated because the decision must be made at a time of personal or family distress – an illness threatens to take the patient's life or that of a loved one. For this reason, in contrast to other items whose consumption can be postponed, the consumer will make a heroic effort and will certainly not hesitate to try all available alternatives. This breaks one of the basic rules for adequate market resource allocation, as there is no symmetry of information. One side – the service provider – supposedly holds the information by accumulating esoteric knowledge  that is inaccessible to the other side.
Another important difference resides in the existence of limits to "rationalizing the production" as in other economic sectors. Every emergency service must always have a neurosurgeon available, even if traumas requiring nocturnal procedures rarely occur. It would be inadmissible to deny treatment to a person with multiple traumas on the basis of statistical evidence that his or her condition accounts for a small incidence of traumas and that it would not be economically justified to keep staff for this purpose. As a further example, even though snakebite is becoming increasingly rare, every health care unit must keep antivenin in stock – duly cooled and periodically checked – most of which is discarded.
Possible functions of standardized production are one more example of major differences. In industry, such possibilities stem from a relative standardization and monotonous industrial processes – the inputs are constant, the processes are repetitive and the outcome is always expected and predicted. In the health sector things are not as simple, as inputs and processes cannot be fully standardized.
An extensive literature shows that distinct health agents or even distinct medical "cultures" come to entirely different diagnoses regarding similar groups of patients submitted to them and prescribe utterly different therapies. According to the patient's social, economic and cultural level, the same disease may be given a totally different treatment. A verminosis, for example, may be treated with broad-spectrum drugs instead of a simple faeces examination or it could be treated using a series of examinations that could lead to a similar therapy (broad-spectrum drugs).
All this happens because there is much subjectivity in the health labour process, which remains basically artisanal, grounded on sight, touch and smell instead of being readily captured in a defined algorithm. Although a great deal of information may be objective – blood pressure, coronary permeability, electrocardiographic waves – other data or even the interpretation of those considered objective are quite subjective. There is always a feeling, a clinical look, based on subjective hints, that may trigger a dozen supplementary examinations. In addition, there is a sort of association between the physician and the patient that has been ontologically constructed during human history, ever since someone's sufferings were first alleviated by someone else.
Hindrances to standardization, however, should be qualified. In some health areas, there are – within certain limits – standard procedures, such as: laboratory procedures; highly standardized procedures in hospital sectors, resulting in serial surgeries; in public health, a number of advancements were achieved by means of standard protocols (e.g. for the treatment of diarrhoea and acute respiratory infections); medical procedures can be partially standardized through detailed classification (as for example, the Diagnosis Related Groups – DRGs).
Such elements, however, encounter a significant limit – the clinical contact controls all other procedures. And the clinical contact is based on much more shifting variables in which there are imponderables and from which it is difficult to construct closed algorithms.
Such observations constitute a source of empirical elements for a substantive theoretical elaboration of the peculiar characteristics of medical care as an economic category. Such peculiarities thus present a series of limitations to the market's ability to provide such services in a quantitatively and qualitatively adequate manner. Arrow had the merit to present this discussion based on an elaborate economic conception.
Arrow's paper  is pedagogically structured. Firstly, the market's working is described in accordance with the neoclassical economic theory that it should lead to the occurrence of a competitive equilibrium and optimum status. Then, the author poses hindrances to medical care marketability. The first basic difference from common commodities is related to the risk-bearing associated with medical care – "to a great extent a disease is an unpredictable phenomenon". A subtle outcome arises: "When there is uncertainty, information or knowledge becomes a commodity ..." But information, in the form of skilled care, is precisely what is being bought from most physicians and, indeed, from most professionals.
The elusive character of information as a commodity suggests that it departs considerably from the usual marketability assumptions about commodities." (p. 183). Thus, he sustains that "all the special features of this industry, in fact, stem from the prevalence of uncertainty". Finally, Arrow considers that "when the market fails to achieve an optimum state, society will, to some extent at least, recognize the gap, and non-market social institutions will arise attempting to bridge it" (p. 184). Therefore, such unique characteristics call for "a special place for medical care in economic analysis" (p. 186).
First, the demand for it is irregular and unpredictable (contrary to the demand for food and clothing, for example). Another important aspect is that the demand for health care is usually associated with an assault on personal integrity. A disease is not just a risk, but a risk associated with a cost per se (reduction or loss of labour capacity, even temporary, obviously affecting one's earning capacity), which is distinct from the specific cost of medical care (p. 187). Furthermore, there is an "opportunity cost": the time lost for earning in the labour market while undergoing treatment.
Second, the physician's behavior cannot be fully known in advance – medical care is one of the activities of which "the product and the activity are identical". In these cases, the commodity purchased cannot be tested before consuming and "there is an element of trust in the relation". The physician's behavior "is assumed to be ruled with concern to the patient's welfare, which is not expected from a salesperson". "In Talcott Parsons's terms, there is a 'collectivity-orientation', which distinguishes medicine and other professions from business, where self-interest on the part of participants is the accepted norm" (p. 187).
Other typical differences from business people would be that: advertisement and price competition are virtually absent among physicians; counseling by doctors concerning other treatment is supposedly given without self-interest; and treatment should be oriented by the needs of the specific case and not restrained by financial considerations (p. 187). Finally, resource allocation in this area is significantly affected by "ethic compulsion" (p. 188).
Third, there is uncertainty concerning the output – recovery from a disease is as unpredictable as its incidence: "Because medical knowledge is so complicated, the information possessed by the physician as to the consequences and possibilities of treatment is necessarily very much greater than that of the patient, or at least so it is believed by both parts. Further, both parts are aware of this informational inequality, and their relation is colored by this knowledge" (p. 190). Information asymmetry proves to have a crucial weight in the physician-patient relationship.
A relevant aspect should be added here: although the physician knows better than the patient, his/her knowledge is still extremely limited, given the extensive ignorance areas of scientific knowledge concerning the working of a human body, aetiology of a number of diseases, etc. Thus, there is a great difference between purchasing a chair from a cabinetmaker and a medical appointment – the cabinetmaker knows how to make the chair that the client desires, but the doctor has every chance of knowing very little about how to treat the patient or even being unable to do so.
Fourth, supply conditions are uncertain – entry into the market is not free, which restrains the assumption of full mobility of the production factors. Doctors must be licensed to provide medical services. Furthermore, medical education costs are high and apparently are only partially incurred by the student (p. 191), which means another dissociation from the requirements for the working of competitive markets, i.e. private benefits granted to students after graduation exceed private costs. Arrow associates the high costs of medical education with the quality requirements imposed by the American Medical Association (AMA) since the Flexner Report (p. 191–192) became known.
Fifth, pricing is uncertain – this topic is not usual in economic texts. There is an extensive price discrimination according to the patient's income, at one extreme reaching zero cost for indigent patients. Price competition is strongly disapproved.
Sixth, there exist indivisibilities – specialists and some sort of equipment constitute significant indivisibilities (p. 194).
As risk (of the disease and its treatment outcome) determines the medical care "market", Arrow calls for the possibility of an insurance market that might be able to organize and distribute such risks. If such a market were possible, the problems identified so far could be solved. However, the analysis of a hypothetical ideal insurance market (pp. 199–207) points to a set of problems:
uncovered population segments (unemployed population, aged people, chronic disease sufferers, low-income population);
differentiated risk pooling (if the market is competitive, high-risk individuals will tend to have to pay higher premiums);
existence of moral hazard, to the extent that individuals covered by the insurance plans would tend to overuse them;
adverse selection – an aspect pointed out by Akerlof , since in case premiums increased so as to cover elderly people, individuals bearing higher risks would be precisely those who would tend to agree to pay for the insurance (therefore, the individuals selected by the insurer would be exactly those with greater health problems: the costlier individuals for the system);
uninsurable diseases (e.g. AIDS at the epidemic outset);
existence of interdependent probabilities (when a problem affecting a person reaches other people, as in epidemics) ;
high administrative costs (which would serve as an argument in favor of quite generalized plans, particularly the compulsory ones).
Such problems determine the market's incapacity to provide a comprehensive insurance for medical care (p. 210). In a recent interview, Arrow  maintained the diagnosis of the 1963 paper, suggesting that funding the system through contributions to a centralized system "can be accomplished in a cheaper way than having many competitive insurance plans". An interesting aspect of this interview is that it confirms the basic elements of a diagnosis accomplished more than 20 years ago.
In the postscript to his original paper , Arrow highlights two aspects as follows: the failure of the market in developing policies of insurance against uncertainty has stimulated the emergence of many social institutions; in these institutions, the usual market premises are "contradicted to a certain extent". He notes that this is not an exclusive problem of the medical profession. All through the text, he emphasizes the role of nonprofit institutions in the sector (e.g. p. 191).