Positive Economics

Positive Economics in Health Economics

Positive economics is the branch of economic analysis that explains and predicts behaviour and outcomes, so its claims can in principle be checked against evidence. In health economics it answers questions such as how far use of medical care falls when patients pay part of the bill, or how much sugar households bought after the UK soft drinks industry levy came in. Those answers carry no verdict on whether a policy is good, which is the task of welfare economics and other normative frameworks. The distinction matters in health technology assessment, where an estimate of a treatment's costs and QALYs and a judgement about whether they are worth paying for rest on different kinds of argument.

Keynes's division and Friedman's account of a positive science

Milton Friedman opened his 1953 essay on the methodology of positive economics by quoting the division John Neville Keynes drew in 1891 between three kinds of inquiry. A positive science is "a body of systematized knowledge concerning what is", a normative or regulative science discusses "criteria of what ought to be", and an art is "a system of rules for the attainment of a given end".

Friedman then set out what positive economics is for. He wrote that it "is in principle independent of any particular ethical position or normative judgments", that its task is "to provide a system of generalizations that can be used to make correct predictions about the consequences of any change in circumstances", and that its performance is to be judged by the precision, scope and conformity with experience of the predictions it yields. In his view the ultimate goal of a positive science is a theory or hypothesis that yields valid and meaningful predictions about phenomena not yet observed.

A statement is positive because of its form, whether or not it turns out to be correct. "Higher cost sharing reduces spending on medical care" is a positive claim that evidence could refute. "Patients should pay part of the cost of their care" is a normative claim, because no observation alone can settle it.

Positive questions answered by health economics research

Two well-known studies show positive analysis in health. Each estimates how people respond to a change in prices, and neither asks whether the change was desirable.

The RAND Health Insurance Experiment was a randomised experiment in which insurance plans differed in cost sharing, the portion of the bill the patient pays. Manning and colleagues reported in 1987 that a catastrophic insurance plan reduced expenditure by 31 per cent relative to care with a zero out-of-pocket price, and that the price elasticity of demand for medical services was approximately minus 0.2. They also rejected the hypothesis that less favourable coverage of outpatient services raises total expenditure. That is hypothesis testing in the sense Friedman meant, with a prediction set against data that could contradict it.

The UK soft drinks industry levy gives a second example. Rogers and colleagues described it as a two-tier tax on manufacturers, announced in March 2016 and implemented in April 2018, with drinks containing at least 8 g of sugar per 100 mL taxed at GBP 0.24 per litre and a lower tier at GBP 0.18. Their controlled interrupted time series of household purchasing found that, one year after implementation and compared with the counterfactual, the volume of all soft drinks bought rose by 188.8 mL per household per week (2.6 per cent) while the sugar in them fell by 8.0 g (2.7 per cent). Purchases of confectionery and alcoholic drinks did not increase. The authors concluded that further studies should determine whether and how these apparently small effects translate into health outcomes, which is itself a further positive question.

Friedman noted that economists must generally rely on the "experiments" that happen to occur, and he judged the inability to run controlled experiments no fundamental obstacle to testing hypotheses, although such evidence is far harder to interpret. A tax introduced for public health reasons is such an event, and the controlled design builds the counterfactual that the comparison needs.

Prediction as the test of a positive theory

The most debated part of Friedman's essay concerns how a positive theory should be judged. He held that a theory, viewed as a body of substantive hypotheses, "is to be judged by its predictive power for the class of phenomena which it is intended to 'explain'", and that the only relevant test of a hypothesis is comparison of its predictions with experience. Evidence can never prove a hypothesis. It can only fail to disprove it, which is what is meant, somewhat inexactly, when a hypothesis is said to be confirmed.

From this he drew his conclusion about assumptions. The relevant question to ask about the assumptions of a theory, he wrote, is "not whether they are descriptively 'realistic,' for they never are, but whether they are sufficiently good approximations for the purpose in hand", and that question can only be answered by seeing whether the theory yields sufficiently accurate predictions. A demand model that treats patients as if they respond smoothly to price can therefore be useful even though no patient reasons that way, provided its predictions of how use of care changes are accurate enough for the decision at hand.

Decision models in health technology assessment face the same question. A model simplifies disease and treatment pathways, so its structural assumptions are never literally true. Comparing its outputs with observed data not used to build it is part of model validation, and reflects the positive view that a simplification earns its place by how well it predicts.

How positive findings feed normative decisions in health

Positive evidence alone settles no policy question, yet no policy can be decided without it. Friedman argued that normative economics cannot be independent of positive economics, because any policy conclusion "necessarily rests on a prediction about the consequences of doing one thing rather than another". He added that two people may agree on the consequences of a piece of legislation and still disagree about whether it is desirable on balance. He judged that, in the Western world of his day, most policy disputes among disinterested citizens stem from different predictions, and called that judgement itself a positive statement to be tested on the evidence.

Welfare economics makes the same boundary explicit from the other side. Brouwer and colleagues, reviewing welfarism and extra-welfarism in health economics, quote Boadway and Bruce's statement that judging the consequences of an event for economic welfare means going beyond positive economics, "which is concerned with the effects of an event on objectively measurable economic variables, such as price and quantity", and assessing desirability against explicitly stated ethical criteria. On this reading the RAND and levy studies supply the effects, and a social welfare function or a decision maker's stated objective supplies the criteria.

NICE's manual for technology appraisal and highly specialised technologies guidance (PMG36) keeps the two roles apart in its own process. Section 6.1.7 says the committee considers advice on the appropriate approach to "making scientific and social value decisions". In section 6.3 the committee compares the most plausible ICER with a range of GBP 25,000 to GBP 35,000 per QALY gained. Section 6.3.1 links the maximum acceptable ICER to the opportunity cost of programmes displaced within the NHS's fixed budget, so even the threshold combines a positive question (what health the displaced spending would have produced) with the normative choice to weigh a technology's health gains against the health it displaces.

Illustrative example of an appraisal split into positive and normative steps

The figures below are illustrative and describe no real technology. A new treatment costs GBP 9,000 more per patient than current care and gains 0.3 QALYs per patient. The first steps are positive, in that they report what the evidence and the model predict.

$$ICER = \frac{\Delta C}{\Delta E} = \frac{9{,}000}{0.3} = 30{,}000$$

Here $\Delta C$ is the incremental cost per patient in GBP and $\Delta E$ is the incremental health gain per patient in QALYs, so the ICER is GBP 30,000 per QALY gained. Better data could change the figure, which is what makes it a positive claim.

The net health benefit at each end of the NICE range shows what the treatment would add to, or take from, population health once displaced care is counted.

$$NHB_{25} = \Delta E - \frac{\Delta C}{\lambda} = 0.3 - \frac{9000}{25000} = 0.3 - 0.36 = -0.06$$

Here $NHB$ is net health benefit in QALYs per patient and $\lambda$ is the value placed on a QALY gain, GBP 25,000 in this line, so the treatment would cause a net loss of 0.06 QALYs per patient treated.

$$NHB_{35} = 0.3 - \frac{9000}{35000} = 0.3 - 0.257 = 0.043$$

Here $\lambda$ is GBP 35,000, and the treatment gives a net gain of about 0.043 QALYs per patient. Both statements remain positive, because each predicts health consequences given a stated value of $\lambda$.

The final step is normative. With an ICER of GBP 30,000 the treatment sits inside the range where PMG36 asks the committee to refer explicitly to the factors listed in its section 6.3.5, which are the certainty around the ICER, benefits and non-health factors the estimate does not capture, and health inequalities. Whether the health gained is worth the health displaced then depends on how much weight the committee gives to those factors. Two committees with identical numbers could reasonably reach different recommendations, and neither would have made a factual error. If the treatment were found to gain 0.25 QALYs instead of 0.3, the ICER would rise to 9,000 / 0.25 = 36,000 and the case would move above the range, which shows how a change in the positive evidence can shift a normative decision without any change in values.

Boundaries with normative economics, value judgements and welfarism

The counterpart of positive economics is normative economics, which asks what ought to be done, and the gap between the two is the space occupied by value judgements. Hausman's Stanford Encyclopedia entry notes that normative economics is standardly called welfare economics.

Utility sits on both sides of the line. Brouwer and colleagues point out that utility can be given a normative use or "a merely predictive or descriptive use as an index of choice", and, citing Van Praag, that the concept has been central to both normative and positive theory. In choice theory used positively, utility maximisation is a model that predicts behaviour. In welfarism, the same utility becomes the measure of how well off people are, and that step is a value judgement. Welfarist and extra-welfarist health economics differ on that normative step, over which good should be maximised, while drawing on the same positive evidence.

Positive economics is broader than econometrics, since it includes the theory that generates predictions as well as the statistical methods that test them. A study that reports effects and then recommends a policy has moved into normative ground at the point of recommendation.

Limits of the positive and normative split

Hausman reports that most economists and methodologists believe there is a reasonably clear distinction between facts and values, and that most of economics should be regarded as a positive science that helps policy makers choose means without bearing on the choice of ends. He calls this view questionable for several reasons. Economists have to interpret and articulate the incomplete goals and constraints that policy makers give them, economic science is itself a human activity governed by values, and much of economics is built around a normative theory of rationality.

These points have health economics counterparts. Choosing to measure a levy's effect through purchases instead of health, or QALYs as a model's outcome, involves judgements about what matters before any data are analysed. On Friedman's treatment of assumptions, Hausman notes that almost all commentators have been critical, and that Friedman refers to several different things as assumptions and means several different things by calling them unrealistic.

Several misunderstandings recur when the distinction is applied in health economics.

• Positive means true. A positive statement is one that evidence could confirm or refute, and the RAND experiment, for instance, rejected the positive hypothesis that less generous outpatient cover raises total spending.
• A cost-effectiveness result is a recommendation. An ICER or net health benefit is a positive output conditional on its inputs, and turning it into a funding decision requires a normative criterion such as a cost-effectiveness threshold.
• Normative disputes are only about values. Friedman argued that many policy disagreements, such as those over minimum-wage legislation, turn on different predictions, which better positive evidence can narrow.

Sources

• Brouwer WBF, Culyer AJ, van Exel NJA, Rutten FFH. Welfarism vs. extra-welfarism. Journal of Health Economics. 2008;27(2):325-338. doi:10.1016/j.jhealeco.2007.07.003. Author version: https://eprints.whiterose.ac.uk/id/eprint/3789/
• Friedman M. The methodology of positive economics. In: Friedman M. Essays in Positive Economics. Chicago: University of Chicago Press; 1953. p. 3-43. Course copy (pp. 3-16 and 30-43): https://personal.lse.ac.uk/robert49/teaching/ph232/pdf/friedman.pdf
• Hausman DM. Philosophy of economics. In: Zalta EN, editor. The Stanford Encyclopedia of Philosophy. First published 12 September 2003; substantive revision 4 September 2018. https://plato.stanford.edu/entries/economics/
• Manning WG, Newhouse JP, Duan N, Keeler EB, Leibowitz A, Marquis MS. Health insurance and the demand for medical care: evidence from a randomized experiment. American Economic Review. 1987;77(3):251-277.
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Published 31 January 2022, last updated 13 October 2026. https://www.nice.org.uk/process/pmg36/chapter/committee-recommendations-2
• Rogers NT, Pell D, Mytton OT, Penney TL, Briggs A, Cummins S, et al. Changes in soft drinks purchased by British households associated with the UK soft drinks industry levy: a controlled interrupted time series analysis. BMJ Open. 2023;13(12):e077059. doi:10.1136/bmjopen-2023-077059