A Framework for the Fair Pricing of Medicines

Abstract

As high-cost medicines put increasing pressure on public health care budgets, the need to identify ‘fair’ prices for medicines has never been greater. This paper proposes a framework, built upon fundamental economic principles, that allows for the consideration of ‘fair’ prices for medicines. The framework incorporates key considerations from conventional supply-side and demand-side approaches for specifying a cost-effectiveness ‘threshold’, including the health opportunity cost borne by other patients ($k$) and society’s willingness to pay for marginal improvements in population health ($v$). The costs incurred by manufacturers in developing and supplying new medicines are also considered, as are the incentives for manufacturers to strategically price up to any common price per unit of benefit (cost-effectiveness ‘threshold’) specified by the payer. The framework finds that, at any ‘fair’ price, a medicine’s dynamically calculated incremental cost-effectiveness ratio (ICER) lies below $k$. When pricing medicines collectively, the framework finds that a common price below $k$ is required to maximize population health (consumer surplus) or to maximize total welfare (consumer and producer surplus). This framework has important policy implications for payers who wish to improve population health outcomes from constrained health care budgets. In particular, existing approaches to ‘value-based pricing’ should be reconsidered to ensure that patients receive a ‘fair’ share of the resulting economic surplus.

FormalPara Key Points for Decision Makers

A framework is proposed that allows for the consideration of ‘fair’ prices for medicines.

At any ‘fair’ price, a medicine’s dynamically calculated incremental cost-effectiveness ratio (ICER) lies below a supply-side estimate of health opportunity cost ($k$).

When pricing medicines collectively, total welfare is maximized when the common price per unit of benefit is below $k$. Maximizing population health (consumer surplus) requires a common price even further below $k$.

Existing approaches to value-based pricing should be reconsidered to ensure that patients receive a ‘fair’ share of the resulting economic surplus.

1 Introduction

A growing challenge faced by public health care payers around the world is deciding which medicines to approve for reimbursement, and at what price. As higher-cost medicines put increasing pressure on public health care budgets [1], the need to identify ‘fair’ prices for medicines has never been greater.

Economic evaluations of medicines form an important component of health technology assessment (HTA) processes [2]. These are increasingly used during negotiations over the prices of new medicines between payers and manufacturers [3]. Conventionally, these economic evaluations involve a comparison of a medicine’s incremental cost-effectiveness ratio (ICER) with a cost-effectiveness ‘threshold’ [4].

The past decade has seen numerous advancements in the theoretical and empirical literature regarding how such a ‘threshold’ should be specified [5,6,7,8]. At present, there are two conceptually different approaches, termed ‘demand-side’ and ‘supply-side’; a demand-side approach assumes that the ‘threshold’ should reflect a socially legitimate aggregation of individuals’ willingness to pay for a marginal improvement in population health, while a supply-side approach assumes that the ‘threshold’ should reflect the health opportunity cost of reimbursing medicines within a budget-constrained health care system (i.e. the health loss experienced by other patients) [9,10,11].

In practice, reimbursement decisions involve a number of considerations that are not taken into account by either approach. If the ‘threshold’ is low, manufacturers may be unable to supply some medicines at a profit and may also choose not to invest in developing new medicines. If the ‘threshold’ is high, some manufacturers may make substantial profits from new medicines, yet the net benefit provided to patients may be small or even negative due to the resulting opportunity cost. Conventional approaches also do not consider how specifying a ‘threshold’ can result in strategic pricing behaviour by manufacturers.

1.1 Purpose

The purpose of this paper is to propose a framework that addresses some important limitations with conventional approaches. The proposed framework incorporates key considerations from both supply-side and demand-side approaches, including the health opportunity cost borne by other patients and society’s willingness to pay for marginal improvements in population health. The framework extends these by also considering the costs incurred by manufacturers in developing and supplying new medicines, and the incentive for manufacturers to strategically price up to any ‘threshold’ specified by the payer.

The framework is built upon fundamental economic principles, allowing for consideration of the ‘consumer surplus’ and ‘producer surplus’ arising from the reimbursement of medicines, and how different prices impact upon the distribution of total welfare between patients (consumers) and manufacturers (producers). This allows for the consideration of a ‘fair’ price for a medicine.

1.2 A ‘Fair’ Price for a Medicine

The concept of ‘fair’ is inherently normative. When pricing medicines, there are numerous stakeholders with different interests and perspectives on what a ‘fair’ price would be.

For the purposes of this framework, a ‘fair’ price is considered to be one that, at a minimum, balances the interests of patients and manufacturers. In other words, there must be a ‘fair’ distribution of total welfare between patients (consumers) and manufacturers (producers).

The framework also recognizes that ‘fairness’ is required in the way that the public health care system treats different patients, and how it rewards different manufacturers:

• A. ‘Fairness’ requires that the payer assigns equal value to equivalent impacts upon the health of patients, regardless of whether those patients benefit from medicines or bear the opportunity cost of their reimbursement.

• B. Larger manufacturers may be in the position to negotiate more favourable prices with the payer than smaller manufacturers. Where medicines are priced collectively (Sect. 4), ‘fairness’ requires that the payer offers the same ‘common price’ for an equivalent unit of health benefit, regardless of the manufacturer.

1.3 Definitions and Assumptions

The definitions and assumptions used in this framework are summarized in Table 1.

Table 1 Definitions and assumptions used in the framework

2 Economic Principles

When considering a ‘fair’ price for any good, it is informative to consider some fundamental economic principles.

At any given price, the ‘total welfare’ from a good is the sum of two parts:

• A. The ‘consumer surplus’, which is the benefit obtained by consumers because they are able to purchase the good at a price lower than their ‘willingness to pay’.

• B. The ‘producer surplus’, which is the benefit obtained by producers because they are able to sell the good at a price higher than their ‘willingness to accept’.

2.1 Standard Models

Microeconomics has a number of standard models that describe how consumers and producers behave under different market conditions. Two standard models are of particular relevance:

• 1. In a perfectly competitive market, an equilibrium price arises at which there is positive consumer surplus and positive producer surplus.

• 2. In a monopoly with a single price, the producer reduces output and raises the price so as to maximize producer surplus. Consumer surplus is diminished but remains positive.

Critically, both the consumer surplus and the producer surplus are strictly positive in each of these standard models.¹ It is therefore reasonable to expect both to be positive when reimbursing a medicine within a public health care system.

This provides a useful starting point for considering a ‘fair’ price for a medicine; the price should be high enough to provide a positive producer surplus, but low enough to provide a positive consumer surplus.

2.2 Basic Framework

Over the following sections, a basic framework will be outlined, grounded in these economic principles. To remain accessible, this basic framework makes a number of simplifying assumptions; the implications of relaxing many of these assumptions will be considered in Sect. 7.

3 Pricing Medicines Independently

First, the basic framework will be constructed under the simplifying assumption that medicines are priced independently. Later this framework will be extended to consider the pricing of medicines collectively.

Considering the consumer and producer surplus arising from reimbursing a medicine, and hence the range of ‘fair’ prices, requires the specification of demand and supply curves.

3.1 Demand Curve and Consumer Surplus

Given a publicly funded health care payer perspective, a reasonable specification of the payer’s demand curve is based on the net impact on the lifetime health of patients associated with reimbursing the medicine within the publicly funded health care system in question, where health is measured using a commonly accepted unit and discounted to a present value.

The use of equity weighting and/or an alternative perspective would imply a different specification of the demand curve; the implications of this are considered in Sect. 7.

Given a constrained budget, the net impact of reimbursing a medicine upon the health of patients is a function of two components:

• 1. The gain in health experienced by patients who receive the medicine.

• 2. The loss in health experienced by other patients whose health care subsequently receives less funding than it would have done if the medicine were not reimbursed.

3.1.1 Health Gain

The gain in health for patients who receive the medicine is routinely estimated as part of economic evaluations conducted by HTA agencies and will hereafter be denoted as $\Delta H$.

There is often uncertainty when estimating $\Delta H$, due to limitations in the clinical evidence base; this has implications for the determination of a ‘fair’ price, which are considered in Sect. 5.

For simplicity, the basic framework will consider medicines with one indication only, and assumes homogeneity within this indication, such that $\Delta H$ remains constant as the quantity of medicine increases; the implications of relaxing these assumptions are considered in Sect. 7.

3.1.2 Health Loss

Since the patients who incur health opportunity costs are typically unidentifiable, the standard approach for estimating the magnitude of the health loss is to divide the incremental cost of the medicine (hereafter denoted as $\Delta C$) by the ‘supply-side threshold’ (denoted as $k$), which reflects the marginal productivity of spending within the relevant health care system budget. Recent and ongoing empirical work has attempted to estimate $k$ for various public health care systems around the world [12]. The methods used for these studies take into account any technical inefficiencies associated with the current budget allocation [7].

As a hypothetical example, if $k$ is estimated to be £15,000 per quality-adjusted life-year (QALY), this means that every incremental £15,000 of cost imposed on the health care budget would be expected to result in a health loss of 1 QALY for other patients. Supposing $\Delta C$ is £60,000, this implies an estimated health loss of 4 QALYs ($\Delta C/k$); reimbursing this medicine would therefore be expected to improve population health (i.e. provide positive consumer surplus) only if $\Delta H$ exceeds 4 QALYs.

There is uncertainty in estimating the health loss, due to uncertainty in both $\Delta C$ and $k$. This is important when determining a ‘fair’ price for the medicine and will be considered in Sect. 5.

For simplicity, it is assumed that medicines have ‘marginal’ net budget impact, and that there is a single budget; the implications of relaxing these assumptions are considered in Sect. 7.

3.1.3 Plotting the Demand Curve

Under the simplifying assumptions above, the demand curve for a medicine may be plotted as a perfectly elastic horizontal line, stretching from the vertical axis to the quantity of medicine for which there is clinical need. This horizontal line is plotted at the price at which the health gain from the medicine ($\Delta H$) is exactly offset by the health loss ($\Delta C/k$), such that the net impact of reimbursing the medicine on population health is zero. That is, the demand curve plots the price at which

$$\Delta H=\Delta C/k$$

(1)

Rearranging equation (1), it follows that the demand curve plots the price at which the ICER of the medicine equals $k$:

$$\Delta C/\Delta H=k$$

(2)

For the hypothetical medicine in Fig. 1, the ICER equals $k$ at a price of P₁, such that the demand curve (D₁) is also plotted at this price.

Fig. 1

Demand curve and consumer surplus for a hypothetical medicine. P₁ and P₂ represent possible prices for the medicine. D₁ represents the demand curve for the medicine. Q₁ represents the quantity of medicine supplied. ICER incremental cost-effectiveness ratio

3.1.4 Consumer Surplus

In Fig. 1, a price of P₁ would result in no consumer surplus since there would be no area between the price and the demand curve (which would exactly overlap). A price below P₁ (e.g. P₂) results in positive consumer surplus, since the health gain exceeds the health loss, such that net population health outcomes are improved. The consumer surplus depends upon the quantity of medicine supplied (Q₁); at a price of P₂, this is represented by the area of the green region.

A price above P₁ would result in negative consumer surplus, since the health loss would exceed the health gain. As noted earlier, a zero or negative consumer surplus is not an outcome that arises in relevant standard models in economics and does not accord with the definition of a ‘fair’ price provided earlier.

3.1.5 More versus Less Effective Medicines

All else equal, the more effective a medicine is at improving health ($\Delta H$), the greater the price at which the ICER equals $k$, and hence the higher the demand curve (e.g. D₂ in Fig. 2); conversely, the less effective a medicine, the lower the demand curve (e.g. D₃ in Fig. 2).

Fig. 2

Implications of a medicine’s effectiveness, and the health opportunity cost of its reimbursement (k), for the medicine’s demand curve. P₁, P₃ and P₄ represent possible prices for the medicine. D₁, D₂ and D₃ represent possible demand curves for the medicine

3.1.6 Differences in k Across Different Countries

Theoretical and empirical research suggests that $k$ is higher in countries with greater health spending per capita [7, 12]. All else equal, a higher value of $k$ results in a higher demand curve. All else equal, a richer country would therefore be expected to have a higher demand curve for any given medicine (e.g. D₂ in Fig. 2) than a poorer country (e.g. D₃ in Fig. 2).

3.2 Supply Curve and Producer Surplus

The supply curve plots the lowest price that a manufacturer would be willing to accept when supplying any given quantity of medicine to the public health care system in question.

Plotting the supply curve is challenging. It is a function of many factors, including (but not limited to) marginal costs of production, and the potential implications for ‘reference pricing’ in other jurisdictions (e.g. a manufacturer might be unwilling to accept a price, even if it covers marginal costs, if doing so results in a lower price in other jurisdictions).

Furthermore, manufacturers have incentives to keep information on the above factors confidential, and make efforts to do so in practice [13,14,15]. Payers therefore generally have little information from which to estimate the supply curve for any given medicine. For the purposes of this framework, the supply curve will therefore be assumed to be private information of the manufacturer that is not observed by the payer.

3.3 Total Welfare

The demand and supply curves may be used to consider the ‘total welfare’ (or ‘economic surplus’) that arises from reimbursement of a medicine at any given price and quantity, and the distribution of this between patients (consumer surplus) and manufacturers (producer surplus).

In Fig. 3, the consumer surplus is illustrated by the green area below the demand curve and above the price, while the producer surplus is illustrated by the blue area above the supply curve and below the price.

Fig. 3

Distribution of consumer surplus and producer surplus for a hypothetical medicine. P₁, P₅ and P₆ represent possible prices for the medicine. D₁ and S₁ represent the demand and supply curves for the medicine, respectively. Q₁ represents the quantity of medicine supplied

In general, where the supply curve lies below the demand curve at the relevant quantity, there is a range of possible ‘fair’ prices at which consumer and producer surplus are both positive (e.g. P₅ in Fig. 3).

The upper bound of this range is a price corresponding to the demand curve (e.g. P₁ in Fig. 3), such that the ICER equals $k$. At this price, the entirety of the economic surplus is allocated to the producer, and consumer surplus is zero. As argued earlier, this is not a ‘fair’ share of the economic surplus; a ‘fair’ price must be below this upper bound, such that consumer surplus is strictly positive.

The lower bound of this range is determined by the location of the supply curve. Since the supply curve is assumed to be private information of the manufacturer that is not observed by the payer, this lower bound is unknown to the payer in practice.

4 Pricing Medicines Collectively

In practice, the inability to observe supply curves limits the payer’s ability to consider a ‘fair’ price for each medicine independently and provides a rationale for extending the framework to consider a ‘fair’ common price for a unit of health that applies to all medicines collectively.²

A collective approach based on a single ‘common price’ has some notable advantages:

• 1. It provides manufacturers with a clear signal as to the payment they will receive for any additional units of health gain provided by the medicines they develop. This reduces uncertainty as to the manufacturer’s return on investment, which will be considered in Sect. 5.

• 2. It rewards efficient research and development (R&D). Since manufacturers can ‘price up’ to the common price regardless of their reserve price, they can receive disproportionately large returns if their production costs are low and/or their medicines are highly effective. Conversely, manufacturers who develop costly and ineffective medicines will receive lower returns, disincentivizing inefficient R&D.

In determining a ‘fair’ common price, it is important to consider the total economic surplus arising from all medicines reimbursed at each potential common price, and the distribution of this between patients (consumers) and manufacturers (producers).

4.1 Additional Assumptions

The collective approach proposed here makes the following assumptions, further to those outlined in Table 1.

• 1. A single common price for a unit of health, $\lambda$, is publicly specified by the payer. Given the assumption of a causal relationship between $\lambda$ and the prices of medicines, reimbursed medicines with a reserve ICER less than or equal to $\lambda$ are supplied and priced such that the ICER equals $\lambda$. Medicines with a reserve ICER greater than $\lambda$ are not supplied.

• 2. Medicines are considered independent, such that reimbursing (or not reimbursing) a medicine has no impact on the reserve ICERs for other medicines.

4.2 Hypothetical Example

Given the confidential nature of manufacturers’ reserve prices in practice, the collective approach will be illustrated using a simple hypothetical example.

There are four medicines, each of which is supplied only if the common price specified by the payer (in £ per QALY) equals or exceeds the reserve ICER reported in Table 2. If a medicine is supplied, it will be reimbursed at the common price and will provide the health gain reported in Table 2. Any costs of reimbursement fall on a public health care budget and diminish the health of other patients; it will be assumed that $k$ is £15,000 per QALY, in line with UK empirical estimates [16].³

Table 2 Hypothetical example

To consider the total welfare at any given common price, consumer and producer surplus must both be valued in a common metric. For this, a ‘demand-side’ threshold ($v$) is used. Recent work by Woods et al. [17] considered two different values of $v$: £60,000 per QALY, based on the value of a statistical life year used by the UK Government [18]; and £30,000 per QALY, a value more closely aligned with recent literature [9, 10, 19]. The same approach is adopted here.

4.3 Consumer Surplus

To begin, the consumer surplus will be calculated at different common prices. To illustrate the key insights of this approach, it is sufficient to consider three common prices within the range of reserve ICERs (£5000, £10,000, and £15,000 per QALY, where the last equals $k$), as well as a common price below the range of reserve ICERs (£0 per QALY) and one above this range (£20,000 per QALY).

The consumer surplus at each common price is summarized in Fig. 4, with more detailed figures provided in the electronic supplementary material (ESM).

Fig. 4

Consumer surplus at each common price in the hypothetical example. k denotes the health opportunity cost of reimbursing medicines. QALY quality-adjusted life-year

4.3.1 Common Price of £0 per QALY

At a common price of $0 per QALY, no medicines are reimbursed because every reserve ICER lies above the common price. No health gain is provided, but there is also no health loss for other patients, so the net impact on population health is zero. This is illustrated by the grey dot at the origin of Fig. 4 and in the ESM (Fig. 10).

4.3.2 Common Price of £5000 per QALY

At a common price of £5000 per QALY, medicine A is supplied because the common price exceeds its reserve ICER.

The health gain for patients receiving A is 30 QALYs. Although its reserve ICER is £2750 per QALY, it is priced up to the common price of £5000 per QALY. With a health gain of 30 QALYs, and an ICER of £5000 per QALY, it follows that the incremental cost of A is £150,000. Since $k$ is £15,000 per QALY, the health loss from reimbursing A is 10 QALYs. Reimbursing A at this common price therefore improves population health by 20 QALYs, comprising a 30 QALY gain and a 10 QALY loss.

The remaining medicines are not supplied, providing no health gain but also no health loss. The consumer surplus is therefore 20 QALYs, illustrated by the yellow dot in Fig. 4 and in the ESM (Fig. 11).

4.3.3 Common Price of £10,000 per QALY

Increasing the common price to £10,000 per QALY has two countervailing effects. Medicine B is now supplied and is priced up to £10,000 per QALY. With a health gain of 60 QALYs, and an ICER of £10,000 per QALY, the incremental cost is £600,000. The health loss is therefore 40 QALYs, such that reimbursing B improves population health by 20 QALYs.

However, by increasing the common price, medicine A is now priced up to £10,000 per QALY. Its incremental cost is now £300,000, so the health loss to other patients is now 20 QALYs (compared with 10 QALYs at a common price of £5000 per QALY). Reimbursing A at this higher common price therefore improves population health by only 10 QALYs (rather than 20 QALYs).

Since medicines C and D are not supplied, the total health gain is 90 QALYs and the total health loss is 60 QALYs. The consumer surplus is 30 QALYs, illustrated by the green dot in Fig. 4 and in the ESM (Fig. 12).

4.3.4 Common Price of £15,000 per QALY

At a common price of £15,000 per QALY, medicine C is now supplied and priced up to £15,000 per QALY. With a health gain of 45 QALYs, the incremental cost is £675,000 and the health loss is also 45 QALYs. Reimbursing C has no net impact on population health since the health gain equals the health loss.

Increasing the common price also results in medicines A and B being priced up to £15,000 per QALY. The incremental costs of A and B are now £450,000 and £900,000, resulting in a health loss of 30 QALYs and 60 QALYs, respectively, in each case equal to the health gain. Reimbursing A and B at this higher common price therefore nullifies their net impact on population health.

Medicine D is still not supplied, so the total health gain is 135 QALYs and total health loss is also 135 QALYs, resulting in no consumer surplus, as illustrated by the black dot in Fig. 4 and in the ESM (Fig. 13).

Note that this finding of zero consumer surplus at a common price of $k$ is generalizable beyond this example: any medicine priced at $k$ provides no net impact on population health, since the health gains are exactly offset by the health losses. This has critical implications for determining a ‘fair’ common price for medicines in practice.

4.3.5 Common Price of £20,000 per QALY

At a common price of £20,000 per QALY, above the entire range of reserve ICERs, all four medicines are supplied and priced up to £20,000 per QALY; since this exceeds $k$, all four medicines displace more QALYs through their reimbursement than they provide to patients.

Given their associated health gains (Table 2), the incremental costs are now £600,000 for medicine A, £1,200,000 for B, and £900,000 for each of C and D; the resulting health loss is now 40 QALYs for A, 80 QALYs for B, and 60 QALYs for each of C and D. The total health gain is 180 QALYs and the total health loss is 240 QALYs, resulting in a net loss in population health (i.e. a negative consumer surplus) of 60 QALYs; this is illustrated by the red dots in Fig. 4 and in the ESM (Fig. 14).

4.3.6 Maximizing Consumer Surplus

Although the greatest consumer surplus in the analyses above arose at a common price of £10,000 per QALY, these five common prices are not exhaustive, and in fact the maximum consumer surplus arises at another common price: £7750 per QALY. This corresponds to the reserve ICER of medicine B, where the consumer surplus is 43.5 QALYs.

It is a general finding that consumer surplus is maximized at a common price below $k$ corresponding to a reserve ICER for a medicine. The trend in the consumer surplus—initially increasing with the common price, reaching a maximum at a common price below $k$, then declining to zero at a common price of $k$—is also a general finding. These general findings are discussed in Sect. 4.6.

4.4 Producer Surplus

The assumption that manufacturers are unwilling to supply at a loss has the following implications for the producer surplus:

• 1. If the common price is lower than a medicine’s reserve ICER, the manufacturer will not supply and the producer surplus will be zero.

• 2. If the common price equals the medicine’s reserve ICER, the medicine will be supplied but the producer surplus will be zero.

• 3. If the common price exceeds the reserve ICER, the medicine will be supplied and will be priced up to the common price, such that producer surplus will be positive. In this case, producer surplus can be calculated by subtracting the incremental cost of the medicine at the reserve ICER (aggregated over all patients) from the incremental cost at the common price. For example, if the incremental cost is £100,000 when priced at the reserve ICER (where producer surplus is zero) but increases to £300,000 when priced up to a higher common price, the producer surplus is £200,000.

The producer surplus at each common price is summarized in Fig. 5, with more detailed figures provided in the ESM.

Fig. 5

Producer surplus at each common price in the hypothetical example. k denotes the health opportunity cost of reimbursing medicines. QALY quality-adjusted life-year

4.4.1 Common Price of £0 per QALY

At a common price of $0 per QALY, no medicines are supplied because each reserve ICER lies above the common price. Producer surplus is zero, illustrated by the grey dot at the origin of Fig. 5 and in the ESM (Fig. 15).

4.4.2 Common Price of £5000 per QALY

At a common price of £5000 per QALY, medicine A is now supplied. At its reserve ICER (£2750 per QALY), medicine A would have an incremental cost of £82,500. Instead, it is priced up to £5000 per QALY, so the incremental cost is £150,000. The increase in its incremental cost (£67,500) represents the producer surplus accruing to the manufacturer at this common price. Since no other medicines are supplied, the total producer surplus is £67,500, illustrated by the yellow dot in Fig. 5 and in the ESM (Fig. 16).

4.4.3 Common Price of £10,000 per QALY

At a common price of £10,000 per QALY, medicine B is now supplied. Both A and B are priced up to £10,000 per QALY. At their reserve ICERs, the incremental costs of medicines A and B are £82,500 and £465,000, respectively. At this common price, their incremental costs are £300,000 and £600,000, resulting in a producer surplus of £217,500 and £135,000, respectively. No other medicines are supplied, so the total producer surplus is £352,500, illustrated by the green dot in Fig. 5 and in the ESM (Fig. 17).

4.4.4 Common Price of £15,000 per QALY

At a common price of £15,000 per QALY, medicine C is now supplied. Medicines A, B, and C are all priced up to £15,000 per QALY. At their reserve ICERs, the incremental costs of medicines A, B, and C are £82,500, £465,000, and £551,250, respectively. At this common price, their incremental costs are £450,000, £900,000, and £675,000, resulting in a producer surplus of £367,500, £435,000, and £123,750, respectively. Since medicine D is not reimbursed, the total producer surplus is £926,250, illustrated by the black dot in Fig. 5 and in the ESM (Fig. 18).

4.4.5 Common Price of £20,000 per QALY

At this common price, above the entire range of reserve ICERs, all four medicines are now supplied and priced up to £20,000 per QALY. At their reserve ICERs, the incremental costs of medicines A, B, C, and D are £82,500, £465,000, £551,250, and £866,250, respectively. At this common price, their incremental costs are £600,000, £1,200,000, and £900,000 for both C and D, resulting in a producer surplus of £517,500, £735,000, £348,750, and £33,750, respectively. The total producer surplus is £1,635,000, illustrated by the red dot in Fig. 5 and in the ESM (Fig. 19).

4.5 Total Welfare

The consumer surplus calculated in Sect. 4.3 can be converted into monetary terms using the demand-side threshold ($v$). The total welfare at each common price is the sum of the monetary value of the consumer surplus and the producer surplus calculated in Sect. 4.4.

Table 3 summarises the consumer surplus, monetary value of the consumer surplus, producer surplus, and total welfare for each common price considered above, for each of $v$ = £30,000 per QALY and $v$ = £60,000 per QALY.

Table 3 Consumer surplus (CS), producer surplus, and total welfare, at each common price in the hypothetical example

In the example here, for each of $v$ = £30,000 per QALY and $v$ = £60,000 per QALY, total welfare is maximized at a common price of £7750 per QALY, below $k$ and coinciding with the reserve ICER for medicine B; this is summarized in Table 3 and highlighted in bold.

This is a generalizable finding (see Sect. 4.6); under the reasonable assumption that $v>k$, total welfare is maximized at a common price below $k$ coinciding with the reserve ICER of a medicine. Note that if $v$ is only slightly greater than $k$, the welfare maximizing common price will coincide with the highest reserve ICER below $k$ (in the example here, if $v$ = £16,000 per QALY, then total welfare is maximized at a common price of £12,250 per QALY). Where the difference between $v$ and $k$ is larger, the welfare-maximizing common price may coincide with another reserve ICER further below $k$, as in this example where $v$ = £30,000 or £60,000 per QALY.

It should also be noted that $v\le k$ is implausible in practice, given the costs associated with administering both taxation and the public health care system, and the unwillingness of individuals to pay as much (or more) in taxes to generate a marginal population QALY as they are willing to pay for a marginal QALY for themselves [20]. Nevertheless, for completeness, the implications of assuming $v\le k$ are considered in the ESM.

4.6 Generalizable Findings

Some of the findings of the example provided here are generalizable. These are summarized in the following sections and in Fig. 6.

Fig. 6

General shape of the consumer surplus, producer surplus, and total welfare curves, assuming the lower bound of the distribution of reserve incremental cost-effectiveness ratios (ICERs) is zero. λ_C and λ_max represent the common prices that maximize consumer surplus and total welfare, respectively. k denotes the health opportunity cost of reimbursing medicines. If the lowest reserve ICER is above zero, the consumer and producer surplus will be zero for all common prices below that reserve ICER (such that the curves will be flat along the horizontal axis until that point). Alternatively, if the lowest reserve ICER is negative (i.e. there are medicines that dominate their comparator at their reserve price), these will be supplied (and provide a positive consumer and producer surplus) even at a common price of zero, such that the curves both intersect the vertical axis above the origin

4.6.1 Consumer Surplus

• 1. Consumer surplus is zero at any common price below the lowest reserve ICER of any medicine. This is because no medicines are supplied, so there is no health gain nor health loss, and so no impact on population health.

• 2. Consumer surplus is positive for common prices above the lowest reserve ICER but below $k$. Within this range, a marginal increase in the common price may give rise to two countervailing effects:

  • a. An increase in consumer surplus due to one or more new medicines being supplied since the common price now equals the respective reserve ICER(s). Since these new medicines are priced below $k$, their reimbursement improves population health.

  • b. A decrease in consumer surplus due to medicines already supplied being priced up to the new common price, diminishing population health.

  At low common prices, where few medicines are supplied, the first effect generally outweighs the second, so a marginal increase in the common price improves consumer surplus (such that the green curve in Fig. 6 slopes upwards).

  As the common price (and the number of medicines supplied) increases, the relative magnitude of the second effect grows, eventually equalling that of the first. At this common price (denoted as ${\lambda }_{\mathrm{C}}$), the consumer surplus is maximized and the green curve in Fig. 6 peaks. This point always corresponds to the reserve ICER of a medicine.⁴

  At common prices above ${\lambda }_{\mathrm{C}}$ but below $k$, marginal increases in the common price result in a declining (but still positive) consumer surplus, with the second effect outweighing the first (causing the green curve in Fig. 6 to slope downwards).

• 3. At a common price of $k$, the consumer surplus is zero. This is a critical finding and arises because the health gains from any supplied medicines are exactly offset by the health losses that result from their reimbursement. The green curve in Fig. 6 intersects the horizontal axis at this point.

• 4. Above a common price of $k$, consumer surplus is negative, and becomes more negative as the common price is increased. This is because all medicines are priced sufficiently highly that their health gains are outweighed by the health losses resulting from their reimbursement, and because any new medicines supplied as a result of increasing the common price are also priced such that they diminish population health outcomes.

4.6.2 Producer Surplus

• 1. Producer surplus is zero at any common price below the lowest reserve ICER of any medicine. This is because no medicines are supplied.

• 2. For common prices above the lowest reserve ICER of any medicine, producer surplus unambiguously increases as the common price increases. This is due to two complementary effects:

  • a. An increase in producer surplus due to one or more new medicines being supplied since the common price now meets or exceeds the respective reserve ICER(s).

  • b. An increase in producer surplus due to higher pricing for medicines already supplied, which are now priced up to the new common price.

4.6.3 Total Welfare

• 1. Where $v>k$, total welfare is maximized at a common price at or above ${\lambda }_{\mathrm{C}}$ but below $k$, coinciding with the reserve ICER for a medicine. This is plotted as ${\lambda }_{\mathrm{max}}$ in Fig. 6.

• 2. If $v$ is only slightly greater than $k$, the welfare maximizing common price coincides with the highest reserve ICER below $k$. If $v$ is considerably greater than $k$, the welfare maximizing common price may coincide with another reserve ICER further below $k$.

4.7 Establishing a ‘Fair’ Common Price

As in earlier sections, a ‘fair’ price requires that consumer and producer surplus are both positive. When pricing medicines collectively, consumer and producer surplus are both positive at all common prices greater than the lowest reserve ICER of any medicine but below $k$.

Nevertheless, an important finding of the framework is that consumer and producer surplus both increase up to a common price of ${\lambda }_{\mathrm{C}}$, so there is mutual interest in setting the common price at least as high as ${\lambda }_{\mathrm{C}}$. The most relevant range for establishing a ‘fair’ common price is therefore between ${\lambda }_{\mathrm{C}}$ (where consumer surplus is maximized) and $k$ (where consumer surplus is zero).

At any ‘fair’ common price, there will generally be some medicines with low reserve ICERs for which the manufacturer receives substantial profits, and others with higher reserve ICERs for which most of the economic surplus for that medicine is allocated to patients. This is inevitable when setting a single common price for all medicines collectively. There will also generally be medicines with reserve ICERs above the common price but below $k$. This raises a number of issues and is considered further in the ESM.

5 Return on Investment

In previous sections, a ‘fair’ price was defined as one that allocates, at a minimum, a positive share of the total welfare from a medicine’s reimbursement to both consumers and producers.

This principle is consistent with standard models of a market operating under either perfect or imperfect competition. Both consumer and producer surplus are positive in competitive markets, and both remain positive under a standard model of a monopoly.

Yet refinements to this definition might be appropriate. This section considers the ‘return on investment’ for both the manufacturer and payer from developing and reimbursing medicines.

5.1 Return for the Manufacturer

A limitation with the supply curve, and hence producer surplus, is that it only considers costs that increase with the quantity supplied. It does not consider fixed costs, including R&D costs.

Manufacturers incur substantial costs and risks when developing medicines [21, 22], driven by the high costs of regulation for demonstrating effectiveness and manufacturing quality. In addition, they incur the costs of medicines that fail during clinical trials [23]. Investing capital in developing medicines has an opportunity cost, since it could generate positive returns elsewhere [24, 25].

A medicine might therefore be supplied at a price that exceeds its marginal costs of production, providing a positive producer surplus, yet have a negative ‘economic profit’ (producer surplus minus fixed costs).

The definition of a ‘fair’ price stated earlier can be refined to address this. Provided the price still results in positive consumer surplus (and a reasonable rate of return for the payer, as considered below), the manufacturer should be allocated a sufficient share of the total welfare to cover fixed costs (including R&D), and enough additional surplus to reward the risk taken in developing the medicine.

In other words, a ‘fair’ price is one at which the producer surplus is not merely positive, but sufficiently positive to provide the manufacturer with a reasonable rate of return on its investment. This is a particularly important consideration for drugs with small target populations, for which the fixed costs of development must be recovered from fewer patients, and for which R&D may be commercially riskier [26].

5.2 Return for the Payer

The payer also takes risks when reimbursing medicines. Both the health gains and health losses are uncertain. As a result, there can be substantial uncertainty as to the net impact of reimbursing a medicine upon population health.

Reimbursing a medicine should therefore be considered an investment by the payer. Since the broad purpose of a public health care system is to improve the health of the public, the return on this investment may be considered in terms of population health (instead of monetary terms), using a pharmacoeconomic model to estimate the health gains and losses in each time period, discounted to a present value. Probabilistic analysis also provides a means to explicitly consider the payer’s risk. It is rational for the payer to seek a return that is commensurate with this risk.

A limitation with the demand curve, and hence consumer surplus, is that it only considers the payer’s willingness to pay for a marginal unit of medicine, and not the fixed costs that apply regardless of the quantity of medicine reimbursed. These fixed costs vary across medicines, and may include (but are not limited to) the costs associated with

• conducting HTA (including personnel, data collection and analysis, and consultation);

• negotiation and contracting;

• guideline development;

• developing infrastructure (including computer systems and administrative support);

• health care provider training;

• patient education;

• legal and contractual costs;

• public engagement and communication costs (including stakeholder engagement);

• ongoing monitoring and evaluation costs.

Many of these costs are irreversible; they cannot be recovered if the medicine turns out to be less effective than expected. If the medicine turns out to be more costly than expected, this has negative implications for the health of other patients. Reversing a decision to reimburse a health technology in such circumstances is also costly.

The definition of a ‘fair’ price can be refined to reflect this risk. Provided the price still results in a reasonable rate of return for the manufacturer (see previous section), the payer should be allocated enough economic surplus to cover its fixed costs, and enough additional surplus to reward the risk taken in reimbursing the medicine.

In other words, a ‘fair’ price is one at which the consumer surplus is not merely positive, but sufficiently positive to provide the payer with a reasonable rate of return on its investment⁵.

5.3 Equalizing Risk-Adjusted Rates of Return

A fundamental principle of financial economics is that a risky investment requires a positive expected rate of return, with a greater risk requiring a greater expected rate of return [30, 31].

A potential approach to specifying a single ‘fair’ price, within the range considered above, is to equalize the risk-adjusted rates of return for both the manufacturer and payer. This would result in a greater nominal return for the party that incurs the greater risk. For example, if the manufacturer incurs a greater risk in developing the medicine than the payer incurs in reimbursing the medicine, then a greater proportion of the economic surplus would be allocated to the manufacturer.

An advantage of this approach is that it would reward manufacturers who come to market with higher quality evidence, since this would reduce the risks faced by the payer.

A significant challenge for implementing this approach in practice is that the manufacturer’s expected rate of return is typically unknown to the payer. Even if it were disclosed, there would likely be disagreement as to how the risks faced by each party would be estimated. Nevertheless, defining the principles needed to characterise a single ‘fair’ price is a necessary first step towards developing the methods by which this could, in future, be calculated.

6 Pricing Over a Medicine’s Lifecycle

The basic framework developed so far has been used to consider the boundaries of a ‘fair’ price for a medicine. In practice, a complicating factor is that a medicine is unlikely to have just one price over its lifecycle. For example, the price of a medicine may be expected to fall after patent expiry, when it is subject to generic competition.

This ‘dynamic’ pricing has implications for the consumer and producer surplus arising from a medicine’s reimbursement over the course of its lifecycle. All else equal, a reduction in the price of a medicine following patent expiry would increase the consumer surplus, and reduce the producer surplus, in later years.

It follows that a ‘fair’ price (or ‘fair’ set of prices, if prices change over time) should, as far as practicable, be established over the entire lifecycle of the medicine.

6.1 Extending the Basic Framework

The basic framework can be extended to consider this. The principles considered earlier remain: the price (or set of prices) for the medicine should be such that the economic surplus arising from its reimbursement across its lifecycle is allocated fairly between consumers and producers, resulting in a reasonable rate of return for each party.

Changes in a medicine’s prices over its lifecycle can be considered through dynamic modelling using multiple cohorts [32]. Expected changes in medicines’ prices would be explicitly modelled for future cohorts, allowing for estimation of the consumer and producer surplus in each period and also the rate of return for each party. Uncertainty in future pricing would be incorporated through probabilistic analysis.

6.2 Factors to Consider in Dynamic Modelling

To calculate a ‘fair’ price (or set of prices) over a medicine’s lifecycle, a dynamic model would need to take into account the following:

• Expected changes in the price of the medicine over time.

• Expected changes in the price of any comparators. For example, the patent on a comparator may expire before that of the medicine, increasing the incremental cost of the medicine over the medium term.

• The potential for ‘evergreening’, which reduces the length of time that the payer benefits from a lower price following patent expiry [33]. The price of a medicine may even increase following the original expiry date of the primary patent [34, 35].

• Expected changes in the future market share of the medicine.

• Time preference, which reduces the present value of any gains in population health arising from a price reduction following patent expiry.

• Uncertainty in the considerations above. Paying a higher price for a medicine today, on the expectation that the price might fall in future, effectively transfers risk onto the payer. The payer’s expected rate of return from their investment would need to increase to reflect this additional risk (Sect. 5).

6.3 Establishing a ‘Fair’ Price Using Dynamic Modelling

The same principles established in the basic framework continue to apply. A ‘fair’ price (or set of prices) must be high enough to result in a reasonable rate of return for the manufacturer, but low enough to result in a reasonable rate of return for the payer, over the medicine’s lifecycle, when discounted to present values.

If the medicine’s ICER is greater than or equal to $k$ when calculated using a dynamic model, this means that the medicine is not expected to improve population health over its lifecycle, even when taking into account the off-patent period, and hence would not be ‘fair’.

Unlike static models, dynamic models allow the payer to consider the possibility of delaying reimbursement until the price of the medicine falls. In a model with dynamic pricing, the ICER for the medicine will generally change over time as new cohorts enter the model. If a ‘fair’ price cannot be found at launch (because the manufacturer will not supply at a price that provides for a reasonable rate of return for the payer, perhaps because of ‘reference pricing’ considerations in other jurisdictions) then the payer may wish to delay reimbursement of the medicine until the price falls, rather than decline to reimburse outright.

7 Extensions to the Basic Framework

There are numerous possible extensions to the basic framework outlined in earlier sections. The following are not exhaustive, and future research has the potential to provide many more.

7.1 Medicines That Cannot Be Supplied at a ‘Fair’ Price

For medicines that have relatively high marginal costs of production and/or provide relatively small incremental benefits to patients, it is possible that the supply curve lies above the demand curve (e.g. S₂ in Fig. 7). In such cases there are no ‘fair’ prices at which both consumer and producer surplus are positive since the total economic surplus is negative.

Fig. 7

Medicines that cannot be supplied at a ‘fair’ price. P₁, P₇ and P₈ represent possible prices for the medicine. D₁ and S₂ represent the demand and supply curves for the medicine, respectively. Q₁ represents the quantity of medicine supplied

The inability to find a ‘fair’ price in such cases is a particularly relevant consideration for medicines with high marginal costs of production, such as cell and gene therapies [36].

A potential solution, building upon the basic framework, could be to subsidize the reimbursement of these medicines by redistributing the economic surplus from other medicines, while ensuring that medicines overall continue to provide both positive consumer and producer surplus. This potential solution is described in more detail in the ESM.

7.2 Multiple Indications and Heterogeneity

The shape and location of a medicine’s demand curve depends upon the health gain ($\Delta H$), its incremental cost ($\Delta C$), the supply-side threshold ($k$), and the quantity of medicine for which there is clinical need (Sect. 3). Where a medicine has multiple indications, one or more of these factors might differ across each indication, resulting in a stepped demand curve (Fig. 8).

Fig. 8

Demand curve for a hypothetical medicine with multiple indications. P₉, P₁₀ and P₁₁ represent possible prices for the medicine. D₄ represents a stepped demand curve for the medicine, comprising all three bold horizontal lines. Q₂ represents the quantity of medicine supplied for indication A. Q₃ minus Q₂ represents the quantity of medicine supplied for indication B. Q₄ minus Q₃ represents the quantity of medicine supplied for indication C

Figure 8 can be used to consider the consumer surplus associated with different approaches to pricing across indications. For example, a single price of P₁₀ across all indications would result in positive consumer surplus from indication A (represented by the area of the green region), zero consumer surplus from indication B (since the price coincides with the demand curve), but negative consumer surplus from indication C (represented by the area of the red region). With a single price across all indications, ensuring that all indications provide a strictly positive consumer surplus would require a price below P₁₁. Alternatively, if separate prices can apply for each indication, then ‘fair’ pricing requires a price below P₉ for indication A, below P₁₀ for indication B, and below P₁₁ for indication C, resulting in a positive consumer surplus within each indication.

Furthermore, where there is heterogeneity in the patient population, the factors above might differ across patients within an indication. Observable heterogeneity can be addressed by allocating patients into subgroups, and the implications for the demand curve are similar to those that arise with multiple indications (with a stepped demand curve across subgroups).

7.3 Non-Marginal Net Budget Impact

The use of $k$ to estimate the health opportunity cost is appropriate only if the net budget impact of reimbursing the medicine is marginal [37]. Where the net budget impact is substantial (e.g. some recent treatments for hepatitis C), reimbursement requires displacement of more than just the marginal health care service, resulting in a disproportionately large health loss (i.e. greater than $\Delta C/k$) [8, 38].

In such cases, the demand curve will slope down as the quantity increases (Fig. 9). A price of $k$ will therefore result in negative consumer surplus, such that the upper limit of the range of ‘fair’ prices is further below $k$ than under the basic framework.

Fig. 9

Demand curve for a hypothetical medicine with non-marginal net budget impact. P₁ represents a possible price for the medicine. D₅ represents the demand curve for the medicine. Q₁ represents the quantity of medicine supplied

7.4 Budgetary Silos

If the budget for the public health care system has been allocated to separate budgetary ‘silos’, a given incremental cost will generally result in a different health loss depending on the specific silo upon which it falls. In other words, each silo has its own $k$.

If the incremental costs of a specific medicine fall on one silo only, the demand curve would look similar to that in the basic framework, and the relevant $k$ would be that for the silo in question.

However, if the incremental costs fall on multiple silos, then the demand curve may become a step function (similar to that in Fig. 8, with the steps caused by a different $k$ for each silo).

7.5 Investments in R&D and Future Innovations

If manufacturers invest some of the producer surplus into R&D, and if this potentially results in innovations that increase future economic surplus, this might justify allocating a larger share of the economic surplus to producers through a higher price.

In setting a higher price, consideration must be given to the dynamic impact on the consumer and producer surplus. To remain ‘fair’, a higher price would need to be mutually beneficial; it would not be ‘fair’ for the payer to give up more consumer surplus to support R&D than they expect to receive back through future innovations, after discounting to a present value.

A ‘fair’ common price in such a dynamic model would also still be lower than $k$. This is because a common price equal to or above $k$ will not only negate all consumer surplus from current medicines but will also result in future innovations being priced up to the same high common price, resulting in no consumer surplus over the longer term.

7.6 Private Insurance Markets

Many of the implications of the framework are relevant for policymakers who regulate private insurance markets. Reimbursement of medicines by private insurers also results in health opportunity costs: incremental costs must be covered either by an increase in premiums and/or copays (resulting in some patients reducing their coverage, or usage of insured services), or by making cutbacks elsewhere within the plan. Either way there is a health loss borne by some patients. Recent work by Vanness et al. estimated $k$ for private insurers in the United States at $104,000 per QALY; such an estimate could be used within this framework [39].

7.7 Societal Perspective

There is a growing understanding of the many non-health effects that arise for patients, their families, and informal caregivers as a result of reimbursing medicines. It is also recognized that costs may fall upon patients or informal caregivers (including co-pays, patient and caregiver time, and transportation costs). Some direct medical costs may also fall on private payers. None of these broader impacts or costs are considered under the publicly funded health care payer perspective adopted in the basic framework.

A societal perspective would impact upon specification of a medicine’s demand curve in two distinct ways:

• 1. It broadens the scope of $\Delta H$ and $\Delta C$ (with $\Delta H$ becoming a measure of societal benefit);

• 2. It broadens the scope of the opportunity cost [40]. In addition to health losses, it now includes non-health benefits to other patients and caregivers that would have arisen had resources not been reallocated to reimbursing the medicine. It includes lost productivity due to foregone health and non-health benefits. It includes costs imposed on private insurers due to other patients’ diminished access to public health care, and any costs imposed on other public sectors, such as a greater reliance on social services or affordable housing. It also includes any out-of-pocket costs imposed on other patients and caregivers due to their diminished access to public health care.

A ‘fair’ price would remain one that provides positive consumer and producer surplus (given this modified demand curve), and a reasonable return to both the manufacturer and payer.

When considering medicines collectively, a societal perspective would require that $k$ be replaced with the broader ‘societal’ measure of opportunity cost described above; the general findings from Sect. 4.6 would then still apply.

7.8 Equity Weighting

There is increasing interest in the use of ‘distributional cost-effectiveness analysis’ [41, 42]. This can be accommodated within the proposed framework. The payer’s objective would now be to maximize equity-weighted population health. In common with a societal perspective (Sect. 7.7), a broader consideration of opportunity cost would be required (with equity weights applied to health losses), but the general findings of the framework would continue to hold.

8 Discussion

A framework has been proposed to support the determination of ‘fair’ prices for medicines. This framework attempts to strike a balance between the interests of patients and manufacturers. It also attempts to ensure that different patients and manufacturers are treated ‘fairly’ by the public health care payer.

In a departure from convention, the proposed framework does not consider $k$ and $v$ as competing approaches for defining a common price for a unit of health (or cost-effectiveness ‘threshold’). Neither are used for this purpose. Instead, $k$ is considered a critical component of a medicine’s demand curve, which allows for estimation of the consumer surplus at any given price, while $v$ is used to value the consumer surplus in monetary terms to allow for calculation of the total welfare. Both $k$ and $v$ are considered important and located coherently within the framework.

Another departure is the assumption of a causal relationship between the common price offered by the payer and the resulting ICERs of medicines (i.e. ‘pricing to the threshold’). When medicines are priced collectively using a common price, the payer must take this into account. Under the reasonable assumption that $v>k$, total welfare is maximized at a common price below $k$, with consumer surplus maximized at an even lower common price. A common price of $k$ or $v$ results in zero or negative consumer surplus, respectively.

A further departure from convention is the recognition that reimbursing medicines is a risky investment by the payer, given the often substantial uncertainty regarding the expected impact on net population health outcomes. Conventionally, research on ‘return on investment’ has focused on the manufacturer; under this framework, ensuring a ‘fair’ price requires that these considerations be extended to the payer.

8.1 Ensuring a Reasonable Rate of Return for Both Parties

Since both the manufacturer and payer incur risks when developing and reimbursing medicines, a ‘fair’ price is one at which both consumer and producer surpluses are sufficiently positive that the manufacturer and payer each receive a reasonable rate of return on their investments.

Therefore, the payer should not expect the manufacturer to accept a price that is barely profitable, even if they have the monopsonistic power to push the price down close to the manufacturer’s reserve price. It also means that the manufacturer should not expect the payer to accept a price that results in little or no population health gain (i.e. where the ICER is close to $k$), even if they can use their monopolistic power to push the price higher than this.

Simply put, both payers and manufacturers should respect the risks and opportunity costs faced by the other party and accept a price that provides each party with a reasonable rate of return.

8.2 Rethinking Conventional Value-Based Pricing

Some health economists have advocated for an approach under which prices are set according to the ‘value’ a medicine provides, given the payer’s ‘willingness to pay’ [43]. In 2019, the US-based Institute for Clinical and Economic Review began a collaboration to develop new methods to guide ‘value-based pricing’ [44]. A resulting white paper on the topic was headlined “Aligning drug price with drug benefits to maximise patient outcomes” [45].

The framework proposed here shows that this approach would not result in ‘fair’ prices. This is because the payer’s ‘willingness to pay’, and the ‘benefits’ provided by medicines, should not be used to directly determine a medicine’s price, but rather a medicine’s demand curve. Conflating the price and the demand curve by pricing according to ‘willingness to pay’ (or by ‘aligning’ medicine prices with their benefits) will inevitably result in no consumer surplus, since there is no region between the price and the demand curve⁶. This is no more correct than pricing according to the developer’s ‘willingness to accept’ (an approach that would result in no producer surplus); neither is a ‘fair’ way to price medicines.

Ensuring a ‘fair’ price instead requires that the price be set somewhere below the demand curve and above the supply curve, such that the payer and manufacturer each receive a ‘fair’ share of the resulting economic surplus.

8.3 Maximizing Population Health

A common price (or cost-effectiveness ‘threshold’) based on $k$ has been posited as a means to ‘maximize’ population health [10, 47, 48]. Yet this is not true in the presence of a causal relationship between the common price and the ICERs of medicines. Far from ‘maximizing’ population health, a common price of $k$ allows for medicines to be priced up to the point where the population health gain is zero and the entire economic surplus is allocated to the producer. Instead, this framework finds that a common price below $k$ is required to maximize population health. Total welfare is also maximized at a common price below $k$, provided $v>k$.

It should be noted that both manufacturers and payers have incentives to depart from the welfare maximizing common price (with population health generally maximized at a lower common price, and profits maximized at a higher common price). The prices negotiated in practice between payers and manufacturers therefore depend upon a number of strategic considerations by both parties. For payers who are interested in maximizing population health, the game theoretic approach proposed by Pekarsky provides useful guidance for navigating these strategic considerations [49, 50]. Pekarsky’s work is of particular value for payers who consider the list prices for medicines in their jurisdiction to be ‘unfairly’ high and who wish to negotiate prices down to within the ‘fair’ range.

Regulators may also be interested in addressing the asymmetry of information between payers and manufacturers with respect to the lower boundary of the range of ‘fair’ prices. While both parties can observe the upper boundary (the common price specified by the payer), the lower boundary (the manufacturer’s reserve ICER) is assumed to be private information that is not observed by the payer. This asymmetry of information provides manufacturers with an advantage during negotiations with the payer. Regulatory efforts to provide greater transparency around manufacturers’ reserve ICERs could support payers in these negotiations, potentially improving population health outcomes.

8.4 Ensuring ‘Fair’ Prices Globally

It is reasonable to expect richer countries to contribute more than poorer countries towards the worldwide costs of developing medicines. At the same time, it is unreasonable to expect any country’s health care system to shoulder such a high price for medicines that reimbursement diminishes population health.

The proposed framework satisfies these principles. Both theoretically and empirically, supply-side thresholds are higher in countries with greater health spending per capita [7, 12]. This framework, and definition of a ‘fair price’, allows for higher prices in richer countries than in poorer countries, while ensuring that prices are not sufficiently high in any country to diminish population health.

8.5 Sustainable Innovation and Public Health Care Systems

Maintaining a sustainable pipeline of new medicines requires a reasonable rate of return for manufacturers. Maintaining sustainable public health care systems requires a reasonable rate of return for payers. The proposed framework, and definition of ‘fair’ pricing, therefore supports sustainable innovation and sustainable public health care systems over the long term.

8.6 Limitations of the Framework

Estimating the location of the supply curve, and in turn the producer surplus, is challenging for payers in practice. Manufacturers have an incentive to exaggerate development costs in order to increase the lower bound of the range of prices considered ‘fair’. If the supply curve is not observed by the payer, the lower bound of the range of ‘fair’ prices remains unknown to the payer in practice. To be considered ‘fair’, the price must be lower than $k$ to provide any consumer surplus; however, a price too far below $k$ would also not be ‘fair’ since it would result in no producer surplus. The payer cannot readily determine how far below $k$ the price can be set while ensuring producer surplus remains positive.

Estimating $k$, and hence the location of the demand curve, is also challenging, although empirical work has been attempted in a number of countries [12]. The methods used for these studies will likely improve, which might result in revised estimates of $k$ over time.

8.7 Future Research

A potential extension to this framework would be to consider whether other mechanisms for reallocating the economic surplus could be used as a supplement to pricing, in a way that increases the total welfare while remaining ‘fair’ under the principles adopted in this framework.

For example, it has been argued that, under strict assumptions, innovation is optimally incentivized when the entire economic surplus is allocated to the producer [51]. Under the basic framework in this paper, this would require a price of $k$. Although such pricing would not be considered ‘fair’ in and of itself (since there would be no consumer surplus), if this approach were to be supplemented with a reallocation of the economic surplus from producers to consumers (e.g. via greater taxation of manufacturers to support increased health care spending), then it might be possible to increase the total surplus while ensuring that the final allocation is ‘fair’ to both parties.

Alternatively, it might be argued that the price of a new medicine should be no higher than that which would arise in a competitive market (i.e. in the absence of a patent monopoly), since a higher price results in a ‘deadweight loss’ and reduced total welfare. However, such a price might be considerably below $k$, and would not be considered ‘fair’ in and of itself if it does not provide the manufacturer with a reasonable rate of return on its investment. Yet if such a price were to be supplemented with a reallocation of the economic surplus from consumers to producers (e.g. via tax breaks for manufacturers of specific medicines, or government grants to support biomedical research), then it might be possible to achieve a greater total surplus while ensuring that the final allocation is ‘fair’.

Caution would be needed when considering any such supplemental approach. The assumptions required for innovation to be optimally incentivized at a price of $k$ are strict and may not apply to the development of medicines [52]. The supplemental mechanism used to reallocate the economic surplus may also be distortionary, potentially reducing the total surplus. And if any supplemental reallocation results in greater or lesser health care spending, this would be expected to change $k$, which would in turn impact upon the determination of a ‘fair’ price [7].

The proposed framework also highlights the need for further research in a number of other areas. How can the common price that maximizes population health, or that maximizes total welfare, be estimated in practice? What does the distribution of reserve ICERs look like? What is the typical shape and location of a medicine’s supply curve? And what is a ‘reasonable’ rate of return for a payer when making risky investments in new medicines? Future research on these and many other considerations has the potential to extend the framework presented in this paper, enriching our understanding of how ‘fair’ prices for medicines can be determined in practice.