Drug shortages: the price of keeping a factory ready

A familiar medicine can become hard to find even when several companies sell it. Those brands may rely on the same factory. In an April 2025 account of a quetiapine shortage, France’s medicines regulator, ANSM, described a Greek factory that normally supplied 60% of the French market through seven marketing companies. A quality problem and partial shutdown at that factory disrupted several commercial supply routes at once. Source: ANSM, 2024-2025 episode.

This raises a straightforward economic question. When a buyer pays for a medicine, does the payment cover only the units delivered, or also the resources needed to keep delivering during a disruption? A usable reserve, an alternative supplier or a backup production line costs money before it is called upon. Its value becomes most visible when ordinary supply fails.

On 8 September 2026, the European Commission approved €400 million in German aid for Sanofi to strengthen insulin supply. The arrangement combines a factory commitment, a production floor, an active-ingredient reserve and priority delivery. It offers a concrete way to examine the difference between buying medicines and paying for an obligation to maintain availability. Approval, however, establishes neither that the money has been disbursed nor that the arrangement will achieve its intended results. Source: European Commission.

Several brands can lead back to one factory

Consider a simplified medicine supply chain. One facility makes the active pharmaceutical ingredient, the substance responsible for the medicine’s effect. Further manufacturing turns it into the form intended for use, followed by packaging and distribution. These operations may be integrated or spread across different businesses. The FDA and the U.S. Government Accountability Office describe fragmented supply chains as one of the challenges in preventing and resolving shortages. Source: GAO’s 9 April 2025 report, “Background”.

Now imagine three competing brands buying from the same manufacturer. Their sales departments compete, but an important part of their production risk is shared. Switching brands adds no factory. Conversely, two genuinely independent manufacturers may have no spare capacity to replace one another’s lost output.

Quetiapine illustrates the first problem. It does not, by itself, establish that prices were too low or that a manufacturer deliberately neglected investment. ANSM describes the quality failure and its effect on deliveries. That account does not disclose the factory’s finances or investment decisions.

A WHO mapping published on 9 September 2026, covering work conducted from October 2024 to June 2025, identifies a related visibility problem in insulin. Its public summary distinguishes 22 full-chain manufacturers and 34 fill-and-finish manufacturers, and highlights opaque relationships with insulin crystal producers. These are company counts, not a count of independent suppliers that could immediately serve Europe. Source: WHO, official summary of “Global mapping of insulin manufacturers”.

Readiness has a cost before it has a customer

Suppose two fictional manufacturers can supply the same compliant product. One runs its equipment close to full capacity. The other maintains additional capacity, keeps materials available and prepares an alternative production route. If they sell identical volumes, the second has more costs to recover.

Reliability can protect sales and reputation, so manufacturers have reasons of their own to invest in it. But they do not necessarily capture its full value. If uninterrupted delivery saves a hospital from rearranging care, that benefit does not automatically become revenue for the supplier. And if the buyer cannot distinguish the two production arrangements, it may compare two prices without seeing that they finance different levels of preparedness.

The FDA’s 2019 report, revised in February 2020, identifies this kind of incentive problem: some less profitable medicines and more advanced quality-management systems may attract insufficient investment. The agency also describes difficulties in restoring supply after a disruption. This is a diagnosis of particular market vulnerabilities, especially for some older products, not a finding that all medicines are underpriced. Source: FDA, report and revised summary.

A small fictional calculation makes the issue tangible. A manufacturer sells ten million units a year. Keeping additional capacity ready costs €500,000 annually. Spread over the sales volume, that is five cents per unit. The calculation says nothing about whether the expenditure is worthwhile: that would require the likelihood and consequences of disruptions, and the effectiveness of the backup. It simply shows how identical products can carry different costs of maintaining availability.

Nor would a five-cent price increase establish that the money had been invested in reliable supply. Payment still needs to be linked to a defined service. Without a measurable commitment, higher revenue does not tell the buyer how many extra units could actually be delivered after a shutdown.

The lead time for replacement supply

In a simplified market story, a shortage attracts new supply. Timing is the missing variable. GAO points to constraints involving capital, technology, active ingredients and approval of new manufacturing sites or suppliers. Those constraints prevent a production increase from being as immediate as a purchase order. Source: GAO, “Causes of Drug Shortages”.

Theoretical capacity is therefore not enough. A business might have a building without a suitable production line, a line without available inputs, or output that still needs to pass the necessary checks before release. Deliverable volume depends on the stage that actually limits the chain.

Compliance and continuity are different attributes. A compliant medicine meets the requirements that apply to it. That does not establish that its manufacturer can supply more of it next week. Dispensing with controls does not create industrial redundancy; it changes the risk borne by patients. European prevention guidance instead combines quality systems, advance planning and coordination. Source: EMA’s 2023 recommendations.

FDA’s historical study provides some evidence of the slow response. Among medicines entering shortage between 2013 and 2017 for which this calculation was possible, 42 out of 139, or about 30%, regained at least their pre-shortage volume at the study’s measurement point: after twelve months, or when the shortage ended if that happened earlier. This is not a measure of shortages in 2026. It shows that lost supply did not reliably trigger a prompt replacement in that historical sample. Source: FDA, revised Appendix F, printed page 121.

Inventory buys time

Inventory and additional production perform different jobs. The first supplies something already manufactured. The second replaces part of a missing flow. They may work well together, but one is not a substitute for the other in every situation.

Imagine a fictional product with demand of 1,000 units a day. Initial inventory is 14,000 units, equivalent to fourteen days of normal demand. An interruption removes 60% of deliveries for sixty days. Unaffected suppliers continue to deliver 400 units daily.

Inventory does not fall by 1,000 units each day. It fills the 600-unit shortfall. It fully covers twenty-three days and part of the twenty-fourth. By the end of the sixty-day disruption, 22,000 units of demand have gone unmet.

Now add a backup delivering 200 units daily from day fifteen. The daily call on inventory falls from 600 units to 400. The first partially unserved day moves to day twenty-nine, while total unmet demand falls to 12,800 units. In this example, the extra supply provides meaningful but incomplete protection.

If the backup instead delivers 600 units a day from the same date, all demand is served during the interruption and 5,600 units remain in stock. Initial inventory, replacement volume and mobilisation time jointly determine the outcome. These are teaching calculations, not Sanofi stock figures or medical doses.

The model deliberately excludes expiry, batch recalls, rising demand and differences between product presentations. It also assumes the backup remains available while the main source is disrupted. Remove that assumption and the backup’s advantage disappears. A supplier list cannot replace an examination of common dependencies.

Insulin puts an availability contract into practice

Germany’s arrangement, approved on 8 September 2026, uses public service compensation for a service of general economic interest. In practical terms, a public authority pays for defined obligations beyond ordinary product sales. The announcement includes a new factory at Frankfurt-Höchst by the end of 2032, annual production of at least 1.1 tonnes through the end of 2042, a one-tonne active-ingredient reserve through the same date, and priority for European Economic Area markets during shortages. Source: European Commission, “Die deutsche Maßnahme”.

These commitments describe different things. A factory is a productive asset. Annual output is a flow. Inventory is an amount held at a particular time. Delivery priority allocates scarce supply. Treating them as a single promise of security would lose the most useful questions: which disruption can each commitment absorb, and which vulnerabilities remain?

A tonne of active ingredient is not a tonne of ready-to-dispense medicine. It may protect against a shortage of material upstream without solving a failure further down the production chain. Converting it into “months of treatment secured” would require the relevant product mix, consumption volumes, conversion yields and available downstream capacity. The announcement does not establish that calculation.

The chronology matters too. On 1 August 2024, Sanofi announced an investment of about €1.3 billion with a 2029 horizon, already conditional on public support and European approval. The new facility was intended to replace existing installations. The announcement therefore does not establish that two independent factories will operate in parallel. Source: Sanofi’s original announcement, a corporate source.

The industrial timetable presented in 2024 and the deadline in the commitment approved in 2026 are not enough to establish a three-year delay without further explanation. Dividing €400 million by €1.3 billion would be similarly premature as a way to calculate a subsidy rate: compensation for net costs associated with obligations over time and an announced investment do not necessarily have matching scopes.

What the Sanofi compensation pays for

It would also be misleading to treat insulin as interchangeable with every low-priced generic. WHO’s mapping distinguishes manufacturing roles and barriers to accessing insulin markets. FDA’s findings on older injectable medicines do not provide the accounts of the Frankfurt operation. The evidence from one market cannot establish the cost structure of another. Source: WHO’s insulin mapping. FDA’s analysis of shortage causes.

A profitable group can own a less profitable activity. That possibility does not establish that a particular public payment is necessary. The opposite shortcut is no better: a medicine’s importance cannot, on its own, determine how much money is needed to secure its supply.

The Commission’s assessment assumes closure without aid and concludes that compensation will not exceed the net costs of the public service, with provisions to prevent and recover overcompensation. These are the Commission’s findings, not calculations that readers can independently reproduce from the announcement. Source: European Commission, “Die Bewertung der Kommission”.

An independent economic assessment would need the costs included, expected revenue, compensation period, permitted return and alternatives considered. Monitoring delivery would require production, usable inventory, milestones and failures. The sources reviewed do not disclose that full picture. They establish announced commitments, not a measured industrial outcome.

Different payment designs leave different risks

A contract can pay only for purchased units, provide visibility over future volumes, or explicitly remunerate an availability obligation. OECD’s analysis considers awards to multiple suppliers, reliability and inventory requirements, and combinations of incentives and penalties. Their effects depend on market structure and contract execution. Source: OECD, chapter 2, section 2.2.1.

The trade-offs become apparent when the commitments are made concrete. Reserving capacity can make it unavailable to another buyer. Guaranteed volumes may give the supplier stability but bind the purchaser if demand changes. Dividing an order may reduce dependence, yet creates no second factory when both winning bidders use the same contractor. A penalty can alter incentives; it cannot manufacture the missing units during an interruption.

Geography is not the whole answer either. In a scenario where two nearby plants share an electricity supply or a critical input, they remain exposed to a common disruption. In another, distant plants have different dependencies but longer delivery times. The relevant comparison concerns the full supply chain and the particular disruption, not simply the location of the final facility.

These limitations do not make prevention pointless. They prevent a price increase, an inventory reserve or a new factory from being treated as a universal guarantee. OECD also notes that evidence on the effectiveness and cost-effectiveness of the available measures is not always sufficient. A plausible mechanism still needs to be checked against the deliveries it actually preserves. Source: OECD’s 2024 report, foreword.

A pack contains a product. Continued access depends on a production system that must still be usable tomorrow. The economic task is to make that system visible, distinguish the protections it provides and identify the obligations being financed. Only then can the price of availability be discussed separately from the price of the pack.

Further reading

For a different industry facing difficult-to-replace production capacity, see aircraft engines and maintenance. The sectors, lead times and health risks differ; the comparison concerns the cost of replacement. The glossary also explains active pharmaceutical ingredients and services of general economic interest.

Sources

Sources checked on 17 September 2026. Historical episodes are dated; future commitments are not described as completed.

FDA. Drug Shortages: Root Causes and Potential Solutions, October 2019, revised February 2020. Summary, economic causes and Appendix F, printed page 121. Shortage-entry cohort: 2013-2017.

GAO. GAO-25-107110, 9 April 2025. Manufacturing and regulatory constraints; “Background”.

ANSM. Quetiapine supply update, first published 24 April 2025; page updated 5 September 2025. The passage used describes dependencies in the episode that began in 2024.

WHO. Global mapping of insulin manufacturers, 9 September 2026. Official summary; mapping conducted from October 2024 to June 2025. Company counts do not measure capacity available to Europe.

European Commission. Approval of German aid for insulin supply, 8 September 2026. The measure and the Commission’s assessment.

Sanofi. Original Frankfurt investment announcement, 1 August 2024. Corporate source used for the company’s own announcement and conditions.

OECD. Securing Medical Supply Chains in a Post-Pandemic World, chapter 2 and foreword, 23 February 2024. Visibility, procurement, diversification and limits to policy evaluation.

EMA. Guidance for industry to prevent and mitigate medicine shortages, 2023. Prevention, quality and coordination recommendations.

Limitations

This article examines manufacturing and contracting mechanisms. It is not a live medicine-availability bulletin, medical advice or a clinical assessment of treatments. The quetiapine episode does not establish the financial cause of the quality problem. WHO’s summary does not allow every manufacturing relationship to be reconstructed. The detailed accounts and full contract needed to audit the Sanofi compensation were not obtained.

The simulator uses fictional product quantities, constant demand and an assumed independent backup. It does not forecast shortages, value health consequences or recommend personal medicine stocks. Demand unmet during the interruption is not assumed to be made up later.

Original l0g text, calculations and graphics: CC BY 4.0. Source documents retain their own reuse terms.