Each week, In Theory takes on a big idea in the news and explores it from a range of perspectives. This week we’re talking about drug pricing. Need a primer? Catch up here.

Jason Shafrin is a senior research economist at Precision Health Economics and director of research at the Innovation and Value Initiative.

Imagine a major medical breakthrough: a cure for Alzheimer’s. Imagine that cure not only would improve the cognitive abilities of the more than 5 million Americans living with Alzheimer’s but also would give these patients additional years of life. Imagine the economic impact of allowing these individuals to live at home, return to work and become productive members of society.

But what should be the price of a cure? Ideally, it should be based on the value that the drug gives to patients.

Let’s get this out of the way: Drug pricing is confusing. The media often focuses on the high sticker price of drugs, but due to varying levels of insurance coverage for patients and to large rebates and discounts given to insurers and pharmacy benefit managers, retail prices differ from what pharmaceutical firms receive or what patients and their insurers typically pay.

But what if drug pricing were simple, with patients buying drugs directly from pharmaceutical companies? Economics 101 tells us that the price of the drug would be set by patient demand for the drug and the cost to supply the treatment.

The production cost schedule (a.k.a. the supply curve) is a strange one. The cost of producing the first pill is large, $802 million on average. This cost comes largely from research and development expenses. The cost of producing the second, third or fourth pill, however, is small. Once you know the recipe, the raw materials are usually cheap. High school students in Australia were able to re-create the drug Daraprim — the expensive HIV/AIDS treatment of “Pharma Bro fame — for only $2 a pill. This is likely to be true for most medications.

Patients rightly argue that because the cost of actually manufacturing drugs is so cheap, drug costs should be low. If the cost patients paid for our hypothetical Alzheimer’s cure were too high, many patients who could benefit from the drug might forgo treatment due to cost considerations.

Drug companies, however, rightly argue that because the cost of the research and development is high, the price of the cure should be high. Like most people, pharmaceutical firms work harder when they expect greater rewards for their efforts. Without sufficiently high prices, no pharmaceutical company would decide to invest in the research necessary to identify a cure for Alzheimer’s and dementia.

Currently, this conflict between what patients should pay (very little) and what innovators should receive (much more) is resolved by the insurance system. Insurance companies reimburse pharmaceutical companies generously for novel treatments, and costs are distributed across the insured population through premiums. Sick patients with insurance coverage who benefit from a drug, on the other hand, typically pay a small share of the cost through co-pays. Economists have found that paying for drugs through the insurance system is economically efficient.

In our current insurance-based system, we want to strongly reward innovators that identify breakthrough treatments such as a cure for Alzheimer’s, but reward more modestly treatments that provide smaller gains in survival or quality of life. Tying a drug’s price to the value it creates will accomplish exactly that goal. Even leaders in the pharmaceutical industry such as the chief executives of Novartis and Allergan have recently proposed pricing drugs based on how they affect real-world patient outcomes.

Is centralized price-setting by the government the solution? In theory this could work, but in practice this is unlikely. Governments generally have little incentive to get prices right, because governments don’t go out of business if they get this wrong. Years of wrangling over Medicare’s reimbursement to physicians — the so-called “doc fix” — have demonstrated that the federal government is not good at setting prices. A market-based approach is imperfect, but it is the least bad tool we have for determining prices.

Approaches in which insurers and governments pay for treatments based on value and patient outcomes are more promising. For instance, Johnson & Johnson and the United Kingdom’s National Health Service agreed on a high price for a cancer treatment (Velcade), provided that J&J would refund the price for any patients for whom the tumor did not shrink. In the United States, Amgen and Harvard Pilgrim have a similar “money back guarantee” for patients taking PCSK9 cholesterol-lowering medications.

These novel pricing and data-sharing structures are still in their infancy, however, and further research is needed. At the Innovation and Value Initiative, where I am director of research, we’re working to create rigorous and transparent value assessments in health care. We’re hoping we can ensure that patients can afford their medicines; that insurers will pay a fair price; and that pharmaceutical firms will be incentivized to create the next cure for Alzheimer’s.

Read more: