Executive Summary

The United States is the world’s largest market for pharmaceuticals, and the world’s leading source of medical innovation. But it also has the highest drug prices by far, putting many of the cutting-edge therapies out of reach for millions of Americans.

The high price of prescription drugs consistently ranks at the top of Americans’ health care concerns, according to polls. The pharmaceutical industry and its allies, however, have long argued that high drug prices are a good thing, because the profits from these price hikes allow companies to spend more on research and development to discover and launch the cures of the future.

Profit growth at the largest pharmaceutical companies—driven by price hikes on older drugs—rarely leads to the development of innovative new medicines.

To test the claim that higher drug prices drive innovation, we gathered data from the industry, individual companies, and the FDA to conduct a counterfactual analysis: what would happen at some of the largest pharmaceutical companies in the world if prices on certain blockbuster drugs had remained constant over the last 10 years?

Specifically, we selected one drug at each of the 17 largest publicly traded drug companies with substantial sales in the United States. Together, these companies represent more than 60 percent of total pharmaceutical sales worldwide and the 17 drugs represent 12.2 percent of net U.S. pharmaceutical sales.

Links to the individual company analyses: AbbVie’s Humira | Amgen’s Enbrel | Astellas’ Xtandi | AstraZeneca’s Lynparza | Biogen’s Tysabri | Bristol Myers Squibb’s Orencia | Eli Lilly’s Alimta | Gilead’s Truvada | GlaxoSmithKline’s Benlysta | Johnson & Johnson’s Stelara | Merck’s Keytruda | Novartis’ Gilenya | Novo Nordisk’s Victoza | Pfizer’s Xeljanz | Roche’s Avastin | Sanofi’s Aubagio | Takeda’s Entyvio

The analysis in this paper builds off of one we published in 2021 for Abbvie Inc., maker of Humira, the largest U.S. drug by revenue. In that analysis, we found that Humira price increases of $42 billion over a decade led to the production of a single new drug.

In our new analysis, we find that the savings to the public from constant unit drug prices would have been substantial: $139.2 billion from 2012–2021, or 36.7 percent of U.S. sales on the drugs sampled. We estimated that such a price freeze would have reduced R&D spending by $25 billion over the same timeframe. But that reduction in R&D spending, we estimate, would have had a minimal impact on the development of innovative new medicines. Under the counterfactual scenario, the number of drugs approved by the U.S. Food & Drug Administration would have only declined by one percent.

It turns out that the largest pharmaceutical companies are extremely inefficient in turning R&D spending into FDA-approved medicines. While researchers have estimated that the pharmaceutical industry spends about $1 billion on average to successfully discover and develop one new drug, inclusive of failures along the way, the 17 companies in our analysis spent 3 to 10 times that amount.

Put another way, we found that it would take a $30 billion decline in revenue among these 17 companies to result in one fewer FDA-approved drug.

In other sectors of the economy, researchers have long understood that most innovation happens at small companies, not large ones. Large companies have bureaucratic cultures that strangle innovative ideas. It should not be surprising that we observe the same phenomenon in the pharmaceutical industry.

The key insight for policymakers is that profit growth at the largest pharmaceutical companies — driven by price hikes on older, branded, monopoly drugs — rarely leads to the development of innovative new medicines.

The prescription drug provisions of the Inflation Reduction Act, enacted by Congress in 2022, reflect these principles. The legislation empowers Medicare to negotiate government subsidies on a small fraction of drugs that drive Medicare’s costs. There is much more to do, however, if we are to achieve a health care system that is fiscally sustainable for taxpayers and affordable for patients.

In particular, Congress needs to unleash the power of market-based competition to reduce prices and costs of prescription drugs, especially for medicines that have been on the market for longer than seven years.

Not all novel drugs are innovative

The term “innovation” is often used loosely when it comes to drug development. Just because a drug has been approved by the FDA does not mean that it is “innovative,” nor that it is a “cure,” nor that it has broad implications for public health.

True cures vs. incremental benefit. Many FDA-approved drugs provide a modest clinical benefit, while a small minority can truly be called cures.

Common diseases vs. rare diseases. Many newly approved drugs treat extremely rare diseases that fewer than 10,000 Americans experience in a given year; today, it is relatively less common for companies to develop drugs that treat diseases that tens of millions of Americans have.

Innovative technologies vs. standard methods. Most newly-approved drugs build on standardized techniques for drug development, just as newly-marketed cars build on standardized techniques for their manufacture. Rare is the drug that represents a conceptual leap forward in drug development, such as the use of chimeric antigen receptor T cells (CAR T) to stimulate a patient’s immune system against cancer cells; rare, too, is the drug that represents a true cure for a common disease, such as the development of the nucleotide analogue Sovaldi (sofosbuvir) for hepatitis C.

Ultimately, the kind of innovation that can deliver the most value for Americans is that which significantly improves outcomes for the biggest public health problems like heart disease, cancer, Alzheimer’s, diabetes, and most recently, COVID-19. Not all newly-approved drugs meet this standard.

For the purposes of this study, however, let us charitably assume that all newly approved drugs are “innovative.” Do price increases on old drugs drive new drug development?

How much do companies spend to produce one new drug?

Several studies have sought to quantify the R&D spending needed to develop a new drug. Two estimates are particularly relevant to his paper.

Woulters et al. (2020) estimated the capitalized R&D cost to bring a new drug to market at $985 million; that is, the total cost including failed drug candidates. On the other hand, another widely cited study (DiMasi et al., 2018) estimated the capitalized cost to develop a new drug at $2.9 billion.

The majority of drugs sampled in the Woulters study were from small firms with publicly available data, while the drugs sampled in the DiMasi study were predominately from large firms, provided confidentially to DiMasi and colleagues.

Does decreased pharma revenue directly correlate with decreased R&D spending?

Implicit in the assumption that a decline in revenue leads to a decline in innovation is that R&D activity is closely linked to revenue. Several researchers have examined this phenomenon and attempted to quantify the magnitude of the association.

Dubois et al. (2015) estimated pharmaceutical companies expect on average $2.5 billion in additional revenue to develop a new drug, based on a large dataset of 630 drugs sold across 14 countries between 1997–2007. The study arrived at this revenue figure based on a calculated innovation to market size elasticity of 0.23, meaning that increasing market size by one percent yields an increase in the number of new products by 0.23 percent. The $2.5 billion figure included R&D costs (fixed capital costs) and marketing and related costs (variable costs) equal to 50 percent of revenue. The authors concluded that an elasticity below one “is plausible in light of other evidence that innovation in pharmaceuticals is becoming more difficult and expensive over time…that the costs of regulatory approval are rising and…that the industry is running out of ‘low hanging fruit.’”

Do rising prices on old drugs help produce new medicines?

Several studies also sought to describe and predict drug pricing behavior among firms of different sizes.

For launch prices, new research from Rome et al. indicates companies have introduced drugs to market at higher prices over time, both for list prices as well as net prices inclusive of rebates and discounts. After adjusting for drug characteristics, mean list price launches for drugs introduced between 2008–2021 grew exponentially by 13 percent per year, while net prices rose 10.7 percent. These higher prices at launch hold even when accounting for the increased complexity of molecules developed by the industry.

As for price changes following a new drug launch, the research overwhelmingly indicates that manufacturers routinely increase list and net prices for branded drugs. An investigation by Hernandez et al. showed that, from 2007–2018, list prices increased by 9.1 percent per year, while net prices increased 4.5 percent per year. Particular classes of drugs had especially high increases. For example, multiple sclerosis drugs rose by 157 percent on net, while lipid-lowering agents rose by 95 percent.

The pharmaceutical industry argues that price increases are needed to fuel future drug innovation. Our previous work challenged this notion, by demonstrating that a bigger driver of new drug development is low up-front R&D costs, rather than high back-end price increases.

Methodology

For this paper, we sought to assess the impact on R&D productivity of a counterfactual scenario in which U.S. unit prices remained constant for the last 10 years on several of the highest revenue drugs at the largest pharmaceutical companies in the world.

For a selected drug manufactured by each company, we constructed a scenario in which net prices did not increase. We sought to measure the degree to which a pharmaceutical company converted R&D spending into novel therapies, and used that measure of efficiency to determine how many novel drugs would have been lost without said price increases.

The analysis was completed in four steps:

Drug sample construction. To construct the sample, we gathered data from the pharmaceutical industry to find the largest pharmaceutical companies by U.S. revenue, using publicly available data. The screening process yielded 17 companies. At each company, we selected one drug that met the following criteria:

1. The drug is a branded drug that experienced market exclusivity due to enforced patents or other legal means;
2. The drug’s exclusivity period (i.e., with no generic or biosimilar competition) lasted for more than half of the time period examined; and
3. Full-year U.S. sales data exists for the drug for a minimum of five years.

The drugs selected for the analysis are described below.

Drug price decomposition. For each year a company reported full-year sales data for the selected drug, we calculated an estimated net price based on net sales revenue and number of units sold. We then decomposed each drug’s U.S. sales growth relative to the first year of sales (baseline year) into two categories: revenue gained by price increases and revenue gained by prescribing volume increases. Using the price-driven revenue figure, we calculated the loss in R&D spending as a result of holding the selected drug’s price constant, assuming that a one-percent reduction in revenue would result in a one-percent reduction in R&D spending.

Company R&D efficiency. To find the impact of lost R&D spending on the number of novel drugs developed, we needed to determine each company’s R&D efficiency, defined as R&D spending per novel drug developed. We defined a novel drug as one that received a “novel drug approval” designation by the FDA, included in the agency’s list published each year. Since there are many ways to “develop” a drug, including in-house or by acquiring the rights to sell a drug after its discovery, we performed a sensitivity analysis under four specifications: 1) the company is awarded full credit for each drug it originates, either on its own or in collaboration with another company, 2) the company is awarded varying levels of credit for each drug based on a sliding scale of clinical development, 3) the company is awarded full credit for each drug owned by the company, whether originated or later acquired, and 4) the company is awarded full credit for each drug in the company’s portfolio — including drugs owned and those for which the company has a license to sell the drug on behalf of the drug’s owner.

The table below summarizes the four specifications used to calculate a range of each company’s R&D efficiency.

Estimating the number of novel drugs lost. Finally, for each drug development specification, we calculated the number of forgone novel drugs developed due to lost R&D spending.

The relationship between revenue and R&D spending

This paper sought to quantify the impact of reduced revenues to novel drug development at large pharmaceutical companies. We report the results of the analysis in several steps. First, we present descriptive statistics of the 17 companies and drugs selected. Second, we present company R&D efficiency reported as a range based on the four specifications under which each company is credited with an novel drug’s development. Third, we present the revenues and R&D expenditures lost for the counterfactual scenario in which each selected drug’s net price remains constant throughout the time period examined. Fourth, we report the estimated number of novel drugs lost based on the calculated loss in R&D spending. Finally, we present each company’s results as individual case studies.

From 2012–2021, the 17 companies in the analysis accounted for 60 percent of global revenues and 56 percent of global R&D spending.

Of the 17 drugs sampled, 16 exhibited net price increases over each drug’s span of full-year sales data within the last 10 years. The only exception was Novo Nordic’s Victoza, which had a net price decrease of 21.6 percent from 2011–2021, due to superior products entering the market. Nevertheless, the drug’s net price increased 32.2 percent from 2011–2017. Each drug amassed net positive revenue over the time period solely from price increases, ranging from $1.5 billion (Eli Lilly’s Alimta) to $49.0 billion (Abbvie’s Humira).

With a few exceptions, U.S.-based revenue derived from price increases on each drug represented a modest 0.7–3.7 percent of each company’s worldwide revenue. United States-based price increase revenue on three drugs stood out: Humira (16.2 percent of worldwide revenue), Amgen’s Enbrel (9.6 percent), and Biogen’s Tysabri (6.6 percent).

In a counterfactual scenario in which prices remained constant on each drug, the corresponding loss in revenue for the 17 companies combined for $139.2 billion. Given the rate of investment in R&D from the 17 companies relative to company revenues, this reduced spending is estimated to reduce R&D investment by about $25.0 billion over the same period, ranging from $255 million less R&D spending at GlaxoSmithKline to $7.2 billion less at Abbvie. The lost R&D spending represented 18 percent of total revenue lost, meaning U.S. taxpayers would gain $5.57 in savings in exchange for $1.00 less in R&D spending.

The following graph summarizes savings achieved and resulting R&D spending lost in the absence of price-driven revenue for each company’s drug in the analysis.

R&D spending per new drug developed

We found that the 17 companies collectively held 219 novel drugs approved by the FDA when each drug selected was marketed in the United States. Of those, 51 percent (111 drugs) were originated by the companies, either alone or in partnership with another firm. In contrast, 49 percent (108 drugs) were acquired later in the development process or in-licensed from an originating company.

With each company’s drug approvals in hand, we calculated each company’s efficiency in developing new drugs, defined by R&D spending per novel drug developed. When considering the four specifications in defining the number of drugs developed by each company, the companies spent $4.9 billion per novel drug at the median (interquartile range [IQR]: $4.1-$6.8 billion). This amount is up to five times higher than the typical amount cited in the academic literature, indicating larger firms are less efficient at converting R&D spending into cures.

The graph below summarizes the range of R&D costs per new drug developed at each company based on drug development specification.

Depending on how one defines novel drug “development,” the largest pharmaceutical companies in the world spend far above the average industry-wide cost to develop a new drug. The sensitivity analysis ran four specifications, with companies receiving varying levels of “credit” for drug development: 1) full credit only for drugs originated in-house, 2) a “Partial Developer” model giving varying levels of credit depending on company’s involvement in developing the drug, 3) full credit for each drug developed, including those currently owned by the company but originated elsewhere, and 4) full credit for all novel drugs in the company’s portfolio, including in-licensed drugs. No matter which model is used, company data confirms that collectively, the largest companies are less efficient than the industry average at turning R&D spending into novel drug approvals.

Impact of lost R&D on new drugs developed

Finally, based on the 17 companies’ efficiency rates in discovering and developing novel drugs under a variety of strategies, we would expect the reduced R&D spending to result in about five fewer novel drugs from 2012–2021 (4.73, IQR: 3.81–5.87). Five fewer drugs represent 1.2 percent fewer novel drugs than were approved by the FDA over the same time period.

Put another way, the companies in our analysis would collectively develop one less novel drug if they earned $29.8 billion less revenue.

Drug price reform and R&D productivity are not mutually exclusive

The results show that significant savings can be achieved with minimal impact to new drug development.

To put the results in context, $139 billion savings represents 4.8 percent of nationwide prescription drug spending from 2012–2021. In contrast, the five drugs lost due to reduced R&D spending at the 17 companies represents just 1.2 percent of all novel drugs approved by the FDA over the same period.

This study does not capture the full extent of savings and impact to R&D if all brand-name drug prices remained flat over the last decade. Several other drugs, including Johnson & Johnson’s Darzalex, Pfizer’s Ibrance, and Roche’s Perjeta exhibited increasing net prices since these products launched, but were not included in the analysis.

We are therefore certain that additional savings would be obtained if net prices did not increase for all branded drugs sold by the companies in our analysis. Conservatively, we estimate that absent price increases on other drugs in the companies’ portfolios, the United States would have spent $200 billion less on prescription drugs from 2012–2021. The impact to R&D spending would result in an estimated seven fewer novel drugs; again, a minimal impact given 430 novel drugs were approved by the FDA over the same 10-year period.

Importantly, the point estimate of 4.73 fewer drugs developed aggregates the decrement to new drugs from each company, many of which would experience fractional losses of one drug developed for which they could easily adjust R&D spending priorities to minimize the impact. In other words, companies that would suffer a drug decrement of less than 0.33 drugs would likely be able to absorb the reduction in R&D spending without harming the company’s drug development pipeline. Indeed, many companies could adjust other expenditures, such as administrative and marketing expenses, to compensate for less R&D spending. Alternatively, companies could simply maintain R&D spending levels, resulting in a somewhat higher percentage of revenues spent on R&D. While companies may suffer a short-term impact to profitability, they are likely to be rewarded with more drugs developed in the future and higher long-term profitability versus the alternative of reducing R&D activity.

Regardless, the paper’s resulting ratio of 1.2 percent drugs lost to 4.8 precent savings represents an novel drug-to-revenue elasticity of 0.27. These results are consistent with academic research. For example, Dubois et al. calculated a nearly identical elasticity of novel drug to market size of 0.23.

Given recent estimates of $1 billion to $2.9 billion to develop a new drug, the results also indicate that the largest pharmaceutical companies fare even worse — punching well below their weight in discovering and developing novel drugs. Shockingly, if the companies in this paper conservatively developed drugs at $2.5 billion per new drug, they would theoretically double new drug development over today’s rate.

While not the subject of this paper, one must ask what the United States could do with the $139 billion in savings from the 17 drugs we analyzed alone. Historically, the National Institutes of Health has an annual budget of $42 billion. $139 billion could provide a 33 percent boost over historic levels for 10 years. But perhaps more importantly, such savings could be returned to the American people in the form of lower insurance premiums and out-of-pocket spending on prescription drugs.

Finally, though this paper shows that lower U.S. drug prices will result in incrementally less money flowing into research and development, the results show this justification to forgo drug price reform is insufficient. After all, with that logic, drug makers should triple the price of every drug in order to incrementally increase R&D spending. Rather, because the data shows the tradeoff between lower prices and innovation is much smaller than believed, making prescription drugs more affordable should take precedence.

Consistent with this idea, the Fair Care Act of 2022 compliments the drug price reforms in the Inflation Reduction Act by increasing competition, particularly in the biologics market, while encouraging innovation by accelerating approval of innovative drugs. Under reforms such as the Fair Care Act, Congress can ensure Americans have greater access to innovative drugs at affordable prices.

Appendix

Individual company analyses

The following chart shows the 17 companies used in our analysis, by order of annual price-driven revenue on the company’s drug selected. For a detailed analysis of each drug’s pricing and effects on company drug development, hover over each company and click on the “Company analysis” link in each popup.

The drug development process

The drug development process begins with basic biological research, including screening thousands of molecules with the potential to treat disease. But unlike the painstaking trial and error of yesteryear, today’s screening processes use powerful computing and artificial intelligence to effectively and quickly screen for drug candidates. More than ever before, startup companies can compete with larger, established industry players through leveraging this technology.

In the screening process, researchers first determine a gene, protein, or chemical process that causes disease. This is the drug target. Based on the target’s characteristics, researchers may run thousands of compounds through test systems — known as assays — for molecular, biochemical, and cellular interactions with the target. Using tools and processes such as high-throughput screening (HTS) and Hit to Lead (H2L), researchers identify the most promising candidates and further modify them to improve efficacy and reduce side effects. At this stage, thousands of potential candidates become a few hundred.

Following lead optimization, companies begin preclinical trials testing a candidate’s therapeutic effects in vivo (mice), in vitro (laboratory), and ex vivo (animal cells or tissue from a non-living animal). A preclinical trial tests the drug’s absorption, metabolic characteristics, and mechanism of action. Researchers also learn the drug’s best dosage and administration route, efficacy, side effects, and how it interacts with other drugs. Through preclinical development, researchers trim potential drugs to a handful of candidates.

Upon successful completion of preclinical trials, drug candidates enter clinical testing in human subjects. Phase I clinical trials involve healthy volunteers that test a drug’s pharmacokinetics ; that is, how the drug moves through the body from absorption to metabolism to elimination. Upon successful completion of Phase I trials, a drug will enter Phase II and/or Phase III testing involving members of the patient population.

Along every step in the process, researchers collaborate with the FDA to ensure studies meet standards of scientific rigor. Upon submission of a New Drug Application (NDA) or Biologic License Application (BLA), the FDA reviews clinical data and determines whether the drug is safe and effective for use.

The drug development process does not end with FDA approval. Following approval, the drug’s sponsor conducts post-market monitoring and further clinical testing (Phase IV) to continually assess efficacy and safety over the long term.

While drug development is best explained as a step-by-step process, it is often not linear. A drug sponsor may need to revisit earlier steps in the process to iteratively develop and improve a candidate until it produces the desired effect.

Industry trends in drug development

Several trends in the industry dictate which diseases are targeted, which companies develop novel drugs for those disorders, and the pace of drug development to treat them.

Historically, most clinical development began with basic research at non-profit research centers, government laboratories, and universities. Then, larger and established for-profit companies would leverage that research to develop suitable drug candidates exclusively in-house, from original discovery through FDA approval and post-market monitoring.

However, the industry changed significantly in recent years. In our paper “High Drug Prices Don’t Accelerate Innovation — Lower R&D Costs Do,” we showed that the majority of novel drugs are now originally conceived at startup companies that are unprofitable and unable to sustain operations without cash from outside sources such as venture capital. Advances in life sciences and computing technology have allowed such companies to make more ground-breaking discoveries than ever before.

On the other hand, large legacy firms play a major role in guiding promising drugs through the latter stages of clinical development. In addition, a wide array of collaborative efforts across firms of all sizes and academic institutions exist to leverage expertise and resources.

Government policy and market dynamics often dictate how companies work to develop novel drugs, including which diseases to treat. For example, the Orphan Drug Act of 1983 was designed to incentivize companies to develop drugs for rare diseases. Decades after the law’s passage, about half of all Phase III clinical trials are for drugs intended to treat rare conditions that collectively afflict less than 0.5 percent of the population. Additionally, companies focus their efforts on treating rare diseases for which competition is low and profit potential is high.

Detailed methodology

For this paper, we hypothesize that if U.S. prices remained constant for the last 10 years on several of the highest revenue drugs at the largest pharmaceutical companies in the world, the savings would be substantial to the American public, without significantly harming innovation.

To test this claim, we constructed a counterfactual scenario in which net price increases did not occur on a selected drug at each of the largest pharmaceutical companies as defined by revenue. We sought to measure the degree to which a pharmaceutical company converted R&D spending into novel therapies and used that measure of “efficiency” to determine how many novel drugs would have been lost without said price increases.

To construct the sample, we gathered data from the pharmaceutical industry. We started with the 23 largest pharmaceutical companies in the world, defined as companies with $10 billion or more in revenue for 2021. To obtain robust sales data, we screened out companies that omit key data from financial reports, are privately owned, specialize primarily in generic drugs, or have U.S. drug sales below 25 percent of worldwide company sales.

The screening process yielded 17 large, publicly traded companies with significant U.S. pharmaceutical portfolios. For each of the 17 companies included in the analysis, we excluded worldwide revenue and R&D spending not related to pharmaceutical products. We then examined one drug in each company’s portfolio based on three criteria: 1) the drug is a brand name; 2) the drug remained exclusive, with no generic or biosimilar equivalents on the market, for more than half of the time period examined; and 3) full-year U.S. sales data exists for a minimum of five years — i.e., data exists at least since 2017.

The drugs selected for the analysis are described below.

Citing data from Symphony Health, we gathered U.S. net sales revenue and the number of units sold in 2011 or the first year with significant U.S. sales to calculate each drug’s baseline sales volume and price, net of market-wide discounts and rebates. We then repeated these calculations for each year the drug was marketed through 2021. We then decomposed each drug’s U.S. sales growth relative to the first year of sales (baseline year) into two categories: revenue gained by price increases and revenue gained by prescribing volume increases.

Using the price-driven revenue figure, we calculated the reduced R&D spending as a result of holding the selected drug’s price constant. To do so, we assumed R&D spending moved in tandem with revenue, such that a one percent reduction in revenue would result in a one percent reduction in R&D spending, as follows:

where for time period (x), RD(x) represents total company R&D spending and D(x) represents the percentage decrement to company revenues from holding prices on the selected drug constant.

To find the impact of reduced R&D spending on the number of novel drugs developed, we needed to determine each company’s R&D efficiency, defined as R&D spending per novel drug developed. But in order to do so, we first need to define what it means to develop a novel drug. For our analysis, we define a novel drug as one that received a “novel drug approval” designation by the FDA, included in the agency’s list published each year.

Determining whether a company “developed” an novel drug is more difficult when considering different types of collaborative agreements between drug companies. A strict definition of development would include only drugs that were discovered and brought through clinical trials from Company A’s own R&D efforts. The broadest definition would also include drugs in-licensed by Company A from Company B, on the assumption that such drugs would not be developed and approved if not for the financing and expertise of Company A.

To account for various meanings of drug “development,” we performed a sensitivity analysis under four specifications (y), defined below, to credit each company with novel drugs discovered.

Specification 1. The company is awarded credit (one point per drug) only for drugs it originated, either on its own or in collaboration with another company. This scenario is the most restrictive, awarding credit to companies only for drugs they discovered in-house rather than obtaining the rights to sell the drug later in the clinical development process.

Specification 2. The company is awarded varying levels of credit for drug development based on a sliding-scale point system. Drugs developed by the company in-house from the time the drug is originally invented is given one point, and drugs developed by the company in-house from the time the drug is originally invented, in collaboration with another company, is given 0.75 points. For drug development rights obtained after successful Phase I trial results, companies that solely own or collaborate on development through the drug’s FDA approval are awarded 0.5 and 0.25 points, respectively. In this way, drugs that are discovered and developed in-house are given greater credit because the company took a greater risk than a company that obtains the rights to sell a drug following successful early-stage trials.

Specification 3. The company is awarded credit (one point per drug) for each drug owned by the company, whether the drug was developed in-house since its original discovery or was first developed by the company during a later phase of clinical development — i.e., the company began developing the drug following successful Phase I trials completed by the drug’s originator. This scenario excludes drugs that are strictly in-licensed by the company, but nevertheless puts a greater emphasis on R&D spending at any point in a drug’s clinical development cycle.

Specification 4. The company is awarded credit (one point per drug) for each drug owned or for which the company has a license to sell the drug — i.e., all novel drugs in the company’s sales portfolio. This method emphasizes the role large companies may play in encouraging startups to develop new drugs on the prospect that a large company purchases the rights to the drug once it shows sales potential.

The table below summarizes the four specifications used to a range of each company’s R&D efficiency.

Therefore, the equation for R&D efficiency under specification (y), RDE(x)^(y), is defined as:

where for time period (x), RD(x) represents total company R&D spending, and ID(x)^(y) represents the number of novel drugs developed under specification (y).

Finally, with R&D spending lost due to constant prices on the company’s selected drug and the company’s R&D efficiency, we calculate the number of novel drugs lost, IDL(x)^(y), for each specification (y) as follows:

where for time period (x), RDL(x) represents R&D spending lost and RDE(x)^(y) represents R&D efficiency under specification (y).

To illustrate, consider the example of Perjeta, one of Roche’s drugs not included in this paper’s analysis. A monoclonal antibody used in the treatment of metastatic HER2+ breast cancer and sold since 2013, the drug has eight years of full-year sales data from 2014–2021; this represents time period (x) in this analysis.

During that time, Roche earned $9.4 billion in net U.S. sales on Perjeta. Of this amount, $2.1 billion came from price increases alone, accounting for 0.58 percent of the company’s $351 billion in worldwide revenue. This yields D(x) = 0.0058. Since we assume Roche would spend 0.58 percent less on $86.3 billion in R&D spending without Perjeta’s revenue from price increases, Roche’s decreased R&D spending over the eight years, RDL(x), is $504 million:

Separately, we credited Roche with developing 6–13 drugs from 2014–2021, depending on how its drug development is measured. For this example, we show calculations for specification #1, which only counts in-house drug discoveries. Given that Roche spent $86.3 billion in R&D and developed 6 drugs under specification #1, Roche’s R&D efficiency is $14.4 billion:

Repeating the sensitivity analysis for all four specifications described above, we determine that Roche spent $6.6–$14.4 billion per novel drug developed from 2014–2021; many multiples over the average industry cost to develop a new drug. Therefore, absent price increases on Perjeta, the loss of $504 million in R&D spending would result in 0.04–0.08 less novel drugs developed at Roche (specification #1 shown below):

The analysis has some limitations. The model used is static, meaning it does not account for changes in company behavior that may result under the counterfactual scenario tested. We also did not include measures of inflation, overestimating the magnitude of revenue saved and R&D spending. On the other hand, we did not include every drug that exhibited net price increases, underestimating company-wide revenue and R&D decrements. In addition, we did not distinguish between different therapeutic classes of drugs that may have different characteristics, leading to different development behaviors by firms. Finally, the analysis does not weight each company’s drug discoveries based on the size of each drug’s patient population, obscuring the magnitude of the effect of fewer drugs developed on public welfare. We plan to address many of these issues in future research.