The group of selected developers includes different types of organizations: nine publicly listed companies, two non-listed private companies, six research or academic centers, and one state-owned company. Partnerships are diverse as well, occurring between publicly listed companies (Pfizer and BioNTech), academic institutions and publicly listed companies (Oxford University and AstraZeneca) or non-listed companies (Bharat Biotech and the Indian Council of Medical Research). Vaccine developers are based in HICs such as the US or Europe as well as in MICs such as China, Cuba, Russia, or India.

The vaccines have a diverse range of platforms. Innovative technologies such as adenoviral vectors (Oxford/AstraZeneca, Janssen, Gamaleya, CanSino) and mRNA platforms (Pfizer/BioNTech, Moderna) coexisted with traditional platforms, such as protein subunit vaccines (Medigen, Finlay Institute, CGEB, Novavax), or whole inactivated viruses (Bharat Biotech, Sinovac, Sinopharm). This has implications for access because the logistic requirements for some vaccines are heavy, for example, mRNA vaccines usually require cold chain and ultra-cold chain logistics, compared to other vaccine platforms [79] (Table 2).

Out of the 14 vaccines selected for analysis, there is evidence of public investments for 12 of them, with CanSino and Sinovac the exceptions. However, there are notable differences between Western and non-Western developers in the type and amount of support received to expedite preclinical and early clinical development (Fig. 2). These differences increase when considering APAs, especially with Western Early Arrivers signing APAs with the US and the European Union. Notably, Pfizer did not receive public grants but its partner BioNTech did receive public grants and loans. The Pfizer/BioNTech vaccine also benefited from publicly-financed APAs, see Figs. 2 and 3 below. In contrast, Janssen and Novavax received funding from other sources, such as the APA between Janssen and the African Union or the agreement between Novavax and COVAX, the global platform tasked with purchasing and delivering COVID-19 vaccines globally. Oxford/AstraZeneca’s APA with CEPI/COVAX represents a large portion of the public support received.

R&D investments for the selected COVID-19 vaccine developers

Value of declared advance purchase agreements of selected COVID-19 vaccines

There is limited evidence about the role of the public sector in funding and facilitating the development of most non-Western COVID-19 vaccines. Both Russian vaccines, Sputnik V and Sputnik Light, were supported by the Russian Direct Investment Fund (RDIF) with at least USD 295 million, as well as by the Russian Ministry of Health (USD 399.5 million) [102, 103]. In addition, the Russian Ministry of Defense supported the development of preclinical and clinical trials [71]. There is little publicly-available information about the economic support provided by Chinese institutions, except for the approximately USD 142 million allocated by the state-owned company Sinopharm to the development of two inactivated COVID-19 vaccines [76]. Sinovac funded the development of its vaccine through an acquisition deal with Sino Biopharm, a Hong Kong based company that acquired 15.03% of Sinovac in December 2020; and through securing USD 15 million in investments from private investors.

Bharat Biotech’s Covaxin was developed in partnership with the Indian Council of Medical Research (ICMR, the main public agency in India for the coordination and promotion of biomedical research) [104] and received funding to expand manufacturing capacity from the Indian Government [105]. Medigen’s vaccine received in-kind support from the US NIH (biological samples to test its vaccine in animals), and received public subsidies from the Taiwanese government to run a Phase I clinical trial [106, 107]. The two Cuban vaccines, Abdala and Soberana 2, were developed in public research centers that are part of the state-owned conglomerate BioCubaFarma [108, 109], although there is no publicly available data on the costs of development and the resources invested by the Cuban government.

In summary, there is evidence of public funding for the development of all vaccines analyzed (including those developed within public institutions), except for Sinovac and CanSino.Footnote 1 The evidence available points towards a higher monetary level of support for Western developers, especially when considering the use of APAs, which fostered the late-stage development of several vaccines. Nevertheless, the lack of publicly available data prevents us from painting a comprehensive picture of the role of public and private funding in the development of all vaccines.

Regulatory approvals can provide some insight into the priorities of vaccine developers. That is, we can look at the countries and regions where developers first achieved regulatory approval, as well as the overall number of places where they filed for registration, to get a sense of their priority markets and the extent of their capacity and/or interest in providing global access. Admittedly, this is an imperfect indicator of a developer’s strategy, particularly if used alone, given that the timeline between the submission of a regulatory dossier and approval depends largely on the regulatory agency. Nevertheless, it remains a useful indicator, particularly given the urgency of the pandemic which led to the use of emergency use authorizations and rolling review processes that fast tracked approvals, likely facilitating developers to file for registration for their products globally [110].

An analysis of the regulatory approvals timeline shows a separation in two groups of vaccine developers, based on the speed to obtain regulatory approval globally (Fig. 4): “early arrivers” (Pfizer/BioNTech, Moderna, Sinovac, Sinopharm, Oxford/AstraZeneca, Sputnik V, Bharat Biotech, and CanSino), which obtained at least 25% of all their regulatory approvals before 228 days (or roughly 7.5 months, the median time of all developers together) and “latecomers” (Novavax, Soberana, Abdala, Medigen, Sputnik Light and Janssen), which obtained all of their approvals after 228 days. The exception is Janssen, which is situated closer to the early arrivers, with 25% of its approvals obtained in 233.5 days, but still above the median threshold of 228 days.

Distribution of regulatory approvals for each vaccine. The Y-axis represents days since the approval of the first vaccine globally (Sputnik V in Russia)

Oxford/AstraZeneca simultaneously had the largest number of global registrations and was the fastest to obtain all of its regulatory approvals, obtaining 75% of all its regulatory approvals in 224 days after the approval of the first vaccine (Fig. 5). Novavax, on the other hand, was the last in the group to achieve its first approval, which came 448 days after the approval of the first vaccine globally, Sputnik V in Russia.

Total number of vaccine approvals by innovation model and income level

Most vaccines we analyzed obtained WHO Emergency Use Listing (EUL), which was necessary to enter into purchase agreements with UNICEF or COVAX. Those that had not received EUL (to date) were Medigen, which submitted an expression of interest, Abdala and Sputnik V, which had been accepted for review, and Sputnik Light and Soberana 2, which had not submitted an expression of interest (Table 2). All Western developers (including Oxford/AstraZeneca) except Novavax obtained WHO EUL within the first four months of 2021. In contrast, Bharat Biotech, Sinopharm, Sinovac, and CanSino obtained EUL in the second half of 2021.

When regulatory approvals are sorted by country income levels, we observe a clear divide in market prioritization among the developers (Fig. 5). Whereas Western developers predominantly obtained approvals in HICs, Major Chinese and Russian Developers focused more on UMICs and LoMICs. However, this broad trend becomes a bit more nuanced when looking at individual companies. For example, out of the 20 countries that granted regulatory approval to Novavax, 80% were HICs, more than Janssen (33.90%, n = 59), Pfizer/BioNTech (38.27%, n = 81), and Moderna (44.90%, n = 49). Sinopharm (n = 70), Sputnik V (n = 75), and Sinovac (n = 42), obtained a higher proportion of approvals in LoMICs (47.14%, 44%, 40.48%) than UMICs (27.14%, 37.33%, 35.71%). Oxford/AstraZeneca (n = 155) had a more even distribution of approvals, with the highest number of approvals in LICs (n = 22, 14.19%), which were otherwise deprioritized by all other developers.

By examining data on the manufacturing networks formed by vaccine developers, we can explore the production needs and capacities of developers during the pandemic, and get a sense of longer-term strategies.

For instance, Western developers and Oxford/AstraZeneca prioritized manufacturing agreements with companies based in HICs (86.19% of their agreements on average). On the other hand, the Russian and Major Chinese Developers frequently focused their activities in LoMICs and UMICs (on average, 45.79% with LoMICs and 42.62% with UMICs). The rest of the developers signed fewer agreements (Fig. 6).

Number of manufacturing agreements by income level of the manufacturing partner’s country

Fill-and-finish agreements were the agreements most frequently announced (37.04% of all agreements). Western developers and Oxford/AstraZeneca also signed several agreements to produce drug substance (24.39% on average). Neither Sinopharm nor Sinovac reported agreements to produce drug substance or adjuvants, and instead rely on fill-and-finish agreements (68.18% of all their agreements).

Gamaleya built the largest network of manufacturers in the group to produce Sputnik V, with 39 different manufacturers, 56.41% of which intended to begin as fill-and-finish, and then progress to end-to-end production agreements and 30.77% categorized as end-to-end. Nevertheless, it is unclear how many of these agreements led to actual production, given the low volume of deliveries and the production and quality issues surrounding the development of Sputnik V [102]. In addition to Gamaleya, Bharat Biotech, Novavax, and both Cuban developers also signed at least 25% of all their agreements as end-to-end agreements, which could potentially signal a lack of in-house capacity to produce large volumes of their vaccines and/or a greater willingness to engage in technology transfer [64] (Fig. 7).

Type of manufacturing agreements

The number of purchased doses by countries can provide an idea of the supply capacity of the different developers and their priority markets. For example, although Sinopharm and Gamaleya were among the developers with the highest number of agreements, they committed fewer doses than Western developers and Oxford/AstraZeneca, potentially illustrating the differences in supply and production capacity of these two groups (Fig. 8).

Doses committed and purchase agreements. Left axis (bar chart) shows the number of doses committed. Right axis (X’s) shows the number of purchase agreements

Whereas Western Early Arrivers committed the largest share of their doses to HICs (66.84% Pfizer/BioNTech and 79.59% Moderna), Western Latecomers showed a different pattern, with Novavax committing 48.33% of its doses to COVAX (and 45.35% to HICs) and Janssen committing 29.95% of its doses to LoMICs through its APA with the African Union (and 47.46% to HICs).

Sinopharm and Sputnik V had a larger number of purchase agreements with LoMICs (59.68% and 59.64% respectively) over UMICs (15.50% and 36.17% respectively), whereas Sinovac seemed to follow a different strategy, with 36.67% of its doses dedicated to COVAX and 29.56% to UMICs. Oxford/AstraZeneca committed roughly half of its doses to LoMICs (52.14%), mostly to India, with the rest of the doses being roughly equally distributed between HICs (18.57%) and COVAX (17.57%), with UMICs receiving a smaller share (11.27%). Oxford/AstraZeneca accounted for the largest proportion of doses committed to LICs (18.5 million doses, 52.13% of all doses purchased by LICs), followed by Sinopharm (11.2 million doses, 31.54%) (Fig. 9).

Share of vaccines committed by income level of purchasing country

When looking at the time distribution of purchase agreements, Pfizer/BioNTech and Moderna sustained a high number of agreements consistently throughout the period of analysis, until 750 and 800 days, respectively, after their first agreement was signed (Fig. 10), showing the sustained demand and market dominance of Western Early Arriver vaccines over other vaccine developers.

Time distribution of vaccine agreements. Axis Y shows the number of days since the first purchase agreement signed by each vaccine developer

Oxford/AstraZeneca and the Major Chinese developers also have a wide distribution of agreements, although most of their agreements were signed within the first year following their first purchase agreement. Western Latecomers, Russian Developer, and the Small MIC Developers signed very few agreements after the first year following their first agreement.

Pfizer/BioNTech and Oxford/AstraZeneca have delivered more vaccine doses than other developers (Fig. 11), followed by Moderna, Sinovac and Sinopharm. However, the number of doses of Sinopharm and Sinovac are likely substantially undercounted, given that the vaccines delivered in China have not been reported publicly (see Limitations and areas for further research). Western Early Arrivers delivered most of their doses in HICs (61.40% Pfizer and 72.49% Moderna), compared with Oxford/AstraZeneca and the Major Chinese developers that prioritized LoMICs above other income groups (69.02% Oxford/AstraZeneca, 73.25% million Sinopharm, and 59.08% Sinovac). Western Latecomers (particularly Novavax) and Russian Developer delivered fewer doses than initially committed. Janssen is among the developers with more deliveries in LICs (208 million doses, 35.57% of all its deliveries) and LoMICs (207 million doses, 35.45%), mostly through COVAX (48.64% of its deliveries to these countries were through the global platform).

Doses by income group of the recipient country

The distribution of vaccine deliveries was staggered by income level, with HICs receiving more vaccines earlier than LoMICs and UMICs, and with LICs accessing very few doses much later than the rest.

Deliveries in HICs occurred in two stages (Fig. 12): the first one peaked early in 2021 and remained constant throughout the rest of the year. The second stage was in 2022, where deliveries slowed down, perhaps due to a situation of market saturation, before rising again as the autumn began in the Northern Hemisphere and guidelines in many HICs - notably in the US, which accounts for much of the trend - recommended booster doses.

Timeline of vaccine deliveries for each innovation model and income level of recipient country. Doses are shown relative to the income group’s population for comparability

Deliveries in UMICs follow a pattern driven by China, which experienced two large peaks - one between April and September 2021, and another between December 2021 and January 2022, followed by a steep decline in 2022. However, it is not clear which vaccines were distributed in China, nor is it clear whether the lack of data for 2022 reflects a lack of vaccine distribution, or a lack of data on vaccine distribution within the country (Fig. 12). The drop in deliveries in China between August and December 2021 seems to correspond to an increase in exported deliveries from Major Chinese developer vaccines to LoMICs.

Deliveries from the Major Chinese Developers comprised a large proportion of supply in UMICs, and even more so in LoMICs, until late 2021. In LoMICs, there is a single, smaller peak between November 2021 and February 2022, with a large proportion of these deliveries being in India. The country had a large portion of deliveries of Oxford/AstraZeneca (manufactured by India-based Serum Institute) between June and October 2021 that could be related to the export bans imposed by the country from March to September 2021, to extend vaccinations nationally.

The pace at which UMICs and HICs accessed enough doses to vaccinate each person with one dose seems to be similar (Fig. 13, left side). Nevertheless, this is skewed by the large number of deliveries in China, which accounts for a large portion of the deliveries in UMICs. When removing China and India, the remaining deliveries in UMICs and LoMICs do not reach one dose per person in the period of analysis (Fig. 13, right side). Deliveries in LICs do not reach one dose per person in the entire period of analysis, demonstrating that access inequalities greatly affected this group of countries over others.

Cumulative deliveries relative to the population of the income group in 2021 and 2022. The left side includes all countries. Right side excludes deliveries in China and India

When looking at the channel chosen for the acquisition and delivery of vaccines, doses purchased through bilateral and regional (deliveries through the African Vaccine Acquisition Trust, AVAT) agreements with developers represent 45.80% of all deliveries, when all “Unknown deliveries” are included, and began earlier than the deliveries from COVAX, which represent 11.90% of all deliveries. In Fig. 14, deliveries via COVAX peak at the end of 2021, and were dedicated to LICs and LoMICs. Despite being the focus of substantial discussion and debate, donations represent less than 3% of all the deliveries. As mentioned in the Limitations and areas for further research section, a large share of the deliveries (39.34%) has no data on the delivery channel.

Vaccine deliveries by type of delivery mechanism

Small MIC Developers mainly delivered doses in their respective countries and regions of origin. Although there is no data on deliveries in Cuba, the two Cuban vaccines were used in the country according to publicly available information [111] and was exported to other Latin American countries (Mexico, Nicaragua, and Venezuela). Bharat Biotech and Medigen delivered most of their doses domestically (97.38% to India and 95.99% to Taiwan, respectively). Regarding CanSino’s deliveries, the percentage of doses delivered in China is not publicly availbale. Nevertheless, for doses outside of China, there is evidence that the company focused its deliveries on Mexico (50.87%) and Pakistan (38.55% of all its deliveries).

Price information has not been systematically disclosed by developers or governments during the pandemic. Nevertheless, the data available shows that Oxford/AstraZeneca had the lowest average prices across the different income groups, ranging between 3.00 and 5.55 USD/dose. However, prices for HICs are lower than those for UMICs and LoMICs, although in some LoMICs such as India or Bangladesh, average prices are higher due to a different price for the private market. Similarly, Janssen, with an average price of around 10 USD/dose across income groups, also seemed to charge lower prices in HICs than in LoMICs and UMICs.

Moderna, Sinopharm, and Pfizer appear to have applied a tiered pricing approach, with clearly different prices for different income groups. The first two vaccines are also the two most expensive vaccines in HICs (40 and 36 USD/dose, respectively), and two of the most expensive vaccines in UMICs (28.88 and 19.98 USD/dose), although their LIC pricing is roughly on par with other developers selling to those markets. Pfizer’s tiered pricing is more of a binary split between pricing in HICs (20.85 USD/dose) and pricing in UMICs and LoMICs (11.2 and 10.58 USD/dose, respectively).

There are no price points in HICs for Sinovac and CanSino (Bharat Biotech did not deliver doses in HICs), and their listed prices for UMICs and LoMICs are higher than most of the other vaccine developers ranging from 15 to 27 USD/dose. Gamaleya, which committed to sell Sputnik V at less than 10 USD/dose [112], was reportedly selling it at almost 20 USD/dose in HICs and LoMICs, according to the data analyzed.

Despite orienting their operations towards supplying MICs, developers such as Sinopharm, CanSino, Gamaleya, Sinovac, and Bharat Biotech priced their products higher in these countries than other developers like Pfizer, Janssen, and Oxford/AstraZeneca.

Prices available for COVAX are on average lower than for all income groups, ranging from 3 USD/dose (Oxford/AstraZeneca and Novavax), to 10 USD/dose (Moderna). Nevertheless, in the publicly-disclosed contract between COVAX and the South African government, the maximum vaccine cost is listed to be 21.10USD/dose, which is more than the price reported by COVAX [113] (Fig. 15).

Average prices per dose per vaccine by income level