Equitable Access to
Vaccines: Myth or
Reality?
Shushanik Hakobyan, Henry Rawlings, and Jiaxiong Yao
WP/22/257
IMF Working Papers describe research in
progress by the author(s) and are published to
elicit comments and to encourage debate.
The views expressed in IMF Working Papers are
those of the author(s) and do not necessarily
represent the views of the IMF, its Executive Board,
or IMF management.
2022
DEC
* The authors would like to thank Aqib Aslam, Papa N’Diaye, Robert Zymek, Can Sever and seminar participants at the International
Monetary Fund for useful comments and suggestions.
© 2022 International Monetary Fund WP/22/257
IMF Working Paper
African Department
Equitable Access to Vaccines: Myth or Reality?
Prepared by Shushanik Hakobyan, Henry Rawlings, and Jiaxiong Yao
Authorized for distribution by Papa N’Diaye
December 2022
I
MF Working Papers describe research in progress by the author(s) and are published to elicit
comments and to encourage debate. The views expressed in IMF Working Papers are those of the
author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management.
ABSTRACT: Fighting the COVID-19 pandemic required vaccinations; however, ending it requires vaccination
equality. The progress in vaccinations varies greatly across countries, with low- and middle-income countries
having much lower vaccination rates than advanced countries. Initially, the limited vaccine supply was in part to
blame for slow pace of vaccinations in low-income countries. But as the supply constraints eased toward the
end of 2021, the focus has shifted to in-country distribution challenges and vaccine hesitancy. This paper
quantifies the importance of various factors in driving vaccination rates across countries, including vaccine
deliveries, demographic structure, health and transport infrastructure and development level. It then estimates
the contribution of these factors to vaccination inequality. We show that much of the vaccination inequality in
2021-22 was driven by the lack of access to vaccines which is beyond countries’ control. And although
vaccination inequality declined over time, access to vaccines remains the dominant driver of vaccination
inequality.
JEL Classification Numbers:
I1
Keywords:
COVID-19; vaccination; access; inequality
Author’s E-Mail Address:
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
3
WORKING PAPERS
Equitable access to vaccines:
Myth or reality?
Prepared by Shushanik Hakobyan, Henry Rawlings, and Jiaxiong Yao
1
1
“The authors would like to thank AFR Regional Studies Division staff and ICD seminar participants for their helpful comments and
suggestions. All remaining errors are ours.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
4
1. Introduction
Fighting the COVID-19 pandemic required vaccinations; however, ending it requires equal
distribution of vaccinations. More than two years after the discovery of the virus, the world
continues to grapple with new variants of the virus and renewed surge in cases. Scientific
consensus on the contagious and mutable nature of the coronavirus means that no one is safe
until everyone is safe, and the only way to stop the pandemic is to stamp out the virus
simultaneously worldwide.
Figure 1. Unequal Distribution of Vaccinations Around the World
(Population-weighted average)
Source: Our World in Data.
Notes: As of August 18, 2022; non-SSA EMDEs = non-sub-Saharan African emerging markets and developing economies; EM = emerging
markets 1/ excludes China due to lack of reporting before August 2021.
The progress in vaccinations, however, varies greatly across countries. Low- and middle-income
countries so far have much lower vaccination rates than advanced countries. Sub-Saharan Africa
(SSA) stands out as the least vaccinated region in the world, even in comparison to low- and
middle-income countries in other regions such as emerging Europe and Latin America (Figure 1).
As of mid-August 2022, only 19.5 percent of the region’s population was fully vaccinated,
compared to 61 percent in other emerging markets and developing economies (excluding China)
and 75 percent in advanced economies.
2
The unequal landscape of vaccinations around the world has raised concerns among policymakers
about equitable distribution. Among the multitude of factors that contribute to disparities in
COVID-19 vaccination coverage, two elements have garnered much attention: access to vaccines
and vaccination hesitancy. Vaccine supplier and recipient countries have disagreed on which is
the dominant factor behind low vaccination rates in developing countries. At the heart of the
2
In Our World in Data, China started reporting the vaccination numbers with a delay, in August 2021, when over 50 percent of its
population had been already fully vaccinated. Hence, China is excluded in the figures to smooth the series.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
5
debate is the question: who is more responsible for the unequal landscape of vaccinations and is
the distribution equitable? The answer has implications for policymakers worldwide.
This paper examines the drivers of COVID-19 vaccination rates across countries during 2021–22
and quantifies the equitable distribution of vaccinations. We use the principle of equality of
opportunity to measure vaccination inequality, whereby each country has equal access to vaccines
and differences in vaccination rates should only stem from factors that are under a country’s
control. Vaccination equality under the principle of equality of opportunity, therefore, does not
imply equal vaccination rates. Instead, it means that vaccination rates should not depend on
circumstances beyond country’s control. Distribution of vaccinations will be considered
inequitable if countries have unequal opportunities to vaccinate.
This paper finds that access to vaccines, as measured by the delivery of vaccine doses, was the
dominant reason behind vaccine inequality, accounting for more than 70 percent of the vaccine
inequality. Other socio-economic factors—per capita income, urbanization, spending on health,
demographic structure, and human development index—play a relatively minor role.
Ensuring timely production and delivery of vaccine doses therefore plays a crucial role in reducing
the unequitable distribution of vaccinations. Improving domestic distribution networks and
promoting proper vaccination campaigns are also key to drive up vaccination rates.
The remainder of the paper proceeds as follows. Section 2 reviews the related literature and
provides overview of vaccine logistics. Section 3 discusses the methodological approach and
describes the data. Section 4 presents the results, and Section 5 concludes.
2. Background
2.1. Related Literature
At the onset of COVID-19, researchers identified potential challenges to global vaccination and
frameworks to guide the vaccination efforts. The challenges focused on the factors impacting the
development, dissemination, and deployment of vaccines, as categorized by Forman (2021) and
Wouters et al (2021). While data were initially limited on deal making between governments and
vaccine manufacturers, studies drew on the experience during the H1N1 outbreak to warn of
vaccine market dominance by rich countries—a phenomenon later observed during the COVID-
19 pandemic (Fidler 2010, So and Woo 2020, Bollyky et al 2020, Deb et al 2021, Agarwal and Reed
2022).
Frameworks for more equitable distribution of vaccinations centered on the economic, health, and
ethical rationale. Leaving low and middle-income countries (LMICs) behind in vaccination efforts
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
6
would lead to large economic losses (Economist 2021, Deb et al 2021). The health benefits have
been outlined many times. To minimize harm, ethical frameworks posited the prioritization of
vulnerable groups first (Persad 2020, Emanuel et al 2020).
Countries started deploying vaccines at the end of 2020, which allowed the direct observation of
the factors affecting campaign efforts. With procurement and financing difficulties for many LMICs
well-documented at this point (Peacocke 2021), new areas of focus such as logistics and vaccine
hesitancy began to emerge. Countries with underdeveloped health infrastructure faced difficulties
in managing the stringent cold-chain requirements of mRNA vaccines (Cherif and Hakobyan 2021).
Common findings regarding hesitancy included concerns of safety, side-effects of the virus,
perceptions of the severity of the virus and demographic factors (men more likely to get the
vaccine) (Lazarus 2021, Sallam 2021, Dabla-Norris et al 2021). Despite the perception that demand
for vaccines in LMICs was low, and thereby justifying them receiving fewer vaccines, the hesitancy
data showed LMICs tending to have at least as high, if not higher rates of acceptance (Solís Arce
et al 2021).
Relatively few studies have empirically analyzed the drivers of COVID-19 vaccination rates globally.
Our work relates closest to that of Goel and Nelson (2021), Deb et al (2021), and Agarwal and
Reed (2022). Looking at subnational data in the United States, Goel and Nelson (2021) analyzed
socio-economic factors influencing vaccination rates across states. The authors separate vaccine
dissemination (vaccines administered) and efficiency of delivery (vaccines administered as a
percent of received). They find that economic prosperity and higher rural population lead to more
vaccines administered. Furthermore, the results show that vaccine efficiency improves with more
nursing homes per capita, more COVID-19 deaths, and more health workers. Deb et al (2021)
provide an empirical assessment of the determinants of vaccine rollouts in a cross-country setting.
They find early procurement, domestic production of vaccines, and health infrastructure to be
important for the pace of vaccination. While the results from our study underscore the actual
delivery of vaccines as the most critical component driving vaccination rates, we share in the
finding that infrastructure matters as well. Agarwal and Reed (2022) dig further into what drives
deliveries of vaccines. They find 60-75 percent of the delay in vaccine deliveries to low- and
middle-income countries is attributable to their signing purchase agreements later than high-
income countries, which placed them further behind in the delivery line. Their study is intrinsically
linked to this paper, given the latter’s focus on vaccine deliveries as a key determinant of the
observed vaccination rates.
This paper contributes to the literature on the drivers of COVID-19 vaccination rates and equitable
distribution of vaccinations. In particular, it builds on the relatively few studies of COVID-19
vaccination determinants in a cross-country setting by (i) examining the key drivers of the
vaccination rates, (ii) constructing a measure of equitable distribution of vaccinations—grounded
in the theory of equality of access (Roemer and Trannoy 2016); and (iii) quantifying the
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
7
contribution of different factors to the gap between the actual distribution and the equitable
distribution. Our decomposition of the factors contributing to vaccination inequality draws on
Shorrocks (2013) for the Shapely value decomposition method using Gini index as a measure of
overall inequality, supplemented by other inequality indices, such as MN-measure (Magdalou and
Nock 2011; Hufe et al 2021). By drawing on the experience of sub-Saharan Africa explicitly, we
shed light on the inequalities LMICs could face when caught up in a global pandemic.
2.2. Vaccine Logistics
Successful vaccination programs are built upon an efficient end-to-end supply chain and logistics
systems that ensure the uninterrupted availability of vaccines from manufacturing to delivery.
Analyzing COVID-19 vaccination inequality and its drivers requires untangling the many intricate
stages of the vaccination process (Figure 2). First, the recipient country finds vaccine suppliers and
signs purchase agreements with them. In case of some countries, particularly low- and middle-
income, there could be an intermediary such as the COVID-19 Vaccine Global Access (COVAX)
Facility,
a global risk-sharing mechanism, that facilitates the procurements and distribution of
vaccines to the recipient countries. Next, the suppliers produce the vaccines and schedule a
delivery—usually with a delay due to production disruptions (e.g., industrial accidents and
unexpected shortages in the supply chains such as packaging inputs) and export restrictions by
major vaccine producers.
3
Figure 2. Vaccine Logistics
Source: The World Health Organization Essential Programme on Immunization.
3
In early 2021, export controls on key ingredients into vaccine production, such as syringes and vials, hampered global vaccine
manufacturing by forcing suppliers to prioritize domestic contracts (for example, in the US and UK). In April 2021, following an
unprecedented surge in COVID-19 cases, India halted the exports of AstraZeneca vaccines produced by the Serum Institute of India,
the main supplier of vaccines to COVAX.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
8
Ahead of receiving the vaccine doses, countries need to develop distribution schemes, ready
health infrastructure, train health workers, and generate demand. Countries form vaccination
plans to prioritize who receives vaccines first (e.g., elderly, immunocompromised, or front-line
workers) based on demographic composition and the evolution of the virus. Some COVID-19
vaccines require cold-chain storage capability, further complicating the in-country distribution of
vaccines in countries with weak health systems. For countries with hard-to-reach populations (e.g.,
rural-remote areas, islands, conflict zones, refugee camps), officials devise plans to address these
challenges through creative means such as drones or flying doctors to remote areas. Doses move
from centralized arrival points to local health centers and into the arms of citizens. As vaccination
campaigns proceed, governments need to build trust, address misinformation, and provide
incentives to maintain the momentum for reaching vaccination goals.
3. Equitable Access to Vaccines
3.1. Delivery and Vaccination
In the vaccine logistics describe above, there are two key milestones: delivery of vaccine doses
and vaccine administration. While the delivery of vaccine doses is predicated mostly on
international circumstances, administering delivered vaccines depends primarily on domestic
factors. Indeed, beyond signing the advance purchase agreements early, recipient countries have
little control over the de facto delivery of vaccine doses, the responsibility for which almost entirely
lies with the suppliers.
4
For example, in April 2021, following an unprecedented surge in COVID-
19 cases, India imposed a ban on the exports of AstraZeneca vaccines produced by the Serum
Institute of India, the main supplier of vaccines to COVAX facility at that time which was the major
source of vaccines for low and middle-income countries.
5
Similarly, several disruptions hampered
global vaccine manufacturing in early months of vaccinations, for example, industrial accidents,
unexpected shortages in the supply chains such as packaging inputs, and export controls on key
ingredients and inputs into vaccine production such as syringes and vials which forced suppliers
to prioritize domestic contracts (e.g., in the US and UK). These events derailed vaccination
campaigns and further delayed the vaccinations in low and middle-income countries, in addition
to delays in signing advance purchase agreements.
The delivery of vaccine doses is therefore a moment when countries’ influence on vaccination
outcomes changes: the recipient countries typically have little influence on the timing of vaccine
4
Agarwal and Reed (2022) find that 60-75 percent of the delay in vaccine deliveries to low- and middle-income countries is
attributable to their signing purchase agreements later than high-income countries.
5
The ban was rescinded in October 2021.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
9
arrival, but once the vaccines are delivered, it is incumbent upon the recipient countries to get
them administered.
Throughout the paper, we focus on vaccination rates at the country level as the main variable of
interest. Ultimately, ending the COVID-19 pandemic requires high vaccination rates across all
countries. It is therefore important to understand the factors that affect vaccination rates.
Two measures of vaccination rates can be used to gauge the level of vaccination across countries.
The first measures the cumulative administered doses and corresponds directly to the cumulative
doses delivered. The second measure captures the percent of population with at least one dose
of vaccine. While the latter measure may be more meaningful from a health protection perspective,
the information on the type of vaccines administered is not publicly available for a large number
of countries, and therefore this measure does not correspond well to the type of vaccines
delivered. We use the first measure in this paper and robustness checks using the second measure
are included in the Annex III.
Specifically, we denote the cumulative administered doses per hundred people by
for country
. Let
denote country ’s cumulative delivered doses per hundred people, and
its average
administering rate at a certain date. We can write the following accounting identity for the
vaccinate rate
:
=
.
In logarithm, we have
log
= log
+ log
. (1)
While
is usually beyond country’s control,
is a key variable that measures the country’s efforts
to vaccinate its population, as well as vaccine hesitancy among population. It is clear though that
a high administering rate
is neither a sufficient nor a necessary condition for a high vaccination
rate
, because of the presence of delivery
in equation (1).
To analyze the drivers of vaccination rates, we initially abstract from administering rates and focus
on a general functional form where vaccination rates depend on delivered doses as well as a range
of fundamental factors:
log
= (log
,
), (2)
where
is a vector of economic, health, and demographic factors. We then analyze the
administering rates separately:
log
=
(
)
. (3)
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
10
Note that in addition to the channel through
in equation (3),
might affect
through
as
well.
3.2. Equitable Access
In this paper, we resort to the principle of equality of opportunity to measure equitable
distribution of vaccinations.
6
Under this principle, the distribution of vaccinations is equitable if
vaccination rates do not depend on circumstances beyond country’s control.
A country’s borders are a natural choice to distinguish between circumstances within and beyond
country’s control. Delivery of vaccine doses,
, originates outside of country’s borders, whereas
vaccine administering rates,
, depend on factors within country’s borders. As such, the
decomposition in equation (1) naturally distinguishes the two stages of the vaccination process
that are beyond and within a recipient country’s control.
For simplicity and tractability, and as discussed in Section 3.1, this paper assumes that vaccine
delivery is the only factor beyond the recipient countries’ control. Under this assumption,
equitable distribution of vaccinations implies that
should vary proportionally to
. In practice,
whether a factor is within a country’s control is less clear-cut for many reasons. For example,
securing vaccine deals and signing purchase agreements may directly affect vaccine delivery
(Agarwal and Reed 2022). Therefore, the delivery of vaccine doses may not be entirely beyond
country’s control. By the same token, domestic factors, such as demographic structure,
infrastructure, and income status, are circumstances inherited from previous governments and
hard to change in a short time span. Hence, the assumption that domestic factors are within
country’s control may also be too strong.
Consider countries = {1,2, , , }, with the observed vaccination rates =
(
,
, ,
)
. Let
=
(
,
, ,
)
be the vaccination rates under the equitable distribution of vaccinations.
When vaccination rates are equal across countries,
=
=
= . Such perfectly egalitarian
distribution of vaccination rates, however, might not be an equitable outcome, as country-specific
factors, such as vaccine hesitancy, affect the eventual vaccination rates. Imagine a hypothetical
scenario, where country A and country B have the same population, but 50 percent of country A’s
population remains vaccine-averse, whereas the entire population of country B is vaccine-
receptive. It is not equitable to send country A twice as many vaccines as country B just to reach
the same vaccination rates. Instead, the equitable distribution of vaccinations proposed here
follows the principle of equality of opportunity, whereby the distribution does not depend on
6
Equality of opportunity can be described as seeking to offset differences in outcomes attributable to luck, but not those differences
in outcomes for which individuals are responsible (Roemer and Trannoy 2016).
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
11
factors beyond a country’s control. Specifically,
=
(
,
, ,
)
does not depend on the
delivery of vaccine doses.
To compute the equitable distribution of vaccinations, a linear regression model of vaccinations
is employed:
l
og
=
+
+
, (4)
where
is a vector of domestic factors within country ’s control, in logarithm, including COVID-
19 cases and deaths, GDP per capita, infrastructure quality, urbanization, government spending
on health, demographic structure, and human development index.
is a vector of factors
beyond country ’s immediate control, also in logarithm. In this paper,
contains only one
variable, namely, the delivery of contracted or agreed vaccine doses.
is assumed to be
independent of
and
, and captures unobservables such as vaccine hesitancy.
The equitable distribution of vaccinations follows:
log
=
+ , (5)
where is a constant such that
=
. Intuitively,
=
(
,
, ,
)
is a redistribution of
vaccination rates based on factors within countries’ immediate control, holding the total
administered doses fixed.
3.3. Inequality decomposition
Having established a measure of equitable distribution of vaccinations, we quantify the
contribution of different factors to the gap between the actual distribution and the equitable
distribution. To this end, we follow Shorrocks (2013) and conduct a Shapely value decomposition,
which can decompose any inequality index in an additive manner. We choose the Gini index as a
measure of overall inequality.
The share of vaccine delivery in the overall vaccination inequality is the ratio of its contribution to
the overall Gini index. Since the delivery of vaccine doses is the only factor assumed to be beyond
country’s control, its contribution to the overall Gini index represents the inequitable distribution
of global vaccinations.
Annex I provides further details on the Shapley value decomposition. We show that the results of
the Shapley value decomposition are robust to using other inequality indices, such as a MN-
measure (Magdalou and Nock 2011), which is a divergence measure of inequality widely used in
the literature. The MN-measure attaches higher weights to shortfalls from the equitable
distribution (Hufe et al 2021) and encompasses mean log deviation—a prevalent measure in the
inequality literature (e.g., Ravallion 2019)—as a special case when the equitable distribution is the
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
12
equal distribution. Since most inequality indices require non-negative inputs, a log transformation
is sometimes needed.
3.4. Data
This paper uses data compiled from various sources as follows.
Vaccine delivery: Data on monthly deliveries of vaccine doses by vaccine type, covering the
period from November 2020 to March 2022, are obtained from Airfinity.
Vaccination rates, COVID-19 cases and deaths: Data on daily vaccination rates, cases and
deaths come from Our World in Data. For many countries in the sample, vaccination rates are
reported infrequently, hence, we use the highest vaccination rate for each month in each country
as the cumulative vaccination rate.
Vaccination uptake: Data on average vaccination coverage come from Airfinity (sourced from
the World Health Organization and UNICEF), and include vaccination against polio, diphtheria
tetanus toxoid and pertussis, hepatitis B, measles, and Hib disease.
Median travel time: Nelson et al (2019) provide estimates of travel time from any location on the
Earth’s surface to the nearest settlement at the pixel level with a 30 arc-second resolution. We use
the travel time to nearest city that has a population between 0.5 and 1 million to compute the
median of all pixel values of a country. This measure is used as a proxy for the difficulty in getting
vaccines to remote populations within a country.
Economic, health, and demographic indicators: We obtain GDP per capita, quality of overall
infrastructure, urban share of population, spending on health, the share of 15+ population aged
15-64, the share of population aged 65 and above from the World Bank World Development
Indicators. Human development index is obtained from the United Nations Development
Programme.
Table 1 shows the summary statistics for the full sample, and the sample of sub-Saharan African
countries, non-SSA EMDEs, and advanced economies. Not surprisingly, vaccination rates and
cumulative delivery (per 100 people) are lowest in SSA and highest in advanced economies.
Among the three country groups, SSA countries have the lowest number of COVID-19 cases and
deaths (per million people), lowest levels of GDP per capita, health expenditure per capita,
infrastructure quality score, human development index and lowest share of urban population. SSA
also boasts the smallest share of 15+ population at 61 percent, compared to 84 percent in
advanced economies and 75 percent in other EMDEs.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
13
Table 1. Summary Statistics, as of 2022Q1
4. Results
4.1. Factors affecting vaccination rates
Vaccine delivery and vaccination rates: are SSA countries different?
As shown in Figure 1, sub-Saharan Africa is one of the least vaccinated regions in the world. To
examine the extent to which the delivery of vaccine doses affects vaccination rates and whether
sub-Saharan Africa is different in this regard, Figure 3 plots cumulative vaccination rates against
cumulative delivery rates by country at the end of 2022Q1, with red circles identifying sub-Saharan
African countries. Cumulative delivery and vaccination rates are highly correlated, with most
observations close to 45-degree line. On average, about 76 percent of delivered vaccine doses
have been administered worldwide. Delivery rates in SSA countries are markedly lower than those
in the rest of the world, reflecting the difficulty for SSA countries to obtain vaccines.
However, there is no strong evidence that SSA countries have been administering the delivered
vaccine doses at a different speed from the rest of the world, particularly since the second half of
2021. To show this formally, we regress the vaccination rates on cumulative delivery rates,
controlling for SSA countries and including an interaction term between SSA dummy and
cumulative delivery rates. We repeat this for several points in time: end-2021Q2, 2021Q3, 2021Q4
and 2022Q1 (Table 2). The interaction term between the SSA dummy and cumulative delivery rate
is negative albeit statistically insignificant in 2021Q2 and 2021Q3 (columns 1 and 2), but turns
positive, remaining statistically insignificant, for the subsequent points in time (columns 3 and 4),
suggesting that SSA countries, despite a slow start, were administering delivered vaccines at a
similar—if not faster—speed compared to the rest of the world.
Mean St Dev Mean St Dev Mean St Dev Mean St Dev
Vaccination rate (doses per 100 people) 115 77 46 54 141 62 212 25
Cumulative delivery (doses per 100 people) 148 88 68 62 177 68 264 39
Cumulative cases (per million people) 93,062 110,286 35,566 80,304 104,548 98,778 254,152 136,507
Cumulative deaths (per million people) 972 1209 347 528 1,327 1,369 850 889
GDP per capita (current 2021 USD) 15,899 20,022 4,832 6,098 18,136 20,574 46,694 18,899
Median travel time (hours) 12 24 14 18 8.0 6.4 32 79
Infrastructure quality score 3.7 1.0 3.2 0.8 3.8 0.9 5.7 0.8
Urban share of population (percent) 58 23 42 16 64 21 87 9
Health expenditure per capita (current 2021 USD) 1,097 1,357 328 425 1,060 788 4,799 1,755
Share of population age 15-64 64 6.9 57 5.2 67 5.1 66 6
Share of population age 65+ 7.2 5.1 3.6 2.0 7.8 4.2 18 5
Human development index 0.7 0.1 0.6 0.1 0.7 0.1 0.9 0.0
Number of observations 95 31 57 7
Non-SSA EMDEs
Advanced
Sources: Airfinity, Our World in Data, Nelson et al. (2019), World Bank World Development Indicators, UNDP, and IMF staff calculations.
Full sample
Sub-Saharan Africa
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
14
Figure 3. Vaccine Delivery and Vaccination Rates, as of 2022Q1
In addition to the number of vaccine doses delivered, the type of vaccine doses and their source
(COVAX, direct purchases from manufacturers, bilateral donations) may also matter for
vaccination rates. In case of the type of vaccine, concerns about safety, side effects, and
effectiveness of different vaccines, especially at the beginning of the vaccination campaigns, have
been widespread and even observed among health care workers.
7
mRNA-based vaccines tend to
give a strong boost to vaccination rates, whereas other types either reduce vaccination rates or
do not have a statistically significant impact on vaccination rates (Annex II). In terms of vaccine
sources, LMICs relied heavily on deliveries from COVAX facility, having limited funds to sign
advance purchase agreements. In particular, in SSA countries only vaccine deliveries from COVAX
have positive and statistically significant impact on vaccination rates, whereas direct purchases
from manufacturers and to a lesser extent bilateral donations are associated with higher
vaccination rates in the rest of the world (Annex II). This suggests that low delivery rates in SSA
reflect a combination of limited funds and insufficient donations.
7
The suspension of AstraZeneca’s rollout in some European countries, the South African data on its effectiveness and the temporary
suspension of the Johnson & Johnson vaccine in the US to evaluate reports of blood clotting, affected confidence in COVID-19
vaccines. Ultimately, AstraZeneca’s vaccine was refused by several countries.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
15
Table 2. Vaccination Rates and Cumulative Delivery of Vaccines
Drivers of vaccination rates
As shown in the previous analysis, delivery of vaccine doses is an important factor that drives
vaccination rates. Once vaccines are delivered to the country, the rate at which they are
administered depends on a multitude of factors such as the health care system, transportation
network, demographic structure, as well as development status.
In the next set of regressions, using the vaccination and cumulative delivery rates as of 2022Q1,
we sequentially add regressors controlling for a range of domestic factors that affect country’s
ability to administer vaccine doses (Table 3). Columns 2-8 show that each of these factors
individually has a statistically significant impact on vaccination rates: higher vaccination rates are
associated with higher COVID-19 cases, higher COVID-19 deaths, higher GDP per capita, better
infrastructure quality, higher health spending, older population, and higher human development
status.
8
The coefficient before each variable can be interpreted as an elasticity. For example,
column 1 implies that when cumulative delivery increases by 10 percent, vaccination rates would
increase by 12.8 percent.
9
The last row calculates that if each individual factor increases from the
25
th
percentile to the 75
th
percentile in the sample, vaccination rates would increase by about 100
percent.
8
Median travel time, infrastructure quality score and urban share of population are proxies for the ease of in-country delivery of
vaccines, and hence controlled for in the same regression. Similarly, the demographic structure is captured by two variables—share
of 15-64 and 65+ population, which are included in the same regression.
9
Note that the elasticity can be above 1. For instance, from a 20 percent vaccination rate of 100 doses delivered to a 40 percent
vaccination rate of 200 doses delivered, the implied elasticity is 2/1.5 = 1.33.
2021Q2 2021Q3 2021Q4 2022Q1
(1) (2) (3) (4)
Log cumulative delivery (doses per 100 people) 1.16*** 1.18*** 1.13*** 1.16***
(18.89) (14.13) (21.93) (20.56)
SSA dummy 0.13 0.01 -1.01 -1.23
(0.39) (0.02) (-1.48) (-1.40)
Log cumulative delivery x SSA dummy -0.12 -0.01 0.24 0.28
(-1.19) (-0.12) (1.42) (1.36)
Observations 146 149 150 150
R-squared 0.85 0.91 0.81 0.88
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Sources: Airfinity, Our World in Data, and IMF staff calculations.
Log vaccination rate (doses per 100 people), as of
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
16
When these factors are controlled for in a single regression, only cumulative delivery and quality
of infrastructure matter statistically significantly for vaccination rates (column 9). This remains the
case even after we drop some covariates that are highly correlated with others (column 10);
cumulative deaths, median travel time, and human development index are highly correlated with
cumulative cases, infrastructure score, and GDP per capita, respectively. If we think of the quality
of infrastructure as a measure of the ease with which vaccines can be distributed within the
country, the two statistically significant factors highlight the importance of vaccine logistics—be
it international or domestic—in getting people vaccinated.
Table 3. Factors Affecting Vaccination Rates
4.2. Factors affecting administering rates
In this section, we focus on equation (3), exploring factors affecting administering rates
. In
addition to the fundamental drivers considered in Section 4.1, one potential factor that affects
administering rates is vaccine hesitancy. If a country has a historically low vaccination uptake, its
population may not be willing to get COVID-19 vaccines, leading to lower administering rates.
Furthermore, if a country has already vaccinated a large proportion of its population, the
administering rates could be low as well.
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Log cumulative delivery (doses per 100 people) 1.28*** 1.38*** 1.35***
(15.67) (9.19) (7.91)
Log cumulative cases (per 1000 people) 0.45*** -0.17 -0.01
(7.79) (-1.29) (-0.38)
Log cumulative deaths (per 1000 people) 0.39*** 0.16
(5.14) (1.39)
Log GDP per capita 0.74*** -0.03 -0.13
(8.39) (-0.28) (-1.57)
Log median travel time 0.03 0.03
(0.31) (0.67)
Log infrastructure quality score 2.41*** 0.60** 0.42**
(5.05) (2.53) (2.62)
Log urban share of population 0.78** 0.26 0.31
(2.09) (1.15) (1.00)
Log health expenditure per capita 0.69*** -0.01 -0.08
(8.92) (-0.04) (-0.70)
Log share of 15-64 population 6.37*** 0.70 0.78
(8.52) (1.07) (1.31)
Log share of 65+ population 0.48*** 0.04 0.01
(5.88) (0.52) (0.18)
Log human development index 4.35*** -0.90
(9.95) (-1.19)
Observations 150 149 148 150 98 145 150 150 95 104
R-squared 0.88 0.45 0.31 0.47 0.449 0.48 0.50 0.57 0.92 0.90
144% 137% 107% 119% 92% 128% 97% 137%
Sources: Airfinity, Our World in Data, Nelson et al. (2019), World Bank World Development Indicators, UNDP, and IMF staff calculations.
Log vaccination rate (doses per 100 people), as of 2022Q1
Changes in vaccination rates if an individual factor increases from 25th percentile to 75th percentile
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1. To calculate the changes in vaccinations rates for individual factors, column (5) only uses infrastructure
quality score and column (7) only uses the share of population age 15-64.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
17
Drivers of administering rates
Panel (a) of Figure 4 shows that countries that have had high vaccination uptake in the past tend
to have higher average administering rates for COVID-19 vaccine. However, there is a great
heterogeneity across countries, with some displaying a high COVID-19 vaccine administering rate
despite low historic vaccination uptake.
Panel (b) compares the marginal administering rates (
/
) over 2022Q1 with vaccination rates
as of 2021Q4 and shows that there is no correlation. In other words, even if a country has a large
share of its population vaccinated by end-2021, it does not imply the marginal administering rates
would be low over the next quarter. Admittedly, vaccine delivery data are not always updated in
a timely fashion, resulting in marginal vaccination rates over 1 in some countries, despite
aggregating the delivery and vaccination data to quarterly frequency. Dropping the observations
with administering rates greater than 1 does not qualitatively alter our results.
Figure 4. Administering Rates vs. Historical vaccine uptake and GDP per capita
Table 4 reports the results from regressions where the dependent variable is the ratio of
administered doses over delivered doses as of 2022Q1, and the economic, health, and
demographic factors are added sequentially as regressors. Columns 1-8 show that higher
historical vaccination uptake, higher COVID-19 cases and deaths, higher GDP per capita, better
infrastructure quality, more per capita spending on health, older population, and higher human
development index are associated with higher administering rates. If each individual factor
increases from the 25
th
percentile to the 75
th
percentile in the sample, the administering rate
increases by about 20 percent. When these factors are controlled for in a single regression, all
coefficients are imprecisely estimated, in contrast to the results in Table 4 where vaccine logistics
0 .2 .4 .6 .8 1
Average administering rate
20
40 60
80 100
Historical vaccination uptake
Data as of 2022Q1
Linear fit
0
5
10 15
Marginal administering rate (2022Q1)
0
50 100
150 200 250
Vaccination rate (2021Q4)
Data
Linear fit
Note: Administering rate is the administered share of delivered doses.
Sources: Airfinity, Our World in Data, and IMF staff calculations.
Panel (a) Panel (b)
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
18
is a significant determinant of vaccination rates. The results are qualitatively unchanged and
robust to altering the date of administering rates, namely 2021Q3 and 2021Q4.
Table 4. Factors Affecting Administering Rates
Excess doses of vaccines
From a country’s perspective, after vaccine doses are delivered, increasing administering rates is
key to raising vaccination rates. However, from a global perspective, the differences in the
distribution of delivered doses can be far more important than the differences in administering
rates, because sharing unused vaccines of one country with another can substantially increase the
vaccination rate of the recipient country at the same administering rate.
To examine the degree of unused vaccines across countries, one possible measure is the
difference between the cumulative number of delivered doses and the cumulative number of
administered doses, calculated as
(1
). However, this measure does not necessarily capture
the excess doses at any point in time for two reasons. First, poor and infrequent data reporting by
many countries does not allow to accurately capture the timing of vaccinations, hence the
computed excess doses might be overstated due to lack of vaccination data. Second, some
countries might have excess demand for vaccines, but it may take them some time to get the
2021Q4 2021Q3
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Log vaccination uptake 0.693*** 0.342 0.116 0.293
(5.07) (1.01) (0.21) (0.60)
Log cumulative cases (per 1000 people) 0.061*** 0.052 -0.225 0.049
(3.93) (1.22) (-1.15) (0.73)
Log cumulative deaths (per 1000 people) 0.058*** -0.026 0.317 0.019
(3.36) (-0.78) (1.58) (0.40)
Log GDP per capita 0.116*** -0.037 0.203 0.301*
(5.06) (-0.40) (1.52) (1.91)
Log median travel time -0.028 -0.005 0.074 -0.018
(-1.16) (-0.21) (1.21) (-0.49)
Log infrastructure quality score 0.459*** 0.208 0.588* 0.114
(3.24) (1.37) (1.79) (0.51)
Log urban share of population 0.058 0.003 0.033 -0.303**
(0.91) (0.04) (0.19) (-2.22)
Log health expenditure per capita 0.116*** 0.041 0.104 0.056
(4.51) (0.58) (0.62) (0.48)
Log share of 15-64 population 0.989*** 0.759 1.211* 0.422
(4.55) (1.16) (1.85) (0.51)
Log share of 65+ population 0.090*** 0.061 0.041 0.073
(2.71) (0.90) (0.48) (0.90)
Log human development index 0.715*** -0.470 -1.593 -1.300
(5.34) (-0.91) (-1.20) (-1.31)
Observations 135 145 143 143 91 138 144 143 84 85 89
R-squared 0.153 0.100 0.068 0.131 0.249 0.148 0.152 0.165 0.311 0.505 0.398
13% 21% 18% 19% 18% 22% 15% 23%
Sources: Airfinity, Our World in Data, Nelson et al. (2019), World Bank World Development Indicators, UNDP, and IMF staff calculations.
Changes in administering rates if an individual factor increases from 25th percentile to 75th percentile
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1. To calculate the changes in vaccinations rates for individual factors, column (5) only uses infrastructure quality score
and column (7) only uses the share of population age 15-64.
Administering rate (administered doses/delivered doses), as of
2022Q1
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
19
shots into arms after the vaccines have been delivered, which would also overstate the number of
excess doses.
With these caveats in mind, Panel (a) of Figure 5 uses the United States as an example for such
calculation. It shows that since April 2021, the excess doses of vaccines have been rising steadily,
reaching almost 200 per 100 people in April 2022. Panel (b) plots excess doses of vaccines for all
countries in our sample. It shows that advanced economies, such as the United States and United
Kingdom, have accumulated far more excess doses than developing countries, despite vaccine
hesitancy being often cited as the reason for low vaccination rates in developing countries.
Figure 5. Excess Vaccine Doses for Selected Countries
4.3. Equitable vaccination rates
Equitable distribution of vaccinations
Figure 6 compares the actual vaccination rates (as of 2022Q1) with the estimated equitable
distribution of vaccinations. Circles above the 45-degree line imply that the equitable vaccination
rates are higher than actual vaccination rates.
First, it is remarkable to notice that the equitable distribution of vaccinations is close to—but not
the same as—equal distribution of vaccinations. This highlights the importance of vaccine delivery
in driving the overall inequality in vaccination rates: if vaccination rates only depend on factors
under a country’s control, they would have been close to equal across countries.
Second, the equitable distribution of vaccinations has most countries at 100 doses per 100 people,
reflecting the vaccine supply capacity of the world in early 2022. Most SSA countries, however,
would have had higher vaccination rates under the equitable distribution.
Sources: Airfinity and IMF staff calculations.
Panel (a) Panel (b)
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
20
Figure 6. Actual and Equitable Distribution of Vaccinations, as of 2022Q1
4.4. Evolution of vaccination inequality
Global vaccination inequality, as measured by the Gini index of vaccination rates across countries,
has been on the decline over time (Figure 7), thanks in part to increased vaccine production.
However, vaccination inequality remained as high as 0.4 even at the end of 2022Q1.
Figure 7. Evolution of Global Vaccination Inequality
Using the Shapley value decomposition, Figure 8 examines the contribution of each factor in
column 10 of Table 3 to global vaccination inequality. Vaccine delivery is clearly the dominant
Sources: Airfinity and IMF staff calculations.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
21
factor that drives global vaccination inequality. Its importance remained the same throughout
2021Q2-2022Q1, accounting for between 70 and 80 percent of global vaccination inequality.
By contrast, other health, economic and demographic factors, such as COVID-19 cases, GDP per
capita, urban share of population, and the share of 15+ population, play a minor role in global
vaccination inequality. Vaccine hesitancy, while not directly observable, is captured in the residual
term in Figure 8. The contribution of the residual term increased slightly over time in 2021, peaking
at only 13 percent of global vaccination inequality. This suggests that vaccine hesitancy is not
nearly as important as vaccine delivery for vaccination rates.
Figure 8. Decomposition of Vaccination Inequality as Measured by Gini Coefficient
5. Conclusion and policy Implications
COVID-19 vaccination rates vary greatly across countries, with low- and middle-income countries
having much lower vaccination rates than advanced countries. This paper provides an empirical
assessment of the importance of various factors in driving vaccination rates across countries,
distinguishing between factors within a country’s control (demographic structure, health and
transport infrastructure and development level) and beyond (vaccine delivery). We show that even
though the delivery of vaccine doses to sub-Saharan African countries lagged behind the rest of
the world, the rates at which these countries administered vaccines is comparable to those of
other countries. Moreover, when we control for factors within countries’ control such as the health
care system, transportation network, demographic structure, and development status, delivery of
vaccine doses and quality of infrastructure appear to matter the most for vaccination rates, which
highlights the importance of vaccine logistics—whether international or domestic—in getting jabs
into people’s arms. Finally, when we quantify the contribution of each factor to the overall
-10% 0% 10% 20% 30% 40% 50% 60% 70% 80%
Cumulative delivery
Cumulative cases
GDP per capita
Infrastructure quality score
Urban share of population
Health expenditure per capita
Share of population age 15-64
Share of population age 65+
Residual
2021Q2
2021Q3
2021Q4
2022Q1
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
22
vaccination inequality, delivery of vaccine doses accounts for about 80 percent, while vaccine
hesitancy at most for only about 13 percent of global vaccination inequality.
As we transition to fighting COVID-19 over the longer term and in preparation of future
pandemics, it is pertinent for international community to acknowledge that strengthening global
collaboration and systems for a faster and better response is critical to reaching the global
vaccination targets. As outlined in Agarwal et al (2022), stronger multilateral institutions with high-
level political support to coordinate the global response to health shocks; rapid and adequate
financing windows for global public goods activated at the onset of pandemics; diversified
regional manufacturing capacity with arrangements to share technology and know-how;
improved regulatory frameworks to allow speedy delivery of existing and novel tools worldwide
are key areas where improvements could lead to lower vaccination inequality. In their turn,
countries will need to invest, and the international community should allocate additional resources
to strengthen health systems to prevent, detect and respond to future threats. Only collectively
can we fight future pandemics better.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
23
Annex I. Measures of Vaccination Inequality
Shapley decomposition of the Gini index
The Shapley decomposition calculates the marginal decline in the Gini index once contributing
factors are removed sequentially.
10
Since the marginal decline depends on the specific elimination
path, the average marginal decline over all possible elimination paths is used as the contribution
of the factor.
F
or example, consider the contribution of delivery of vaccine doses to the overall vaccination
inequality. Suppose there are factors in total that drive the overall inequality. One specific
elimination path is to remove vaccine delivery while keeping all other
(
1
)
factors. Let
denote the set of all other ( 1 ) factors and
vaccine delivery,
{
}
the marginal
contribution of the delivery of vaccine doses. It follows that the contribution in this specific
elimination path is:
(
vaccine delivery,
)
=
(va
ccine delivery,
{
}
) (
{
}
).
However, one could consider eliminating vaccine delivery from different subsets of all other
factors. Each subset has factors with = 0, , 1. In other words,
|
|
= . The average
marginal contribution of vaccine delivery is then:
(
vaccine delivery
)
=
(
)
!!
!
|
|
(
vaccine delivery,
)
.
The main analysis in this paper uses the Gini index as the baseline measure of inequality. In
addition to the Gini index, we also consider an alternative measure of global vaccination inequality,
which shows a similar result: global vaccination inequality declined over time and vaccine delivery
remained the dominant driver of vaccine inequality.
MN-Measure
The MN-measure (Magdalou and Nock 2011) is a divergence measure of inequality widely used
in the literature. It attaches higher weights to shortfalls from the equitable distribution (Hufe et al
2021) and therefore complements the Gini index when vaccination rates are highly uneven across
countries.
10
Removal here means that the differences of a factor are removed—the factor is assumed to take the mean value of different
countries.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
24
To assess the difference between the observed distribution of vaccinations and the equitable
distribution, a scalar inequality measure that aggregates the divergence between the two
distributions is used:
(|
|
)
=
. (1.1)
This is a decomposable divergence measure of distributions that has two desirable properties for
the purpose of analyzing vaccine inequality. First, a progressive transfer of vaccines from a country
with a high vaccination rate to a country with a low vaccination rate reduces the inequality. Second,
the measure allows the quantification of the contribution of different factors to the overall
inequality.
The overall MN-measure of global vaccine inequality is given by (||), where is the average
vaccination rate across countries. It encompasses both the equitable and unequitable components
of inequality. To compute the equitable distribution of vaccinations, a linear regression model of
vaccinations is employed similar to equation (1.1),
log
=
+
+
. (1.2)
However, different from equation (1.1), the left-hand side of equation (1.2) is the logarithm of
vaccination rates. This is to ensure that
in the MN-measure is positive and well defined in
subsequent calculations. All variables on the right-hand side of equation (1.2) are also in logarithm
so that the coefficients can be interpreted as elasticities.
Following the formulation in Almås et al (2011), the equitable distribution of vaccinations is
obtained by assuming that it does not depend on factors beyond country’s control and takes the
following form:
=
exp
exp
.
As before, a S
hapley value decomposition is conducted to quantify the contribution of each factor
to the MN-measure of global vaccine inequality.
The left panel of Figure 8 compares the evolution of global vaccine inequality as measured by the
Gini index and the MN-measure. Both show that vaccination inequality declined over the course
of 2021 and into 2022. The right panel, however, shows that the inequitable component of global
vaccine inequality—or the component driven by vaccine delivery—remained persistently high
under the MN-measure, accounting for about 90 percent of the overall global vaccination
inequality.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
25
Both measures of inequality therefore depict the same landscape of global vaccine inequality: it
is declining over time but remains high, and vaccine delivery continues to be the dominant driver
of such inequality.
Figure 1.1. Alternative Measures of Global Vaccination Inequality and Its Inequitable Component
Sources: Airfinity and IMF staff calculations.
0%
20%
40%
60%
80%
100%
120%
2021Q2 2021Q3 2021Q4 2022Q1
Inequitable Share of Vaccination Inequality
MN divergence Gini
0.0
0.2
0.4
0.6
0.8
1.0
1.2
2021Q2 2021Q3 2021Q4 2022Q1
Alternative Measures of Vaccination Inequality
MN divergence Gini
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
26
Annex II. Vaccine Types and Sources
The effectiveness of various types of vaccines has attracted much media coverage, which in turn
may have affected people’s willingness to get vaccinated. The effectiveness of vaccines is also a
factor that could shape vaccine hesitancy. To gauge the importance of vaccine type for vaccination
rates, we regress vaccination rates on cumulative delivery and shares of various vaccines delivered
as of 2022Q1 (Table 2.1). We group COVID-19 vaccines into four major types—mRNA (Pfizer and
Moderna), AstraZeneca, Chinese vaccines, and other types of vaccines, and calculate the share of
each type in total doses delivered to the country.
Column 1 shows that higher share of mRNA vaccines is associated with higher vaccination rates.
This positive association increases when the United States, as the major producer of mRNA
vaccines, is excluded in column 2. By contrast, columns 3 and 5 show that higher shares of
AstraZeneca and Chinese vaccines are associated with lower vaccination rates. Compared to
columns 3 and 5, excluding vaccine producer countries in columns 4 and 6 reduces the magnitude
of the coefficient, suggesting that AstraZeneca and the Chinese vaccines are used more often in
producer countries than abroad. However, none of the vaccine types affect vaccination rates in a
statistically significant way.
Table 2.1. Vaccination Rates and Type of Vaccines
Taking a step further, we examine if the results in Table 2.1 hold for SSA countries, the region with
the lowest vaccination rates, by including an SSA dummy and interaction terms of the SSA dummy
and shares of various types of vaccines. Table 2.2 shows that vaccination rates in SSA countries
depend on the type of delivered vaccines in a similar way to those in other countries. The
All
countries
excl. US
All
countries
excl. UK
and India
All
countries
excl. China
All
countries
All
countries
(1) (2) (3) (4) (5) (6) (7) (8)
Log cumulative delivery (doses per 100 people) 1.261*** 1.265*** 1.280*** 1.282*** 1.280*** 1.279*** 1.275*** 1.269***
(14.73) (14.76) (15.94) (15.84) (15.46) (15.49) (15.54) (11.78)
Share of mRNA 0.153 0.186 0.107
(1.20) (1.48) (0.50)
Share of AstraZeneca -0.059 -0.085 -0.058
(-0.36) (-0.50) (-0.28)
Share of Chinese vaccines -0.104 -0.117 -0.065
(-0.54) (-0.58) (-0.23)
Share of other types of vaccines 0.188
(1.54)
Observations 150 149 150 148 150 149 150 150
R-squared 0.877 0.878 0.876 0.875 0.876 0.875 0.876 0.877
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Sources: Airfinity, Our World in Data, and IMF staff calculations.
Log vaccination rate (doses per 100 people), as of 2022Q1
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
27
coefficient on the interaction term of the SSA dummy and the share of mRNA vaccine is statistically
significant, indicating that mRNA vaccine might boost vaccination rates in SSA.
Table 2.2. Vaccination Rates and Type of Vaccines in Sub-Saharan Africa
The source of vaccine deliveries may also matter for vaccination rates. As shown in Figure 2.1, SSA
countries predominantly received their vaccine doses from COVAX, reflecting the fact that they
did not sign advance purchase agreements with vaccine manufacturers due to financing
constraints. In contrast, advanced economies relied only on direct purchases from manufacturers.
Even non-SSA emerging market and developing economies sourced most of their vaccine doses
directly from manufacturers.
More formally, we regress the vaccination rates on cumulative delivery rates by source of vaccines
(COVAX, bilateral donations and direct purchases from manufacturers),
11
controlling for SSA
countries and including an interaction term between SSA dummy and cumulative delivery rates
by source (Table 2.3). The interaction term between the SSA dummy and COVAX cumulative
11
Deliveries from African Union’s African Vaccine Acquisition Trust (AVAT) are counted toward COVAX. Since many countries have
only deliveries from one or two sources, to maintain the sample size, we add 1 dose before taking logarithm of the variables for
vaccine deliveries.
(1) (2) (3) (4) (5)
Log cumulative delivery (doses per 100 people) 1.256*** 1.257*** 1.276*** 1.279*** 1.247***
(15.29) (13.95) (14.06) (14.02) (12.58)
SSA -0.228 -0.197
(-1.36) (-0.85)
Share of mRNA -0.218 -0.082 0.081 0.014 -0.205
(-1.65) (-0.50) (0.51) (0.17) (-1.01)
SSA × Share of mRNA 1.018* 0.870*
(1.84) (1.77)
Share of AstraZeneca 0.117 0.081
(0.85) (0.38)
SSA × Share of AstraZeneca 0.319 0.244
(0.50) (0.45)
Share of Chinese vaccines 0.011 -0.001
(0.06) (-0.01)
SSA × Share of Chinese vaccines -0.410 -0.142
(-0.77) (-0.35)
Share of other types of vaccines 0.196
(1.55)
SSA × Share of other types -2.214
(-0.58)
Observations 150 150 150 150 150
R-squared 0.879 0.877 0.878 0.876 0.880
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Sources: Airfinity, Our World in Data, and IMF staff calculations.
Log vaccination rate (doses per 100 people), as of 2022Q1
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
28
delivery rate is positive and statistically significant, whereas the interaction terms between the SSA
dummy and deliveries from the other two sources of vaccines is statistically insignificant,
suggesting that that low delivery rates in SSA countries reflect a combination of limited funds and
insufficient donations.
Figure 2.1. Cumulative Vaccine Deliveries by Source
(Per 100 people, as of 2022Q1)
Source: Airfinity and IMF staff calculations.
Notes: EMDEs = Emerging markets and developing economies.
Table 2.3. Vaccination Rates and Sources of Vaccines in Sub-Saharan Africa
(1) (2) (3)
Log cumulative delivery (doses per 100 people)
COVAX 0.
00 0.10 0.09*
(0.07) (1.58) (1.69)
Manufacturers 0.33*** 0.32*** 0.32***
(5.52) (3.89) (3.79)
Bilateral donations 0.20** 0.15** 0.15**
(2.61) (2.16) (2.28)
SSA -3.59*** -4.22*** -4.30***
(-3.09) (-3.44) (-3.49)
SSA × Log COVAX delivery 0.82** 1.30*** 1.32***
(2.48) (3.87) (3.86)
SSA × Log manufacturers delivery -0.01 -0.12 -0.16
(-0.11) (-0.79) (-1.11)
SSA × Log bilateral donations delivery 0.14 -0.23 -0.19
(0.98) (-1.36) (-1.08)
Observations 150 95 104
R-squared 0.71 0.87 0.87
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Sources: Airfinity, Our World in Data, and IMF staff calculations.
Log vaccination rate (doses per 100 people), as of 2022Q1
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
29
Annex III. Robustness Checks
In Table 3, we analyze the factors that affect vaccination rates across countries. In this section, we
conduct two robustness checks.
First, we rerun the regressions in Table 3 with population weights. By assigning more weights to
more populous countries, this can alleviate a possible concern that potentially fast progress in
vaccination in a few smaller countries drive the results. Table 3.1 shows that the coefficients on
vaccine delivery in columns 1, 9 and 10 remain statistically significant and quantitatively similar to
those in Table 3, suggesting that the finding that delivery is an important factor driving vaccination
rates is robust. Interestingly, when weighted by population and examined individually, the
coefficients before cumulative cases and deaths are no longer statistically significant, which
indicates that the high correlation between the spread of COVID-19 and vaccination rates might
be driven by only a few countries with small population. When all or most covariates are included
in columns 9 and 10, infrastructure quality score also remains statistically significant, highlighting
again the important role of logistics in vaccination rates.
Table 3.1. Factors Affecting Vaccination Rates, Weighted by Population
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Log cumulative delivery (doses per 100 people)
1.28*** 1.37*** 1.30***
(15.44) (11.03) (9.11)
Log cumulative cases (per 1000 people) 0.06 -0.18* 0.01*
(0.41) (-1.84) (1.75)
Log cumulative deaths (per 1000 people) 0.05 0.17**
(0.34) (2.17)
Log GDP per capita 0.81*** 0.13 0.12
(4.88) (1.09) (0.80)
Log median travel time -0.35* -0.01
(-1.85) (-0.24)
Log infrastructure quality score 2.73*** 0.68*** 0.44**
(3.70) (3.42) (2.08)
Log urban share of population 0.42 0.13 0.21
(1.20) (0.57) (0.78)
Log health expenditure per capita 0.49*** 0.10 -0.29**
(3.13) (0.75) (-2.31)
Log share of 15-64 population 7.64*** 1.41** 0.95
(6.03) (2.32) (1.27)
Log share of 65+ population 0.53*** 0.05 -0.03
(4.25) (0.74) (-0.25)
Log human development index 4.45*** -2.55***
(5.85) (-3.52)
Observations 150 149 148 150 98 145 150 150 95 104
R-squared 0.92 0.02 0.01 0.48 0.50 0.33 0.73 0.54 0.96 0.95
Sources: Airfinity, Our World in Data, Nelson et al. (2019), World Bank World Development Indicators, UNDP, and IMF staff calculations.
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
Log vaccination rate (doses per 100 people), as of 2022Q1
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
30
Second, we take into account vaccine types by adjusting vaccination rates and delivered vaccines.
In Table 3, we use administered doses per hundred people as a dependent variable, which does
not fully reflect immunization coverage. In this exercise, we use the share of people who received
at least one dose of COVID-19 vaccine as a dependent variable. To adjust the number of delivered
vaccines accordingly, we divide deliveries of all vaccines by 2, except J&J which required only one
dose. Results in Table 3.2 show that the coefficients on adjusted cumulative delivery in columns 1,
9 and 10 are statistically significant and also quantitatively similar to those in Table 3. The
coefficients on infrastructure quality score continue to be statistically significant in columns 9 and
10.
Table 3.2. Factors Affecting Vaccination Rates, Accounting for Types of Vaccines
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Log cumulative delivery (doses per 100 people) 1.21*** 1.36* ** 1.33* **
(12.31) (8.77) (7.15)
Log cumulative cases (per 1000 people) 0.36* ** -0.19 0.00
(7.06) (-1.53) (0.06)
Log cumulative deaths (per 1000 people) 0.32* ** 0.18*
(4.62) (1.68)
Log GDP per capita 0.57*** -0.09 -0.15
(6.84) (-0.82) (-1.65)
Log median travel time 0.04 0.01
(0.37) (0.20)
Log infrastructure quality score 1.94*** 0.52* * 0.44**
(4.35) (2.26) (2.57)
Log urban share of population 0.64* 0.33 0.38
(1.83) (1.49) (1.27)
Log health expenditure per capita 0.53* ** -0.08 -0.15
(7.30) (-0.68) (-1.29)
Log share of 15-64 population 4.98* ** 0.54 0.85
(7.17) (0.94) (1.58)
Log share of 65+ population 0.36* ** -0.01 0.03
(5.17) (-0.11) (0.49)
Log human development index 3.38* ** -0.09
(8.08) (-0.11)
Observations 150 149 148 150 98 145 150 150 95 104
R-squared 0.83 0.39 0.28 0.40 0.40 0.41 0.42 0.49 0.89 0.87
Sources: Airfinity, Our World in Data, Nelson et al. (2019), World Bank World Development Indicators, UNDP, and IMF staff calculations.
Log vaccination rate (doses per 100 people), as of 2022Q1
Note: t-statistics in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
31
References
Agarwal, Ruchir, Jeremy Farrar, Gita Gopinath, Richard Hatchett, and Peter Sands. 2022. “A Global Strategy
to Manage the Long-Term Risks of COVID-19.” IMF Working Paper No. 2022/068.
Agarwal, Ruchir, and Tristan Reed. 2022. “Financing vaccine equity: Funding for day-zero of the next
pandemic.” IMF Working Paper No. 2022/099.
Almås, Ingvild, Alexander Cappelen, Jo Lind, Erick Sørensen, and Bertil Tungodden. 2011. “Measuring
unequitable (in) equality.” Journal of Public Economics 95 (7-8): 488–99.
Bollyky, Thomas J., Lawrence Gostin, Margaret Hamburg. 2020. “The Equitable Distribution of COVID-19
Therapeutics and Vaccines.” JAMA. 323 (24): 2462–63.
Cherif, Reda and Shushanik Hakobyan. 2021. “Vaccinating against COVID-19 to Avert a Disaster in Africa.”
IMF Special Series on COVID-19.
Dabla-Norris, Era, Habah Khan, Frederico Lima, Alexandre Sollaci. 2021. “Who Doesn’t Want to be
Vaccinated? Determinants of Vaccine Hesitancy During COVID-19.” IMF Working Paper No. 2021/130.
Deb, Pragyan, Davide Furceri, Daniel Jimenez, Siddarth Kothari, Jonathan Ostry and Nour Tawk. 2021.
“Determinants of COVID-19 Vaccine Rollouts and Their Effects on Health Outcomes.” IMF Working
Paper No. 2021/247.
________. 2021. “The Effects of COVID-19 Vaccines on Economic Activity.” IMF Working Paper No.
2021/248.
The Economist Intelligence Unit. 2021. “How much will vaccine inequity cost?”
Emanuel, Ezekiel, Govind Persad, Adam Kern, et al. 2020. “An ethical framework for global vaccine
allocation.” Science 369 (6509): 1309–12.
Fidler, David P. 2010. “Negotiating equitable access to influenza vaccines: global health diplomacy and the
controversies surrounding avian influenza H5N1 and pandemic influenza H1N1.” PLoS Med 7 (5).
Forman, Rebecca, Soleil Shah, Patrick Jeurissen, et al. 2021. “COVID-19 vaccine challenges: What have we
learned so far and what remains to be done?” Health Policy 125 (5): 553–567.
Goel, Rajeev K., and Michael A. Nelson. 2021. “Drivers of COVID-19 vaccinations: vaccine delivery and
delivery efficiency in the United States.” NETNOMICS: Economic Research and Electronic Networking
22: 53–69.
Hufe, Paul, Ravi Kanbur, and Andreas Peichl. 2021. “Measuring Unequitable Inequality: Reconciling
Equality of Opportunity and Freedom from Poverty.” The Review of Economic Studies 89 (6): 3345–80.
Lazarus, Jeffrey, Scott C. Ratzan, Adam Palayew, et al. 2021. “A global survey of potential acceptance of a
COVID-19 vaccine.” Nat Med 27: 225–228.
IMF WORKING PAPERS
Equitable access to vaccines: Myth or reality?
INTERNATIONAL MONETARY FUND
32
Magdalou, Brice and Richard Nock. 2011. “Income distributions and decomposable divergence
measures.” Journal of Economic Theory 146 (6): 2440–54.
Nelson, Andy, Daniel J. Weiss, Jacob van Etten, et al. 20
19. "A suite of global accessibility indicators." Sci
Data 6 (1): 266.
Peacocke, Elizabeth F., Lieke F. Heupink, Katrine Frønsdal, et al. 2021. “Global access to COVID-19 vaccines:
a scoping review of factors that may influence equitable access for low and middle-income countries.”
BMJ Open 11 (9).
Persad, Govind, Monica E. Peek, and Ezekiel J. Emanuel. 2020. “Fairly Prioritizing Groups for Access to
COVID-19 Vaccines.” JAMA 324 (16): 1601–2.
Ravallion, Martin. 2019. “Global inequality when unequal countries create unequal people.” European
Economic Review 111 (C): 85–97.
Roemer, John E. and Alain Trannoy. 2016. “Equality of opportunity: Theory and measurement.” Journal of
Economic Literature 54 (4): 1288–1332.
Sallam,
Malik. 2021. “COVID-19 Vaccine Hesitancy Worldwide: A Concise Systematic Review of Vaccine
Acceptance Rates.” Vaccines 9 (2): 160.
Shorrocks, Anthony. 2013. “Decomposition procedures for distributional analysis: a unified framework
based on the Shapley value.” Journal of Economic Inequality 11 (1): 99.
So, A
nthony D., and Joshua Woo. 2020. “Reserving coronavirus disease 2019 vaccines for global access:
Cross sectional analysis.” BMJ 371.
Solís Arce, Julio S., Shana S. Warren, Niccoló F. Meriggi, et al. 2021. “COVID-19 vaccine acceptance and
hesitancy in low- and middle-income countries.” Nat Med 27: 1385–94.
Wouters, Oliver, Kenneth Shadlen, Maximillian Salcher-Konrad, et al. 2021. “Challenges in ensuring global
access to COVID-19 vaccines: production, affordability, allocation, and deployment.” The Lancet 397
(10278): 1023–34.
