1 What vaccination rate(s) minimise total societal costs after 'opening up' to COVID-19?

2 Age-structured SIRM Results for the Delta variant in Australia (New South Wales, Victoria
3 and Western Australia)
4

5 Long Chu1, R. Quentin Grafton1,*, Tom Kompas1,2

6

7
1
8 Crawford School of Public Policy, Australian National University, Canberra, Australia

9 2 Centre of Excellence for Biosecurity Risk Analysis, School of Biosciences and School of Ecosystem and
10 Forest Sciences, University of Melbourne, Melbourne, Australia

11
*
12 Corresponding author: Tom Kompas, Centre of Excellence for Biosecurity Risk Analysis,
13 School of Biosciences and School of Ecosystem and Forest Sciences, University of
14 Melbourne, Melbourne, Australia
15
16 Email: tom.kompas@unimelb.edu.au
17

18 Data and source code is availabe at: https://osf.io/zu3bn/

19

20

21

22

23

24

25 18 October 2021
26

Electronic copy available at: https://ssrn.com/abstract=3944437

27 Abstract
28 Using an age-structured, stochastic SIRM model, calibrated to Australian data post July 2021 with
29 community transmission of the Delta variant, we project possible public health outcomes (daily cases,
30 hospitalisations, ICU beds, ventilators and fatalities) and economy costs for three states: New South
31 Wales (NSW), Victoria (VIC) and Western Australia (WA). NSW and VIC have had on-going
32 community transmission since July 2021 and have been in ‘lockdown’ to suppress transmission. WA
33 did not have on-going community transmission nor was it in lockdown at the model start date (11
34 October 2021) but did maintain strict state border controls. We project the public health outcomes and
35 the economic costs of ‘opening up’ (relaxation of lockdowns in NSW and VIC or fully opening the state
36 border for WA) at alternative vaccination rates (70%, 80% and 90%), compare peak patient demand for
37 ICU beds and ventilators to staffed state-level bed capacity, and calculate a ‘preferred’ vaccination rate
38 that minimises societal costs that varies by state. We find that the preferred vaccination rate for all states
39 exceeds 80% and that the preferred population vaccination rate is increasing with: (1) the speed of
40 vaccination; (2) the effectiveness (infection, hospitalisation and fatality) of the vaccine; (3) the lower is
41 the daily lockdown cost; (4) the larger are the public health costs from COVID-19; (5) the higher is the
42 rate of community transmission before opening up; and (6) the less effective are the public health
43 measures after opening up.
44 Keywords: SARS-CoV-2, Hospitalisation, ICU admission, pandemic, societal costs

45

Electronic copy available at: https://ssrn.com/abstract=3944437

46 1. Introduction
47 Most national governments are seeking to vaccinate their populations as fast as possible to reduce the
48 likelihood of hospitalisations and fatalities from COVID-19 caused by the SARS-CoV-2 virus. For
49 countries with widespread community transmission and public health measures in place intended to
50 reduce transmission, a key public policy question is: What is the preferred vaccination rate(s) to trigger
51 relaxation of public health measures (‘opening up’) that may include school closures, work-from-home
52 advisories, temporary closure of most retail shops, etc.? From a total societal cost perspective this
53 involves a trade-off. That is, a higher vaccination rate prior to opening up will, all else equal, reduce
54 public health costs but the delay to achieve a higher vaccination rate may increase cumulative economy
55 lockdown costs. We calculate these trade-offs to estimate a ‘preferred’ population vaccination rate for
56 opening up an economy that has either stringent public health measures in place, that we label a
57 ‘lockdown’, or strict border controls, such as supervised 14 days quarantine, for all arrivals.
58 We used data for Delta variant community transmission and from 1 July 2021and for three states in
59 Australia: New South Wales (NSW), Victoria (VIC) and Western Australia (WA). We project using
60 separate state-level Susceptible-Infected-Recovered-Mortality (SIRM) models, the state-level public
61 health outcomes associated with opening up after 70%, 80% and 90% of the eligible (12 years and
62 older) state populations are fully vaccinated, from a model start date (11 October 2021). At the model
63 start date for our SIRM projections, NSW had widespread community transmission of the Delta variant
64 but declining new daily cases, VIC had widespread community transmission of the Delta variant with
65 increasing new daily cases, and WA had no community transmission.
66 The NSW and VIC epidemics began from an initial outbreak first identified in Sydney on 16 June, 2021
67 [1]. In response to their state-wide epidemics, the NSW and VIC state governments imposed a variety
68 of public health measures including school closures, work-from-home advisories, mandated mask
69 wearing, among other measures. As the model start date, WA did not have community transmission
70 (zero COVID), and only had minimal public health measures (no lockdown). Nevertheless, it did have
71 strict arrival protocols, including an entry permit for all arrivals, COVID-19 testing, and 14-days
72 supervised quarantine for all international and domestic arrivals from Australian jurisdictions with
73 community transmission.
74 Australia provides a valuable comparison to other countries when assessing the preferred population
75 vaccination rate to begin relaxation of stringent public health measures to reduce COVID-19
76 community transmission. This is because of: (1) the diversity of public health approaches within
77 Australia and across its state jurisdictions; (2) differences in community transmission across Australian
78 states; (3) a low rate of COVID-19 infection in Australia such that almost the entire population is highly
79 susceptible to COVID-19 in the absence of vaccination; (4) an agreed-to-National Plan in relation to
80 opening up and the use of lockdowns based on 70% and 80% national vaccination levels of those aged
81 16 years and above [2]; (5) multiple state and/or national epidemiological models to compare projected
82 health outcomes [3-7]; and (6) ex-ante opportunity to determine the preferred vaccination rate by state.
83 Our contributions are three-fold. First, we develop separate state-based age-structured SIRM models
84 for NSW, Victoria and WA calibrated to the Delta variant. Second, our projections from the SIRM
85 models provide an opportunity to assess public health capacity (ICU beds and ventilators) to projected
86 public health demand under multiple vaccination rate scenarios. Third, we use Australian Treasury
87 estimates of lockdown costs [8], estimates of health care costs [9], including losses associated with
88 those who recover, and estimates of the dollar loss of fatalities, adjusted for age with a Value of
89 Statistical Life (VSL) used by the Australian government [10]. Using these costs and SIRM models we
90 estimated the preferred vaccination rate that minimises total societal costs from opening up. Our
91 methods where suitable data are available, can be applied to any jurisdiction, both ex-ante and ex-post,
92 to determine a preferred population vaccination rate(s) to relax stringent public health measures.

Electronic copy available at: https://ssrn.com/abstract=3944437

93 2. Model specification
94 Australia has six states and two territories. Each jurisdiction is responsible for its own public health
95 measures and has the ability to control entry and exit from their borders. We separately estimate a
96 separate age-structured SIRM for NSW, Vic and WA. Collectively, these three states represent more
97 than two thirds of the Australian population and economy.

98 2.1. Age cohorts and population

99 Table 1 summarises the population of Australia and its two most populous states NSW and VIC and its
100 largest state by area, WA. There are ten age groups, following the age group classifications of the
101 Australian Bureau of Statistics (ABS), for each of the three-age structured SIRM models estimated for
102 each state.
103 Table 1. Australian, NSW, VIC, and WA age cohorts (millions of people)
Group Age Australia NSW VIC WA
index Total 12+ Total 12+ Total 12+ Total 12+
(i)
1 0-9 3.18 0.00 1.01 0.00 0.82 0.00 0.35 0.00
2 10-19 3.09 2.44 0.97 0.77 0.77 0.61 0.33 0.26
3 20-29 3.62 3.62 1.15 1.15 1.01 1.01 0.35 0.35
4 30-39 3.76 3.76 1.19 1.19 1.03 1.03 0.4 0.4
5 40-49 3.30 3.30 1.04 1.04 0.86 0.86 0.35 0.35
6 50-59 3.12 3.12 0.98 0.98 0.79 0.79 0.33 0.33
7 60-69 2.70 2.70 0.87 0.87 0.68 0.68 0.27 0.27
8 70-79 1.88 1.88 0.61 0.61 0.47 0.47 0.18 0.18
9 80-89 0.85 0.85 0.28 0.28 0.22 0.22 0.08 0.08
10 90+ 0.21 0.21 0.07 0.07 0.06 0.06 0.02 0.02
104 Notes: The population in each age group is extracted from ABS statistics in millions of people together with 12+
105 people who are eligible for Covid-19 vaccination, rounded to the nearest 2-decimal places.
106 Each group 𝑖 ∈ [1,2, … ,10] is further classified into two categories, unvaccinated and vaccinated. Each
107 category has its own SIRM compartment. The two categories are linked by the flow of vaccination from
108 the unvaccinated to the vaccinated category. Our model only considers fully vaccinated (with two
109 doses) people labelled 'vaccinated' and those who have either no dose or only one dose labelled
110 'unvaccinated'.

Electronic copy available at: https://ssrn.com/abstract=3944437

111
112 Fig 1. Age-structured SIRM compartments

113 2.2. Multi-age-group infections

114 Susceptible people in group 𝑖, either unvaccinated or vaccinated, may get infected during contact with
115 infected people, i.e., the source of infection. The source of infection can be in any of the age groups,
116 𝑗 ∈ [1,2, . . 10], including the same group of the susceptible people (group 𝑖). The probability of a
117 susceptible person in group 𝑖 getting infected depends on: (i) the relative susceptibility of group 𝑖 which
118 is presented by parameter 𝛾! , (ii) whether the susceptible person (in group 𝑖) has been vaccinated –
119 vaccination can reduce the susceptibility of people in age group 𝑖 by 𝜎! ∈ (0,1), and (iii) the
120 contagiousness of the source of infection in age group 𝑗 toward susceptible people in age group 𝑖 - the
121 relative contagiousness of unvaccinated source of infection (denoted as 𝛼!" ) and the vaccine
122 effectiveness with respect to onward transmission, i.e., how much vaccination can reduce onward
123 transmission (denoted as 𝛽!" ).

124 The dynamics of each age group 𝑖 ∈ [1,2. .10] are formalised in equations (1)-(8) following the SIRM
125 diagram in Fig 1. The first four equations present the unvaccinated compartments for each age group,
126 and the remaining four equations represent the vaccinated compartments noting we do not account for
127 any monthly population growth.
128 The parameter 𝑅# is the average (unmitigated) basic reproduction rate, 𝑁! is the population of age group
129 𝑖, 𝑇! is the average unmitigated exposure time of an infection source, Γ! is the effectiveness of control
130 measures (i.e., how effective are public health measures at reducing community transmission within
131 each age cohort), 𝑟𝑢! and 𝑟𝑣! are the finalisation rates (either recovery or death) of unvaccinated and
132 vaccinated patients in age group 𝑖. Further, 𝑚𝑢! and 𝑚! are, respectively, the mortality rate of
133 unvaccinated patients in age group 𝑖 and the vaccine effectiveness on mortality (i.e., how vaccination
134 reduce mortality risk). Compartmental variables are defined in Fig 1.

Electronic copy available at: https://ssrn.com/abstract=3944437

 $%&! ( )! *"
135 = −𝑉! − ∑#"! (1 − Γ! ) 𝑈𝑆! ∑)
"-.A𝛼!" 𝑈𝐼" + (1 − 𝛽!" )𝛼!" 𝑉𝐼" D (1)
$' # (! )! ,!

$%/! ( )! *"
136 = ∑#"! (1 − Γ! )𝑈𝑆! ∑)
"-.A𝛼!" 𝑈𝐼" + (1 − 𝛽!" )𝛼!" 𝑉𝐼" D − 𝑈𝐼! 𝑟𝑢! (2)
$' # (! )! ,!

$%*!
137 = 𝑈𝐼! 𝑟𝑢! (1 − 𝑚𝑢! ) (3)
$'

$%0!
138 = 𝑈𝐼! 𝑟𝑢! 𝑚𝑢! (4)
$'

$1&! ()*
139 = 𝑉! − (1 − 𝜎! ) ∑#"! ( !)", (1 − Γ! )𝑉𝑆! ∑)
"-.A𝛼!" 𝑈𝐼" + (1 − 𝛽!" )𝛼!" 𝑉𝐼" D (5)
$' # ! ! !

!"#! &'(
140 !$
= (1 − 𝜎% ) ∑#"! & !'"* (1 − Γ% )𝑉𝑆% ∑'
+,-+𝛼%+ 𝑈𝐼+ + (1 − 𝛽%+ )𝛼%+ 𝑉𝐼+ 2 − 𝑉𝐼% 𝑟𝑣% (6)
# ! ! !

$1*!
141 $'
= 𝑉𝐼! 𝑟𝑣! +1 − 𝑚𝑢% (1 − 𝑚% )2 (7)
$10!
142 $'
= 𝑉𝐼! 𝑟𝑣! 𝑚𝑢! (1 − 𝑚! ) (8)

143 We assume a fixed reduction in community transmission associated with opening up, but we vary this
144 parameter in sensitivity analysis with respect to the preferred vaccination rate.

145 2.3. Hospitalisations, ICU admissions, and ventilation requirements

146 We assume the number of hospitalisations due to COVID-19 is a fixed proportion of the number of
147 infected patients, where the fraction varies across age groups and whether the patients have been
148 vaccinated. The estimated number of hospitalisations is formalised in equation (9) for unvaccinated
149 patients (𝑈𝐻! ) and in equation (10) for vaccinated patients (𝑉𝐻! ). In these equations, ℎ! and 𝑣ℎ! are the
150 average hospitalisation rate of unvaccinated cases and the vaccine effectiveness on hospitalisation (i.e.,
151 how much vaccination reduces the risk of hospitalisation) of age group 𝑖. Similarly, the estimated ICU
152 admissions are formalised in equations (11) and (12) where 𝑖𝑐𝑢! and 𝑣𝑖𝑐𝑢! are the rate of ICU
153 admissions for unvaccinated cases and the vaccine effectiveness on ICU admissions, respectively. The
154 estimated ventilation requirements are formalised in equations (13) and (14) where 𝑣𝑒𝑛! and 𝑣𝑣𝑒𝑛! are
155 the average rate of ventilation requirements for unvaccinated cases and the vaccine effectiveness on
156 ventilation requirements.
157 𝑈𝐻! = 𝑈𝐼! × ℎ! (9)
158 𝑉𝐻! = 𝑉𝐼! × ℎ! (1 − 𝑣ℎ! ) (10)

159 𝑈𝐼𝐶𝑈! = 𝑈𝐼! × 𝑖𝑐𝑢! (11)

160 𝑉𝐼𝐶𝑈! = 𝑉𝐼! × 𝑖𝑐𝑢! (1 − 𝑣𝑖𝑐𝑢! ) (12)
161 𝑈𝑉𝑒𝑛! = 𝑈𝐼! × 𝑣𝑒𝑛! (13)
162 𝑉𝑉𝑒𝑛! = 𝑉𝐼! × 𝑣𝑒𝑛! (1 − 𝑣𝑣𝑒𝑛! ) (14)

163 3. Data and parameterisation

164 3.1. Summary of key COVID19 data
165 Table 2 summarises COVID-19 outcomes at the model start date, 11th October 2021, in Australia, and
166 separately for NSW, VIC and WA. This table shows that NSW and VIC have accounted for more than
167 95% of the total cases. Epidemiological indicators vary across the states with substantial differences in

Electronic copy available at: https://ssrn.com/abstract=3944437

168 terms of the hospitalisation rates, the ICU admission rate, and the ventilation rate since the start of the
169 pandemic.
170 The progress of vaccination has varied across the states. For example, about 58% of the eligible (12+)
171 population in NSW have been fully vaccinated as of the model start date (11 October 2021), and the
172 number of fully vaccinated people has been increasing by about 67,000 per day in NSW over the
173 preceding 14-day period. By comparison, on the model start date, about 46% of VIC's eligible
174 population was fully vaccinated and it had a daily vaccination rate of 45,000 over the preceding 14
175 days.
176 Table 2. Summary of key COVID-19 Statistics in NSW, VIC and WA, as of 11 October 2021

NSW VIC WA
Total cases (1000 people) 69.21 54.47 1.11
Active cases (1000 people) 6.75 19.01 0
Hospital cases (1000 people) 769 677
0
[Hospitalization rate] [11%] [3.7%]
ICU cases (1000 people) 0.153 0.133
0
[ICU admission rate] [2.2%] [0.7%]
Cases on ventilators (1000 people) 0.071 0.094 0
[Ventilation requirement rate] [1.1%] [0.5%] [NA]
Recovered (1000 people) 61.96 34.546 1.1
Total deaths (1000 people) 499 913 0.01
[Mortality Rate] [0.8%] [0.6%] [NA]
14-day new-case trend with control
0.72 1.616 0
measures (1000 people/day)
Fully vaccinated (million people) 4.92 3.23 1.10
[Full vaccination/eligible population] [70.7%] [56.6%] [49.1%]
14-day vaccination rate (million
0.063 0.045 0.011
people /day)
177 Notes:
178 1. Data sources are reported in Appendix 2.
179 2. 14-day new case trend, the 14-day vaccination rates, the rates of hospitalisation, ICU admission, and
180 ventilation are estimated using the average of the 14-day period ending 11 October 2021. The mortality
181 rates only are estimated using data from 1 July 2021 that coincident with the Delta epidemic in NSW
182 and VIC.

183 3.2. Epidemiological parameters

184 Epidemiological parameters vary significantly across age groups. Table 3 provides the age distribution
185 of the epidemiological parameters from [3: table S3]. For each jurisdiction (NSW, VIC and WA), we
186 scale the age distribution in Table 3 to match the aggregate values observed in actual data in Table 2.
187 In addition, we assume the age distribution of the ventilation requirement is similar to that of the ICU
188 admissions.
189 COVID-19 is highly infectious if not mitigated. We assume a daily transmission rate of COVID-19
190 equivalent to the reproduction number that varies between 4.0 and 8.0 [11], i.e., (𝑅# ~[4,8]). We also
191 assume that a low level of restrictions (social distancing, masks, hygiene) can reduce the transmission
192 by 0%-20% (Γ! ~[0,0.2])
193 Under the most recent lockdowns, community transmission in NSW and VIC is assumed to generate
194 daily new cases between 80% and 120% of what has been observed during the 14-day period ending at
195 the model start date, noting that WA had no community transmission at the model start date. The

Electronic copy available at: https://ssrn.com/abstract=3944437

196 average recovery time for COVID-19 patients, if they can recover and do not suffer from 'long COVID'
197 symptoms, is 14 days [12], i.e., 𝑟𝑣! = 𝑢𝑣! = 365/14.
198 Table 3. Age distribution of key epidemiological parameters
Age groups Relative susceptibility Hospitalisation ICU rate Mortality rate
(Ordinal susceptibility rate (% of cases) (% of cases)
units) (% of cases)
0-9 0.34 0.001 6E-05 4E-05
10-19 0.66 0.004 1E-04 4E-05
20-29 1.00 0.019 8E-04 2E-04
30-39 1.00 0.052 0.002 7E-04
40-49 1.00 0.074 0.004 0.002
50-59 1.00 0.171 0.019 0.006
60-69 1.24 0.283 0.071 0.016
70-79 1.47 0.413 0.122 0.049
80-89 1.47 0.449 0.031 0.158
90+ 1.47 0.449 0.031 0.287
199 Source: [3: table S3]

200 3.3. Vaccination targets, vaccination progress, and vaccine efficacy

201 For each jurisdiction, we consider three vaccination targets for removing lockdown measures, namely
202 70%, 80% and 90% of the eligible population (12 years and older) are fully vaccinated. The vaccination
203 progress is assumed to be similar to the average vaccination rate during the 14-day period ending at the
204 model start date of 11 October 2021 and until 90% of the eligible (12 years and older) population is
205 fully vaccinated in the respective states of NSW, VIC and WA. In all cases, we assume that opening up
206 is irreversible and, thus, we do not consider lockdowns of alternative stringency or any subsequent re-
207 introduction of lockdowns [13].
208 We assume the numbers of people becoming fully vaccinated across the ten age groups each day are
209 proportional to the number of people waiting for vaccinations across the age groups, i.e., those who
210 want to be vaccinated but are not yet fully vaccinated. We also specify that the proportion of the
211 Australian population (and for each state for each age cohort) that is vaccine resistant such that they
212 would ‘Definetly not’ get vaccinated is 5.5% [14]. We note that the proportion of those unwilling to be
213 vaccinated in Australia declined from 9.4% to 8.2% in September 2021 and this trend is expected to
214 continue [15].
215 In relation to vaccine effectiveness, we use the parameters published by [4: table S2.3 and S2.5] for
216 Australia, as given in Table 4. When there are significant differences across vaccines in relation to
217 effectiveness, we calculate their average levels (𝜎! = 0.7, 𝛽!" = 0.65, 𝑣ℎ! = 0.87, 𝑣𝑖𝑐𝑢! =
218 0.88, 𝑣𝑣𝑒𝑛! = 0.88, 𝑚! = 0.9)
219 Table 4. Australian (NSW, VIC, and WA) Vaccine Effectiveness
Susceptibility Onward Hospitalisation ICU Ventilation Mortality rate
transmission rate rate rate
0.7 0.65 0.87 0.88 0.88 0.9
220 Source: Adapted from [4: table S2.3 and S2.5]

Electronic copy available at: https://ssrn.com/abstract=3944437

221 3.4. Health care, welfare, and economy costs
222 Parameters for the health care and welfare costs are taken from [9]. The cost of a normal hospital bed
223 is 1,839 AUD/day, and the cost of an ICU bed is 4,885 AUD/day. The value of a statistical life year
224 (VSLY) is 217,000AUD [10]. If a COVID-19 patient can recover without long-COVID symptoms, we
225 assume the welfare loss of the patient is equivalent to the value of time required for recovery as a
226 proportion of the VSLY. We assume that 10% of patients have long COVID symptoms, i.e., they are
227 no longer infectious, but some of their symptoms persist for up to 12 weeks.
228
229 We specify that the VSL is 5.0 million AUD [10] for people below 60 years. For those from 60 years
230 and above, we calculate an adjusted-VSL that is assumed to decrease linearly to 10% of the VSL (more
231 than twice of the estimated VSLY) until 89-years of age. For 90-year-olds and above, the adjusted-VSL
232 is fixed at 10% of the unadjusted VSL, or 500,000 AUD.
233
234 Our measures of economy costs of lockdowns are derived from the Australian Treasury which
235 categorises costs at a national level into: (1) strict lockdown, (2) moderate lockdown, (3) low-level
236 restrictions and (4) baseline (minimum public health) restrictions. Their associated costs are 3.2, 2.35,
237 0.65 and 0.1 billion AUD per week, respectively, at a national level. These costs to do not include
238 additional mental health or other costs associated with lockdowns [16].
239
240 We scaled the Australian Treasury costs to a state-level cost based on the Gross State Product as a
241 proportion of Australia's Gross Domestic Product. We assumed that between a 70% and 80%
242 vaccination rate at a state level when there is community transmission (NSW and VIC) that the
243 lockdown costs range from low-level restrictions to moderate lockdown; at 80% to 90% vaccination
244 rate the public health measures cost are the equivalent of low-level restrictions; and with a 90%
245 vaccination rate and above baseline restrictions apply.
246
247 Table 5. Australian Public Health Measures and Economy Costs by Stringency of Public Health
248 Measures

Moderate Low level Baseline

Strict lockdowns
lockdowns restrictions restrictions

Economy cost $3.2 billion/week $2.35 billion/week $0.65 $0.1

billion/week billion/week
Stay at home • Stay-at-home • Stay-at-home • No stay-at- • No stay-at-
orders except essential except for work, home orders home orders
purposes study and essential
purposes
Density • 4 sqm rule • 2 sqm rule • 2 sqm rule • 2 sqm rule
restrictions
Retail trade • Non-essential • Increased retail • Social • Social
retailers and venues activity, subject to distancing rules distancing rules
closed to public density restrictions apply apply
• Take away and • Seated dining for • Larger groups
home delivery only small groups at allowed
cafes/restaurants
Work • Only workplaces • Work from • Return to work, • 1.5 sqm rule
categorised as home, if possible, but social
permitted work capacity limits and distancing and
allowed to operate restrictions on capacity
on-site and subject office space apply restrictions on
to restrictions office space apply

Electronic copy available at: https://ssrn.com/abstract=3944437

 Schools and • Closed, remote • Closed or • Open • Open
childcare learning only graduated return

Capacity • No gatherings, • Indoor venues • Recreational • Large sporting

restrictions non-essential venues closed activities allowed venues to
etc. closed • Capacity limits and venues open operate at 70%
restricted to small but social capacity
groups outdoors distancing and
capacity limits
apply
Travel • Essential • Non-essential • No travel • No travel
restrictions movements only travel limited; no restrictions restrictions
within 5 or 10 km intra or inter-state • Interstate travel • Interstate
radius travel allowed travel allowed
• No intra- or inter-
state travel
• Other • Curfew • 5 visitors to • Requirements
• No household household and for record-
visitors and 2-person limited outdoor keeping, COVID-
limit on exercise gatherings e.g., 10 safe plans
people
249 Sources: Extracted from [8: tables 4-5]

250 3.5. Calibration process

251 The system of differential equations (9)-(13) is numerically solved with the finite difference technique
.
252 [17, 18]. The time unit is one year, and the time step is set at (one day) to match the frequency of
234
253 reported data. Time zero, the model start date, is specified as 11 October, 2021. The maximum
254 simulation time horizon is 4 years (1,460 days). The demand for health care services (hospitalisations,
255 ICU, and ventilation) were calculated using equations (9)-(14). The rates of hospitalisation, ICU, and
256 ventilation in each state are reported in the appendix. These state-specific rates are estimated, using the
257 age distribution in Table 3, but scaled to match the aggregate values observed in actual data in Table 2.

258 4. Model Results

259 4.1. Vaccination progress
260 At a national level, the daily vaccination rate is estimated to be nearly 160,000 people a day, as
261 approximated for the 14-day period ending at the model start date of 11 October 2021. At this
262 vaccination rate, it would take another 16, 29, and 43 days for Australia to reach the 70%, 80% and
263 90% targets, respectively, as shown in Fig 2. After the 90% target is reached, the daily vaccination rate
264 is reduced by 20%, and the black line, which represents the number of fully vaccinated people, become
265 slightly flatter. This is consistent with a flattening of the speed of vaccination at high vaccination rates
266 [19].

10

Electronic copy available at: https://ssrn.com/abstract=3944437

267
268 Fig 2. Australian population (millions of people) vaccination targets (%) from 11 October 2021

269 4.2. NSW

270 The projected outcomes for NSW are presented in Fig 3 and summarised in Table 6. The Fig
271 summarises six key public health outcomes for three scenarios: (i) lockdown ends when 70% of the
272 eligible population is fully vaccinated (red colour), (ii) lockdown ends when 80% of the eligible
273 population is fully vaccinated (yellow colour), and (iii) lockdown ends when 90% vaccination of the
274 eligible population is fully vaccinated (green colour). The solid curves are the mean levels of the
275 projected outcomes, and the bands represent their 95% confidence intervals. All six panels project what
276 would happen before and after a vaccination level (70%, 80% and 90%) is achieved, which is presented
277 by the vertical lines.
278 On 11 October 2021, NSW achieved the 70% vaccination target. If the vaccination rate is maintained
279 at the average of the 14-day period ending 11 October 2021 (i.e., around 67,000 people become fully
280 vaccinated a day), it will take NSW around 10-11 days to achieve the 80% target and another 10-11
281 days to achieve the 90% target. These time periods are shorter than the vaccination progress for
282 Australia to achieve the same vaccination rates.

283
284 Fig 3. NSW: Projected Public health outcomes after opening up
285 Notes:

286 1. Day zero is specified to be 11 October 2021.

11

Electronic copy available at: https://ssrn.com/abstract=3944437

287 2. The red, yellow, and green colours represent scenarios where lockdowns are removed after the 70%,
288 80% and 90% targets, respectively.
289 3. The (red, yellow, and green) vertical lines represent the days when the 70%, 80% and 90% targets are
290 reached.
291 4. The (red, yellow, and green) curves represent the mean of all projection outcomes, and the bands
292 represent their 95% confidence intervals.
293 5. The black horizontal lines in panels (d) and (e) are the capacity of the health care system, i.e., staffed
294 ICU beds and ventilators (data sources reported in Appendix 2).
295 6. The red horizontal dashed lines in panels (d) and (e) are the capacity net the non-COVID19 demand,
296 which is estimated via the occupation rates of ICU beds and ventilators in 2018/2019 (data sources
297 reported in Appendix 2).
298
299 Table 6. NSW: Projected outcomes at different vaccination rates for opening up (1000’s people)

Vaccination rates for opening up

70% 80% 90%
451
Cumulative cases [429.9- 375.4 338.2
473.1] [354.2-398.2] [315.2-361.7]
34.2 21.1 15.7
Peak active cases
[31.3-37.2] [18.4-24] [13.1-18.4]
2.4 1.3 0.9
Peak hospitalisation
[2.3-2.6] [1.2-1.5] [0.8-1.1]
Peak ICU 0.6 0.3 0.2
(Capacity= 1.024; Average non- [0.5-0.6] [0.3-0.4] [0.2-0.3]
Covid demand=0.578; Net
capacity=0.446)
Peak ventilation 0.3 0.2 0.1
(Capacity=2.447; Average non-Covid [0.3-0.3] [0.1-0.2] [0.1-0.1]
demand=0.197; Net capacity=2.250)
Cumulative fatalities 2.4 2.1 1.9
[2.4-2.5] [2-2.2] [1.8-2]
300 Notes:

301 1. Outside brackets are the mean, inside brackets are the 95% CI. Numbers are rounded to the nearest 1-decimal
302 place.
303 2. ICU capacity is estimated at 1024 beds, including 884 available, staffed ICU beds plus 140 additional staffed ICU
304 beds available. Average demand for ICU beds from non-COVID-19 patients is estimated at 578, i.e., ~211,000 bed
305 days in 2018/19 (data sources reported in Appendix 2). ICU net capacity (446 beds) is the difference
306 between the total capacity and the non-COVID-19 demand.
307 3. Ventilation capacity includes 2,447 ventilators. Average demand for ventilators from non-COVID-19 patients is
308 estimated at 197 in 2018/19, i.e., ~34% of the total admission (data sources reported in Appendix 2). Ventilation
309 net capacity (2,250 ventilators) is the difference between the total capacity and the non-COVID-19
310 demand.

311 Community transmission increases when lockdowns are relaxed. We highlight that the number of ICU
312 beds and ventilators available for COVID-19 patients depends on the staffing capacity to maintain the
313 high quality of care needed by patients in ICU. While there is an estimated ‘surge’ capacity for
314 additional beds in ICU of about 800 for Australia, only half of this surge capacity could be staffed with
315 suitably qualified and experienced medical personnel [20]. Thus, a capacity limit on suitably qualified
316 and experienced staff places an upper limit on the net capacity of ICU beds and ventilators available for
317 non-COVD-19 and COVID-19 patients. When patient demand is close to or exceeds this net capacity
318 for ICU beds and ventilators, the fatality rate for non-COVD-19 and COVID-19 patients will likely rise.

12

Electronic copy available at: https://ssrn.com/abstract=3944437

319 Table 7 compares the cost of health care services and patient welfare losses across different vaccination
320 targets from opening up in NSW. The numbers in this table are costs beginning on 11th October 2021,
321 and all costs incurred before that day, (including health service cost and welfare losses for recovered
322 and non-recovered patients) are considered ‘sunk’ in that they would be incurred regardless of the
323 opening-up decision.
324 The costs of health care services would be some 7%-8% of the total costs in NSW. Opening up at a
325 higher vaccination rate would reduce the total loss by 1.9 billion AUD when the vaccination rate is
326 increased from 70% to 80%, and this would be further reduced by 0.9 billion AUD if lockdowns were
327 maintained until the 90% vaccination rate is achieved.
328 Table 7. NSW: Estimates of health care and welfare losses at different vaccination rates for opening
329 up, billions AUD

Vacc Health care services Welfare losses Total

inati Non- ICU Total Recover Non- Total Health care
on ICU people recovered and welfare
rate losses
0.4 0.3 0.7 5.1 4.2 9.4 10.1
70%
[0.4-0.4] [0.3-0.3] [0.7-0.8] [4.7-5.6] [4.1-4.4] [8.8-10] [9.4-10.8]
0.3 0.3 0.6 4.1 3.5 7.6 8.2
80%
[0.3-0.4] [0.2-0.3] [0.5-0.6] [3.6-4.7] [3.3-3.6] [6.9-8.3] [7.4-9]
0.3 0.2 0.5 3.6 3.1 6.7 7.3
90%
[0.2-0.3] [0.2-0.3] [0.4-0.6] [3.1-4.2] [2.9-3.3] [6-7.5] [6.5-8.1]
330 Notes:
331 1. All costs are counted from 11 October 2021.
332 2. Outside brackets are the mean, inside brackets are the 95% CI. Numbers are rounded to the nearest 1-decimal
333 place.
334
335 Fig 4 provides three cost curves: (1) Health care costs and welfare losses from COVID-19; (2) Economy
336 lockdown costs; and (3) Total costs, the sum of both health care costs and welfare losses and economy
337 lockdown costs. The costs of health care services and welfare losses are declining in the vaccination
338 rate before opening up because a greater level of vaccination means fewer hospitalisations and fatalities,
339 all else equal. The economy lockdown costs are increasing in the vaccination rate as the higher is the
340 vaccination rate at opening up the longer are the cumulative daily costs associated with a lockdown.
341 From a societal costs perspective, the minimum of the sum of the two costs (health care and welfare
342 losses and economy lockdown costs) is the preferred vaccination rate for opening up noting a higher
343 vaccination rate will always deliver lower health care costs and welfare losses but higher lockdown
344 costs. To the extent that vulnerable communities that have a higher risk of contracting COVID-19 [21]
345 and/or have a higher rate of hospitalisation or mortality, other than differences based on age, and a lower
346 vaccination rate than the state average, our estimates will underestimate the health care costs and
347 welfare losses at the preferred vaccination rate.
348 The economy lockdown cost in NSW is estimated by scaling down the national economy cost of
349 lockdowns per week (Table 5) using the share of NSW in the total GDP of Australia (31.7%) and the
350 time required for NSW to reach a target for vaccination rate, i.e., 10-11 days to increase the vaccination
351 rate from 70% to 80% or from 80% to 90%. During this period, the economy lockdown cost would be
352 around 0.77 billion AUD. Fig 4 suggests that the preferred vaccination target for opening up in NSW
353 is 85%.

13

Electronic copy available at: https://ssrn.com/abstract=3944437

354
355 Fig 4. NSW: Societal Costs of Different Lockdown ‘Opening Up’ Vaccination Rates
356 Notes:

357 1. All costs are counted from 11 October 2021.

358 2. Total costs = health care costs and welfare losses + economy lockdown costs
359
360 We compared the three common vaccination targets for removing the lockdown in NSW with the
361 preferred vaccination rate for NSW (85%) in Table 8. Results in this table are converted to per-capita
362 and percentage changes for comparison purposes. Opening up NSW at the 70% vaccination rate (with
363 low-level to moderate public health restrictions in place) would incur additional health care and welfare
364 losses of 284 AUD/person (about 30% higher than the cost at the economically justifiable target) and
365 save approximately 149 AUD/person from not having a lockdown. Thus, the net societal loss from
366 premature opening at a 70% vaccination rate, rather than at the preferred rate of 85%, is $135
367 AUD/person. Identical calculations are also provided in Table 7 for the 80% (10 AUD/person loss) and
368 90% (also 10 AUD/person loss) vaccination rates noting the lowest total societal costs are at the
369 preferred vaccination rate of 85%.
370 Table 8. NSW: Per capita (AUD and %) differences from Opening Up at alternative vaccination rate
371 to the preferred vaccination rate (84%)

Health care and welfare Lockdown

Vaccination rate loss AUD/person Economy cost Total cost
70% +284 (+29.7%) -149 (-100%) +135 (+12.2%)
80% +60 (+6.3%) -50 (-33.3%) +10 (+0.9%)
90% -40 (-4.2%) +50 (+33.3%) +10 (+0.9%)
372 Notes:

373 1. Numbers are per-capita deviations from the economically justifiable vaccination rate, '+' = higher
374 and '-' =lower.
375 2. All numbers are counted from 11/10/2021
376 3. Outside brackets are the absolute value of the deviation rounded to the nearest $/person
377 4. Inside brackets are the relative deviation in percentage rounded to the nearest 1-decimal place.

378 Table 9 reports the sensitivity analysis to the preferred vaccination target with respect to effectiveness
379 of reducing community transmission (at 50%, 150% and 200% of the baseline) after opening up and

14

Electronic copy available at: https://ssrn.com/abstract=3944437

380 the speed of vaccination (at 100% and 120% of the baseline). Table 9 shows the preferred vaccination
381 rate is increasing: (1) the less effective are public health measures after opening up and (2) the faster is
382 the speed of vaccination. This second finding is consistent with [5] who used an agent-based modelling
383 framework to show that, for Australia, the more effective are public health measures after opening up
384 then the lower is the vaccination rate required to avoid adverse public health outcomes.
385 Table 9. NSW: Sensitivity of preferred vaccination rate to the effectiveness of public health measures
386 after opening up at reducing community transmission and the speed of vaccination.

Daily vaccination rate

(Baseline =67,000 people/day)
110% 120%
Baseline baseline baseline
Reduction in community
transmission after Opening 50% baseline 86% (11.4) 86% (11.3) 87% (11.1)
Up (Baseline level =10% Baseline 85% (9) 85% (8.9) 85% (8.8)
reduction in community
transmission with public 150% baseline 85% (7) 85% (6.8) 85% (6.7)
health measures) 200% baseline 82% (5.3) 82% (5.2) 82% (5.1)
387 Notes: 1. Outside brackets are the preferred vaccination rate to open up. 2. Inside brackets are the associated
388 total societal cost from 11 October 2021 in billion dollars, rounded to the nearest 1-decomal point
389 associated with that preferred vaccination rate.

390 Table 10 reports the sensitivity analysis of the preferred vaccination rate in relation to the effectiveness
391 of the public health measures during lockdown and state-level lockdown costs (80% and 120% of
392 baseline). Table 10 shows that higher economy lockdown cost increases the total societal cost, but it
393 has insignificant impacts on the preferred vaccination rate.
394 Table 10. NSW: Sensitivity of preferred vaccination rate to economy lockdown cost and community
395 transmission during lockdown
Lockdown cost compared to low-level restriction cost
Baseline level = (2.35-0.65)x31.7% billion AUD/week
80% baseline Baseline 120% baseline
Daily community 80%
88% (8.4) 85% (8.7) 83% (8.9)
transmission during baseline
lockdown Baseline 88% (8.8) 85% (9) 83% (9.2)
(Baseline level =14-day 120%
average prior to 11/10/2021) 88% (9.1) 84% (9.4) 82% (9.6)
baseline
396 Notes:

397 1. Outside brackets are the economically justifiable vaccination target for removing the lockdown.
398 2. Inside brackets are the associated total cost from 11 October 2021 in billion dollars, rounded to the
399 nearest 1-decimal place.

400 4.3. Victoria

401 Public health outcomes for VIC are summarised in Fig 5 and Table 11. If the vaccination rate is
402 maintained at the average of the 14-day period ending 11 October 2021 (i.e., around 45,000 people
403 become fully vaccinated a day), it will take Victoria around 18, 30 and 43 days to achieve the 70%,
404 80%, and 90% rates, respectively. Our projected peak hospitalisations and ICU admissions are
405 comparable to those of [4], respectively, at 1,200-2,500 and 260-550 under various scenarios from
406 October to December 2021. As with NSW, opening up at a higher vaccination rate reduces cases,

15

Electronic copy available at: https://ssrn.com/abstract=3944437

407 hospitalisations, ICU numbers and fatalities. Unlike NSW, VIC is very close after around 45-50 days
408 to reaching its maximum ICU capacity should it open up at a 70% vaccination rate.

409
410 Fig 5. VIC: Projected public health outcomes
411 Notes:

412 1. Day zero is specified to be 11 October 2021.

413 2. The red, yellow, and green colours represent scenarios where lockdowns are removed after the 70%,
414 80% and 90% targets, respectively.
415 3. The (red, yellow, and green) vertical lines represent the days when the 70%, 80% and 90% targets are
416 reached.
417 4. The (red, yellow, and green) curves represent the mean of all projection outcomes, and the bands
418 represent their 95% confidence intervals.
419 5. The black horizontal lines in panels (d) and (e) are the capacity of the health care system, i.e., staffed
420 ICU beds and ventilators (data sources reported in Appendix 2).
421 6. The red horizontal dashed lines in panels (d) and (e) are the capacity net the non-covid demand, which
422 is estimated via the occupation rates of ICU beds and ventilators in 2018/2019 (data sources reported in
423 Appendix 2).

424 Table 11. VIC: Projected outcomes at different vaccination rates for opening (1000’s people)

Vaccination rates for opening up

70% 80% 90%
Cumulative cases 741.3 484.2 387.4
[662.4-827.9] [440.5-533.6] [356.1-423.3]
96.2 48.4 30.2
Peak active cases
[81.5-111.9] [41-56.9] [25.3-35.7]
2.3 1.1 0.7
Peak hospitalisation
[1.9-2.6] [0.9-1.2] [0.7-0.7]
Peak ICU 0.6 0.3 0.1
(Capacity= 0.543; Average non- [0.5-0.7] [0.2-0.3] [0.1-0.2]
Covid demand=0.340; Net
capacity=0.203)
Peak ventilation 0.4 0.2 0.1
(Capacity=2.023; Average non- [0.3-0.4] [0.1-0.2] [0.1-0.1]
Covid demand=0.141; Net
capacity=1.882)
Cumulative fatalities 3.4 2.5 2.1
[3.2-3.8] [2.3-2.6] [2-2.2]
425 Notes:

16

Electronic copy available at: https://ssrn.com/abstract=3944437

426 1. Outside brackets are the mean, inside brackets are the 95% CI. Numbers are rounded to the nearest 1-decimal
427 place.
428 2. ICU capacity is estimated at 543 beds, including 476 available, staffed ICU beds plus 67 additional staffed ICU beds
429 available. Average demand for ICU beds from non-COVID-19 patients is estimated at 340, i.e., ~124,000 bed days
430 in 2018/19 (data sources reported in Appendix 2). ICU net capacity (203 beds) is the difference between
431 the ICU capacity and non-COVID-19 demand.
432 3. Ventilation capacity is estimated at 2,203. Average demand for ventilators from non-COVID-19 patients is estimated
433 at 141 in 2018/19, i.e., ~42% of the total admission (data sources reported in Appendix 2). Ventilation net
434 capacity (1,882 ventilators) is the difference between the ventilation capacity and non-COVID-19
435 demand.
436
437 Table 12 summarises the cost of health care services and patient welfare losses for VIC. Maintaining
438 lockdowns while waiting for the vaccination rate to reach a higher level would reduce health care
439 services costs and welfare losses. These avoided costs would be around 5.7 billion AUD if the
440 lockdowns could be extended from a 70% vaccination rate to the 80% vaccination rate and would be
441 2.1 billion from the 80% target to the 90% target.
442 Table 12. VIC: Estimates of health care and welfare losses at different vaccination rates for opening
443 up, billions AUD
Vacc Health care services Welfare losses Total
inati Non- ICU Total Recover Non- Total Health care
on ICU people recovered and welfare
rate losses
70% 0.2 0.2 0.4 9.4 5.6 15 15.4
[0.2-0.3] [0.2-0.2] [0.4-0.5] [8.2-10.7] [5-6.4] [13.2-17.1] [13.5-17.5]
80% 0.1 0.1 0.3 6 3.5 9.4 9.7
[0.1-0.2] [0.1-0.1] [0.2-0.3] [5.2-6.8] [3.2-3.8] [8.4-10.6] [8.7-10.9]
90% 0.1 0.1 0.2 4.7 2.7 7.4 7.6
[0.1-0.1] [0.1-0.1] [0.2-0.2] [4.1-5.3] [2.6-2.9] [6.6-8.3] [6.8-8.5]
444 Notes:
445 1. Only costs after 11/10/2021 are counted.
446 2. Outside brackets are the mean, inside brackets are the 95% CI. Numbers are rounded to the nearest 1-decimal
447 place.
448
449 Fig 6 summarises the costs of different vaccination rates for VIC. It shows that the preferred vaccination
450 rate that minimises total societal costs in the state is 93%.

17

Electronic copy available at: https://ssrn.com/abstract=3944437

451

452
453 Fig 6. VIC: Societal Costs of Different Lockdown ‘Opening Up’ Vaccination Rates
454 Notes:

455 1. All costs are counted from 11 October 2021.

456 2. Total costs = health care costs and welfare losses + economy lockdown costs

457 Table 13 compares vaccination rates (70%, 80% and 90%) for opening up with the preferred vaccination
458 rate for VIC (93%). Results in this table are converted to per-capita and percentage changes for
459 comparison purposes. Opening up VIC at the 70% vaccination rate (with low-level to moderate public
460 health restrictions) would incur additional health care costs and welfare losses of approximately 1,200
461 AUD/person, nearly five times the per capita cost savings from not having a lockdown of about 250
462 AUD/person, compared to the preferred vaccination rate of 93%. Thus, the net societal loss from
463 premature opening at 70% vaccination rate, rather than at the preferred rate of 93%, $962 AUD/person.
464 Identical calculations are also provided in Table 13 for an 80% (about 208 AUD/person loss) and 90%
465 (6 AUD/person loss) vaccination rate noting that the lowest total societal costs are at the preferred
466 vaccination rate of 93%.
467 Table 13. VIC: Per capita (AUD and %) cost differences of alternative vaccination rates to the
468 preferred vaccination rate (93%)
Health care Costs Economy
Total cost
Vaccination rate and Welfare Losses Lockdown Cost
70% +1216 (+110.9%) -254 (-62.3%) +962 (+63.9%)
80% +355 (+32.4%) -147 (-36%) +208 (+13.8%)
90% +46 (+4.2%) -40 (-9.8%) +6 (+0.4%)
469 Notes:
470 1. Numbers are per-capita deviations from the economically justifiable vaccination rate, '+' = higher and '-'
471 =lower
472 2. Outside brackets are the absolute value of the deviation rounded to the nearest $/person.
473 3. Inside brackets are the relative deviation in percentage rounded to the nearest 1-decimal place.
474

18

Electronic copy available at: https://ssrn.com/abstract=3944437

475 Tables 14 provides sensitivity analyses with respect to the effectiveness of reducing community
476 transmission (at 50%, 150% and 200% of the baseline) after opening up and the speed of vaccination
477 (at 110% and 120% of the baseline).
478
479 Table 14. VIC: Sensitivity of the preferred vaccination rate to the effectiveness of public health
480 measures after opening up and speed of vaccination.
Daily vaccination rate
(Baseline=45,000/day)
110%
Baseline baseline 120% baseline
Reduction in community 50% baseline 95% (11.6) 95% (11.5) 95% (11.2)
transmission after Baseline 93% (10.1) 93% (9.9) 94% (9.6)
opening up (Baseline
150% baseline 90% (9) 90% (8.7) 91% (8.5)
level =10%)
200% baseline 90% (8.1) 90% (7.8) 90% (7.5)
481 Notes:

482 1. Outside brackets are the economically justifiable vaccination target for removing the lockdown.
483 2. Inside brackets is the associated total cost from 11/10/2021 in billion dollars, rounded to the nearest 1-
484 decimal place.

485 Table 15 provides a sensitivity analysis with respect to the preferred vaccination rate in relation to the
486 effectiveness of the public health measures during lockdown (80% and 120% of the baseline) and state-
487 level lockdown costs (80% and 120% of baseline).
488 Table 15. VIC: Sensitivity of preferred vaccination rate to economy lockdown costs and rate of
489 community transmission during lockdown

Lockdown cost compared to low-level restriction cost

Baseline level = (2.35-0.65)x23.6% billion AUD/week
80% baseline Baseline 120% baseline
Daily community 80%
93% (9) 91% (9.5) 90% (10)
transmission during baseline
lockdown (Baseline level
Baseline 95% (9.5) 93% (10.1) 91% (10.6)
=14-day average prior to
11/10/2021) 120%
95% (10) 94% (10.6) 92% (11.2)
baseline
490 Notes:

491 1. Outside brackets are the economically justifiable vaccination target for removing the lockdown.
492 2. Inside brackets is the associated total cost from 10/11/2021 in billion dollars, rounded to the nearest 1-
493 decimal place.

494 4.4. Western Australia

495 The COVID-19 epidemic in WA has been different to NSW and VIC. As of 11 October 2021, WA had
496 no cases of community transmission with all active cases associated with arrivals from abroad or inter-
497 state in supervised quarantine. WA is one of the most isolated jurisdictions in the world and almost all
498 out-of-state arrivals come by air. This isolation, along with very strict protocols for entry and supervised
499 quarantine, has helped WA to maintain its zero-Covid status.
500 As of the model start date, WA did not have any significant restrictions within the state, i.e., people are
501 requested to check-in when entering venues and follow some restrictions in relation to age care if they
502 are symptomatic. Our estimates of the vaccination rate in WA to 11 October 2021 is 43%, with around

19

Electronic copy available at: https://ssrn.com/abstract=3944437

503 11,000 people becoming fully vaccinated per day. At this vaccination rate it would take WA about 43,
504 64, and 84 days to achieve the 70%, 80% and 90% vaccination rates.
505 When WA opens its borders, incoming passengers can be seeding even if they are required to be fully
506 vaccinated and tested because many fully vaccinated cases are asymptomatic [22, 23]. The
507 epidemiology outcomes depend on how the state would respond if an outbreak occurred. Two important
508 considerations of response to an outbreak are whether the state would close its border again (and if yes,
509 when) and whether the state would impose restrictions should community transmission occur (and if
510 yes, when).
511 We assume that the state would not open its borders if it expected to close its border again in a short
512 period of time. As the vaccination progress is foreseeable, the state only chooses to open early, before
513 the vaccination progress is complete, if it would not reverse the policy, at least until all eligible and
514 willing people are fully vaccinated. In addition, we assume that if an outbreak continues 90 days after
515 the vaccination progress is complete, the state will impose more restrictive health measures to reduce
516 community transmission.
517 The projected public health outcomes should WA opens its borders at 70%, 80% and 90% vaccination
518 rates, assuming that WA has no community transmission until it its state border fully opens up, are
519 provided in Fig 7 and Table 16 provides summary statistics. When its state border is opened, we assume
520 that interstate travel would return to its 2020/21 level (around 1,600 incoming passengers a day); all
521 incoming passengers would be fully vaccinated, and 0.5% of the passengers would be asymptomatically
522 infectious. The effectiveness of the current (and minimal) public health restrictions in WA until opening
523 up are able to reduce community transmission by 5%. As there is no available WA data for the Delta
524 variant, we assume the hospitalisation rate, ICU rate and mortality rate in the state when community
525 transmission occurs is the average of NSW and VIC.

526
527 Fig 7. WA: Projected public health outcomes after opening up
528 Notes:

529 1. Day zero is specified to be 11 October 2021.

530 2. The red, yellow, and green colours represent scenarios where WA border is opened after the 70%, 80%
531 and 90% targets, respectively.
532 3. The (red, yellow, and green) vertical lines represent the days when the 70%, 80% and 90% targets are
533 reached.
534 4. The (red, yellow, and green) curves represent the mean of all projection outcomes, and the bands
535 represent their 95% confidence intervals.
536 5. The black horizontal lines in panels (d) and (e) are the capacity of the health care system, i.e., staffed
537 ICU beds and ventilators (data sources reported in Appendix 2).

20

Electronic copy available at: https://ssrn.com/abstract=3944437

538 6. The red horizontal dashed lines in panels (d) and (e) are the capacity net the non-covid demand, which
539 is estimated via the occupation rates of ICU beds and ventilators in 2018/2019 (data sources reported in
540 Appendix 2).
541
542 Fig 7 shows that COVID-19 outbreaks in WA would increase at a lower growth rate than projected for
543 NSW and VIC. This is because when the state opens up there is no community transmission and the
544 vaccination rate is at least 70% for those aged 12 years and older. By contrast to NSW and VIC, Zero
545 (or very low) COVID-19 in WA, allows for better epidemic control without lockdowns [24, 25]. In all
546 three scenarios, however, outbreaks would continue after the vaccination progress is complete, and the
547 state would need to apply more restrictive health measures to control community transmission to try to
548 achieve zero or very low COVID.
549 Table 16. WA: Projected outcomes at different vaccination rates for opening up

Vaccination rates for opening up

70% 80% 90%
Cumulative cases (rounded to the 28.7 9.5 3.3
nearest 1000) [25.9-31.7] [8.7-10.4] [3.1-3.6]
Peak active cases (rounded to the 2.6 0.8 0.2
nearest 1000) [2.3-3] [0.7-0.8] [0.2-0.2]
104 31 8
Peak hospitalisation (people)
[92-116] [28-35] [7-9]
Peak ICU (people)
(Capacity= 190; Average non-Covid 25 8 2
demand=102; Net capacity=88) [22-29] [7-9] [2-2]
Peak ventilation (people)
(Capacity=518; Average non-Covid 14 4 1
demand=42; Net capacity=476) [12-15] [4-5] [1-1]
Cumulative fatalities (people) 154 54 21
[141-168] [50-60] [20-22]
550 Notes:

551 1. Outside brackets are the mean, inside brackets are the 95% CI. Numbers are rounded to the nearest 1-decimal
552 place.
553 2. ICU capacity is estimated at 190 beds, including 159 available, staffed ICU beds plus 31 additional staffed ICU beds
554 available. Average demand for ICU beds from non-COVID-19 patients is estimated at 102, i.e., ~37,300 bed days
555 in 2018/19 (data sources reported in Appendix 2). ICU net capacity (88 beds) is the difference between
556 the ICU capacity and non-COVID-19 demand.
557 3. Ventilation capacity is estimated at 518. Average demand for ventilators from non-COVID-19 patients is estimated
558 at 42 in 2018/19, i.e., ~43% of the total admission (data sources reported in Appendix 2). Ventilation net
559 capacity (476 ventilators) is the difference between the ventilation capacity and non-COVID-19
560 demand

561 We assume that WA’s state border closure reduces the ‘accommodation and food’ sector valued at
562 about $2 billion AUD per annum [26] by 50%, or equivalent to 40 million AUD per week. In Table 17,
563 if the WA border is opened at a 70% vaccination rate the average health care costs and welfare losses
564 would be 729 million AUD (95%CI~ [652-810]). If WA’s strict state border controls were maintained
565 until it reached an 80% vaccination rate, the health care costs, and welfare losses would reduce to 224
566 million AUD (95%CI~ [201-251]). If the state border controls were maintained until a 90% vaccination
567 rate, these costs would further reduce to 59 million AUD (95%CI~ [53-67]).
568 Table 17. WA: Projected health care costs and welfare losses at different vaccination rates for opening
569 up, millions AUD

Health care services Welfare losses

21

Electronic copy available at: https://ssrn.com/abstract=3944437

 Vacc Non- ICU Total Recover Non- Total Total
inati ICU people recovered Health care
on and welfare
rate losses
70% 21 18 40 366 324 690 729
[19-24] [16-21] [35-44] [323-410] [294-355] [617-765] [652-810]
80% 7 6 12 111 101 212 224
[6-7] [5-6] [11-14] [99-125] [91-112] [190-237] [201-251]
90% 2 2 3 29 27 56 59
[2-2] [1-2] [3-4] [26-33] [24-30] [50-63] [53-67]
570 Notes:
571 1. Only counting costs after 11 October 2021.
572 2. Outside brackets are the mean, inside brackets are the 95% CI. Numbers are rounded to the nearest million dollars.
573
574 Fig 8 summarises the projected costs of different vaccination rates for WA. It shows that the preferred
575 vaccination rate that minimises total societal costs in the state is 88%. If its state border remains closed
576 until a higher vaccination rate were reached, the cost of health care services and welfare loss would be
577 lower, but the border closure costs would be higher. At the current vaccination progress (11,000 people
578 becoming fully vaccinated per day), it would take WA approximately 43 days from 11 October 2021 to
579 reach the 70% vaccination target, and the state border closure cost during this period would be some
580 246 million AUD.
581

582
583 Fig 8. WA: Societal Costs of Different Vaccination Rates for State Border Reopening
584 Notes: Including costs that vary with vaccination targets, counted from 11/10/2021. Costs appliable to ALL
585 vaccination targets are not included (e.g., vaccination cost, the cost of maintaining the minimal restriction, the
586 cost of policy interventions after the vaccination target is achieved)

587 Table 18 compares three vaccination rates (70%, 80% and 90%) with the preferred vaccination rate
588 (88%) that minimises societal costs for the state. Opening the state border at a 70% vaccination rate
589 would incur additional health care costs and welfare losses of approximately 245 AUD/person, three-
590 fold larger than the cost saving of 79 AUD/person from not having a state border closure, compared to
591 the preferred vaccination rate (88%).

22

Electronic copy available at: https://ssrn.com/abstract=3944437

592 Table 18. WA: Per capita (AUD and %) cost differences of alternative vaccination rates to the
593 preferred vaccination rate (88%)
Health Care Costs State Border
Total cost
and Welfare Losses Costs
+166
70% +245 (+617.6%) -79 (-46.2%) (+78.5%)
80% +55 (+137.5%) -35 (-20.5%) +19 (+9.2%)
90% -8 (-19%) +9 (+5.1%) +1 (+0.6%)
594 Notes:
595 1. Numbers are per-capita deviations from the economically justifiable vaccination rate, '+' = higher and '-'
596 =lower
597 2. Outside brackets are the absolute value of the deviation rounded to the nearest $/person.
598 3. Inside brackets is the relative deviation in percentage rounded to the nearest 1-decimal place.
599
600
601 Table 19 reports the sensitivity analysis with respect to the speed of vaccination (100% and 120% or
602 baseline) rate and the effectiveness (50% and 150% of baseline) of public health measures after opening
603 up.
604
605 Table 19. WA: Sensitivity of preferred vaccination rate to the effectiveness of public health measures
606 after opening up at reducing community transmission and speed of vaccination.
Daily vaccination rate
(Baseline=11,000/day)
110% 120%
Baseline baseline baseline
Effectiveness of reduction 50% baseline 89% (577) 89% (525) 90% (484)
in community transmission
after opening up (Baseline Baseline 88% (562) 87% (511) 88% (473)
level =5%) 150% baseline 86% (546) 86% (497) 86% (460)
607 Notes: 1. Outside brackets are the economically justifiable vaccination target for opening the border.
608 2. Inside brackets are the associated total cost from 11 October 2021, rounded to the nearest million
609 dollars.

610 Table 20 provides a sensitivity analysis to border closure costs (80% and 120% of baseline state border
611 closure costs) and the probability of COVID-19 seeding in the community after the state border opens
612 up (150% and 200% of the baseline). Table 20 shows that the assumed state border closure costs have
613 a substantial impact on the preferred vaccination rate. In terms of the probability of COVID-19 seeding
614 from out-of-state arrivals, a much higher proportion of fully vaccinated arrivals could be asymptomatic
615 [22, 23, 27]. Thus, our baseline assumption that 0.5% of fully vaccinated arrivals are asymptomatic
616 after opening up of the state border is likely to be a ‘best case’. Consequently, we compared the effect
617 on the preferred vaccination rate of a higher probability of seeding (150% and 200% of the baseline)
618 and observed that the lower is the state border costs and the higher is the seeding rate of COVID-19
619 from arrivals, the higher is the preferred vaccination rate.
620 Table 20. WA: Sensitivity of preferred vaccination rate to opening up with the probability of seeding
621 and state border closure costs
Border closure costs
Baseline =40 million AUD/week
80% 120%
baseline Baseline baseline

23

Electronic copy available at: https://ssrn.com/abstract=3944437

 Baseline 90% (470) 88% (562) 86% (651)
Probability of seeding
(Baseline = 0.5% asymptomatic 150%
91% (498) 90% (595) 90% (691)
fully vaccinated passengers baseline
after opening border) 200%
94% (514) 92% (620) 90% (719)
baseline
622 Notes:

623 1. Outside brackets are the economically justifiable vaccination target for opening the border.
624 2. Inside brackets is the associated total cost from 11/10/2021, rounded to the nearest million dollars.

625 5. Discussion
626 Our results provide a number of important insights for decision-makers when determining the
627 vaccination rate that minimises total societal costs from opening up an economy that is in lockdown to
628 reduce community transmission of COVID-19 or has imposed strict border controls. First, our projected
629 public health outcomes are comparable, adjusted for differences between state and the national
630 population, to the high seeding in [28, p:8-10], with baseline public health and safety measures and
631 partially effective testing, tracing, isolation and quarantine measures scenario. Our results are consistent
632 with both the Doherty Institute and the Grattan Institute [7] results that find that the public health
633 outcomes from opening up at a 80% vaccination rate are much better than at a 70% vaccination rate.
634 Second, we provide a method to calculate the preferred vaccination rate to open up a locked down
635 economy with community transmission (NSW and VIC) and show that this preferred vaccination rate
636 depends on several key factors (see Fig 9). We find that the preferred population vaccination rate is
637 increasing with: (1) the speed of vaccination; (2) the lower is the daily lockdown cost; (3) the larger are
638 the public health costs from COVID-19; (4) the higher is the rate of community transmission before
639 opening up; and (5) the less effective are the public health measures after opening up.

640
641 Fig 9. Summary of the sensitivity of the economically justifiable vaccination target

24

Electronic copy available at: https://ssrn.com/abstract=3944437

642 Third, in the case of a state with no community transmission (WA), we find the preferred vaccination
643 rate is increasing in: (1) the speed of vaccination; (2) the less effective are the public health measures
644 after opening up; (3) the lower are border closure costs; and (4) the greater is the probability of COVID-
645 19 seeding into the community from out-of-state arrivals.
646 Fourth, the costs of opening up too soon are asymmetric such that the losses from opening prematurely
647 are greater than the losses from opening too late for the same percentile error difference in the preferred
648 vaccination rate. In the context of Australia, we also find that opening up prematurely, for example at a
649 70% vaccination rate, imposes substantial per capita costs estimated at 135 AUD (NSW), 962 AUD
650 (VIC) and 166 AUD (WA). Importantly, we project that the combined demand for staffed ICU beds by
651 COVID-19 and non-COVID-19 patients would likely exceed the state-level current and surge capacity
652 at 70% vaccination rate for both NSW and VIC and may also exceed the state-level capacity of VIC at
653 an 80% vaccination rate..
654 Fifth, our estimated preferred vaccination rates for NSW (85%), VIC (93%) and WA (88%) all exceed
655 the Phase C vaccination targets (16 years and older) under Australia’s National Plan to transition the
656 response to COVID-19 [2]. This finding is consistent with the Grattan Institute’s recommendation,
657 based on projected public health outcomes, to fully vaccinate 80% of all Australians (and 95% of those
658 70 years and older) before opening up the international borders and no longer using lockdowns [7].
659 Under Australia’s National Plan, Phase B commences when 70% of those 16 years and older are fully
660 vaccinated. Phase B coincides with eased restrictions on vaccinated persons and has as a key goal to
661 minimise on-going community cases with low-level restrictions. Phase C of the National Plan
662 commences when 80% of those 16 years and older are fully vaccinated. Under Phase C, fully vaccinated
663 people would be exempt from travel restrictions and a key goal would be to have only highly targeted
664 lockdowns. The differences between our estimated preferred vaccination rates (NSW 85%, VIC 93%,
665 and WA 88%) and the 70% and 80% targets in the National Plan are more pronounced than they appear
666 because a 70% and 80% vaccination of those 12 years and older is ‘equivalent’ to a 74% and 85%
667 vaccination rate of those 16 years and older. Importantly, our preferred state-level vaccination rates
668 would be higher again should there be vulnerable populations, with high probabilities of morbidity and
669 mortality from COVID-19 independent of age, and if these sub-populations were vaccinated at less than
670 the preferred state-level average vaccination rate.
671 While our model and results were developed for three Australian states there are, nevertheless, of ex-
672 ante relevance to other jurisdictions in Australia, such as Tasmania and the Northern Territory, and
673 elsewhere, such as New Zealand [29], which have yet to either open up their respective borders and/or
674 open up their locked down economies. Our approach to estimating total societal costs and the preferred
675 vaccination rate also allows for an ex-post assessment of decisions of other countries that have already
676 opened up. For instance, the United Kingdom opened up on the 19 July 2021 when 67% of its population
677 16 years and older was vaccinated while Denmark opened up on 10 September, 2021 when 80% of its
678 population 16 years and older was vaccinated [30].

679 6. Conclusions
680 As countries transition in their responses to the COVID-19 pandemic through vaccinations, a
681 fundamental question is: What is the preferred vaccination rate that minimises societal costs for opening
682 up public health measures or state border controls? Using a separate age-structured SIRM model for
683 three jurisdictions in Australia (New South Wales, Victoria and Western Australia), we estimated a
684 preferred vaccination rate for opening up, by state, that minimises the sum of health care costs, welfare
685 losses from fatalities and those recovering from COVID-19, and economy lockdown costs and/or state
686 border control costs.

25

Electronic copy available at: https://ssrn.com/abstract=3944437

687 Our results show that the target vaccination rates under Australia’s National Plan to transition its
688 COVID-19 response are lower than the vaccination rate we estimate that would minimise state-level
689 societal costs. We also find that opening up at a lower than preferred vaccination rate, such as 70%
690 under Phase B of the National Plan, would impose substantial per capita societal costs. In the states of
691 New South Wales and Victoria, opening up at a 70% vaccination rate the projected ICU patient demand
692 is expected to exceed the available state-level staffed ICU bed capacity.
693 The methods we used to estimate a preferred vaccination rate can be applied to any jurisdiction where
694 there are available data. Our results also provide useful guidance as to the qualitative effects of different
695 factors, such as the speed of vaccination, the effectiveness of low-level public health measures, among
696 others, on the preferred vaccination rate that minimises total societal costs. In sum, our modelling
697 provides a valuable tool that can used by decision makers to help determine the cost-minimising
698 vaccination rate to open up an economy in lockdown and/or with strict border controls on arrivals.

699

26

Electronic copy available at: https://ssrn.com/abstract=3944437

700 References
701 1. Cockburn, P. How NSW's COVID-19 outbreak grew from one limo driver to more than 300
702 cases. 2021 [cited 2021 7 July 2021]; Available from: https://www.abc.net.au/news/2021-07-
703 07/nsw-covid-19-outbreak-explained/100271128.
704 2. Department of Prime Minister and Cabinet National Plan to transition Australia’s National
705 COVID-19 Response. 2021, Australia Government.
706 3. Abeysuriya, R., D. Delport, R. Sacks-Davis, M. Hellard, and N. Scott, COVID-19
707 mathematical modelling of the Victoria roadmap 2021. 2021, Burnet Institute.
708 4. Doherty Institute, Doherty Modelling Report for National Cabinet 30 July 2021. 2021.
709 5. Milne, G.J., J. Carrivick, and D. Whyatt, Non-pharmaceutical interventions and vaccinating
710 school children required to contain SARS-CoV-2 Delta variant outbreaks in Australia: a
711 modelling analysis. medRxiv, 2021: p. 2021.10.03.21264492.
712 6. Blakely, T., et al., Determining the optimal COVID-19 policy response using agent-based
713 modelling linked to health and cost modelling: Case study for Victoria, Australia. medRxiv,
714 2021: p. 2021.01.11.21249630.
715 7. Duckett, S., et al., Race to 80: our best shot at living with COVID. 2021, Grattan Institute:
716 ISBN: 978-0-6450879-7-0.
717 8. Australian Treasury, National Plan to Transition to Australia's National COVID-19
718 Response: Economic Impact Analysis, ed. The Treasury - Australian Government. 2021,
719 Australia.
720 9. Kompas, T., R.Q. Grafton, T.N. Che, L. Chu, and J. Camac, Health and economic costs of
721 early and delayed suppression and the unmitigated spread of COVID-19: The case of
722 Australia. PLoS One, 2021. 16(6): p. e0252400.
723 10. Department of Prime Minister and Cabinet Best Practice Regulation Guidance Note Value of
724 statistical life. 2021, Australia Government.
725 11. Liu, Y. and J. Rocklöv, The reproductive number of the Delta variant of SARS-CoV-2 is far
726 higher compared to the ancestral SARS-CoV-2 virus. Journal of Travel Medicine, 2021.
727 12. Raveendran, A.V., R. Jayadevan, and S. Sashidharan, Long COVID: An overview. Diabetes &
728 metabolic syndrome, 2021. 15(3): p. 869-875.
729 13. Blakeley, T., et al., Determining the optimal COVID-19 policy response using agent-based
730 modelling linked to health and cost modelling: Case study for Victoria, Australia. MedRxiv
731 pre-print 2021.
732 14. Edwards, B., N. Biddle, M. Gray, and K. Sollis, COVID-19 vaccine hesitancy and resistance:
733 Correlates in a nationally representative longitudinal survey of the Australian population.
734 PLOS ONE, 2021. 16(3): p. e0248892.
735 15. Melbourne Institute. Vaccine Hesitancy Tracker. 2021 [cited 2021 12/10/2021]; Available
736 from: https://melbourneinstitute.unimelb.edu.au/publications/research-
737 insights/ttpn/vaccination-report.
738 16. Meyerowitz-Katz, G., et al., Is the cure really worse than the disease? The health impacts of
739 lockdowns during COVID-19. BMJ Global Health, 2021. 6(8): p. e006653.
740 17. Miranda, M.J. and P.L. Fackler, Applied Computational Economics and Finance. 2002: MIT
741 Press.
742 18. Kenneth, L.J., Numerical Methods in Economics. 1998: MIT Press.
743 19. Schmelz, K. and S. Bowles, Overcoming COVID-19 vaccination resistance when alternative
744 policies affect the dynamics of conformism, social norms, and crowding out. Proceedings of
745 the National Academy of Sciences, 2021. 118(25): p. e2104912118.
746 20. Litton, E., et al., Increasing ICU capacity to accommodate higher demand during the
747 COVID-19 pandemic. Medical Journal of Australia, 2021.
748 21. AIHW, The first year of COVID-19 in Australia: direct and indirect health effects. . 2021,
749 Australian Insititute of Health and Welfare: ISBN: 978-1-76054-865-0.
750 22. Alene, M., et al., Magnitude of asymptomatic COVID-19 cases throughout the course of
751 infection: A systematic review and meta-analysis. PLOS ONE, 2021. 16(3): p. e0249090.
752 23. Sah, P., et al., Asymptomatic SARS-CoV-2 infection: A systematic review and meta-analysis.
753 Proceedings of the National Academy of Sciences, 2021. 118(34): p. e2109229118.

27

Electronic copy available at: https://ssrn.com/abstract=3944437

754 24. Contreras, S., et al., Low case numbers enable long-term stable pandemic control without
755 lockdowns. Science Advances, 2021. 7(41): p. eabg2243.
756 25. Shonchoy, A.S., K.S. Ishtiaq, S. Talukder, N.U. Ahmed, and R. Chowdhury, A novel index-
757 based decision support toolkit for safe reopening following a generalized lockdown in low
758 and middle-income countries. Scientific Reports, 2021. 11(1): p. 14108.
759 26. Government of Western Australia, WESTERN AUSTRALIA ECONOMIC PROFILE –
760 September 2021. 2021, Government of Western Australia, Department of Jobs, Tourism,
761 Science and Innovation.
762 27. Mendez, R. Most fully vaccinated people who get Covid delta infections are asymptomatic,
763 WHO says. 2021 [cited 2021 12/7/2021]; Available from:
764 https://www.cnbc.com/2021/07/12/most-fully-vaccinated-people-who-get-covid-delta-
765 infections-are-asymptomatic-who-says-.html.
766 28. Doherty Institute, Doherty Modelling Interim Report to National Cabinet, 17 September
767 2021. 2021.
768 29. Skegg, D., et al. Strategic COVID-19 Public Health Advisory Group. Advice to Minister
769 Verrall. 2021 [cited 2021 7/10/2021]; Available from:
770 https://covid19.govt.nz/assets/reports/Independent-Advisory-Groups/Strategic-COVID-19-
771 Public-Health-Advisory-Group-Advice-to-Minister-Verrall-June-2021.pdf.
772 30. Byrne, B. Freedom Day: Denmark lifted all COVID restrictions. 2021 [cited 2021
773 10/9/2021]; Available from: https://insiderpaper.com/freedom-day-denmark-lifts-all-covid-
774 restrictions/.
775 31. ANZICS, Intensive Care Resources and Activity Report 2018/19. 2020, Australian and New
776 Zealand Intensive Care Society.

777

28

Electronic copy available at: https://ssrn.com/abstract=3944437

778 Appendices
779 Appendix 1: Data-matched age distributions of hospitalisation, ICU
780 admission, ventilation, and fatality rates
781 NSW
Age Hospitalisation ICU Ventilation Fatality
group admission
0-9 2.552E-03 1.469E-04 7.160E-05 6.419E-05
10-19 7.229E-03 3.276E-04 1.597E-04 6.419E-05
20-29 3.375E-02 1.836E-03 8.950E-04 3.370E-04
30-39 9.257E-02 5.288E-03 2.578E-03 1.075E-03
40-49 1.307E-01 1.097E-02 5.346E-03 3.274E-03
50-59 3.038E-01 4.702E-02 2.293E-02 8.826E-03
60-69 5.021E-01 1.737E-01 8.468E-02 2.535E-02
70-79 7.337E-01 2.975E-01 1.450E-01 7.932E-02
80-89 7.976E-01 7.581E-02 3.696E-02 2.540E-01
90+ 7.976E-01 7.581E-02 3.696E-02 4.599E-01

782 Victoria
Age ICU
Hospitalisation Ventilation Fatality
group admission
0-9 8.265E-04 5.159E-05 3.393E-05 4.996E-05
10-19 2.341E-03 1.151E-04 7.568E-05 4.996E-05
20-29 1.093E-02 6.449E-04 4.242E-04 2.623E-04
30-39 2.998E-02 1.857E-03 1.222E-03 8.369E-04
40-49 4.233E-02 3.852E-03 2.534E-03 2.548E-03
50-59 9.838E-02 1.652E-02 1.086E-02 6.870E-03
60-69 1.626E-01 6.101E-02 4.013E-02 1.974E-02
70-79 2.376E-01 1.045E-01 6.873E-02 6.174E-02
80-89 2.583E-01 2.663E-02 1.752E-02 1.977E-01
90+ 2.583E-01 2.663E-02 1.752E-02 3.580E-01

783 Western Australia (the average of NSW and Victoria)

Age ICU
Hospitalisation Ventilation Fatality
group admission
0-9 1.689E-03 9.923E-05 5.277E-05 5.708E-05
10-19 4.785E-03 2.213E-04 1.177E-04 5.708E-05
20-29 2.234E-02 1.240E-03 6.596E-04 2.996E-04
30-39 6.127E-02 3.572E-03 1.900E-03 9.560E-04
40-49 8.652E-02 7.409E-03 3.940E-03 2.911E-03
50-59 2.011E-01 3.177E-02 1.689E-02 7.848E-03
60-69 3.324E-01 1.174E-01 6.241E-02 2.254E-02
70-79 4.856E-01 2.010E-01 1.069E-01 7.053E-02
80-89 5.280E-01 5.122E-02 2.724E-02 2.259E-01
90+ 5.280E-01 5.122E-02 2.724E-02 4.090E-01

29

Electronic copy available at: https://ssrn.com/abstract=3944437

784 Appendix 2: Data sources
785
786 1. Daily data for total COVID19 cases, active, recovery, fatalities, fully vaccinated:
787 https://www.covid19data.com.au (assessed 15/10/2021),
788 2. Daily data for hospitalisation, ICU admission, and ventilation requirements:
789 https://covidlive.com.au/states-and-territories (assessed 15/10/2021)
790 3. Staffed ICU and ventilator capacity [20: box 1 - column 4, box 4 - column 5, box 3 - column
791 6]
792 4. Non-COVID19 (2018/2019) ICU bed days and ventilation rates: [31]
793 5. Population data: https://www.abs.gov.au/statistics/people/population/national-state-and-
794 territory-population/mar-2021/31010do002_202103.xls
795
796

30

Electronic copy available at: https://ssrn.com/abstract=3944437