彼得森经济研究所-加速接种新冠疫苗的经济成本和效益（英文）-2021.5-16正式版.ppt

POLICY BRIEF21-11 Economic Costs andBenefits of AcceleratedCOVID-19 V accinationsJoseph E.Gagnon,Steven Kamin,and John KearnsM ay 2021Note:The authors thank Olivier Blanchard,Chad Bown,Jim Capretta,Monica deB olle,Jacob Kirkegaard,and M arcus Noland for helpful comments and suggestions.A ll errors and opinions are the authors own and do not necessarily reflect the viewsof the American Enterprise Institute,the Peterson Institute,or other members of theirmanagement and research staf f.I.INTRODUCTIONJust as COVID-1 9 produced a global economic shock,the COVID-1 9 vaccinationprograms are generally understood to be a prerequisite for a return to normalcyin our social and economic lives.Emergency measures to research,test,produce,and distribute vaccines have been expensive,but increases in GDP resulting fromthe vaccines are expected to exceed those costs by wide margins.Few studieshave quantified the economic costs and benefits of dif ferent rates of COVID-1 9vaccination,however.This P olicy Brief focuses on developing such a quantitativeassessment for the United States;the approach may be applied to othercountries as well.Joseph E.Gagnon is seniorfellow at the PetersonInstitute for InternationalEconomics.Steven Kamin is residentscholar at the AmericanEnterprise Institute.John Kearns is researchassistant at the AmericanEnterprise Institute.T wo illustrative scenarios support the conclusion that most plausible optionsto accelerate vaccinations would have economic benefits that far exceed theircosts,in addition to their more important accomplishment of saving lives.ThisP olicy Brief shows that if,for example,the United States had adopted a moreaggressive policy in 2020 of unconditional contracts with vaccine producers,the up-front cost would have been higher but thousands of lives would havebeen saved and economic growth would have been stronger.Instead,thefederal government conditioned its contracts on the vaccines being proven safeand ef fective.The projections presented here show that even if unconditionalcontracts led to support for vaccines that failed the phase III trial and ultimatelywere not used,the cost would have been worth it.1 750 M assachusetts A venue,NW|W ashington,DC 200 36-1 90 3 USA|+1.202.32 8.9000|PB 21-11|MAY 2021 2The few studies that have addressed this issue find that vaccinations lead tovery substantial increases in GDP relative to their cost.Juan Castillo and colleagues(2021)calculate that a baseline path of global vaccinations,in which it would taketwo years to vaccinate 70 percent of the worlds population,would boost globalGDP by$8.7 trillion compared with an alternative without vaccinations;a fasterpace of vaccination that hastens the end of the pandemic by 10 months wouldlead to an additional$970 billion in world GDP.But this study relies on outdated(mid-2020)estimates of the effect of the pandemic on global GDP.It also doesnot account for adjustments by households and businesses that have moderatedthe economic effect of the disease.Cem akmakli and colleagues(2021)do notcalculate the economic benefits of vaccination per se,but they find that failureto vaccinate the worlds developing economies could cost the world between$1.5 trillion and$6.1 trillion;impacts on the United States alone could be as highas$670 billion.Their analysis for the most part relies on a calibrated generalequilibrium model,and it is unclear how well it mirrors real-world behavior.Our study differs from this and other research in a number of ways.First,ourestimates of the economic effects of the pandemic are based on an empirical modelof the responses of GDP to COVID-19 deaths and lockdown regulations.Second,wefocus primarily on the effects of vaccination on the US economy.And,finally,ratherthan attempting to quantify the economic effects of an all-or-nothing vaccinationeffort,we consider the effects of faster or slower paths of vaccination,comparedwith a baseline path projected by public health experts.This narrower focus is moreuseful because it narrows the spotlight to allocation decisions that can yet be madeby policymakers or that could be made in a future pandemic.It should also be morerealistic,in that moving from no vaccination program to a rapid program wouldlikely induce substantial changes in economic and personal behavior that would notoccur with incremental differences in the pace of vaccination.As previous studies have found,vaccinations are indisputably cost-efficient,but owing in part to the more marginal changes in the pace of vaccination underconsideration here the effects are less dramatic.For example,one scenarioassumes that,prior to the launch of phase III vaccine trials last year,the federalgovernment had signed unconditional contracts to purchase vaccines from theproducers instead of the conditional contracts it actually used.The unconditionalcontracts would have exposed the government to the risk of paying for somevaccines that were not effectivebut it would have enabled producers to beginlarge-scale production three to four months earlier than they actually did,withthe immediate distribution of 4 million doses of vaccine per day on the grantingof Emergency Use Authorization,saving 35,000 lives and boosting 2021 GDP by$64 billion.The estimated incremental cost of this scenario is no more than$20billion and potentially much less.The fiscal effects alone(higher taxes and lessentitlement spending)would exceed the cost of faster vaccinations by at least$12 billion,meaning that there would actually be no net cost to the governmentof accelerating vaccinations in this manner.Thus,all told,the approach toprocurement specified in this scenario not only saves many lives but also raisesUS GDP and reduces the fiscal deficit.We also examine a slower-than-baseline scenario,based on the pace ofvaccinations that has occurred and is projected for the European Union.Thisleads to 16,000 more deaths and$36 billion in lower GDP,compared with directcost savings of only around$1 billion to$2 billion and a net increase in thePB 21-11|MAY 2021 3fiscal deficit of at least$16 billion.We believe other plausible scenarios in whichvaccines are accelerated or retarded would have similar cost-benefit results.Although our methodology is more empirically grounded than other studies,it is still subject to a large range of uncertainty.Nevertheless,we find widemargins of benefits over costs despite conservative assumptions.Accordingly,itis clear that policymakers should be searching much more intensely for ways toaccelerate vaccinations both now and in future pandemics.Saving lives is surelythe most important reason to do so,but it is comforting to know that there arelarge net economic benefits.Our focus is on the United States,but our empirical estimates also applyto other advanced economies.We leave the analysis of developing economiesfor future work,but we are confident that benefits exceed costs there as wellespecially if the less expensive vaccines are used.II.ALTERNATIVE VACCINATION PATHS AND HEALTH OUTCOMESOur baseline vaccination path is that assumed by the Institute for Health Metricsand Evaluation(IHME)as of April 19,2021.1 Baseline:The pace of daily vaccinations ramps up sharply in April and thenreturns to near zero by late May.IHME assumes that 175 million Americanadults get at least one shot and 155 million are“effectively”vaccinated bysummer,after which no more vaccines are sought by the US population.Thisassumption is based on surveys of peoples willingness to accept the vaccine(about 69percent of adult Americans would have received at least one shot,which IHME assumes is sufficient to drive COVID-19 deaths to minimal levels).We consider two alternatives to the baseline path.The“preproduction”path assumes that,during the summer of 2020,the federal government hadcommitted with the various vaccine producers to make unconditional purchasesof their vaccines as soon as they were ready to start phase III trials(whichinvolve testing on large numbers of subjects),but before those trials had provensuccessful and garnered approval by the Food and Drug Administration(FDA).By the start of the phase III trials,the final composition of the vaccine was settledand companies could have begun mass production.For Moderna and Pfizer,thetwo companies providing the vast majority of vaccines for Americans,productioncould have started in August 2020 instead of November 2020.21 IHME projections are available at www.healthdata.org/covid/updates.Detailed data onvaccination history can be found at https:/ourworldindata.org/covid-vaccinations.2 The US government contracted to purchase 100 million doses of Pfizers vaccine on July 22,2020,and 100 million doses of Modernas vaccine on August 11,2020,but these contractswere contingent on a successful phase III trial and FDA approval.Moderna and Pfizer beganto ramp up large-scale production in November,ahead of FDA approval,based on positivetrial results.In our preproduction scenario,Moderna and Pfizer would have begun the ramp-up in August and would have had enough doses on hand by the middle of December todistribute 4 million doses a day indefinitely.See U.S.Government Engages Pfizer to ProduceMillions of Doses of COVID-19 Vaccine,HHS Press Office,July 22,2020;Pfizer Update On OurU.S.COVID-19 Vaccine Candidate Distribution Preparedness,Pfizer press release,November16,2020;and Moderna Reports Third Quarter 2020 Financial Results and Provides BusinessUpdates,Moderna press release,October 29,2020.Johnson&Johnsons vaccine developmentwas somewhat behind those of Moderna and Pfizer,but it too waited for FDA approval beforelarge-scale production.See Johnson&Johnson confident in 1B dose goal for COVID-19 vaccinenext year,looking ahead to 2022,Fierce Pharma,November 12,2020.PB 21-11|MAY 2021 4This arrangement differs from the baseline(that is,what actually transpired)in that the actual contracts signed by the government in 2020 were conditionalon a successful phase III trial and FDA approval.Although the vaccine companiesstarted to secure facilities and line up suppliers immediately after the contractswere signed,they waited until after phase III results became available to startlarge-scale production.As a consequence,relatively little vaccine had beenproduced at the time of FDA approval.In the preproduction scenario,total purchases through the end of 2021 wouldbe for as many as four times the number of doses assumed in the baseline path.The excess purchases would be made in anticipation that some of the vaccineswould fail or be markedly less successful in phase III trials.But insofar as everyproducer would have been paid for a contracted number of doses,whether ornot they were successful,the companies would have ramped up production3capacity in advance of the trials.begin immediate large-scale distribution of vaccines with FDA Emergency UseAuthorization.We assume the federal government,coordinating with state and4Supplies would thus have been available to5local authorities as well as the companies,would have set up the distributionsystem in advance of vaccine approvals and production.To provide a sense of the range of plausible vaccination paths and theirconsequences,we also consider a slower-than-baseline path that mimics theexpected trajectory of vaccinations in the European Union(scaled in proportionto the US adult population),based on the latest IHME projections.The EuropeanUnion was less aggressive than the United States in securing vaccines during2020.Figure1 displays the baseline and two alternative vaccination paths.Preproduction:This path assumes that 4 million doses are administered eachday starting at the beginning of baseline vaccinations(December 16,2020)and continuing until mid-March 2021.The same share of the population isultimately covered as in the baseline path,but coverage is achieved morethan two months earlier.European:Vaccinations in this path start two weeks later than baseline andramp up more slowly,reaching the same plateau as in the baseline by thebeginning of July 2021.This path is based on actual EU data through April619 and IHME projections thereafter.The path is scaled in proportion to the USadult population.3 The number of doses actually contracted by the federal government is nearly four times asmany as assumed in the baseline path.It is not clear how many will be delivered in time to beuseful for Americans(or delivered at all because they may not receive FDA approval),but thegovernment is likely to be able to sell or donate them for use abroad.Thus,excess doses in thepreproduction path may not have a very high net cost.45In the event that a producers vaccine does not prove effective,the government and theproducer could bargain for a reduction in the number of doses produced in exchange for somesharing of the cost savings.Although production would have begun three to four months earlier than otherwise,it is likelythat cumulative production by mid-December 2020 would not have equaled actual productionas of late March 2021,owing to bottlenecks and congestion effects.To distribute 4 milliondoses a day starting in mid-December,as assumed in the preproduction scenario,cumulativeproduction would have needed to be accelerated by only two months.See Moderna And PfizerOn Track To Hit Vaccine Production Goals,NPR,March 11,2021.6 IHME assumes a slightly higher plateau in Europe based on surveys of vaccine acceptance.Weapply the IHME American plateau in this scenario,as our intent is to gauge what would happenin the United States had the government chosen a slower path of vaccinations.PB 21-11|MAY 2021 5Figure 1Path of vaccinations under baseline and alternative scenarios,December 2020December 2021cumulative doses administered(millions)300200100April 200December2020February2021April2021June2021August2021October2021December2021BaselinePreproductionEuropeanNote:Projections for the baseline scenario start on April 20,2021.Sources:Institute for Health Metrics and Evaluation(IHME);authors calculations.The baseline path of deaths is from the IHMEs reference scenario and isconsistent with the baseline path of vaccination.The IHME projects deaths todecline steadily to 179 per day by August 1,2021.This path assumes that nopeople are vaccinated after late May.We assume that the decline in deathsfrom late May to late July reflects a lag in the buildup of immunity followingvaccination plus a lag between infection and death but that these lags are mostlyplayed out by August 1.Thus,we extrapolate deaths at the August 1 rate throughthe end of the year.The results are not sensitive to different assumptions aboutdeaths in the second half of the year because all the differences in the alternativescenarios occur before August 1.For each of the alternative paths of vaccinations,we adjust daily deaths basedon the difference between baseline and alternative vaccination rates.We assumethat each vaccine dose reduces the risk of death by 47.5percent,based on thetwo-dose efficacy of the Pfizer and Moderna vaccin