COVID-19 Economic Impact And Recovery Framework (working .
COVID-19Economic Impact andRecovery Framework(working paper)1
Table of Contents1. CONTEXT AND APPROACH . 32. ECONOMIC IMPACT MODEL . 42.1. Magnitude of Response . 42.2. Industry Susceptibility . 52.3. Response Scenario Timeline . 62.4. GDP Forecast . 73. SOCIAL CONTACT AS THE NEW SCARCE RESOURCE . 84. DETERMINING THE SOCIAL CONTACT BUDGET . 105. SPENDING SOCIAL CONTACT BUDGET . 105.1. Maximizing GDP . 105.2. Additional Recovery Objectives. 136. INCREASING SOCIAL CONTACT BUDGET . 146.1. Increasing Social Contact Budget . 146.2. The Importance of a Conservative Approach . 157. CONCLUSION . 168. APPENDIX: ECONOMIC IMPACT MODEL. 178.1. Approach to Economic Recovery and Industry Scarring . 178.2. GDP Forecast . 189. APPENDIX: RECOVERY POLICY . 209.1. Alternative Views for R Values . 209.2. Consumer Spending Recovery Curve . 212
Rolf Harms, Greg Detwiler, David Morrell, Diana Wang, Jenny YeCorrespondence: rolf.harms@microsoft.comIn this paper we outline an economic model that quantifies the impact of government and self-initiatedresponses to COVID-19. We also created an app that shows regularly updated estimates and allows a user todefine custom scenarios and then calculates GDP impact. Policymakers are focused on shaping GDPoutcomes, not merely predicting them. Based on our work with epidemiology teams and policymakers (aswell as our own economic modeling), we outline a policy framework to help governments think throughrecovery and reopening strategy.1. CONTEXT AND APPROACHCOVID-19 requires policymakers to strike a careful balance between managing infection curves andeconomic recovery. The former is informed by current data on infections as well as modelled projectionsof infection and disease transmission rates. The latter is informed by estimates of the economic impact ofgovernment mitigation measures as well as the impact of self-initiated changes in human behavior (e.g.,fear of going to stores). Keeping the economy closed for a long time will cause significant economicdamage. Insufficient social distancing may cause a resurgence that will cause more deaths and moreeconomic damage. This is a unique dilemma that requires a combined perspective from epidemiologymodels and economic models.At Microsoft and GitHub, we are notepidemiology experts, but we dounderstand the power of software, theability of the cloud to solve complexproblems, and the value in platforms thatempower others. In an effort to help, wehave been assisting some of the leadingepidemiology teams with open sourcingtheir code, running and calibrating theirmodels in the cloud, and building out UserInterfaces.This work has given us a betterappreciation of the epidemiology aspects. Separately we have been building an economic model toestimate GDP impact by industry, using our understanding of the disease models to layer in differentscenarios. Given the unusual nature of this economic disruption, we have applied some of the sametechniques we used in assessing past technological disruptions such as our “economics of the cloud”work.1In talking to policymakers in the U.S. and U.K., it is clear the lines between the epidemiology andeconomic impact work are blurring. Today the epidemiology models and economic models tend to live inisolation. This has led to a debate between healthcare outcomes and economic impact. We do not take aposition in this debate. We argue that to some degree this is a false trade-off: There are manyopportunities to improve the outcome for one without compromising the other.1This modeling work helped create the case for Microsoft’s investment in cloud computing about a decade ago.3
In this paper, we bridge the experience gained from working with epidemiology teams as well as our owneconomic modeling to outline a policy framework (and data) to support policymakers in this difficult time.In Section 2 we provide an overview of our economic impact model. In Section 3 we introduce the notionof social contact as a new scarce economic resource. In Section 4 we discuss the notion of a “SocialContact Budget.” Section 5 outlines a data-driven approach on how to best spend it and achieve thehighest “Return on Social Contact.” Finally, in Section 6 we frame up how the budget can be expandedover time and show data that quantifies the recovery.2. ECONOMIC IMPACT MODELThis section will provide a summary of the economic model we have developed that quantifies the impactto U.S. GDP under several scenarios. Although this paper focuses on U.S. numbers, our model covers the15 largest economies in the world representing 75% of global GDP, all of which will eventually be madeavailable in our Power BI Dashboard (U.S., Canada, U.K., Italy, and Australia currently available). We willbriefly highlight our approach below with more details provided in the appendix.The model has three key modules:1) The magnitude of response as measured by changes in people’s mobility/social contact, whetherdriven by government mitigation measures or by self-initiated behavior change2) The degree to which industries are susceptible to disruption from government intervention andself-initiated behavior change3) The duration over which we expect government interventions and behavior change to last2.1 Magnitude of ResponseFor our model we use a continuous scale for magnitude of response. A country’s current and historicallevel of response is determined using Google’s publicly released cell phone mobility data 2, which showsthe amount of foot traffic in key areas (e.g., stores, transit, work) relative to normal over time and bygeography. We initially used the Oxford Stringency Index data3 but found mobility data to be a better fitfor our use case as it captures the impact of government interventions as well as compliance andvoluntary measures taken by individuals. In general, we see five levels of response based on the variationin interventions seen worldwide (Fig. 2).Google LLC "Google COVID-19 Community Mobility Reports". https://www.google.com/covid19/mobility/ Accessed: Jun8, 2020.3Hale, Thomas, Sam Webster, Anna Petherick, Toby Phillips, and Beatriz Kira (2020). Oxford COVID-19 GovernmentResponse Tracker, Blavatnik School of Government.24
2.2 Industry SusceptibilityWe have assessed industries based on their susceptibility to social distancing and lockdown. To do this,we have studied data from 15 countries and used it to build a data cube of over 1,000 individualeconomic impact datapoints4 that cross references economic impact in 30 industries with the magnitudeof response in place at the time. To the extent possible we rely on official economic reporting data buthave also incorporated more real-time or high-frequency data sources (e.g., consumer spending, shippingvolumes). Using this data, we determine the degree of impact for each industry under different degreesresponse. Figure 3 shows a sample of industries.We then aggregate the economic impact across industries based on each industry’s contribution to U.S.GDP. Figure 4 shows the high-level breakdown of U.S. GDP by industry (value added) that we used forweighting purposes in this rollup, with rough color coding to indicate the relative impact to industriesunder a ‘medium’ level of response (i.e., what U.S. mobility looked like on average in May).4Sources: U.S. Bureau of Economic Analysis; U.S. Census Bureau; U.S. Department of Agriculture; U.S. Department ofEducation; Board of Governors of the Federal Reserve System (US), Industrial Production Index [INDPRO], retrieved fromFRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/INDPRO, May 5, 2020; Ministry of Trade,Industry and Energy (Korea); Statistics Korea; El Ministerio de Trabajo y Economía Social (Spain); Instituto Nacional deEstadística (Spain) Own compilation with data taken from the INE website: www.ine.es; Office for National Statistics(U.K.) Contains public sector information licensed under the Open Government Licence v3.0; National Institute ofStatistics and Economic Studies (France); Banque de France; Statistics Canada; China Statistical Information Networkwww.stats.gov.cn; Italian National Institute of Statistics; Effects of the corona pandemic on the economy and businessactivity in Germany, Statistisches Bundesamt (Destatis), 2020. ona/Economy/context-economy own calculation; Parliament of Australia; Association of American Railroads;Associated General Contractors of America; Mortgage Bankers Association (U.S.); Trading Economics;; Ibis World; Statista;NPD Group; STR; analyst research; etc.5
2.3 Response Scenario TimelineThe final variable is how the magnitude of response varies over time as the pandemic evolves. Our modelallows for complete flexibility to define custom scenarios around mitigation strategies down to a weekby-week level. For historical parts of the timeline we have primarily used Google mobility data todetermine the actual level of response (supplemented by Apple mobility data5 where Google mobilitydata is unavailable).For the U.S. we have evaluated four scenarios that capture variation across three dimensions: 1) when dowe expect widespread availability of a vaccine, 2) will a second wave occur and when, and 3) howaggressive will the recovery be. Figure 5 summarizes our four scenarios and the associated responsetimelines. Optimistic Case: We have modeled out the current U.S. response based solely on announcedmeasures as of June 8 with an end to most statewide shelter-in-place/stay-at-home orders inJune. In the optimistic case, we assume economic recovery will be relatively fast, with consumerconfidence and expenditure rebounding quickly. We also assume lighter mitigation measuresremain in place (together with contact tracing and widespread testing) as we will still have5Apple Mobility Trends Reports. https://www.apple.com/covid19/mobility.6
infections we need to manage until a vaccine is broadly available. We assume broad vaccineavailability in early 2021 at which point magnitude of response returns to normal. This is generallyviewed as optimistic, but this scenario provides the upper bound of our estimates. No Second Wave: We have also modeled out the same scenario with a moderate economicrecovery and a later vaccine timing of mid-2021. This is more consistent with the consensus onwhen a vaccine will be widely available. Fall Second Wave: Most epidemiology models predict the virus will see a second wave. Thisscenario takes the prior case but adds a second wave in the fall due to a combination of poorongoing mitigation, lack of herd immunity, and/or viral mutations. We also assume that economicrecovery will be slower driven by increased bankruptcy and unemployment risk from the secondwave. Winter Second Wave: This represents a more bearish second wave scenario by combining asecond wave (beginning in early 2021 as opposed to fall of 2020) with slower recovery as well asdelayed vaccine timing to 2022. This provides a lower bound to our estimates from a two-yeargrowth perspective (2019-21 CAGR).2.4 GDP ForecastThe above produces our forecast of U.S. Real GDP growth (Fig. 6). In our “Optimistic” scenario, we see a2020 decline of 5% and 2021 growth of 9%. Given the early vaccine timing and fast recovery thisrepresents our most optimistic scenario.For the “No second wave”scenario we assume a moreconsensus-based broadvaccine availability of mid2021. This scenario thereforepushes out the recovery dueto longer mitigation andslower return to “business asusual,” which results in anestimated GDP impact of 6% for 2020.Both second wave scenarios result in a “W-shaped” or “double-dip” recession, with an initial downturnfollowed by false signs of recovery and eventually a second downturn. These second wave scenarios areour most bearish in terms of GDP impact and result in a -6% to -7% impact to 2020 GDP. Because the“Winter Second Wave” scenario’s impact extends deeper into next year and involves a later vaccinetiming, we see a much slower return to growth in 2021 for this case (0% growth in 2021 versus 4% growthin 2021 for the earlier “Fall Second Wave” scenario).Over time we continue to track how our model performs against actual economic data as well as analystforecasts, and at the moment our scenarios are in line with the analyst consensus. We have alsobenchmarked our model output against the released Q1 GDP Second Estimate6 resulting in 0.3%difference.6Bureau of Economic Analysis7
Although we present U.S. data in this paper, we have expanded this work to a global model that coversabout 75% of global GDP. This has also enabled us to further calibrate the model across countries. Wehave also been getting feedback from policymakers and non-governmental organizations (NGO’s). Weplan to refine this work in several ways: Regional views: For the U.S. we plan to add a view by state; for the U.K. we plan to add regions.Industry-specific reopening: Given much of the conversation has shifted toward opening theeconomy, we are adding the capability to specify a reopening timeline specific to each industry.Indirect / spillover effects: We plan to add the impact of spillover effects across industries toaccount for the fact that the output of some industries provides the input for others. This willensure we are properly modeling supply-side effects (e.g., supply chain disruptions).Covering low- and middle-income countries: We are discussing a LMIC view with NGO’s (whichmay require additional model changes given their structural differences).Long-term scarring: Although we have modeled variable recovery driven by factors such asbankruptcy risk and long-term unemployment by industry, our current model does notincorporate structural “scarring” or the long-term shifts in growth trajectory for certain industries,which we plan to incorporate in the future.Policymakers are not merely in the business of predicting GDP, but also help to shape it. In the rest of thispaper we outline an overall framework and a data-driven approach on how to think through the recoverydecisions that minimize GDP impact.3. SOCIAL CONTACT AS THE NEW SCARCE RESOURCEOur modeling of the impact to GDP from various intervention scenarios shows that preventing a secondwave of infections has benefits for both public health and the economy. As such, there will be a continuedneed to manage the spread of COVID-19 until a vaccine has become broadly available or herd immunityhas been built. Until then, policymakers face the difficult task of determining how and when portions ofthe economy can be opened in such a way that reduces the risk of a second wave of infections.As this pandemic has highlighted, containing the spread of the infectious disease required trade-offs inthe form of reduced economic activity. However, economic activity and the disease transmission rate “R”are not directly linked. They are both driven by a third factor that we never consider under “normal”circumstances: social contact.Without the threat of a pandemic, social contact has essentially no cost (in fact, it is usually seen as apositive), and therefore we have had little reason to limit it. However, the emergence of COVID-19 hassuddenly put a high cost on social contact in the form of higher rates of disease transmission. Becausesocial contact does not have a “price” that would allow the market to find an efficient equilibrium betweeneconomic activity and disease transmission, reducing social contact relies on government interventionand/or self-initiated behavior changes. In economics this is referred to as a negative externality, a(negative) byproduct of economic activity not accounted for in its cost.7To control the spread of infections and limit the negative impact of social contact, governments haveenacted a variety of restrictions on businesses and individuals. In the upper part of Figure 7, we chart theimpact of these restrictions and self-imposed behavior change on social contact (as measured by Google7This is very similar to pollution, which also has no inherent price. To reduce pollution, some governments have imposedartificial prices that attempt to reflect the cost of its negative externalities.8
mobility data as a proxy) and transmission (asmeasured by R)8 over the initial containmentperiod of the crisis (early March to the troughof U.S. mobility in mid-April). As the data show,transmission rate has declined as social contacthas been reduced. This relationship remainsdirectionally consistent regardless of whichestimates of R used and we have providedalternative estimates of R in the appendix.9Because social contact is also a criticalcomponent of the economy, these restrictionshave also led to a reduction in economicactivity (as measured by Consumer Spending 10in the lower part of Fig. 7). As we will discusslater, the relationship between social contactand economic ac
economic impact datapoints4 that cross references economic impact in 30 industries with the magnitude of response in place at the time. To the extent possible we rely on official economic reporting data but have also incorporated more real-time or high-frequency data sources (e.g., consumer spending, shipping volumes).
COVID-19 Mental health impact COVID-19 Impact on Sleep COVID-19 Positive Impacts University of California, San Dr. Ariel J. Lang ajlang@health.ucsd.edu ID: 21877 COVID-19 Household Environment Scale (CHES) - English COVID-19 Household Environment Scale (CHES) - Spanish COVID-19 Social Distancing and Symptoms COVID-19 on Family .
The RER No. 17 is a collection of notes on the Economic and Social Impact of COVID-19 that will be pub-lished in three parts. The first part was launched on April 29 and focused on the macroeconomic impact of COVID-19. This second part shows how the macroeconomic impact affects the people in the region. It discusses the social impact of COVID-19 in the Western Balkans in six separate RER notes .
3.1. The Covid-19 global and regional context 3 3.1.1. Impact on global and regional growth 4 3.1.2. Covid-19 effects on the global and regional employment 7 3.1.3. Global Covid-19 Responses 8 3.2. Impact of the Covid-19 on the Malawi economy. 10 3.2.1. Impact on the overall economy 10 3.2.2. Sectoral analysis of the effects of Covid-19 12
4.2 Impact of Covid-19 on the Employment 16 4.3 Impact of Covid-19 on the Business Revenue 19 4.4 Impact of Covid-19 on the operations of Businesses 20 4.5 Significant challenges faced by business markets due to Covid-19 22 4.7 Support needed by business 29 4.8 Support by Businesses to the Government in its fight against Covid-19 31
1 Policy Brief Articulating the Pathways of the Socio-Economic Impact of the Coronavirus (COVID-19) Pandemic on the Kenyan Economy1 Summary - This policy brief assesses the possible vulnerabilities and impacts on Kenya of the COVID-19 pandemic. Although it is too early to predict the socio-economic impact of the COVID-19 pandemic on the Kenyan econ0my,
SRH Sexual and Reproductive Health UNAIDS Joint United Nations Programme on HIV/AIDS . 8 Socio-Economic impAct ASSESSmEnt of coViD-19 on ApuA p nEw GuinEA COVID-19 and its Socio-Economic Impact . impact, and policy recommendations. The SEIA proposes recommendations on how to 'build back better', i.e. it provides options for
The Socio-Economic Impact Assessment of COVID-19 in Thailand was commissioned by the UN Country Team . and Public Health Measures 12 Figure 3: Impact Pathway Analysis Depicting Measures to Mitigate the Socio-Economic Impact 23 . Intelligence Unit and a report on the social impact of the crisis by Oxford Policy Management. It also incorporates
THe GLoBAL ANd reGIoNAL eCoNoMIC CoNTeXT 6 3.1 The global economic context 6 3.2 The regional economic context 8 4. IrAQ’S eCoNoMIC STATUS BeFore THe CoVId-19 PANdeMIC 9 5. THe eCoNoMIC IMPACT .
