June 2021 – Since the onset of the pandemic, a group of Queen’s Economics Department (QED) researchers have had a remarkable influence on public discourse and government policy around COVID-19 lockdowns both in Canada and abroad. This influence has largely come through their STUDIO Economic Model, developed to help policymakers and public health officials estimate the economic costs associated with alternative disease mitigation and reopening scenarios.

With the onset of the COVID-19 pandemic in early 2020, there was significant uncertainty about how governments should respond to slow or stop the spread of the disease. How effective were alternative lockdown policies at protecting the population from COVID? And, how costly were such policies in terms of their impacts on economies, education disruptions, inequality, and even other aspects of health and mental health? While epidemiologists and public health officials had real time data on COVID-19 cases and deaths and could presented estimates of how many lives might be saved by “flattening the curve” or through stay-at-home orders, we were less prepared with estimates of the costs of such lockdown policies to society. This led to a perception that public health policies were being chosen without regard to their broader and long-term costs.

This led economists from the QED and Limestone Analytics to develop the Short-Term Under-capacity Dynamic Input-Output (STUDIO) model [1]. The STUDIO model allows researchers to estimate the economic costs of alternative lockdown and recovery strategies proposed by policymakers and public health officials, breaking the costs down by sector at the local level. It allowed industry leaders and local communities to understand the impact of the lockdown policies to advocate for relaxing restrictions in low-risk industries and locations. Over the past year, the early capabilities of the model have been expanded to provide a tool for public health officials and policymakers who want to compare the economic costs under alternative lockdown and reopening plans.

  The team is led by QED Professors Huw Lloyd-Ellis and Christopher Cotton, in partnership with Brett Crowley, Dr. Bahman Kashi, and Post Doc Frédéric Tremblay. The research has led to interdisciplinary academic research collaborations with epidemiologists and public health researchers, and a broad range of economists from other institutions. It has also led to community and industry partnerships, as the team members collaborated with local, provincial and national governments, private companies, major think tanks, and national research networks to disseminate their results and make their model and analysis accessible to decision makers.  

The Model

The STUDIO model adapts, calibrates, and implements a dynamic, seasonally adjusted, input-output framework that uses regional Input-Output tables, commonly provided by statistical agencies, to represent and model the evolving production structure of the economy. Unlike standard IO models, STUDIO incorporates supply-side constraints on production. This framework allows STUDIO to dynamically model the effects of severe shocks like those caused by COVID-19 and the ensuing lockdowns by modelling the economy’s evolving production dynamics and allows researchers to provide forecasts of the economic impacts of COVID-19 going into the future. By incorporating different assumptions to represent alternative reopening strategies, the model is then able to provide projections for the economic effects of these different strategies as the economy recovers from the initial shocks.

To calibrate the parameters of the model and the specific IO structure of each provincial economy, the STUDIO model uses Symmetric Provincial Input-Output (IO) Summary Tables based on annual Supply-Use production data, inter-provincial and international trade flow data, and labour market data (wages and employment) for each province, provided by Statistics Canada. The STUDIO model then calculates production outcomes for key industries in each region following the COVID-19 shocks by explicitly imposing labour constraints in the production function, replicating observed changes in hours worked and employment for each of the industries studied using data from the monthly Labour Force Survey (LFS). In addition, it allows for endogenous demand-side effects coming from increased unemployment via a simple household expenditure model.

The STUDIO model studies an intra-year period, and its results are thus reported relative to counter-factual levels of economic metrics for each point in the year calculated as estimates for what would have occurred in no-pandemic scenario, to account for seasonal adjustments in intra-year economic activity.

Initial Findings

STUDIO’s initial modelling considered five different reopening strategies and their economic effects for each Canadian province over the course of 2020. This initial analysis demonstrated several of the model’s key findings that have been influential for public policy in Canada. The initial five possible scenarios ranged from the economic effects that would have occurred if each province adhered to their originally stated reopening plans for spring and summer 2020, to scenarios where new lockdown measures may have been required in fall 2020. This initial modelling showed just how costly COVID-19 was for the economy; even the optimistic scenario predicted significant deviations from seasonal norms until the end of 2020. A large range of possible outcomes based on the initial scenarios was shown, due to their alternative policies. Not surprisingly, scenarios that either introduced negative demand shocks or further negative supply shocks resulted in greater losses across the board than the Optimistic scenario. Of those that introduce demand shocks, the “Dr. Doom scenario”, which assumed household’s propensity to consume out of current income declined by 20% by May 2020 and subsequently recovered only gradually, implied the largest and most persistent losses.

The STUDIO model also implied that the scenarios most devastating for the economy were scenarios in which a second lockdown was required. It was modeled that a second wave scenario in the fall that looked like the first would impact PEI, Newfoundland and Quebec most severely relative to seasonal norms, and Manitoba the least severely. According to these scenario estimates, the overall GDP loss due to combined first and second waves and consequent lock-downs would exceed $285 billion in 2020. For descriptions of each initial scenario and precise numerical results, see [1]. These results suggested to the researchers at the JDI and Limestone Analytics that the least costly restriction policy would be to lockdown earlier and get the pandemic under control as opposed to maintaining less strict lockdowns for a longer period of time.

STUDIO Provides Instructive Further Analysis

The STUDIO model was also applied in December 2020 to analyze the effects of different recovery strategies then under consideration [3]. A policy brief produced by Queen’s economists applied the STUDIO model to study the effects of three different strategies: a Continued Mitigation strategy, that avoided strict economic lockdowns other than in cases when necessary to protect hospitals from becoming overwhelmed, and restricting high-risk large social gatherings; the Melbourne Model, where the imposition of immediate and wide-reaching lockdown restrictions would be undertaken in an effort to bring local transmission rates of COVID-19 close to zero, offering a promise of a fully-open economy before wide-spread vaccination is feasible, which was based off of the Australian zero-COVID strategy; and the Canadian Shield Proposal, which pursued the same eventual goal as the Melbourne Model, but assumes moderately-less-restrictive lockdown measures are used for a longer period of time compared to the Melbourne Model.

               The analysis using the STUDIO model found that there was variation in optimal policy across location. As defined, a Canadian Shield policy was shown to be consistently less-costly than a Melbourne Model policy, with this relationship holding across all locations. In most locations the Continued Mitigation policy was shown to be most costly for the economy, even when widespread vaccination and permanent relaxation of lockdown restrictions by fall 2021 were assumed. However, this was not the case in Quebec or Atlantic Canada, however, where a Melbourne Model policy was shown to be more economically costly. For precise numerical findings, see [3]. Overall, the effects of the alternative policies over the course of the year were graphically depicted [3]:

The analysis illustrated that a more intense early year lockdown could be less costly for the economy than less-intense, shorter duration lockdowns spread across a longer period of time, even under an expectation that vaccinations will be widely available in the second half of the year, facilitating a more robust recovery in the fall and winter even without an early-year lockdown. This was because full economic activity does not immediately resume with the lifting of economic restrictions; rather the economy takes time to fully recover. This meant an intense lockdown allowing for full reopening afterwards may have resulted in fewer overall jobs lost and a lower decline in GDP than on-again, off-again lockdowns where new restrictions would be required every few months before the economy had a chance to fully recover from the previous round of restrictions. Additionally, the briefing suggested that the worst of the economic impacts of the pandemic are now behind us, although the recovery still has a ways to go.

STUDIO Model Influences Policy in Canada

The STUDIO model has had a large influence on public policy in Canada since its creation. The model has been used to author numerous policy briefs at the request of provincial and federal policymakers, and it was the foundation for the Economic Impact Explorer dashboards released by Limestone Analytics and Local Intel, as well as the Eastern Ontario Leadership Council’s COVID-19 economic impact tool and Canada’s Digital Technology Supercluster Looking Glass project.

The STUDIO model provided the economic analysis for Global Canada’s COVID Strategic Choices Group, a task force of business leaders, policymakers, and academics brought together to assess the scientific evidence around the epidemiological and economic impacts of alternative lockdown strategies to recommend a path forward for provincial and federal governments. Their “Canadian Shield” proposal, [4] gained a lot of media attention and helped shape policy decisions at the provincial and national levels. Dr. Cotton served as the economic modeling lead for the group.

The model is also the primary macroeconomic model for the One Society Network, a nationwide network of researchers funded by the Public Health Agency of Canada (PHAC) and the Natural Sciences and Engineering Research Council of Canada (NSERC). The network is working towards increasing Canada’s capacity at infectious disease modeling, bringing together experts in both economics and epidemiology. Dr. Cotton is the co-director of the network, and Dr. Lloyd-Ellis is leading the network’s macroeconomic modeling efforts. The STUDIO model’s use by the One Society Network places it at the forefront of the Canadian effort to collectively develop a national capacity for understanding the impacts of COVID-19, and preparing for the next pandemic. Funding for the One Society Network came through NSERC’s Emerging Infectious Disease Modeling Initiative, an initiative launched by the Minister of Innovation, Science, and Industry, and the Minister of Health to “strengthen ties between the academic community, industry and the public sector [and…] significantly improve the coordination of infectious disease modelling to help Canada better respond to future public health threats” [5]. Situated as the only NSERC funded research network focusing on the broad social and economic impacts of the pandemic and public health policy, as the Initiative mainly focused on epidemiological research networks, the One Society Network, with the STUDIO model at its core, will play an important role in supporting “short-, medium- and long-term public health decisions [and…] produc[ing] tools to evaluate alternative pandemic policy responses for all sectors of the economy and aspects of society, including marginalized groups” [6].

STUDIO Applied in International Analysis

The influence of the STUDIO model has not been limited to Canada and Canadian policy. Demonstrating its adaptability and versatility, the STUDIO Model has contributed to economic analysis in multiple international contexts. Specifically, the STUDIO model has been applied to model the economic costs of COVID-19 in Malawi and Rwanda. For example, it was used as a quantitative method that added estimates of national income and employment income to four different modeled scenarios of how Malawi could emerge from the COVID-19 crisis by 2025 in analysis undertaken by a team of economists from the QED and Limestone Analytics in partnership with the Malawi National Planning Authority and the Copenhagen Consensus Center [7]. It has also been used to compare outcomes under alternative stimulus spending policies in Malawi [8].

References

[1] Cotton, C., Crowley, B., Kashi, B., Lloyd-Ellis, H., and Tremblay, F. (2020). Quantifying the Economic Impacts of COVID-19 Policy Responses on Canada’s Provinces in (Almost) Real-Time. QED Working Paper Number 1441.

[2] Avelino, A. F. T., and Hewings, G.J. D.  (2019). The Challenge of Estimating the Impact of Disasters: Many Approaches, Many Limitations and a Compromise. in Y. Okuyama and A. Rose (eds.), Advances in Spatial and Economic Modeling of Disaster Impacts, Advances in Spatial Science. Springer Nature Switzerland AG.

[3] Cotton, C., Crowley, B., Kashi, B., Lloyd-Ellis, H., and Tremblay, F. COVID-19 Planning for 2021: Comparing the Economic Impact of Alternative Recovery Scenarios. Limestone Analytics and JDI Policy Paper 20-1202, Dec. 2020

[4] Agnew, M., Ayinde, T., Beaulieu, A., Colijn, C., Cotton, C., Crowe, M., Dhalla, I., Ferbey, J., Greenhill, R., Haggart, B., House, B., Imgrund, R., Jebwab, J., Khangura, J., Kwong, J., McCabe, C., Morris, A., Soucy, J.P. R., and Tuite. A. (2020). Building the Canadian Shield: A New Strategy to Protect Canadians from COVID and from the Fight Against COVID. Global Canada, COVID Strategy Choices Group.

[5] https://www.canada.ca/en/innovation-science-economic-development/news/2021/04/government-of-canada-invests-in-infectious-disease-modelling-to-support-canadas-covid-19-response.html

[6] https://www.canada.ca/en/innovation-science-economic-development/news/2021/04/government-of-canada-invests-in-infectious-disease-modelling-to-support-canadas-covid-19-response.html

[7] Brady, C., Cotton, C., Crowley, B., Davis, S., Farquharson, C., Kashi, B., Lloyd-Ellis, H., and Tremblay, F. (2020). Emerging from Crisis: Applying Scenario Planning in Malawi. Limestone Analytics and JDI Policy Paper 20-1201.

[8] Crowley, B., and Tremblay, F. (2021). Optimizing COVID-19 Stimulus Spending in Malawi. Limestone Analytics and JDI Policy Paper 21-0301.

This article was written by Brock Mutic, JDI Intern, with Christopher Cotton, JDI Director. For questions on this research, contact Christopher Cotton.