By Huw Lloyd Ellis, Queen’s University
Huw Lloyd Ellis is a Professor of Economics at Queen’s University. Here, he discusses the new STUDIO model developed by Queen’s University economists and Limestone Analytics for assessing the impact of COVID-19.
We all know it’s bad. COVID-19 and the lockdowns needed to counter it have created a global economic storm whose impact on Ontario since mid March has been more disruptive than any downturn that most of us have seen in our lifetimes. We’ve seen large downturns in the level of employment. A large fraction of those still employed are working from home and many of those still employed were working reduced hours.
Understanding the economic costs in terms of lost production from these adjustments is important for many reasons. Firstly, these costs translate into major losses in household incomes that may never be recouped. These losses are far from equally distributed and depend crucially on where people live and the industries in which they work. Secondly, the resulting loss in the tax base adds an additional strain on government finances over and above those created by increased spending to offset the size and impacts of layoffs and business distress. The ongoing losses in production today represent a permanent loss in economic wealth that will impact our future after-tax incomes for many years.
No matter how distasteful, we must ultimately come to terms with the trade offs between the economic costs of lockdowns and the potential loss of life implied by relaxing them. This ugly cost-benefit analysis is crucial in guiding policy relating to the nature and rate of relaxation of restrictions and the design of future approaches to additional waves of COVID-19 and other possible future pandemics. Estimates of the economic production costs of lockdowns are a crucial component of this analysis.
Unfortunately, accurate measures of value added by industry and aggregate GDP by province are typically available from Statistics Canada only with a significant time lag. While employment statistics are available more quickly via the Labour Force Survey (LFS), their implications for overall production is not clear, especially during periods of rapid change. In order to assess the effects of shocks and policies in real time and to forecast the likely implications of different possible scenarios, economists must typically rely on various quantitative economic models that reflect as much as possible the actual constraints faced and choices made by households, producers and governments.
Economists at Queen’s University in collaboration with Limestone Analytics are building on recent innovations in inter-industry modelling to develop a Short-Term Under-capacity Dynamic Input-Output (STUDIO) model of the Ontario economy to generate real-time estimates of the economic impacts of COVID-19 and the associated lockdown, based on quickly available data. STUDIO uses regional Input-Output tables, provided by Statistics Canada, to realistically account for cross-industry, dynamic interactions that result from each industry’s use of intermediate inputs produced by others. It is used to develop short term (month-to-month) dynamic scenarios and, unlike standard Input-Output models, it incorporates supply-side constraints on production.
STUDIO provides current estimates and forecasts for hours worked, employment, labour income and the contribution to GDP by each industry under alternative scenarios for a given region. The output and employment of a given producer depends on the demand for its goods and services from downstream producers and the supply of inputs available from upstream ones. Consequently, the speed of recovery for any one industry in Ontario depends on its input-output interactions with others that may continue to be constrained, even if that industry is not. STUDIO captures these complex interactions to allow a full assessment of the impacts of each scenario.
So, how bad is it so far? Preliminary estimates of the impacts of the production shutdowns across the province suggest that the loss in Ontario’s GDP relative to predicted seasonal norms were 9.4% in March, 23.7% in April, and 26.0% in May. This is much bigger than the percentage reductions in employment in those months, largely due to the substantial reduction in average hours worked.
Together, these estimated losses amounted to over $40 billion, or about $7500 per household over the first two months. This loss in the value of production translates into an immediate loss of wages for many as well as losses to firm owners, shareholders and lenders. While government interventions can offset the immediate impact of these losses on current incomes, their effects will be felt for years to come through taxes or public expenditure reductions.
And how bad could it get? Estimates, of course, depend on what we believe will happen under various future scenarios. How quickly will the province relax restrictions on various industries after Stage 1? How quickly can firms, restaurants and retailers ramp up production while also maintaining safe working environments? How will household expenditure patterns change in the face of new epidemiological risks? What if there is a “second wave” that is worse than the first causing further shutdowns?
Even under a very optimistic scenario in which Ontario reopens rapidly and the economy largely recovers by the end of 2020, the preliminary estimates are sobering: a 9.9% loss in the province’s annual GDP, which amounts to over $83 billion or almost $15,000 per household. This is likely a lower bound.
These losses are not and will not be equally distributed. We already know that certain industries, such as Food and Accommodation Services and Transportation, and their employees have been hit much harder than others. These are also industries in which average wages are already relatively low. Given likely persistence in reduced demand for these services by consumers, it is hard to see how their wages and employment will recover quickly. Productive reallocation of labour during a period in which other industries are also constrained also seems unlikely. Developing STUDIO and other models further to assess the likely costs of these impacts and their distribution remains a crucial objective.