Mobility, Not Lockdown Easing, Will Drive The Recovery

Changing Behaviour is More Important Than Easing Lockdown Measures

The COVID-19 pandemic has dealt both health and economic blows. As new data emerges, many researchers aim to understand not only how severe the impact has been, but also how mitigating measures have influenced the outcomes. The challenge, however, is that many macroeconomic indicators are low frequency, making it difficult to isolate changes in economic activity to fiscal policies. To address such issues, a new IMF working paper uses high-frequency data on electricity usage, unemployment insurance claims and location-based mobility to track ‘real-time’ economic activity.

It asks:

(1) Does the timing of Non-Pharmaceutical Interventions (NPIs), such as school and business closures, explain differences in the economic consequences across European countries and US states?
(2) How does people’s mobility respond to the outbreak?
(3) What makes an economy more vulnerable to COVID-19?

The main result of the paper is that the timing of de jure NPIs has no discernible effect on economic outcomes. Otherwise stated, country- and state-wide differences in the timing of lockdown measures cannot explain variations in economic fallouts in Europe and the United States. Instead, the authors argue, mobility can help us understand the heterogeneous economic consequences of COVID-19.

They find that changes in people’s mobility happened prior to governments either imposing or easing lockdown measures. In addition, mobility is positively related to economic activity, which suggests that it was the reduction in peoples’ movements prior to US and European lockdowns that caused economic activity to fall. As mobility dropped, so did economic activity, regardless of NPI timings. The authors emphasize that, going forward, the main driver of economic recovery will be mobility returning to pre-pandemic levels, rather than the timing of lockdown removals. Furthermore, their results show that poorer US states (which have a lower share of workers that can work from home) have been more vulnerable to the economic consequences of COVID-19.

Data Sources and Availability

To monitor economic activity across European countries and US states, the authors collect a variety of indicators, depending on data availability, from January to early May 2020.

• Electricity usage
  • Why? Electricity is an input in most economic activities and is difficult to substitute in the short run.
  • When? Data is available within the same day.
  • Where? The European Network of Transmission System Operators for Electricity (ENTSO-E) for 32 European countries, and the U.S. Energy Information Administration for 64 Balancing Authorities in the US.
  • Other info: Energy consumption data exhibits substantial day-of-the-week fluctuations, so the authors measure electricity usage with respect to the same day of the same week in 2019.
• Unemployment insurance (UI)
  • Why? Data closely tracks labour market developments, so an increase in UI claims is one of the earliest signs of rising unemployment.
  • When? Weekly frequency with only a one-week lag.
  • Where? Only available for the US, from the U.S. Department of Labor.
  • Other info: Analysis is complemented with data on hours worked, number of employees and number of businesses from the time-tracking tool Homebase.
• COVID-19 cases and deaths
  • Why? Data tracks the severity of the pandemic in each country or region.
  • When? Daily data.
  • Where? European Centre for Disease Prevention and Control (ECDC) for European countries, and COVID Tracking Project for US states.
• Mobility
  • Why? Captures de facto social distancing.
  • When? Daily basis.
  • Where? Google Community Mobility Index for both European countries and US states.
  • Other info: Data is based on the number of times individuals visit certain places. Daily change in mobility is computed with respect to the median value in the corresponding date of the week during the period 3 Jan. to 6 Feb. 2020. Data on transit stations, workplaces, retail stores and recreation places, and groceries and pharmacies are used as measures of social and economic life.
• Non-Pharmaceutical Interventions
  • Why? Timing of adoption for social distancing, closure of nonessential services, closure of public venues, school closures and shelter-in-place orders.
  • When? Daily data.
  • Where? University of Oxford Coronavirus Government Response Tracker for Europe and Keystone Strategy for the US.
  • Other info: Index is a score based on overall government response, containment and health, economic support and stringency index.

It is worth noting that Macro Hive has been tracking data on mobility and NPIs for a number of weeks. You can follow our weekly reports in the Mobility Report and Stringency Tracker. We also track flight data, which you can find here.

Correlations

The paper looks at a number of correlations. Energy consumption and the labour market (US) are proxies for economic activity and are considered in relation to COVID-19 severity, mobility (a proxy for de facto mitigation efforts) and NPIs (a proxy for de jure mitigation efforts). They divide the results by geographic regions:

Europe

• COVID-19 and energy consumption: A doubling of deaths leads to a decrease in energy consumption of approximately 2.3%. Otherwise stated, countries with a higher number of deaths per capita and a sharper decline in people’s mobility have also reduced their energy consumption more.
• De facto mitigation efforts (mobility) and energy consumption: There is a positive association between electricity usage and mobility.

Charts 1 & 2: EU Energy Usage Is Linked to COVID-19 Deaths and Mobility

Source: Page 14 of ‘Tracking the Economic Impact of COVID-19’

• De jure mitigation efforts (NPIs) and energy consumption: There was a statistically significant correlation between stringency mitigation policies and electricity in early weeks of the pandemic, but not in April. The weak correlation is reaffirmed by the result that electricity usage falls sizeably before the introduction of stay-at-home orders.
• De jure mitigation efforts (NPIs) and mobility: There is a weak correlation between mobility and the timing of lockdown measures. Mobility falls significantly before the introduction of NPIs. It also starts improving about two weeks before reopening of lockdown measures.

United States

• COVID-19 and energy consumption: States with more COVID-19 deaths experienced larger falls in energy usage. Furthermore, within states, as the number of COVID-19 cases increases, electricity usage decreases – a doubling of the number of cases leads to a decrease in energy consumption of 0.8%.
• COVID-19 and labour market: States with more COVID-19 deaths experienced more job losses and a lower number of hours worked. In addition, as the number of cases within states increases, the number of UI claims increase – a doubling of the number of cases is associated with 11% more claims, although this elasticity declined over time. Job losses have increased more in these areas:

A. Poorer states.
B. States with a lower employment share in hotels and leisure.
C. States with a lower share of jobs that can be done at home.
D. States that do not have in place laws for paid sick leave.

• De facto mitigation efforts (mobility) and energy consumption: There is a positive association between electricity usage and mobility – the lower mobility is in US states, the lower energy consumption is and thus the lower economic activity is also.
• De facto mitigation efforts (mobility) and labour market: There is a negative association between UI claims and mobility – the lower mobility is, the higher unemployment claims are and thus the lower economic activity is.

Charts 3 & 4: Higher Mobility Equals Higher Economic Activity

Source: Page 15 of “Tracking the Economic Impact of COVID-19”

• De jure mitigation efforts (NPIs) and energy consumption: There is a weak association between NPIs and energy consumption. The US were early NPI adopters and perform no worse than late adopters in terms of electricity usage. The timing of NPIs is weakly related to energy consumption.
• De jure mitigation efforts (NPIs) and labour market: The timing of de jure NPIs is not significantly associated with the number of UI claims per capita, whether they control for the size of the local outbreak or other state-level characteristics. Furthermore, hours worked declined sharply well before the introduction of NPIs at a state level and recovered earlier than NPIs were eased.
• De jure mitigation efforts (NPIs) and mobility: There is a weak correlation between mobility and the timing of lockdown measures. Prior to lockdown in the US, voluntary distancing and closures happened even more significantly than in Europe, perhaps as individuals followed news coverage on the COVID-19 crisis. Post lockdown, in the 45 US states that allowed nonessential businesses to reopen since late April, mobility showed a gradual recovery starting two weeks before reopening.

Charts 5 & 6: Consumer Confidence Is Tenuously Linked to Lockdowns

Source: Page 17 of “Tracking the Economic Impact of COVID-19”

Bottom Line

Two main findings stand out from both the US and Europe. First, there are strong anticipation effects in both mobility and economic activity (as proxied by UI and energy usage). Second, there is no clear evidence of any acceleration in economic activity soon after the restrictions abated. It is therefore important to emphasise that, for economic recovery, behaviour matters more than the timing of reopening.

The paper is useful because it emphasises the scale of the response required to bring economic activity back to pre-COVID-19 levels. A return to the status quo – that is, the easing of lockdown measures – is now unlikely to be a sufficient stimulus for consumers and workers. In pursuit of a quick and strong recovery, policymakers need to undertake an unprecedented level of intervention and, crucially, restore confidence by minimising health risks. Only once we see mobility return to normal levels can we expect to see economic growth do the same.

To view the full paper, please click here.