- Joseph Paul Byrne, Heriot Watt University, firstname.lastname@example.org
- Mario Cerrato, Adam Smith Business School, email@example.com
- Xuan Zhang, Institute of Economics and Finance, Nanjing University, firstname.lastname@example.org
This blog presents the first empirical results from an ongoing project on COVID-19 in UK and its economic impact. We present and discuss some preliminary results that can help us to better understand the dynamics of the spread of the virus across UK.
The COVID-19 pandemic of 2020 has had an unprecedented global and country level-impact upon public health and economic activity. Country level data may mask our understanding of the causes and consequences of the crisis, therefore in this study we use more granular UK national and regional data to investigate the transmission mechanisms within the UK using data for daily lab-confirmed COVID-19 cases and death rates for different regions: East Midlands; East of England; London; North East; North West; South East; South West; West Midlands; and Yorkshire and The Humber. The data are taken from Public Health England and span the period in the first half of 2020. We consider Non-Pharmaceutical Interventions (NPI), for example social distancing, by public health authorities and indicators of behavioural responses around the introduction of the virus. We report several interesting results that, in our view, can help us to better understand: (i) how the virus spreads across the UK regions; (ii) the speed of diffusion of the virus across the regions; (iii) the impact and the effectiveness of social distancing rules imposed in UK on the spread of the virus.
How the virus spread: implications for policy interventions
Following the peak of the virus in London (2nd April), the North East and North West both followed London rapidly but reached their peaks with a slight delay (30th and 22nd April respectively). England overall peaked on 22nd April. The temporal relationship between London and the North of England raises the question of London been largely connected with these areas. We show that the spread of the virus in London is most correlated with the East and West Midlands (i.e. rho = 0.85 and 0.82 respectively); and least correlated with Yorkshire and The Humber, and Scotland (i.e. rho = 0.63 and 0.56 respectively). Also, the North East and North have a high correlation of 0.98. The South East and East England are also close to one another (i.e. rho = 0.97), as are the South East and South West (i.e. rho = 0.99). Our results imply that while there were temporal gaps between regions when the epidemic was rapidly expanding, when it was contracting in May, the regions behaved in a more similar fashion.
Our research found that there are regions where the diffusion of the virus happens very fast (for example the South East and East England). Measured lockdowns in these regions should be imposed sooner rather than later. However, English regions could potentially come out of lockdown at the same speed, in the absence of evidence of localised infection spikes. Therefore, our result reinforces the case for lockdowns to be coordinated at central government level when appropriate and suggests that to make lockdowns work more successfully, there could be a need of central government intervention and coordination between UK nations.
If infection rates decline at similar slow rates, and asymmetric non-pharmaceutical interventions play little role in this decline, then excessively different policy at national level may only confuse the public health message. In contrast, when COVID-19 begins to take hold it does so at a rapid rate requiring equally rapid, and potentially bespoke, geographical responses within the UK.
Did lockdowns work?
To assess the impact and the effectiveness of the lockdown measures imposed in UK, we use Google Mobility Data and the Oxford Stringency Index. With widening public and government knowledge of the unfolding epidemic, and the real prospect that the NHS had insufficient capacity, on 23rd March 2020 the UK Prime Minister advised citizens to stay at home. We investigate the extent to which this unprecedented lockdown impacted the transmission of the virus and death rates. One widely known indicator of lock-down is measured by the Oxford COVID-19 Government Response Tracker. This measures the strictness of the lockdown measures introduced across countries. We use this to compare the UK response to that of China, for example, in responding to their own epidemics based upon their particular infection curves.
We also use Google Mobility Data. There are several different Google Mobility Measures. Our Google mobility data include retail activity at supermarkets, activity near places of work, travel on the transportation network and also recreational activity in cinemas, restaurants and theatres. We present some interesting first empirical evidence on the impact of government NPI and behaviour responses to the UK COVID-19 epidemic. Although more strict policy and reduced mobility shall likely decrease the infection rate, and therefore confirmed deaths, more evidence to the virus decreasing is associated with greater lock-down strictness and bigger falls in mobility for longer. Our results suggest retail measures of activity in this period (first half of 2020, January to June 2020) is strongly and positively associated with the infection rate. This suggests that as the public began to ‘panic buy’ in the onset of the lockdown period, this may have inadvertently been associated with reduced social distancing and it helped the spread of the disease.
We also find more evidence that reduced mobility around work is associated with a decline in infection and death rates. We present evidence that the Oxford measure of Lockdown Stringency has a negative and statistically significant impact upon the virus. Mobility to go to work has an impact upon infection rate and retail activity is positively impacting infection and death rates. The overall picture seems to suggest that government stringency policy works in reducing the infection rate and social distancing is also effective in keeping the spread of the virus under control.
Image credit: Matt Brown. Published on Flickr | CC BY 2.0