Categories

Did the Covid-19 local lockdowns reduce business activity in the UK?

James Hurley and Danny Walker

In 2020 governments around the world responded to Covid-19 (Covid) by introducing lockdown measures that were designed to slow the spread of the virus. Business activity fell materially. But it is difficult to isolate the impact of the local lockdown measures on business activity, given that business activity was affected by other factors such as voluntary social distancing at the same time. In this post we compare UK small and medium enterprises (SMEs) located close to the borders of – but not within – local lockdowns with similar businesses just inside, and conclude that the local lockdown measures causally reduced turnover growth by 8 percentage points relative to businesses outside of the lockdowns, driven by restaurants and non-food retail. Average turnover growth over the period was around -20%, which implies that the lockdowns accounted for only two fifths of the overall drop in business activity at most.

This post analyses the impact of local lockdowns on economic activity, where existing evidence is sparse.

This paper analyses Covid lockdowns: measures such as business closures and restrictions on household mixing that have been introduced in the UK to protect public health. Although a large literature has emerged on the impact they have on health outcomes, evidence on economic outcomes is still relatively sparse. In this post we describe the results from a recent Staff Working Paper that attempts to answer the following question: did the local lockdowns in the UK reduce business activity, or would activity have fallen by just as much even in the absence of the measures?

Isolating the causal impact of local lockdowns on economic activity is not straightforward. We focus on the borders of local lockdowns in the UK in mid-2020.

A simple comparison of the activity of businesses that were subject to the lockdown measures with those that were not would not identify the causal effect of the lockdown measures. The results would be biased because lockdown measures are endogenous: they respond to the pandemic, which affects business activity via spending and mobility. Businesses in areas affected by local lockdowns would be very likely to face lower activity even if there were no lockdown. We get around this problem by comparing the turnover and costs of businesses that happen to have been located on either side of the boundaries of local lockdown measures in the UK. More formally, we use a regression discontinuity design (RDD) with a ‘running variable’ that captures the distance to local lockdowns.

To take one specific example, a pub that happened to be 500 metres inside the Leicester border was required to close for business, while one that was 500 metres outside was not. Besides their location, the two pubs are likely to be very similar to one another and have overlaps in clientele. But note that our method captures both the negative impact of the local lockdowns and any positive spillover effects of local lockdowns on nearby businesses.

We focus on the 60,000 SMEs in the UK that have addresses within two kilometres of local lockdowns.

Figure 1 summarises the timing of the public health measures in England, based on information taken from Government and local authority websites. In this post we analyse the local lockdowns that required businesses to close, which are shown in green on the chart. The local lockdowns affected businesses in places like Leicester and Manchester. We use data on the current accounts of two million limited SMEs, obtained via Experian, which we introduced in a previous post. A full definition of the SMEs used in the paper can be found in our previous Staff Working Paper. This data includes a postcode for each business, which we convert to precise geographical co-ordinates, and match to Companies House to obtain data on other business characteristics like the sector they operate in. We use an algorithm to choose the optimal geographical area to analyse around the local lockdown boundaries (known as the ‘bandwidth’), which is two kilometres. This radius covers 60,000 businesses.

Figure 1: Public health measures in England since January 2020

We use a measure of SME turnover growth that followed a similar path to GDP over the Covid period.

We analyse a measure of business turnover proxied using total current account inflows, and costs proxied by total outflows. To strip out seasonality we compute a year-on-year growth rate. To build confidence in the data, we have run some simple comparisons to growth in GDP and aggregate corporate profits. As shown in Figure 9 of this post, the new data tracks macroeconomic aggregates relatively closely.

We find that the local lockdowns reduced SME turnover growth by around 8 percentage points on average and reduced costs growth by 4 percentage points.

Figure 2 visualises the headline result: SMEs that were just inside local lockdown borders had significantly lower turnover growth than those just outside. We find that local lockdowns reduce year-on-year turnover growth by 8 percentage points on average, and costs by 4 percentage points on average. The results are robust to a number of standard placebo checks, including running the analysis using incorrect lockdown boundaries, earlier months or widening the bandwidth in the regressions.

The estimated 8 percentage point impact from local lockdowns could represent ‘lost’ output in aggregate terms, or it could be that it just represents switching of expenditure from inside to outside local lockdowns. So the estimates should be seen as upper bounds on the overall reduction in output caused by local lockdowns.

Figure 2: Impact of local lockdowns on SME turnover growth at the boundary

These effects appear to be much larger for restaurants and non-food retail than other types of business. These are the businesses that were most directly affected by the measures.

We split the data to produce separate results for SMEs in different sectors. The estimated effect of local lockdowns on turnover growth is statistically insignificant for most sectors. But firms in the accommodation and food sector appear to have seen a large and statistically significant effect, of around -12 percentage points. Within that sector, licensed restaurants were hit the hardest, with an impact of -40 percentage points on average (Figure 3). Within wholesale and retail, there was a big divergence between non-food retail and wholesale (Figure 4). These results are unsurprising because restaurants and non-food retail, such as clothes shops, were both directly targeted by the lockdown measures.

Figure 3: Impact of local lockdowns on SME turnover growth for different subsectors in accommodation and food

Figure 4: Impact of local lockdowns on SME turnover growth for different subsectors in wholesale and retail

We estimate that the local lockdowns account for two fifths of the fall in SME growth over the period. There is some evidence that the impact then reversed.

On average, SME turnover growth in the UK was around -20% at the time of the local lockdowns when we take into account all SMEs, including those that were not subject to the measures. The estimated 8 percentage point fall that was caused by the local lockdowns therefore accounts for two fifths of the overall reduction in business activity, so is relatively small in comparison. This suggests that turnover growth is likely to have fallen substantially even in the absence of the local lockdown policies. Taken at face value, the fact that less than half of the drop in turnover among SMEs subject to local lockdowns was caused by the policy measures could imply a major role for other factors in explaining the overall drop in output last year, including voluntary social distancing. Figure 5 also shows that the local lockdown impact was short-lived and may have even fully reversed a couple of months later, perhaps owing to pent up demand.

Figure 5: Impact of local lockdowns on SME turnover growth at different time horizons (months since local lockdown)

There are a number of reasons to interpret these results with caution, but they help to improve the evidence base for analysis of Covid and future pandemic policy.

Our method is not able to strip out the spillover effects of local lockdowns on unaffected businesses. This means our estimates should be seen as upper bounds on the overall reduction in output caused by local lockdowns. This also limits the read across to national lockdowns, where positive ‘spillover effects’ wouldn’t be possible. We also do not quantify the impact of lockdowns on Covid cases, which other studies show was likely to have been large. And it is worth bearing in mind that one objective of the measures was to limit activity at businesses like restaurants and clothes shops: on this criterion they were successful. That said, the results help to inform analysis of the scale of voluntary social distancing during Covid and improve the evidence base for policy responses to future pandemics.


James Hurley and Danny Walker work in the Bank’s Macro-Financial Risks Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.

Source