Who is Most Vulnerable? Exploring Job Vulnerability, Social Distancing and Demand During COVID-19

A crisis can be generally defined as an event that has a significant outcome on the function and performance of the future interests of a business (Regester and Larkin, 2002). Crisis management is required in order to return an organisation to “normal function” during or preceding disruption (Breier et al. 2021; Kraus et al. 2020). Past crises such as the 2003 SARS global outbreaks (Goddard et al. 2006), and global recessions (1975, 1982, 1991 and 2009) are exemplars of how government policy and actions are imperative to alleviating the impact on businesses and society as a whole, both in the short and long term. Due to limited real time data, prior research on crisis management often focuses on the aftermath of a crisis, with limited studies focusing on a crisis whilst it is still ongoing (Williams et al. 2017). However, literature does identify that crises often have an impact on employment within regions. Depending on the nature of the shock (e.g. economic, health, natural disaster or technological development), it can result in both job insecurity and job displacement. Job displacement can have long ranging impacts, where highly educated individuals often avail of lower skilled jobs, which leads to the displacement of lower skilled individuals (Gesthuizen and Wolbers, 2010; Kalleberg, 1996). Job switching is an option during times of crisis however, it is reliant upon employees having the necessary transferrable cross sectoral skills. Neal (1998) cautions that workers often build human capital and sector specific knowledge and skills, which will incur high costs to switch to a completely new occupation. Furthermore, prior research identifies the long ranging impact of job losses due to crises, which can last for several decades (Wachter et al. 2020; Papadopoulos, 2016).

It is suggested that crises do not impact all individuals equally, which consequently can lead to higher levels of job insecurity and vulnerability across specific sectors and types of individuals. For example, O’Higgins (2014) and Verick (2009) identify that the 2008 economic recession had a larger impact on sectors such as manufacturing, construction, services and tourism compared to others. Research also suggests that young people across all socioeconomic settings appear to be most impacted by prior recessions (Dietrich, 2013; Papadopoulos, 2016; Arpaia and Curci, 2010). O’Higgins (2014) identify that 40% of those unemployed in 2009 were individuals between 15 and 24. Sironi (2018) suggests that unemployment can have a long-term impact on young people, resulting in lower pay and poorer working conditions for those who do manage to find work. Furthermore, it is widely reported that individuals with lower levels of formal qualification have more limited employment stability overall (Cairo and Cajner, 2018) and are more at risk of unemployment (Gebel and Giesecke, 2011). Gebel and Giesecke, (2011) identify that less skilled individuals often have to take on more temporary roles. However, temporary roles are suggested to be the first to lose their jobs during a recession (Peck and Theodore, 2007).

The impact of a crisis on the demand for goods and services is also widely documented within prior literature. Indeed, prior research by Pettigrew et al. (2014) and Ashford and Ashford (2012) suggest that even the threat of potential job loss or reduction will negatively influence consumers spend and consumption patterns. Consequently, at times of recessions or other crises where there is job insecurity or vulnerability, demand for particular goods and services may be reduced, further increasing the risk of unemployment within particular sectors (Eugenio Martin and Campos-Soria, 2014; Papatheodorou, Rosselló and Xiao, 2010). However, some authors suggest that switching to part-time working models or placing individuals on short term work schemes, such as those evident in Germany after the 2008 financial crisis, can help alleviate the impact that crises can have on the short term negative demand changes to goods and services (Boeri and Bruecker, 2011). This may limit unemployment. However, this type of strategy is thought to best suit occupations which have high recruitment and firing costs, are reliant upon specialised human capital and are in countries with strong employment legislation (Lydon et al. 2019).

Unemployment and demand reduction can have severe negative consequences to society and regions overall. McGann (2020) identify that it took many regions over a decade to return to pre-2008 economic recession employment rates and caution that the number of workers which will be displaced by COVID-19 is potentially much higher. Consequently, a timely and effective government and policy response is imperative to mitigate the potential short and long-term impacts of Covid-19. Prior research has resulted in mixed findings on whether centralised or decentralised governance structures will result in more positive effects (Schatz and Berlin, 2011; Jiang et al., 2020). These mixed findings could be explained by the variations in context and magnitudes of a crisis; which make decision making and indeed policy making inherently difficult. During a health pandemic, policy makers are faced with difficult competing values relating to protecting the health and well-being of society versus limiting the impact on the economy (Sharma et al. 2021). Drawing on the public value literature, it can be derived that policy makers have a duty and are accountable for decision making. Their laws, regulations, services and other actions should create value for society (Faulkner and Kaufman, 2017). This becomes difficult during a health crisis, where paradoxical and competing values arise (Yang, 2020). Protecting health at the expense of the economy are ultimately intertwined, where an economic downturn may increase job vulnerability, which in turn can lead to unemployment and negative healthcare outcomes. Rising unemployment in turn has a negative impact upon consumers purchasing power which in turn will limit economic recovery (Ashford et al. 2012; Sharma et al. 2021; Sheth, 2020).

In the case of Covid-19, the initial speed of action by governments varied across the globe due to issues associated with the collation of data on the impact and spread of the virus. By the middle of March 2020, many countries had enforced strict measures to reduce travel within and amongst countries, established social distancing regulations, required businesses to cease trading temporarily or move operations online (UK Government 2020). This has had a considerable impact on economies globally where stock markets slumped by as much as 35% in March 2020 and unemployment levels increased globally by between 3.8% and 12.7% (IMF, 2020). The enforcement of social distancing is a novel and unique characteristic of the COVID-19 policy response, which differs from prior crises, which makes it difficult to use predictive modelling and forecasts (Panovska-Griffiths, 2020). Furthermore, policy decision making in a crisis is also compounded by an unavoidable time-lag between the crisis onset and the publication of data on real time impacts. This demands innovative approaches and analytical insights which combines expert knowledge and cognitive and normative frames (Surel, 2000; Schiliro, 2013). Following McKinsey and Company (2020) we argue that point-in-time estimates at a regional level may be useful to analytically explore the potential impact of a crisis on job vulnerability. This will provide data which can inform immediate policy responses to alleviate the impact of a crisis.

This research is focused on NI which is a peripheral region of the UK and shares a land border with the Republic of Ireland. Historically, the region has experienced social and economic challenges (Rowthorn, 1981; Birnie et al. 2019). Some of these are attributed to political conflict which started back in the 1960s, referred to as ‘The Troubles’. This political conflict resulted in an increase in unemployment which was compounded by a declining manufacturing sector (PWC, 2016). As a post conflict society, it has made significant progress however, it is still one of the most disadvantaged regions of the UK (HM Treasury, 2011; PWC, 2016). NI was one of the hardest hit UK regions in the 2008/09 recession, where the economy contracted by 7.8% (Magill and McPeake 2016). Whilst the economy has been slowly recovering, it has not yet reached 2008 pre-recession levels (PWC, 2016) and still remains dependent on a large public sector, where a large proportion of the private sector is lower, value-added business activity in the non-tradable sector (Teague, 2016). Furthermore during 2019, NI was ranked as the least innovative region in the UK (BEIS, UK Innovation Survey, 2019) which is a position it had maintained for most of the last decade. This signals the ongoing challenges the region is facing in improving productivity and economic growth. Pre-COVID-19, the region has also been faced with economic uncertainties relating to Brexit, where it is predicted that it will be the most negatively impacted UK region. A hard border could undermine the Good Friday Agreement and the 1998 NI Act (European Parliament Briefing, 2018). Considering the historical and unique contextual factors surrounding NI, it could be suggested that COVID-19 will severely impact the region, and result in an economic contraction far greater than the 2008/09 recession. Consequently, there is a need for creative modelling to aid understanding of potential impacts of Covid-19 in order to develop an appropriate policy response.

A variety of data sources were used to inform the assumptions which took into account NI's characteristics. First, the UK, Office for National Statistics’ Labour Force Survey (LFS) occupational data was used to create a ‘pre-COVID’ labour market outlook for NI (ONS, 2019). This was developed by averaging the four quarters of the LFS in 2019. Other data sources deployed were data from the Department for Labour O*NET Occupational Characteristics (Onet, 2020) which provided an indexation for relevant characteristics; the NI Economic Forecasting Model

An internal forecasting model used by UUEPC informing sectoral trends.

In line with the methodology followed by McKinsey and Company (2020), this study utilised Department of Labor O*NET (2021) occupational characteristics data to estimate the vulnerability of all 967 US occupations

These are the standard occupational classifications used by the United States Department of Labor and are comparable with UK standard occupational classifications (ONS, 2021a).

Determination of vulnerability levels were based on a weighted average of scores from ‘proximity’, ‘exposure to disease’, and ‘working with the public’. Greater weighting was given to ‘exposure to disease’ and ‘working with the public’. A statistical adjustment was made to control for high scores across most occupations for ‘proximity’. If this was left unadjusted, it would have overemphasised levels of vulnerability in the labour market. Occupations with a standardised score below 55/100 were categorised as ‘low’ risk; occupations with a standardised score between 55/100 and 65/100 were categorised as ‘medium’ risk; and occupations with a standardised score higher than 65/100 were categorised as ‘high’ risk. Table 2 shows the results of this analysis. At 2 metres of social distancing, 10% of occupations with a ‘low risk’ score are categorised as vulnerable, compared to 40% of occupations deemed at ‘medium risk’ being vulnerable, and 90% of occupations at ‘high risk’ were vulnerable. These numbers were then recalculated to apply a reduction from 2 metre to 1 metre social distancing regulations. Assumptions of reduced vulnerability were based on both survey data (ONS, 2021b) and evidence from industry figures, including those in hospitality and retail (McKinsey and Company, 2020; NI Chamber of Commerce, 2020). Vulnerability classifications can be found in Annex 1.

The impacts of social distancing were then inserted into the ‘baseline’ model to create a new social distancing baseline, before considering impacts on demand. Impacts on demand were considered to occur as a second-stage effect, where social distancing requirements materially constrain demand. This is evidenced in the hospitality, retail and recreation sectors where reduced capacity will reduce demand. Furthermore, wider economic factors were considered, which considered that there could be a reduction in household incomes due to restricted trading. Furthermore, it was acknowledged that expenditure levels may be lower due to reduced confidence in the economy (Bank of England, 2020). In this sense, demand can be classified as both consumer demand and labour demand, both of which are likely to be adversely affected by the COVID-19 crisis.

The calculation of these demand impacts were again influenced by McKinsey and Company (2020) who assessed demand on a sectoral basis. This had to be adjusted for NI using emerging evidence from surveys (ONS, 2021b), UK sectoral Gross Value-Added forecasts (OBR, 2020), local forecasts and labour market insights (Magill and McPeake, 2020). Demand impacts were also adjusted based the breakdown of an occupation into part-time or full-time; where it is identified in prior research that part-time work is typically more insecure than full-time work. A central, or most likely scenario provided the baseline, and the upper and lower bounds were provided to illustrate the most likely optimistic and pessimistic outcomes. Demand-side impacts were presented as ‘shares of jobs in each sector that are vulnerable’. The baseline demand impacts (based on a central scenario), are detailed in Annex 2. This accounts for demand losses at 1 metre and 2 metre social distancing, and whether the job is carried out on a part-time or full-time basis.

Demand impacts were inserted into a Standard Occupational Code: Standard Industrial Code (SOC: SIC) matrix, which identifies a 4-digit occupation's employment share in each 2-digit sector of NI. This data was from the 2011 Census for NI (NISRA, 2011) and was used as a best estimate for employment structures, in the absence of a more recent census. This is regarded as a secure assumption. Structural industrial change has occurred in the last decade, however the types of occupations carrying out work within sectors are not expected to have changed to the same extent. This modelling allows vulnerability of an occupation to social distancing requirements, and to demand impacts to be identified.

Demand-side vulnerability was calculated for the remainder of jobs across occupations that were not affected by social distancing. This allowed a demand-side vulnerability figure for each occupation. When this was aggregated with social distancing vulnerabilities, it presented an estimate of total vulnerability across occupations. Social distancing and demand impacts were modelled at a social distancing requirement of 2 metres and then at a 1 metre requirement.

The model outputs are reported on an occupational, sectoral, part-time and full time, geographical, gender, income and qualifications basis. This was achieved using Labour Force Survey baseline evidence and the Office of Budget Responsibility analysis of Coronavirus Job Retention Scheme (CJRS) claimants (OBR, 2020). Other comparisons, including ethnicity and religious background were not possible due to a lack of granular occupational data in NI.

The most significant policy developments in the UK to help alleviate job vulnerability were those of the Coronavirus Job Retention Scheme (CJRS), Self-Employment Income Support Scheme (SEISS) and rates relief from the NI Executive for NI businesses. Social distancing measures were relaxed from a minimum distance of two-metres to one-metre on 29^th June 2020. Based on our occupational vulnerability model, it was estimated that this change reduced the number of vulnerable jobs from 260,000 to 230,000 (−12%). While a pre-COVID estimate was unavailable due to lack of data, 230,000 jobs being vulnerable is suggested to be much higher than ‘normal’ times. As such, it could be derived that there was a greater risk of unemployment due to social distancing requirements constricting activity in a significant proportion of the labour market. This has implications for several sectors, both from a labour and customer perspective. In production or customer facing sectors, halving of the two-metre social distancing rule increased the number of people that can work on a factory floor, shop or office, which targets those deemed as non-essential workers. This change then provides a reduction in CJRS support requirements and a boost to output (or GVA). In retail and hospitality, loosening of social distancing provided a dual benefit in terms of mitigating the labour market impact, as well as the ability to boost capacity within the retail and hospitality sectors. This should then improve the potential for these places to generate demand and revenues that are closer to pre-pandemic levels. Figure 2 shows the impact social distancing requirements had on the job vulnerability of certain occupations. Figures 2, 3 and 4 show that the largest relative impacts were expected to occur in occupations such as leisure, travel and related personal services. It was found that up to one in five posts would no longer be considered vulnerable with a reduction in social distancing to 1-metre. This is equivalent to approximately 3,000 jobs overall and approximately 2,500 jobs within sectors such as textiles, printing and other skilled trades. The reduction in social distancing could also remove about one fifth of the workforce from vulnerability.

Figure 2

Figure 3

Figure 4

The vulnerability reductions are a transposition of the occupational trends using the SOC: SIC matrix. Those working in hospitality and sales occupations were more heavily represented in the accommodation and food, and wholesale and retail sectors respectively. At the last census in 2011, elementary administration and services occupations made up 44% of all jobs in the accommodation and food and wholesale and retail sectors. While social distancing measures disproportionately impacted these sectors at the outset, they would be expected to see the most significant gains from social distancing relaxations.

Figure 5 shows that accommodation and food services, which had approximately 59% of employee jobs being supported via the CJRS across the UK

Statistics obtained from the HMRC, 2020.

Figure 5

It could be suggested that there is a relative logistical flexibility for most firms in responding to the social distancing rule. Many jobs can be carried out at home, while others can be continued at a distance. The modelling suggests that the most vulnerable are highly exposed to job loss, however, they ultimately make up a minority of the workforce. The greatest risk to the economy does appear to result from demand constriction, where issues with demand will likely continue in certain industries such as hospitality and leisure until social distancing measures relax completely. Slowing economic activity has structural impacts on the economy, which will hurt consumer confidence and lead to a collective tightening in discretionary spending. This ‘paradox of thrift’ (Keynes et al. 2018) across society could prove much more damage to the economy and labour market in the long term.

From our analysis, it was identified that demand reductions for goods and services from consumers, and for labour from employers, would be the most significant driving factor of vulnerability (almost two thirds: one third ratio of demand to social distancing). However, patterns differ by occupation and sector. This is illustrated by the 90:10 relationship in sales occupations and retail and wholesale; and by a 20:80 relationship in leisure and travel. This illustrates how individual sectors will be influenced differently by demand stimulating or social distancing policy (retail and hospitality respectively). Our modelling illustrates clearly that retail and wholesale accounts for the largest numbers of vulnerable jobs. However, whilst social distancing relaxations will be welcomed, as mentioned, demand is much more important and therefore is the key risk. As such, other supports may be necessary for those occupations and sectors that are less sensitive to changes in social distancing requirements, and also for those workers that are not returned from furlough.

Figure 6 demonstrates that a very vulnerable occupation, elementary administration and services, is impacted by social distancing requirements. It was identified that those within this occupational group for example bar staff, working in customer-facing, high capacity roles had 67% job vulnerability at the 2-metre social distancing restriction, and 62% at the 1-metre social distancing restriction. Conversely, sales occupations, which are heavily exposed to peripheral demand impacts, were found to have 88% job vulnerability at the height of the crisis.

Figure 6

Figure 7 shows occupational vulnerability and suggests that up to 85% of wholesale and retail trade jobs are likely to be vulnerable to demand impacts. This is likely to be driven by limited capacity and expected falls in discretionary spending and consumer confidence.

Figure 7

Between March and May 2020, there was a strong rise in the number of people not employed in their previously “normal” roles. This took the form of unemployment (64,000 at time of analysis in May 2020)

NISRA, 2020, available at: https://www.nisra.gov.uk/publications/claimant-count-tables

Those individuals who have lower levels of formal qualifications are estimated to be the most negatively impacted group in society. They are more likely to be in occupations that have been supported through the CJRS or have become unemployed or economically inactive. In addition, the qualification levels of those who are already economically inactive or unemployed are skewed towards those with lower educational attainment on average (Magill and McPeak, 2016). As shown in figure 8, those with lower levels of formal qualification were likely to find it more difficult to secure employment prior to the COVID-19 crisis (ONS, 2017) and if employed, are more likely to have been in vulnerable occupations and sectors as the crisis unfolded.

Figure 8

Due to those with lower levels of qualifications being disproportionately impacted by social distancing requirements, it is unsurprising that they are likely to realise the largest relative reduction in employment vulnerability by reducing the 2-metre social distancing rule to 1-metre (see figure 9). Occupations that would benefit from this include;

Elementary administration and services sector (47% of workers possess less than level 2 qualifications under the NQF);

Figure 9

Collectively, these occupations account for the removal of approximately 13,000 jobs from vulnerability as a result of the reduction in social distancing. Whilst these lower qualified workers are likely to have other skills (for example experience and aptitudes not captured by a formal qualification), this can be difficult to track formally in a job application process and in related data. Consequently, this increases the risk that those with lower levels of formal qualifications may find it difficult to secure employment as policy supports unwind.

In terms of age groups, it is estimated that 32 – 38% of 16–24-year-old workers were vulnerable to the impacts of COVID-19 in June 2020, compared to 22–26% of 35 – 44-year-olds, who are the least affected age band (see figure 10). Official data indicates that 29% of elementary administration and services occupations are carried out by 16–24-year-olds, while one-in-three people working in sales are in this age band (ONS, 2020). Social distancing relaxations to 1- metre would most benefit the 16–24-year-old age group by removing 2–8% or 2,000–8,000 individuals from vulnerability.

Figure 10

Higher levels of formal qualifications and average earning power are negatively correlated to job vulnerability. The modelling suggested that the lowest earners are disproportionately impacted by job vulnerabilities and are most likely to become less vulnerable due to the reduction in social distancing requirements. Around two thirds of the jobs that are no longer estimated to be vulnerable due to a reduction in social distancing requirements are in occupations in which the median wage is less than £20,000 (see figure 11).

Figure 11

From the point-in-time estimate, figures 12 and 13 indicate that males would be more likely to suffer the negative economic impacts of COVID. 29–33% of male-occupied jobs are suggested to be vulnerable, compared to 20–25% of female occupied jobs. Recent increases in health worker demand, which is generally female dominated, has helped ease the impact for females. However, conversely, male dominated sectors such as manufacturing and construction are some of the early returners from CJRS. Interestingly, when the impact of social distancing relaxations are considered, females are slightly more likely to be removed from vulnerability than males. 5% – 20% of female held jobs are likely to see vulnerability reduce, compared to 5 – 15% for male held jobs. This is a product of the two occupations that benefit most from the reduction in social distancing (elementary administration and services and sales) comprising mainly of female employees.

Figure 12

Figure 13

It was found that part-time workers are more likely to be vulnerable to COVID-19 (see figure 14 and 15). This could be driven by the fact that they are more likely to work in the two occupations that are considered to be most vulnerable - elementary administration and services and sales (55% vs 45% for full time workers). Furthermore, they make up a large proportion of workers in wholesale and retail and accommodation and foods services, two of the most heavily impacted sectors. Relaxations to social distancing could also benefit these part-time workers to a greater extent too. Job vulnerability among part-time workers could reduce by 5–20% as a result of the measures, compared to 5–15% for full-time workers. In addition, it is likely that employers will favour flexible or phased return to work, to try to align labour availability with lower levels of demand in the short to medium term.

Figure 14

Figure 15

It was found that the benefits of reducing social distancing requirements are relatively even across NI's council areas, ranging from a 10% to 13% reduction in the proportion of vulnerable jobs. As can be seen in figure 16, Mid-Ulster remains the most vulnerable council, with 29% of jobs considered vulnerable under the one-metre rule and 32% under the two-metre rule. Lisburn and Castlereagh remains the least vulnerable council area, with 24% of jobs considered vulnerable under the one-metre rule and 27% under the two-metre rule.

Figure 16

Importantly, and unlike trends exhibited in earnings and qualifications, the reduction in social distancing results in larger proportions of jobs remaining vulnerable in peripheral areas. Belfast, which was the third least vulnerable council area at the onset of the pandemic (with 29% of jobs considered vulnerable) is expected to see the greatest gains from reduction of social distancing requirements, with the number of vulnerable jobs falling by 13%. In contrast, Mid Ulster, the most vulnerable council area at the start of the pandemic, is expected to see the second smallest reduction in vulnerable jobs, of 10%. This is shown in figure 17.

Figure 17

Our modelling showed that individuals who have lower levels of formal qualification, lower incomes, young people, males and part-time workers benefit most from the reduction in social distancing requirements. However, despite this, these individuals remain the most vulnerable groups in society and are more at risk now than prior to COVID-19. Consequently, policy interventions are needed to target the most vulnerable not only at times of crises but during stability.

It is noteworthy that council areas around Belfast, which were relatively more robust in economic terms prior to COVID-19, are those that benefit most from the reduction in social distancing, due to the concentration of retail and hospitality in those areas. More peripheral areas (that already lagged in economic terms) will benefit less from the reduction in social distancing. Therefore, it is apparent that the recovery could widen the economic gap between the strongest and weakest sub-regions of NI. This concurs with Zenka et al. (2019) who identifies that peripheral areas are more at risk during a crisis. Table 3 provides a summary demographic and geographic overview of those most impacted by COVID-19. Those based in customer facing roles with high sensitivity to demand constraints are most vulnerable (on an occupational and sectoral basis). These areas make up a significant share of the NI labour market and this will accordingly have implications for the wider NI economy. Individuals who are working in vulnerable areas tend to be young, earn less than the median wage and typically possess low levels of formal qualifications. Drawing from prior research on the impact crises can have on the young (Sironi, 2018; O’Higgins, 2014), it can be suggested that any job loss that ensues from this vulnerability is therefore more likely to result in long-term unemployment. Therefore, policy should implement pre-emptive interventions in order to prevent large proportions of these groups from losing their job in a short space of time.

Whilst job vulnerabilities and the associated risk to the economy is mitigated partially by the reduction in social distancing requirements, our research concurs with prior research on recession crises, where demand is the key driver of vulnerability during a crisis (Myant et al. 2016; Ashford et al. 2012). Therefore, stimulating demand is crucial to facilitate the recovery of the economy in the longer term. To illustrate this risk, if just 10% of those vulnerable posts do not have the demand to sustain them, the outcome would translate to a claimant unemployment rate of 10%.

The UK Summer Budget 2020 resulted in the implementation of UK wide policies to boost consumer demand. This included the Kickstarter programme, CJRS returners bonus, VAT reductions for the most impacted sectors, a Stamp Duty holiday for house buyers and the “Eat Out to Help Out” voucher scheme. However, consumer habits change during a recession (Ashford et al. 2012), where it has been suggested that consumers dedicate more time to shopping, however, search for lower prices and identify substitutes (McKenzie et al. 2011). Indeed, the contagion of COVID-19 has led to an acceleration of online spending which impacts local companies. Based on our findings, we suggest that further mechanisms are needed to stimulate local demand in order to reduce job vulnerability. These could take the form of:

An NI-based voucher scheme to encourage consumption in the hardest hit sectors;

It is clear from this research that in general, vulnerable groups will benefit most from the relaxation of social distancing requirements. Public policy should support the most vulnerable groups and those most in need in order minimise the risks of longer-term damage and scarring effects as a result of non-employment (Martin and Sunley, 2015; Benner, 2012). The winding down of the CJRS, and introduction of the Job Support Scheme (JSS) signalled a movement towards supporting businesses that are deemed ‘viable’. From November 2020, additional supports for NI workers and businesses were added. This included:

Coronavirus Job Support Scheme;

However, despite this, there will be several issues left unresolved by these measures. This includes freelancers and the self-employed, those reskilling, and individuals forced to remain in hibernation due to health concerns lasting longer than expected. Therefore, targeted policy supports are needed for these individuals. Furthermore, we suggest that that economic and labour market shifts will impact the economy for some time. This will require the redistribution of skills and training; especially in digital technologies to cope with job displacement (Kraus et al. 2021b; Gesthuizen and Wolbers, 2010). It is suggested that individuals will be going back to work in an economy that has experienced rapid structural change as the recession accelerated existing trends associated with the fourth industrial revolution.

Like all research, this study was bound by some limitations. Whilst we illustrate that our point-in-time model was useful to predict job vulnerability during a crisis, it cannot comment on exactly which jobs will be lost. This is due in large part to the intervention of CJRS and a lack of comparable crisis and policy interventions for reference. There is also a lack of data to be able to provide comparable job vulnerability levels prior to Covid-19. Furthermore, the point-in-time approach, which is adjusted for contextual regional characteristics of NI, limits this model's exact replication in future studies. Future use of this point-in-time approach and methodology will need to consider fluctuations relating to sectoral demand, innovation, technological progress, and government support. These elements will need to be adapted across regions and allow for temporal attributes. Furthermore, whilst this study is discussing the economic implications of a health pandemic, it was out of the scope of this study to be able to model the trade-off between social distancing reductions and potential increases in virus transmission or public health implications. The characterisation of occupations utilised in this study could be expanded beyond analysis of health risks and draw comparisons between the skill sets of those more or less exposed to the risks of COVID-19. This would allow a preemptive assessment of likely patterns of labour re-allocation in response to COVID-19. Labour re-allocation between sectors is historically rare, with Davis and Haltiwanger (1992) finding that intra-industry reallocation accounted for only 13% of annual excess job reallocation in the 1970s and 1980s in US manufacturing. However, we are operating within unique times and the growth of automation suggests that skills are increasingly transferrable in the modern economy, which should ease some allocative frictions. Future analysis may also compare vulnerability findings with actual redundancy data. This will allow for baseline models that can give an idea of vulnerability in the economy in normal times. Such a study would allow policymakers to respond more quickly to shocks in the future. It will also be important to assess vulnerability in a ‘with-COVID world’. While removal of social distancing would theoretically reduce vulnerability in this model to zero, the impacts of economic scarring and changing consumer behaviours will ultimately leave a significant number of jobs in long-term vulnerability or redundancy. New analysis for long-term vulnerability could focus on sectoral demand patterns, skill loss, labour supply, educational attainment, and consumer confidence and behaviours, among others.