Research and analysis

Employment Status in the UK: 2023 survey

Published 27 May 2025

Prepared by NatCen for HM Revenue and Customs

Curtis Jessop, Jerome Swan, Jessie Reddin, Marta Mezzanzanica, Olivia Sexton (NatCen Research)

Research report number: 773

August 2024

The views in this report are the authors’ own and do not necessarily reflect those of HM Revenue and Customs

1. Glossary

1.1 Employment terms

Term	Definition
Construction Industry Scheme (CIS)	Under this scheme, lead building contractors deduct money from a subcontractor’s invoice. These deductions go to HMRC to count towards the subcontractor’s tax and National Insurance liabilities.
Employment law	Employment law is an area of the legal system and covers a large number of topics that are involved in the employment relationship. Employment law is dynamic, jurisdiction specific and subject to case law developments. It constitutes contract law and statutory rights such as: pay, holiday entitlement, working hours, maternity or paternity leave, etc. This study focuses on elements of a working relationship that are helpful to determine the employment status of an individual for rights purposes.
Employment status	Employment status is the classification of a working relationship between a person providing work and a person carrying out that work. There are two separate employment status regimes: for tax purposes and for employment rights. For rights, it determines an individual’s entitlement to statutory employment rights and provides the employer with a set of responsibilities. For tax, it establishes the tax regime that applies to the income earned by the individual, which determines the amount of tax and National Insurance contributions individuals (and if appropriate, the businesses they work for) pay. It is important to note that an individual’s employment status for tax and rights purposes will not always be the same. See Chapter 4 for further information.
Engager	The person or organisation an individual does work for.
Limb (a) worker (or ‘employee’)	This group are also referred to as ‘employees’. Section 230 of the Employment Rights Act 1996 states that ‘employee’ describes an individual who has entered into or works under a contract of employment. Individuals with this employment rights status normally have a high degree of certainty over where, when and how work is carried out. The individual would likely work regularly and be paid for completing those hours of work. They tend to be individuals in permanent full-time or part-time work but can also, in some circumstances, include individuals on casual contracts. They are entitled to the full suite of employment rights after relevant qualifying periods. A person may be a limb (a) worker (employee) in employment law but have a different status for tax purposes. Employers must work out each worker’s status in both employment law and tax law.
Limb (b) worker	Limb (b) workers are entitled to core statutory employment rights but have increased freedom in how much, where and when they work. ‘Worker’ is often a term used to describe limb (b) workers only. However, this report uses the statutory meaning of ‘worker’, covering both limb (a) (employees) and limb (b) workers. All limb (b) workers are ‘workers’, but not all ‘workers’ are limb (b) workers. Please also see the definition for ‘worker’ in this glossary.
PAYE (Pay as You Earn)	The system HMRC uses to collect Income Tax and National Insurance contributions from employment income. Employers deduct payments directly from employees’ pay before paying the employee and HMRC.
Platform worker	Individuals who both obtain work through an online platform or ‘app’ and get paid via the same platform or ‘app’.
Self Assessment tax return	The system HMRC uses to collect Income Tax that is not automatically deducted from wages, pensions and savings. For example, self-employed individuals are responsible for declaring their taxable income and filing a tax return by completing and submitting their Self Assessment after the end of the relevant tax year.
Self-employed	Generally, a person is self-employed if they run their business for themselves and take responsibility for its success or failure. Self-employed individuals are not paid through PAYE (unless they own a business and run their own PAYE scheme), instead they typically pay tax through Self Assessment. Self-employed individuals do not have the employment rights of limb (a) workers (also known as ‘employees’) or limb (b) workers. Self-employment is not defined in statute for employment rights or tax purposes.
Umbrella company	An umbrella company is a business often used by recruitment agencies to pay temporary workers. In most cases, the umbrella company employs a temporary worker and pays their wages through PAYE. It does not find temporary work for the worker; this is done by the recruitment agency (also known as an ‘employment business’). Although the umbrella company is technically the employer of the temporary worker and will pay them, the work they carry out will be for one of the recruitment agency’s clients. This work is often on a short-term basis.
Worker	Under section 230(3) of the Employment Rights Act 1996, a ‘worker’ is an individual who has entered into or works under: A) a contract of employment (in other words, they are an employee or limb (a) worker) B) a contract, whereby the individual undertakes to do or personally perform any work or services for another party to the contract whose status is not by virtue of the contract that of a client or customer. This group are known as limb (b) workers. ‘Worker’ is often a term used to describe limb (b) workers only. However, this report uses the statutory meaning of ‘worker’, covering both limb (a) workers (employees) and limb (b) workers. Please note: all limb (b) workers are ‘workers’, but not all ‘workers’ are limb (b) workers.

1.2 Study specific terms

Term	Definition
Assigned status workforce	This term is used by this report to describe those in paid employment who, based on their answers to the upfront screening questions, were assigned an employment rights status for all their three highest paid jobs. This group therefore comprises those who were categorised as either a ‘limb (a) worker (employee)’ or as ‘self-employed/incorporated’ following the initial survey screening section for all of their three highest paid jobs. As such, none of the individuals in this group were routed to the main survey section which aimed to investigate jobs with an employment status that could not be determined quickly by the short set of screener questions.
Screened limb (a) workers	Those in the assigned status workforce who based on the answers to the screening questions were found to be an obvious limb (a) worker for at least one of their jobs.
Unassigned status workforce	This term is used by this report to describe those in paid work whose employment rights status for at least one of their three highest paid jobs was unable to be assigned by the upfront screener. They were therefore left as ‘unassigned’. This could either reflect having less typical self-employed practices, or less typical limb (a) worker practices and therefore may be a limb (b) worker. It does not mean that an individual’s employment status is ‘unassigned’ in real life but instead that this study was unable to determine it based on their initial survey answers (survey screener). Given the complexity of employment status, and the limited questions used to screen participants, not all limb (a) workers and self-employed individuals will have been screened out (as reflected by the feature analysis group outcomes). Conversely, it is possible that some limb (b) workers may have been accidentally attributed to the ‘assigned status workforce’. The analysis presented in this report for this group of individuals therefore relate to the first of their three highest paid jobs that they were left with an unassigned status for following the upfront screening questions.

1.3 Analysis terms

Term	Definition
Feature analysis	This refers to a two-step analysis methodology used in this study to determine worker status for rights purposes. Firstly, by identifying several working patterns and behaviours, the type of association respondents have with each of the four working relationship features is established (see definition above). An association with a feature can be positive (or ‘present’), negative (or ‘absent’) or undetermined. The combinations of associations are then used to form predictive discrete groups. Specifically, they are combined to determine the likelihood that an individual is either ‘highly likely to be a worker’, ‘fairly likely to be a worker’, ‘likely self-employed’ or ‘undetermined’. The criteria for each of these outcomes is outlined in section 1.2 of Appendix B.
Features	In the context of this report, the term ‘features’ refers to the qualities of a working relationship. More specifically, this aligns with the ‘features’ of a working relationship that are often taken into account by the courts to help determine worker status for an individual’s employment rights. The 4 features chosen for analysis in this report are: control, enterprise, integration and personal service.

2. Executive summary

HMRC commissioned The National Centre for Social Research (NatCen) to undertake research on individuals’ employment statuses for rights purposes in the UK labour market. This included a survey to produce an updated estimate of the total ‘worker’ population in the UK. It also included follow-up qualitative research with individuals who had responded to the survey about their employment status. The findings from quantitative research are included in this report. The findings from the follow-up qualitative research can be found in our sister report.

2.1 Outline

A nationally representative survey of 8,767 UK adults aged 18 to 64 was carried out using a sample from the probability-based NatCen Opinion Panel. Survey interviews were conducted online and over the phone, and fieldwork ran from 18 May 2023 to 18 June 2023. The analysis and conclusions drawn from this research may therefore not reflect the current UK labour market and should be viewed in the context discussed in this report.

This research follows on from previous research commissioned by HMRC, which examined employment rights statuses in the UK during early 2019. This report provides an updated look on these same employment statuses and a revised set estimates of the proportion of the population that are ‘workers’ for April 2023.

The term ‘workers’ refers to individuals who under the Employment Rights Act 1996 are in either of these two categories: limb (a) workers (otherwise known as ‘employees’) and limb (b) workers. All other individuals in paid work would fit under the third and final category for employment rights status: self-employed.

2.2 Methodology

Survey participants answered a set of screening questions aimed at establishing the employment status for their paid jobs. Participants looped through the screener a maximum of 3 times, depending on the number of paid jobs they had. If the screening questions failed to assign an employment status to at least one of the respondents’ three highest paid jobs, then the respondents were deemed eligible for the main survey. Participants with multiple jobs were eligible for the main survey based upon the highest paid job that was not assigned a status during the upfront screening. This group of respondents is referred to as the ‘unassigned status workforce’ in this report. To further determine their status for that specific job, the main survey asked them a series of questions based on employment status from case law.

Those who were in paid work and for whom the survey screener was able to assign an employment status for all their three highest paid jobs were ‘screened out’ of the main survey. These individuals are referred to as the ‘assigned status workforce’ in this report. This is because these individuals already had given a clear indication of their rights status during the upfront screening questions and were thereby categorised early on in the survey as either limb (a) workers or self-employed individuals. For participants with multiple jobs, this meant they had been identified as either a limb (a) worker for one of their jobs or as self-employed for each of their jobs.

Those who were not in any form of paid work were also ‘screened out’ as there was no employment status to establish.

The main survey focused on asking those in the ‘unassigned status workforce’ some further questions about the job that was not assigned a status during the upfront screening. To establish the employment status for this unassigned status job, 4 working relationships features were examined: control, enterprise, integration and personal service. Each of these features comprised a range of individual components, selected based on references to them in case law. However, it is important to note that this analysis does not examine all the factors that might be examined if a court were to consider the employment status in each individual case. Individual answers to sets of specific questions were deemed to demonstrate an indication towards the presence or absence of the relevant feature. There could be more than one indication in either direction or none at all, and many answers did not indicate either way. The answers given to the questions relevant to each feature were looked at together to determine whether the overall relationship with that feature was categorised as being positive (or ‘present’) or negative (or ‘absent’). For more information, see the Technical Report. In brief:

Control

Control looked at the extent to which a respondent’s engager could determine the actual work or task or how they would perform the work, as well as the timing and location of the work and whether they could take on work from other organisations. The presence of control indicated an increased likelihood of being a ‘worker’.

Enterprise

Enterprise looked at the extent to which the respondents’ work was arranged such that they could increase their profit level according to their working practices, but also ran the risk of making a loss. The absence of enterprise indicated an increased likelihood of being a ‘worker’.

Integration

Integration looked at the extent to which respondents were an integrated part of the business or the party they contracted with. The presence of integration meant that an individual was less likely to be in business on their own account, and more likely to be a ‘worker’.

Personal service

Personal service looked at the extent to which respondents are obliged to perform the work for their job themselves, without the right to pay someone else to do the work for them. The presence of personal service indicated an increased likelihood of being a ‘worker’.

These 4 features were the conceptual inputs for the ‘feature analysis’ – the type of analysis used in this study to determine worker status. Firstly, the feature analysis determined the associations respondents had with each of the 4 features, and then combined these to form predictive discrete groups (‘highly likely to be a worker’; ‘fairly likely to be a worker’; ‘undetermined’; ‘likely self-employed’). For more information on these two steps of the analysis, see Chapter 4 and Chapter 5.

Figure 1: Employment status in April 2023: overview of key populations

Figure 1 – Employment status in April 2023: overview of key populations

Figure 1 illustrates the key populations of interest in this study and consists of a pie chart showing of the individuals who completed the paid work survey screener:

• 20% were not in paid work
• 57% were assigned a status (assigned status workforce, made up of 52% who were identified as a limb (a) worker (employees) for at one of their screened jobs and 4% self-employed for all of their screened jobs)
• 24% were not assigned a status (unassigned status workforce)

The unassigned status workforce completed the main survey where the feature analysis categorised these individuals into 4 discrete groups: Highly likely workers (9%), Fairly likely workers (46%), Undetermined (38%) and Likely self-employed (6%)

2.3 Key findings

In April 2023, 80% of the UK adults aged 18 to 64 were in some type of paid work, leaving 20% who were not in any form of paid work. Specifically, following a screening stage, a large proportion (57%) was assigned an employment status for all three (should they have multiple jobs) of their highest paid jobs, while the remaining 24% was left with an ‘unassigned status’ for at least one of their jobs. See figure 13 for further details.

52% of all individuals were identified as being in the assigned status workforce and also as being an obvious limb (a) worker (or employee) for at least one of their jobs.

Extrapolated to the UK adult population, this means that: in April 2023, the number of UK adults aged 18 to 64 in the assigned status workforce who were limb (a) workers (or employees) for at least one job was between 20.6 and 22.0 million.

For those with at least one of their three highest paid jobs left with an unassigned employment status after a short set of screening questions — referred to as the ‘unassigned status workforce’ — the feature analysis was used to determine their likely employment status for that job. Overall, 2% were identified as ‘highly likely to be workers’ and 11% as ‘fairly likely to be workers’.

Extrapolated to the UK adult population, this means that: in April 2023, the number of UK adults aged 18 to 64 who were ‘highly likely to be a worker’ ranged from 0.7 to 1.1 million; and the number of those who were ‘fairly likely to be a worker’ ranged from 4.0 to 4.8 million.

Those belonging to the ‘highly likely’ worker group plus the screened limb (a) workers of the ‘assigned status workforce’ formed the smaller estimate for those who were likely to have worker status: between 21.6 and 22.9 million. Adding those who were ‘fairly likely to be workers’, and therefore determining the larger estimate, the number of workers was estimated to be between 25.9 and 27.2 million^{[footnote 1]}.

Within the unassigned status workforce – excluding those who were either ‘highly likely’’ or ‘fairly likely’ to be a worker (55% in total) – there were two more different categories of individuals: a small proportion (6%) were deemed as being ‘likely self-employed’, while a large proportion were allocated to an ‘undetermined status’ group (38%) for that job.

Extrapolated to the UK adult population, this means that: for a number ranging from 3.7 million to 4.1 million of UK adults aged 18 to 64, the feature analysis approach was not able to clearly determine their work status and further investigation would be required.

As some people within the ‘undetermined status’ group would likely to be found as workers upon further investigation, the estimate of the number of workers found within from this research is likely to still be an underestimate of the true figure. Further research into how to assign an employment status to each of the workforce groups is recommended to improve the precision of estimates.

3. Introduction

HM Revenue and Customs (HMRC) commissioned the National Centre for social research (NatCen) to undertake research on employment status in the UK, to advance existing knowledge and provide robust up-to-date estimates of the number of people with ‘worker’ status for their employment rights. In particular, these estimates attempt to account for workers with a less typical set of working practices (referred to in this report as ‘the unassigned status workforce’).

3.1 Research questions

This report includes four chapters. Each chapter focuses on one of the four overarching research questions the study aimed to explore:

• research question 1: what is the prevalence of those in the unassigned status workforce? And what are the demographic characteristics of this population?

• research question 2: what type of work do those in the unassigned status workforce carry out?

• research question 3: for each of the four features of a working relationship that can indicate worker status for an individual’s employment rights, how many of those in the unassigned status workforce display behaviours associated to them?

• research question 4: what proportion of the unassigned status workforce is likely to have worker status, based on the feature analysis? And what employment status is likely to be assigned to these individuals by their engagers?

3.2 Background

An individual’s employment status determines their employment rights and the tax that the person, their clients, or the businesses they work for (engagers) must pay. Section 230 of the Employment Rights Act 1996 defines the employment status of ‘worker’ as either a ‘limb (a) worker’ (employee) or ‘limb (b) worker’. The aim of the limb (b) worker status is to extend the coverage of certain employment rights to a wider group of individuals who are in dependent working relationships but who are not limb (a) workers (employees).

In 2018, HMRC commissioned quantitative research to estimate the number and characteristics of individuals who were workers for rights purposes. By evaluating the engagements of people with less typical working practices (referred to as the ‘unassigned status workforce’ in this report), it estimated that the total number of workers in the overall population was between 23.6m and 26.3m as of early 2019 ^{[footnote 2]}.

Against a backdrop of greater public awareness of limb (b) worker status and a growth in less traditional working practices^{[footnote 3]}, HMRC commissioned new research to update existing estimates of workers in the UK labour market and advance existing knowledge of their characteristics.

3.3 Populations of interest

This report discusses two key populations of interest: those in the ‘assigned status workforce’ and those in the ‘unassigned status workforce’. These populations were identified early in the survey by assessing specific working practices via a short set of screening questions, such as asking individuals how they were paid for their work. These groups reflect the specific research objectives of this study, and do not exist within the wider literature.

The ‘assigned status workforce’ describes individuals in paid work who, based on their initial answers, were obviously either a limb (a) worker or a self-employed individual for all of their three highest paid jobs. The ‘unassigned status workforce’ describes those who, based on the initial evidence from the screening questions, were not assigned an employment status for at least one of their three highest paid jobs. This greater uncertainty for the ‘unassigned status workforce’ meant that further survey questions were required to try to determine their employment status. It is important to note that this does not mean that an individual’s employment status is ‘unassigned’ in real life but instead that this study was unable to determine it using only their answers to the screening questions.

These two labels have been used in this report because they are a ‘best fit’ for communicating the working practices of the populations of interests to a non-technical audience. However, the complex nature of employment status, plus the wide variety of existing working practices, mean that no two labels can be fully comprehensive. Some important caveats for these labels to keep in mind when reading this report are summarised below.

The assigned status workforce

This subpopulation did not include those who described themselves as ‘partners in a business or professional practice’^{[footnote 4]}. This is because the screening sample was too small to provide meaningful data. It also excluded those who said that none of the listed work types in the survey could describe their jobs.

The unassigned status workforce

This group included those who did not provide their National Insurance number for their job, but also those who could not remember whether they provided it or not. Since not providing a National Insurance number represents a less typical working practice for a limb (a) worker, those who said they could not remember providing it were required to answer further survey questions to clarify their status. This means that those individuals who provided it and just could not remember doing so have been given the unassigned status workforce survey route out of caution, but may actually have typical limb (a) worker practices.

An overview of the process for categorising the workforce populations is shown in Figure 2.

Figure 2: Questionnaire overview showing how the target population were established

Figure 2 – Questionnaire overview showing how the target populations were established

3.4 Study context

The primary objective of this research was to produce an updated estimate of the ‘worker’ population in the UK. In addition to the methodological considerations (see appendices), it is important to understand the specific context surrounding this study and how this has changed since 2019 when the previous research was carried out.

Fieldwork was conducted between May and June 2023, asking participants to report on their employment status during April 2023. Therefore, this study only provides a snapshot view of working behaviours during this one month.

Recent years have been a notable period of change for the UK. The formal exit from the European Union in January 2020 was followed almost immediately by the COVID-19 pandemic and the related measures put in place to safeguard livelihoods. After this, Russia’s invasion of Ukraine in February 2022 led to disruption in the global economic system and in particular to supplies of natural gas to western Europe. One consequence of these events was a steep rise in prices, which led to what is currently known as the ‘cost-of-living crisis’. These factors had, and continue to have, many diverse societal, economic and cultural effects – including on the UK labour market. This report lists only a few key points that are related to levels of employment and unemployment:

EU exit

Since 1 January 2020, EU citizens no longer have the automatic right to work in the UK and must apply for visas alongside people from the rest of the world. Research showed these conditions impacted the labour force in the low-skilled economy and contributed to a shortfall of people in the UK labour force.

COVID-19 pandemic

From 2020 to 2021 there were a series of COVID-19 pandemic lockdowns, which led to changes in the UK population’s working patterns. There has also been a rise in the number of working age people who are economically inactive owing to ill health – the total number of these rose from 2 million in spring 2019 to more than 2.5 million in spring 2023, though more detailed analysis suggests that COVID-19 is not the only or main reason for this rise.

Cost-of-living crisis

The cost-of-living crisis differs from some previous episodes of economic downturn in that, to date, it has not seen a dramatic rise in unemployment ^{[footnote 5]} but has instead led to changes in the working habits such as more individuals taking on extra hours, if not changing job.

3.5 Summary of methodology

Survey

The questionnaire was split into two overarching sections: an ‘upfront screening’ questionnaire for all respondents and the ‘main survey’ for people who were unable to be assigned an employment status for one of their jobs based on their answers during the upfront screener (also known as the ‘unassigned status workforce’).

The initial screening section of the questionnaire was designed to determine whether the respondent was in paid work or not during April 2023. If respondents were in paid work, they were then asked about the type of work, employment conditions and pay structure of their job. Respondents were asked about up to three jobs in the screening sections (in descending pay order) to determine whether they were eligible for the main survey. 

The main questionnaire focused on collecting contextual information on the job that respondents in the unassigned status workforce were ‘screened in’ for, including their work pattern, contractual obligations, and income. Further detail regarding the purpose of each questionnaire section can be found in Appendix A.

Sample design

Following previous research commissioned by HMRC on employment status, a high-quality and replicable survey methodology was required to ensure that estimates of   ‘worker’ employment status in the labour market were accurate and could enable comparisons to be made over time with a known level of confidence. For these reasons, a probability-based sampling approach was used for this research.

A sample of adults aged 18 to 64 living in the UK was recruited using the NatCen Opinion Panel, a random-probability research panel owned by NatCen. The random-probability approach means that each member of the population has a known and non-zero chance of selection for the study. Therefore, the findings discussed in this report can be inferred to the target population, thus enabling the estimate of the number and proportion of 18 to 64 year olds in the UK with worker status. The random-probability approach allows the application of statistical testing to establish where differences are statistically significant and the provision of confidence intervals around the estimates.

More details on the sample design can be found in Appendix A.

Fieldwork and response rate

The data were collected over a 4 week fieldwork period with a mixed-mode (Computer-Assisted Web Interviewing and Computer-Assisted Telephone Interviewing) fieldwork design: all study participants were initially invited to take part online, with those choosing not to (or unable to) complete online being followed up by an interviewer from the NatCen Telephone Unit.

The survey was completed by 8,984 NatCen Opinion Panel members across the UK (6.4% overall NatCen Opinion Panel response rate ^{[footnote 6]} ahead of data cleaning. More details on response rates at various stages of recruitment are included in Appendix A.

How to interpret the findings in this report

Data used in this analysis have been weighted to the target population based on ONS mid-2021 population estimates. All findings have been tested for statistical significance, and all differences reported are statistically significant unless stated otherwise. Statistical testing was conducted at the 95% level. For more information about the weighting and analysis approaches, see Appendix A:

• estimates have been calculated to 13 decimal places and rounded to the nearest percentage point (for example, 46.50% = 47% while 46.49% = 46%)
• due to rounding to the nearest whole number, reported percentages might total 99% or 101%
• if the value of a cell in a table is less than 0.5% but greater than zero, the cell shows ‘0%’, whilst values in some cells have been supressed to prevent disclosure (N<25)
• tables with non-ordinal dependent variables are ordered by overall prevalence
• table and chart present only unweighted bases, whilst the figures and the statistical tests are run on the weighted counts

3.6 Report overview

Chapter 1 begins by outlining the prevalence of the paid workforce population among UK adults aged 18 to 64. The demographic differences of the two workforce populations (the assigned and unassigned status workforce) are statistically tested in this section, to illustrate the distinguishing traits between these populations and ultimately provide further insight into the defining features of the unassigned status workforce. An understanding of the unassigned status workforce is then further developed by Chapter 2, which examines the types of work they carried out.

After establishing the demographic and work type qualities of the unassigned status workforce, Chapter 3 and Chapter 4 focuses on analysing the type of relationship that these respondents showed to four features of employment status (control, enterprise, integration and personal service). Chapter 3 begins by outlining each of the dimensions that contribute towards these four features and reports on participant responses to the corresponding survey questions. By subsequently grouping the responses related to each dimension, the final analysis of this chapter illustrates the overall positive, negative and undetermined associations with each feature.

The feature association outlined in Chapter 3 are then used as the input for Chapter 4. The combinations of positive, negative and undetermined associations across all four features are used to assign respondents to four different employment rights status predictions: ‘highly likely to be a worker’, ‘fairly likely to be a worker’, ‘likely to be self-employed’ and those still ‘undetermined’. Using these outcomes, Chapter 4 calculates an estimated range for the overall number of workers in the UK as of April 2023. Lastly, this chapter explores two other elements of employment that may be indicative of employment status: holiday pay and how Income Tax is paid. These dimensions help to provide the bigger picture on employment status, by identifying the employment status that was actually assigned to (highly and fairly) likely workers by their engagers, both with regards to their employment rights and tax.

4. Demographic characteristics of the unassigned status workforce

This chapter addresses research question number 1: What is the prevalence of those in the unassigned status workforce, and what are the demographic characteristics of this population?

This chapter begins by outlining the proportion of adults in the UK aged 18 to 64 who were in paid work during April 2023. Specifically, the prevalence of two workforce populations are highlighted: the ‘unassigned status workforce’ and the ‘assigned status workforce’. Once these have been identified, a comparative analysis between the two workforces focuses on their demographic differences with regard to the total number of jobs they had, their working arrangements, age, sex, ethnicity and socioeconomic classification. Through this analysis, the chapter shows the unique demographic characteristics of individuals within the ‘unassigned status workforce’, who are the target population for the wider report.

4.1 Prevalence in the population

Workforce type

Overall, 80% of the total population were in some form of paid work during April 2023, leaving 20% who were not in any form of paid work. Specifically, 24% of individuals were classified as being in the ‘unassigned status workforce’, while 57% were classified as being in the ‘assigned status workforce’. These results are presented in Figure 3.

Figure 3: Proportion of the paid workforce types during April 2023

Figure 3 – Proportion of the paid workforce types during April 2023

Base: Adults aged 18 to 64 in the UK
Unweighted: n=8,767

It should be noted that respondents who at an initial question reported their working status was ‘Not in paid work’ (21%) were asked a follow-up question which checked if they had done any paid work at all in April. Those who said they had (1% of all respondents) were included in the overall ‘paid workforce’ group even if they did not primarily identify with a paid working category.

Number of jobs

The majority of individuals in paid work reported only having one job (92%), while 7% had two jobs and 1% had more than two jobs in April 2023. Those in the assigned status workforce were more likely to report only having one job (97%) than the unassigned status workforce (79%). Inversely, individuals in the unassigned status workforce were more likely to have two jobs (17%) and more the two jobs (4%) than those in the assigned status workforce (2% and 0% respectively). 

4.2 Demographic Analysis

Overall working arrangements

An individual’s workforce type was associated with their overall working arrangements. The majority of individuals who did some form of paid work were in full-time positions (76%), but this proportion did vary across the two workforce types. Individuals in the unassigned status workforce were less likely to be working full-time (62%) compared to those in the assigned status workforce (82%). Those in the unassigned status workforce were more likely to either be working part-time for 8 to 29 hours a week (25%) or in both paid work and education but not on a government scheme (6%) when compared against those in the assigned status workforce (16% and 1% respectively).

Figure 4: Working arrangements by workforce type

Figure 4 – Working arrangements by workforce type
	Assigned status workforce	Unassigned status workforce	All in paid work
Full-time paid work	82%	62%	76%
Part-time paid work (8-29 hours)	16%	25%	18%
Part-time paid work (under 8 hours)	1%	4%	2%
On a government scheme	0%	1%	0%
On a paid government scheme and also in education	0%	0%	0%
In paid work and education but not on a government scheme	1%	6%	2%
Not primarily a ‘paid working’ category, but did some paid work	0%	2%	1%

Base: Adults aged 18 to 64 in the UK who were in paid work during April 2023
Unweighted: n=7,039 (Assigned status workforce: 5,088; Unassigned status workforce: 1,951)

Age

Overall, workforce type was associated with age. Figure 5 illustrates the differences in age for the two paid workforce populations. A minority of both workforce populations were aged 18 to 24 (8% overall), but those in the unassigned status workforce were more likely to be aged 18 to 24 (14%) than those in the assigned status workforce (5%). The unassigned status workforce was less likely to be aged between 35 to 44 (21%) or 45 to 54 (20%) compared to those with assigned status workforce (25% in each group).

Figure 5: Age group by workforce type

Figure 5 – Age group by workforce type
	Assigned status workforce	Unassigned status workforce	All in paid work
18-24	5%	14%	8%
25-34	27%	26%	27%
35-44	25%	21%	27%
45-54	25%	20%	23%
55-64	18%	19%	18%

Base: Adults aged 18 to 64 in the UK who were in paid work during April 2023
Unweighted: n=7,039 (Assigned status workforce: 5,088; Unassigned status workforce: 1,951)

Sex

Overall, there was an even distribution of the sexes among those who were in any form of paid work in April 2023, with 50% of individuals reporting as male and 50% as female. There were no statistically significant differences between the workforce types and the sex of their respective members.

Ethnicity

The analysis found that workforce type was associated with ethnicity. Those in the unassigned status workforce were more likely to be from an ethnic minority group when compared to those with the assigned status workforce. Specifically, people in the unassigned status workforce were less likely to be ‘White’ (81%) than those in the assigned status workforce (88%). By contrast, those in the unassigned status workforce were more likely to have ‘Black’ (5%) and ‘Asian’ (9%) ethnicities, compared to those in the assigned status workforce (2% and 6% respectively).

Figure 6: Ethnicity by workforce type

Figure 6 – Ethnicity by workforce type
	Assigned status workforce	Unassigned status workforce	All in paid work
Any ‘White’ ethnicity	88%	81%	86%
Any ‘Black’ ethnicity	2%	5%	3%
Any ‘Asian’ ethnicity	6%	9%	7%
Mixed or other ethnicity	3%	4%	4%

Base: Adults aged 18 to 64 in the UK who were in paid work during April 2023
Unweighted: n=7,039 (Assigned status workforce: 5,088; Unassigned status workforce: 1,951)

Socio-economic classification

Workforce type was also found to be associated with socio-economic classification. Among all individuals in paid work, the most common socio-economic classification was ‘managerial and professional occupations’, with this group representing three-fifths of individuals (60%). Although this was the clear majority of both workforce types, those in the unassigned status workforce were less likely to be in this classification (54%) compared to those in the assigned status workforce (62%).

Those in the unassigned status workforce were more likely to work for small employers and own account workers (11%) compared to people in the assigned status workforce (3%). Additionally, the unassigned status workforce was less likely to be in lower supervisory and technical occupations (6%) and more likely to be in semi-routine and routine occupations (14%) compared to the assigned status workforce (9% and 11% respectively). It is also worth noting that those in the unassigned status workforce were more likely to be unclassifiable by this measure (4%) when compared with the assigned status workforce (2%).

Figure 7: Socio-economic classification by workforce type

Figure 7 – Socio-economic classification by workforce type
	Assigned status workforce	Unassigned status workforce	All in paid work
Managerial and professional occupations	62%	54%	60%
Intermediate occupations	13%	10%	12%
Semi-routine and routine occupations	11%	14%	12%
Lower supervisory and technical occupations	9%	6%	8%
Small employers and own account workers	3%	11%	5%
Not classifiable	2%	4%	3%

Base: Adults aged 18 to 64 in the UK who were in paid work during April 2023
Unweighted: n=7,039 (Assigned status workforce: 5,088; Unassigned status workforce: 1,951)

5. Type of work for the unassigned status workforce

This chapter addresses research question number 2: What is the type of work that those in the unassigned status workforce carry out?

The chapter provides an overview for the working practices of those in the ‘unassigned status workforce’, to further understand the behaviours that characterise this target subpopulation. Specifically, the lens of analysis in this chapter focuses solely on answers about individuals’ highest paid job that was not assigned an employment status during the upfront screening. Through this job-specific lens, the sections explore areas such as the industry and sector they worked in, their agreed work pattern, working hours, gross pay, length of time in the job, scenarios for ending the job and whether it involved online platform work.

5.1 Work type

Industry

Those in the unassigned status workforce operated across a wide range of fields for their unassigned status job. Across 16 industries presented to participants, the most commonly reported were ‘other services’ (20%), ‘human health and social work activities’ (14%) and ‘education’ (13%). All 16 industries’ codes (including ‘other services’^{[footnote 7]} which is a category in its own right) were taken from the Standard Industrial Classification (SIC) list used by the ONS to determine the type of economic activity performed by the business.

Industry representation was also relatively high for ‘administrative and support services’ (8%), ‘professional, scientific and technical activities’ (7%), ‘accommodation and food services’ (7%) and ‘information and communication’ jobs (6%). A full breakdown of the industries reported by individuals is shown in Table 1.

Table 1: Breakdown of work industries for those in the ’Unassigned status workforce’

Industry description	Proportion among the unassigned status workforce
Other services	20%
Human health and social work activities	14%
Education	13%
Administrative and support services	8%
Professional, scientific and technical activities	7%
Accommodation and food services	7%
Information and communication	6%
Construction	4%
Wholesale, retail and repair of motor vehicles	4%
Manufacturing	4%
Financial and insurance activities	4%
Transport and storage	4%
Public admin and defence; social security	3%
Agriculture, forestry and fishing	2%
Real estate activities	1%
Mining, energy and water supply	0%

Base: Adults aged 18 to 64 in the UK who were in the ‘Unassigned status workforce’ during April 2023
Unweighted: n = 1,951

Engager sector

Over half of those in the unassigned status workforce (51%) stated that they worked for private sector organisations. There was a further 31% who reported that they worked for public sector engagers, 14% who worked for private individuals (such as providing services in domestic settings) and 9% who worked for third sector organisations. One in twenty individuals (5%) also said that they worked for ‘other’ types of engagers.

Work pattern

Nearly two thirds of individuals in the unassigned status workforce worked fixed or agreed hours (65%) for the job they did not have an assigned status for. The remaining portion was split between people who worked whenever work is offered to them (16%), people who worked when they want to accept work (16%) and, lastly, those that had some other arrangement (4%).

Work hours

The ONS defines full-time as employees working more than 30 paid hours per week. Around half of those in the unassigned status workforce said they worked 31 hours or more a week (50%), with over a third specifically having worked between 31 to 40 hours per week (35%).

For those who worked 30 hours or less a week, most of them worked between 21 to 30 hours per week (13%), while only 4% worked less than 4 hours per week.

There were also one in ten individuals who reported that they had no regular hours (10%). A full breakdown of the weekly hours worked is summarised in Table 2.

Table 2: Breakdown of the job’s weekly working hours for the ‘Unassigned status workforce’

Weekly working hours for the job	Proportion among the unassigned status workforce
Less than 4 hours	4%
4 to 8 hours	7%
9 to 15 hours	9%
16 to 20 hours	8%
21 to 30 hours	13%
31 to 40 hours	35%
41 to 50 hours	12%
51 to 60 hours	2%
Over 60 hours	1%
No regular hours	10%
31+ hours (NET)	50%

Base: Adults aged 18 to 64 in the UK who were in the ‘Unassigned status workforce’ during April 2023
Unweighted: n = 1,951

Length of time in job

More than half of those in the unassigned status workforce had started their job over two years ago (55%). The next largest proportion was for those who had started this job between one and two years ago (19%), followed by those who started between six months and a year ago (13%) and those who started between one month and six months ago (10%). There was only a very small minority of individuals who had started their job less than a month ago (3%), and even fewer who did not know when they had started doing this job (1%).

Gross pay

Over half of individuals in the unassigned status workforce population (53%) earned no more than £20,000 per year with a sizeable proportion earning £5,000 or less per year (23%).

For those who did earn over £20,000, this was mostly made up of those earning between £20,000 to £30,000 (18%) or £30,000 to £40,000 (10%). Only a small proportion of individuals reported a larger gross pay of over £40,000 (12%), while 8% preferred not to say. A full breakdown of all of the pay bands is shown in Table 3.

Table 3: Breakdown of the job’s gross pay for those in the ‘Unassigned status workforce’

Gross pay by band	Proportion among the unassigned status workforce
Up to £5,000	23%
Over £5,000 but no more than £12,570	16%
Over £12,570 but not more than £20,000	13%
Over £20,000 but no more than £30,000	18%
Over £30,000 but no more than £40,000	10%
Over £40,000 but no more than £50,270	4%
Over £50,270 but no more than £60,000	2%
Over £60,000 but no more than £70,000	1%
Over £70,000 but no more than £100,000	2%
Over £100,000 but no more than £125,140	1%
Over £125,140	1%
Prefer not to say	8%

Base: Adults aged 18 to 64 in the UK who were in the ‘Unassigned status workforce’ during April 2023
Unweighted: n = 1,951

Scenarios for ending their work

An additional question was asked of those in the unassigned status workforce and were still working in the same job selected as reference for this study at the time of completing the survey, to reveal what scenarios might result in them ending this job.

The most commonly cited reason for ending the ‘unassigned’ work was if they were to have lower take home pay (46%), followed by less guarantee of work (33%), less choice over when or where they could work (28%) and a reduced entitlement to employment rights (28%). Those who had multiple jobs were also asked whether if their other jobs started earning more than the study reference job, would this result in them ending the latter, 7% said that it would.  Further to this, 30% of individuals said that none of the listed scenarios would result in them ending the job for which they had an ‘unassigned’ working status.

Online platform work

Within the unassigned workforce, a sub-set of individuals had been asked whether they obtained work and got paid through, an online platform or ‘app’. This subset were those who were either paid by the organisation they worked for, paid a salary or wage by an employment agency or umbrella company, ran a business or professional practice, worked for themselves or did freelance work. Among all of those in the unassigned status workforce, 86% did not both obtain work and receive payment via an online platform or ‘app’, while 10% said they did. This left 4% of the unassigned status workforce who were not eligible for this question as it was highly unlikely that they obtained work, and got paid through, an online platform or ‘app’. These were those who described themselves as a director of their own limited business or a subcontractor in the construction industry scheme.

5.2 Continuing relationship

Future work requirements

Three-fifths of individuals in the unassigned status workforce reported that their engager was not required to offer them further work once they have completed the work they are contracted to do (62%), while a further fifth were required to be offered further work (21%) and around a final fifth did not know (17%).

Acceptance of work

Similarly, those in the unassigned status workforce were asked if they were required to accept any work that was offered to them by their engager, around three-fifths reported that they did not have to accept the work (59%). This left 27% who were required to accept the work and 14% who did not know.

6. Feature groups

This chapter addresses research question number 3: For each of the four features of a working relationship that indicates worker status for an individual’s employment rights, how many of those in the unassigned status workforce displayed behaviours associated to it?

This chapter examines the unassigned status workforce by exploring the working practices that relate to employment rights. It focuses on four working ‘features’ which align with those in case law that are highly relevant for determining whether an individual is likely to have worker status or not, namely: control, enterprise, integration, and personal service^{[footnote 8]}.

Although this study made use of the considerations made by the courts when determining employment status, the methodology did not perfectly replicate what would happen at a tribunal. There may be other identifiers of each of these features that were not assessed within this survey and the questions asked do not represent a complete list. Ultimately, there will be many factors taken into account by the courts and each case will be decided based upon its own merits. Instead, this study’s approach identified the dimensions that the courts have access to when determining an individual’s employment status and grouped these dimensions into features that were intended to take a similar approach to that of the courts.

For each feature, this chapter first lists each feature’s key dimensions, followed by the distribution of answers for the survey questions related to the dimensions. Finally, it discusses the results of the feature analysis that was used to assess the extent to which UK adults aged 18 to 64 in the unassigned status workforce displayed behaviours associated with the feature.

Specifically, the feature analysis was used to establish whether an individual had a positive, negative or undetermined association with each feature:

• positive association: If a set of behaviours indicated that an individual was more likely to be a worker, whether that was limb (a) or limb (b)
• negative association. If a set of behaviours indicated that an individual was less likely to be a worker
• undetermined association: If a set of behaviours indicated both the presence and absence of a feature, or gave no indication at all ^{[footnote 9]}, meaning that neither a positive nor negative association could be established

Depending on the feature considered, a positive association was determined by either a strong presence or clear absence of a feature. In the example of control, integration and personal service, an increased presence of the related behaviours led to a positive association with the feature. By contrast, for enterprise, a notable absence of this feature would result in a positive association. A higher incidence of positive associations across all of the four features then suggested that the individual was more likely to be a worker^{[footnote 10]}.

Accordingly, at the opposite end, this meant that a negative association with a feature indicated being further from the behaviours of a typical worker and that such an individual was more likely to be self-employed. However, these associations are not always clear-cut and thus a proportion of individuals were deemed to have an undetermined association with a given feature. This does not mean that these individuals did or did not have a certain employment status, just that the answers given to the questions asked did not clearly demonstrate a positive or negative relationship. If their cases went to the courts, it is plausible to assume that some limb (a) and limb (b) workers would be found.

6.1 Control

What does ‘control’ mean for employment status?

The level of ‘control’ in an individual’s employment refers to the extent to which aspects of their work are controlled by an engager. That is, the extent to which the person or organisation that an individual does work for is in charge of the work the individual does.

This section analyses the behaviours that indicate the presence or absence of ‘control’ in the work of the unassigned status workforce. In case law, the presence of ‘control’ has been shown to be one of the key conditions for a working relationship to be assigned the status of limb (a) worker (employee). A level of ‘control’ is a common feature of limb (b) workers’ employment too, but unlike limb (a) worker relationships it is not a pre-requisite of this status. The clear presence of engager ‘control’ would therefore equate to a positive association with the feature, thereby suggesting that an individual is more likely to have worker status.

There are several elements that can be used to determine the presence of ‘control’ in an individual’s employment. For this study, five key dimensions were assessed:

• whether their engager could move an individual to work on an alternative task, without renegotiating the amount they are paid
• whether the engager had the right to decide how the individual performs the work
• whether the individual was told when they have to work
• whether the individual was told where they have to work
• whether the individual could take on similar work from other engagers at the same time

Key dimensions of control

Each dimension was assessed using a dedicated survey question, so in total five survey questions measured the extent to which ‘control’ could be detected.

Moving tasks without pay renegotiation

Those in the unassigned status workforce were asked whether they could be assigned different tasks without pay renegotiation for their unassigned status job. Around a third (29%) of this subsample said they could be moved to a different task without pay renegotiation, while a further 24% could be moved tasks but only if they were to agree to a pay renegotiation. Another third (33%) reported that they could not be moved to a different task at all. However, there was some difficulty either understanding or answering the question: 14% said they did not know.

Dictating how tasks should be done

For the majority of the unassigned status workforce their engager had either complete or partial say in how they did their work (78%). Specifically, 30% reported that their engager was able to decide how the work was done without their input, while 47% said they agreed with others how the work needs to be done or they could decide some parts of the work but not others. The remaining 22% reported they could decide how the work is done (of these: 19% said they cover a highly skilled role; 4% said they do not need to listen to anyone else’s view for any parts of their work).

Deciding when to work

Most individuals working in the unassigned status workforce were told in some capacity when they had to work (70%). This was made up of both those who simply worked ‘when I am told to’ (36%), as well as those who were told when to work but that this was driven by the nature of the work they did, as it had to be done at a specific time (34%). One in four said they were able to choose when they worked (25%), while the remaining few had either some other arrangement (4%) or did not know (1%).

Deciding where to work

Similarly to individual arrangements for when to work, most individuals in the unassigned status workforce were also told where they had to work (74%), whether this was because they simply worked where they are told to (42%) or because the nature of the work meant it needed to be done in a specific place (32%). Around one fifth (21%) were able to work where they wanted, while a minority (4%) said they either had some other arrangements or did not know (1%).

Taking on similar work 

Over half of the unassigned status workforce individuals reported they were allowed to take on similar work while doing their job (57%), while one in four (25%) said they were not allowed. A sizeable group of individuals were unsure and said they did not know (18%).

6.2 Feature analysis

The distribution of answers to the five questions outlined in the previous section was used to establish the association between the feature of ‘control’ and an individual’s work. The qualifying criteria for each type of association is outlined in Appendix B.

Overall, the majority of individuals in the unassigned status workforce were classified as ‘undetermined’ (53%). This suggests that it was challenging to establish the extent to which their engagers controlled the work of individuals in this population. A small minority (9%) were negatively associated with this feature, meaning very few individuals showed low levels of engager ‘control’ in their work. The remaining 39% demonstrated a positive association with ‘control’, which is the outcome associated with a greater likelihood of worker status. These results are presented in Figure 8.

Figure 8: Associations with ‘control’

Figure 8 – Associations with ‘control’

Base: Adults aged 18 to 64 in the UK in the ‘Unassigned status workforce’
Unweighted: n= 1,951

6.3 Enterprise

What does ‘enterprise’ mean for employment status?

The levels of ‘enterprise’ in an individual’s employment describes the extent to which their work is arranged such that they can increase profit levels or risk losses based on their working practices.

This section analyses the prevalence of specific characteristics that demonstrate ‘enterprise’. The absence of ‘enterprise’ is linked to a positive association and would subsequently suggest that an individual is more likely to have worker status. Specifically, a low level of ‘enterprise’ reveals that an individual is more likely to be a limb (a) worker (employee). However, some high-profile cases have shown that individuals with considerable ‘enterprise’ present in their work could still have been in a limb (b) worker relationship with their engager, so the positive and negative associations with ‘enterprise’ are less of an indicator for limb (b) status.

In total four key dimensions were used to assess the enterprise feature:

• whether the individual had to provide any of the items that are essential to performing the work
• how the individual was paid for work
• whether the individual had to correct poor-quality work without additional pay
• whether the individual could increase profits by working more efficiently or by reducing costs

Key dimensions of enterprise

The following four questions were used to measure the extent to which the work of the unassigned status workforce showed the absence of enterprise.

Essential items for work

Just under half of the unassigned status workforce were required to provide essential items to complete their work (45%). The most commonly reported item individuals needed to provide was ‘a phone, tablet or laptop’ (30%), followed by ‘tools or other small equipment’ (15%), ‘vehicle essential to providing the services’ and ‘materials to complete the task’ (both at 13%).

Among those who said they were required to provide essential items, 69% estimated their value to be up to £5,000, 8% above £5,000 up to £10,000, 9% above £10,000 up to £25,000, 5% above £25,000 up to £50,000 and 1% above £50,000 (while 8% did not know the estimated value) ^{[footnote 11]}.

There were a few different reasons as to why individuals did not need to provide essential items. In the unassigned status workforce 40% said this was because the person they work for provides the essential items for them, a further 13% claimed that no essential items were required at all, while 2% did not know. 

Type of payment pattern

Almost half of those in the unassigned status workforce were paid a fixed amount each pay day (46%), while over a third said they were mainly paid in relation to the time worked (36%). Others were paid a fixed sum for a job or task (11%), by commission or bonus (2%) or by some other arrangement (4%).

Rectifying their work

When asked whether they were required to correct poor quality work in their own time without additional compensation, 60% of the unassigned status workforce said they were not, while 24% said they were. The remaining 16% did not know.

Increasing profit

Almost three quarters of the unassigned status workforce said they were unable to increase their profits by working more efficiently or reducing their costs (73%). This left 17% who said they were able to do so and 10% who did not know if they could.

Feature analysis

The range of answers to the four questions set out in the previous section facilitated an understanding for the association between the feature of ‘enterprise’ and an individual’s work. The qualifying criteria for each type of association is outlined in Appendix B.

Overall, the feature analysis revealed that over half of the unassigned status workforce were positively associated with the characteristics of enterprise (51%). The rest displayed a roughly equal divide between negative (23%) and undetermined association (26%). These results are presented in Figure 9.

Figure 9: Associations with ‘enterprise’

Figure 9 – Associations with ‘enterprise’

Base: Adults aged 18 to 64 in the UK in the ‘Unassigned status workforce’
Unweighted: n= 1,951

6.4 Integration

What does ‘integration’ mean for employment status?

The levels of ‘integration’ in an individual’s employment describes the extent to which a person is an involved part of the business or the party that they are contracted with.

This section analyses the characteristics that build towards ‘integration’ being present for an individual in the unassigned status workforce. It should be noted that there are many different ways to be integrated into work, making it particularly challenging to determine that an individual is definitely not integrated. For this reason, the threshold for attaining a negative association for this feature is relatively high and so the negative group is fairly small compared to the positive group for this feature. A high level of integration leads to a positive association with this feature, which then suggests an individual is more likely to be a worker – either a limb (a) or (b) – and less likely to be operating a business of their own (in other words, self-employed).

In total four key dimensions were used to assess the presence of the integration feature:

• whether the individual line managed someone in the business
• whether the individual had to follow staff rules or guidelines, were subject to a formal performance management process, or were managed by someone else who works for the engager
• whether the engager had to give more than the statutory week’s notice^{[footnote 12]} before terminating contract
• whether the individual generated work through advertising or other marketing^{[footnote 13]}

This section also outlines whether or not an individual was required to wear a uniform chosen by the persons or organisations they worked for. Although this question was not an input for the final feature analysis, respondent answers help to add useful contextual detail on the extent to which an individual was integrated into the organisation they worked for.

Key dimensions of integration

The four questions outlined in this section were used to measure the extent to which the work of the unassigned status workforce showed the presence of ‘integration’.

Line management of others

Around three quarters of the unassigned status workforce were not responsible for the line management of others as a part of their job (74%), while 22% did line manage others and 4% did not know.

Notice period length

Around half of the unassigned status workforce had a notice period of a week or more (51%). Specifically 27% said the length was up to 30 days, with 24% requiring notice of more than 30 days. A further quarter had either less than a week or no notice period at all (2% and 24% respectively). Finally, a sizeable proportion said that they did not know the length of their notice period (22%).

Rules and guidelines at work

The majority of those operating in the unassigned status workforce did have rules that applied to them and their work (79%). The most commonly cited practice was needing to follow staff rules or guidelines (61%), followed by being managed by someone else (45%) and being subject to a formal performance management process (40%).

Just under one in five said that none of the listed practices applied to them (18%), however it is important to acknowledge that there may be other requirements asked of them beyond the ones included in the survey. Only a minority of individuals (3%) did not know if any of the listed rules applied to them.

Generating work through advertising or marketing

When those in the unassigned status workforce were asked if they had ever generated work for this job through advertising or marketing, around three quarters reported that they had never done this (76%), with 21% reporting that they had done so.

Work uniforms

The majority of those in the unassigned status workforce were not required to wear a uniform (64%), while 36% did have to wear one and less than 1% did not know.

Feature analysis

Using the distribution of responses to the four questions in the previous section, the analysis was able to establish the association between the feature of ‘integration’ and the work carried out by the individual. The qualifying criteria for each type of association is outlined in Appendix B.

The findings of the feature analysis showed that most of the unassigned status workforce had a positive association with the integration feature (60%). Those with this positive association (high levels) of ‘integration’ in their job have an increased likelihood of having worker status. This left 15% who were negatively associated with this feature and 25% remained undetermined for their association with integration. These results are presented in Figure 10.

Figure 10: Association with ‘integration’

Figure 10 – Association with ‘integration’

Base: Adults aged 18 to 64 in the UK in the ’Unassigned status workforce’
Unweighted: n = 1,951

6.5 Personal service

What does ‘personal service’ mean for employment status?

The levels of ‘personal service’ in an individual’s employment describes to the extent to which individuals are obliged to perform the work for their job themselves, without the right to pay someone else to do the work for them.

This section analyses the dimensions that indicate the presence of ‘personal service’ in the employment of the unassigned status workforce. If ‘personal service’ is found within a working relationship, this would then point towards an individual having worker status. To satisfy the criteria for both limb (a) and limb (b) worker statuses, individuals must be obliged to perform work themselves, without the right to pay someone else to do the work for them. However, it is worth noting that although ‘personal service’ is a pre-requisite of being a worker, a negative association with personal service is not definitive for an individual being self-employed. For this reason, personal service has a unique role in the final step of the feature analysis status classification, which is outlined in Chapter 4.

In total two key dimensions were used to classify the levels of the personal service feature in an individual’s work:

• whether the engager had the right to insist that the individual does the work
• whether the individual had ever paid someone else to do the work

Key dimensions of ‘personal service’

The two questions used to examine the extent to which people in the unassigned status workforce had ‘personal service’ present in their work are outlined in this section.

Ability to pay others to do the work

Among those in the unassigned status workforce, just under half of the group (46%) reported that their engagers could insist that they do the work themselves, while 27% said their engager could not insist that they did so^{[footnote 14]}. A further 27% said they did not know whether the person or organisation could insist on this, so it is worth noting that there was seemingly some uncertainty surrounding this characteristic of their work.

History of paying others to do the work

Despite the fact that over a quarter of the unassigned status workforce said they had an unrestricted right to choose someone else to carry out their work, only a very small minority had ever actually paid someone else to do their work (3%). The majority (96%) never paid someone else to carry out their work, while 1% could not remember.

Those who had previously paid someone else to do the work were subsequently asked how frequently they had done this^{[footnote 15]}. The results showed that this was a relatively infrequent practice – 64% reported doing this ‘very rarely’, a further 17% only did so ‘every now and then’ with the remaining 19% adopting this practice ‘often’ or ‘most of the time’ (8% ‘and 11% respectively).

Feature analysis

The combinations of answers to the two dimensions discussed in the previous section were used to determine whether the feature of ‘personal service’ was present (and thus positively associated) with an individual’s work overall. The qualifying criteria for each type of association is outlined in Appendix B.

The analysis found that in general, associations with ‘personal service’ were challenging to establish among the unassigned status workforce. As a result, the majority of individuals were categorised as having an undetermined association with this feature (53%). The high levels of undetermined status for this feature suggests that the assessment of personal service would likely benefit from some additional development work, but this evaluation was beyond the scope of this study. Outside of those with an undetermined association, there was a sizeable proportion who were positively associated with this feature (45%) and therefore had an increased likelihood of worker status. Only a very small minority (2%) were found to have had a negative association. These results are presented in Figure 11.

This low proportion of negative features might be explained by the fact that an individual had to have actually substituted their work to get a negative association. Moreover, the questionnaire design should be taken into account: those individuals who said they were freelancers, sub-contractors or directors of their own limited company, as well as those who said they worked for themselves, were asked whether they employed anyone (excluding family members) to do work for them in their categorisation screener section. Those who did were directly categorised into the ‘assigned status workforce’ group. Given the overlap between the question in the screener section and the criteria for this feature group, this could be a reason for the smaller size of the negative group, as some of those who would have given a negative personal service response were categorised early on.

Figure 11: Associations with ‘personal service’

Figure 11 – Associations with ‘personal service’

Base: Adults aged 18 to 64 in the UK in the ’Unassigned status workforce’
Unweighted: n = 1,951

7. Establishing employment status

This chapter addresses research question number 4: What proportion of the unassigned status workforce were likely to have worker status, based on the feature analysis? And what employment status was seemingly assigned to these individuals by their engagers?

By looking at the interactions between the four key features that are associated with an individual ‘worker’ status^{[footnote 16]} (namely: control, enterprise, integration and personal service^{[footnote 17]}, this chapter provides estimated ranges for the overall number of workers in the unassigned status workforce in the UK as of April 2023.

This chapter also explores two other key elements of employment status: holiday pay and Income Tax. These were analysed as indicators of the employment status that was actually assigned to likely workers by their engager. As holiday pay is a statutory right for both limb (a) and limb (b) workers, but not for those self-employed, it can be helpful for assessing whether an individual is being treated as a worker by their engager. Similarly, the way Income Tax was paid can be used as a proxy for what employment status was assigned to individuals for tax purposes. Being paid net of Income Tax can mean that payment goes through payroll and tax is deducted through PAYE, which would suggest that the individual is employed for tax purposes^{[footnote 18]}. Used together, these two elements are useful for highlighting the differences between a person’s actual working status and their work status through the lens of their engager.

7.1 Interactions between the feature groups

The dimensions of the four features collected in the survey were selected based on references made to them in the case law for employment status. An individual’s answers regarding each of these dimensions have then been used to assess the existence of the features. In other words, their answers have been used to determine a positive or negative feature association. However, as seen in Chapter 3, providing a valid response to such questions did not always guarantee being able to establish the presence or absence of a feature. This is because in some cases a respondent’s set of answers may give indications in both directions or may provide no direction at all. In such scenarios, individuals were assigned an undetermined status for that feature.

Untangling the complex relationships between the presence and absence of each feature can help researchers better understand employment status. For example, those who have positive associations with all four feature groups are the most likely to be a worker (either limb (a) or limb (b) worker). Conversely, those sitting at the opposite end of this spectrum – that is, those with very few or no positive features – are the most likely to be self-employed.

The final results of the feature analysis are displayed in Figure 12.

Figure 12: Feature analysis interactions results and groups for the ‘Unassigned status workforce’

Feature analysis interactions	Percentage of the ‘Unassigned status workforce’
4 Positive features (Highly likely workers)	9%
3 Positive features, 0 negative features (Fairly likely worker)	22%
2 Positive features, 0 negative features (Fairly likely worker)	21%
3 Positive features, 1 negative features* (Fairly likely worker)	3%
0 or 1 features determined or a 1:1 mix ** (Undetermined)	26%
Mixed features (2:2, 2:1, 1:2, 3:1)*** (Undetermined)	13%
0 Positive features, 2 negative features (Likely self-employed)	2%
3 or 4 Negative features***’ (Likely self-employed)	5%

Not including personal service as the negative feature
** 1:1 means one positive and one negative feature
** For more details on the meaning of the ratios, refer to Appendix B
***‘Personal service must be the positive feature if three features are negative
Base: Adults aged 18 to 64 in the UK in the ‘Unassigned status workforce’
Unweighted: n = 1,951

Overall, just under one in ten of those in the unassigned status workforce had all four features positively associated with their work for their unassigned status job (9%) and were therefore identified as being ‘highly likely to be a worker’.

A further 46% of the unassigned status workforce were classified as ‘fairly likely to be workers’ based on their responses. An individual qualified for this group if they met any of the following conditions: three positive features and no negative features (22%), three positive features and one negative feature that was not ‘personal service’ ^{[footnote 19]} (3%), or two positive features and no negative features (21%). Overall, this meant that among those in the unassigned status workforce, there was a combined group of (‘highly’ and ‘fairly’) likely workers that summed to 55%.

Of the remaining 45% of the unassigned status workforce, who were neither ‘highly likely’’ nor ‘fairly likely’ to be a worker, there were two different categories of individuals. The largest proportion of these were those assigned an overall ‘undetermined’ status (38%). To be deemed as ‘undetermined’, there were several feature group interactions that a respondent could have. These combinations meant that an individual either indicated a mixture of both positive and negative associations, or none at all, with the four key features signifying worker status. A breakdown of the feature interactions that qualified an individual as being undetermined can be found in Appendix B.

However, this is not to say that those who were ‘undetermined’ had an uncertain employment status but rather that the bounds of the feature analysis could not categorise this group. As a result, some of this ‘undetermined’ group could still have been workers. A key recommendation for further research would be to conduct additional detailed analysis of this group to help solidify the likely employment status for these individuals.

The remaining 6% of the unassigned status workforce showed either one^{[footnote 20]} or zero positive features and negative features in their working relationship and thus were deemed as being ‘likely self-employed’. Since positive associations with the four features signify an increased likelihood of being a worker, either limb (a) or limb (b), it follows that those who were furthest from these criteria were likely to belong to the self-employment category.

7.2 Worker population estimates

The results of the feature analysis can be used to estimate the number of workers (limb (a) and limb (b) workers) aged 18 to 64 in the UK during April 2023^{[footnote 21]}. The results of the feature analysis are shown in Figure 13 alongside the other groups that were identified during the upfront screening, namely: screened limb (a) workers, those who were self-employed for all of their screened jobs and those who were not in any paid work during April 2023.

The worker estimates provided at the end of this section include a combination of those in the assigned status workforce who had at least one job identified as having limb (a) worker status during the upfront screening, plus those in the unassigned status workforce identified as either ‘highly likely’ or ‘fairly likely’ to be a worker through the feature analysis. Some caveats to this approach are outlined in the next later section.

Figure 13: Likely employment status among adults aged 18 to 64

Likely employment status among adults aged 18 to 64	Percentage of adults aged 18 to 64 in the UK
Highly likely to be a worker	2%
Fairly likely to be a worker	11%
Likely self-employed	2%
Undetermined	9%
Screened limb (a) workers	52%
Self-employed for all screened jobs	4%
Not in paid work	20%

Base: Adults aged 18 to 64 in the UK
Unweighted: n = 8,757

Workers are made up of two different categories of individuals – limb (a) workers (also referred to as ‘employees’), and limb (b) workers. Therefore, in order to accurately estimate the pool of workers within the UK population, this section combines totals for likely workers found in the feature analysis together with those in the assigned status workforce who were already  screened as limb (a) workers via the limb (a) screener section of the survey (52%). This earlier identified group of screened limb (a) workers  formed a proportion of the assigned status workforce outlined in Chapter 1.

Those belonging to the ‘highly likely’ worker group, plus the assigned status workforce limb (a) workers identified in the screener, represented the smaller estimate for those who were likely to have worker status in the sample. This can be used as the lower estimate for an estimated range. By adding in the ‘fairly likely’ worker category to this, a greater upper estimate is produced. The results of these calculations are summarised in Table 4.

Table 4: Worker population estimates (rounded to the nearest 100,000)

Population group	Proportion of sample (%)	Population estimate (millions)	Confidence interval 95% (millions)
Highly likely to be worker group	2%	0.9	0.7 – 1.1
Fairly likely to be worker group	11%	4.4	4.0 – 4.8
Screened limb (a) workers (employees)	52%	21.3	20.6 - 22.0
Undetermined status group	9%	3.7	3.3 – 4.1
Highly likely and fairly likely worker group	13%	5.3	4.8 – 5.7
Smaller worker population estimate (Highly likely workers, plus the screened limb (a) workers)	55%	22.2	21.6 - 22.9
Larger worker population estimate (highly likely workers, fairly likely workers and the screened limb (a) workers)	65%	26.6	25.9 - 27.2

Please note as some categories are subsets of other categories in this table, rows do not add to 100%.
Base: Adults aged 18 to 64 in the UK
Unweighted: n=8,767

Extrapolating the survey results to the UK adult population enables the estimation of the total number of workers, aged 18 to 64, in April 2023. The analysis of those in the unassigned status workforce showed that 5.3 million (95% confidence interval: 4.8 to 5.7 million) were ‘highly likely’ or ‘fairly likely’ to be a worker. By adding this to those who were screened as limb (a) workers (employees), the estimate of the total number of workers added up to 26.6 million (95% confidence interval: 25.9 to 27.2 million).

Caveats to the estimated size of the overall worker population

The figures discussed in this report may systematically underestimate the size of the overall worker population, for two main reasons.

Firstly, the estimate does not account for the 3.7 million individuals with an undetermined status by the feature analysis. While it is likely that some of these would be found to be a worker for their unassigned status job, it was not possible to quantify this with the data available and the type of analysis carried out. Predictive methodologies such as factor analysis or cluster analysis could be used to attempt to categorise these people but this was out of the scope of this research.

Secondly, the steps followed to calculate the smaller and larger estimates do not take into account that some individuals have multiple jobs for which they have different employment statuses. In particular, in the sample there were some individuals who – for their unassigned status job following the feature analysis – were not found to be workers but who also had other jobs that were identified as having limb (a) status during the upfront screening questions. Although this subgroup could be quantified with the survey data, HMRC required this study to follow the same analysis procedures carried out for the 2019 wave. Maintaining comparability with the previous data collection was therefore prioritised over adjustments to the existing procedures to improve accuracy^{[footnote 22]}. While accounting for this group would be unlikely to have a large impact on the overall estimates as only a minority of the population had an unassigned status workforce and multiple jobs (6%), for future research we would recommend that the approach is reviewed.

Further detail on how these estimates could be expanded to avoid underestimations in the future is outlined in the Technical Report.

7.3 Establishing assigned employment status of likely workers

While the feature analysis helped to estimate the proportion of the population who were (‘highly’ or ‘fairly’) likely to operate with worker status in April 2023, it cannot provide an indication of the employment status assigned to the worker by the engager. The Taylor review of modern working practices (2017) highlighted that individuals can have misconceptions about their own employment status. For this reason, other working practices can be used as proxies to help determine their ‘real’ employment status – for rights and tax purposes – in the eyes of their engagers

Employment status for rights purposes

Whether or not someone receives holiday pay can be an indicator of their employment status with regards to their employment rights because holiday pay is a statutory right for all workers, but it is not for those who are self-employed. This dimension was not used in the feature analysis assignment criteria because holiday pay is a consequence of having (being assigned by an engager) a certain employment status, rather than a working characteristic leading to an employment status verdict. However, as with all proxies, it is not perfect. This is particularly true for rolled up holiday pay as it is given as a percentage of additional pay on top of the individual’s normal hourly rate. Although respondents were provided with a definition for rolled up holiday pay, a potential lack of understanding around this practice leaves a small risk that some individuals may not realise they are being paid it since it is included in their regular income. 

Table 5: How holiday pay was received by feature analysis groups

Holiday pay status	Unassigned status workforce (total)	Highly likely to be a worker	Fairly likely to be a worker	Likely self-employed	Undetermined
Receive holiday pay as annual leave	51%	84%	68%	4%	30%
Receive rolled-up holiday pay	9%	7%	12%	1%	8%
Do not receive holiday pay	34%	7%	14%	91%	55%
Don’t know	6%	1%	6%	4%	8%

Base: Adults aged 18 to 64 in the UK in the ‘Unassigned status workforce’.
Unweighted: n = 1,951 (Highly likely to be a worker: 184; Fairly likely to be a worker: 865; Likely self-employed: 167; Undetermined: 735)

Table 5 illustrates that individuals with more positive associations with a worker status were more likely to report having holiday pay. Instances of receiving holiday pay fell with individuals’ positive associations with worker status. Among those categorised as ‘highly likely’ to be workers, 91% said they received holiday pay through either paid annual leave or rolled up holiday pay. 80% of ‘fairly likely’ workers had any type of holiday pay, and only 38% of those with undetermined employment status reported the same. This suggests that individuals whose status was more inclined towards being either a limb (a) or limb (b) worker were also most likely to receive holiday pay, indicating they were more likely treated like a worker by their engager, with regards to their employment rights.

Employment status for tax purposes

To help establish an individual’s employment status for tax purposes, respondents were asked whether they had received payment for their work net of tax or as a gross sum. The underlying reasoning here was that being paid net of tax could mean that the payment is put through payroll and tax is deducted through PAYE, which would suggest that an individual is an employee for tax purposes. However, this could also mean an individual’s pay goes through Construction Industry Scheme (CIS) deductions, which is a withholding scheme to collect Income Tax from self-employed workers in the construction sector. Therefore, this dimension should be only considered as a proxy for employment status in practice, rather than as a confirmed determiner. However, this only applies to a very small proportion of individuals, meaning this dimension can be suitably used as a proxy.

Table 6: Self Assessment tax returns completion by feature analysis groups

Self Assessment tax return completion	Unassigned status workforce (total)	Highly likely to be a worker	Fairly likely to be a worker	Likely self-employed	Undetermined
Self Assessment tax return completed	30%	9%	16%	88%	41%
Self Assessment tax return not completed as pay tax in other way (for example, PAYE)	50%	79%	64%	6%	34%
Self Assessment tax return not completed as do not earn enough income	12%	8%	11%	6%	15%
Don’t know	8%	4%	9%	0%	10%

Base: Adults aged 18 to 64 in the UK in the ‘Unassigned status workforce’
Unweighted: n = 1,950 (Highly likely to be a worker: 184; Fairly likely to be a worker: 864; Likely self-employed: 167; Undetermined: 735)

Table 7: How National Insurance and Income Tax were paid by feature analysis group^{[footnote 23]}

How National Insurance and Income Tax were paid	Unassigned status workforce (total)	Highly likely to be a worker	Fairly likely to be a worker	Likely self-employed	Undetermined
Pay both own NI and Income Tax	36%	23%	21%	88%	48%
Pay own NI but not Income Tax	1%	Answer not selected by any respondents	0%	0%	3%
Pay own Income Tax but not NI	2%	0%	1%	7%	3%
Both NI and Income Tax deducted by organisation they work for	61%	76%	78%	5%	46%

Base: Adults aged 18 to 64 in the UK in the ‘Unassigned status workforce’ who earn enough income to pay Income Tax.
Unweighted: n = 1,766 (Highly likely to be a worker: 172; Fairly likely to be a worker: 791; Likely self-employed: 153; Undetermined: 650)

When looking at the variation in tax status, the responses suggested that those who were positively associated with more features (and therefore who had a greater likelihood of worker status) were most likely to have typical ‘employed’ tax status. A majority of 79% of ‘highly likely’ workers did not complete a Self Assessment tax return and instead their tax is paid in another way such as PAYE, as shown in Table 6. Further to this, 76% of ‘highly likely’ workers reported that their National Insurance contributions and Income Tax were deducted from their pay by the organisation they work for, as illustrated in Table 7.

By comparison, the ‘fairly likely’ workers showed lower levels of people being paid net tax: 64% did not complete a Self Assessment tax return and paid their tax in some other way. 78% also said they had their National Insurance and tax deducted by the organisations they worked for, in line with the figures for ‘highly likely’ workers. These proportions fell further again for those in the undetermined group, to 34% and 46% respectively.

Considering these two components of Income Tax, the results suggested that a higher number of positive feature associations, and therefore an increased likelihood of worker status, aligned with individuals being more likely to be viewed as employed (rather than self-employed) by engagers with regards to tax.

Finally, 30% of individuals, across all work statuses, reported they completed a Self Assessment for their work. The reasons for them doing so are outlined in Table 8

Table 8: Reasons for completing a Self Assessment tax return by employment group

Reasons for completing a Self Assessment tax return	Unassigned status workforce (total)	Highly likely to be a worker	Fairly likely to be a worker	Likely self-employed	Undetermined
To pay tax on self-employment income as I am a sole trader, sub-contractor or freelancer	71%	Suppressed to prevent disclosure	39%	84%	84%
To pay tax on dividend income as I am a company shareholder	10%	Suppressed to prevent disclosure	5%	15%	12%
I earn over £100,000 per year	4%	Suppressed to prevent disclosure	7%	1%	3%
Reasons independent of my employment status, such as to report bonus earnings	8%	Suppressed to prevent disclosure	15%	2%	6%
Don’t know	15%	Suppressed to prevent disclosure	40%	8%	6%

Base: Adults aged 18 to 64 in the UK who completed a Self Assessment tax return.
Unweighted: n = 691 (Highly likely to be a worker: 22; Fairly likely to be a worker: 165; Likely self-employed: 144; Undetermined: 360)

8. Conclusion

An individual’s employment status determines the employment rights they are entitled to, and the taxes that they and the organisation they work for must pay, and these two systems are separate. Employment status is therefore at the core of both employment law and the tax system, but it can be highly complex to assess in some cases.

In the UK there are two employment status categories for tax purposes: self-employed or employee. Whereas there are three categories for employment rights purposes. The three categories for employment rights are detailed below:

Limb (a) workers (employees)

Limb (a) workers (employees) are entitled to the full suite of employment rights (after the relevant qualifying periods) and work under a contract of employment.

Self-employed individuals who work for themselves

Self-employed status is not defined in employment law. Self-employed people have very few employment rights.

Limb (b) workers

Limb (b) workers have a separate status between employment and self-employment and have some, but not all, of the employment rights that limb (a) workers have.

Behind these concise labels there are a lot of multifaceted and moving parts that mean employment rights status can be complicated to establish. The employment status of each individual depends on what actually happens when they are at work – that is, it depends on the terms of the contract, how the working arrangements operate in practice and whether the different legal tests are satisfied.

Given the complexity of the matter, individuals are not always well informed about their own employment status and not always capable of assessing how different factors that make up their employment status apply to them. A direct question asking people to self-categorise themselves might lead to inaccurate estimates.

Drawing on the current body of case law underpinning employment status, a set of questions were designed for this research to test the presence or absence of four features (control, enterprise, integration and personal service) of participants’ working relationship. The type of association that respondents showed to these four features were then used to estimate the number of UK adult aged 18 to 64 who were likely to be workers in April 2023.

The approach taken for this study provided some up-to-date insight on the employment status, socio-demographic characteristics and type of work of those with less typical working practices (the ‘unassigned status workforce’ group) ^{[footnote 24]}. Compared to the ‘assigned status workforce’, the ‘unassigned status workforce’ were more likely to be in the youngest age group (18 to 24) and have an ethnic minority background. Just over half of the individuals in the unassigned working status worked in private organisations and almost a third in public sector organisations. The top three common industries which employed them were: ‘Other services’, ‘Human health and social work activities’ and ‘Education’.

As for employment status, this study estimated that the number of workers in the UK population aged 18 to 64 was between 25.9 and 27.2 million as of April 2023. This total range estimate is made up of two groups of individuals: an estimated 20.6 to 22.0 million limb (a) workers within the assigned status workforce, plus 4.8 to 5.7 million unassigned status workforce individuals who were identified as ‘highly likely’ or ‘fairly likely’ to be a worker.

Using two proxy variables, the study also explored whether the feature analysis categorisation aligned with the employment rights and tax status that seemed to actually be being applied to the individuals in the unassigned status workforce. These analyses are relevant as they help understand if everyone who should be receiving full worker employment rights is receiving them. With regards to rights, using holiday pay as a proxy, it was found that those categorised as ‘highly’ or ‘fairly’ likely to be a worker were more likely than others in the unassigned status workforce to have their engagers applying worker status rights during their employment. Therefore, it is suspected that most individuals were receiving correct treatment for rights purposes due to this alignment with the feature analysis findings, although this research suggests there is still a sizeable minority who may not be. Thus, this slight discrepancy emphasises the work still to be done with regard to ensuring employment rights are clear for both individuals and their engagers. With regards to how tax is paid, when looking at Income Tax and National Insurance contributions, the majority of the ‘highly’ and ‘fairly’ likely workers were seemingly assigned employee tax status. This suggests that most of the people deemed ‘likely to be workers’ via the survey are being treated by their engagers correctly for rights and tax purposes.

Although this research can provide more information about the incidence and characteristics of workers among those with employment statuses that are unable to be identified by the screener questions, there are some limitations that should be considered.

Firstly, the results should always be referred to the survey’s reference period context (April 2023). This area of law is likely to evolve in the next few years, both as a result of the courts’ response to the new working practices amplified by the gig economy (such as zero-hours contracts) and the COVID-19 pandemic, and new legislation. This means that the levels of workers is in flux, so to get updated and reliable estimates the survey instruments used to measure the prevalence of different workforce should continually adapt, thus posing challenges for the evaluation of trends. Further questionnaire content development and cognitive testing is therefore recommended for future iterations of the survey to prevent measurement error. This is especially important for questions used as input for the feature analysis as it is key that the questionnaire stays relevant and responsive to the dimensions that characterise worker status. Please see section 5 of the qualitative report for further discussions surrounding content development.

Secondly, this study focused on the group that makes up the bulk of the tax base: adults in paid work aged 18 to 64. As this doesn’t encompass the entirety of the working population, the estimates found are an underestimate of the total number of workers.

Finally, as the analysis did not cover the full range of possible answer combinations across each feature, the status for a large proportion of individuals remined undetermined. While outside of the scope of this research project, further analysis would be recommended to attempt to classify this group as it makes up a sizeable proportion of the unassigned status workforce (38%). Further research could also be employed to distinguish between limb (a) and limb (b) workers and get more granular insights on the types of work and motivations of less traditional workers such as limb (b) workers.

9. Appendix A. Methodology

9.1 Quantitative Methodology

Questionnaire development

Questionnaire design

The questionnaire design for this study was complex. For respondents, it can be both cognitively difficult and sensitive to answer questions about working practices, particularly for those with a less typical set of working practices. As such, designing a questionnaire with clear wording and accessible terminology that ensures respondents interpreted questions consistently (and in the same way as researchers) was vital.

The Employment Status survey was previously conducted in 2019. The first wave of this survey used face-to-face fieldwork and a questionnaire completed using a CAPI (Computer Administered Personal Interview) system. To develop the questionnaire for the 2023 wave, the previous questionnaire underwent desk review by experienced researchers. The review focused on several key elements. Firstly, the content of the questionnaire was reviewed to ensure that it was up to date, reflected current policies and was relevant for the 2023 research objectives. Secondly, the technical details including routing instructions, text fills and multi-code items were reviewed to ensure they were used appropriately. Finally, the questions themselves were reviewed to ensure they were appropriate for online CAWI^{[footnote 25]} and telephone CATI^{[footnote 26]} interviews, given that the previous wave used a CAPI design.

During the initial questionnaire development meeting with HMRC and after reviewing the questionnaire, it became apparent that some of the questions from the previous wave were not suitable to ask in 2023. NatCen therefore optimised existing survey wording where applicable, and developed new questions on topics such as who individuals were paid by, work type and involvement in the Construction Industry Scheme, with these new questions undergoing cognitive testing. Given the aim of the review (and ultimately the survey) was to create a questionnaire that was easy to understand and accessible, optimisation of the survey took precedent over comparability with the previous wave of the survey. This should be considered when comparing estimates from this wave to previous estimates.

An exception to this was the feature analysis questions and methodology which were kept consistent with the research conducted in 2019. Although for the 2023 study it was a priority to use the same analysis questions and procedures, we would recommend further development of these questions for future research (see Appendix C).

Questionnaire overview

The questionnaire was split into three macro sections: an ‘upfront screening’ for all respondents, the ‘main survey’ for those in the unassigned status workforce and ‘closing questions’ that were asked of all respondents again. An overview of the questionnaire structure is illustrated in Figure 1.

The upfront screening component was made of four smaller parts. Firstly, all respondents answered the ‘paid work screener’, which assessed whether or not the respondent was in any form of paid work during April 2023. Those who were in paid work were asked how many jobs they worked, and they were able to record up to three of their highest paid jobs. Those who reported doing no paid work in April were screened out of the survey, progressing straight to the ‘closing questions’.

Figure 1: Questionnaire structure overview

Figure 1 - Questionnaire structure overview

Those who were in some form of paid work then entered the looped part of the upfront screening, starting with questions about their highest paid job^{[footnote 27]}. This loop was made up of three parts. Firstly, respondents began by answering questions in the ‘categorisation screener’, which asked them about their payment structure, their work type and other employment conditions. The answers to the ‘categorisation screener’ for the job in question then determined whether a respondent would move onto either the ‘limb (a) screener’ or the ‘incorporated and self-employed screener’ for that job. This routing was based upon whether the categorisation screener had deemed them as being more likely to be a limb (a) worker or more likely to be incorporated and self-employed.

These final two screeners were where a job was either classified as an unassigned status workforce job or the assigned status workforce job. If a respondent was deemed to be in the assigned status workforce and they had more than one job, they were sent back to the start of the upfront screening loop (beginning again at the categorisation screener) and were re-asked screening questions that were now for their next highest paid job.

This looping process repeated for each job until either: there were no more remaining jobs to ask a respondent about, or they had answered about three jobs, and so respondents were sent to the ‘closing questions’ section; or one of their jobs met the qualifying criteria for being in the unassigned status workforce and they entered the ‘main survey’.

As soon as a respondent first met the qualifying criteria for being in the unassigned status workforce for one of their jobs, they entered the ‘main survey’. This main section of the questionnaire focused on collecting information about the specific job that a respondent was screened in for, including their working pattern, their contractual obligations and income. Many of these questions were used to assess the four working features that were analysed within the feature analysis (see Appendix B). After completing the ‘main survey’, the unassigned status workforce respondents then passed through the final ‘closing questions’.

Emulating the Labour Force Survey

As a part of the questionnaire development, the 2022 Labour Force Survey (LFS) was used as reference for two key areas: an individual’s work type and tax payment.

Work type

Like the 2023 Employment Status study, the 2019 wave made use of preliminary screening to focus on people within the scope of the survey (those in the unassigned status workforce). Part of that screening included a question adapted from the LFS study about general working arrangements, with several permitted answers. There were some key differences between the questions, as shown in Table 1.

The LFS did not suggest ‘Paid a salary or wage’ by an employer, having instead a separate question to distinguish in general terms between employed and self-employed. Crucially, the 2019 survey only permitted a single answer, whereas the LFS allowed multiple answers, reflecting the potential for overlapping categories. The 2019 survey also allowed an open-ended answer option, but several respondents just wrote in ‘self-employed’.

The 2023 wave of the Employment Status study emulated the LFS design, again with further adaptations. The routing for this question was further tightened: individuals who had earlier said they were employed^{[footnote 28]} (rather than self-employed) were asked a second question to assess whether they were paid by the organisation they worked for, a different organisation, or whether they were a director in a limited company that pays them a salary or dividends. Both those saying self-employed and those employed but who were a director or paid by someone else were then allowed to choose several of the options (as shown in Table 1).

The 2023 survey also added ‘umbrella company’ as an intermediary of particular interest and narrow ‘sub-contractor’ because only the Construction Industry Scheme has tax implications for sub-contractors. Other trades that might be sub-contracted could still find valid options. In line with the LFS, ‘Paid by an employer’ was moved to an earlier question, now tightened to only filter through those who were both employed and paid by the employer for whom the respondent actually worked. There was also additional help link text for respondents that added to code two and three, which is further detailed in Appendix D.

Table 1: Adaptations made to the Labour Force Survey (LFS) ‘work type’ question answer list in the 2019 and 2023 HMRC Employment Status study

Labour Force Survey 2022	Employment Status 2019	Employment Status 2023
MULTICODE	SINGLE CODE	MULTICODE
1. Paid a salary or wage by an employment agency	1. Paid a salary or wage by an employer	1. Paid a salary or wage by an employment agency
2. A sole director of your own limited business	2. Paid a salary or wage by an agency	2. Paid a salary or wage by an umbrella company
3. Running a business or professional practice	3. Director of own limited company	3. A director of my own limited business
4. A partner in a business or professional practice	4. Partner in a business or professional practice	4. Running a business or professional practice
5. Working for yourself	5. Work for myself	5. A partner in a business or professional practice
6. A sub-contractor	6. Work as a sub-contractor	6. Working for myself
7. Or doing freelance work	7. Do freelance work	7. Doing freelance work
8. None of the above	8. I work in some other way (specify)	8. A sub-contractor in the Construction Industry Scheme
		9. None of the above

These adjustments to the ‘employment type’ and ‘work type’ questions brought several advantages to the survey.

Where several options applied, respondents could give them all and researchers could apply a set of rules to decide which was most relevant for the survey.

Individuals working through their own companies often consider themselves self-employed while also being employees of the company. However individuals considered themselves, there was now the option to report that they were directors. Cognitive testing had likewise suggested that it was worth adding an explicit option for directors to the question for employees about who they were paid by.

The alignment between the 2023 answer options and those in the LFS allows researchers to look up answers to many work-related questions in the LFS from people giving the same response to that question in the 2023 employment status survey.

Considerations for future research

Being responsive and adaptive towards questionnaire design will be essential for ensuring that the feature analysis outcomes (further detailed in Appendix B) remain accurate for individuals’ working practices over time. For example, since the research was first conducted in 2019 that has been a paradigm shift in working practices after global events like the COVID-19 pandemic. Key feature analysis questions that rely upon responses on the control over when and where individuals work, for example, now hold a new meaning in the post-pandemic working context that enables greater flexible working. Moreover, employment case law itself, which is used to determine worker status in more ambiguous cases, is by nature an ever-changing area. As more cases go to the courts, new dimensions of working relationships may become more relevant to employment rights status and become areas that need to be asked about. Therefore, since both the employment rights landscape and legislation in the UK is in constant flux, the research used to assess it needs to be continuously adapting in equal measure.

Further discussion of the recommendations for improving the feature analysis methodology beyond questionnaire development is included in Appendix C.

Cognitive testing

Cognitive interviews were used to test the questionnaire ahead of mainstage fieldwork.

Cognitive interviewing uses ‘think aloud’ and probing techniques to give insight into the thought processes respondents go through when answering survey questions. This approach helps researchers identify problems with question wording and questionnaire design by exploring, for example: comprehension of key terms within the questions; whether respondents were able to select a suitable response option; or sensitivity of questions.

A total of 15 interviews were carried out by NatCen researchers. Participants were sampled purposively to cover different less typical work types, as a well as a range of sexes and ages. This is show in ­Table 1. Participants were given a £30 voucher as an incentive for their time and help.

Table 2: Demographic characteristics of the cognitive interview participants

	Characteristics	Number achieved
Gender	Male	7
	Female	8
Age	18-34	4
	35-49	5
	50+	6
Employment type	Self-employed (sole trader only, excludes partnership)	4
	Agency worker	4
	Small business owner (limited company)	3
	Online platform or app-based worker	4
Number of jobs or sources of income	2 or more	12

Base: 15 participants who took part in the cognitive interviews.

To replicate the mode of administration used in fieldwork, a cognitive testing version of the ‘Employment Status Survey’ was programmed, to allow completion of the survey in cognitive testing to occur through CAWI (Computer Assisted Web Interview). Priorities for cognitive testing included checking that the screening questionnaire filtering into the main survey worked as intended, and that all key terms in the survey were understood by participants.

Interviews were recorded and summarised in a thematic matrix alongside participants’ answers to questions and probes, and interviewer’s observations, allowing for the systematic analysis of the qualitative data. Once all cognitive interviews had been completed and analysed, the findings were discussed in depth with HMRC, from which recommendations were made to the survey prior to mainstage fieldwork.

Across all the tested questions, several overarching themes emerged. In the questionnaire, three primary issues were identified that applied to multiple questions. Firstly, within the main questionnaire, participants were not always sure of which job or source of income they had been screened in for, and therefore which job they should answer the main questionnaire based on. Secondly, within the main questionnaire, the tense used within the questions did not always align with the context they were asked in. For example, some questions asked about past experiences but were written in the present tense^{[footnote 29]}. Finally, across all parts of the survey, the language used within questions did not always feel applicable to self-employed individuals, particularly when questions asked employer-related questions.

Following the identification of these issues, several changes were made to the questionnaire. Namely, one key methodological improvement that was made was adding a text banner at the top of every question in the ‘main survey’, using text fill to indicate which job participants’ answers should be based on.

As it was anticipated that most of the questionnaire content would be kept the same as in 2019, only one round of cognitive testing was planned for this study. A second round of cognitive interviews would have allowed to check whether key changes implemented led to the desired improvements in participant understanding. A relevant example is the revision of the definition of ‘job’ at question A3. However, a second round of cognitive testing was not possible due to the budget and time constraints of this study.

Piloting

Piloting the questionnaire aimed to: (i) provide estimates of the incidence rate of respondents being screened into the main survey; (ii) test whether the screener questions were working, creating clean data, and running to the correct length; and (iii) provide insights into how the screener questions were being answered.

The main recommendations that were acted on following the pilot study were:

• to drop the ‘Prefer not to say’ option at A2 and A3, to reduce item missingness due to satisficing
• to add a separate answer code to the question regarding payment that was specifically for directors of limited companies, so they were clear which code to select and were not screened out unintentionally

9.2 Data collection

Sampling

The  NatCen Opinion Panel is a probability-based panel of people from across the UK recruited through the British Social Attitudes (BSA) survey and the Consumer Protection Study (CPS), for which participants are selected at random from the general population using the Postcode Address File (PAF) as a sample frame. 

People interviewed for the BSA or CPS survey were asked whether they wanted to join the panel at the end of survey, and those who agreed were asked to confirm their contact details. Once recruited, all panellists were sent a letter confirming that they had joined the panel, with an information leaflet providing more detailed information on about what being part would involve. Through this procedure the NatCen Opinion Panel sample is refreshed at least once a year

For this wave, all panel members aged 18 to 64 were selected to be issued for a total of 13,449 cases. Overall, the survey was completed by 8,984 UK adults aged 18 to 64. After conducting quality checks to identify potentially fraudulent cases, 217 cases were dropped. This gave a final sample size of 8,767 respondents for this research.

Fieldwork

Fieldwork was conducted using a sequential mixed-mode web and telephone design over a four-week period. Respondents were initially invited to take part online, and web fieldwork ran from 18 May 2023 to 18 June 2023. Those not taking part online were issued to telephone fieldwork which ran from 25 May 2023 to 18 June 2023. Fieldwork resources were also targeted to improve sample quality – for example, participants who have taken part in the past but never taken part online were issued to telephone fieldwork one week early.

Participants were sent multiple invitations and reminders by letters, emails, and text messages to encourage participation. As a thank you for their time, an incentive worth £5 was offered to participants who completed the full interview (upfront screening and main survey), plus to those who went through a long upfront screening process, even if they were eventually screened out for being in the assigned status workforce.  Participants who were screened out of the survey at the very beginning (question A2) due to not being in any form of paid work during April 2023 (see Appendix C) did not receive an incentive. For those who were not screened out at A2 but who were still later screened out for being in the assigned status workforce, NatCen offered to make a £1 donation to the charity National Numeracy on their behalf^{[footnote 30]}.   

Response rates

The response rates for the current survey are summarized in ­Table 3 below. Amongst the panellists invited to take part in the survey, the survey achieved a 65% response rate. When taking account of non-response at BSA and CPS, refusal at recruitment to the NatCen Opinion Panel and subsequent attrition, the estimated overall response rate was 6.4%. 

Table 3: NatCen Opinion Panel response rates at each recruitment stage for the HMRC Employment Status 2023 survey

Recruitment stage	Response
Number of issued panel members	13,449
Number of deadwood cases identified during	2
Number of productive cases	8,984
Survey response rate (%)	65%
Overall response	Response
Estimated number of issued BSA/CPS cases	234,052
Estimated number of deadwood addresses at BSA/CPS stage	23,129
Number of BSA/CPS productive and validated cases	36,397
Number of cases recruited to the NatCen Opinion Panel	25,153
Estimated BSA/CPS response rate	17%
NatCen Opinion Panel recruitment rate (%)	69%
Number of deadwood cases in the NatCen Opinion Panel	187
Estimated overall panel response rate (%)	6.4%

Weighting

All estimates in this report are based on weighted data. A respondent-level weight that adjusts for design, recruitment and non-response bias of the panel sample was used to match the sample to those aged 18 to 64 in the UK population.

Non-response for NatCen’s probability panel surveys can occur at three stages: (i) non-response at the survey used for recruitment (BSA and CPS), (ii) refusal to join the panel at the end of that interview and (iii) non-response in the survey of panel members itself. A weight to account for non-response at each of these three stages was computed. This three-stage system was used because the variables underlying non-response could be different at each stage. The final weight is the product of these three weights.

Recruitment survey weight

Panel members are recruited from the BSA 2015 to BSA 2022, and the 2021 CPS. BSA weights are derived to make the sample of BSA respondents’ representative of the general British population in terms of gender, age and Government Office Region (GOR). CPS weights are derived to make the sample of CPS respondents’ representative of the general Northern Ireland population in terms of gender, age and Government Office Region (GOR).

Panel weight

This weight accounts for nonresponse at the panel recruitment stage. A logistic regression model (weighted by recruitment survey weights) was used to derive the probability of response of each panel member. The resulting panel weight is multiplied by the recruitment survey weights, so the panel is representative of the population.

Survey weight

This weight is to adjust the bias caused by non-response to this panel survey. The final survey weight is the result of multiplying the survey non-response weight by the panel weight.

Respondent-level grossing weight

The final survey weight is scaled to the final responding sample size (n=8,767). The final weight was also ‘grossed’ (scaled up) to produce counts at population-level (estimates for numbers of people with worker status in the UK population), rather than in the responding sample. The grossing weight therefore produces estimates for numbers of people in the UK aged 18 to 64, rather than numbers in the sample.

9.3 Analysis

Rounding and summing percentages

Figures in the report are rounded to the nearest whole percentage point to improve accessibility and to avoid giving a false sense of precision. As a result, figures may not sum to 100%. For some questions participants are also able to select multiple options, so in this scenario figures may also not sum to 100%.

Bases

All reported base sizes (the number of cases on which the analysis is based) are unweighted.

Significance testing

Statistical testing was conducted when comparing estimates between different groups. Our analysis was based on 95% confidence intervals and p-value thresholds of 0.05. Unless specified otherwise, all reported differences are statistically significant.

Where a result is not statistically significant this does not necessarily mean that the difference between groups is not ‘real’; it may also mean that we lacked sufficient statistical evidence to confirm that it is a ‘real’ difference.

Tables

If the value of a cell in a table is less than 0.5 per cent but greater than zero, the cell shows ‘0%’.  An empty cell with an asterisk indicates the figures have been supressed to prevent disclosure (N<25). While the sign ‘–‘ indicates the specific answer was not selected by any respondents.

Tables with non-ordinal dependent variables are ordered by overall prevalence.

10. Appendix B. Feature analysis

10.1 Premise of the feature analysis

The goal of the feature analysis is to categorise and predict individuals’ employment status. In particular, it determines the likelihood that an individual may or may not have ‘worker’ status with regards to their employment rights.

This analysis works by assessing four key features of a working relationship: control, enterprise, integration and personal service. All four of these features have been identified by case law as being central to determining ‘worker’ status.

There are multiple different dimensions that contribute towards any one of these features. In other words, there are several working practices and behaviours that can be helpful in indicating whether a feature may be present or absent in an individual’s work. For every dimension, respondents were asked a corresponding survey question that indicated the extent to which that dimension was a part of their work.

By combining the responses for each dimension for a given feature, the feature analysis then calculates whether an individual was:

• positively associated with a feature: If a set of behaviours indicated that an individual was more likely to be a worker, whether that was limb (a) or limb (b)
• negatively associated with a feature: if a set of behaviours indicated that an individual was less likely to be a worker
• undetermined association with a feature: if a set of behaviours indicated both the presence and absence of a feature, or gave no indication at all, meaning that neither a positive nor negative association could be established

Once the associations with each of the four features has been established, the combinations of all the associations are then used to determine an individual’s likely employment status. All individuals are assigned to one of four outcomes: highly likely to be a worker, fairly likely to be a worker, likely to be self-employed, and undetermined. The criteria for each of these outcomes is outlined in section 1.2 of Appendix B.

10.2 Definitions of the feature groups

Control

• Input variables list: WrkMove; C4, C4Ck, WrkHow; C5, WrkWhen; WrkWhere; SimWrk
  For this feature the input variables come from both screener and main survey sections. For the screener questions, the feature analysis calculations only considered the screener answers that were given for respondents’ qualifying job loop.  For example, if QualJob=1, then consider Job1_C4=1.

Rules for derivation:

Positive:
If at least 3 of the below are selected:
WrkMove =1 OR C4 = 1 OR C4Ck=1
WrkHow =1,2 OR C5 = 1,2
WrkWhen=1
WrkWhere =1
SimWrk =2

AND NOT ((WrkMove =3 OR C4=3 OR C4Ck=3) AND (WrkHow =4 OR C5=4)

Negative:
If at least 3 of the below are selected.
WrkMove =3 OR C4 =3 OR C4Ck =3
WrkHow =4 OR C5 =4
WrkWhen=3
WrkWhere =3

AND NOT (SimWrk =2 OR (WrkHow =1 OR C5=1))

Undetermined:
All other cases not yet categorized

Enterprise

• Input variables list: EssItems; HowPaid; PaidPoorWork; IncrProfit

Rules for derivation:

Positive – if the condition below is met.
((At least 3 of:
EssItems=8
HowPaid=1,3
PaidPoorWrk=2
IncrProfit=2)

OR

(EssItems=9 AND PaidPoorWrk=2 AND IncrProfit=2))

AND NOT EssItems=1,3,6

Negative – if the condition below is met.
((EssItems=1,3,6)

OR

(HowPaid=2 AND PaidPoorWork=1 AND IncrProfit=1))

AND NOT EssItems=8

Undetermined
All other cases not yet categorized

Integration

• Input variables list: LineManag; WrkRules; Notice

Rules for derivation:

Positive – if the condition below is met.
IF (WrkRules=2 AND Notice=2), then at least 3 of the below:
LineManag=1
WrkRules=1,2,3
Notice=1,2

OR

IF ELSE, then at least 2 of the below:
LineManag=1
WrkRules=1,2,3
Notice=1,2

Negative:
If at least two of the below are selected.
LineManag=3
WrkRules=4
Notice=4
AdvMark=1

AND NOT (LineManag=1 OR WrkRules=1,3 OR Notice=1)

Undetermined
All other cases not yet categorized

Personal Service

• Input variables list: PaidElse; PaidElseEver

Rules for derivation:

Positive – if both of the below are selected.
PaidElse=1
PaidElseEver=2

Negative – if both of the below are selected.
PaidElse=2
PaidElseEver=1

Undetermined
All other cases not yet categorized

10.3 Feature Analysis Outcomes:

• Highly likely to be a worker
  All four features are positive (4:0)

• Fairly likely to be a worker
  Three positive features AND no negative features.
  Three positive features AND one negative feature that is not PersonalService.
  Two positive features AND no negative features

• Likely self-employed
  Two negative features AND no positive features
  Three negative features AND if any feature is positive, it must only be PersonalService
  Four negative features

• Undetermined
  Three positive features AND PersonalService is negative.
  Two positive features AND two negative features
  Two positive features AND one negative feature
  Two negative features AND one positive feature
  One positive feature AND one negative feature
  One positive feature AND no negative features
  One negative feature AND no positive features
  No positive features and no negative features

11. Appendix C. Worker population estimates  – limitations and further developments

The feature analysis outcome groups and the screened limb (a) workers in the assigned status workforce were combined to produce a set of estimates for the likely number of workers in the UK by extrapolating the proportions of these groups within the sample to the UK 18 to 64 population. Those in the unassigned status workforce who belonged to the ‘highly likely’ worker group, plus the assigned status workforce limb (a) workers identified in the screener, represented the smallest estimate for workers. By adding in the ‘fairly likely’ worker individuals to this, a greater upper estimate is produced (known in this report as the largest estimate).

It was decided that, for the purposes of this study, the feature analysis methodology and approach to producing population estimates were kept consistent with the previous research conducted in 2019. Although changes to the feature analysis calculations were out of the scope for this 2023 research, we would recommend considering further development work to be carried out to improve the precision and accuracy of the estimates.

The factors contributing towards an overall loss of accuracy and precision in estimates of the number of workers can be broken down into those readily quantifiable using the 2023 data, and those that would require changes to the study design to be resolved. When considering the latter, it is important to try to quantify the potential gain in accuracy, as well as assess whether an increase in accuracy is worth an increase in the complexity of the survey (for example, number of jobs investigated and thus survey loops) or of the analysis. Most factors discussed below are, for example, related to a specific and relatively small subgroup of the population: individuals who have multiple paid jobs with different employment status.

11.1 Quantifiable loss of accuracy or precision in the current approach

Screened limb (a) workers within the unassigned status workforce

In the current methodology, the population estimate begins by counting those who were found to be as a ‘highly likely’ or ‘fairly likely’ worker after the feature analysis. This figure is then topped up by those in the assigned status workforce who were identified as a limb (a) worker during the upfront screening questions, for at least one of their jobs. This approach misses those in the unassigned status workforce who were assessed to be either self-employed or undetermined following the feature analysis but were also identified as a limb (a) worker for one of their higher earning jobs during the survey screener.

Although this is likely to only apply to a small subgroup within the paid workforce population, these individuals should still be considered to improve the accuracy and precision of the estimates. They are ‘known’ workers and could be counted as such. Future research that utilises the feature analysis approach could begin by counting all individuals who were identified as screened limb (a) workers for any of their jobs, in both the assigned and unassigned status workforce. This figure could then be topped up by looking to those in the unassigned status workforce who were not screened as a limb (a) for another job but who have been identified as a ‘highly likely’ or ‘fairly likely’ worker for their unassigned status job.

‘Fairly likely’ workers with other screened limb (a) jobs

All ‘highly likely’ and ‘fairly likely’ workers in the unassigned status workforce were included in both the larger and smaller worker estimate, regardless of their status for their other jobs. However, those determined as being ‘fairly likely’ workers will not be included in the smaller estimate, and some of these ‘fairly likely’ workers could have already been identified as a typical limb (a) worker for one of their other jobs ahead of being asked about their unassigned status workforce job. When producing the smaller estimate this group could be included in the calculation as having a worker status increasing the smaller estimate and the precision of the overall estimates.

11.2 Non-quantifiable loss of accuracy/precision in the current approach

Large proportion of undetermined individuals

Currently, a notable proportion of individuals, in the unassigned status workforce, get left with an ‘undetermined’ employment rights status by the feature analysis. In reality all individuals in paid work have an employment status if they are in paid work and this category of individuals, if further analysed, would likely contain workers, and their exclusion will result in an under-estimate of the total number of workers. Further improvements could be made to the feature analysis outcome categories to reduce the number of individuals who are not assigned to either self-employed, highly likely to be a worker or fairly likely to be a worker.

Unanswered loops for lower-earning jobs

The survey routing screened individuals into the main survey based on their first job that showed a less typical set of working practices and was thus left with an unassigned status by the screening questions. The study therefore, by design, disregards the status of any other jobs individuals might have. Considering the example of an individual with three paid jobs, if their first job qualified them as part of the unassigned status workforce then they would not be asked about their other two lower earning jobs. The subsequent feature analysis may then determine that this person was not a ‘highly likely’ or ‘fairly likely’ worker for that job; however, one of their jobs left unasked about could have been identified as having limb (a) worker status during the screener, or ‘highly likely’ or ‘fairly likely’ status during the follow-up questions.

Too few loops to cover all jobs

The upfront screener only assessed the highest three paid jobs that an individual had during April 2023. This was done to prevent the survey becoming overly-burdensome for affected individuals. Although relatively uncommon, it is likely that some individuals will have had more than three jobs. Theoretically, someone could have been determined as having self-employed or ‘undetermined’ status for their first three jobs but had a further job that had obvious or ‘likely’ worker status.

Age of sampled individuals

This survey had a target population of people aged 18 to 64. Although there was clear rationale for restricting the ages that were sampled for this survey^{[footnote 31]}, this means that there will have been workers in the UK excluded from the sample. This means that population estimates are limited to the number of workers in the UK aged 18 to 64 during April 2023, rather than the total number of workers in the UK overall.

12. Appendix D. Questionnaire specification

13. Appendix D: Questionnaire

This is a modified version of the questionnaire script used in this research and aims to show all the questions that could potentially be asked to participants. Routing of individual questions has been simplified to aid comprehension and in-text explanations of the aims of each section have been added.

Please contact personaltaxes.socialresearch@hmrc.gov.uk if you require the original questionnaire script.

13.1 Screener

Screener A

A1. Which of the following best describes your working status in April?

• if you work variable hours, please answer considering the number of hours worked in April
• if you have more than one type of employment or job, please answer considering the total number of hours you worked across all jobs in April
• if you are self-employed, please answer considering the number of hours worked per week

One answer only allowed

1. full-time paid work (30+ hours per week)
2. part-time paid work (8 to 29 hours per week)
3. part-time paid work (under 8 hours per week)
4. on a paid Government scheme (for example, apprenticeships, kickstart, traineeships, industry placements)
5. on a paid Government scheme (for example, apprenticeships, kickstart, traineeships, industry placements), and also in education
6. in both paid work and education, but not on a Government scheme 7.  not in paid work (for example, unemployed, retired, in education only, permanently sick or disabled)

DK/PNTS NA

A2. Have you done any paid work at all in April?

One answer only allowed

1. yes
2. no

DK/PNTS NA

If an individual had not done any paid work in the last month, then they were screened out and not asked any further questions.

A3. How many different jobs did you have in April?

By ‘job’, we mean any form of work you do that is paid. A ‘job’ can include one type of work with multiple clients (for example: working as babysitter for different families counts as one job).

• please disregard income you receive from any investments you have (for example, property, interest, or shares)
• please include work you get via different apps or organisations

One answer only allowed

1. one job
2. two jobs
3. more than two jobs

DK/PNTS NA

From this question, the number of screener loops required was calculated, with a maximum of 3 loops. Participants with 1 job were asked to complete 1 loop, those with 2 jobs were asked to complete up to 2 loops, and those with more than 2 jobs were asked to complete up to 3 loops. Participants entered the main survey based on the first eligible job, regardless of whether they had other jobs to loop through.

JobLabel: you said you had (“one job”, “two jobs”, “three or more jobs”) in April.

Please assign a short label to (“this job”, “each of these jobs”) so that we can refer to (“it”, “them”) throughout the survey.

Highest paid job in April = [TEXT BOX – char limit 25]  (LABEL1)

Those with two loops were also asked, Second highest paid job in April: [TEXT BOX - char limit 25] LABEL2

Those with three loops were also asked, Third highest paid job in April: [TEXT BOX (character limit of 25)] LABEL3

ERROR MESSAGE IF CHAR LIMIT >25: “Label too long. You can type a maximum of 25 characters. The label doesn’t need to be an exact description of the job. Please assign a short label so that I can remind you which job to think about in the next sections.

Categorisation Screener

Loop section starts

At the beginning of each loop, participants were reminded which job the upcoming set of questions referred to using the assigned label and indicating whether the job was their first, second, or third highest paid.

IF Loop=1: “The next set of questions is about your paid job: LABEL1.”

IF Loop=2, “Thanks for your answers about your highest paid job. The next set of questions is about your second highest paid job: LABEL2.”

IF Loop=3: “Thanks for your answers about your first and second highest paid job. We would now like to ask some questions about your third highest paid job: LABEL3.”

And before every question if the individual was too complete more than one loop, they were reminded to answer about the job referenced at the beginning of the loop eg

“Thinking about your job as: LOOP =2: “LABEL2”; LOOP=3: “LABEL3”

EmpSelf. Did you work as an employee or self-employed?

One answer only allowed

1. employee
2. self-employed

DK/PNTS NA

EmpPaid. How were you paid – or will you be paid – for the work you did in April?

One answer only allowed

1. by the organisation or company I did work for
2. by a different organisation or company (for example, an agency, umbrella company, or my own company)
3. I am a director in a limited company that pays me a salary or dividends

DK/PNTS NA

WrkTypQ. Which of the following options best describe this type of work?

Please select all that apply

1. paid a salary or wage by an employment agency
2. paid a salary or wage by an umbrella company [HELP LINK: An umbrella company is a business often used by recruitment agencies to pay temporary workers. In most cases, the umbrella company employs you and pays your wages through PAYE. It does not find temporary work for you, this is done by the recruitment agency (also known as an ‘employment business’).]
3. a director of my own limited business [HELP LINK:  A director is responsible for running the business and making sure all legal and reporting requirements are met, either by you or someone you hire to do this on your behalf]
4. running a business or professional practice
5. a partner in a business or professional practice
6. working for myself
7. doing freelance work
8. a sub-contractor in the Construction Industry Scheme
9. none of the above [EXCLUSIVE]

DK/PNTS NA

SOFT CHECK: IF WrkTypQ = ‘None of the above’: You said none of the options available describe your work. Please, take a moment to consider the options again before proceeding. Knowing your type of work is fundamental for this survey. HMRC won’t be able to identify you.

IF WrkTypQ = ‘None of the above’, SCREEN OUT for this job and return to the categorisation screener to assess the next highest paid job.

IF only ‘A partner in a business or professional practice’ selected, SCREEN OUT for this job and return to the categorisation screener to assess the next highest paid job.

PaidBy. Which of these paid – or will pay - your earnings to you personally for the work you did in April?

One answer only allowed

1. an employment agency
2. the umbrella company
3. my own limited business
4. the partnership

DK/PNTS NA

If partnership pays earnings, SCREEN OUT for this job. If individual has multiple jobs, return to the categorisation screener to assess the next highest paid job.

A5. Did you get this work and get paid through an online platform or ‘app’ (for example, CitySprint, Uber, Deliveroo, TaskRabbit, Bubblesitter etc.)?

One answer only allowed

1. yes, I got the work and got paid through an online platform or ‘app’
2. no

DK/PNTS NA

A6. In general, when you do work, are payments made to a limited company you own, or in which you have a significant shareholding?

By ‘significant share’ we mean you own more than 5% of the shares.

One answer only allowed

1. yes
2. no

The variable “WrkTyp” was coded as a ‘SINGLE CODE’ variable indicating the type of work of interest for HMRC (i.e. the type of work ‘selected’ for the survey). This variable was computed by considering ‘WrkTypQ’ and ‘PaidBy’. This variable was then used to route individuals into Screener B (for those likely to be employees) or Screener C (for those likely to be self-employed). The answer codes of ‘WrkTyp’ are the same of ‘WrkTypQ’.

Limb (a) or employee screener

The questions in Screener B looked to assign the status limb (a) (employees). If the job could be assigned the status limb (a) (employee), then the individual was screened out for that job and looped back to the categorising screener if needed. If the job was not able to be assigned limb (a) (employee), individuals were screened into the Main Survey.

The questions in this screener aimed to assess whether individuals had limb (a) (employees) status. If the nature of the job indicated limb (a) (employee) status, the individual was screened out for that job and looped back to the categorising screener to assess any other jobs. If the nature of the job did not indicate limb (a) (employee) status, individuals were screened into the ‘Main Survey’ for that job.

B2. Which of these best describes the work you undertake?

One answer only allowed

1. permanent work with no fixed term (i.e. a contract with no end date)
2. seasonal work (for example, work that only occurs at certain times of the year)
3. contract for a fixed period or a fixed task
4. agency temping­
5. casual work (for example, no formal contract, you choose when you work)
6. under a zero-hours contract
7. an apprentice 8.none of the above [EXCLUSIVE]

DK/PNTS NA

SOFT CHECK: IF B2 = ‘None of the above’: “Please take a moment to think again about the type of contract for your (second, third) highest paid job. Please check your answer before proceeding.

B2check. You said the work you undertake is: {“a permanent work with no fixed term (i.e. contract with no end date)”; {“an apprentice”}. Do any of these other options also describe this work?

Please select all that apply

1. yes, Seasonal work (for example, work that only occurs at certain times of the year)
2. yes, Contract for a fixed period or a fixed task
3. yes, Agency temping
4. yes, Casual work (for example, no formal contract, you choose when you work)
5. yes, Under a zero-hours contract
6. no other options describe this work [EXCLUSIVE]

DK/PNTS NA

B3. When you started your work, were you required to provide your National Insurance Number (NINO)?

One answer only allowed

1. yes
2. no
3. I can’t remember

DK/PNTS NA

Incorporated and Self-employed Screener

The questions in this screener  aimed to assess whether individuals had self-employed status. If the nature of the job indicated self-employed status, the individual was screened out for that job and looped back to the categorising screener to assess any other jobs. If the nature of the job did not indicate self-employed status, individuals were screened into the Main Survey for that job.

C1. Does your work mainly involve making or buying goods to sell for your own profit? By ‘making or buying goods to sell’, we mean you make or purchase physical items or products to sell, rather than selling your labour or services.

One answer only allowed

1. yes, it involves mainly making or buying goods to sell for my own profit
2. yes, it involves mainly making or buying good to sell but not for my own profit
3. no, it doesn’t involve making or buying goods to sell (because this work requires my labour or services)

DK/PNTS NA

C2. Does your company/Do you hire and pay people to work for or with you?

Please select all that apply

1. I pay individuals (not family members) to work for or with me
2. I pay family members to work for me or with me
3. I get people to work for or with me, including family, but I do not pay them
4. I do not hire or pay anyone to work for or with me [EXCLUSIVE]

DK/PNTS NA

C3. When your company is/you are paid for work, typically is the payment for labour or services, or for materials or goods?

• labour or services are usually associated with the effort, time and skills of the individual providing the service
• materials or goods could be, for example, purchasing grout as a tiler, cosmetics as a makeup artist, or costs of ingredients as a private chef. This would not include capital assets or one-off purchases such as a car or a laptop/computer used to do the work

One answer only allowed

1. only labour or services
2. mostly labour or services
3. roughly equal split of labour or services and materials or goods
4. mostly materials or goods

DK/PNTS NA

C4. Can the person(s), organisation(s), or client(s) you complete work for move you to work on a different task than the one you originally agreed to do, without renegotiating the amount you are paid?

One answer only allowed

1. yes, they can move me to a different task without renegotiating my pay
2. yes, they can move me to a different task but only if I agree to the renegotiation
3. no, they cannot move me to a different task at all
4. don’t know

DK/PNTS NA

C4Ck. Your answer to this question is very important. Please take a moment to think about it. Can the person(s), organisation(s), or client(s) you complete work for move you to work on a different task than you originally agreed to do, without renegotiating the amount you are paid?

One answer only allowed

1. yes, they can move me to a different task without renegotiating my pay
2. yes, they can move me to a different task but only if I agree to the renegotiation
3. no, they cannot move me to a different task at all
4. I really don’t know

DK/PNTS NA

C5. Once you start the work, can the person(s), organisation(s) or client(s) you complete work for dictate how you perform the work?

For example, require you to follow specific instructions or guidelines to achieve the specific end result (for example, healthcare assistant required to follow a specific process; taxi or delivery driver required to follow a specific route).

One answer only allowed

1. yes – they can decide how the work needs to be done without my input
2. partly – I agree with others how the work needs to be done, or I can decide how to do some parts of the work myself but not other parts
3. no – I decide how the work is done because it’s a highly skilled role
4. no – I decide how all the work is done and don’t need to listen to anyone else’s view if I don’t want to

DK/PNTS NA

C7. Typically, how does your company/do you get paid? Please choose the option which applies most often.

One answer only allowed

1. clients pay directly (including via PayPal, cash, direct bank transfer or cheque)
2. clients pay indirectly (for example, through an agency, or an app or platform such as Deliveroo or Uber that I get work through)

DK/PNTS NA

C8. How many different people, organisations or clients has your company/have you received income from in the last six months?

One answer only allowed

1. 1
2. 2 to 5
3. 6-10
4. 11+

DK/PNTS NA

LOOP SECTION ENDS 

13.2 Main Survey

Where any of an individual’s 3 highest paid jobs couldn’t be assigned Limb (b) (employee) or Self-employed, they entered the ‘Main Survey’. The ‘Main Survey’ aimed to attain more information on four working features: control, enterprise, integration and personal service. The Main Survey concerned only one Job –the first one that couldn’t be assigned a status by the screener. Throughout the survey, participants were reminded which job to answer the questions on and that the survey only concerned this one job.

IntroMain. For the remaining questions please think only about your (second/third) highest paid job which you labelled as (LABEL1/LABEL2/LABEL3)

If you have previously answered about your business: where the questions refer to ‘you’, please think about your business when answering.

InJob. You said you had your job as (LABEL1/LABEL2/LABEL3) in April. As of today, are you still in this job?

1. yes
2. no

DK/PNTA NA

InJobNo.

If an individual was not in the job the main survey concerns, they were presented with a note stating: ‘Some of the next questions use the present tense assuming you are still in this job. Please answer considering your experience for the whole time you have done it.’



As with the screener participants were reminded through which job the survey concerned eg ‘Thinking about your job LABEL1/LABEL2/LABEL3)’

WrkArea. Below is a list of economic activities. Which one best describes the sort of work you do in this job?

One answer only allowed

1. accommodation and food services
2. administrative and support services
3. agriculture, forestry and fishing
4. construction
5. education
6. financial and insurance activities
7. human health and social work activities
8. information and communication
9. manufacturing
10. mining, energy and water supply
11. professional, scientific and technical activities
12. public admin and defence; social security
13. real estate activities
14. transport and storage
15. wholesale, retail and repair of motor vehicles
16. other services

WrkAreaOth. You said the work you do or did in this job is: “Other services”. Please specify the type of services in the box below.

[TEXT BOX]

EngagerSec. Which of the following best describes the person(s), organisation(s) or client(s) you do work for in this job?

Please select all that apply

1. private sector organisations (for example, working for a business)
2. private individuals (for example, providing services in domestic settings, like  plumbing or cleaning )
3. public sector organisations (for example, working for a government-funded organisation , for example, the NHS)
4. third sector organisations (for example, working for charities or non-profit organisations)
5. Other (please specify)

AdvMark. Have you ever generated work through advertising or other marketing, for example, a website, social media (not including LinkedIn) or word of mouth for this job?

One answer only allowed

1. yes
2. No
3. Don’t know

WrkPattern. We would now like to ask you some questions about your work pattern. Which of the following best describes your working pattern for your job?

One answer only allowed

1. fixed or agreed hours
2. I work whenever work is offered
3. I work when I want to accept work
4. I have some other arrangement (please specify)

WeekHrs. How many hours do you work in a typical week for this job?

One answer only allowed

1. less than 4 hours
2. 4 to 8 hours
3. 9 to 15 hours
4. 16 to 20 hours
5. 21 to 30 hours
6. 31 to 40 hours
7. 41 to 50 hours
8. 51 to 60 hours
9. over 60 hours
10. I have no ‘typical’ week – my hours are very irregular

HowLong. When did you first start doing this job?

One answer only allowed

1. less than one month ago
2. between one month to 6 months ago
3. between 6 months and 1 year ago
4. between 1 year and 2 years ago
5. over 2 years ago
6. don’t know

EndWrk. Which, if any, of the following scenarios would likely result in you ending this job?

Please select all that apply

1. less guarantee of work
2. lower take home pay
3. reduced entitlement to employment rights (for example, not getting holiday pay or the national minimum wage)
4. less choice over when or where I can work
5. if I started earning more in my other job(s) than this job
6. none of these would result in me ending this job [EXCLUSIVE]

SimWrk. We would now like to ask you some questions about your employment rights. Are you allowed to take on similar work from other people or organisations while doing this job?

Please think about the people or organisations who you directly do the work for.

One answer only allowed

1. yes
2. no
3. don’t know

HolPay. When on holiday, are you paid by the person(s) or organisation(s) that you do/your company does work for?

By this question, we mean that your contract with the person or organisation that pays you clearly states you are entitled to holiday pay, either as paid annual leave or rolled-up holiday pay.

This includes holiday pay that you receive from your employment agency or umbrella company.

HELP LINK PREVIEW: “What does rolled-up holiday pay mean?” HELP LINK TEXT: “One receives rolled-up holiday pay if their holiday pay entitlement is included in the hourly rate pay they receive for doing the work”.

One answer only allowed

1. yes, I receive holiday pay as paid annual leave
2. yes, I receive rolled-up holiday pay
3. no, I don’t receive any holiday pay
4. don’t know

Notice. Once you have started the work, how much notice does the person(s) or organisation(s) paying you have to give you before terminating your contract so you could no longer do work for them (ignoring gross misconduct)?

One answer only allowed

1. 30 days or more
2. between 1 week and 30 days
3. less than 1 week
4. I have no notice period
5. don’t know

WrkOfferReq. Once you have completed the work that you are/your company is contracted to do, is the person(s) or organisation(s) that you have done the work for required to offer you further work?

By ‘the person(s) or organisation(s) that you do work for’ we mean the agency or company you have the contractual relationship with, rather than the end client.

One answer only allowed

1. yes, they are required to offer further work
2. no, they are not required to offer further work
3. don’t know

WrkAcceptReq. If offered work by the person(s) or organisation(s) that pay you/your company, are you required to accept it?

One answer only allowed

1. yes
2. no
3. don’t know

PaidElse. Can the person(s) or organisation(s) you do work for insist that you do the work yourself, rather than someone else you may choose to do the work?

If this work involves using an online platform or app, the platform or app may insist you do the work yourself by checking your identity before you do the work.

One answer only allowed

1. yes, the person, organisation or online platform/app can insist I do the work myself
2. no, the person, organisation or online platform/app cannot insist I do the work myself
3. don’t know

PaidElseEver. Have you ever paid someone else to do the work you are/your company is contracted to do for this job?

One answer only allowed

1. yes
2. no
3. can’t remember

PaidElseOcc. How often have you paid someone else either to do the work with you, or to do the work instead of you? If unsure, please give your best estimate.

One answer only allowed

1. very rarely (roughly less than 5% of the time)
2. every now and then (roughly between 5% and 25% of the time)
3. often (roughly between 25% and 50% of the time)
4. most of the time (roughly more than 50% of the time)

WrkMove. Can the person(s),organisation(s) or client(s) you complete work for move you to work on a different task than you originally agreed to do, without renegotiating the amount you are paid?

By ‘the person(s) or organisation(s) that you work for’ we mean the agency or company you have the contractual relationship with, rather than the end client.

One answer only allowed

1. yes, they can move me to a different task without renegotiating my pay
2. yes, they can move me to a different task but only if I agree to the renegotiation
3. no, they cannot move me to a different task at all
4. don’t know

WrkHow. Once you start the work, does the person(s) or organisation(s) you do work for dictate how you perform the work?

For example, require you to follow specific instructions or guidelines to achieve the specific end result (for example, builder required to follow specific process, healthcare assistant required to follow a specific process)).

One answer only allowed

1. yes – they can decide how the work needs to be done without my input
2. partly – I agree with others how the work needs to be done, or I can decide how to do some parts of the work myself but not other parts
3. no – I decide how the work is done because it’s a highly skilled role
4. no – I decide how all the work is done and don’t need to listen to anyone else’s view if I don’t want to

WrkWhen. Are you told when you have to work (for example, you have prescribed hours, core hours, minimum hours)?

One answer only allowed

1. yes – I work when I am told to
2. yes – but this is determined by the nature of the work, which has to be done at a specific time, for example, opening hours, standard performance or event times)
3. no – I can work when I want
4. I have some other arrangement [SPECIFY]
5. don’t know

WrkWhere. Are you told where you have to work?

One answer only allowed

1. yes – I work where I am told to
2. yes – but this is determined by the nature of my role and the work has to be done at a specific place (for example, builder has to work on the building site, a nurse who has to work in the hospital)
3. no – I can work where I want
4. I have some other arrangement [SPECIFY]
5. don’t know

WrkRules. When performing your work, which if any of the following applies to you?

Please select all that apply

1. I’m subject to a formal performance management process (for example, performance discussion, end of task/year appraisal, meeting set sales quotas)
2. I have to follow staff rules/guidelines
3. I’m managed by someone else who works for the person(s) or organisation that I work for
4. none of the above [EXCLUSIVE]
5. don’t know [EXCLUSIVE]

LineManag. Do you line manage other individuals when carrying out this job?

By that we mean whether you directly manage and are responsible for the work of other member of staff.

One answer only allowed

1. yes
2. no
3. I don’t work for a business or organization

Uniform. Are you required to wear a uniform that is chosen by the person(s) or organisation(s) you do work for?

One answer only allowed

1. yes
2. no
3. don’t know

EssItems. Which if any of the following essential items do you have to provide to do this job? Please only refer to items that you provide yourself and are not reimbursed for.

Please select all that apply

1. vehicle essential to providing the services (for example, transport or delivery services)
2. vehicle to aid you providing the service (for example, a car for a salesperson to travel between customers, but not including a vehicle required for commuting from home to work)
3. heavy machinery, industrial vehicles (for example, a lorry) or other high-cost specialist equipment
4. phone, tablet or laptop
5. tools or other small equipment
6. materials to complete the task
7. I am required to provide other essential items [SPECIFY]
8. none – the person I do work for provides essential items [EXCLUSIVE]
9. none – no items are required [EXCLUSIVE]
10. don’t know [EXCLUSIVE]

EssItemValue. What is the total estimated value of these items? Please only include the value which you paid for the items themselves, and not any costs associated with running or replacing these items.

One answer only allowed

1. £0 – £1,000
2. £1,001 – £5,000
3. £5,001 - £10,000
4. £10,001 - £25,000
5. £25,001 - £50,000
6. more than £50,000
7. don’t know

HowPaid. Which best describes how you are/your company is paid for this work? If more than one option applies, please select the best description for the majority of your income from this job.

One answer only allowed

1. mainly a fixed amount each pay day
2. mainly a fixed sum for a job/task
3. mainly paid in relation to time worked
4. mainly commission/bonus payment
5. I have some other arrangement (specify)

PaidPoorWrk. If you performed poor quality or incorrect work, are you required to rectify this in your own time without being paid extra or covering any additional costs?

One answer only allowed

1. yes
2. no
3. don’t know

IncrProfit. Are you able to increase your profits by working more efficiently or by reducing costs (for example, through better sourcing of materials, or investing in better equipment)?

One answer only allowed

1. yes
2. no
3. don’t know

IntroTax. We would now like to ask you some questions about taxes. Any information you provide will be anonymised and HMRC will not be able to identify you. Your answers will not be linked to your tax affairs with HMRC.

We are interested in asking these questions because we want to understand more about income on a population level. If you are unsure of any answers, please give your closest estimation.

CorpTax. Have you paid, or will you pay, any corporation tax in relation to the income from this job in the last 12 months ?

HELP LINK: “Corporation Tax is paid on profits from doing business as a limited company, a foreign company within a UK branch or office, a club, co-operative or other unincorporated associated (like a community sports group or club)”

One answer only allowed

1. yes, I have paid, or will pay, corporation tax
2. no, I have not and will not pay corporation tax
3. don’t know

SATax. Do you, or someone else on your behalf, complete a Self Assessment tax return in relation to this job?

One answer only allowed

1. yes, a Self Assessment tax return is completed, or will be completed
2. no, a Self Assessment tax return is not completed, because I pay tax in some other way (for example, PAYE)
3. no a Self Assessment tax return is not completed because I do not earn enough income to pay tax
4. don’t know

SATaxWhy. Why do you, or someone else on your behalf, complete a Self Assessment tax return for this job?

Please select all that apply

1. to pay tax on self-employment income from this job as a sole trader, sub-contractor, or freelancer
2. to pay tax on dividend income from this job as a company shareholder
3. I earn over £100,000 per year
4. for a reason independent of my employment status, such as to report bonus earnings
5. don’t know [EXCLUSIVE]

NITax. Do you pay your own National Insurance and Income Tax for this job, or are these deducted by the organisation(s) you work for (for example, your client, employer, agency)?

One answer only allowed

1. pay both my own National Insurance and Tax
2. pay my own National Insurance, but not my own Tax
3. pay my own Tax, but not my own National Insurance
4. both National Insurance and tax deducted by the organisation(s) I work for (for example: my client, employer, agency)

CISChk. Are you a self-employed subcontractor working under the Construction Industry Scheme for this job?

HELP LINK: “What is the Construction Industry Scheme (CIS)?”; HELP LINK TEXT: “Under the Construction Industry Scheme (CIS), contractors deduct money from a subcontractor’s payments and pass it onto HMRC. The deductions count as advance payments towards the subcontractor’s tax and National Insurance.

Contractors must register for the scheme. Subcontractors do not have to register, but deductions are taken from their payments at a higher rate if they’re not registered.

One answer only allowed

1. yes
2. no
3. don’t know

IntroIncome1. We would now like to ask you some questions about your income. Any information you provide will be anonymised and HMRC will not be able to identify you. Your answers will not be linked to your tax affairs with HMRC.

We are interested in asking these questions because we want to understand more about income on a population level. If you are unsure of any answers, please give your closest estimation.

For all these questions, please answer in relation to your job: (LABEL1/LABEL2/LABEL3)

GrossPay. Approximately, what was the gross pay you received from this job in the last 12 months?

The gross pay is your pay before tax and expenses.

1. {NUMERIC 0-10,000,000}
2. Prefer not to say

GrossPay_Band. Your answers are crucial to understand more about income on a population level.Your answers will not be linked to your tax affairs, and HMRC won’t be able to identify you.

Approximately, what was your gross pay from this job in the last 12 months by band? The gross pay is your pay before tax and expenses.

One answer only allowed

1. up to £5,000
2. over £5,000 but not more than £12,570
3. over £12,570 but not more than £20,000
4. over £20,000 but not more than £30,000
5. over £30,000 but not more than £40,000
6. over £40,000 but not more than £50,270
7. over £50,270 but not more than £60,000
8. over £60,000 but not more than £70,000
9. over £70,000 but not more than £100,000
10. over £100,000 but not more than £125,140
11. over £125,140
12. prefer not to say

PreTaxProfit. Approximately, what was your  total amount of income you were paid after expenses but before tax in the last 12 months?

This is your pre-tax profit, and depending on your circumstances, it might be the same as your gross pay.

1. {NUMERIC 1-10,000,000}
2. Prefer not to say

PreTaxProfit_Band. Your answers are crucial to understand more about income on a population level. Your answers will not be linked to your tax affairs, and HMRC won’t be able to identify you.

Approximately what was your total amount of income you were paid after deducting expenses but before tax for this job in the last 12 months by band?

One answer only allowed

1. up to £5,000
2. over £5,000 but not more than £12,570
3. over £12,570 but not more than £20,000
4. over £20,000 but not more than £30,000
5. over £30,000 but not more than £40,000
6. over £40,000 but not more than £50,270
7. over £50,270 but not more than £60,000
8. over £60,000 but not more than £70,000
9. over £70,000 but not more than £100,000
10. over £100,000 but not more than £125,140
11. over £125,140 but not more than £150,000
12. over £150,000 but not more than £250,000
13. over £250,000
14. prefer not to say

Turnover. Approximately, what was your company’s turnover in the last 12 months? By ‘turnover’ we mean the company’s total income before deducting expenses and tax.

1. {NUMERIC 1-10,000,000}
2. Prefer not to say

Turnover_Band. Your answers are crucial to understand more about income on a population level. Your answers will not be linked to your tax affairs, and HMRC won’t be able to identify you.

Approximately, what was your company’s turnover in the last 12 months by band?

One answer only allowed

1. up to £5,000
2. over £5,000 but not more than £12,570
3. over £12,570 but not more than £20,000
4. over £20,000 but not more than £30,000
5. over £30,000 but not more than £40,000
6. over £40,000 but not more than £50,270
7. over £50,270 but not more than £60,000
8. over £60,000 but not more than £70,000
9. over £70,000 but not more than £100,000
10. over £100,000 but not more than £125,140
11. over £125,140 but not more than £150,000
12. over £150,000 but not more than £250,000
13. over £250,000
14. prefer not to say

PreTaxProfitComp. Approximately, what was your company’s total amount of income after expenses but before tax, not including any reliefs or allowances you may claim as part of your business in the last 12 months?

This is your company’s pre-tax profit.

1. {NUMERIC 1-10,000,000}
2. prefer not to say

PreTaxProfitComp_Band. Your answers are crucial to understand more about income on a population level. Your answers will not be linked to your tax affairs, and HMRC won’t be able to identify you.

Approximately, what was your company’s total amount of income after expenses but before tax, not including any reliefs or allowances you may claim as part of your business in the last 12 months by band?

This is your company’s pre-tax profit.

One answer only allowed

1. up to £5,000
2. over £5,000 but not more than £12,570
3. over £12,570 but not more than £20,000
4. over £20,000 but not more than £30,000
5. over £30,000 but not more than £40,000
6. over £40,000 but not more than £50,270
7. over £50,270 but not more than £60,000
8. over £60,000 but not more than £70,000
9. over £70,000 but not more than £100,000
10. over £100,000 but not more than £125,140
11. over £125,140 but not more than £150,000
12. over £150,000 but not more than £250,000
13. over £250,000
14. prefer not to say

HowPaidSelf. How did you pay yourself in the last 12 months?

One answer only allowed

1. wholly by wages and salary
2. mixed dividends and wages/salary
3. wholly by dividends
4. other [SPECIFY]

PreTaxWage. Approximately what was your pre-tax wage from this job in the last 12 months, excluding dividends?

1. {NUMERIC 1-10,000,000}
2. Prefer not to say

PreTaxWage_Band. Your answers are crucial to understand more about income on a population level. Your answers will not be linked to your tax affairs, and HMRC won’t be able to identify you.

Approximately, what was your pre-tax wage from this job in the last 12 months, excluding dividends, by band?

One answer only allowed

1. up to £5,000
2. over £5,000 but not more than £12,570
3. over £12,570 but not more than £20,000
4. over £20,000 but not more than £30,000
5. over £30,000 but not more than £40,000
6. over £40,000 but not more than £50,270
7. over £50,270 but not more than £60,000
8. over £60,000 but not more than £70,000
9. over £70,000 but not more than £100,000
10. over £100,000 but not more than £125,140
11. over £125,140 but not more than £150,000
12. over £150,000 but not more than £250,000
13. over £250,000
14. prefer not to say

PaymentPrd. We would now like to ask you about the period that your most recent income from this job covered.

Would you like to answer this in weeks or in months?

One answer only allowed

1. weeks
2. months

PaymentWk. What period did your most recent payment cover?

[NUMERIC] WEEK(S)

PaymentMth. What period did your most recent payment cover?

[NUMERIC] MONTH(S)

PreTaxDiv. Approximately, what was your income from dividends pre-tax from this job in the last year?

1. {NUMERIC 1-10,000,000}
2. prefer not to say

PreTaxDiv_Band. Your answers are crucial to understand more about income on a population level. Your answers will not be linked to your tax affairs, and HMRC won’t be able to identify you.

Approximately what was your pre-tax dividend income from this job in the last 12 months by band?

One answer only allowed

1. up to £5,000
2. over £5,000 but not more than £12,570
3. over £12,570 but not more than £20,000
4. over £20,000 but not more than £30,000
5. over £30,000 but not more than £40,000
6. over £40,000 but not more than £50,270
7. over £50,270 but not more than £60,000
8. over £60,000 but not more than £70,000
9. over £70,000 but not more than £100,000
10. over £100,000 but not more than £125,140
11. over £125,140 but no more than £150,000
12. over £150,000
13. prefer not to say

SHARES. What percentage of the shares do you currently control in the company you are a director of/your clients make payments to/you are a director of and that your clients make payments to?

One answer only allowed

1. less than 5%
2. 5% to 25%
3. 25% to 50%
4. over 50%
5. don’t know

This next set of questions was asked to both those screened out at the screener stage and those screened into the main survey. Slightly different intros to this section were given depending on whether the individual had experienced the full survey.

**IntroIncome2. We would now like to ask you some questions about your income. Any information you provide will be anonymised and HMRC will not be able to identify you. Your answers will not be linked to your tax affairs with HMRC.

We are interested in asking these questions because we want to understand more about income on a population level. If you are unsure of any answers, please give your closest estimation.

These questions are about all your earnings.

GenIntro. The questions specific to your (second/third) highest paid job which you labelled as (LABEL1/LABEL2/LABEL3) have now ended.

The last three questions are about all your earnings.

GrossPay_Year. What was your gross income, that is your pay before tax and expenses, you received (including all earnings from work, investments, benefits, and rental income) in the last 12 months?

One answer only allowed

1. up to £5,000
2. over £5,000 but not more than £12,570
3. over £12,570 but not more than £20,000
4. over £20,000 but not more than £30,000
5. over £30,000 but not more than £40,000
6. over £40,000 but not more than £50,270
7. over £50,270 but not more than £60,000
8. over £60,000 but not more than £70,000
9. over £70,000 but not more than £100,000
10. over £100,000 but not more than £125,140
11. over £125,140 but no more than £150,000
12. over £150,000
13. prefer not to say

Benefits. Are you, or any spouse or partner you live with, currently receiving any of the following state benefits: Universal Credit, Working Tax Credits or Child Tax Credits?

One answer only allowed

1. yes
2. no
3. don’t know
4. prefer not to say

Emp12Months. Of the last 12 months, how many calendar months were you working in any paid job?

[NUMERIC, LIMIT 0-12] month(s)

SOFTCHECK: “You previously said you did no paid work in April and therefore have worked less than 12 calendar months in the past 12 months. Please check your answer before continuing.

SOFTCHECK: “You previously said you did paid work in April. Therefore, for at least 1 calendar month in the past 12 months you were in some form of paid job. Please check your answer before continuing.

«NatCen Opinion Panel - standard demographic questions» 

For information on this section, please email OpinionPanel@natcen.ac.uk or call 020 7250 1866.

End of Survey.

---

1. Due to the way they are calculated, these figures may systematically underestimate the size of the overall worker population. Please refer to Chapter 4 and the technical report for more information. ↩

2. It is important to note that the 2019 study was run among UK adults aged 16 or over, unlike this study which was run with UK adults aged 18 to 64. In addition, the 2019 study had a fieldwork screening period of three months during late 2018 and early 2019, rather than the one month across May and June 2023 in this study. ↩

3. The growth of food delivery services such as Deliveroo and Just Eat have been a key industry towards this change in working practices and propelling the growth of the gig economy. (Source: Transport Planning Society, Food for thought: The rise of on-demand food delivery services and growing need to switch these journeys from motors to muscle, a case study of London, December 2019) ↩

4. The automatic inclusion of ‘partners in a professional business or practice’ into the assigned workforce is a change since the 2019 study. ↩

5. Vacancies and jobs in the UK - Office for National Statistics (ons.gov.uk) ↩

6. The survey achieved a 65% response rate among those panellists invited to participate, However, when accounting for all stages of non-response, including participation in the recruitment survey and recruitment to the NatCen Opinion Panel, the overall response rate for the NatCen Opinion Panel was 6.4%. More information on this is available in Appendix A. ↩

7. SIC code definitions were not explained to respondents. Due to the varied nature of the service sector, respondents who selected ‘Other Service Activities’ were asked to provide a free-text explanation of their job, which included responses such as ‘Creative services’, ‘Garden maintenance’, and ‘Cleaning’. ↩

8. The four features mentioned in this report may be referred to using alternatives terms in other literature. ↩

9. This included scenarios where participants genuinely did not know the answer to the question or selected an answer option which did not discriminate enough for the given feature. ↩

10. This status categorisation is developed further in Chapter 4. ↩

11. The value of the essential items used to complete an individual’s work was not an input of the Feature Analysis but instead provides some additional contextual detail. ↩

12. Employees with continuous employment of at least one month but less than two years are entitled to at least one week’s notice from the employer. ↩

13. This particular element is only used in the criteria for negative associations, rather than assessing both positive and negative associations. ↩

14. In a scenario where the individual has said that their engager cannot force them to perform the work themselves, they are implying that they are able to pay someone else to do it. However, this unrestricted right remained untested for any who have not actually done this. Therefore, it was crucial to ask a follow-up question that could assess the reality of this claim in practice. ↩

15. The frequency of which individuals had paid someone else to do their work was not an input of the Feature Analysis but instead provides some additional contextual detail. ↩

16. This can be either as a limb (a) (employee) or limb (b) worker. Please see the glossary for further information on the definitions of these categories. ↩

17. The determining attributes of these four features have been outlined in detail in chapter 3. ↩

18. For tax purposes the working population is split into two categories: employed or self-employed. Please note that this differs from employment status for rights purposes, which is made up of three separate categories: limb (b) worker, limb (a) worker (employee) and self-employed. Please refer to the glossary for further information on the definitions of each of these categories. ↩

19. For further detail on why a negative association with personal service has been excluded from this condition, refer to section 3.4. ↩

20. An individual was only categorised as likely to be self-employed when they had one positive feature if this positive feature was personal service. ↩

21. Consideration should be made for the fact that there also may be more ‘workers’ outside of the 18 to 64 age category since the working age in the UK begins at 16. This age of working extends onwards until an individual chooses to retire, which may be older than 64 for some individuals. Therefore, there may be more workers in the UK than the estimates within this report suggest. ↩

22. This is in keeping with limitations during the questionnaire development phase for redesigning the existing feature analysis questions. Rationale behind this decision can be found in the accompanying technical report. ↩

23. In some scenarios, an individual may have paid their own Income Tax but not National Insurance contributions (NI). For example, those above the state pension age do not pay NI but do pay Income Tax if they are earning above the Personal Allowance. Conversely, an individual may have paid their own NI but not Income Tax. For example, an individual on lower pay may exceed the Class 2 NI threshold but not reach the Personal Allowance for Income Tax. In 2022, the government made changes to more closely align the thresholds at which individuals start paying Income Tax and NI, starting from the tax year 2023 to 2024 onwards. As the Self Assessment deadline for work undertaken from 5 April 2023 will be in 2025, some of the scenarios reported in Table 7 may no longer apply when individuals come to pay tax by Self Assessment for paid work undertaken from April 2023. ↩

24. The ‘unassigned status workforce’ does not describe individuals with no employment rights status in real life but instead describes those whose employment status for rights purposes was unable to be determined by the survey screening questions. All those in paid work have an employment rights status in real life. For further detail on the definitions for these target populations, see the ‘populations of interest’ section in the introduction and the glossary. ↩

25. ‘CAWI’ stands for Computer Automated Web Interview. ↩

26. ‘CATI’ stands for Computer Automated Telephone Interview. ↩

27. One key methodological limitation that should be mentioned here is that this approach meant that a respondent would go on to only answer they ‘main survey’ questions for their highest paid ‘unassigned status workforce’ job. Some respondents may have had other qualifying jobs that were not asked about because they had already been screened in for being in the unassigned status workforce for a higher paid job. ↩

28. See question ‘EmpSelf’ in Appendix D. ↩

29. In particular, specific considerations were made for respondents who may no longer be in the unassigned status workforce job that they were in during April 2023 during the time of interview. ↩

30. National Numeracy was chosen independently by NatCen without the involvement of HMRC. NatCen selected this charity because the objectives of the two organisations were complementary. National Numeracy’s mission is ‘to improve how people understand and work with numbers’, which supports with the quantitative research work carried out at NatCen. ↩

31. The NatCen Opinion Panel only includes individuals aged 18 and above, thereby restricting the lower age limit. The upper bound was then set at 64 to concentrate the sample on the age groups who form the majority of the tax base and accordingly ensure enough members of these age groups were surveyed. ↩