where have 600,000 trade union members gone? using the ... · margaret pargeter ; and paul sellers,...

38
Where have 600,000 Trade Union Members Gone? Using the Labour Force Survey to Estimate Changes in Trade Union Membership between 2001 and 2015 By Roger Sealey The author can be contacted at: [email protected] © 2018

Upload: buidung

Post on 26-Apr-2018

214 views

Category:

Documents


1 download

TRANSCRIPT

Where have 600,000 Trade Union Members Gone? Using the LabourForce Survey to Estimate Changes

in Trade Union Membershipbetween 2001 and 2015

By Roger Sealey

The author can be contacted at:[email protected]

© 2018

able of Contents 1

1.0 Acknowledgements 5 1.1 Acknowledgements 1.2 About the author

2.0 Executive Summary 6

3 The Labour Force Survey 7 3.1 Background to the Labour Force Survey 7

3.2 Imputation of does not apply responses 8

3.3 Sampling Variance and the LFS 9

4.0 Overview of Change in Trade Union Membership 12 4.1 Introduction 9

4.2 Changes in Trade Union Density and Composition 141

4.3 Change in Gender 152

4.4 Full-time vs Part-time 163

4.5 Establishment Size 17

4.6 Industry 18

4.6.1 Manufacturing 16 4.7 UK Regions 21

4.8 Occupation 26

4.9 Change in trade union membership due to age 29

4.10 Public vs Private Sector 31

4.11 Trade Union Coverage in the Private and Public Sector 31

4.12 Trade Union Presence 32

5.0 Conclusion 33 5.1 Overview 33

5.2 Change in trade union membership 33

5.3 Where trade union density is higher 34

6.0 Bibliography 36

1

Table of Contents 1

1.0 Acknowledgements 51.1 Acknowledgements 51.2 About the author 5

2.0 Executive Summary 6

3 The Labour Force Survey 73.1 Background to the Labour Force Survey 73.2 Imputation of does not apply responses 83.3 Sampling Variance and the LFS 9

4.0 Overview of Change in Trade Union Membership 124.1 Introduction 94.2 Changes in Trade Union Density and Composition 144.3 Change in Gender 154.4 Full-time vs Part-time 164.5 Establishment Size 174.6 Industry 184.6.1 Manufacturing 214.7 UK Regions 214.8 Occupation 264.9 Change in trade union membership due to age 294.10 Public vs Private Sector 314.11 Trade Union Coverage in the Private and Public Sector 314.12 Trade Union Presence 32

5.0 Conclusion 335.1 Overview 335.2 Change in trade union membership 335.3 Where trade union density is higher 34

6.0 Bibliography 36

1

Tables

Table 1: Variables in the LFS that may be or may have been, affected by non-responding DNAs

Table 2: Outcome for Different Methods of Imputation for Trade Union Membership '000.

8

9

Table 3: Trade Union density and trade union membership levels. In employment, 2015. All in employment in the United Kingdom. 10

Table 4: Change in Trade Union Membership by gender and employment status 2001-2015. 13

Table 5: Change in trade union membership by gender 2001-15. 14

Table 6: Change in gender union density and employment 2001-2015. 15

Table 7: Trade union density and employment composition 2001 and 2015. 15

Table 8: Change in full-time and part-time union density and employment 2001-2015. 16

Table 9: Trade union density and employment composition 2001 and 2015. 17

Table 10: Change in establishment size union density and employment 2001-2015. 17

Table 11: Trade union density and employment composition 2001 and 2015 by establishment size. 17

Table 12: Change trade union membership and employees by industry between 2001 and 2015. 18

Table 13: Trade union density and employment composition 2001 and 2015 by industry. 19

Table 14: Change in trade union membership and employment between 2001 and 2015 by manufacturing and services. 20

2

Tables

Table 1: Variables in the LFS that may be or may havebeen, affected by non-responding DNAs 8

Table 2: Outcome for Different Methods of Imputation forTrade Union Membership ‘000. 9

Table 3: Trade Union density and trade union membershiplevels. In employment, 2015. All in employment in theUnited Kingdom. 10

Table 4: Change in Trade Union Membership by gender andemployment status 2001-2015. 13

Table 5: Change in trade union membership by gender2001-15. 14

Table 6: Change in gender union density and employment2001-2015. 15

Table 7: Trade union density and employment composition2001 and 2015. 15

Table 8: Change in full-time and part-time union density andemployment 2001-2015. 16

Table 9: Trade union density and employment composition2001 and 2015. 17

Table 10: Change in establishment size union density andemployment 2001-2015. 17

Table 11: Trade union density and employment composition2001 and 2015 by establishment size. 17

Table 12: Change trade union membership and employeesby industry between 2001 and 2015. 18

Table 13: Trade union density and employment composition 2001and 2015 by industry. 19

Table 14: Change in trade union membership and employmentbetween 2001 and 2015 by manufacturing and services. 20

2

Table 15: The change in trade union density and employment composition for manufacturing and service sectors 2001 2015. 20

Table 16: Change in manufacturing trade union membership by two figure SIC between 2001 and 2015 21

Table 17: Change in manufacturing employment by two figure SIC between 2001 and 2015. 22

Table 18: Change in manufacturing trade union density by two figure SIC between 2001 and 2015. 23

Table 19: Change in trade union membership and employment between 2001 and 2015 by UK Regions. 24

Table 20: Union density by region 2001-15 ranked for highest trade union density in 2001. 20

Table 21: The change in trade union density and employment composition by UK regions 2001 and 2015. 25

Table 22: Change in trade union membership and employees by North and South of England between 2001 and 2015. 25

Table 23: The change in trade union density and employment composition by North and South regions in England 2001 and 2015. 26

Table 24: The change in trade union density and employment by occupation 2001 and 2015. 26

Table 25: The change in trade union density and employment composition by occupation 2001 and 2015. 27

Table 26: The change in trade union density and employment composition by occupation 2001 and 2015. Source: Oct-Dec 2001 and 2015. 28

Table 27: The change in trade union membership and employment by manual and non-manual occupations 2001 and 2015 28

Table 28: The change in trade union density and employment composition for manual and non-manual occupations 2001 and 2015 28

Table 29: The change in trade union membership and employment by age groups 2001 and 2015. 29

3

Table 15: The change in trade union density and employmentcomposition for manufacturing and service sectors 20012015. 20

Table 16: Change in manufacturing trade union membershipby two figure SIC between 2001 and 2015 21

Table 17: Change in manufacturing employment by two figureSIC between 2001 and 2015. 22

Table 18: Change in manufacturing trade union density by twofigure SIC between 2001 and 2015. 23

Table 19: Change in trade union membership and employmentbetween 2001 and 2015 by UK Regions. 24

Table 20: Union density by region 2001-15 ranked for highest tradeunion density in 2001. 20

Table 21: The change in trade union density and employmentcomposition by UK regions 2001 and 2015. 25

Table 22: Change in trade union membership and employees byNorth and South of England between 2001 and 2015. 25

Table 23: The change in trade union density and employmentcomposition by North and South regions in England 2001and 2015. 26

Table 24: The change in trade union density and employmentby occupation 2001 and 2015. 26

Table 25: The change in trade union density and employmentcomposition by occupation 2001 and 2015. 27

Table 26: The change in trade union density and employmentcomposition by occupation 2001 and 2015. Source:Oct-Dec 2001 and 2015. 28

Table 27: The change in trade union membership and employmentby manual and non-manual occupations 2001 and 2015 28

Table 28: The change in trade union density and employmentcomposition for manual and non-manual occupations2001 and 2015 28

Table 29: The change in trade union membership and employmentby age groups 2001 and 2015. 29

3

Table 30: The change in trade union density and employment composition by age groups 2001 and 2015

Table 31: Change in core working age group 2001-15

Table 32: The change in trade union membership and employment by sector 2001 and 2015.

30

31

31

Table 33: The change in trade union density and employment composition for private and public sectors 2001 and 2015 31

Table 34: Trade union coverage and employees in the private and public sector 2001 and 2015. 32

Table 35: The change in trade union coverage and employment composition for private and public sectors 2001 and 2015. 32

Table 36: Trade union presence and employees in the private and sector 2001 and 2015. 33

Table 37: The change in trade union presence and employment composition for private and public sectors 2001 and 2015. 33

Table 38: Change in trade union membership by selected categories 2001-15. 34

Table 39: Summary of results of analysis 34

Figures

Figure 1: Trade union membership totals by Certification Officer and BEIS Employees. 12

Figure 2: Employee trade union members by gender 2001-15 13

4

Table 30: The change in trade union density and employmentcomposition by age groups 2001 and 2015 30

Table 31: Change in core working age group 2001-15 31

Table 32: The change in trade union membership and employmentby sector 2001 and 2015. 31

Table 33: The change in trade union density and employmentcomposition for private and public sectors 2001and 2015 31

Table 34: Trade union coverage and employees in the private andpublic sector 2001 and 2015. 32

Table 35: The change in trade union coverage and employmentcomposition for private and public sectors 2001 and 2015. 32

Table 36: Trade union presence and employees in the private andsector 2001 and 2015. 33

Table 37: The change in trade union presence and employmentcomposition for private and public sectors 2001 and 2015. 33

Table 38: Change in trade union membership by selectedcategories 2001-15. 34

Table 39: Summary of results of analysis 34

Figures

Figure 1: Trade union membership totals by Certification Officerand BEIS Employees. 12

Figure 2: Employee trade union members by gender 2001-15 13

4

1.0 Acknowledgements

1.1 Acknowledgements

I am grateful to the following for their help and advice with this paper: Ivan Bishop, Statistician, Labour Market Directorate, Department for Business, Innovation & Skills; Dr Sharon Bolton, Data Curation Manager, UK Data Ser►ice (UCL); Professor Alex Bryson, Department of Social Science, University College London; John Forth, Fellow, NIESR; Phil Fry, Office of National Statistics, Newport; Professor Francis Green, UCL Institute of Education, University College London; Da►id Horsfield, Geoff Pargeter, Margaret Pargeter ; and Paul Sellers, Policy Officer, Trades Union Congress.

1.2 About the author

Roger Sealey was a transport researcher for the Transport and General Workers' Union, now Unite the Union, between 2000 and 2009, when he retired.

His currently research interests are in labour market issues and logistics.

Education: Ruskin College Oxford 1974-76, Labour Studies Diploma. Oxford Polytechnic — Now Oxford Brookes University — 1988-91 BSc Economics and Accounting and Finance. He was awarded a PhD in economics in 2003 from Oxford Brookes University.

He can be contacted at [email protected].

5

1.0 Acknowledgements

1.1 Acknowledgements

I am grateful to the following for their help and advice with this paper: Ivan Bishop, Statistician, Labour Market Directorate, Department for Business, Innovation & Skills; Dr Sharon Bolton, Data Curation Manager, UK Data Service (UCL); Professor Alex Bryson, Department of Social Science, University College London; John Forth, Fellow, NIESR; Phil Fry, Office of National Statistics, Newport; Professor Francis Green, UCL Institute of Education, University College London; David Horsfield, Geoff Pargeter, Margaret Pargeter ; and Paul Sellers, Policy Officer, Trades Union Congress.

1.2 About the author

Roger Sealey was a transport researcher for the Transport and General Workers’ Union, now Unite the Union, between 2000 and 2009, when he retired.

His currently research interests are in labour market issues andlogistics.

Education: Ruskin College Oxford 1974-76, Labour Studies Diploma.Oxford Polytechnic – Now Oxford Brookes University – 1988-91 BScEconomics and Accounting and Finance. He was awarded a PhD ineconomics in 2003 from Oxford Brookes University.

He can be contacted at [email protected].

5

2.0 Executive Summary

This paper uses data from the UK Labour Force Survey to examine changes in trade union membership of employees between 2001 and 2015 in the United Kingdom.

The paper starts by explaining the methodology used to calculate trade union membership used by BIS. It then explains the importance of using imputation when calculating trade union membership. It then explains the sampling variance and how this needs to be taken into account when using the LFS

It then looks at changes in trade union membership using trade union density and the composition of the workforce for a number of variables. These variables are: gender; full-time vs part-time; establishment size; industry; UK regions; occupations; changes due to age; private vs public sector; trade union coverage; and trade union presence.

However, it warns against relying solely on using trade union density to explain changes in trade union membership.

The initial analysis showed that the manufacturing sector was most effected by loss of union membership and employment. Therefore this sector was analysed in greater detail. This analysis showed that 637,420 union members were lost from this sector, and 1.2 million jobs were also lost during the same period — see 4.6.1 Manufacturing

The paper concludes by stating that we can say the decline in union membership came primarily from the core working age group and the manufacturing occupations.

6

2.0 Executive SummaryThis paper uses data from the UK Labour Force Survey to examine changes in trade union membership of employees between 2001 and 2015 in the United Kingdom.

The paper starts by explaining the methodology used to calculate trade union membership used by BIS. It then explains the importance of using imputation when calculating trade union membership. It then explains the sampling variance and how this needs to be taken into account when using the LFS

It then looks at changes in trade union membership using trade union density and the composition of the workforce for a number of variables. These variables are: gender;full-time vs part-time; establishment size; industry; UK regions; occupations; changes due to age; private vs public sector; trade union coverage; and trade union presence.

However, it warns against relying solely on using trade union density to explain changes in trade union membership.

The initial analysis showed that the manufacturing sector was most effected by loss of union membership and employment. Therefore this sector was analysed in greaterdetail. This analysis showed that 637,420 union members were lost from this sector,and 1.2 million jobs were also lost during the same period – see 4.6.1 Manufacturing

The paper concludes by stating that we can say the decline in union membership came primarily from the core working age group and the manufacturing occupations.

6

3 The Labour Force Survey

3.1 Background to the Labour Force Survey

The United Kingdom Office of National Statistics (ONS) in an information paper described the primary purpose of the Labour Force Survey (LFS) as "providing good quality point in time and change estimates for various labour market outputs and related topics" (ONS: 2015a:1). In this case the "labour market" covers all aspects of people's work, including the education and training needed to equip them for work, the jobs themselves, job-search for those out of work, and income from work and benefits. There are three economic activity groups — employed [either employee or self-employed], and we use employee in this paper, unemployed and economically inactive - used in the LFS and are in accordance with the standard International Labour Organisation (ILO) definitions.

One of the strengths of the LFS is that it has the largest sample size of any household survey in the UK, and can therefore generate robust statistics at regional level. In addition, the sampling errors are small, because the LFS has a single-stage, random sample of addresses. The survey covers a large range of employment-related variables and non-employment-related variables, allowing cross-linking analyses to be undertaken (eg, earnings against educational attainment)' (ONS: 2015a:3).

A limitation of the LFS is that the sample design provides no guarantee of adequate coverage of any industry, as the survey is not stratified by type of industry. The LFS coverage also omits communal establishments, excepting NHS housing and students in boarding schools and halls of residence. Members of the armed forces are only included if they live in private accommodation. In this paper we exclude members of the armed forces in line with the Department for Business Innovation and Skills (BIS) methodology (BIS 2015a:p55). Also, workers under 16 years of age are not included.

The sample size of the LFS is approximately 40,000 responding UK households per quarter. The survey is undertaken quarterly and each quarter's sample is made up of five waves. Respondents are interviewed for five successive waves at three-monthly intervals. Approximately 20% of the sample is replaced every quarter. The LFS is intended to be representative of the entire population of the UK. In this paper we use the data from the October — December LFS between 2001 and 2015. The reason for this that the October — December is the only calendar quarter where the trade union membership question is asked.

The LFS uses a system of calibration weighting. The weights are formed using a population weighting procedure which involves weighting data to sub-regional population estimates and then adjusting for the estimated age and sex composition by region (the income variables have their own weights).

If a household is unavailable for interview, but was interviewed in the previous wave, responses from the previous wave are rolled forward. This is referred to as imputation. Imputation is a 'process for estimating missing values using the non-missing information available for a subject' (Everitt 2002:186). Imputation is carried out to minimise non-response bias in estimates, while simultaneously improving

7

3 The Labour Force Survey

3.1 Background to the Labour Force Survey

The United Kingdom Office of National Statistics (ONS) in an information paper described the primary purpose of the Labour Force Survey (LFS) as "providing good quality point in time and change estimates for various labour market outputs and related topics” (ONS: 2015a:1). In this case the “labour market” covers all aspects of people's work, including the education and training needed to equip them for work, thejobs themselves, job-search for those out of work, and income from work and benefits.There are three economic activity groups – employed [either employee or self-employed], and we use employee in this paper, unemployed and economically inactive- used in the LFS and are in accordance with the standard International Labour Organisation (ILO) definitions.

One of the strengths of the LFS is that it has the largest sample size of any household survey in the UK, and can therefore generate robust statistics at regional level. In addition, the sampling errors are small, because the LFS has a single-stage, random sample of addresses. The survey covers a large range of employment-related variables and non-employment-related variables, allowing cross-linking analyses to be undertaken (eg, earnings against educational attainment)’ (ONS: 2015a:3).

A limitation of the LFS is that the sample design provides no guarantee of adequate coverage of any industry, as the survey is not stratified by type of industry. The LFS coverage also omits communal establishments, excepting NHS housing and students in boarding schools and halls of residence. Members of the armed forces are only included if they live in private accommodation. In this paper we exclude members of the armed forces in line with the Department for Business Innovation and Skills (BIS) methodology (BIS 2015a:p55). Also, workers under 16 years of age are not included.

The sample size of the LFS is approximately 40,000 responding UK households per quarter. The survey is undertaken quarterly and each quarter’s sample is made up of five waves. Respondents are interviewed for five successive waves at three-monthly intervals. Approximately 20% of the sample is replaced every quarter. The LFS is intended to be representative of the entire population of the UK. In this paper we use the data from the October – December LFS between 2001 and 2015. The reason for this that the October – December is the only calendar quarter where the trade union membership question is asked.

The LFS uses a system of calibration weighting. The weights are formed using a population weighting procedure which involves weighting data to sub-regionalpopulation estimates and then adjusting for the estimated age and sex composition byregion (the income variables have their own weights).

If a household is unavailable for interview, but was interviewed in the previous wave, responses from the previous wave are rolled forward. This is referred to as imputation. Imputation is a ‘process for estimating missing values using the non-missing information available for a subject’ (Everitt 2002:186). Imputation is carried out to minimise non-response bias in estimates, while simultaneously improving

7

precision by boosting the sample size. The rationale is that most LFS variables do not change from one quarter to another for most people. Responses are rolled forward for one wave only. Data is not rolled forward after a second consecutive non-response. In the_LFS Performance and Quality Monitoring Report (PQM) tables and charts (at person or household level) containing responses which have been rolled forward from the previous wave are denoted by the term 'including imputed'. Tables and charts which do not contain responses which have been rolled forward from the previous wave are denoted by the term 'excluding imputed' (ONS 2015a:5).

3.2 Imputation of does not apply responses

When analysing the LFS we may find that the does not apply (DNA) category may be exceptionally large. This can be due to the question not being asked in every quarter, a non-core question e.g. trade union member. And some respondent are not contacted in every survey. Table 1 gives a list of variables in the LFS that may be or may have been, affected by non-responding DNAs.

Examination of Table 1 identifies three LFS variables that we will use in this paper. They are: TUCOV, TUPRES, and UNION.

ACCDAY4 EDINS11 MATLVE SMOCCT TRONJB WCHJB ACCURH(1-4) FEEIR(1-5) NOBACK9 SMSOCIO TRVDRV WCHJB3 AWARE GOBCK9 NOCUST SMSOC101 TRVMTH BANK HOLS NUMILL SMSOC103 TRVTME BHNOTA HOLSB QAPL11 SMSOC104 TUCOV BHNOTB ILCURR REASOFF9 TFEE10(1-5) TUPRES BHNOTC ILLWRK ROAD TIMECODE TYPILL BHPAID LANG SMEARNER TIMEDAYS TYPINJ BNKH11(01-11) LANGDI SMHCOMP TMEOFF UNION BNKHOLF LANGD2 SMOCCD TRHR11 VOCQPLII

Table 1: LFS variables which may be affected by non-responding DNAs. Source: Labour Force Survey User Guide Vol 1p 73.

The problem of DNAs in regards to calculating trade union membership has been recognised by both the ONS and BIS, and there are two possible methods of estimating union membership levels taking into account the problem of DNAs. One method, a general one, used by the ONS, relies on calculating an adjusted weight, whereby cases with missing data which are assumed to have the same distribution as valid responses. The process of imputing the non-responding DNAs is as follows:

(i) calculate: Valid response total

(Valid response total) - (non-responding DNAs)

(ii) multiply each of the valid responses by this factor (excluding the DNAs)

8

precision by boosting the sample size. The rationale is that most LFS variables do not change from one quarter to another for most people. Responses are rolled forward for one wave only. Data is not rolled forward after a second consecutive non-response. In the LFS Performance and Quality Monitoring Report (PQM) tables and charts (at person or household level) containing responses which have been rolled forward from the previous wave are denoted by the term 'including imputed'. Tables and charts which do not contain responses which have been rolled forward from the previous wave are denoted by the term 'excluding imputed' (ONS 2015a:5).

3.2 Imputation of does not apply responses

When analysing the LFS we may find that the does not apply (DNA) category may be exceptionally large. This can be due to the question not being asked in everyquarter, a non-core question e.g. trade union member. And some respondent are notcontacted in every survey. Table 1 gives a list of variables in the LFS that may be ormay have been, affected by non-responding DNAs.

Examination of Table 1 identifies three LFS variables that we will use in this paper. They are: TUCOV, TUPRES, and UNION.

ACCDAY4 EDINS11 MATLVE SMOCCT TRONJB WCHJBACCURH(1-4) FEEIR(1-5) NOBACK9 SMSOC10 TRVDRV WCHJB3AWARE GOBCK9 NOCUST SMSOC101 TRVMTHBANK HOLS NUMILL SMSOC103 TRVTMEBHNOTA HOLSB QAPL11 SMSOC104 TUCOVBHNOTB ILCURR REASOFF9 TFEE10(1-5) TUPRESBHNOTC ILLWRK ROAD TIMECODE TYPILLBHPAID LANG SMEARNER TIMEDAYS TYPINJBNKH11(01-11) LANGD1 SMHCOMP TMEOFF UNIONBNKHOLF LANGD2 SMOCCD TRHR11 VOCQPL11

Table 1: LFS variables which may be affected by non-responding DNAs.Source: Labour Force Survey User Guide Vol 1p 73.

The problem of DNAs in regards to calculating trade union membership has been recognised by both the ONS and BIS, and there are two possible methods of estimating union membership levels taking into account the problem of DNAs. One method, a general one, used by the ONS, relies on calculating an adjusted weight, whereby cases with missing data which are assumed to have the same distribution as valid responses.The process of imputing the non-responding DNAs is as follows:

(i) calculate:Valid response total

(Valid response total) - (non-responding DNAs)

(ii) multiply each of the valid responses by this factor (excluding theDNAs)

8

In this paper, we are only examining trade union members who are employees, so these results will be a slightly different from the ONS methodology which uses six variables, and by filtering on these groups it is possible to produce a table of UNION by !OUTCOME. — see the Labour Survey User Guide 2015 Vol 1:p72.

A second method was developed — we assume by BIS statistician(s) in the Labour Market Directorate —`to closely approximate the results from an adjusted weights approach but also allow a consistent time series with the calendar quarter information to be estimated. It consists of union density multiplied by the population (as estimated by the LFS). The method is improved by making the same calculation but by detailed age, gender and regional disaggregations and then aggregating them back up to national and regional levels. Age, gender and region were specifically chosen as they are the basis on which the LFS is weighted (see the Labour Survey User Guide Vol 1.)' (Department for Business Innovation and Skills 2015a:p55).

Between 2001 and 2014 the regional variable used by BEIS was GOVTOR (Government Office regions), this was withdrawn by the ONS in January 2015 and replaced by GOVTOF2. GOVTOF2 combines Government Office Region 2 and 3.

The reason for the dropping of the GOVTOR variable from the LFS datasets was that it could be used to potentially disclose individual responses. BIS were advised by their disclosure control team to introduce the GOVTOF2 variable to minimise the risk of individual data being revealed.

The importance of using an imputation method to calculate trade union membership is shown in Table 2. Without using an imputation methodology we under-estimate trade union membership in total by just over a million members.

Original Output ONS Imputation BEIS Imputation Year male female Total male female Total male female Total 2015 2,417 3,006 5,423 2,885 3,593 6,478 2,896 3,582 6,478

Table 2 Outcome for Different Methods of Imputation for Trade Union Membership `000. Source: LFS Oct-Dec 2015.

Table 2 shows the different outcomes using the two different methodologies previously described. Although both the ONS and BEIS totals are the same there are differences between the genders.

3.3 Sampling Variance and the LFS

`The labour force survey (LFS) is a sample survey, which provides estimates of various measures of the population that reflect the particular sample selected. As any valid sample from the population is likely to produce different estimates for the population than other valid samples, there is uncertainty around the population estimates resulting from the selection of a particular sample (as opposed to any other valid sample)' (BIS 2013:3).

This is known as the sampling error, 'which is the difference between the sample result and the population characteristic being estimated. In practice, the sampling error can rarely be determined because the population characteristics is not usually known'

9

In this paper, we are only examining trade union members who are employees, so these results will be a slightly different from the ONS methodology which uses six variables, and by filtering on these groups it is possible to produce a table of UNION by IOUTCOME. – see the Labour Survey User Guide 2015 Vol 1:p72.

A second method was developed – we assume by BIS statistician(s) in the LabourMarket Directorate –‘to closely approximate the results from an adjusted weights approach but also allow a consistent time series with the calendar quarter information to be estimated. It consists of union density multiplied by the population (as estimatedby the LFS). The method is improved by making the same calculation but by detailedage, gender and regional disaggregations and then aggregating them back up tonational and regional levels. Age, gender and region were specifically chosen as they are the basis on which the LFS is weighted (see the Labour Survey User Guide Vol 1.)’ (Department for Business Innovation and Skills 2015a:p55).

Between 2001 and 2014 the regional variable used by BEIS was GOVTOR (Government Office regions), this was withdrawn by the ONS in January 2015 and replaced by GOVTOF2. GOVTOF2 combines Government Office Region 2 and 3.

The reason for the dropping of the GOVTOR variable from the LFS datasets was that it could be used to potentially disclose individual responses. BIS were advised by their disclosure control team to introduce the GOVTOF2 variable to minimise the risk of individual data being revealed.

The importance of using an imputation method to calculate trade union membership is shown in Table 2. Without using an imputation methodology we under-estimate trade union membership in total by just over a million members.

Table 2: Outcome for Different Methods of Imputation for Trade Union Membership ‘000. Source: LFS Oct-Dec 2015.

Table 2 shows the different outcomes using the two different methodologies previously described. Although both the ONS and BEIS totals are the same there are differences between the genders.

3.3 Sampling Variance and the LFS

‘The labour force survey (LFS) is a sample survey, which provides estimates of variousmeasures of the population that reflect the particular sample selected. As any validsample from the population is likely to produce different estimates for the populationthan other valid samples, there is uncertainty around the population estimatesresulting from the selection of a particular sample (as opposed to any other valid sample)’ (BIS 2013:3).

This is known as the sampling error, ‘which is the difference between the sample resultand the population characteristic being estimated. In practice, the sampling error can rarely be determined because the population characteristics is not usually known’

9

Original Output ONS Imputation BEIS ImputationYear male female Total male female Total male female Total2015 2,417 3,006 5,423 2,885 3,593 6,478 2,896 3,582 6,478

(Everitt 2002: 332). However, in 'the methodology user guide', ONS advise that users can calculate their own confidence intervals by estimating their own design factors by applying a 'complex samples' analysis in a statistical package' (Department for Business Innovation and Skills 2013: 4).

'As the LFS uses random sampling of households it is possible to provide estimates of the sampling error (measured as the standard error), which forms the basis of the confidence interval placed around estimate. Normally a 95% confidence interval would be used, which means that in 19 out of 20 different samples we would expect the true (population) level or rate considered to be within the 95% confidence interval for the sample estimate. Where a simple random sample is used, the 95% confidence interval for an estimate is set at ±1.96 standard errors around the sample estimate' (BIS 2013:3).

Fortunately BIS has calculated 95% confidence intervals for trade union membership levels and density estimates be found for the trade union membership statistics 2012 to 2015 on the United Kingdom Government web site. Table 3 shows the trade union density and trade union membership levels for those in employment in the UK in 2015.

Union density (%) Union membership levels (`000) Population

estimate Lower bound

Upper bound

Population estimate

Lower bound

Upper bound

All in employment - United Kingdom 21.9 21.4 22.4 5,702 5,358 6,045 Sex Male 18.8 18.1 19.4 2,591 2,424 2,757 Female 25.5 24.8 26.2 3,111 2,916 3,306

Table 3: Trade Union density and trade union membership levels. In employment, 2015. All in employment in the United Kingdom. Source: BIS 2015b Table 5.

The BIS tables show the estimate, and the upper and lower bounds of the estimate at the 95% confidence levels. These tables cover:

Trade union density and membership levels for employees.

Trade union presence for employees.

Collective agreement coverage for employees.

Average hourly earnings by trade union status for employees.

Trade Union density and trade union membership levels for those in employment.

10

(Everitt 2002: 332). However, in ‘the methodology user guide’, ONS advise that users can calculate their own confidence intervals by estimating their own design factors by applying a ‘complex samples’ analysis in a statistical package’ (Department for Business Innovation and Skills 2013: 4).

‘As the LFS uses random sampling of households it is possible to provide estimatesof the sampling error (measured as the standard error), which forms the basis of theconfidence interval placed around estimate. Normally a 95% confidence interval would be used, which means that in 19 out of 20 different samples we would expect the true (population) level or rate considered to be within the 95% confidence interval for thesample estimate. Where a simple random sample is used, the 95% confidence interval for an estimate is set at ±1.96 standard errors around the sample estimate’ (BIS 2013:3).

Fortunately BIS has calculated 95% confidence intervals for trade union membershiplevels and density estimates be found for the trade union membership statistics 2012to 2015 on the United Kingdom Government web site. Table 3 shows the trade uniondensity and trade union membership levels for those in employment in the UK in 2015.

Table 3: Trade Union density and trade union membership levels. In employment, 2015. All in employment in the United Kingdom. Source: BIS 2015b Table 5.

The BIS tables show the estimate, and the upper and lower bounds of the estimate at the 95% confidence levels. These tables cover:

Trade union density and membership levels for employees.

Trade union presence for employees.

Collective agreement coverage for employees.

Average hourly earnings by trade union status for employees.

Trade Union density and trade union membership levels for those in employment.

10

Union density (%) Union membership levels (‘000)Population

estimateLowerbound

Upperbound

Populationestimate

Lowerbound

Upperbound

All inemployment - United Kingdom 21.9 21.4 22.4 5,702 5,358 6,045SexMale 18.8 18.1 19.4 2,591 2,424 2,757Female 25.5 24.8 26.2 3,111 2,916 3,306

If the confidence intervals of two estimated values do not overlap then they are considered statistically significantly different at the 5% significance level. The confidence intervals are based on the methodology described in 'Sampling variance in the Trade Union Membership Statistics' BIS 2013.

11

If the confidence intervals of two estimated values do not overlap then they are considered statistically significantly different at the 5% significance level. The confidence intervals are based on the methodology described in ‘Sampling variance in the Trade Union Membership Statistics’ BIS 2013.

11

4.0 Overview of Change in Trade Union Membership

4.1 Introduction

In this section we will critically examine how trade union membership has changed between 2001 and 2015. At the start of an analysis on trade union membership it is not unusual to find a chart showing the change in trade union membership over a given time e.g. Forth and Bryson 2015:2. Where they use data on trade union membership from the Certification Officer. These figures include: unemployed; retired, and self-employed trade union members — see Certification Officer (2015). This figure is greater than the actual number of trade union members who are employees in employment, using the LFS, which is the base for this study.

Figure 1: Trade union membership totals by Certification Officer and BEIS Employees. Source: BIS (2015a) Table 1.

Mem

bers

'00

0

8,000

7,800

7,600

7,400

7,200

7,000

6,800

6,600

6,400

6,200

6,000

—01" -.16 )̀ -.<54)‘ —041) "Q" ".).1* 1.)',1%) ,V ;V 1\) 't) ,'V 't) 19 '19 '196;., 46,1, ACC & ct , <,1; N.0)'

'19 '19 '17-cso,)

'19 '19 '19 '19 '19

Year

Certification Officer — BEIS

In Figure 1 we show the difference between using the Certification Officer's data for trade union membership and BIS (now Department for Business, Energy & Industrial Strategy (BEIS)) data for employees who were in a trade union. The examination of the chart shows a significant difference between the two data sets.

Both of the elements of the chart shows a long-term overall decline in trade union membership. However, if we analyse the data by gender we come to a different conclusion — See Figure 2.

12

4.0 Overview of Change in Trade Union Membership

4.1 Introduction

In this section we will critically examine how trade union membership has changed between 2001 and 2015. At the start of an analysis on trade union membership it is not unusual to find a chart showing the change in trade union membership over a given time e.g. Forth and Bryson 2015:2. Where they use data on trade union membership from the Certification Officer. These figures include: unemployed; retired, and self-employed trade union members – see Certification Officer (2015). This figure is greaterthan the actual number of trade union members who are employees in employment,using the LFS, which is the base for this study.

Figure 1: Trade union membership totals by Certification Officer and BEIS Employees. Source: BIS (2015a) Table 1.

In Figure 1 we show the difference between using the Certification Officer’s data for trade union membership and BIS (now Department for Business, Energy & IndustrialStrategy (BEIS)) data for employees who were in a trade union. The examination ofthe chart shows a significant difference between the two data sets.

Both of the elements of the chart shows a long-term overall decline in trade union membership. However, if we analyse the data by gender we come to a different conclusion – See Figure 2.

12

Figure 2: Employee trade union members by gender 2001-15 Source: LFS Oct-Dec 2001-2015

3,700,000 ...,,.....„

3,300,000

2,900,000

2,500,000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Male Female

From Figure 2 we can see that while male trade union membership declined by 795,000 between 2001 and 2015. During the same period, female union membership increased by 299,000. Overall there was a net decrease of 566,000 trade union members.

Further examination of the chart also reveals that in 2005 for the first time there were more female trade union members in the UK than male trade union members.

Disaggregating the data further by employment status we can see that male full-time trade union employees accounted for the whole of the decline in trade union membership between 2001 and 2015 — see Table 4. Male FT trade union membership declined by 854,000 between 2001 and 2015. The total change in employment status for all trade union members was -599,000.

Employment Status Gender Full-time '000 Part-time '000 Female 120 78 Male -854 57

Table 4: Change in Trade Union Membership by gender and employment status 2001-2015. Source: LFS Oct-Dec 2001-2015

A number of observers use trade union density as the main method of analysis of changes in trade union membership e.g. Green 1992. Trade union density is obtained by taking the number of trade union members di►ided by the relevant working population and then multiply by one-hundred. This union density figure has also been

13

Figure 2: Employee trade union members by gender 2001-15 Source: LFS Oct-Dec 2001-2015

From Figure 2 we can see that while male trade union membership declined by 795,000 between 2001 and 2015. During the same period, female union membershipincreased by 299,000. Overall there was a net decrease of 566,000 trade unionmembers.

Further examination of the chart also reveals that in 2005 for the first time there were more female trade union members in the UK than male trade union members.

Disaggregating the data further by employment status we can see that male full-timetrade union employees accounted for the whole of the decline in trade unionmembership between 2001 and 2015 – see Table 4. Male FT trade union membershipdeclined by 854,000 between 2001 and 2015. The total change in employment statusfor all trade union members was -599,000.

Table 4: Change in Trade Union Membership by gender and employment status 2001- 2015. Source: LFS Oct-Dec 2001-2015

A number of observers use trade union density as the main method of analysis of changes in trade union membership e.g. Green 1992. Trade union density is obtained by taking the number of trade union members divided by the relevant working population and then multiply by one-hundred. This union density figure has also been

13

Employment StatusGender Full-time ‘000 Part-time ‘000Female 120 78Male -854 57

used by some observers as a proxy for trade union power. The higher the trade union density the more power the trade union has, in collective barraging terms.

However, using trade union density alone for analysing time-series data can be problematical — see Table 5.

Gender Full-time Part-time Female Change in Trade Union

Membership 5.4% 7.0%

Change in Employment 13.7% 4.5% Change in Trade Union Density -7.4% 2.4%

Male Change in Trade Union Membership -24.0% 42.2% Change in Employment 2.3% 46.1%

Change in Trade Union Density -25.7% -2.7% Table 5: Change in trade union membership by gender 2001-15. Source LFS 2001 and 2015

If we look at full-time female trade union membership this increased by 5.4 per cent between 2001 and 2015 from 2.2m to 2.4m. This represents an actual increase of 120,000, however, at the same time employment increased by 922,000 from 6.7m to 7.7m. In 2001 trade union density was 33.3 per cent, yet despite an increase in membership the trade union density had fallen by -7.4 per cent from 33.3 to 30.7 per cent in 2015. This illustrates the risks of using only changes in trade density alone, without taking into account changes in the actual data, in assessing changes in trade union membership.

4.2 Changes in Trade Union Density and Composition

During the period covered by this study, there has been a change in the composition of the workforce. In previous studies, the change in the composition of the workforce has been linked to decline in trade union density (Bryson and Gomez 2002:51). The reason for this is that the change in the composition of the workforce has taken place in traditional trade union strongholds, such as the manufacturing sector, and in manual occupations. These sectors where trade unions were traditionally strong have declined in relevant importance as a percentage of the total workforce.

Over the years a number of studies have taken place to identify where trade union density was higher. 'The broad consensus of such studies is that union density tends to be higher: (a) for males than for females; (b) for full-timers than for part-timers; (c) in larger than in smaller establishments; (d) in manufacturing industries than in service industries; (e) in the North than in the South; (f) for manual than for non-manual workers; (g) for older than for younger workers. (Green 1992: 446).

14

used by some observers as a proxy for trade union power. The higher the trade union density the more power the trade union has, in collective barraging terms.

However, using trade union density alone for analysing time-series data can be problematical – see Table 5.

Table 5: Change in trade union membership by gender 2001-15. Source LFS 2001 and 2015

If we look at full-time female trade union membership this increased by 5.4 per cent between 2001 and 2015 from 2.2m to 2.4m. This represents an actual increase of 120,000, however, at the same time employment increased by 922,000 from 6.7m to7.7m. In 2001 trade union density was 33.3 per cent, yet despite an increase inmembership the trade union density had fallen by -7.4 per cent from 33.3 to 30.7 percent in 2015. This illustrates the risks of using only changes in trade density alone,without taking into account changes in the actual data, in assessing changes in tradeunion membership.

4.2 Changes in Trade Union Density and Composition

During the period covered by this study, there has been a change in the composition of the workforce. In previous studies, the change in the composition of the workforcehas been linked to decline in trade union density (Bryson and Gomez 2002:51). Thereason for this is that the change in the composition of the workforce has taken placein traditional trade union strongholds, such as the manufacturing sector, and in manualoccupations. These sectors where trade unions were traditionally strong have declinedin relevant importance as a percentage of the total workforce.

Over the years a number of studies have taken place to identify where trade union density was higher. ‘The broad consensus of such studies is that union density tends to be higher:(a) for males than for females;(b) for full-timers than for part-timers;(c) in larger than in smaller establishments;(d) in manufacturing industries than in service industries;(e) in the North than in the South;(f) for manual than for non-manual workers;(g) for older than for younger workers.(Green 1992: 446).

14

Gender Full-time Part-timeFemale Change in Trade Union

Membership5.4% 7.0%

Change in Employment 13.7% 4.5%Change in Trade Union Density -7.4% 2.4%

Male Change in Trade Union Membership -24.0% 42.2%Change in Employment 2.3% 46.1%

Change in Trade Union Density -25.7% -2.7%

In addition to these variables, we will use: public and private sectors, trade union coverage, and trade union presence. The assumption we are using in regards to the public/private sector is that trade union density will be higher in the public sector compared to the private sector.

The trade union coverage and presence questions were used in the LFS for the first time in autumn 1996. However, UK Data Archives only have data going back to October-December 2001. With these two variables we have no prior assumptions, however, we will analyse the changes which have taken place.

One of the objects of this study is to find out if these assumptions are still true, or have any of them changed.

In the following sections, we will analyse the changes that have taken in these variables.

4.3 Change in Gender

Previously we had established that although overall trade union membership had declined between 2001 and 2015 by 599,000, using FT and PT employment status and gender, however, during the same period female trade union membership had increased by 229,000, an increase of 6.8 per cent — See Table 6. Overall trade union membership declined by 599,000 a decrease of eight per cent.

Union membership Change in workforce Gender Actual % Change Actual % Change Female 229,000 6.8 1,153,146 9.8 Male -794,881 -21.5 760,860 6.1

Table 6: Change in gender union density and employment 2001-2015. Source: LFS Oct —Dec 2001 and 2015.

The size of the workforce for both male and female employees increased between 2001 and 2015. The greatest increase was for female employees which increased by just under 1.2m an increase of 9.8 per cent. Male employees increased by just under 0.8m an increase of 6.1 per cent.

In Table 7 we give the estimates of trade union density and employment composition for 2001 and 2015 by gender.

Union Density % Employment Composition % Gender 2001 2015 2015-2001 2001 2015 2015-2001

Female 28.5 27.4 -1.1 48.6 49.2 0.6 Male 29.7 21.5 -8.2 51.4 50.8 -0.6

Table 7: Trade union density and employment composition 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

15

In addition to these variables, we will use: public and private sectors, trade union coverage, and trade union presence. The assumption we are using in regards to the public/private sector is that trade union density will be higher in the public sector compared to the private sector.

The trade union coverage and presence questions were used in the LFS for the first time in autumn 1996. However, UK Data Archives only have data going back toOctober-December 2001. With these two variables we have no prior assumptions,however, we will analyse the changes which have taken place.

One of the objects of this study is to find out if these assumptions are still true, or have any of them changed.

In the following sections, we will analyse the changes that have taken in these variables.

4.3 Change in Gender

Previously we had established that although overall trade union membership had declined between 2001 and 2015 by 599,000, using FT and PT employment status and gender, however, during the same period female trade union membership had increased by 229,000, an increase of 6.8 per cent – See Table 6. Overall trade union membership declined by 599,000 a decrease of eight per cent.

Table 6: Change in gender union density and employment 2001-2015. Source: LFS Oct –Dec 2001 and 2015.

The size of the workforce for both male and female employees increased between2001 and 2015. The greatest increase was for female employees which increased byjust under 1.2m an increase of 9.8 per cent. Male employees increased by just under0.8m an increase of 6.1 per cent.

In Table 7 we give the estimates of trade union density and employment composition for 2001 and 2015 by gender.

Table 7: Trade union density and employment composition 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

15

Union membership Change in workforceGender Actual % Change Actual % ChangeFemale 229,000 6.8 1,153,146 9.8Male -794,881 -21.5 760,860 6.1

Union Density % Employment Composition %Gender 2001 2015 2015-2001 2001 2015 2015-2001Female 28.5 27.4 -1.1 48.6 49.2 0.6Male 29.7 21.5 -8.2 51.4 50.8 -0.6

Male trade union density declined from 29.7 per cent to 21.5 per cent between 2001 and 2015. During the same period, female trade union density remained fairly stable, declining from 28.5 per cent to 27.4 per cent.

In regards to the change in the composition of the workforce, there was an increase in the female composition of the workforce. Increasing by 0.6 per cent from 48.6 to 49.2 per cent - see Table 7.

From the above analysis, we can see that the proposition that trade union density is likely to be greater for males than for females is no longer true, and has not been since 2002. Also, there is now near parity between females and males in terms of the composition of the workforce

4.4 Full-time vs Part-time

Between 2001 and 2015 full-time trade union membership declined by 734,000 from 5.8m to 5.1m a decrease of 12.7 per cent. For part-time employees increased by 135,000 a increase of eleven per cent - see Table 8.

Union membership Change in workforce Status Actual % Change Actual % Change FT -734,002 -12.7 760,860 6.1 PT 134,583 10.7 1,153,146 9.8

Table 8: Change in full-time and part-time union density and employment 2001-2015. Source LFS Oct -Dec 2001 and 2015.

Both full-time and part-time employment increased during this time, full-time employment increased by 6.1 per cent, an increase of under 0.8m. Part-time employment increased 9.8 per cent, an increase just under 1.2m.

During this period trade union density for full-time employees declined by 6.2 per cent, from 32.1 per cent to 25.9 per cent. Part-time employee density remained static at 20.4 per cent.

In regards to changes in the composition of the workforce during the period covered, although full-time employment increased by 1.2m, it declined by 0.7 per cent to 73.9 per cent of employees.

From the above analysis, we can see that the proposition that trade union density is still greater for full-time employees than for part-timer employees. Although the difference in density between full-time and part-time has narrowed considerably. In 2001 the difference was 11.6 per cent, by 2015 it had decreased to 5.5 per cent - see Table 9.

16

Male trade union density declined from 29.7 per cent to 21.5 per cent between 2001and 2015. During the same period, female trade union density remained fairly stable, declining from 28.5 per cent to 27.4 per cent.

In regards to the change in the composition of the workforce, there was an increase in the female composition of the workforce. Increasing by 0.6 per cent from 48.6 to 49.2 per cent – see Table 7.

From the above analysis, we can see that the proposition that trade union density is likely to be greater for males than for females is no longer true, and has not been since2002. Also, there is now near parity between females and males in terms of thecomposition of the workforce

4.4 Full-time vs Part-time

Between 2001 and 2015 full-time trade union membership declined by 734,000 from 5.8m to 5.1m a decrease of 12.7 per cent. For part-time employees increased by 135,000 a increase of eleven per cent – see Table 8.

Table 8: Change in full-time and part-time union density and employment 2001-2015. Source LFS Oct –Dec 2001 and 2015.

Both full-time and part-time employment increased during this time, full-time employment increased by 6.1 per cent, an increase of under 0.8m. Part-time employment increased 9.8 per cent, an increase just under 1.2m.

During this period trade union density for full-time employees declined by 6.2 per cent, from 32.1 per cent to 25.9 per cent. Part-time employee density remained static at 20.4 per cent.

In regards to changes in the composition of the workforce during the period covered, although full-time employment increased by 1.2m, it declined by 0.7 per cent to 73.9 per cent of employees.

From the above analysis, we can see that the proposition that trade union density is still greater for full-time employees than for part-timer employees. Although the difference in density between full-time and part-time has narrowed considerably. In 2001 the difference was 11.6 per cent, by 2015 it had decreased to 5.5 per cent – see Table 9.

16

Union membership Change in workforceStatus Actual % Change Actual % ChangeFT -734,002 -12.7 760,860 6.1PT 134,583 10.7 1,153,146 9.8

Total Union Density Employment Composition % Status 2001 2015 2015-2001 2001 2015 2015-2001 FT 32.1 25.9 -6.2 74.6 73.9 -0.7 PT 20.5 20.4 -0.1 25.4 26.1 0.7

Table 9: Trade union density and employment composition 2001 and 2015. Source: Labour Force Survey Oct-Dec 2001 and 2015.

4.5 Establishment Size

In establishment with employee numbers between 1-49, the number of trade union members declined by 93,000 between 2001 and 2015. This represents a decline of 4.6 per cent - See Table 10. During the same period for establishments with over fifty employees the decrease was 523,000. A decrease of 10.5 per cent. This 523,000 decrease accounts for 66 per cent of the 794,881 total decline in trade union membership between 2001 and 2015.

Union membership Change in workforce Size Actual % Change Actual % Change 1-49 -93,209 -4.6 1,353,810 12.3 50 and over -523,163 -10.5 753,318 5.8

Table 10: Change in establishment size union density and employment 2001-2015. Source LFS Oct -Dec 2001 and 2015.

At the same time, employment in establishments with employee numbers between 1 and 49 increased by 1.36m and increase of 12.3 per cent. For establishments with fifty plus employees the increase was 753,000, an increase of 5.8 per cent.

It is worth noting that the increase in employment in establishments with 1-49 employees, and the loss of trade membership does not necessarily reflect some of the current thinking that trade unions find it more difficult to retain membership in smaller establishments, given the loss of over 0.5m in establishments with fifty plus employees. Although this may be true for establishments with less than ten employees,

The change in trade union density for establishments with 1- 49 employees decreased by 2.8 per cent from 18.4 per cent to 15.6 per cent. Over the same time period establishments with 50 and over employees had a decline of 5.2 per cent from 38.3 to 33.1 per cent - see Table 11.

Union Density Employment Composition % Size 2001 2015 2015-2001 2001 2015 2015-2001 1-49 18.4 15.6 -2.8 45.9 47.0 1.1 50 and over 38.3 33.1 -5.2 54.1 53.0 -1.1

Table 11: Trade union density and employment composition 2001 and 2015 by establishment size. Source: LFS Oct-Dec 2001 and 2015.

There was a slight increase in the number of smaller establishments, increasing by 1.1 per cent from 45.9 per cent to 47.0 per cent.

17

Table 9: Trade union density and employment composition 2001 and 2015. Source: Labour Force Survey Oct-Dec 2001 and 2015.

4.5 Establishment Size

In establishment with employee numbers between 1-49, the number of trade union members declined by 93,000 between 2001 and 2015. This represents a decline of 4.6 per cent – See Table 10. During the same period for establishments with over fifty employees the decrease was 523,000. A decrease of 10.5 per cent. This 523,000 decrease accounts for 66 per cent of the 794,881 total decline in trade union membership between 2001 and 2015.

Table 10: Change in establishment size union density and employment 2001-2015. Source LFS Oct –Dec 2001 and 2015.

At the same time, employment in establishments with employee numbers between 1 and 49 increased by 1.36m and increase of 12.3 per cent. For establishments with fifty plus employees the increase was 753,000, an increase of 5.8 per cent.

It is worth noting that the increase in employment in establishments with 1-49 employees, and the loss of trade membership does not necessarily reflect some of the current thinking that trade unions find it more difficult to retain membership in smallerestablishments, given the loss of over 0.5m in establishments with fifty plusemployees. Although this may be true for establishments with less than tenemployees,

The change in trade union density for establishments with 1- 49 employees decreased by 2.8 per cent from 18.4 per cent to 15.6 per cent. Over the same time period establishments with 50 and over employees had a decline of 5.2 per cent from 38.3 to 33.1 per cent – see Table 11.

Table 11: Trade union density and employment composition 2001 and 2015 by establishment size. Source: LFS Oct-Dec 2001 and 2015.

There was a slight increase in the number of smaller establishments, increasing by 1.1 per cent from 45.9 per cent to 47.0 per cent.

17

Total Union Density Employment Composition %Status 2001 2015 2015-2001 2001 2015 2015-2001FT 32.1 25.9 -6.2 74.6 73.9 -0.7PT 20.5 20.4 -0.1 25.4 26.1 0.7

Union membership Change in workforceSize Actual % Change Actual % Change1-49 -93,209 -4.6 1,353,810 12.350 and over -523,163 -10.5 753,318 5.8

Union Density Employment Composition %Size 2001 2015 2015-2001 2001 2015 2015-20011-49 18.4 15.6 -2.8 45.9 47.0 1.150 and over 38.3 33.1 -5.2 54.1 53.0 -1.1

We can confirm that trade union density is still higher in larger than in smaller establishments.

4.6 Industry

When undertaking a time-series analysis it is not unusual to find that one or more variables change over time due to the updating of international classification, and this is the case for standard industrial classifications (SIC). The SIC2007 codes were first used in the LFS in 2009. To keep as closely as possible to Green (1992) we have convert the SIC2007 codes to SIC92 codes between 2009 and 2015. This conversion variable gives reasonable comparisons, but not an exact conversion. The LFS User Guide Volume 5 (LFS Classifications) gives full industry breakdowns for both SIC92 & SIC2007.

If we examine Table 12 we see that the four industries who had the largest actual loss in trade union membership are: Manufacturing; Public administration and defence; Transport, Storage and Communications; and Construction. The decline in manufacturing trade union membership is considerably larger than the three subsequent sectors. Trade union membership in manufacturing declined from 1.1m in 2001 to 0.5m in 2015, a decline of 0.7m. The total decline in trade union membership for the following three sectors was 0.6m in total.

Union membership Change in workforce Actual % Change Actual % Change

Manufacturing -652,061 -58.5 -1,270,800 -30.2 Public administration and defence -290,539 -27.9 15,972 -17.1 Transport, Storage and Communications -187,098 -25.8 -69,851 4.0 Construction -84,375 -33.9 -43,312 -31.7 Other community social and personal -27,789 -11 304,553 27.5 Mining & Quarrying -11,268 -47.8 26,214 25.9 Hotels and Restaurants -7,707 -13 394,967 36.5 Agriculture & Fishing -2,460 -16.5 30,948 16.1 Electricity Gas and Water Suppliers -197 -0.2 42,948 21.5 Real Estate, Renting and Business Activities 3,962 1.5 907,608 35.1 Wholesale, Retail and Motor Trade 30,423 7.1 102,234 0.5 Financial Intermediation 151,583 -49.8 -29,148 -2.5 Health and Social Work 205,489 16.1 1,010,618 35.6 Education 365,761 32.4 840,449 39.5 Total Change -506,276 -11.6 2,263,400 9.4

Table 12: Change trade union membership and employees by industry between 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

18

We can confirm that trade union density is still higher in larger than in smaller establishments.

4.6 Industry

When undertaking a time-series analysis it is not unusual to find that one or more variables change over time due to the updating of international classification, and this is the case for standard industrial classifications (SIC). The SIC2007 codes were first used in the LFS in 2009. To keep as closely as possible to Green (1992) we haveconvert the SIC2007 codes to SIC92 codes between 2009 and 2015. This conversionvariable gives reasonable comparisons, but not an exact conversion. The LFS UserGuide Volume 5 (LFS Classifications) gives full industry breakdowns for both SIC92 &SIC2007.

If we examine Table 12 we see that the four industries who had the largest actual loss in trade union membership are: Manufacturing; Public administration and defence; Transport, Storage and Communications; and Construction.The decline in manufacturing trade union membership is considerably larger than the three subsequent sectors. Trade union membership in manufacturing declined from 1.1m in 2001 to 0.5m in 2015, a decline of 0.7m. The total decline in trade union membership for the following three sectors was 0.6m in total.

Table 12: Change trade union membership and employees by industry between 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

18

Union membership Change in workforceActual % Change Actual % Change

Manufacturing -652,061 -58.5 -1,270,800 -30.2Public administration anddefence -290,539 -27.9 15,972 -17.1Transport, Storage andCommunications -187,098 -25.8 -69,851 4.0Construction -84,375 -33.9 -43,312 -31.7Other community social andpersonal -27,789 -11 304,553 27.5Mining & Quarrying -11,268 -47.8 26,214 25.9Hotels and Restaurants -7,707 -13 394,967 36.5Agriculture & Fishing -2,460 -16.5 30,948 16.1Electricity Gas and WaterSuppliers -197 -0.2 42,948 21.5Real Estate, Renting andBusiness Activities 3,962 1.5 907,608 35.1Wholesale, Retail and MotorTrade 30,423 7.1 102,234 0.5Financial Intermediation 151,583 -49.8 -29,148 -2.5Health and Social Work 205,489 16.1 1,010,618 35.6Education 365,761 32.4 840,449 39.5Total Change -506,276 -11.6 2,263,400 9.4

Manufacturing employment for the same period declined from 4.2m to 2.9m, a decline of 1.3m. Employment for the following three groups declined by 97,191.

Five sectors had an increase in trade union members, a total of 0.8m, of which education accounted for 0.4m. Employment in these five sectors increased by 2.8m, of which Health and Social Work accounted for 1.0m.

The greatest decrease in trade union density occurred in public administration and defence industries. Again this illustrates the danger of relying solely on trade union density when analysing changes in trade union membership. These industries had a decline trade union density from 59.9 per cent to 42.8 per cent. Mining and quarrying were next with a decline from 23.3 per cent to 9.7 per cent. In the case of financial intermediation it declined from a density of 25.6 to 13.2 per cent. For manufacturing, the decline of trade union density was from 26.5 per cent to 15.8 per cent. The only industry to experience an increase in trade union density was the wholesale, retail and motor trade, where trade union density increased from 11.7 per cent to 12.2 per cent - see Table 13.

Union Density Employment Composition % Industry 2001 2015 2015-2001 2001 2015 2015-2001 Agriculture & Fishing 7.7 5.6 -2.2 0.8 0.8 0.0 Mining & Quarrying 23.3 9.7 -13.7 0.4 0.5 0.1 Manufacturing 26.5 15.8 -10.7 17.4 11.1 -6.3 Electricity Gas and Water Suppliers 50.3 41.4 -9.0 0.8 0.9 0.1 Construction 18.8 12.8 -6.0 5.5 4.9 -0.6 Wholesale, Retail and Motor Trade 11.7 12.2 0.5 15.3 14.4 -0.9 Hotels and Restaurants 5.5 3.5 -2.0 4.5 5.6 1.1 Transport, Storage and Communications 41.8 32.3 -9.5 7.2 6.3 -0.9 Financial Intermediation 25.6 13.2 -12.4 4.9 4.4 -0.5 Real Estate, Renting and Business Activities 10.5 7.9 -2.6 10.7 13.2 2.5 Public administration and defence 59.9 42.8 -17.1 7.2 6.7 -0.6 Education 53.1 50.4 -2.7 8.8 11.2 2.4 Health and Social Work 44.9 38.4 -6.4 11.8 14.6 2.8 Other community social and personal 22.7 15.8 -6.9 4.6 5.4 0.8

Table 13: Trade union density and employment composition 2001 and 2015 by industry.

In regards to the changing employment composition of employees by industry manufacturing had the greatest negative change from 17.4 per cent of employees to 11.1 per cent of employees. The largest positive increase in employees was experienced by the health and social work sector. This increased from 11.8 per cent of the employee population to 14.6

19

Manufacturing employment for the same period declined from 4.2m to 2.9m, a decline of 1.3m. Employment for the following three groups declined by 97,191.

Five sectors had an increase in trade union members, a total of 0.8m, of whicheducation accounted for 0.4m. Employment in these five sectors increased by 2.8m,of which Health and Social Work accounted for 1.0m.

The greatest decrease in trade union density occurred in public administration and defence industries. Again this illustrates the danger of relying solely on trade union density when analysing changes in trade union membership. These industries had a decline trade union density from 59.9 per cent to 42.8 per cent. Mining and quarrying were next with a decline from 23.3 per cent to 9.7 per cent. In the case of financialintermediation it declined from a density of 25.6 to 13.2 per cent. For manufacturing,the decline of trade union density was from 26.5 per cent to 15.8 per cent. The onlyindustry to experience an increase in trade union density was the wholesale, retail andmotor trade, where trade union density increased from 11.7 per cent to 12.2 per cent – see Table 13.

Table 13: Trade union density and employment composition 2001 and 2015 by industry.

In regards to the changing employment composition of employees by industry manufacturing had the greatest negative change from 17.4 per cent of employees to 11.1 per cent of employees. The largest positive increase in employees was experienced by the health and social work sector. This increased from 11.8 per cent of the employee population to 14.6

19

Union Density Employment Composition %Industry 2001 2015 2015-2001 2001 2015 2015-2001Agriculture & Fishing 7.7 5.6 -2.2 0.8 0.8 0.0Mining & Quarrying 23.3 9.7 -13.7 0.4 0.5 0.1Manufacturing 26.5 15.8 -10.7 17.4 11.1 -6.3Electricity Gas andWater Suppliers 50.3 41.4 -9.0 0.8 0.9 0.1Construction 18.8 12.8 -6.0 5.5 4.9 -0.6Wholesale, Retail andMotor Trade 11.7 12.2 0.5 15.3 14.4 -0.9Hotels and Restaurants 5.5 3.5 -2.0 4.5 5.6 1.1Transport, Storage andCommunications 41.8 32.3 -9.5 7.2 6.3 -0.9FinancialIntermediation 25.6 13.2 -12.4 4.9 4.4 -0.5Real Estate, Rentingand Business Activities 10.5 7.9 -2.6 10.7 13.2 2.5Public administrationand defence 59.9 42.8 -17.1 7.2 6.7 -0.6Education 53.1 50.4 -2.7 8.8 11.2 2.4Health and Social Work 44.9 38.4 -6.4 11.8 14.6 2.8Other communitysocial and personal 22.7 15.8 -6.9 4.6 5.4 0.8

However, in this study, we are testing the proposition that trade union density is greater in manufacturing industries than in the service industries. To do this we have aggregated the first four industrial sectors into one variable which we call manufacturing sector, and the remaining sectors are aggregated into a variable called service sector.

Even by 2001, the trade union density for the service industries was greater than the manufacturing industries. Therefore, between the late 1970s and 1980's and 2001 there had already been a shift in trade union membership towards the services industries sectors.

The service industries trade union density was 30.3 per cent compared with 25.0 per cent for the manufacturing sectors. By 2015 the services sectors trade union density had decreased to 25.2 per cent and the manufacturing sector to 15.7 per cent.

Between 2001 and 2015 the total manufacturing sector trade union membership decreased by 0.75m from 1.5 to 0.75m, a decrease of fifty per cent. During the same period the service industries sectors trade union membership declined by 59,000 a decline of 1.1 per cent - see Table 14.

Union membership Change in workforce Industry Actual % Change Actual % Change Manufacturing -750,362 -49.9 -1,214,002 -20.2 Services -59,080 -1.1 3,477,402 19.2

Table 14: Change in trade union membership and employment between 2001 and 2015 by manufacturing and services.

In regards to the change in the composition of the workforce, the manufacturing sectors declined by 1.2m a decline of twenty per cent, while the services sectors employment increased by 3.5m an increase of nineteen per cent.

As we had previously stated, between 2001 and 2015 trade union density in the manufacturing industries had decreased from 20.5 per cent to 15.7 per cent, a decrease of 4.8 per cent. While the services sectors trade union density declined from 30.3 per cent to 25.2 per cent, a decrease of 5.1 per cent - see Table 15

Union Density Employment Composition % Industry 2001 2015 2015-2001 2001 2015 2015-2001 Manufacturing 20.5 15.7 -4.8 25.0 18.2 -6.7 Services 30.3 25.2 -5.1 75.0 81.8 6.7

Table 15: The change in trade union density and employment composition for manufacturing and service sectors 2001 and 2015.

The employment composition for the manufacturing sectors declined from 25 per cent to 18 per cent a change of 6.7 per cent. For the service industries the composition of the workforce increased from 75 per cent to 82 per cent.

From the above analysis we can say that trade union density for the service industries is now greater than the manufacturing industries.

20

However, in this study, we are testing the proposition that trade union density is greater in manufacturing industries than in the service industries. To do this we have aggregated the first four industrial sectors into one variable which we callmanufacturing sector, and the remaining sectors are aggregated into a variable calledservice sector.

Even by 2001, the trade union density for the service industries was greater than the manufacturing industries. Therefore, between the late 1970s and 1980’s and 2001 there had already been a shift in trade union membership towards the services industries sectors.

The service industries trade union density was 30.3 per cent compared with 25.0 per cent for the manufacturing sectors. By 2015 the services sectors trade union density had decreased to 25.2 per cent and the manufacturing sector to 15.7 per cent.

Between 2001 and 2015 the total manufacturing sector trade union membership decreased by 0.75m from 1.5 to 0.75m, a decrease of fifty per cent. During the sameperiod the service industries sectors trade union membership declined by 59,000 adecline of 1.1 per cent – see Table 14.

Table 14: Change in trade union membership and employment between 2001 and 2015 by manufacturing and services.

In regards to the change in the composition of the workforce, the manufacturing sectors declined by 1.2m a decline of twenty per cent, while the services sectors employment increased by 3.5m an increase of nineteen per cent.

As we had previously stated, between 2001 and 2015 trade union density in the manufacturing industries had decreased from 20.5 per cent to 15.7 per cent, a decrease of 4.8 per cent. While the services sectors trade union density declined from 30.3 per cent to 25.2 per cent, a decrease of 5.1 per cent – see Table 15

Table 15: The change in trade union density and employment composition for manufacturing and service sectors 2001 and 2015.

The employment composition for the manufacturing sectors declined from 25 per cent to 18 per cent a change of 6.7 per cent. For the service industries the composition of the workforce increased from 75 per cent to 82 per cent.

From the above analysis we can say that trade union density for the service industries is now greater than the manufacturing industries.

20

Union membership Change in workforceIndustry Actual % Change Actual % ChangeManufacturing -750,362 -49.9 -1,214,002 -20.2Services -59,080 -1.1 3,477,402 19.2

Union Density Employment Composition %Industry 2001 2015 2015-2001 2001 2015 2015-2001Manufacturing 20.5 15.7 -4.8 25.0 18.2 -6.7Services 30.3 25.2 -5.1 75.0 81.8 6.7

4.6.1 Manufacturing

From the above it is obvious that the manufacturing sector has taken the greatest hit in regards to both the decline in trade union membership and the number of employees in employment. In Table 16 we have analysed this sector at the two figure level of SIC. Looking at the changes in trade union membership, we have ranked them from the highest to the lowest loss. The totals for this table and Tables 17 and 18 differ from previous tables totals due in part to rounding errors and the less precise allocation of the two figure SIC code.

SIC92 Union 15:Food, beverage manufacture -77,095 34:Motor vehicle, trailer, etc. manufacture -74,094 22:Printing, publ ishi ng, recorded media -66,381 28:Fabric-metal prod (not machine equipment) manufacture -47,693 24:Chemicals,chemical products man -38,278 35:Other transport equipment manufacture -36,730 29:Mach,equipment manufacture -34,368 26:Other non-metallic products man -34,296 27:Basic metals manufacture -32,349 32:Radio,TV,communication equipment man -26,665 31:Elec machine ,equipment manufacture -26,313 36:Furniture etc. manufacture -23,941 25:Rubber,plastic products manufacture -23,892 17:Textile manufacture -20,616 23:Coke,petrol products, nuclear fuel manufacture -14,877 21:Pulp,paper,paper prods manufacture -14,588 30:Office machine, computer manufacture -13,243 18:Clothing,fur manufacture -12,377 33:Medical,precision,optical equipment manufacture -10,316 19:Leather,leather goods manufacture -4,808 20:Wood,straw,cork,wood prods(not furniture -4,600 16:Tobacco products manufacture -1,518 37:Recycling 1,687 Total -637,420

Table 16: Change in manufacturing trade union membership by two figure SIC between 2001 and 2015

From Table 16 we can see that the three largest sectors of loss of trade union membership were: food, beverage manufacture; motor vehicle, trailer, etc. manufacture; and printing, publishing, recorded media. These accounted for 217,570 of the total loss of membership, or just over one third.

Only one sector, recycling, had an increase of membership, but even this was nominal increase.

21

4.6.1 Manufacturing

From the above it is obvious that the manufacturing sector has taken the greatest hit in regards to both the decline in trade union membership and the number of employees in employment. In Table 16 we have analysed this sector at the two figure level of SIC. Looking at the changes in trade union membership, we have ranked them from the highest to the lowest loss. The totals for this table and Tables 17 and 18 differ from previous tables totals due in part to rounding errors and the less precise allocation of the two figure SIC code.

Table 16: Change in manufacturing trade union membership by two figure SIC between 2001 and 2015

From Table 16 we can see that the three largest sectors of loss of trade union membership were: food, beverage manufacture; motor vehicle, trailer, etc. manufacture; and printing, publishing, recorded media. These accounted for 217,570 of the total loss of membership, or just over one third.

Only one sector, recycling, had an increase of membership, but even this was nominal increase.

21

SIC92 Union15:Food, beverage manufacture -77,09534:Motor vehicle, trailer, etc. manufacture -74,09422:Printing,publishing,recorded media -66,38128:Fabric-metal prod (not machine equipment) manufacture -47,69324:Chemicals,chemical products man -38,27835:Other transport equipment manufacture -36,73029:Mach,equipment manufacture -34,36826:Other non-metallic products man -34,29627:Basic metals manufacture -32,34932:Radio,TV,communication equipment man -26,66531:Elec machine ,equipment manufacture -26,31336:Furniture etc. manufacture -23,94125:Rubber,plastic products manufacture -23,89217:Textile manufacture -20,61623:Coke,petrol products, nuclear fuel manufacture -14,87721:Pulp,paper,paper prods manufacture -14,58830:Office machine, computer manufacture -13,24318:Clothing,fur manufacture -12,37733:Medical,precision,optical equipment manufacture -10,31619:Leather,leather goods manufacture -4,80820:Wood,straw,cork,wood prods(not furniture -4,60016:Tobacco products manufacture -1,51837:Recycling 1,687Total -637,420

In Table 17 we show the change in manufacturing employment at the two figure level of SIC. The three highest losses in employment were from: printing; publishing; and recorded media. These sectors accounted for 416,655 job losses, which in percentage terms is 32.8 per cent.

SIC92 Employment 22:Printing, publishing, recorded media -162,269 28:Fabric-metal prod (not machine equipment) manufacturing -141,038 34:Motor vehicle, trailer, etc. manufacture -113,348 32:Radio,TV,communication equipment manufacture -111,688 31:Electircal machine equipment manufacture -94,870 36:Furniture etc. manufacture -79,736 25:Rubber,plastic products manufacture -79,433 24: Chemicals,chemical products man -75,419 30:Office machine, computer manufacture -66,035 17:Textile manufacture -65,911 27:Basic metals manufacture -46,684 21: Pulp,paper,paper prods manufacture -43,835 35:Other transport equipment manufacture -38,128 26:Other non-metallic products man -36,128 18:Clothing,fur manufacture -29,207 23:Coke,petrol products, nuclear fuel manufacture -28,691 15:Food,beverage manufacture -23,640 33: Medical, precision,optical equipment manufacture -22,451 20:Wood,straw,cork,wood prods(not furniture) -19,427 29:Machine,equipment manufacture -10,571 19: Leather,leather goods manufacture -5,956 16:Tobacco products manufacture -1,574 37:Recycling 25,239 Total -1,270,800

Table 17: Change in manufacturing employment by two figure SIC between 2001 and 2015.

If we compare the three largest losses of trade union membership in manufacturing —Table 16 — and the loss of manufacturing employment — Table 17 — we see that food, beverage manufacture lost 3.2 times more trade union members than the number of employees lost in that sector. In contrast both motor vehicle, trailer, etc. manufacture and motor vehicle, trailer, etc. manufacture lost proportionally less trade union members than the number of employees, 0.65 and 0.41 respectively.

Table 18 shows the change in manufacturing trade union density by two figure SIC between 2001 and 2015. The three largest declines in actual union density were in: leather, leather goods manufacture — -24.1; other non-metallic products manufacture - -23.8; and tobacco products manufacture — 19.7.

22

In Table 17 we show the change in manufacturing employment at the two figure levelof SIC. The three highest losses in employment were from: printing; publishing; andrecorded media. These sectors accounted for 416,655 job losses, which in percentage terms is 32.8 per cent.

Table 17: Change in manufacturing employment by two figure SIC between 2001 and2015.

If we compare the three largest losses of trade union membership in manufacturing – Table 16 – and the loss of manufacturing employment – Table 17 – we see that food,beverage manufacture lost 3.2 times more trade union members than the number ofemployees lost in that sector. In contrast both motor vehicle, trailer, etc. manufactureand motor vehicle, trailer, etc. manufacture lost proportionally less trade unionmembers than the number of employees, 0.65 and 0.41 respectively.

Table 18 shows the change in manufacturing trade union density by two figure SIC between 2001 and 2015. The three largest declines in actual union density were in: leather, leather goods manufacture – -24.1; other non-metallic products manufacture - -23.8; and tobacco products manufacture – 19.7.

22

SIC92 Employment22:Printing, publishing, recorded media -162,26928:Fabric-metal prod (not machine equipment) manufacturing -141,03834:Motor vehicle, trailer, etc. manufacture -113,34832:Radio,TV,communication equipment manufacture -111,68831:Electircal machine equipment manufacture -94,87036:Furniture etc. manufacture -79,73625:Rubber,plastic products manufacture -79,43324:Chemicals,chemical products man -75,41930:Office machine, computer manufacture -66,03517:Textile manufacture -65,91127:Basic metals manufacture -46,68421:Pulp,paper,paper prods manufacture -43,83535:Other transport equipment manufacture -38,12826:Other non-metallic products man -36,12818:Clothing,fur manufacture -29,20723:Coke,petrol products, nuclear fuel manufacture -28,69115:Food,beverage manufacture -23,64033:Medical,precision,optical equipment manufacture -22,45120:Wood,straw,cork,wood prods(not furniture) -19,42729:Machine,equipment manufacture -10,57119:Leather,leather goods manufacture -5,95616:Tobacco products manufacture -1,57437:Recycling 25,239Total -1,270,800

SIC92 Change in

Density % change in density

19:Leather, leather goods manufacture -24.1 -76.5 26:Other non-metallic products manufacture -23.8 -60.7 16:Tobacco products manufacture -19.7 -40.7 32:Radio,TV,communication equipment man No data No data 15:Food,beverage manufacture -16.9 -53.6 37:Recycling -15.8 -62.9 27:Basic metals manufacture -14.6 -34.1 34:Motor vehicle, trailer, etc., manufacture -14.0 -31.3 18:Clothing,fur manufacture -13.4 -67.0 22:Printing,publishing,recorded media -11.6 -46.6 17:Textile manufacture -11.2 -55.4 23:Coke,petrol products, nuclear fuel manufacturing -10.9 -26.6 36:Furniture etc manufacture -9.8 -67.3 24:Chemicals,chemical products man -9.0 -40.7 28:Fabric-metal products (not machine, equipment) manufacture -8.6 -42.3 35:Other transport equipment manufacture -8.6 -18.3 30:Office machines, computer manufacture -8.0 -59.4 29:Mach,equipment manufacture -7.7 -30.1 31:Elec machine, equipment manufacture -7.4 -40.2 33:Medical,precision,optical equipment man -6.5 -45.8 25:Rubber,plastic products manufacture -5.5 -27.4 20:Wood,straw,cork,wood prods(not furniture) -3.8 -30.9 21:Pulp,paper,paper prods manufacture -1.9 -6.1

Table 18: Change in manufacturing trade union density by two figure SIC between 2001 and 2015.

4.7 UK Regions

Between 2001 and 2015 the UK region which had the greatest loss in trade union members was the North West, with a decline of 113,634 - see Table 19. This represented a loss of 12.3 per cent, but this was not the largest percentage loss. This was the North East with a decline of 18.1 per cent. Only Wales had an increase in trade union membership, an increase of 7,553.

23

Table 18: Change in manufacturing trade union density by two figure SIC between 2001 and 2015.

4.7 UK Regions

Between 2001 and 2015 the UK region which had the greatest loss in trade union members was the North West, with a decline of 113,634 – see Table 19. This represented a loss of 12.3 per cent, but this was not the largest percentage loss. Thiswas the North East with a decline of 18.1 per cent. Only Wales had an increase in trade union membership, an increase of 7,553.

23

SIC92Change in

Density% changein density

19:Leather, leather goods manufacture -24.1 -76.526:Other non-metallic products manufacture -23.8 -60.716:Tobacco products manufacture -19.7 -40.732:Radio,TV,communication equipment man No data No data15:Food,beverage manufacture -16.9 -53.637:Recycling -15.8 -62.927:Basic metals manufacture -14.6 -34.134:Motor vehicle, trailer, etc., manufacture -14.0 -31.318:Clothing,fur manufacture -13.4 -67.022:Printing,publishing,recorded media -11.6 -46.617:Textile manufacture -11.2 -55.423:Coke,petrol products, nuclear fuel manufacturing -10.9 -26.636:Furniture etc manufacture -9.8 -67.324:Chemicals,chemical products man -9.0 -40.728:Fabric-metal products (not machine, equipment)manufacture -8.6 -42.335:Other transport equipment manufacture -8.6 -18.330:Office machines, computer manufacture -8.0 -59.429:Mach,equipment manufacture -7.7 -30.131:Elec machine, equipment manufacture -7.4 -40.233:Medical,precision,optical equipment man -6.5 -45.825:Rubber,plastic products manufacture -5.5 -27.420:Wood,straw,cork,wood prods(not furniture) -3.8 -30.921:Pulp,paper,paper prods manufacture -1.9 -6.1

Union membership Change in workforce Actual Density Actual %

North East -71,804 -18.1 98,538 10.0 North West -113,634 -12.3 236,895 8.7 Yorkshire & Humberside -49,411 -7.8 139,060 6.9 East Midlands -56,473 -11.3 148,836 8.4 West Midlands -75,458 -11.7 101,486 4.7 East of England -39,211 -7.3 207,209 8.9 London -92,901 -13.0 529,304 17.6 South East -24,264 -3.1 205,720 5.9 South West -43,060 -8.2 187,487 9.4 Wales 7,553 1.8 133,796 12.6 Scotland -3,002 -0.4 220,643 10.5 Northern Ireland -3,654 -1.5 108,483 18.2

Table 19: Change in trade union membership and employment between 2001 and 2015 by UK Regions. Source: LFS Oct-Dec 2001 and 2015

Table 20 shows the change in trade union density by UK region for the years 2001 and 2015. This table has been sorted to show the regions which had the highest union density in 2001. By 2015 the ranking of the top four in 2001 had changed. Northern Ireland declined to second place. While Wales moved up from being in third place to first place. The North East declined from second place to fourth place, and Scotland moved from fourth place to third place. The remaining eight regions remain relatively stationary in the rankings.

UK Region 2001 UK Region 2015 Northern Ireland 41.0 Wales 34.9 North East 40.2 Northern Ireland 34.1 Wales 38.6 Scotland 31.6 Scotland 35.0 North East 30.0 North West 34.1 North West 27.5 Yorkshire & Humberside 31.3 Yorkshire & Humberside 27.0 West Midlands 29.7 West Midlands 25.1 East Midlands 28.2 East Midlands 23.1 South West 26.1 South West 21.9 London 23.8 South East 20.6 East of England 23.1 East of England 19.7 South East 22.6 London 17.6

Table 20: Union density by region 2001-15 ranked for highest trade union density in 2001. Source: LFS Oct-Dec 2001 and 2015.

In regards to the employment composition of the workforce, this remained relatively stable between 2001 and 2015 - see Table 21.

24

Table 19: Change in trade union membership and employment between 2001 and 2015 by UK Regions. Source: LFS Oct-Dec 2001 and 2015

Table 20 shows the change in trade union density by UK region for the years 2001 and 2015. This table has been sorted to show the regions which had the highest union density in 2001. By 2015 the ranking of the top four in 2001 had changed. Northern Ireland declined to second place. While Wales moved up from being in third place to first place. The North East declined from second place to fourth place, and Scotland moved from fourth place to third place. The remaining eight regions remain relatively stationary in the rankings.

Table 20: Union density by region 2001-15 ranked for highest trade union density in 2001. Source: LFS Oct-Dec 2001 and 2015.

In regards to the employment composition of the workforce, this remained relatively stable between 2001 and 2015 – see Table 21.

24

Union membership Change in workforceActual Density Actual %

North East -71,804 -18.1 98,538 10.0North West -113,634 -12.3 236,895 8.7Yorkshire & Humberside -49,411 -7.8 139,060 6.9East Midlands -56,473 -11.3 148,836 8.4

West Midlands -75,458 -11.7 101,486 4.7East of England -39,211 -7.3 207,209 8.9London -92,901 -13.0 529,304 17.6South East -24,264 -3.1 205,720 5.9South West -43,060 -8.2 187,487 9.4Wales 7,553 1.8 133,796 12.6Scotland -3,002 -0.4 220,643 10.5Northern Ireland -3,654 -1.5 108,483 18.2

UK Region 2001 UK Region 2015Northern Ireland 41.0 Wales 34.9North East 40.2 Northern Ireland 34.1Wales 38.6 Scotland 31.6Scotland 35.0 North East 30.0North West 34.1 North West 27.5Yorkshire & Humberside 31.3 Yorkshire & Humberside 27.0West Midlands 29.7 West Midlands 25.1East Midlands 28.2 East Midlands 23.1

South West 26.1 South West 21.9London 23.8 South East 20.6East of England 23.1 East of England 19.7South East 22.6 London 17.6

Total Union Density Employment Composition % 2001 2015 2015-2001 2001 2015 2015-2001

North East 40.2 30.0 -10.2 4.1 4.1 -0.1 North West 34.1 27.5 -6.6 11.2 11.1 -0.2 Yorkshire & Humberside 31.3 27.0 -4.3 8.3 8.1 -0.1 East Midlands 28.2 23.1 -5.1 7.3 7.2 -0.4 West Midlands 29.7 25.1 -4.6 9.0 8.6 -0.1 East of England 23.1 19.7 -3.4 9.6 9.5 0.9 London 23.8 17.6 -6.2 12.4 13.4 -0.5 South East 22.6 20.6 -1.9 14.3 13.9 0.0 South West 26.1 21.9 -4.2 8.3 8.3 0.1 Wales 38.6 34.9 -3.7 4.4 4.5 0.1 Scotland 35.0 31.6 -3.5 8.7 8.7 0.2 Northern Ireland 41.0 34.1 -6.8 2.5 2.7 0.0

Table 21: The change in trade union density and employment composition by UK regions 2001 and 2015. Source LFS Oct - Dec 2001 and 2015.

So far we have examined the changes which have taken place on a region basis for trade union membership and density by UK Regions. However, we need also to look specifically at the changes which have taken place between the North and the South of England. The North of England is composed of the North East, the North West, and Yorkshire & Humberside. The South of England is composed of the East Midlands, the West Midlands, the East of England, London, the South East, and the South West. Trade union membership in the North declined from 1.95m to 1.72m, s representing a decline of 12 per cent, compared with the decline of 8.9 per cent for the South.

Trade union membership decreased by 234,849 in the North, and by 331,367 in the South. Employment of employees in the North increased 474,493 an increase of 8.3 per cent. In the South employment of employees increased by 1.4m an increase of 9.4 per cent - see Table 22.

Union membership Change in workforce Region Actual % Change Actual % Change North -234,849 -12.0 474,493 8.3 South -331,367 -8.9 1,380,042 9.4

Table 22: Change in trade union membership and employees by North and South of England between 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

Trade union density in the North declined from 34.2 per cent to 25.1 per cent, while the South declined from 25.1 to 20.9 per cent - see table 23

Between 2001 and 2015 the employment composition of both regions remain relatively stable.

25

Total Union Density Employment Composition %

North East North West Yorkshire &

2001 2015 2015-2001 2001 2015 2015-200140.2 30.0 -10.2 4.1 4.1 -0.134.1 27.5 -6.6 11.2 11.1 -0.2

Humberside 31.3 27.0 -4.3 8.3 8.1 -0.1East Midlands

West Midlands East of England LondonSouth EastSouth WestWales

ScotlandNorthern Ireland

28.2 23.1 -5.1 7.3 7.2 -0.429.7 25.1 -4.6 9.0 8.6 -0.123.1 19.7 -3.4 9.6 9.5 0.923.8 17.6 -6.2 12.4 13.4 -0.522.6 20.6 -1.9 14.3 13.9 0.026.1 21.9 -4.2 8.3 8.3 0.138.6 34.9 -3.7 4.4 4.5 0.135.0 31.6 -3.5 8.7 8.7 0.241.0 34.1 -6.8 2.5 2.7 0.0

Table 21: The change in trade union density and employment composition by UK regions 2001 and 2015. Source LFS Oct – Dec 2001 and 2015.

So far we have examined the changes which have taken place on a region basis for trade union membership and density by UK Regions. However, we need also to look specifically at the changes which have taken place between the North and the Southof England. The North of England is composed of the North East, the North West, andYorkshire & Humberside. The South of England is composed of the East Midlands,the West Midlands, the East of England, London, the South East, and the South West. Trade union membership in the North declined from 1.95m to 1.72m, s representing a decline of 12 per cent, compared with the decline of 8.9 per cent for the South.

Trade union membership decreased by 234,849 in the North, and by 331,367 in the South. Employment of employees in the North increased 474,493 an increase of 8.3 per cent. In the South employment of employees increased by 1.4m an increase of 9.4 per cent – see Table 22.

Table 22: Change in trade union membership and employees by North and South of England between 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

Trade union density in the North declined from 34.2 per cent to 25.1 per cent, while the South declined from 25.1 to 20.9 per cent – see table 23

Between 2001 and 2015 the employment composition of both regions remain relatively stable.

25

Union membership Change in workforceRegion Actual % Change Actual % ChangeNorth -234,849 -12.0 474,493 8.3South -331,367 -8.9 1,380,042 9.4

Union Density Employment Composition % Region 2001 2015 2015-2001 2001 2015 2015-2001 North 34.2 25.1 -6.4 27.9 27.7 -0.2 South 25.1 20.9 -4.2 72.1 72.3 0.2

Table 23: The change in trade union density and employment composition by North and South regions in England 2001 and 2015. Source LFS Oct - Dec 2001 and 2015.

The above analysis shows that both in 2001 and 2015 the proposition that trade union density would be greater in the North than in the South of England is supported. Although the North had the largest percentage decrease in both trade union membership and trade union density.

4.8 Occupation

As with the SIC standard occupation classifications (SOC) have changed over time due to the updating of international classification. In this case, we used SOC2000 occupational classification. This allows us again to keep as close as we can to the SOC used by Green (1992:450). The LFS User Guide Volume 5 (LFS Classifications) gives full occupational breakdowns for both SOC2000 & SOC2011.

Union membership Change in Workforce

Actual % Change Actual

%

Change Managers 1,829 0.3 622,692 1.5 Professional 213,464 15.8 982,081 3.0 Associate Professional 112,862 8.5 642,404 1.5 Administrative and Professional -291,898 -33.8 -522,014 -3.0 Skilled trades occupations -334,338 -47.7 -249,439 -1.6 Caring Leisure and other services 171,434 33.0 729,225 2.3 Sale and other customer service occupations 46,833 17.4 55,696 -0.4 Process plant and other operators -329,991 -44.0 -447,011 -2.3 Elementary occupations -145,148 -21.6 31,719 -0.8 Total -554,953 -7.9 1,845,353

Table 24: The change in trade union density and employment composition by occupation 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

Between 2001 and 2015 the greatest decrease in trade union membership took place in the skilled trade occupations. The membership decreased from 701,000 to 367,000 a decrease in density of 47.7 per cent. The second largest decease occurred in the

26

Table 23: The change in trade union density and employment composition by North and South regions in England 2001 and 2015. Source LFS Oct – Dec 2001 and 2015.

The above analysis shows that both in 2001 and 2015 the proposition that trade union density would be greater in the North than in the South of England is supported. Although the North had the largest percentage decrease in both trade union membership and trade union density.

4.8 Occupation

As with the SIC standard occupation classifications (SOC) have changed over time due to the updating of international classification. In this case, we used SOC2000 occupational classification. This allows us again to keep as close as we can to theSOC used by Green (1992:450). The LFS User Guide Volume 5 (LFS Classifications) gives full occupational breakdowns for both SOC2000 & SOC2011.

Table 24: The change in trade union density and employment composition by occupation 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

Between 2001 and 2015 the greatest decrease in trade union membership took placein the skilled trade occupations. The membership decreased from 701,000 to 367,000 a decrease in density of 47.7 per cent. The second largest decease occurred in the

26

Union Density Employment Composition %Region 2001 2015 2015-2001 2001 2015 2015-2001North 34.2 25.1 -6.4 27.9 27.7 -0.2South 25.1 20.9 -4.2 72.1 72.3 0.2

Union membership Change in Workforce

Actual % Change Actual%

ChangeManagers 1,829 0.3 622,692 1.5Professional 213,464 15.8 982,081 3.0AssociateProfessional 112,862 8.5 642,404 1.5Administrative andProfessional -291,898 -33.8 -522,014 -3.0Skilled tradesoccupations -334,338 -47.7 -249,439 -1.6Caring Leisure andother services 171,434 33.0 729,225 2.3Sale and othercustomer serviceoccupations 46,833 17.4 55,696 -0.4Process plant andother operators -329,991 -44.0 -447,011 -2.3Elementaryoccupations -145,148 -21.6 31,719 -0.8Total -554,953 -7.9 1,845,353

process plant and other operators occupations. Where trade union membership declined by 330,000 from 749,881 to 419,890 - see Table 24.

The largest actual increase in trade union membership occurred in the caring, leisure and other services occupations. They increased by 171,000 from 519,000 to 690,000, an increase of trade union density of 33.0 per cent. The second largest increase was in the associate professional occupations which increased by 113,000 from 1.3m to 1.4m an increase in trade union density of 8.5 per cent - see Table 25.

Union membership Change in Workforce Actual % Change Actual % Change

Managers 1,829 0.3 622,692 1.5 Professional 213,464 15.8 982,081 3.0 Associate Professional 112,862 8.5 642,404 1.5 Administrative and Professional -291,898 -33.8 -522,014 -3.0 Skilled trades occupations -334,338 -47.7 -249,439 -1.6 Caring Leisure and other services 171,434 33.0 729,225 2.3 Sale and other customer service occupations 46,833 17.4 55,696 -0.4 Process plant and other operators -329,991 -44.0 -447,011 -2.3 Elementary occupations -145,148 -21.6 31,719 -0.8 Total -554,953 -7.9 1,845,353

Table 25: The change in trade union density and employment composition by occupation 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

During the same period the largest decline in trade union density was in the skilled trades occupations where it decreased by 12.5 per cent from 30.1 to 17.7. The second largest decreased was in the process plant and other operators, which decreased by 10.3 per cent, from 36.1 per cent to 25.8 per cent - see Table 26. The only occupational group to experience an increase in trade union density was the sale and other customer service occupations, where it increased from 12.9 per cent to 14.8 per cent, an increase of 1.9 per cent .

In this section we are trying to establish if trade union membership and density is greater for manual than for non-manual employees. To do this we created a variable called manual which comprises of: skilled trade occupations; process plant and other operators; and elementary occupations. The other variable is made up of: managers; associate professional; administrative and professional; caring leisure and other services; and sale and other customer service occupations. Examination of the LFS

27

process plant and other operators occupations. Where trade union membership declined by 330,000 from 749,881 to 419,890 – see Table 24.

The largest actual increase in trade union membership occurred in the caring, leisureand other services occupations. They increased by 171,000 from 519,000 to 690,000,an increase of trade union density of 33.0 per cent. The second largest increase wasin the associate professional occupations which increased by 113,000 from 1.3m to1.4m an increase in trade union density of 8.5 per cent – see Table 25.

Table 25: The change in trade union density and employment composition by occupation 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

During the same period the largest decline in trade union density was in the skilled trades occupations where it decreased by 12.5 per cent from 30.1 to 17.7. The secondlargest decreased was in the process plant and other operators, which decreased by 10.3 per cent, from 36.1 per cent to 25.8 per cent – see Table 26.The only occupational group to experience an increase in trade union density was thesale and other customer service occupations, where it increased from 12.9 per cent to14.8 per cent, an increase of 1.9 per cent .

In this section we are trying to establish if trade union membership and density is greater for manual than for non-manual employees. To do this we created a variable called manual which comprises of: skilled trade occupations; process plant and other operators; and elementary occupations. The other variable is made up of: managers; associate professional; administrative and professional; caring leisure and other services; and sale and other customer service occupations. Examination of the LFS

27

Union membership Change in WorkforceActual % Change Actual % Change

Managers 1,829 0.3 622,692 1.5Professional 213,464 15.8 982,081 3.0AssociateProfessional 112,862 8.5 642,404 1.5Administrative andProfessional -291,898 -33.8 -522,014 -3.0Skilled tradesoccupations -334,338 -47.7 -249,439 -1.6Caring Leisure andother services 171,434 33.0 729,225 2.3Sale and othercustomer serviceoccupations 46,833 17.4 55,696 -0.4Process plant andother operators -329,991 -44.0 -447,011 -2.3Elementaryoccupations -145,148 -21.6 31,719 -0.8Total -554,953 -7.9 1,845,353

data reveals that even by 2001 in trade union density and absolute terms non-manual trade union membership was greater than manual trade union membership. In 2001 there were 2.1m manual trade union members, compared with 4.9m non-manual trade union members.

Union Density Employment Composition 2001 2015 Change 2001 2015 Change

Managers 17.2 14.5 -2.8 13.2 16.2 3.0 Professional 49.2 42.0 -7.2 11.4 14.1 2.7 Associate Professional 41.2 37.3 -3.9 13.4 14.6 1.3 Administrative and Professional 24.2 18.8 -5.4 14.7 11.5 -3.3

30.1 17.7 -12.5 9.6 7.8 -1.8 29.0 27.4 -1.6 7.4 9.5 2.1

Skilled trades occupations 12.9 14.8 1.9 8.6 8.1 -0.5 Caring Leisure and other services 36.1 25.8 -10.3 8.6 6.1 -2.4 Elementary occupations 21.1 16.4 -4.7 13.1 12.1 -1.0

Table 26: The change in trade union density and employment composition by occupation 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

By 2015 manual trade union membership had declined 809,000 to 1.3m. During the same period non-manual trade union membership had increased by 254,523 to 5.1m. Meanwhile manual employees declined from 7.6m to 6.9m a decrease of 665,000. Non-manual employment increased from 16.6m to 19.1m an increase of 2.5m.

Union membership Change in workforce Occupation Actual % Change Actual % Change Manual -809,477 -38.1 -664,731 -8.8 Non-manual 254,523 5.2 2,510,084 15.1

Table 27: The change in trade union membership and employment by manual and non-manual occupations 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The trade union density was 28.0 per cent for the manual occupations in 2001 and had declined to 19.0 per cent by 2015 - see Table 28.

Union Density Employment Composition % Occupations 2001 2015 2015-2001 2001 2015 2015-2001 Manual 28.0 19.0 -9.0 74.2 74.3 0.1 Non-manual 29.4 26.9 -2.5 25.8 25.7 -0.1

Table 28: The change in trade union density and employment composition for manual and non-manual occupations 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

28

data reveals that even by 2001 in trade union density and absolute terms non-manual trade union membership was greater than manual trade union membership. In 2001there were 2.1m manual trade union members, compared with 4.9m non-manual tradeunion members.

Table 26: The change in trade union density and employment composition by occupation 2001 and 2015. Source: LFS Oct-Dec 2001 and 2015.

By 2015 manual trade union membership had declined 809,000 to 1.3m. During thesame period non-manual trade union membership had increased by 254,523 to 5.1m.Meanwhile manual employees declined from 7.6m to 6.9m a decrease of 665,000.Non-manual employment increased from 16.6m to 19.1m an increase of 2.5m.

Table 27: The change in trade union membership and employment by manual andnon-manual occupations 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The trade union density was 28.0 per cent for the manual occupations in 2001 and had declined to 19.0 per cent by 2015 – see Table 28.

Table 28: The change in trade union density and employment composition for manual and non-manual occupations 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

28

Union Density Employment Composition2001 2015 Change 2001 2015 Change

Managers 17.2 14.5 -2.8 13.2 16.2 3.0Professional 49.2 42.0 -7.2 11.4 14.1 2.7AssociateProfessional 41.2 37.3 -3.9 13.4 14.6 1.3Administrativeand Professional 24.2 18.8 -5.4 14.7 11.5 -3.3

30.1 17.7 -12.5 9.6 7.8 -1.829.0 27.4 -1.6 7.4 9.5 2.1

Skilled tradesoccupations 12.9 14.8 1.9 8.6 8.1 -0.5Caring Leisureand other services 36.1 25.8 -10.3 8.6 6.1 -2.4Elementaryoccupations 21.1 16.4 -4.7 13.1 12.1 -1.0

Union membership Change in workforceOccupation Actual % Change Actual % ChangeManual -809,477 -38.1 -664,731 -8.8Non-manual 254,523 5.2 2,510,084 15.1

Union Density Employment Composition %Occupations 2001 2015 2015-2001 2001 2015 2015-2001Manual 28.0 19.0 -9.0 74.2 74.3 0.1Non-manual 29.4 26.9 -2.5 25.8 25.7 -0.1

During the same period the employment composition remained relatively stable.

In this section we have demonstrated that trade union membership and density is not greater for manual than for non-manual employees, and has been so before 2001. Both in regards to trade union density and trade union membership, manual trade unionism by density and actual numbers has declined significantly since 2001.

4.9 Change in trade union membership due to age

The change in the age profile of trade union members between 2001 and 2015 can be seen clearly in Table 29. For the age groups between 16 and 49 the decrease in trade union membership was 1.03m, a decrease of -20.2 per cent. The majority of this decrease came from the 30 to 44 age group which decreased by 830,491, a decrease of 27.6 per cent. For the age group 50 to over 70 trade union membership increased by 461,416, an increase of 23.5 per cent.

Union membership Change in workforce Age Change % Change %

16 to 19 -52,002 -67.0 -400,790 -27.6 20 to 24 -26,535 -7.8 333,807 14.0 25 to 29 -54,145 -8.7 447,814 16.1 30 to 34 -205,586 -23.1 -57,610 -1.8 35 to 39 -310,664 -30.1 -412,408 -12.6 40 to 44 -314,242 -28.8 -38,570 -1.3 45 to 49 -63,563 -6.2 553,640 21.2 50 to 54 93,596 9.3 499,369 19.5 55 to 59 138,284 20.9 501,242 27.0 60 to 64 137,505 51.1 514,157 60.8 65 to 69 75,960 413.3 277,308 145.5 Over 70 16,072 219.7 99,498 112.3 Table 29: The change in trade union membership and employment by age groups

2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

In regards to the size of the workforce, for the age groups between 16 and 49 increased by 425,883, and increase of 2.3 per cent. For the age group 50 to over 70 employment increased by 1.9m, an increase of 34.1 per cent.

In contrast to changes in trade union membership, only two age groups had an increase in trade union density - see Table 30. These were the 65-69 and over 70 age groups, the 65-69 age group had the largest increase, an increase of 9.6 per cent from 10.6 per cent to 20.2 per cent. The age group which experienced the greatest decline in trade union density was the 40 to 44 age group. Trade union density declined from 36.5 per cent to 26.3 per cent. This was followed by the 45 to 49 age group where trade union density declined from 39.2 per cent to 30.3 per cent.

29

During the same period the employment composition remained relatively stable.

In this section we have demonstrated that trade union membership and density is not greater for manual than for non-manual employees, and has been so before 2001. Both in regards to trade union density and trade union membership, manual trade unionism by density and actual numbers has declined significantly since 2001.

4.9 Change in trade union membership due to age

The change in the age profile of trade union members between 2001 and 2015 can be seen clearly in Table 29. For the age groups between 16 and 49 the decrease in trade union membership was 1.03m, a decrease of -20.2 per cent. The majority of this decrease came from the 30 to 44 age group which decreased by 830,491, a decreaseof 27.6 per cent. For the age group 50 to over 70 trade union membership increasedby 461,416, an increase of 23.5 per cent.

Union membership Change in workforceAge Change % Change %16 to 19 -52,002 -67.0 -400,790 -27.620 to 24 -26,535 -7.8 333,807 14.025 to 29 -54,145 -8.7 447,814 16.130 to 34 -205,586 -23.1 -57,610 -1.835 to 39 -310,664 -30.1 -412,408 -12.640 to 44 -314,242 -28.8 -38,570 -1.345 to 49 -63,563 -6.2 553,640 21.250 to 54 93,596 9.3 499,369 19.555 to 59 138,284 20. 501,242 27.060 to 64 137,50565 to 69 75,960

951.413.3

514,157 60.8277,308 145.5

Over 70 16,072 219.7 99,498 112.3Table 29: The change in trade union membership and employment by age groups

2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

In regards to the size of the workforce, for the age groups between 16 and 49increased by 425,883, and increase of 2.3 per cent. For the age group 50 to over 70employment increased by 1.9m, an increase of 34.1 per cent.

In contrast to changes in trade union membership, only two age groups had an increase in trade union density – see Table 30. These were the 65-69 and over 70 agegroups, the 65-69 age group had the largest increase, an increase of 9.6 per cent from10.6 per cent to 20.2 per cent. The age group which experienced the greatest declinein trade union density was the 40 to 44 age group. Trade union density declined from36.5 per cent to 26.3 per cent. This was followed by the 45 to 49 age group wheretrade union density declined from 39.2 per cent to 30.3 per cent.

29

Union Density Employment Composition %

Age 2001 2015 2001-2015 2001 2015 2015-2001 16-19 5.4 2.4 -2.9 6.0 4.0 -2.0 20-24 14.3 11.6 -2.7 9.9 10.2 0.4 25-29 22.6 17.6 -4.9 11.5 12.2 0.7 30 to 34 27.9 21.9 -6.1 13.1 11.8 -1.4 35 to 39 31.5 25.2 -6.3 13.5 10.8 -2.7 40 to 44 36.5 26.3 -10.2 12.3 11.1 -1.2 45 to 49 39.2 30.3 -8.9 10.8 12.0 1.1 50 to 54 39.5 36.0 -3.5 10.6 11.5 1.0 55 to 59 35.7 34.0 -1.7 7.7 8.9 1.2 60 to 64 32.0 29.9 -2.1 3.5 5.1 1.6 65-69 10.6 20.2 9.6 0.8 1.8 1.0 Over 70 7.8 12.4 4.7 0.4 0.7 0.3 Table 30: The change in trade union density and employment composition by age

groups 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The overall employment composition of employees by individual age groups remained fairly constant, however, the 16 to 49 age group declined from 77.1 per cent to 72.0 per cent. Conversely, the 50 to over 70 age group increased from 22.9 per cent to 28.0 per cent.

4.9.1 Change in Core Workforce

We now examine what has happened to the core workforce group of employees between 25 and 54.

The decline in the core group trade union membership impact on the overall trade union membership is shown in Table 31. In this core age group - 25 to 54 - trade union membership declined by 855,000, a decline of 15.1 per cent. During the same period employment increased by just under six per cent. From 17.4m to just under 18.0m.

Year Union

members Density Employment Composition 2001 5,665,928 32.6 17,394,430 71.9 2015 4,811,325 26.2 18,386,665 69.3

Table 31: Change in core working age group 2001-15 Source LFS Oct-Dec 2001 and 2015.

Trade union density declined by 6.4 per cent from 32.6 per cent to 26.2 per cent.

The analysis of the various age groupings supports the proposition that trade union density is higher for older than for younger employees. However, of most concern to the trade unions will be the 855,000 decline in the core workforce. Unless this is reversed it does not bold well for the future of the movement.

30

Union Density Employment Composition %Age 2001 2015 2001-2015 2001 2015 2015-200116-19 5.4 2.4 -2.9 6.0 4.0 -2.020-24 14.3 11.6 -2.7 9.9 10.2 0.425-29 22.6 17.6 -4.9 11.5 12.2 0.730 to 34 27.9 21.9 -6.1 13.1 11.8 -1.435 to 39 31.5 25.2 -6.3 13.5 10.8 -2.740 to 44 36.5 26.3 -10.2 12.3 11.1 -1.245 to 49 39.2 30.3 -8.9 10.8 12.0 1.150 to 54 39.5 36.0 -3.5 10.6 11.5 1.055 to 59 35.7 34.0 -1.7 7.7 8.9 1.260 to 64 32.0 29.9 -2.1 3.5 5.1 1.665-69 10.6 20.2 9.6 0.8 1.8 1.0Over 70 7.8 12.4 4.7 0.4 0.7 0.3Table 30: The change in trade union density and employment composition by age

groups 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The overall employment composition of employees by individual age groups remainedfairly constant, however, the 16 to 49 age group declined from 77.1 per cent to 72.0per cent. Conversely, the 50 to over 70 age group increased from 22.9 per cent to 28.0per cent.

4.9.1 Change in Core Workforce

We now examine what has happened to the core workforce group of employees between 25 and 54.

The decline in the core group trade union membership impact on the overall trade union membership is shown in Table 31. In this core age group – 25 to 54 - trade unionmembership declined by 855,000, a decline of 15.1 per cent. During the same periodemployment increased by just under six per cent. From 17.4m to just under 18.0m.

Table 31: Change in core working age group 2001-15 Source LFS Oct-Dec 2001 and2015.

Trade union density declined by 6.4 per cent from 32.6 per cent to 26.2 per cent.

The analysis of the various age groupings supports the proposition that trade union density is higher for older than for younger employees. However, of most concern to the trade unions will be the 855,000 decline in the core workforce. Unless this is reversed it does not bold well for the future of the movement.

30

YearUnion

members Density Employment Composition2001 5,665,928 32.6 17,394,430 71.92015 4,811,325 26.2 18,386,665 69.3

4.10 Public vs Private Sector

Between 2001 and 2015 private sector trade union membership declined from 3.3m to 2.7m a decrease 0.6m, this represents a decline of 18 per cent. During the same period public sector trade union membership increased from 3.71m by 12,780 to 3.72m, an increase of 0.3 per cent - see Table 32.

Union membership Change in workforce Sector Actual % Change Actual % Change Private -590,733 -18.0 1,715,342 9.6 Public 12,780 0.3 556,657 8.9

Table 32: The change in trade union membership and employment by sector 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

At the same time the private sector workforce increased by 1.7m from 17.9m to 19.6m, a growth of 9.6 per cent. Public sector employment increased from 6.2m to 6.8m an increase of 0.6m.

Trade union density in the private sector declined from 18.4 per cent to 13.8 per cent between 2001 and 2015. During the same period public sector trade union density declined from 59.5 to 54.8. The decline in trade union density was remarkably similar at 4.6 for the private sector, and 4.7 for the public sector - see Table 33.

Union Density Employment Composition % Sector 2001 2015 2015-2001 2001 2015 2015-2001 Private 18.4 13.8 -4.6 74.2 74.3 0.1 Public 59.5 54.8 -4.7 25.8 25.7 -0.1

Table 33: The change in trade union density and employment composition for private and public sectors 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The employment composition of the two sectors remained constant during this period.

The above analysis confirms that trade union density is greater in the public sector than the private sector.

4.11 Trade Union Coverage in the Private and Public Sector

In this section we are not measuring the actual number of employees who are trade union members, but the number of employees who are covered by trade union collective agreements in the private and public sectors. Although the workforce population remains the same, the calculation for trade union coverage is different, within the LFS there is a specific variable called TUCOV. The question 'applies to all respondents excluding those in certain government schemes and unpaid workers' (ONS 2015:171).

The number of employees in the private sector reporting being covered by a trade union collective agreement decreased from 3.9m to 3.2m between 2001 and 2015. A decrease of 19.4 per cent. During the same period the number employees in the public

31

4.10 Public vs Private Sector

Between 2001 and 2015 private sector trade union membership declined from 3.3m to 2.7m a decrease 0.6m, this represents a decline of 18 per cent. During the same period public sector trade union membership increased from 3.71m by 12,780 to 3.72m, an increase of 0.3 per cent – see Table 32.

Table 32: The change in trade union membership and employment by sector 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

At the same time the private sector workforce increased by 1.7m from 17.9m to 19.6m, a growth of 9.6 per cent. Public sector employment increased from 6.2m to 6.8m an increase of 0.6m.

Trade union density in the private sector declined from 18.4 per cent to 13.8 per cent between 2001 and 2015. During the same period public sector trade union densitydeclined from 59.5 to 54.8. The decline in trade union density was remarkably similarat 4.6 for the private sector, and 4.7 for the public sector – see Table 33.

Table 33: The change in trade union density and employment composition for private and public sectors 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The employment composition of the two sectors remained constant during this period.

The above analysis confirms that trade union density is greater in the public sector than the private sector.

4.11 Trade Union Coverage in the Private and Public Sector

In this section we are not measuring the actual number of employees who are trade union members, but the number of employees who are covered by trade union collective agreements in the private and public sectors. Although the workforce population remains the same, the calculation for trade union coverage is different, within the LFS there is a specific variable called TUCOV. The question ‘applies to all respondents excluding those in certain government schemes and unpaid workers’(ONS 2015:171).

The number of employees in the private sector reporting being covered by a trade union collective agreement decreased from 3.9m to 3.2m between 2001 and 2015. A decrease of 19.4 per cent. During the same period the number employees in the public

31

Union membership Change in workforceSector Actual % Change Actual % ChangePrivate -590,733 -18.0 1,715,342 9.6Public 12,780 0.3 556,657 8.9

Union Density Employment Composition %Sector 2001 2015 2015-2001 2001 2015 2015-2001Private 18.4 13.8 -4.6 74.2 74.3 0.1Public 59.5 54.8 -4.7 25.8 25.7 -0.1

sector reporting being covered by a trade union collective agreement decreased from 4.5m to 4.1m - see Table 34.

Trade Union Coverage Change in Workforce Sector Actual % Change Actual % Change Private -759,956 -19.4 1,715,342 9.6 Public -402,603 -8.9 556,657 8.9

Table 34: Trade union coverage and employees in the private and public sector 2001 and 2015. Source LFS Oct-Dec 2001 and 2015

In regards to employees in the public and private sector, employment increased by a similar percentage. For the private sector the increase was 1.7m from 17.9m to 19.6m.

The change in trade union coverage was smaller for the private sector, the coverage decreased by 5.8 per cent, compared with the public sector decline of 11.9 per cent between 2001 and 2015 - see table 35.

Trade Union Coverage Employment Composition % Sector 2001 2015 2015-2001 2001 2015 2015-2001 Private 21.9 16.1 -5.8 74.2 74.3 0.1 Public 72.6 60.7 -11.9 25.8 25.7 -0.1

Table 35: The change in trade union coverage and employment composition for private and public sectors 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The employment composition over the same period remained constant, changing by only 0.1 per cent.

The analysis shows clearly that the decline in employees being covered by a trade union collective agreement was in the public sector, this decline was twice the decline in the private sector. Although the actual decline in coverage was greater in the public sector, it still had the greater coverage than the private sector by a considerable amount.

4.12 Trade Union Presence

In regards to trade union presence, the LFS asks the question 'Whether other people at workplace are members of a trade union or staff association" (Office for National Statistics 2015: 171). And the question is asked of all respondents in employment who are not members of a staff association or trade union.

For the private sector the number of people reporting that other people in the workplace were union members decreased by 0.6m, a decrease of 9.9 per cent from 6.0m to 5.5m - see Tale 29. In the public sector the number of employees reporting the presence of a trade union increased by 3.9 per cent - see Table 36.

32

sector reporting being covered by a trade union collective agreement decreased from 4.5m to 4.1m – see Table 34.

Table 34: Trade union coverage and employees in the private and public sector 2001 and 2015. Source LFS Oct-Dec 2001 and 2015

In regards to employees in the public and private sector, employment increased by a similar percentage. For the private sector the increase was 1.7m from 17.9m to 19.6m.

The change in trade union coverage was smaller for the private sector, the coverage decreased by 5.8 per cent, compared with the public sector decline of 11.9 per cent between 2001 and 2015 – see table 35.

Table 35: The change in trade union coverage and employment composition for private and public sectors 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

The employment composition over the same period remained constant, changing by only 0.1 per cent.

The analysis shows clearly that the decline in employees being covered by a trade union collective agreement was in the public sector, this decline was twice the decline in the private sector. Although the actual decline in coverage was greater in the public sector, it still had the greater coverage than the private sector by a considerable amount.

4.12 Trade Union Presence

In regards to trade union presence, the LFS asks the question “Whether other people at workplace are members of a trade union or staff association” (Office for NationalStatistics 2015: 171). And the question is asked of all respondents in employment who are not members of a staff association or trade union.

For the private sector the number of people reporting that other people in the workplace were union members decreased by 0.6m, a decrease of 9.9 per cent from 6.0m to 5.5m – see Tale 29. In the public sector the number of employees reporting the presence of a trade union increased by 3.9 per cent – see Table 36.

32

Trade Union Coverage Change in WorkforceSector Actual % Change Actual % ChangePrivate -759,956 -19.4 1,715,342 9.6Public -402,603 -8.9 556,657 8.9

Trade Union Coverage Employment Composition %Sector 2001 2015 2015-2001 2001 2015 2015-2001Private 21.9 16.1 -5.8 74.2 74.3 0.1Public 72.6 60.7 -11.9 25.8 25.7 -0.1

Trade Union Presence Change in Workforce Sector Actual % Change Actual % Change Private -600,582 -9.9 1,715,342 9.6 Public 212,585 3.9 556,657 8.9

Table 36: Trade union presence and employees in the private and public sector 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

In regards to employees in the public and private sector, employment increased by a similar percentage. For the private sector the increase was 1.7m from 17.9m to 19.6m - see Table 37.

Trade Union Presence Employment Composition % Sector 2001 2015 2015-2001 2001 2015 2015-2001 Private 34.0 28.0 -6.0 74.2 74.3 0.1 Public 88.1 84.0 -4.1 25.8 25.7 -0.1

Table 37: The change in trade union presence and employment composition for private and public sectors 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

Those reporting a trade union presence in a private sector workplace decreased from 34.0 per cent to 28.0 per cent a decrease of 6.0 per cent. For the public sector the decrease was slightly less, decreasing from 88.1 per cent to 84.0 a decrease of 4.1 per cent.

In regards to changes in the composition of the workforce, the largest positive change was in the size of establishment, just over five per cent of the variation was due to the change in the establishment size. Conversely, changes in occupations had a negative effect on the composition of the workforce.

5.0 Conclusion

5.1 Overview

In this paper we have critically examined the changes which have taken place in trade union membership in the UK between 2001 and 2015 using data from the LFS. In so doing we have warned about relying solely on using trade union density to explain changes in trade union membership, if trade union density is used on its own it can mask what is really happening to the actual trade union membership.

5.2 Change in trade union membership

The title of the paper ask the question where have 600,000 trade union members gone? This 600,000 is a net figure, in Table 38 we show the change in the number of trade union members by selected categories, and from this we can say that the actual number of trade union members declined by 0.85m between 2001 and 2015.

33

Table 36: Trade union presence and employees in the private and public sector 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

In regards to employees in the public and private sector, employment increased by a similar percentage. For the private sector the increase was 1.7m from 17.9m to 19.6m – see Table 37.

Table 37: The change in trade union presence and employment composition for private and public sectors 2001 and 2015. Source LFS Oct-Dec 2001 and 2015.

Those reporting a trade union presence in a private sector workplace decreased from 34.0 per cent to 28.0 per cent a decrease of 6.0 per cent. For the public sector the decrease was slightly less, decreasing from 88.1 per cent to 84.0 a decrease of 4.1 per cent.

In regards to changes in the composition of the workforce, the largest positive changewas in the size of establishment, just over five per cent of the variation was due to thechange in the establishment size. Conversely, changes in occupations had a negativeeffect on the composition of the workforce.

5.0 Conclusion

5.1 Overview

In this paper we have critically examined the changes which have taken place in trade union membership in the UK between 2001 and 2015 using data from the LFS. In so doing we have warned about relying solely on using trade union density to explain changes in trade union membership, if trade union density is used on its own it can mask what is really happening to the actual trade union membership.

5.2 Change in trade union membership

The title of the paper ask the question where have 600,000 trade union members gone? This 600,000 is a net figure, in Table 38 we show the change in the number oftrade union members by selected categories, and from this we can say that the actualnumber of trade union members declined by 0.85m between 2001 and 2015.

33

Trade Union Presence Change in WorkforceSector Actual % Change Actual % ChangePrivate -600,582 -9.9 1,715,342 9.6Public 212,585 3.9 556,657 8.9

Trade Union Presence Employment Composition %Sector 2001 2015 2015-2001 2001 2015 2015-2001Private 34.0 28.0 -6.0 74.2 74.3 0.1Public 88.1 84.0 -4.1 25.8 25.7 -0.1

Category Loss Core Working Age Group 854,603 All Manual Occupations 809,477 Male 794,881 All Manufacturing Sectors 750,362 Full-time 734,002 Private Sector 590,733 50 and over 523,163 All Southern Regions 331,367

Table 38: Change in trade union membership by selected categories 2001-15. Source LFS Oct-Dec 2001 and 2015.

Generally we can say that the decline in trade union membership came primarily from: the core male working age group, manufacturing occupations, male members, all the manufacturing sectors, and full-time trade union members. At the same time this loss was compensated for by some degree by the increase of all female and male part-time trade union members, which accounted for in total 245,000 members — see Table 6.

5.3 Where trade union density is higher

In this paper we have also critically examined the propositions from earlier studies, Green 1992:446 referring to Bain and Elsheikh 1979, or Booth 1986, where trade union density was higher. We have compared these propositions with the data for 2015 and Table 39 summaries the results of this analysis. In addition to the variables used by Green, we have used the public and private sector.

Proposition Result Trade union density is greater for males than for females False Trade union density is greater for full-timers than for part-timers True Trade union density is greater in larger than in smaller establishments True Trade union density is greater in manufacturing industries than in service industries False Trade union density is greater in the North than in the South True Trade union density is greater for manual than for non-manual workers False Trade union density is greater for older than for younger workers True Trade union density is greater in the public sector than the private sector True

Table 39: Summary of results of analysis

Since 2001 there have been significant changes in the composition of trade union membership in the UK. Possibly the most significant of these is that there are now more female trade union members than males in the UK.

Trade union membership in the UK is now predominately in the service sectors rather than the manufacturing sectors. However, between the late 1970s and the 1980's and 2001 there had already been a shift in trade union membership towards the services industries sectors. Consequently trade union density is now greater for non-manual employees.

34

Table 38: Change in trade union membership by selected categories 2001-15. Source LFS Oct-Dec 2001 and 2015.

Generally we can say that the decline in trade union membership came primarily from: the core male working age group, manufacturing occupations, male members, all themanufacturing sectors, and full-time trade union members. At the same time this losswas compensated for by some degree by the increase of all female and male part-time trade union members, which accounted for in total 245,000 members – see Table6.

5.3 Where trade union density is higher

In this paper we have also critically examined the propositions from earlier studies, Green 1992:446 referring to Bain and Elsheikh 1979, or Booth 1986, where trade union density was higher. We have compared these propositions with the data for 2015 and Table 39 summaries the results of this analysis. In addition to the variables used by Green, we have used the public and private sector.

Table 39: Summary of results of analysis

Since 2001 there have been significant changes in the composition of trade union membership in the UK. Possibly the most significant of these is that there are now more female trade union members than males in the UK.

Trade union membership in the UK is now predominately in the service sectors rather than the manufacturing sectors. However, between the late 1970s and the 1980’s and 2001 there had already been a shift in trade union membership towards the services industries sectors. Consequently trade union density is now greater for non-manual employees.

34

Category LossCore Working Age Group 854,603All Manual Occupations 809,477Male 794,881All Manufacturing Sectors 750,362Full-time 734,002Private Sector 590,73350 and over 523,163All Southern Regions 331,367

Proposition ResultTrade union density is greater for males than for females FalseTrade union density is greater for full-timers than for part-timers TrueTrade union density is greater in larger than in smaller establishments TrueTrade union density is greater in manufacturing industries than in serviceindustries FalseTrade union density is greater in the North than in the South TrueTrade union density is greater for manual than for non-manual workers FalseTrade union density is greater for older than for younger workers TrueTrade union density is greater in the public sector than the private sector True

In this paper we have concentrated only on the highest aggregated data, except for the manufacturing sector, future research will need to look in more at the disaggregated data, in the way which we disaggregated the manufacturing sector —see Table 4 - to explain more fully by employment status where 600,000 trade union members have gone!

35

In this paper we have concentrated only on the highest aggregated data, except for the manufacturing sector, future research will need to look in more at the disaggregated data, in the way which we disaggregated the manufacturing sector – see Table 4 - to explain more fully by employment status where 600,000 trade union members have gone!

35

6.0 Bibliography

Certification Officer (2015) 'Annual Report of the Certification Officer 2014-2015' https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/53793 3/12450_C.0._15_Ann._Rep._2015-16_Appendix_4.pdf. (Accessed 26 September 2017).

Department for Business Innovation and Skills (8 April 2013) 'Sampling Variance in the Trade Union Membership Statistics' https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/18469 6/13-732-sampling-variance-in-the-trade-union-membership-statistics.pdf. (Accessed 26 September 2017).

Department for Business Innovation and Skills (2015a) TRADE UNION MEMBERSHIP 2015 Statistical Bulletin MAY 2016.

Department for Business Innovation and Skills (2015a) 'Trade Union Membership 2015 - Confidence Intervals' https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/52490 3/Trade_Union_Membership_Statistics_2015_-_Confidence_Intervals.xlsx. (Accessed 26 September 2017).

Everitt B S (2002) The Cambridge Dictionary of Statistics, Second Edition Cambridge University Press.

Green, Francis. (1992) 'Recent Trends in British Trade Union Density: How Much of a Compositional Effect'. British Journal of Industrial Relations, 30:3 September 1992 pp. 445-458.

Office for National Statistics. Social Survey Division and Northern Ireland Statistics and Research Agency. Central Survey Unit, Quarterly Labour Force Survey, October - December, 2001 [computer file]. 3rd Edition. Colchester, Essex: UK Data Archive [distributor], November 2014. SN: 5419 , http://dx.doi.org/10.5255/UKDA-SN-5419-2

Office for National Statistics (2015a) 'Quality and Methodology Information'. Newport: Office for National Statistics 13th January.

Office for National Statistics (2015b) LABOUR FORCE SURVEY USER GUIDE VOLUME 3 — DETAILS OF LFS VARIABLES 2015 VERSION 2 AUGUST 2015

Office for National Statistics. Social Survey Division, Northern Ireland Statistics and Research Agency. Central Survey Unit. (2016). Quarterly Labour Force Survey, October - December, 2015. [data collection]. UK Data Service. SN: 7902, http://dx.doi.org/10.5255/UKDA-SN-7902-1.

Office for National Statistics (2016) LABOUR FORCE SURVEY USER GUIDE VOLUME 1 — LFS BACKGROUND AND METHODOLOGY 2016.

36

6.0 BibliographyCertification Officer (2015) ‘Annual Report of the Certification Officer 2014-2015’https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/537933/12450_C.O._15_Ann._Rep._2015-16_Appendix_4.pdf. (Accessed 26 September2017).

Department for Business Innovation and Skills (8 April 2013) ‘Sampling Variance in the Trade Union Membership Statistics’https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/184696/13-732-sampling-variance-in-the-trade-union-membership-statistics.pdf. (Accessed 26 September 2017).

Department for Business Innovation and Skills (2015a) TRADE UNION MEMBERSHIP 2015 Statistical Bulletin MAY 2016.

Department for Business Innovation and Skills (2015a) ‘Trade Union Membership2015 - Confidence Intervals’https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/524903/Trade_Union_Membership_Statistics_2015_-_Confidence_Intervals.xlsx.(Accessed 26 September 2017).

Everitt B S (2002) The Cambridge Dictionary of Statistics, Second Edition Cambridge University Press.

Green, Francis. (1992) ‘Recent Trends in British Trade Union Density: How Much of aCompositional Effect’. British Journal of Industrial Relations, 30:3 September 1992 pp. 445-458.

Office for National Statistics. Social Survey Division and Northern Ireland Statistics and Research Agency. Central Survey Unit, Quarterly Labour Force Survey, October - December, 2001 [computer file]. 3rd Edition. Colchester, Essex: UK Data Archive [distributor], November 2014. SN: 5419 , http://dx.doi.org/10.5255/UKDA-SN-5419-2

Office for National Statistics (2015a) ‘Quality and Methodology Information’. Newport: Office for National Statistics 13th January.13th Jan

Office for National Statistics (2015b) LABOUR FORCE SURVEY USER GUIDE VOLUME 3 – DETAILS OF LFS VARIABLES 2015 VERSION 2 AUGUST 2015

Office for National Statistics. Social Survey Division, Northern Ireland Statistics and Research Agency. Central Survey Unit. (2016). Quarterly Labour Force Survey, October - December, 2015. [data collection]. UK Data Service. SN: 7902,http://dx.doi.org/10.5255/UKDA-SN-7902-1.

Office for National Statistics (2016) LABOUR FORCE SURVEY USER GUIDE VOLUME 1 – LFS BACKGROUND AND METHODOLOGY 2016.

36

37 37