Earnings top-up evaluation: effects on unemployed people

Department of Social Security Research Report No 131 Earnings Top-up Evaluation: Effects on Unemployed People Part One Surveys of Unemployed People Alison Smith and Richard Dorsett Part Two Econometric Analysis Abigail McKnight A report of research carried out by the Policy Studies Institute (PSI) and the Institute for Employment Research (IER) on behalf of the Department of Social Security Corporate Document Services © Crown Copyright 2001. Published for the Department of Social Security under licence from the Controller of Her Majesty’s Stationery Office by Corporate Document Services, Leeds. Application for reproduction should be made in writing to The Copyright Unit, Her Majesty’s Stationery Office, St Clements House, 2-16 Colegate, Norwich NR3 1BQ. First Published 2001. ISBN 1 84123 289 0 Views expressed in this report are not necessarily those of the Department of Social Security or any other Government Department. Printed by The Charlesworth Group (Huddersfield, UK). CONTENTS Acknowledgements The Authors xi xii List of abbreviations Preface xv Summary 1 xiii Part One - Surveys of Unemployed People 1 Introduction 13 1.1 Overview 13 1.2 The ETU evaluation 13 1.3 The progress of ETU 15 1.3.1 Trends in ETU claims 16 1.3.2 Introduction of the National Minimum Wage 1.4 Summary 21 2 The characteristics of the unemployed sample 2.1 Introduction 23 2.2 Gender, age and ethnicity 24 2.3 Marital status 26 2.4 Household type 27 2.5 Housing tenure 28 2.6 Education and training 30 2.7 Health 32 2.8 Area differences 34 2.9 Summary 34 3 Income and benefits 37 3.1 Introduction 37 3.2 Receipt of benefits 37 3.2.1 Benefit receipt at first interview 37 3.2.2 Benefit receipt at second interview 39 3.3 Average benefit received 40 3.4 Housing costs 41 3.5 Savings 42 3.6 Material and financial well-being 43 3.7 Summary 45 4 Contact with the labour market 47 4.1 Introduction 47 4.2 Economic status at interview 47 4.2.1 Economic status at first interview 4.2.2 Economic status at second interview 19 23 47 52 iii 4.3 4.4 4.5 4.6 5 6 7 iv Previous work experience 56 Looking for work 59 4.4.1 Jobsearch activity 59 4.4.2 Jobsearch methods 66 4.4.3 Work and wage expectations 67 4.4.4 Aspirations for the future 71 Movements into work 73 4.5.1 Looking for work at second interview 4.5.2 Workers at first interview 75 4.5.3 Workers at second interview 76 Summary 78 4.6.1 Movements into work 78 4.6.2 Looking for work 79 4.6.3 Aspirations for the future 79 74 Labour market outcomes 81 5.1 Introduction 81 5.2 Labour market transitions 82 5.2.1 An overview of the sample 82 5.2.2 Modelling the move away from unemployment 5.3 Wages 89 5.4 Expected wages 91 5.5 Summary and discussion 94 The experiences of Earnings Top-up 95 6.1 Introduction 95 6.2 Awareness of ETU 95 6.2.1 Awareness of ETU at first interview 6.2.2 Awareness of ETU at second interview 6.3 Experiences of ETU 102 6.3.1 Experience of ETU at first interview 6.3.2 Experience of ETU at second interview 6.4 Attitudes towards an ETU benefit 104 6.5 Summary 106 Conclusions 109 7.1 First interview - summer 1998 109 7.2 Second interview - summer 1999 111 95 100 102 103 84 Part Two - Econometric Analysis 8 9 Introduction 115 8.1 The ETU pilot: aims and objectives 8.2 Design of the pilot 116 115 Methodologies for social programme evaluation 10 Design of the ETU evaluation 123 10.1 Technical detail of evaluation methodology 11 Description of the data 119 125 127 12 Flows into and out of unemployment in the ETU pilot and comparison areas 135 12.1 Flows into and out of unemployment - urban areas 137 12.2 Flows into and out of unemployment - large towns 138 12.3 Flows into and out of unemployment - seaside areas 140 12.4 Flows into and out of unemployment - rural areas 141 12.5 Flows into and out of unemployment - high skilled and unskilled claimants 142 13 Statistical analysis of unemployment benefit claims 145 13.1 The impact of ETU on all unemployment benefit claimants 145 13.2 The impact of ETU on single unemployment benefit claimants 146 13.3 The variation in the impact of ETU on unskilled unemployment benefit claimants by age group and partnership status 149 13.4 The variation in the estimated impact of ETU on unskilled single male and female claimants by area type 152 14 Movements between unemployment benefit and Earnings Top-up 155 15 Conclusions 159 Appendix A - Unemployed Survey 1998 sampling information 161 Appendix B - Results from the logistic regression model 167 Appendix C - Results from the models in Chapter 5 169 References 175 Other research reports available 177 v LIST OF TABLES Table 1.1 Table 1.2 Table 1.3 Table 1.4 Table 1.5 Table 2.1 Table 2.2 Table 2.3 Table 2.4 Table 2.5 Table 2.6 Table 2.7 Table 2.8 Table 2.9 Table 2.10 Table 2.11 Table 2.12 Table 2.13 Table 2.14 Table 2.15 Table 3.1 Table 3.2 Table 3.3 Table 3.4 Table 3.5 Table 3.6 Table 3.7 Table 3.8 Table 3.9 vi ETU payments and thresholds - 1999/2000 13 ETU pilot area groups and types 14 Number and type of ETU awards by area (September 1998) 18 Average gross earnings, hours worked and ETU received (employees, September 1998) 19 Average gross earnings, hours worked and ETU received (employees, November 1999) 21 Gender of unemployed samples 24 Age of unemployed samples 24 Age of unemployed samples by gender 25 Age of unemployed samples by benefit type at sample 26 Marital status of unemployed samples by gender 27 Household type of unemployed samples by gender 28 Change in household type between first and second interview (1998 sample) 28 Housing tenure of unemployed samples by gender 29 Housing tenure of unemployed sample by age group in 1999 30 Highest academic qualification of unemployed samples by gender 30 Vocational qualifications of unemployed samples by gender 31 Qualifications of 1998 unemployed sample by age group 31 Health status of unemployed samples by gender 32 Health status of 1998 unemployed sample by age group 33 Health status of unemployed sample by benefit type 33 Which of these benefits are you (or your partner) receiving at this moment? 37 Type of benefit received by those claiming disability benefits 39 Which of these benefits are you (or your partner) receiving at this moment? 40 Average benefit received per week by household type 40 Average housing costs per week by tenure 41 Housing costs per week 42 Percentage of respondents with savings 43 Using credit to buy things 44 Which of the phrases best describes how you are managing financially these days? 45 Table 4.1 Table 4.2 Table 4.3 Table 4.4 Table 4.5 Table 4.6 Table 4.7 Table 4.8 Table 4.9 Table 4.10 Table 4.11 Table 4.12 Table 4.13 Table 4.14 Table 4.15 Table 4.16 Table 4.17 Table 4.18 Table 4.19 Table 4.20 Table 4.21 Table 4.22 Table 4.23 Table 4.24 Table 4.25 Table 4.26 Table 4.27 Economic status at time of first interview by gender 48 Economic status at time of first interview by age group 49 Economic status at time of first interview by gender and marital status 50 Economic status at time of first interview by benefit type at time of sampling 51 Economic status at time of first interview by ETU type 51 Economic status at both interviews by gender 52 Economic status at both interviews by age group 53 Economic status at second interview by gender and marital status 54 Economic status at time of interview by benefit type at time of sampling 55 Economic status at time of second interview by ETU type 56 When did you last have a paid job or work as self-employed? 57 Time last worked by age group – 1998 survey 57 Average proportion of last five years spent in each activity by gender 58 Occupational group of last job 58 Why did your most recent job end? 59 Work status at time of first interview 62 Work status at time of first interview by benefit type at sampling 62 Jobsearch status of non-working respondents by ETU area 61 Jobsearch status of non-working respondents by gender and marital status 62 Reasons for not looking for work by benefit type at sampling 63 Reasons for not looking for work by ETU type 64 Expected jobsearch among respondents not looking for work by benefit type at sampling 64 Expected jobsearch among respondents not looking for work by age group 65 Jobsearch methods in four weeks before interview 66 Jobsearch activity by ETU type 67 Median minimum wages for those looking for work at first interview 68 Median target and minimum wages by ETU area 69 vii Table 4.28 Average expected and minimum weekly wages for jobseekers in pilot areas by awareness of ETU 70 Table 4.29 What is the most likely thing to happen to you over the next couple of years? 71 Table 4.30 Most likely thing to happen over the next couple of years by ETU area type 72 Table 4.31 Most likely thing to happen over the next couple of years by benefit type at sampling 72 Table 4.32 Most likely thing to happen over the next couple of years by gender and partnership status 73 Table 4.33 Work status at time of second interview by benefit type at sampling 74 Table 4.34 Work status at first and second interview 75 Table 4.35 Average (median) wages and hours worked for respondents employed at 16 or more hours per week 76 Table 4.36 Percentage in work at second interview 77 Table 5.1 Employment status of the 1999 re-interviewees 83 Table 5.2 Changes in employment status 1998-1999 for the 1999 re-interviewees 84 Table 5.3 Expected hourly wage 93 Table 6.1 Awareness of ETU at first interview 96 Table 6.2 Where did you hear about the introduction of ETU? 100 Table 6.3 Awareness of ETU at second interview 101 Table 6.4 Percentage accepting a wage top-up by socio-demographic characteristics 106 Table 8.1 ETU pilot and comparison areas 117 Table 11.1 Age of unemployment benefit claimants at start of claim 128 Table 11.2 Destination of claimants leaving unemployment 130 Table 11.3 Definition of skill levels 131 Table 11.4 Distribution of skill among unemployment benefit claimants 131 Table 11.5 Distribution of unemployment benefit claimants across ETU pilot and comparison areas 133 Table 13.1 Flows into unemployment in ETU pilot areas relative to comparison areas 146 Table 13.2 Outflows as a share of inflows in ETU pilot areas relative to comparison areas 146 Table 13.3 Flows into unemployment in ETU A areas relative to ETU C areas by skill group - single males 147 Table 13.4 Flows into unemployment in ETU B areas relative to ETU C areas by skill group - single males 147 Table 13.5 Outflows as a share of inflows by ETU area type and skill group - single males 148 viii Table 13.6 Flows into unemployment in ETU A areas relative to ETU C areas by skill group - single females 148 Table 13.7 Flows into unemployment in ETU B areas relative to ETU C areas by skill group - single females 148 Table 13.8 Outflows as a share of inflows by ETU area type and skill group - single females 149 Table 13.9 Flows into unemployment for unskilled single males and females by age at start of claim 150 Table 13.10 Flows into unemployment for unskilled non-single males and females by age at start of claim 151 Table 13.11 Outflows as a share of inflows for unskilled single males and females by age at start of claim 151 Table 13.12 Outflows as a share of inflows for unskilled non-single males and females by age at start of claim 152 Table 13.13 Flows into unemployment for unskilled single males and females by area type 152 Table 13.14 Outflows as a share of inflows for unskilled single males and females by area type 153 Table 14.1 Total number of claims for ETU by gender 155 Table 14.2 Proportion of claims for unemployment benefit ending that involved a transition to ETU by gender and area type single claimants 157 Table A.1 Analysis of response rate - unemployed survey 1998/99 162 Table A.2 Characteristics of respondents in 1998 at each interview 164 Table A.3 Sample characteristics in 1998 at each interview 165 Table B.1 Logistic regression model: probability of being able to name ETU 167 Table C.1 Modelling exits from unemployment since introduction of ETU 169 Table C.2 Modelling exits from unemployment since 1998 interview 170 Table C.3 Modelling wages 171 Table C.4 Modelling whether job provides training 172 Table C.5 Modelling expected wages 173 LIST OF FIGURES Figure 1.1 Figure 1.2 Figure 1.3 Figure 1.4 Figure 1.5 ETU evaluation surveys 15 Number of ETU awards by claimant type 16 Number of ETU awards by Scheme type 17 Percentage of ETU awards by Scheme and client type (September 1998) 17 Total income for ETU claimants earning £3.60 per hour by Scheme and client type 20 ix Figure 4.1 Figure 4.2 Figure 4.3 Figure 5.1 Figure Figure Figure Figure 6.1 6.2 6.3 11.1 Figure 11.2 Figure 11.3 Figure 11.4 Figure 11.5 Figure 12.1 Figure 12.2 Figure 12.3 Figure 12.4 Figure 12.5 Figure 12.6 Figure 12.7 Figure 12.8 Figure 12.9 Figure 12.10 Figure 12.11 Figure 12.12 Figure 12.13 Figure 12.14 Figure 12.15 Figure 14.1 x Jobsearch status at interview by age group 62 Jobsearch status at interview by type of qualifications 63 Most likely situation over next couple of years by age group 73 Movements away from unemployment, by destination 87 Awareness of ETU by age group 97 Awareness of ETU by household type 98 Awareness of ETU by qualification type 99 The Earnings Top-up caseload from October 1996 to January 2000 by claimant type 127 Number of male unemployment benefit claimants according to age at start of claim and marital status 129 Number of female unemployment benefit claimants according to age at start of claim and marital status 129 Distribution of skill among unemployment benefit claimants by area type - claims starting before the start of the ETU pilot 132 Distribution of skill among unemployment benefit claimants by Scheme A, Scheme B and comparison areas – claims starting before the ETU pilot 132 Monthly inflows into unemployment in ETU Scheme A, Scheme B and comparison areas 135 Monthly outflows from unemployment in ETU Scheme A, Scheme B and comparison areas 136 Flows into unemployment - urban areas 137 Flows out of unemployment - urban areas 137 The ratio of inflows to outflows - urban areas 138 Flows into unemployment - large towns 138 Flows out of unemployment - large towns 139 The ratio of inflows to outflows - large towns 139 Flows into unemployment - seaside areas 140 Flows out of unemployment - seaside areas 140 The ratio of inflows to outflows - seaside areas 141 Flows into unemployment - rural areas 141 Flows out of unemployment - rural areas 142 The ratio of inflows to outflows – rural areas 142 Unemployment inflows and outflows for high skilled and unskilled claimants 143 Claims for unemployment benefit ending in a claim for ETU - single claimants 156 ACKNOWLEDGEMENTS Part One We would like to thank our colleagues at PSI involved in the evaluation of ETU, particularly Alan Marsh and Michael White for their helpful comments on this report. Thanks also go to Rebecca Stanley, Elaine Squires and Daphne White at the Social Research Branch of the Analytical Services Division of the Department of Social Security, for their support and patience. Part Two I am grateful to Nettie Roberts and Jane Edgeley at the Office for National Statistics for producing the unemployment (JUVOS) data series and to Nicola Croden and Rebecca Stanley at the Department of Social Security for having the stamina and courage to persist with what appeared to be an impossible task. I am grateful to a number of people who provided useful comments on earlier drafts. In particular, I would like to thank Elaine Squires and officials at the Department of Social Security and the Treasury and Michael White at the Policy Studies Institute. I have benefited from interesting discussions with my colleagues at the Policy Studies Institute, the Institute for Employment Research and the Centre for Research in Social Policy throughout the course of this evaluation. xi THE AUTHORS Part One Alison Smith is a Research Fellow at the Policy Studies Institute. Her research interests include social security policy, work incentives, unemployment, low-income families and living standards. Recent research includes the evaluation of Jobseeker’s Allowance and the 1999 Survey of Low-Income Families. Richard Dorsett is a Senior Research Fellow in the Employment Group at the Policy Studies Institute. He is interested in programme evaluation and the modelling of labour market transitions. Part Two xii Abigail McKnight is a Research Fellow at the Centre for Analysis of Social Exclusion at the London School of Economics. She is a Labour Economist and her research interests include labour market inequality, evaluation of welfare to work programmes, education, low pay and poverty. LIST OF ABBREVIATIONS ETU Earnings Top-up HB Housing Benefit IS Income Support JSA Jobseeker’s Allowance NMW National Minimum Wage UB Unemployment Benefit xiii PREFACE Earnings Top-up (ETU) was an in-work benefit available to low paid workers without children. ETU was piloted from October 1996 to October 1999 in eight areas across Britain. This volume is part of a set of seven final reports from the evaluation of the ETU pilot. (Baseline statistics were published in 1999, in DSS Research Report No. 95, and interim evaluation findings were published in March 2000, DSS Research Reports Nos. 112 and 113). The evaluation was conducted by researchers at the Policy Studies Institute (PSI), the Centre for Research in Social Policy (CRSP) at Loughborough University and the Institute for Employment Research (IER) at the University of Warwick. Outline details of the evaluation are provided in this report. Further information on the evaluation can be found in the six other final reports from the ETU evaluation: Earnings Top-up Evaluation: The Synthesis Report (Marsh, A., 2001, Department of Social Security Research Report No. 135). This report draws together the main results of the evaluation in one volume. The aim of this report is to provide a relatively short and non-technical overview of the evaluation’s conclusions drawn from all strands of the evaluation. It is intended that this will help readers identify the sources to which they can turn for fuller information on the evaluation. Earnings Top-up Evaluation: Employers’ reactions (Lissenburgh, S., Hasluck, C and Green A., 2001, Department of Social Security Research Report No. 132). This report is in two parts. The first presents findings from the surveys with employers carried out by PSI during the ETU pilot. It explores employer’s experiences of ETU focusing on wage effects and hours worked. The second part is econometric analysis, undertaken by IER, which considers the effects of ETU on employers’ behaviour and the recruitment process. Earnings Top-up Evaluation: Effects on Low Paid Workers (Marsh, A., Stephenson, A., Dorsett, R and Elias, P., 2001, Department of Social Security Research Report No. 134). This report is in two parts. The first section, by PSI, presents findings of the surveys conducted with low paid workers and ETU recipients throughout the pilot. It explores the characteristics of these workers and the effect that ETU had on their lives and examines the reasons for non take-up of ETU among eligible workers. The second part, by IER, analyses the same data to explore the wider labour market and potential long-term effects of ETU. Earnings Top-up Evaluation: Qualitative Evidence (Heaver, C. Roberts, S. Stafford, B. and Vincent, J. 2001, Department of Social Security Research Report No. 133). This report presents the findings of qualitative research conducted by CRSP as part of the evaluation of ETU. The report has three parts focusing on ex-recipients of ETU, self-employed recipients and unsuccessful ETU applicants. xv Earnings Top-up Evaluation: Staff Views (Vincent J., Heaver, C., Roberts, S. and Stafford, B., 2001, Department of Social Security In-house Research Report No. 74). This report presents the findings of the staff panels drawn from the eight pilot areas, and from central administrative/ processing staff from the Benefits Agency and Employment Service staff over the three years of the pilot. The report focuses on staff’s views of ETU and the way in which it operated within the pilot areas. It also considers changes over time from the beginning of the pilot to its end in 1999. Earnings Top-up Evaluation: Labour Market Conditions (Green, A. 2001, Department of Social Security In-house Research Report No. 75) This report, by IER, draws out the contrasts and similarities in labour market conditions across local areas included in the ETU pilot. Previously published research in the ETU series include: Low Paid Work in Britain (Marsh, A., Callender, C., Finlayson, L., Ford, R., Green, A and White, M., 1999, Department of Social Security Research Report No. 95). This report presents the findings from the first surveys conducted prior to the introduction of Earnings Top-up, with employers low paid workers and medium term unemployed people. Baseline data on the characteristics of these groups are presented, including health and education, wage expectations, earnings, wage setting behaviour and recruitment. Preliminary information on the labour market profiles of the evaluation areas is also included. The First Effects of Earnings Top-up (Finlayson, L., Ford, R., Marsh, A., Smith, A., and White, M., 1999, Department of Social Security Research Report No. 112). This report presents the findings from surveys conducted in 1997, almost one year after the introduction of ETU with employers, low paid workers, medium term unemployed people and ETU recipients. The report presents interim analysis of the first effects of ETU over this period. Piloting Change (Vincent, J., Abbott, D., Heaver, C., Maguire, S., Miles, A., Stafford, D., 1999, Department of Social Security Research Report No. 113). This report presents the interim findings from three components of the ETU qualitative research: two group discussions with Employment Service and Benefits Agency staff; face-to-face interviews with ETU recipients; and telephone interviews with employers. This report consists of two parts, both of which provide crucial evidence that informs the conclusions of the evaluation of ETU: Part One, by Alison Smith and Richard Dorsett contains the findings of surveys of unemployed people conducted by PSI throughout the ETU pilot. This outlines unemployed people’s experiences of ETU, benefit receipt, their income, contact with the labour market and labour market outcomes. Part Two, by Abigail McKnight contains findings from econometric analysis, carried out by IER, to assess the overall impact of ETU on unemployment and the effect on individual groups of unemployed people. xvi SUMMARY Part One - Surveys of Unemployed People Earnings Top-up was introduced in October 1996 in eight areas of the country for a three-year pilot period. It was an in-work benefit for people without dependent children. There were two different rates of benefit (Scheme A and Scheme B) and it was available to employed and self-employed people who worked for 16 or more hours per week in jobs lasting at least five weeks. It was paid at a fixed rate for a period of 26 weeks and the maximum amount of benefit payable was reduced by 70 pence for each pound of income above the threshold. The two main objectives of ETU were to improve the incentive for unemployed people to take low-paid work of 16 or more hours a week and to encourage low-paid workers to avoid unemployment by raising their incomes relative to out-of-work support. The programme of evaluative research was designed to compare eight test areas with four more areas chosen as ‘control’ areas at different points over the three-year period. The evaluation of ETU included field surveys of low-paid workers, unemployed people, ETU recipients and employers together with ongoing analysis of official administrative statistics, studies of local labour market conditions, and in-depth interviews with key participants. The focus here is on one part of the evaluation: the surveys of unemployed people. The initial sample was selected in spring 1996, and interviewed in summer 1996 and summer 1997. This report discusses the later sample selected in spring 1998 and interviewed in summer 1998 and summer 1999. It comprised people without dependent children who had been claiming either Jobseeker’s Allowance (73 per cent) or Income Support for between 26 and 65 weeks. Their experiences of ETU and particularly whether ETU played any role in helping them into work are explored. Comparisons are made between: • The interviews carried out with the second sample in 1998 and in 1999. • The 1998/9 survey and the earlier sample interviewed in 1996/7. By the time of the 1998 interview, ETU was well established in the pilot areas and claimants were typically young, employed, single people, working for around 30 hours a week but for very low wages (on average £2.90 per hour). Characteristics of the 1998 unemployed survey Gender and age As in 1996, most of the unemployed sample were men (69 per cent). However, the age distribution of the 1998 sample was significantly different from the 1996 survey. Whereas in 1996 32 per cent of the sample were aged under 25, by 1998 this proportion was 19 per cent. There was a 1 corresponding increase in the proportion of respondents aged 45 or over from 37 per cent in 1996 to 47 per cent in 1998 (Section 2.2). Marital status, household type and housing tenure Almost half of respondents were single (48 per cent), 23 per cent were widowed, separated or divorced and 29 per cent were married or cohabiting. In 1998, female respondents were far more likely to be living with a partner (40 per cent) than were men (27 per cent). The proportion of women with partners almost doubled from 22 per cent in 1996 to 40 per cent in 1998 (Section 2.3). Thirty six per cent of respondents lived alone in 1998, compared with 27 per cent in 1996 (Section 2.4). Just under a quarter (24 per cent) lived with their parents in 1998, down from 36 per cent in 1996. Thirty per cent of respondents were living in accommodation where their parents paid the housing costs (compared with 42 per cent in 1996). Another nine per cent were living in a property they owned outright while 10 per cent were buying a property with a mortgage. More respondents were renting accommodation in the 1998 survey – 44 per cent compared with 29 per cent in 1996. These were important changes as ETU was most attractive to out-of-work people with low housing costs (Section 2.5). Educational qualifications On average, respondents in the 1998 survey had fewer educational qualifications than did those in the 1996 survey (Section 2.6). Almost two-thirds of respondents had no academic qualifications (65 per cent compared with 55 per cent in 1996) and just eight per cent had qualifications at A level or above. Women tended to be less well qualified than men. Vocational qualifications were held by only 37 per cent of respondents and almost half of the sample (47 per cent) had no recognised qualifications at all. Health and caring responsibilities Forty three per cent of the sample reported a long-standing illness or disability, an increase of seven percentage points from the 1996 sample (Section 2.7). Sixteen per cent of respondents said they were in receipt of a disability benefit. More than half of women said they had a longstanding health problem (52 per cent compared with 36 per cent of men) and 21 per cent received a disability benefit. Ten per cent of respondents said they spent time caring for someone with a long-standing illness or disability. More than three-quarters of those sampled as receiving Income Support (77 per cent) reported a long-standing illness or disability and 52 per cent said they were receiving disability benefits. In contrast, 30 per cent of those sampled as receiving Jobseeker’s Allowance had health problems and just three per cent said they were receiving disability benefits. Area differences 2 The characteristics of respondents in the control areas should have matched the characteristics of respondents in ETU areas as closely as possible in order to assess the impact of ETU (Section 2.8). Overall, there were few differences; the most noticeable deviation is that fewer respondents lived alone in control areas (29 per cent compared with 37 per cent in Scheme A and 42 per cent in Scheme B areas). Linked to this was the higher proportion of respondents with partners in control areas (36 per cent compared with 29 per cent in Scheme A and 30 per cent in Scheme B areas). Income and benefits Receipt of benefits Few people reported receipt of ETU at either interview (see Chapter 6 for further details). At first interview, most people were receiving Jobseeker’s Allowance (49 per cent) or Income Support (31 per cent) (Section 3.2). More people in the 1998 survey were receiving Council Tax Benefit (35 per cent) and Housing Benefit (43 per cent) than in the 1996 survey (22 per cent and 27 per cent respectively). Almost all people in rented accommodation said they received Housing Benefit (91 per cent) and 43 per cent of people with a mortgage received help in the form of Income Support with their mortgage interest payments. Sixteen per cent of respondents were receiving a disability benefit and the most common type was Incapacity Benefit claimed by 57 per cent of these people. At first interview, the mean amount of benefits received per week was £62.24 (excluding Housing Benefit, Mortgage Interest premium and any disability benefits) (Section 3.3). Housing costs The average contribution made by respondents who lived in accommodation where their parents paid the housing costs was about £20 per week. Most tenants received Housing Benefit that met the full costs of their housing, an average of £42 per week (Section 3.4). Average mortgage payments were similar at around £45 per week and Income Support assistance averaged £27 per week. Material and financial well-being Almost half the sample (48 per cent) said there were things they needed to buy at present that they did not have the money for (Section 3.6). Nineteen per cent of respondents responsible for paying household bills said they were behind with their payments and almost eight out of ten respondents (78 per cent) said they had worried about money in the last few weeks. Fifty three per cent said they had trouble repaying debts over the previous two years. Seventeen per cent admitted to financial difficulties and 13 per cent thought they did not manage very well. Only 27 per cent of respondents had money saved and the median amount was only £100 (Section 3.5). Many relied on family and friends for support. They were more likely to borrow money from friends or relatives than more formal sources and more than a fifth of those living with parents felt they could not afford to move out even though they wanted too. Contact with the labour market There was an average gap of 70 days between the 1998 sample being selected and respondents being interviewed. During that time 11 per cent of the sample had moved into paid work: seven per cent were working for 16 or more hours per week, three per cent were working less than 16 hours per week and one per cent were self-employed. Six per cent were Economic status at first interview in 1998 3 undergoing some form of training or education and 63 per cent said they were unemployed and claiming benefit. One in ten people said they had been sick or disabled for more than six months at the time of first interview (Section 4.2.1). Economic status at second interview in 1999 Few people had moved into work in between first and second interviews (an average of 10 months) (Section 4.2.2). By 1999, still only 13 per cent were working 16 or more hours a week, three per cent were working less than 16 hours and two per cent were unemployed at second interview. Just under half said they were unemployed and claiming benefit (49 per cent) and 15 per cent said they had long-term ill health. Ten per cent were undertaking training or education. At both interviews, young people and those with educational qualifications were more likely to have moved into work (Section 4.5). By second interview, 26 per cent of respondents aged under 25 were working 16 or more hours a week as were 22 per cent of those with qualifications at A level or higher. Return-to-work wages were low at around £3.78 per hour for a 37 hour week. Although few people had moved into work, some of those who had were claiming ETU (13 per cent at first interview and 23 per cent at second). Previous work experience The 1998 sample of unemployed people did not have a great deal of recent work experience (Section 4.3). Forty four per cent of the sample said they had not worked in the previous five years and another 14 per cent said they had never had a paid job. On average, respondents had spent 15 per cent of the previous five years in full-time work and over half (54 per cent) of the time unemployed and claiming benefit. Only one in eight respondents who had worked in the previous five years had work experience in a professional, managerial or technical occupation and the majority worked in personal sales and services, in craft occupations, clerical work or as plant operatives. More than one-fifth of respondents said they had left their previous job because of ill health. Looking for work Of the non-working respondents at first interview, 38 per cent had not looked for work in the previous four weeks (Section 4.4). However, one-third of these said they would have liked a job if a suitable one were available. Sixty per cent of these said they were not currently looking for work because of health problems, 10 per cent had caring responsibilities, and 11 per cent were undertaking further training or education. The majority of people not looking for work were Income Support claimants when sampled as only 18 per cent of non-working respondents who had been in receipt of Jobseeker’s Allowance at sampling said they were not looking for work when interviewed. Male respondents (72 per cent), single people (70 per cent) and those with educational qualifications (76 per cent of people with both academic and vocational qualifications) were more likely to be looking for work. 4 In contrast, only 44 per cent of people living with a partner were looking for work as were 43 per cent of respondents aged 55 or over and 42 per cent of those who reported a long-standing illness or disability. The most commonly used jobsearch methods were looking at advertisements in newspapers (89 per cent) and in the Jobcentre (81 per cent). Aspirations for the future Most people looking for work at first interview were seeking employment (69 per cent) while 10 per cent particularly wanted self-employment and 21 per cent were prepared to consider either. One in ten people specifically wanted to work less than 16 hours per week. The minimum wages jobseekers were prepared to accept were £3.50 per hour on average and by second interview these had risen to £3.75 per hour (Section 4.4.4). Women were prepared to accept lower wages as were those living with parents. Above average wages were sought by people with partners, respondents aged 55 or over, those with degree level qualifications, and those with mortgages. Few people expected to be much better off if they got a job paying their minimum acceptable wage (14 per cent), and a similar proportion expected to be worse off (13 per cent). There was no difference between ETU Scheme areas and control areas overall, but respondents in some areas (for example, Sunderland at £3.16 per hour) gave lower minimum wages than those in other areas (such as Southend at £4.25 per hour). However, there was no strong indication that ETU had suppressed aspiration wages for unemployed people in the individual Scheme A and B areas compared with their control areas. Just over half of respondents (54 per cent) thought they would be working more than 16 hours per week over the next few years and 77 per cent of these believed they would no longer be claiming benefit. Almost one in three thought they would remain unemployed (29 per cent). Respondents sampled as receiving JSA were more optimistic about working in the future (65 per cent) than those sampled as receiving Income Support (22 per cent). Single people (61 per cent) and younger respondents (81 per cent of those aged under 25) were also more likely to think that they would be working over the next couple of years. Labour market outcomes Chapter 5 uses multivariate analysis to explore the effect of ETU on labour market outcomes. The main findings can be summarised as follows: • People living in an ETU pilot area were no more likely to enter work than those in the control areas (Section 5.2.2). • Those in the pilot areas who found work since their 1998 interview were more likely to earn lower wages than those in the control areas (Section 5.3). • Respondents with lower expected wages were more likely to find work, but expected wages were not influenced by ETU (Section 5.4). 5 Therefore, people with lower wage expectations were more likely to find work. As ETU seemed ineffective in reducing expected wages, it was unsurprising to find that ETU was similarly ineffective in helping the unemployed back to work. However, the concept of expected wages is, to some extent, notional and the level of these expected wages may get revised during the jobsearch process, particularly if people become aware of in-work benefits while looking for work. One the other hand, the reservation wage (or minimum wage that someone will accept) exists as a fixed constraint at the point of job offer. As there may not be an exact match between expected and reservation wages, the result that ETU had no effect on expected wages should not be interpreted as evidence that it had no effect on reservation wages, particularly in the context of low overall awareness of ETU (see Chapter 6). The lower levels of wages among those who had found work in the pilot areas might indicate some effect of ETU on the reservation wage. However, this effect, if it existed at all, was not translating into an increased rate of job entry. The experience of Earnings Top-up Awareness of ETU Almost two years after its introduction, awareness of ETU was low among this sample as only 29 per cent of respondents said that they had heard of the introduction of the benefit and just over half of these (16 per cent of all respondents) were able to name the benefit as ETU (Section 6.2). People aged 25 to 34 had the highest level of recalled knowledge: 36 per cent said they had heard of the benefit and 21 per cent could correctly name it. Respondents with a partner were less likely than single people to recall ETU: 24 per cent said they had heard of the benefit and just ten per cent could name ETU. Overall, there was little difference in awareness between Scheme A and Scheme B areas, but respondents in some individual areas had greater levels of awareness than others. Levels of awareness tended to follow the pattern of ETU awards, in that areas with the most ETU claimants (Sunderland, Newcastle, and Doncaster) also tended to be the areas where survey respondents had the best knowledge of the benefit. The lowest level of awareness was in Southend where 20 per cent of respondents said they had heard of the benefit and just seven per cent named ETU. Respondents most commonly had heard of the benefit from official sources. Twenty per cent of people said it had been recommended at the Jobcentre or Employment Service office, 32 per cent said it had been mentioned there and nine per cent said they had seen a publicity display in the benefit office. Eighteen per cent said they had heard of ETU from friends, relatives or neighbours. In 1999, awareness of ETU was slightly higher, though probably only as a result of the first interview in 1998. 6 Experiences of ETU Few people were claiming ETU at either interview (24 respondents at first interview and 30 at second). Similar numbers had previously received ETU (32 at first interview and 34 at second) and 23 had applied for ETU and been turned down. The majority of people who were receiving, or had previously received, ETU said that working and claiming ETU was a better life for them than not working and claiming Income Support or Jobseeker’s Allowance (Section 6.3). Attitudes towards an Earnings Top-up benefit Respondents were asked to imagine they were able to get earnings topped up with benefits while working 16 hours a week or more, and then asked whether they would consider a lower paid job than they wanted or hold out for a higher paid job. Overall, 59 per cent of respondents not in work said they would take a lower paid job and get an Earnings Top-up (Section 6.4). Women were more likely to find an Earnings Top-up acceptable (63 per cent) as were those aged 25 to 34 (66 per cent). People with a partner tended to be more reluctant to accept a top-up (53 per cent) as were those who said they had a long-standing illness or disability (54 per cent). People in Scheme A areas were slightly more likely to say they would accept a top-up (62 per cent) than those living in Scheme B areas (57 per cent) or control areas (58 per cent). People who were working at least 16 hours a week at interview were asked if they would apply for a similar benefit to ETU (that topped up their earnings) if one was available. Two-thirds of those currently working at least 16 hours per week (but not claiming ETU) said they would apply for such a benefit but 69 per cent of these said they would maintain their present working hours even if it meant they did not get a top-up. The majority of people who were working less than 16 hours per week said they would increase their working hours to get a top-up benefit. Views were unchanged a year later at second interview. Conclusions In the gap between selection and first interview, a minority of respondents had found work, most of it fairly low paid. A few of those in the pilot areas had also claimed ETU when entering work, but too few to be counted as an influence on people’s rate of return to work. Of the remainder, many respondents seemed likely to experience continued difficulties getting and keeping work. Typically they were poorly educated and often had only little recent previous work experience. Significant numbers reported health problems or caring responsibilities that restricted their participation in paid work. At first interview, the prospects for this sample seemed less encouraging even than those faced by a similar sample interviewed in 1996. A crucial distinction lay between those paying or not paying for their accommodation. For people paying rent, working and claiming ETU usually incurred a substantial loss of Housing Benefit leaving them with little additional income as workers. In the 1998 sample, 44 per cent of respondents were tenants and another ten per cent had a mortgage to 7 pay. This was much higher than in the corresponding 1996 survey and as a result, the minimum wages they said they would accept were higher, which tended to place more of them beyond the reach of ETU. A serious obstacle to ETU helping this unemployed sample back into work was the lack of awareness of the benefit itself. At first interview, recalled knowledge was disappointingly low, considering the sample was comprised of people with recent experience of claiming benefit for at least six months and living in areas where ETU had been available for around two years. Furthermore, not everyone liked the idea of a top-up to potential wages in work even when such a scenario was put to them (41 per cent). Qualitative research on ETU indicates that some people felt that there should be no need for a benefit top-up as employers ought to pay a ‘decent wage’ in the first place (Vincent et al, 1999). The introduction of the National Minimum Wage can be seen, in part, as a reflection of this view and indeed was enough to move many people out of eligibility for ETU. Clearly, if some type of wage supplementation were to be introduced for workers without dependent children, it would need to allow for higher wages than the ETU pilot did. Few respondents had moved into employment of 16 or more hours a week by the time of second interview (13 per cent) and ETU appeared to have had no significant influence on movements into work. However, it is important to remember that this sample was not representative of all unemployed people in the pilot areas at that time. It may be possible that ETU had a measurable effect on movements into work for people who had been unemployed for shorter periods of time. It is also possible that there were too few movements into work for us to be able to reliably capture any ETU effect unless it was particularly large. The aim of the evaluation was to compare the pilot areas with the control areas and to attribute any difference to ETU. But, for the size of the effect to be measurable there would need to be widespread knowledge of the benefit. The evidence from this survey is that awareness was poor and as advertising for ETU was stopped just six months after the benefit was introduced this is not surprising. ETU also appeared to have no influence on the wage expectations of those who remained unemployed. However, the concept of expected wages is, to some extent, notional and the level of expected wages may get revised during the jobsearch process, particularly if people become aware of in-work benefits while looking for work. In contrast, the reservation wage is the minimum wage that someone will accept at the point of job offer. In the context of low levels of awareness of ETU, the result that ETU had no effect on expected wages should not be interpreted as evidence that it did not, or could not, have lowered reservation wages. 8 Part Two - Econometric Analysis This part of the evaluation provides evidence on the impact of Earnings Top-up on claims for Unemployment Benefit. Earnings Top-up (ETU), an in-work benefit available to low paid single people and couples without children, has been piloted in a number of areas around Britain for a period of three years (October 1996 to October 1999). A pilot of this scale provides the scope to collect and analyse a substantial amount of data and to look in detail at the impact of an in-work benefit in the short and longer term. The focus of this part of the evaluation is to assess the overall impact of ETU on unemployment in the ETU pilot areas and to estimate the impact of ETU on individual groups of Unemployment Benefit claimants. The objective was to assess the wider labour market impact of ETU, not to assess the impact of ETU on an individual’s labour market experience. The basic question to be addressed in this part of the evaluation is whether ETU led to a decrease in flows into unemployment and whether or not the presence of ETU increased flows out of unemployment. An examination of the experience of different skill and demographic groups is made to assess the differential impact of ETU and to test for the presence of substitution effects. To answer these questions we turned to large scale administrative datasets covering all Unemployment Benefit claims from January 1995 to December 1998 and all claims for ETU in the pilot and control areas from the start of the pilot in October 1996 until December 1999. This provided us with an extremely rich source of information against which we could test a range of hypotheses. The great advantage of the administrative data over the survey data is that the coverage is complete and we do not have to worry about sample selection or response rate bias (whether or not the sample is representative of the population), attrition (loss of contact with individuals over time) or small sample size. The large number of observations available mean that we are able to compare outcomes in each of the eight pilot and four comparison areas. The disadvantage is that there is only limited information on the personal and household characteristics of individuals stored in these databases. Overall, this means that there are a limited number of questions we can address but we can be confident about the answers we obtain. An innovation in this part of the evaluation was the development of a skill classification which can be applied to unemployed job seekers via the information they provide on the occupations they usually work in or the occupations in which they are seeking work. This proved to be extremely useful in understanding flows into and out of unemployment, in identifying an ETU target population and in the assessment of potential substitution effects. 9 Key findings The impact of ETU on unemployment The results indicate that after the introduction of ETU, inflows to unemployment in the pilot areas fell relative to comparison areas. There is evidence that ETU led to an increase in net outflows from unemployment in the pilot areas. It is shown, with the aid of a skill classification of Unemployment Benefit claimants, that the impact of ETU on inflows and outflows was greater among an identifiable ETU target population – single, unskilled benefit claimants in young (16-24 years) and older (males 55+, females 45-54 years) age groups. There is evidence that the overall impact was lessened due to substitution effects. That is, some of the gains of the target population were at the expense of other groups (non-single claimants aged 35-44 with slightly higher levels of skill). ETU appears to have had a greater impact on single women than men and individuals living in rural areas where low pay is prevalent. The role of ETU in the transition between unemployment and work An analysis of transitions between Unemployment Benefit and claims for ETU corroborates the findings for all unemployed job seekers. A larger proportion of women than men completing spells of unemployment moved into a job supported by ETU. These transitions are greater in the ETU Scheme B areas (where the more generous version of ETU was being piloted) and for lower skilled benefit claimants. ETU claimants moving from unemployment go on to make a larger number of subsequent claims compared with all ETU claimants. The large number of subsequent claims may explain why the introduction of ETU has led to a reduction in flows into unemployment as well as flows out of unemployment. Overall the results suggest that an in-work benefit for single people and couples without dependent children can not only raise incomes of low paid workers but also reduce the harmful churning at the lower end of the labour market. Recent research (Gregg, 2000) has shown that the experience of unemployment early in an individual’s career (even after controlling for individual specific characteristics) is associated with poorer outcomes in later life. The benefits of ETU to the younger age group (16-24 years) suggest that any future employment tax credit could benefit this group in the short and longer term. 10 Department of Social Security Research Report NO 131 Earnings Top-up Evaluation: Effects on Unemployed People Part One • Surveys of Unemployed People Alison Smith and Richard Dorsett 1 1.1 Overview INTRODUCTION This chapter provides a brief outline of the Earnings Top-up (ETU) pilot evaluation and the part played by this survey of unemployed people. ETU was introduced in October 1996 in eight areas of the country for a three-year pilot period. It was an in-work benefit for people without dependent children. The two main objectives of ETU were to improve the incentive for unemployed people to take low-paid work of 16 or more hours a week and to help low-paid workers to avoid unemployment by raising their incomes relative to out-of-work support. There were two different rates of benefit (Scheme A and Scheme B) for each of three groups of clients: couples, single people aged 18 to 24, and single people aged 25 or over (Table 1.1). It was available to employed and self-employed people who worked 16 or more hours per week in jobs lasting at least five weeks. Like Family Credit, it was paid at a fixed rate for a period of 26 weeks and an additional amount was payable to those working 30 or more hours a week (£11.05 per week). The maximum amount of benefit payable was reduced by 70 pence for each pound of income above the threshold. It was not available to full-time students or people with savings of more than £8,000. Eligibility for those in couples was based on both persons’ income (excluding certain benefits), and only one member of a couple could claim. Table 1.1 ETU payments and thresholds – 1999/2000 Single 18-24 Single 25+ Couples Scheme A Maximum ETU payment £24.40 £30.00 £49.85 Earnings threshold £51.70 £62.45 £80.65 Scheme B Maximum ETU payment £24.40 £30.00 £60.15 Earnings threshold £80.65 £80.65 £80.65 Note: the maximum ETU payment is reduced by 70 pence for each pound of income above the earnings threshold. 1.2 The ETU evaluation The programme of evaluative research was designed to compare eight test areas with four more areas chosen as control areas of corresponding type at different points over the three year period (Table 1.2). The areas were selected because they had high levels of unemployment, a high number of job vacancies and a high proportion of low-paid vacancies and so were areas where ETU was expected to have the most impact. Four types of labour markets were also selected: major urban areas, large towns, seaside areas, and rural areas. 13 The two main target groups for ETU were existing low-paid workers and unemployed people. For the first group, ETU may have encouraged them to remain in work rather than returning to unemployment, whereas for unemployed people ETU could have allowed them to consider a job that paid less than they would normally have accepted. If people were more able to accept low-paid work then this could, in turn, have impacted on the decisions employers made about recruitment and wages. The evaluation of the effects of ETU therefore included corresponding field surveys of low-paid workers and unemployed people. Alongside these were similar field surveys of ETU recipients and telephone surveys of employers (Figure 1.1). The evaluation programme also included analysis of official administrative statistics, studies of local labour market conditions, and in-depth interviews with key participants. Table 1.2 ETU pilot area groups and types Area Scheme A Scheme B Control Area Major urban area Newcastle upon Tyne Sunderland Middlesbrough, Hartlepool and Stockton Large town Barnsley, Castleford, Pontefract, Doncaster Rotherham and Worksop Wakefield and Dewsbury Seaside area Southend Bournemouth Southampton and the Isle of Wight Rural area North Wales (Bangor and Perth and Crief, South Wales (Hay on Wye Caenarfon, Conwy and Colwyn, Denbigh, Dolgellau and Barmouth, Dumbarton, Stirling Brecon, Llanwrtyd Wells, Tredegar, Ebbw Vale, Holyhead, Porthmadog and Ffestiniog, Pwllheli, Shotton, Flint Pontypool, Monmouth, Abergavenny and and Rhyl, Wrexham) Cricklehowell, Cwmbran, Llanelli, Burry Port, Llandeilo and Llandovery) The focus in this report is on just one part of the evaluation: the surveys of unemployed people (Figure 1.1). The first sample were selected in spring 1996, and interviewed in autumn 1996 (before ETU) and summer 1997 (Marsh et al 1999, Finlayson et al 2000). This report analyses the second sample who were selected in spring 1998, interviewed for the first time in summer 1998 and for the second time in summer 1999. Therefore, ETU had been available for almost two years by the time the sample were first interviewed and was coming to an end by the time of the second interview. There were two main reasons for selecting a new sample of unemployed people. First, even if other things remained the same, the effect of ETU may have varied over time as the benefit became more established and 14 people became more aware of it. The expectation was that most of those claiming in 1996 to early 1997 would be workers who already had low paid jobs. It would be a while before the benefit began to speed longerterm unemployed people into work because this relied partly on a response by employers in finding new opportunities to people willing to work for less than they might otherwise have expected or needed. Also, substantial changes in the labour market and benefit system over the previous two years may well have resulted in compositional changes in the target group itself; this is discussed more fully in Chapter 2. As before, the 1998 sample comprised people without dependent children who had been claiming benefit for between 26 and 65 weeks. Throughout this report they are referred to as ‘the unemployed survey’ and the majority were claiming Jobseeker’s Allowance (JSA) (73 per cent) and therefore actively looking for work. Other people were in receipt of Income Support (IS) (usually because of health problems) and these respondents may not have been seeking work at this time (see Appendix A for sampling information). However, ETU could have helped or encouraged some of these people to get paid employment also, particularly if the extent of their illness or impairment did not qualify them for Disability Working Allowance. Figure 1.1 ETU evaluation surveys 1996 1997 Low-paid workers Re-interview lowpaid workers 1998 Postal/telephone follow-up ETU recipients Unemployed people Re-interview unemployed people Employers Employers 1.3 The progress of ETU 1999 New sample lowpaid workers New sample ETU recipients New sample unemployed people Re-interview unemployed people Employers It is worth briefly reviewing DSS administrative data on ETU to consider how take-up of the benefit progressed over the three-year period and to illustrate some basic information on the type of people who claimed ETU. 15 1.3.1 Trends in ETU claims Take-up of ETU over time In the first six months after the introduction of ETU in October 1996, the number of awards rose rapidly to reach 13,454 by April 1997 (Figure 1.2). Thereafter, the number of awards increased more gradually (to around 23,000 awards by June 1998) and levelled off somewhat after that. The majority of awards were to single people, half of whom were under 25 years old. The number of recipients who were part of a couple reached 2,296 by April 1997 and remained under 3,500 after. Area differences As shown in Table 1.1, the maximum amount of ETU paid to single people was the same under both Schemes but those in Scheme B areas had a higher earnings threshold (the same threshold as couples) and so could earn more and still be eligible for the benefit. The earnings threshold was the same for couples under both Schemes but a higher rate of ETU was paid to claimants in Scheme B areas. Therefore, it was expected that more people would claim ETU under Scheme B and the size of the areas was chosen to reflect the fact that Scheme B areas would have a higher proportion of eligible people. Despite this, there were more claims in Scheme B areas, with the gap widening in the first year after the introduction of ETU but remaining fairly constant after then (Figure 1.3). Figure 1.2 Number of ETU awards by claimant type 16 Figure 1.3 Number of ETU awards by Scheme type Client differences Despite the higher rate of benefit, fewer people with partners claimed ETU in Scheme B areas, both as a proportion of all awards in the area (Figure 1.4) and in absolute numbers (1,710 in Scheme A and 1,454 in Scheme B by September 1998). Almost half of awards in Scheme B areas (47 per cent) were to people aged 25 or under, and in absolute terms there were around twice as many awards to this age group in Scheme B areas (6,258 in September 1998) as in Scheme A areas (3,130 in September 1998). For single people aged over 25, the number of claims was around 10 to 15 per cent higher in Scheme B areas (5,515 in September 1998) than in Scheme A areas (4,969 in September 1998). However, awards to this client group were a smaller proportion of all cases in Scheme B areas (42 per cent) than in Scheme A (51 per cent). Figure 1.4 Percentage of ETU awards by Scheme and client type (September 1998) 17 Relatively few claimants were self-employed so the majority of ETU awards were made to employees (91 per cent in September 1998). Of all awards in September 1998, 54 per cent were renewals of existing claims while 38 per cent were new awards (Table 1.3). The area with the largest number of ETU awards was Sunderland with 29 per cent of all awards and the highest rate of renewals (58 per cent). Except for rural areas, there were more awards in Scheme B areas than in the equivalent Scheme A areas. Generally, there were proportionately more new awards in Scheme A areas than under Scheme B. Overall, the volume of new claims fell over time from 56 per cent of all awards in October 1997 to 32 per cent by April 1999. Table 1.3 Number and type of ETU awards by area (September 1998) Row percentages Area Scheme type No of awards New awards Renewals Subsequent Major Urban Area Newcastle upon Tyne A 3,603 39 52 9 Sunderland B 6,443 33 58 9 Large Town Barnsley, Castleford, Pontefract, Wakefield and Dewsbury A 2,401 39 54 7 Doncaster B 3,596 35 55 10 Seaside Area Southend A 797 41 55 4 Bournemouth B 1,457 39 51 10 Rural Area North Wales A 3,008 46 45 9 Perth and Crief, Dumbarton, Stirling B 1,731 40 51 9 All Scheme A All Scheme B A B 9,809 13,227 42 35 51 56 7 9 Total ETU awards All 23,036 38 54 8 Notes: new awards are made to people who have never received ETU before. Renewal awards were those following on immediately from a previous award without a break. Subsequent awards were those where there had been a previous award but a break between claims. Source: Earnings Top-up Statistical Enquiry, 30 September 1998, ASD, DSS. Levels of award Forty per cent of employees claiming ETU received the maximum award (Table 1.4). Recipients with a partner were much less likely to receive the maximum amount (24 per cent) than single claimants (42 per cent) and claimants under Scheme B were more likely to receive the maximum award (43 per cent) than were those under Scheme A (35 per cent). On average, people claiming ETU worked 29 hours per week and typically for very low weekly wages: £85.17 a week in September 1998, an average of just £2.90 per hour. The self-employed reported longer hours and lower wages - on average £43.85 for a 36 hour week in September 1998, just £1.23 per hour. 18 Table 1.4 Average gross earnings, hours worked and ETU received (employees, September 1998) Average Employees Percentage Average Average gross hours worked receiving ETU award earnings per week maximum ETU All £25.55 £85.17 29 40 Couples Single < 25 £37.46 £23.20 £107.09 £80.15 30 32 24 42 Single 25+ £24.43 £83.82 27 42 Scheme A Couples £24.84 £35.55 £74.60 £102.64 28 30 35 27 Single < 25 Single 25+ £21.59 £23.59 £62.52 £73.84 31 25 31 40 Scheme B £26.03 £92.48 31 43 Couples Single < 25 £39.58 £23.98 £112.00 £88.67 31 33 22 47 Single 25+ £25.14 £92.25 28 44 Source: Earnings Top-up Statistical Enquiry, 30 September 1998, ASD, DSS. 1.3.2 Introduction of the National Minimum Wage The National Minimum Wage (NMW) was introduced in April 1999, six months before the end of the ETU pilot. At this time, the NMW was £3.60 per hour for employees aged 22 or over and £3.00 per hour for younger workers, considerably higher than average wages for ETU claimants. Therefore, it was expected that this would change the typical pattern of ETU awards. For this sample, it may also have influenced the ability of unemployed people to return to work. Other things being equal, a rise in earnings would have resulted in a smaller award of ETU for most claimants. In fact, for some people, a wage rate of £3.60 per hour would have left them ineligible for ETU, depending on the number of hours worked. Therefore, there may have been an incentive (particularly for current ETU claimants who saw their wages rise under the NMW) to reduce working hours to get a higher ETU award. This is illustrated graphically in Figure 1.5. For single ETU claimants in Scheme A, the flat slope of the income line at below 25 hours for those aged under 25 and below 29 hours per week for older claimants shows that working hours could be reduced towards 16 with little change in net income. In fact, those aged under 25 in Scheme A became ineligible for ETU if they worked 25 or more hours a week at £3.60 per hour. For single people in Scheme B areas, the flattest slopes of the curves were between 23 and 29 hours where working extra hours contributed little to total income. There is a jump in the income curves at 30 hours because of the inclusion of the 30 hour premium but the lines remain fairly flat after 30 hours per week. Couples tended to do better from 19 ETU as the level of the benefit was more generous than for single claimants and so at £3.60 per hour they were still eligible for some ETU even when working more than 40 hours a week. Figure 1.5 Total income for ETU claimants earning £3.60 per hour by Scheme and client type Overall, there was little evidence before the end of the pilot that the NMW had affected ETU to any great extent. Figures 1.2 and 1.3 illustrated the slight fall in the number of ETU awards in the last few months of the pilot (particularly in claimants aged under 25). Some of this may have been related to the effect of the NMW but there could also have been an anticipatory affect as people might not have claimed because they knew the benefit was being withdrawn. However, it is difficult to assess trends in numbers of claims as the administrative data is for the whole caseload and so some of the cases in payment for each month would have referred to claims that commenced up to six months earlier. The time between the introduction of the NMW and the end of ETU is probably too small to accurately assess the impact of the NMW on ETU. As new claims were no longer being accepted, the caseload of ETU claims had diminished by November 1999 but those claiming would have commenced their claim after the introduction of the NMW in April 1999. The information available suggests that the average awards, earnings and hours worked in November 1999 were similar to those in September 1998 although average earnings had increased slightly and average working hours had decreased slightly (Table 1.5). This, in itself, cannot be taken as evidence that people were reducing their hours to maintain the level of an ETU award, as it may simply have been that those working longer hours had stopped claiming ETU leaving those in the caseload who were working for fewer hours. Average hourly earnings in November 1999 were around NMW levels – for single people aged 25 or over claiming ETU they were £3.65 per 20 hour. However, average wages in Scheme A areas were below NMW levels (£2.38 per hour for single people aged under 25 and £3.49 per hour for those aged 25+). There are some cases where the NMW is not paid (those aged 22 or over undertaking accredited training in the first six months of a new job would receive £3.20 per hour) but it is concerning that so many employees appeared to be receiving wages below NMW level. Table 1.5 Average gross earnings, hours worked and ETU received (employees, November 1999) Average Percentage Average Average gross hours worked receiving ETU award earnings per week maximum ETU All Couples £23.75 £36.23 £89.86 £110.59 26 28 36 25 Single < 25 Single 25+ £20.46 £22.71 £85.91 £87.02 29 24 35 39 Scheme A £24.29 £77.46 24 33 Couples Single < 25 £34.59 £20.63 £106.43 £62.99 28 26 29 26 Single 25+ £22.70 £74.99 21 37 Scheme B Couples £23.39 £37.99 £98.04 £115.05 28 29 38 20 Single < 25 Single 25+ £20.40 £22.71 £95.05 £96.77 30 26 39 41 Employees Source: Earnings Top-up Statistical Enquiry, 30 November 1999, ASD, DSS. 1.4 Summary The remainder of this report describes a sample of unemployed people first interviewed in summer 1998 and re-interviewed in summer 1999. Their experiences of ETU and particularly whether ETU played any role in helping them into work are explored. This design is similar to the survey of unemployed people carried out in 1996 and re-interviewed in 1997 to test the effects of the introduction of ETU. It is useful to compare the two first-wave surveys from 1996 and 1998 to see if ETU addressed the same potential customers two years after its introduction as it did at first. By the time of the first survey interview in 1998, ETU was well established in the pilot areas and claimants were typically employed, single people (many aged under 25), working for around 30 hours a week but for very low wages, on average. 21 2 2.1 Introduction THE CHARACTERISTICS OF THE UNEMPLOYED SAMPLE This chapter describes the characteristics of the sample of unemployed people in 1998 and compares them to the earlier survey in 1996. Where appropriate, it also considers how the characteristics of the 1998 sample had changed by the time of the second interview, around a year later. At the time of sampling, the respondents had been claiming benefit for between 26 and 65 weeks, had no dependent children living with them, and resided in one of the eight ETU or four control areas. Response rates for both surveys for the 1998 sample were satisfactory (in excess of 70 per cent) and there was no obvious response bias introduced by the loss of some respondents between first and second interviews. Appendix A contains more detailed information on the sample, response rates, and response bias. Living in areas characterised by high unemployment and low entry-level wages, these medium term unemployed people were selected as those likely to benefit from assistance into work. ETU had been available for almost two years at the time this second sample of unemployed people were first interviewed in autumn 1998. However, having already claimed benefit for at least six months, they were likely to experience difficulties returning to the labour market, and the extent of their disadvantage needs to be considered when assessing the impact of ETU. There have also been considerable changes in the labour market and benefit system since the first sample of unemployed people were selected in April 1996; three particularly that are likely to affect the characteristics of this later sample of unemployed people. First, is the national fall in claimant count unemployment over the two years, from 2,101,300 in March 1996 (7.7 per cent of the labour force) to 1,376,000 in March 1998 (4.9 per cent of the labour force). Second, Jobseeker’s Allowance (JSA) was introduced in October 1996 and replaced Unemployment Benefit (UB) and Income Support (IS) for unemployed people. This new system, with its reduction in the length of contribution-based benefit and more stringent emphasis on actively seeking work, is in itself credited with reducing claimant unemployment by between 100,000 and 200,000 nationally (Sweeney and McMahon, 1998). Lastly, another programme that may have affected the stock of unemployed people is the introduction of New Deal for Young People in April 1998 that gave training and work experience to 18 – 24 year olds who had usually been unemployed for at least six months. Although this was introduced at about the same time as this survey was sampled, there may have been an anticipatory effect. Also, one of the ETU areas, Newcastle, was a Pathfinder area for New Deal and so the programme would have been in effect there from January 1998. Because of these changes, the characteristics of respondents 23 in this most recent survey are compared with those of the 1996 survey to see how the target group for ETU may have changed over time. Compositional changes in the two surveys could also result from ETU having moved certain groups of people out of unemployment. If this was the case though substantial differences between ETU areas and control areas would be expected and, as is shown in Section 2.8, these have not been found. As in the 1996 survey, some people had moved off benefit by the time of interview and this is discussed further in later chapters. However, to avoid confusion, all sample members are referred to as the ‘unemployed survey’ and references to benefit type relate to their status at time of sample selection. 2.2 Gender, age and ethnicity The majority of the sample were men (69 per cent) and this was largely unchanged from the situation in 1996 (Table 2.1). More IS claimants were female (43 per cent) than were those receiving JSA (27 per cent). People from ethnic minority groups made up a very small percentage of the sample (just one per cent) and this was similar to the 1996 sample. Table 2.1 Gender of unemployed samples Column percentages Unemployed survey 1996 Male Female Base: all respondents Unemployed survey 1998 70 30 69 31 1991 2187 The age distribution of the 1998 sample was significantly different from the 1996 survey (Table 2.2). Whereas in 1996 32 per cent of the sample were aged under 25, by 1998 this proportion was 19 per cent. There was a corresponding increase in the proportion of respondents aged 45 or over from 37 per cent in 1996 to 47 per cent in 1998. By second interview, the 1998 sample had aged another year reducing the proportion of those aged under 25 from 19 per cent to 16 per cent. Table 2.2 Age of unemployed samples Column percentages Age group Unemployed Unemployed Unemployed survey 1996 survey 1998 survey 1999 18 - 24 25 - 34 32 18 19 17 16 18 35 - 44 45 - 54 12 20 16 26 16 26 55 - 64 17 21 24 1991 2187 1309 Base: all respondents Note: columns may not sum to 100 because of rounding. 24 The reduction in the numbers of young people occurred equally for both men and women (Table 2.3). For men, there was an increase in people aged 35 or over whereas for women the increase was in respondents aged 45 or over. Table 2.3 Age of unemployed samples by gender Column percentages Age group Unemployed survey 1996 Unemployed survey 1998 Men Women Men Women 18 - 24 25 - 34 32 20 32 13 19 21 18 11 35 - 44 45 - 54 12 19 13 23 18 23 14 34 55 - 64 17 18 19 23 1398 593 1499 688 Base: all respondents Note: columns may not sum to 100 because of rounding. To some extent, this change in the age distribution of the surveys reflects the reduction in national claimant unemployment in young people between 1996 and 1998, particularly among those who were unemployed for more than six months. In April 1996, people aged 18 to 24 formed 30 per cent of all those unemployed for between 26 and 52 weeks, similar to our 1996 survey. By April 1998 this proportion had fallen to 27 per cent. These figures were for all unemployed people. As this sample was based on unemployed people without dependent children it contained proportionally more of those aged under 25 and over 45 as these age groups were more commonly those without children of dependent age. Therefore, this sample was particularly likely to be affected by changes in the younger age groups. Newcastle, being a Pathfinder area for New Deal for Young People, did not seem to have affected the proportion of unemployed people in the survey aged 18 - 24. In fact, 25 per cent of respondents in Newcastle were aged under 25, slightly higher than the average of 19 per cent. The 1996 sample was drawn from benefit records for UB/IS whereas the 1998 sample was drawn from JSA/IS records. It is not known whether the proportion of IS claimants in the 1996 and 1998 surveys was the same as it has not been possible to identify benefit type for the 1996 sample. Certainly, the IS claimants in the 1998 survey tended to be older and if the proportion of IS claimants in the 1998 survey was higher than in the 1996 survey this could have explained the change in age distribution. Excluding the IS claimants does increase the proportion aged under 25 in the 1998 sample but not by any great extent (Table 2.4). It is, therefore, unlikely that this result is due to any sampling differences. 25 Table 2.4 Age of unemployed samples by benefit type at sample Column percentages Age group JSA IS All 18 - 24 22 11 19 25 - 34 35 - 44 20 18 12 11 17 16 45 - 54 55 - 64 26 14 29 37 26 21 1576 603 2187 Base: all respondents Note: columns may not sum to 100 because of rounding. In summary, it has not been possible to completely explain the change in the age distribution for the 1998 survey. In part, it can be attributed to the reduction in claimant unemployment among people aged under 25 and there were no technical reasons related to sampling that should have caused this result. However, it is important to remember that the ETU surveys have been based on specific areas of the country and so may not reflect the national picture. Therefore, it may be that there has been a larger than average drop in claimant unemployment among young people in the ETU and control areas that could account for this change. It is an important finding that the age distribution of one of the main target groups for ETU changed between 1996 and 1998, but more importantly, for this evaluation, there was no statistically significant difference in age between respondents in the ETU areas and the control areas. 2.3 Marital status Respondents were questioned both on their legal marital status and present ‘partnership’ status. Almost half were single, just under a quarter were widowed, separated or divorced and 29 per cent were married or cohabiting (Table 2.5). However, 31 per cent were currently living with a partner (regardless of their legal status), three per cent higher than in 1996, which reflects the older age range of the 1998 sample. IS claimants were more likely to live with a partner (43 per cent) than were those receiving JSA (27 per cent). In 1998, female respondents were far more likely to be living with a partner (40 per cent) than were men (27 per cent). They were also more likely to be widowed, separated or divorced (29 per cent compared with 20 per cent). For women, accompanying the change in the age distribution since 1996 was a corresponding increase in the proportion of women with partners, which almost doubled from 22 per cent in 1996 to 40 per cent in 1998. For men, there were actually slightly fewer with partners in 1998 than in 1996 and more who were divorced (15 per cent compared with eight per cent in 1996). 26 A year on, there had been little change in partnership status among the 1998 sample. Just eight per cent of those without a partner in 1998 had acquired one by 1999. Nine per cent of those with a partner in 1998 were single again by 1999. Table 2.5 Marital status of unemployed samples by gender Cell percentages Unemployed survey 1996 Marital status Unemployed survey 1998 Men Women All Men Women All Married Cohabiting 26 4 18 4 24 4 22 3 32 6 25 4 Separated from marriage 3 5 3 3 4 3 Separated from cohabiting 1 * 1 1 * 1 Divorced Widowed 8 1 18 7 11 3 15 1 21 4 17 2 Single 57 48 54 55 33 48 Has partner in household 30 22 28 27 40 31 Base: all respondents 593 1991 1499 688 2187 1398 *: less than 0.5 per cent. 2.4 Household type Related to the differences in age and marital status discussed so far, are changes in the type of household respondents lived in (Table 2.6). Thirty six per cent of respondents lived alone in 1998, compared with 27 per cent in 1996. Just under a quarter lived with their parents in 1998, down from 36 per cent in 1996. Around one-fifth of respondents lived with a partner only as almost a third of those living with a partner (32 per cent) lived within a larger household (26 per cent in 1996). IS claimants were more likely to be living with a partner only (32 per cent) and less likely to be living with parents (12 per cent) than JSA recipients (17 per cent and 28 per cent respectively). Just over half (52 per cent) of single respondents living with their parents resided in a household where at least one other person was working (compared with 61 per cent in 1996). Half of other households also contained a worker (57 per cent in 1996). Just 17 per cent of partnered respondents had a working partner (36 per cent in 1996). Few female respondents lived with their parents (13 per cent compared with 29 per cent of men), 35 per cent lived alone and three in ten lived with a partner only (compared with 17 per cent of men). More men were living alone in 1998 (36 per cent) than in 1996 (22 per cent) and fewer were living with their parents (down ten per cent to 29 per cent). 27 Table 2.6 Household type of unemployed samples by gender Column percentages Unemployed survey 1996 Marital status Unemployed survey 1998 Men Women All Men Women All 22 37 27 36 35 36 Lives with partner only 21 Lives with parent/s, 18 20 17 30 21 no partner Other 39 17 29 16 36 17 29 18 13 22 24 19 1398 593 1991 1499 688 2187 Lives alone Base: all respondents Although few people had changed partnership status between 1998 and 1999, more had changed the type of household they lived in (Table 2.7). Ten per cent of those who had been living alone in 1998 were living in a larger household and six per cent of those who had been living with parents in 1998 were no longer doing so. Seven per cent of respondents in 1999 had a child living with them but most of these were of nondependent age. Just one per cent of respondents had a child born between first and second interview. Table 2.7 Change in household type between first and second interview (1998 sample) Column percentages Household type 1998 Lives with Household Lives Lives with parents, no type 1999 alone partner only partner Other 1999 4 85 2 2 8 11 31 23 90 Lives alone Lives with partner only 2 All in Lives with parents, no partner 1 4 Other 7 7 94 2 13 68 18 409 297 334 269 1309 Base: all respondents 28 Note: figures in bold indicate the percentage of each group in 1998 whose situation was unchanged in 1999. 2.5 Housing tenure Three in ten respondents were living in accommodation where their parents paid the housing costs (compared with 42 per cent in 1996) (Table 2.8). Most of these people (70 per cent) made a contribution to the expenses. Another nine per cent were living in a property they owned outright while ten per cent were buying a property with a mortgage. Overall, more respondents were renting accommodation in the 1998 survey – 44 per cent compared with 29 per cent in 1996, and the 28 proportion receiving Housing Benefit (HB) equal to their total rent doubled from 13 per cent in 1996 to 26 per cent in 1998. IS claimants particularly were more likely to be living in rented accommodation (53 per cent compared with 41 per cent of JSA recipients) and over half of female respondents were tenants (53 per cent compared with 37 per cent in 1996). These were important changes as ETU was likely to be most attractive to out-of-work people with low housing costs. At higher rents, receiving ETU did little more than replace entitlement to HB that would otherwise continue in work. Table 2.8 Housing tenure of unemployed samples by gender Column percentages Unemployed survey 1996 Men Unemployed survey 1998 Women All Men Women All Parents pay housing costs 12 Parents pay housing costs – 11 12 10 6 9 makes contribution Owns outright 33 10 22 10 30 10 25 8 13 10 21 9 Owns with mortgage Rents – 100% HB 12 11 14 17 12 13 8 24 12 32 10 26 Rents – no or partial HB 15 20 16 17 21 18 Other, missing data 7 7 7 8 6 7 1398 593 1991 1499 688 2187 Base: all respondents Despite the changes in household type, there were few overall changes to housing tenure in the year or so between first and second interview for the 1999 sample (Table 2.9). One in twelve of those living with parents in 1998 were tenants (seven per cent) or mortgagees (one per cent) by 1999. Most of those aged under 25 were living with parents (80 per cent). The proportion of tenants rose from 15 per cent among those aged under 25 to the majority of those aged 35 or over. The relationship between age and tenure was unchanged from 1998. 29 Table 2.9 Housing tenure of unemployed sample by age group in 1999 Column percentages Age group in 1999 Under 25 25-34 35-44 45-54 55+ All Parents pay housing costs 19 Parents pay housing costs – 8 9 4 3 8 makes contribution Owns outright 61 * 41 1 23 2 9 11 1 22 23 9 Owns with mortgage Rents – 100% HB * 7 4 21 12 23 12 34 18 29 11 24 Rents – no or partial HB 8 19 22 22 19 19 Other, missing data 5 5 9 8 9 8 209 228 212 343 316 1309 Base: all respondents 2.6 Education and training On average, respondents in the 1998 survey had fewer educational qualifications than did those in the 1996 survey. Almost two-thirds of respondents had no academic qualifications (65 per cent compared with 55 per cent in 1996) and just eight per cent had qualifications at A level or above (Table 2.10). Women tended to be less well qualified: seven out of ten had no academic qualifications at all compared with 62 per cent of men. The position of female respondents had worsened on average since 1996 when 52 per cent of women had no academic qualifications. Table 2.10 Highest academic qualification of unemployed samples by gender Column percentages Unemployed survey 1996 Unemployed survey 1998 Men Women All Men Women All Degree A-Level 5 6 7 8 6 7 4 5 3 3 4 4 O-Level/equiv. CSE-level/equiv. 17 15 20 14 18 15 15 14 15 9 15 13 None 57 52 55 62 70 65 1398 593 1991 1499 688 2187 Base: all respondents Vocational qualifications were held by only 37 per cent of respondents (Table 2.11). Almost half of the sample (47 per cent) had no recognised qualifications at all (62 per cent of IS claimants and 42 per cent of JSA recipients). Again, women were more disadvantaged: 68 per cent had no vocational qualifications and 56 per cent had neither academic nor vocational qualifications (compared with 42 per cent of women in 1996). There were no substantial changes in educational levels between 1998 and 1999. 30 Table 2.11 Vocational qualifications of unemployed samples by gender Cell percentages Unemployed survey 1996 Men Unemployed survey 1998 Women All Men Women All Apprenticeship 9 1 7 8 1 6 RSA or similar City and Guilds 5 21 17 13 8 19 5 20 12 10 8 17 ONC or OND HNC or HND 4 3 1 1 3 2 3 3 1 1 2 2 TEC or BTEC Other, including HGV 5 4 5 2 5 3 3 3 2 1 3 14 Professional/nursing None 5 61 7 65 6 62 3 61 5 68 4 63 vocational qualifications 41 42 41 43 56 47 Base: all respondents 593 1991 1499 688 2187 Neither academic or 1398 Note: an individual may have more than one type of qualification and so be counted several times. The lower educational attainment of the 1998 sample was probably related to the change in the age distribution between the surveys. The 1996 survey found that younger people tended to have more educational qualifications than older respondents. In 1998, only 27 per cent of young people aged under 25 had no qualifications at all compared with 63 per cent of respondents aged 55 or over (Table 2.12). In contrast, three out of ten young people had both academic and vocational qualifications compared with only one in ten of the over 55s. Table 2.12 Qualifications of 1998 unemployed sample by age group Column percentages 18-24 25-34 35-44 45-54 55+ All 27 33 45 60 63 47 qualifications only 14 Academic qualifications 13 15 22 23 18 only Academic and 35 27 18 6 4 16 vocational qualifications 30 27 21 12 10 19 Base: all respondents 385 360 575 449 2179 No qualifications Vocational 410 One fifth of respondents had started a course or period of training but said they had not completed it. A similar proportion (22 per cent) of respondents had attended a government training/Employment Service programme in the year before first interview and 36 per cent had been 31 on another education or training course to develop job-related skills. Unemployed people were generally keen to develop their skills, though 36 per cent thought they would not attend a training course in the future. One in ten respondents admitted to having problems with basic reading, 16 per cent to problems with writing and spelling and nine per cent to problems with numbers or simple arithmetic. Eighty per cent of people had no problems with any of these basic skills but 10 per cent had problems with one, six per cent had problems with two and four per cent had problems with all three. Forty two per cent of the sample had a driving licence and 34 per cent said they currently had access to a car, van or motorcycle for their personal use. This was substantially less than for the population as a whole. By 1997, 70 per cent of British households had access to a car (Social Trends 1998). 2.7 Health Less than half of the sample (45 per cent) defined their health as good (seven percentage points lower than the 1996 survey). Forty three per cent said they had a long-standing illness or disability; an increase of seven percentage points from the 1996 sample (Table 2.13) and substantially higher than in the population as a whole (30 per cent of all adults aged 16 – 64, General Household Survey 1995-6). Sixteen per cent of respondents said they were in receipt of a disability benefit. More than half of women said they had a long-standing health problem (52 per cent compared with 36 per cent of men) and 21 per cent received a disability benefit. Table 2.13 Health status of unemployed samples by gender Cell percentages Unemployed survey 1996 Men Unemployed survey 1998 Women All Men Women All Health over last 12 months Good 55 46 52 49 34 45 Fairly good Not good 24 21 24 31 24 24 25 26 26 40 25 30 illness or disability 33 42 36 38 52 43 Receipt of disability benefit 13 18 14 13 21 16 1398 593 1991 1499 688 2187 Has long-standing Base: all respondents Ill health and disability was more common among older respondents (Table 2.14) and this may be the reason for the apparent increase among the 1998 sample compared to the 1996 sample. 32 Table 2.14 Health status of 1998 unemployed sample by age group Column percentages 18-24 25-34 35-44 45-54 55+ All Health over last 12 months Good Fairly good 67 21 57 24 46 26 32 27 28 28 45 25 Not good 12 19 28 41 44 30 Has long-standing illness or disability 25 28 43 53 57 43 8 9 14 20 25 16 410 385 360 575 449 2179 Receipt of disability benefit Base: all respondents With regard to health, the largest differences were apparent between respondents sampled receiving JSA and those claiming IS (Table 2.15). More than three-quarters of IS claimants reported a long-standing illness or disability and 52 per cent said they were receiving disability benefits. In contrast, three in ten JSA recipients had health problems, much in line with the national picture, and just three per cent said they were receiving disability benefits. Table 2.15 Health status of unemployed sample by benefit type Cell percentages Unemployed survey 1998 JSA IS All Good Fairly good 56 28 15 18 44 25 Not good Has long-standing 16 67 30 illness or disability Receipt of disability benefit 30 3 77 52 43 16 1584 599 2187 Health over last 12 months Base: all respondents Ten per cent of respondents said they spent time caring for someone with a long-standing illness or disability, which included one fifth of people living with a partner only and nine per cent of those living with parents. Over half of these people (52 per cent) said that caring responsibilities restricted their participation in paid work. Sixty three per cent of the people being cared for received Attendance Allowance or Disability Living Allowance and 30 per cent of the carers said they received Invalid Care Allowance for looking after this person. 33 The proportions experiencing health problems did not change in the year or so between first and second interview for the 1998 sample. Most people who said they had a long-standing illness or disability in 1998 also gave the same answer in 1999 (78 per cent). Likewise, 84 per cent of those without a long-standing illness or disability in 1998 were in the same position in 1999. 2.8 Area differences It was important that the characteristics of respondents in the control areas matched the characteristics of respondents in ETU areas as closely as possible in order to assess the impact of ETU. Overall, there were few differences in the characteristics of respondents according to the area they lived in. The most noticeable deviation is that fewer respondents lived alone in control areas (29 per cent compared with 37 per cent in Scheme A and 42 per cent in Scheme B areas). Linked to this is the higher proportion of respondents with partners in control areas (36 per cent compared with 29 per cent in Scheme A and 30 per cent in Scheme B areas). Other slight variations are summarised below: • There were more female respondents in control areas – 34 per cent compared with 31 per cent in Scheme A areas and 29 per cent in Scheme B areas. • There were fewer people aged 18 – 24 in Scheme B areas (16 per cent compared with 18 per cent in control areas and 21 per cent in Scheme A areas). • Control areas had more people without any academic qualifications (68 per cent compared with 63 per cent in ETU areas). • Fewer people in control areas were paying rent on their accommodation (15 per cent compared with 19 per cent in Scheme A and 20 per cent in Scheme B). 2.9 Summary The most important difference in the 1998 survey was that there were fewer people aged 18 to 25 (19 per cent) compared with the 1996 survey (32 per cent). Related to this were corresponding changes in marital status, household type, housing tenure and educational level of respondents. For example, in 1998 there were fewer single people and more who were divorced or living with a partner. A greater proportion of respondents lived alone (36 per cent compared with 27 per cent in 1996) and fewer lived with parents (24 per cent compared with 36 per cent in 1996). More people were tenants (44 per cent compared with 29 per cent in 1996) and therefore would have to take housing costs into consideration when accepting a job. Respondents had already spent time out of the labour market as they had been claiming benefit for at least six months when sampled. Many were likely to experience additional problems getting a job because of a lack of skills or ill health. Almost half of the sample (47 per cent) had no recognised educational qualifications (compared with 41 per cent in 1996) and there was a seven per cent increase in the proportion reporting a long-standing 34 illness or disability since 1996 (43 per cent). Women often fared worse in these areas: they were more likely to have health problems and typically were less well educated than men. They were also far more likely to be living with a partner (40 per cent compared with 27 per cent of men). The majority of IS claimants reported health problems and over half said they were in receipt of disability benefits reflecting the substantial differences of this client group compared with those sampled as receiving JSA. By the time of second interview a substantial minority of people had experienced changes in their marital status, household type and housing tenure. This degree of change is typical of younger, single people and has been previously found among samples of people in the range of ETU (Finlayson et al, 2000). 35 3 3.1 Introduction INCOME AND BENEFITS This chapter briefly describes the financial situation of the unemployed sample at first interview. As so few people were in paid work at this time the largest component of income for most respondents was benefits. At second interview, the majority of people were still claiming benefits so the situation was largely unchanged. Clearly, the decision to leave unemployment to take a job will involve a consideration of potential income from work with current income from benefits and so the type and average amounts of benefit received by respondents is discussed in Sections 3.2 and 3.3. However, housing costs and other financial aspects such as savings and debts are also likely to influence the decision to work and these are outlined in later sections. 3.2 Receipt of benefits 3.2.1 Benefit receipt at first interview Respondents were shown a list of benefits and asked which they or their partner were receiving at the time of interview (Table 3.1). Just under two per cent of respondents said they were receiving ETU (eight per cent of those in paid work of at least 16 hours a week, discussed further in Chapter 6). Table 3.1 Which of these benefits are you (or your partner) receiving at this moment? Cell percentages Unemployed survey 1996 Income Support Unemployed survey 1998 64 31 JSA (Contribution-based) JSA (Income-based) N/A N/A 10 20 JSA (Type unknown) Unemployment Benefit N/A 11 19 N/A Council Tax Benefit Training Allowance 22 3 35 1 Earnings Top-Up Employers Pension N/A 0 2 3 State Pension Widows Benefit 3 * 2 * Other social security benefit None of these 0 22 7 9 27 5 43 5 1985 2187 Other benefits Housing Benefit Mortgage Interest payment with IS Base: all respondents Notes: respondents could give more than one answer and so be counted several times. *: less than 0.5 per cent. N/A: benefit not available at that time. 37 The most common types of benefit received were JSA (quoted by 49 per cent of respondents) and IS (31 per cent). At time of sampling 73 per cent of respondents were claiming JSA and by interview 59 per cent of these people said they were receiving it. With an average gap of 70 days between sampling and interview it is unlikely that 41 per cent of people had stopped claiming JSA in that time. There could have been some degree of confusion if respondents thought their benefit was called Unemployment Benefit (not on the list in 1998) or thought that it was IS they received while claiming benefit as unemployed (15 per cent of these people said they were receiving IS at interview). In fact, eight per cent of those who had, at the start of the interview, defined themselves as unemployed and claiming benefit later said they did not receive JSA or IS and two per cent (32 respondents) stated that they did not receive any social security or disability benefit. Likewise, of the 27 per cent of respondents who were claiming IS when sampled, 56 per cent said they were receiving it at interview (another 19 per cent said their partner received it). Linked to the changes in housing tenure of sample respondents discussed in Chapter 2, more people in the 1998 survey were receiving Council Tax Benefit (35 per cent) and HB (43 per cent) than in the 1996 survey (22 per cent and 27 per cent respectively). Almost all people in rented accommodation said they received HB (91 per cent) and 59 per cent received Council Tax Benefit. Fifty four per cent of those owning their property received Council Tax Benefit, as did 58 per cent of people with a mortgage. Forty three per cent of people with a mortgage received help in the form of IS with their mortgage interest payments. Ten per cent of respondents said they had been part of a household that had received Family Credit at some time previously, many of them will have been children at the time (24 per cent of people aged 18 to 24 had previous experience of Family Credit). As shown in Table 2.11 in Chapter 2, 16 per cent of respondents said they were in receipt of a disability benefit. Most of these people (84 per cent) had been sampled as claiming IS. Overall, 28 per cent of the IS sample said they received Incapacity Benefit, 17 per cent claimed Attendance Allowance or Disability Living Allowance, seven per cent got a Disability Premium with their IS, and five per cent received Severe Disablement Allowance. Table 3.2 illustrates the types of disability benefit claimed by respondents who were receiving a disability benefit. There were few changes from the 1996 survey; the most common type was still Incapacity Benefit (57 per cent), followed by Attendance Allowance or Disability Living Allowance (36 per cent). 38 Table 3.2 Type of benefit received by those claiming disability benefits Cell percentages Unemployed survey 1996 Unemployed survey 1998 Men Women All Men Women All Incapacity Benefit 70 53 64 69 40 57 Severe Disablement Allowance 7 13 10 6 16 10 Invalid Care Allowance 5 Disability Premium with 6 5 3 8 5 10 25 16 12 16 14 Allowance Disability Working 31 34 32 31 42 36 Allowance Statutory Sick Pay 3 2 2 1 3 2 1 1 0 1 1 1 Other 0 0 0 6 10 8 167 104 271 192 146 338 Income Support/ Housing Benefit Attendance Allowance/ Disability Living Base: all respondents Note: an individual may claim more than one type of benefit and so be counted several times. Men were more likely to claim Incapacity Benefit (69 per cent) than Attendance Allowance or Disability Living Allowance (31 per cent) whereas the situation was reversed for women (40 per cent received Incapacity Benefit and 43 per cent got Attendance Allowance or Disability Living Allowance). Fewer female respondents were claiming Incapacity Benefit in 1998 (40 per cent) than in the 1996 survey (53 per cent). Examining benefit receipt by work status at interview also suggested that some respondents were not clear which benefits they were receiving (Table 3.3). A minority of those working 16 or more hours a week at interview said they, or their partner, were receiving IS or JSA. 3.2.2 Benefit receipt at second interview There were few overall changes in benefit receipt among the 1998 sample between the interviews (Table 3.3). Twelve per cent of those in work at second interview said they were receiving ETU (discussed further in Chapter 6). 39 Table 3.3 Which of these benefits are you (or your partner) receiving at this moment? Cell percentages First interview Second interview In work Not In work Not (16+ hours) in work (16+ hours) in work Income Support 9 34 * 39 JSA (Contribution-based) JSA (Income-based) 1 3 10 22 0 * 13 19 JSA (Type unknown) New Deal Allowance 6 4 20 2 * 1 13 3 Council Tax Benefit Earnings Top-Up 14 8 37 * 9 12 46 * Employers Pension State Pension 2 * 3 2 2 * 3 4 Widows Benefit Other social security benefit 0 2 * 8 0 1 1 8 None of these 62 5 70 6 Other benefits Disability benefits 6 16 4 17 Housing Benefit Mortgage Interest 15 46 7 45 payment with IS * 5 * 5 156 1950 165 1083 Base: all respondents Notes: respondents could give more than one answer and so be counted several times. *: less than 0.5 per cent. 3.3 Average benefit received The mean amount of benefits received per week was £62.24 (£50.00 median) (Table 3.4). This figure excludes HB or Mortgage Interest premium and any disability benefits received by the respondent. Average disability benefits were £85.19 (mean) or £72.00 (median) a week giving an average total received by respondents claiming disability benefits of £170.61 (mean) or £149.39 (median) a week. Single people living with their parents received the least in benefit, on average about £47 per week. Respondents living alone averaged slightly more at £58 per week, partly because of the inclusion of Council Tax Benefit, while those with a partner received around £88 per week. Table 3.4 Average benefit received per week by household type 40 Base Unemployed survey 1998 Lives alone Lives with partner only Lives with parents, no partner Other 609 287 410 301 £58.20 £88.39 £46.74 £66.78 All respondents receiving benefits 1607 £62.24 Note: this table excludes disability benefits and HB/Mortgage Interest premium Average previous earnings for people who had worked in the five years before first interview were £124 per week (median) or £136 per week at 1998 levels. While the average amount of benefit received was substantially lower than average previous earnings the inclusion of housing costs (discussed in Section 3.4) could narrow the gap quite considerably. 3.4 Housing costs Thirty per cent of respondents lived in accommodation where their parents paid the housing costs, although 70 per cent of these made a contribution to the expenses. The average contribution made was about £20 per week (Table 3.5). Almost all tenants said they received HB (91 per cent) and most received benefit that met the full costs of their housing, an average of £42 per week. Average mortgage payments were similar at around £45 per week but only 43 per cent of people with a mortgage received help with their mortgage interest payments and this assistance averaged £27 per week. Table 3.5 Average housing costs per week by tenure Percentage Base receiving Average Average assistance assistance housing cost with costs received Parents paying housing costs – respondent making contribution 465 £19.61 0 £0.00 Owns with mortgage Rents – 100% HB 211 557 £45.36 £41.94 37 100 £27.10 £41.94 Rents – no or partial HB 396 £50.39 76 £44.17 Because of assistance from parents or in the form of benefits, 66 per cent of respondents had housing costs of less than £10 per week (Table 3.6). However, when HB and Mortgage Interest premium is excluded housing costs rise substantially and only 27 per cent of respondents would have housing costs of less than £10 per week. 41 Table 3.6 Housing costs per week Column percentages Net housing costs Gross housing costs per week per week £0 - £10 66 27 £11 - £20 £21 - £30 17 7 14 11 £31 - £40 £41 - £50 4 3 18 14 £51 - £60 £61 - £70 1 * 5 4 £71 - £80 £81 - £90 * * 3 2 £91 - £100 £101+ * * 1 1 1888 1888 All respondents *: less than 0.5 per cent. Around four in ten people living with their parents wanted to move away: 29 per cent wanted to live alone and 11 per cent wanted to live with others. The remaining 60 per cent said they preferred this arrangement. Forty per cent of those who wanted to live elsewhere said they would experience financial difficulties if they did not live with their parents and another 15 per cent thought they would not manage very well financially. Seven per cent of respondents responsible for their housing costs were behind with their payments. For these people, the mean length of time behind with mortgage payments was 30 weeks and the mean debt was £1370. For tenants, on average, they were behind six weeks with the rent or £295. 3.5 Savings Overall, 27 per cent of respondents said they had some savings (Table 3.7). Respondents in Scheme B areas were less likely to have savings (22 per cent) compared with Scheme A areas (28 per cent) or control areas (31 per cent), and in individual areas rates of savings ranged from 15 per cent of respondents in Perth to 39 per cent in Southampton. Having savings appeared to be related to respondents’ stage in the lifecycle. People with a partner and in owner-occupied accommodation had above average rates of savings as did respondents aged 55 or over (38 per cent had savings). People who defined themselves as unemployed and claiming benefit at time of interview were less likely to have savings than people who had moved into work or economic inactivity. Most people kept their savings in the bank (73 per cent) or Building Society (44 per cent) and 14 per cent had Premium Bonds. The mean 42 amount of savings was £2027 but this was distorted by the presence of some people with large amounts of savings as the median amount of savings was only £100. Three per cent of all respondents (11 per cent of those with savings) had savings of more than £3,000. Just four per cent said they had shares, unit trust, bonds or securities. For those with savings, 24 per cent said they had more savings six months previously, 62 per cent said their savings were about the same and 14 per cent said their savings were less six months earlier. There was little evidence that people had used up their savings while unemployed as only four per cent of those without any savings said they had more saved six months previously. Table 3.7 Percentage of respondents with savings Cell percentages Base Has savings ETU area Scheme A 766 28 Scheme B Control areas 717 704 22 31 Owns house Owns with mortgage 189 213 43 42 Renting With parents 971 654 18 28 Other 160 25 Partnership status Has partner 686 33 No partner 1501 24 Economic activity at first interview In work (any hours) 235 35 Unemployed and claiming benefit Unemployed but not claiming 1374 135 22 46 Other 430 31 Base 2187 27 Housing tenure 3.6 Material and financial wellbeing Almost half the sample (48 per cent) said there were things they needed to buy at present that they did not have the money for. Of these people, 30 per cent said they needed to buy clothes and 17 per cent said footwear. Other common responses were electrical equipment or repairs (25 per cent), bed or bedding (17 per cent), carpets or curtains (16 per cent), home improvements (16 per cent), furniture (15 per cent), a car (ten per cent), and a holiday (nine per cent). 43 Nineteen per cent of respondents responsible for paying household bills said they were behind with their payments. Thirty nine per cent of these people were behind with water payments, 28 per cent owed Council Tax, 28 per cent were behind with gas payments, and 20 per cent owed money for electricity. Unemployed people often find it difficult to obtain credit and so relatively few respondents had borrowed money from established sources like a bank overdraft (seven per cent) or a finance company loan (six per cent) to buy things (Table 3.8). They were more likely to owe money to friends or relatives (16 per cent). Around one fifth of people who said they had borrowed money to buy things had trouble keeping up with the repayments. Almost eight out of ten respondents (78 per cent) said they had worried about money in the last few weeks. More than a third (34 per cent) said they worried about money all the time. Fifty three per cent reported that they had trouble repaying debts over the previous two years. Table 3.8 Using credit to buy things Cell percentages Percentage of people using the Percentage using method who were unable method to keep up repayments Bank overdraft 7 19 Fixed term loan from bank Loan from finance company 2 4 13 26 Loan from money lender Loan from friend/relative 2 16 23 18 Loan from employer * 0 Any borrowing 26 22 Notes: *: less than 0.5 per cent.. Bases: first column base is all respondents (2187), second column base is those in first column. Almost half of people (49 per cent) described their financial situation as ‘getting by alright’ (Table 3.9). Just over one-fifth thought they were managing quite or very well. Thirteen per cent thought they did not manage very well and 17 per cent admitted financial difficulties. The majority of people who had moved into work by first interview (85 per cent) were at least able to say they ‘got by alright’ compared with 68 per cent of people who were unemployed and claiming benefit. 44 Table 3.9 Which of the phrases best describes how you are managing financially these days? Column percentages Unemployed survey 1996 Unemployed survey 1998 I manage very well 8 5 I manage quite well I get by alright 16 46 16 49 I don’t manage very well I have some financial difficulties 12 14 13 14 I am in deep financial trouble 4 3 1985 2187 Base: all respondents 3.7 Summary This chapter has outlined the benefits respondents said they received at time of interview. As expected, the most common types of benefit reported at first interview were JSA (49 per cent) and IS (31 per cent). Substantial groups also received HB (43 per cent) and Council Tax Benefit (35 per cent) and the potential loss or reduction of these benefits may have reduced the attractiveness of ETU for these people. Sixteen per cent of the sample received a disability benefit. Few people thought they were managing well financially (21 per cent). Half of the sample thought they ‘got by alright’ and 30 per cent admitted to financial difficulties. Almost half said there were things they needed to buy at present that they did not have the money for and one-fifth were behind with payments for household bills. Seventy eight per cent of people said they had worried about money in the last few weeks. Few people had savings (27 per cent) and the median amount for those who did was only £100. Many relied on family and friends for support. They were more likely to borrow money from friends or relatives than more formal sources and more than a fifth of those living with parents felt they could not afford to move out even though they wanted too. 45 4 4.1 Introduction CONTACT WITH THE LABOUR MARKET This chapter mainly describes the extent of contact with the labour market of respondents at first interview. By doing so, it acts as an introduction for the more detailed analysis in Chapter 5 that focuses on the longitudinal dimension of this survey, particularly movements out of unemployment. This chapter starts by providing a detailed description of the economic activity of respondents at both first and second interview (Section 4.2). As relatively few people were in work at first interview, the main focus after this is on contact with the labour market for the non-working group, particularly their previous work experience (Section 4.3) and current jobsearch methods (Section 4.4). An important finding in Section 4.4 is that not all non-working respondents in the survey were actively seeking work at the time of first interview, for a multitude of reasons. Linked to this is the discussion of respondents’ aspirations for the future at first interview, which includes the minimum wage they would accept and what they thought was most likely to happen to them in the next few years (Section 4.5). Lastly, the chapter briefly looks at the experiences of those who had moved into work by first or second interview (Section 4.6). 4.2 Economic status at interview This section considers how respondents defined their economic activity at both interviews. 4.2.1 Economic status at first interview There was a gap, on average, of 70 days between the 1998 sample being selected and respondents being first interviewed. During that time 11 per cent of the sample had moved into paid work: seven per cent were working for 16 or more hours per week, three per cent were working less than 16 hours per week and one per cent were self-employed (Table 4.1). In the 1996 survey more people had moved into work of 16 or more hours (11 per cent) and self-employment (two per cent) by first interview but the gap between sampling and interview was longer for the earlier survey: an average of 107 days. Another three per cent of 1998 respondents were on a New Deal programme: this was 11 per cent of people aged under 25 (Table 4.2). Two per cent were attending a training programme and one per cent had gone into full-time education. Sixty three per cent of respondents defined their activity as unemployed and claiming benefit in the 1998 survey, compared with 57 per cent in 1996. Around half of women said they were unemployed and claiming benefit compared with 69 per cent of men. Female respondents more 47 often said they were unemployed but not claiming (10 per cent) than did men (four per cent). One in ten people said they had been sick or disabled for more than six months at time of interview, 13 per cent of women and eight per cent of men. Table 4.1 Economic status at time of first interview by gender Column percentages Unemployed survey 1996 Unemployed survey 1998 Men Women All Men Women All Employed 16+ hours per week 10 12 11 8 6 7 Employed <16 hours per week 1 Self-employed 3 New Deal programme N/A 7 3 2 6 3 1 N/A 2 N/A 1 3 * 3 1 3 2 57 2 69 1 49 2 63 Government training Claimant unemployed 2 62 2 47 Unemployed but not claiming 8 7 7 4 10 6 Full-time education Temporarily sick 1 1 1 1 1 1 (< 6 months) Long-term ill-health 1 3 2 1 1 1 (> 6 months) Care of home 10 14 11 8 13 10 and family Retired * 1 1 5 * 2 * * 4 3 2 1 Other 1 1 1 * 2 * 1398 593 1991 1499 688 2187 Base: all respondents Notes: *: less than 0.5 per cent. N/A: not available. Columns may not sum to 100 because of rounding. Age group Those aged under 25 were more likely to have got a job of at least 16 hours a week by the time of first interview whereas those aged 35 or over were less likely (Table 4.2). Fourteen per cent of those aged 45 or over had long-term health problems. 48 Table 4.2 Economic status at time of first interview by age group Column percentages Unemployed survey 1996 Unemployed survey 1998 18-24 25-34 35-44 45-54 55+ 18-24 25-34 35-44 45-54 55+ Employed 16+ hours 18 12 12 8 8 13 10 5 5 4 Employed <16 hours Self-employed 2 1 3 2 2 1 7 3 3 2 1 * 3 * 2 * 4 1 4 1 New Deal programme Government training N/A 3 N/A 3 N/A 1 N/A 3 N/A 1 11 3 1 4 * 2 1 * * * Claimant unemployed Unemployed but not claiming 65 3 67 4 65 6 47 10 38 16 60 3 71 2 74 5 60 8 55 12 Full-time education Temporarily sick 2 1 1 * 0 3 2 1 0 0 (< 6 months) Long-term ill-health 2 1 2 3 2 1 2 * 1 1 (> 6 months) Care of home/family 4 * 7 0 10 * 18 * 21 * 4 0 4 * 10 1 14 4 13 2 Retired Other 0 0 0 0 0 0 1 0 9 0 0 * 0 0 0 0 * 1 6 1 618 335 237 382 324 408 384 357 571 446 Base: all respondents Notes: *: less than 0.5 per cent. N/A: not available. Columns may not sum to 100 because of rounding. Partnership status Single people (those without a partner in the household) were more likely to say they were unemployed and claiming benefit at time of interview than were those with a partner (Table 4.3). Seventy three per cent of single men and 61 per cent of single women said they were unemployed and claiming benefit at interview compared with 61 per cent of partnered men and just 32 per cent of partnered women. In fact, respondents in couples made up almost all of those unemployed but not claiming. Almost a quarter of women with partners defined themselves as unemployed but not claiming, 15 per cent said they had long-term health problems and nine per cent were looking after the home and family. Less than one-fifth (17 per cent) of the partners of respondents were in paid work themselves (compared with 36 per cent in 1996). Just eight per cent of the partners of respondents who were sampled as claiming IS were working compared with 23 per cent of partners of JSA recipients. 49 Table 4.3 Economic status at time of first interview by gender and marital status Column percentages No partner Has partner Men Women All Men Women All hours per week Employed <16 8 6 8 7 5 6 hours per week Self-employed 2 1 5 * 3 * 2 2 6 1 4 1 New Deal programme Government training 3 3 4 2 3 2 2 1 1 1 1 1 Claimant unemployed Unemployed but 73 61 70 61 32 49 not claiming Full-time education 2 1 1 1 2 1 10 1 24 1 17 1 Temporarily sick (< 6 months) 1 1 1 1 1 1 Long-term ill-health (> 6 months) 6 12 8 12 15 13 Care of home and family * 1 * 1 9 4 Retired Other * * 4 1 1 * * * 3 1 1 * 1085 406 1491 403 277 680 Employed 16+ Base: all respondents Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. Benefit type at sampling Twice as many people who had been receiving JSA were in paid work by the first interview (12 per cent) as respondents recruited from IS (Table 4.4). Seven out of ten of the JSA sample defined themselves as unemployed and claiming benefit and six per cent said they were unemployed but not claiming. Interestingly, almost half of the IS claimants (46 per cent) defined themselves as unemployed and claiming benefit at interview and just under a third (31 per cent) said long term health problems kept them from work. 50 Table 4.4 Economic status at time of first interview by benefit type at sampling Column percentages Unemployed survey 1998 JSA IS All Employed 16+ hours per week 8 5 7 Employed <16 hours per week Self-employed 4 1 1 * 3 1 New Deal programme Government training 4 3 * * 3 2 Claimant unemployed Unemployed but not claiming 70 6 46 6 63 6 Full-time education Temporarily sick 1 1 1 (< 6 months) Long-term ill-health 1 2 1 (> 6 months) Care of home and family 1 1 31 3 10 2 Retired Other * * 4 1 1 * 1577 599 2187 Base: all respondents Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. ETU areas There were no significant differences in respondent’s economic activity between the ETU areas and the control areas (Table 4.5). Table 4.5 Economic status at time of first interview by ETU type Column percentages Unemployed Survey 1998 Scheme A Employed 16+ hours per week 6 Employed <16 hours per week 3 Self-employed 1 New Deal programme 4 Government training 2 Claimant unemployed 63 Unemployed but not claiming 5 Full-time education 1 Temporarily sick (< 6 months) 1 Long-term ill-health (> 6 months) 10 Care of home and family 1 Retired 1 Other * Base: all respondents 766 Scheme B Control All 9 7 7 3 1 2 2 63 5 1 1 10 1 2 * 3 * 2 1 63 9 1 1 9 2 1 1 3 1 3 2 63 6 1 1 10 2 1 * 717 704 2187 Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. 51 4.2.2 Economic status at second interview There was an average duration of 10 months between first and second interview. Most people had a gap between interviews that ranged between eight and 12 months (94 per cent). More people had got jobs by the second interview but the majority were still not in work (Table 2.6). Almost half (49 per cent) said they were unemployed and claiming benefit, down from 63 per cent at first interview. Fifteen per cent said they had long-term health problems, up from 10 per cent at first interview. Thirteen per cent were working 16 or more hours a week, three per cent were working for fewer than 16 hours and two per cent were self-employed. Table 4.6 Economic status at both interviews by gender Column percentages First interview Employed 16+ hours per week Second interview Men Women All Men Women All 8 6 7 14 11 13 Employed <16 hours per week 2 6 3 1 7 3 Self-employed New Deal programme 1 3 * 3 1 3 2 3 1 2 2 3 Government training Claimant unemployed 2 69 1 49 2 63 1 57 1 34 1 49 Unemployed but not claiming 4 10 6 5 8 6 Full-time education Temporarily sick 1 1 1 1 1 1 (< 6 months) Long-term ill-health 1 1 1 2 1 2 (> 6 months) Care of home 8 13 10 11 21 15 and family Retired * * 4 3 2 1 * 2 7 6 3 3 Other * 2 * 1 1 1 1499 688 2187 859 449 1308 Base: all respondents Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. Despite the improved economic conditions, fewer of the 1998 sample of unemployed people had moved into work by the time of second interview compared with the 1996 sample. Thirty five per cent of the 1996 unemployed sample were unemployed and claiming benefit and 28 per cent were working 16 or more hours a week. Proportions in other economic activities were the same as for the 1998 sample. 52 Gender At second interview, the differences between men and women in this sample followed the same pattern as at first interview. Women were less likely to be claimant unemployed (34 per cent) and more likely to have health problems (21 per cent) or be in work of less than 16 hours a week (seven per cent) than were men. Age group Again, following the pattern at first interview, younger respondents were more likely to be working 16 or more hours a week at second interview (26 per cent). Another seven per cent were on a New Deal programme (Table 4.7). Few of those aged over 55 had moved into work between first and second interview. A smaller proportion said they were unemployed and claiming benefit but more said they were retired or had long-term health problems. Table 4.7 Economic status at both interviews by age group Column percentages First interview Second interview 18-24 25-34 35-44 45-54 55+ 18-24 25-34 35-44 45-54 55+ Employed 16+ hours Employed <16 hours 13 1 10 3 5 2 5 4 4 4 26 2 18 * 10 1 9 4 6 4 Self-employed New Deal programme Government training * 11 3 * 1 4 * * 2 1 1 * 1 * * 2 7 1 1 2 1 2 3 1 1 2 1 2 6 1 Claimant unemployed Unemployed but not claiming 60 3 71 2 74 5 60 8 55 12 42 4 62 3 58 6 48 8 40 8 Full-time education Temporarily sick 3 2 1 0 0 4 1 * * 0 (< 6 months) Long-term ill-health 1 2 * 1 1 2 1 3 1 1 (> 6 months) Care of home/family 4 0 4 * 10 1 14 4 13 2 5 2 8 1 13 * 20 4 22 4 Retired Other 0 * 0 0 0 0 * 1 6 1 0 1 0 * 0 1 1 1 11 1 408 384 357 571 446 209 228 212 343 1308 Base: all respondents Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. Partnership status Single people were more likely to be working 16 or more hours a week at second interview than were those with a partner (15 per cent compared with eight per cent) (Table 4.8). More than a fifth of partnered respondents said they had long-term health problems and 15 per cent of partnered women said they were caring for the home or family. This may reflect caring responsibilities for their partner or other family member. Only 53 one fifth of partnered women said they were unemployed and claiming benefit at second interview and a similar number said they were unemployed and not claiming. Single women were less likely than single men to be unemployed and claiming benefit at second interview (43 per cent compared with 61 per cent). They were more likely to have long-standing ill health (20 per cent compared with eight per cent) or to be working less than 16 hours a week (seven per cent compared with less than 0.5 per cent). Fifteen per cent of partners at second interview were working 16 or more hours a week, four per cent were working for fewer than 16 hours and two per cent were self-employed. Table 4.8 Economic status at second interview by gender and marital status Column percentages No partner Has partner Men Women All Men Women All hours per week Employed <16 15 15 15 11 4 8 hours per week Self-employed * 2 7 * 2 1 2 2 7 2 4 2 New Deal programme Government training 3 2 3 1 3 1 1 1 1 1 1 1 Claimant unemployed Unemployed but 61 43 56 45 20 35 not claiming Full-time education 3 2 2 1 2 2 11 0 19 0 14 0 Temporarily sick (< 6 months) 2 1 2 1 1 1 Long-term ill-health (> 6 months) Employed 16+ 8 20 12 20 22 21 Care of home and family 0 Retired 1 3 5 1 2 * 4 15 6 7 5 Other 1 * 1 2 2 2 618 274 892 241 175 414 Base: all respondents Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. Benefit type at sampling People sampled as receiving JSA were more likely to have moved into work but overall only 15 per cent were working 16 or more hours a week by the time of second interview (Table 4.9). The proportion in work among those sampled as receiving IS barely changed between first and second interviews. By the second interview, 40 per cent of those sampled as receiving IS said they had long-term ill health. 54 Table 4.9 Economic status at time of interview by benefit type at time of sampling Column percentages First interview Second interview JSA IS All JSA IS All hours per week Employed <16 8 5 7 15 5 13 hours per week Self-employed 4 1 1 * 3 1 3 2 2 1 3 2 New Deal programme Government training 4 3 * * 3 2 3 2 0 * 3 1 Claimant unemployed Unemployed but 70 46 63 55 32 49 not claiming Full-time education 6 1 6 1 6 1 6 1 6 1 6 1 Temporarily sick (< 6 months) 1 2 1 2 1 2 Long-term ill-health (> 6 months) 1 31 10 6 40 15 Care of home and family 1 Retired * 3 4 2 1 2 2 4 7 3 3 Other * 1 * 1 2 1 1577 599 2187 962 346 1308 Employed 16+ Base: all respondents Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. ETU area Again, there were no significant differences in the economic activity of respondents in the ETU and control areas (Table 4.10). 55 Table 4.10 Economic status at time of second interview by ETU type Column percentages Second interview Employed 16+ hours per week Scheme A Scheme B Control All 13 12 13 13 Employed <16 hours per week 2 2 4 3 Self-employed New Deal programme 3 2 1 3 1 2 2 3 Government training Claimant unemployed 1 50 1 49 1 48 1 49 Unemployed but not claiming 6 6 6 6 Full-time education Temporarily sick 1 1 1 1 (< 6 months) Long-term ill-health 1 2 2 2 (> 6 months) Care of home and family 16 2 16 3 12 3 15 2 Retired Other 2 1 3 1 4 2 3 1 462 410 436 1308 Base: all respondents Notes: *: less than 0.5 per cent. Columns may not sum to 100 because of rounding. 4.3 Previous work experience 56 On average, the 1998 sample of unemployed people did not have a great deal of recent work experience. Despite claiming their current benefit for less than two years, 45 per cent of the sample said they had not worked in the previous five years, compared with 22 per cent of the 1996 survey (Table 4.11). In fact, 14 per cent said they had never had a paid job. A large proportion of these were young people (45 per cent were aged under 25) but not exclusively so (Table 4.12). There was also a fairly even balance of men and women who said they had never worked and no significant differences according to whether respondents had been sampled as receiving JSA or IS. People who had previously worked but not in the five years before interview were more likely, on average, to be aged 45 or over (69 per cent), have health problems (55 per cent had a long-standing illness or disability), and no educational qualifications (60 per cent). Table 4.11 When did you last have a paid job or work as selfemployed? Unemployed survey 1996 Column percentages Unemployed survey 1998 All Non-working All Non-working In work at interview In the last five years 16 62 N/A 73 11 44 N/A 49 5 – 10 years ago More than 10 years ago 5 6 6 7 17 14 20 16 Never worked 11 14 14 15 1985 1645 2187 1952 Base Note: N/A – not applicable. Table 4.12 shows previous work experience by the age group of respondents. Comparison between age groups is not straightforward as clearly older respondents would have more opportunity to have worked at some time than younger ones. Table 4.12 Time last worked by age group – 1998 survey Column percentages Age group at 1998 interview 18-24 25-34 35-44 45-54 55+ All In work at interview 15 13 7 10 9 11 In the last five years 5 – 10 years ago 48 4 51 12 47 22 37 25 41 22 44 17 More than 10 years ago N/A Never worked 33 3 21 16 8 22 6 23 5 14 14 Base 385 360 575 449 2179 410 Note: N/A – not applicable. Proportion of time spent working in the previous five years On average, 1998 survey respondents had spent just 15 per cent of the previous five years in full-time work (compared with 29 per cent for the 1996 survey) and over half (54 per cent) of the previous five years unemployed and claiming benefit (36 per cent in 1996) (Table 4.13). There were no differences between those in ETU areas and control areas in terms of previous work experience, nor between respondents receiving JSA compared with IS when sampled. However, on average, IS claimants had spent 24 per cent of the last five years coping with health problems and 41 per cent unemployed and claiming benefit. 57 Table 4.13 Average proportion of last five years spent in each activity by gender Column percentages Unemployed survey 1996 Unemployed survey 1998 Men Women All Men Women All hours per week Paid work <16 28 32 29 16 14 15 hours per week Self-employment 1 3 6 1 3 3 1 3 5 1 2 2 Claimant unemployment 39 29 36 59 43 54 Paid work 16+ Non-claimant unemployment 5 4 5 4 9 6 Full-time education Ill-health/disability 11 6 12 7 11 6 4 6 5 11 4 8 Other 7 9 7 7 13 9 Base 1118 480 1598 1234 549 1783 Base: all respondents who gave at least two years activity history information. For women, seven per cent of their time in the previous five years was, on average, spent looking after the home and family, compared with less than one per cent of men’s time. They had also spent less time on average unemployed and claiming benefit (43 per cent compared with 59 per cent of men’s time) and more time unemployed but not claiming (nine per cent) and in part-time work (five per cent). Previous occupation Only one in eight respondents who had worked in the previous five years had work experience in a professional, managerial or technical occupation (Table 4.14). Twenty three per cent of respondents had worked in personal sales and services, 21 per cent were plant operatives, 14 per cent had been in craft occupations and one in ten were in clerical work. Twenty one per cent of employees said they had supervised or managed people in their last job. These results were similar to that of the 1996 survey. Table 4.14 Occupational group of last job Column percentages Managerial/Professional Clerical Craft Personal services and sales Plant operatives Other Unemployed survey 1996 Unemployed survey 1998 15 12 16 20 18 18 12 10 14 23 21 20 Base: those who worked in last five years 1504 58 988 Reason for leaving last job Almost a quarter of respondents had previously been employed on a temporary contract that had come to an end and 27 per cent had been made redundant or their employer had closed down (Table 4.15). More than one in five respondents said they had left their previous job because of ill health, half of people sampled as claiming IS and 12 per cent of people receiving JSA. Poor health may have placed people beyond the financial incentive provided by ETU but ETU could also have helped them work shorter hours without too great a loss of income. Table 4.15 Why did your most recent job end? Column percentages Unemployed survey 1996 Unemployed survey 1998 Firm/site closed down 7 8 Made redundant End of temporary job 25 25 19 23 Dismissed Left because of sickness 8 17 7 22 Left – other reason Other, e.g. moved areas 18 0 17 4 Base 981 988 Base: respondents who were not working at time of interview but had worked as an employee in the previous five years. 4.4 Looking for work At first interview, the majority of this sample were not in work (89 per cent). This section considers the extent to which these non-working respondents were looking for a job. This includes their jobsearch methods, their reasons for not looking and their aspirations for the future regarding work. 4.4.1 Jobsearch activity Looking for work at first interview Fifty five per cent of the sample said they were actively looking for work at the time of first interview (Table 4.16). Twelve per cent were working or waiting to start a job. Eleven per cent said they had not looked for work in the four weeks before interview but would like a job. Just under a quarter of respondents (23 per cent) said they had not looked for work in the previous four weeks and did not currently want a job. In 1998, more people were not looking for work when interviewed (34 per cent) than in the 1996 survey (26 per cent). 59 Table 4.16 Work status at time of first interview Column percentages Unemployed survey 1996 Unemployed survey 1998 In work 16 11 Waiting to start job Has looked for work in 1 1 previous four weeks Not looked for work in 57 55 last four weeks but would like a job Not looking for work N/A 26 11 23 Base: all respondents 1985 2187 Notes: N/A: question not asked in 1996 survey. All respondents receiving JSA should have been actively seeking work in order to meet the conditions of benefit receipt. It is not possible to know accurately who was claiming JSA at time of interview as respondents were sometimes confused about the name of the benefit they received (discussed further in Chapter 3, Section 3.2). However, only 16 per cent of people sampled as receiving JSA said they were not looking for work when interviewed about two to three months later (Table 4.17). In contrast, 82 per cent of those sampled as claiming IS said they were not looking for work. Overall, 62 per cent of non-working respondents had looked for work in the four weeks before interview: 82 per cent of the JSA sample but only 13 per cent of sampled IS claimants. Another 12 per cent said they would like a job but had not looked in the previous four weeks. Table 4.17 Work status at time of first interview by benefit type at sampling Column percentages JSA IS All In work Waiting to start job 12 1 6 * 11 1 Has looked for work in previous four weeks 71 12 55 Not looked for work in last four weeks but would like a job 7 21 11 Not looking for work 9 61 23 Base: all respondents at first interview 1585 599 2187 Notes: * Less than 0.5 per cent. Slightly fewer non-working respondents were looking for work in Scheme B areas (60 per cent) than in Scheme A (64 per cent) or control areas (63 per cent) (Table 4.18). There was also some variation at individual areas. 60 Middlesborough had the highest proportion of people looking for work (69 per cent) and South Wales the lowest (54 per cent). One third of respondents in Doncaster said they were not looking for work and did not want a job compared with 22 per cent in Newcastle and Southampton. This may reflect local variation in job opportunities. Table 4.18 Jobsearch status of non-working respondents by ETU area Row percentages Not looking Looking but would Not looking Base for work like a job for work ETU Scheme A area ETU Scheme B area 684 625 64 60 11 12 25 28 Control area 624 63 13 24 Newcastle Castleford 167 167 68 65 10 6 22 29 Southend North Wales 171 179 61 62 16 11 23 27 Sunderland Doncaster 166 158 57 59 14 8 29 33 Bournemouth Perth 147 154 58 65 16 12 26 23 Middlesborough Southampton 176 118 69 64 11 14 20 22 South Wales Rotherham 166 164 54 65 16 11 30 24 All non-working respondents 1930 62 12 26 More male respondents said they were looking for work (72 per cent) than were female respondents (41 per cent). In the 1996 survey, the equivalent figures were 74 per cent of men and 54 per cent of women. In contrast, 44 per cent of women and 18 per cent of men said they were not looking for work and did not want a job. There were also differences according to the marital status of respondents (Table 4.19). Overall, seven out of ten single people said they were looking for a job at time of first interview compared with 44 per cent of those with a partner (in the 1996 survey 61 per cent of non-working respondents with partners were looking for work). However, just one fifth of women with a partner said they were looking for work compared with 61 per cent of men with a partner. A similar proportion of partnered women (19 per cent) said they had not looked for work in the previous four weeks but would like a job and 61 per cent were not looking for work at all (compared with 27 per cent of partnered men). 61 Table 4.19 Jobsearch status of non-working respondents by gender and marital status Column percentages No partner Looking for work Has partner Men Women All Men Women All 76 55 70 61 20 44 Not looking but would like a job 10 13 11 12 19 15 Not looking for work 14 32 19 27 61 41 Base: all respondents 970 359 1329 356 245 601 Younger people appeared more likely to be seeking work at time of interview (Figure 4.1). Almost half (46 per cent) of respondents aged 55 or over said they were not looking for work and did not want a job. Figure 4.1 Jobsearch status at interview by age group Respondents with educational qualifications had a greater attachment to the labour market on average. The greatest contrast lay between those with some educational qualifications and those with none: around half of people without any educational qualifications said they were looking for work compared with 76 per cent of people with both academic and vocational qualifications (Figure 4.2). 62 Figure 4.2 Jobsearch status at interview by type of qualifications Reasons for not looking for work The most common explanations as to why people were not looking for work were ill-health or disability (64 per cent), looking after family (14 per cent), getting near to retirement age (six per cent) and going into full-time education or training (seven per cent) (Table 4.20). Health problems were mentioned by 76 per cent of people not looking for work who had been claiming IS and 42 per cent of those sampled as receiving JSA. Almost a fifth of the JSA recipients who were not looking for work said they were attending a training or educational course. Just two per cent of respondents not currently looking for work were concerned that they would lose benefits from taking a job. Table 4.20 Reasons for not looking for work by benefit type at sampling Multiple response percentages JSA IS All Ill health/disability 42 76 64 Looking after family Retired/too old 10 6 15 5 14 6 Full-time education or training Would lose money/benefits 19 5 2 1 7 2 No work available Other reasons 5 13 1 3 2 5 Base: respondents not looking for work 244 486 730 Respondents could give more than one answer so columns need not sum to 100. The pattern of responses was similar in ETU and control areas (Table 4.21). Slightly more people in Scheme B areas said health problems were stopping them looking for work (67 per cent) but fewer said family commitments (10 per cent) compared with Scheme A and control areas. 63 Table 4.21 Reasons for not looking for work by ETU type Multiple response percentages Scheme A Scheme B Control Ill health/disability 62 67 61 Looking after family Retired/too old 16 3 10 9 14 4 Full-time education or training Would lose money/benefits 10 1 5 2 8 2 No work available Other reasons 1 8 2 6 4 9 244 252 234 Base: respondents not looking for work Respondents could give more than one answer so columns need not sum to 100. One third of people who had not looked for work in the previous four weeks said they would like a job if a suitable one were available (Table 4.22). Sixty per cent of these said they were not currently looking for work because of health problems, 10 per cent had caring responsibilities, and 11 per cent were undertaking further training or education. Eighty per cent said they had not looked for work in more than a year. Forty four per cent thought they might look for a paid job one day but twothirds could not estimate how long it would be before they looked for work. Respondents who had been receiving JSA when sampled were more likely than IS claimants to say they would take a suitable job (44 per cent) and 57 per cent expected to return to work at some point. They had also more recently been looking for work than people who were claiming IS at time of sampling. There were no significant differences between respondents in ETU areas and control areas. Table 4.22 Expected jobsearch among respondents not looking for work by benefit type at sampling Column percentages JSA IS All Would you like to have a job if a suitable one was available? 64 Yes No 44 44 26 61 32 56 Not sure 12 13 12 How long has it been since you last looked for work? 1 –2 months 20 2 8 3 –6 months 7 –12 months 13 8 2 7 5 7 More than 1 year 59 89 80 Do you think you might look for a paid job one day? Yes 57 38 44 No 43 62 56 Base: respondents not looking for work 244 486 730 Young respondents not currently looking for work were more likely to have looked in the last year (46 per cent) and the majority expected to work at some time in the future (91 per cent) (Table 4.23). The majority of those aged 45 or over did not expect to work again. Table 4.23 Expected jobsearch among respondents not looking for work by age group Column percentages 18-24 35-44 45-54 All Would you like to have a job if a suitable one was available? Yes 45 38 52 29 19 No Not sure 38 10 57 14 73 8 47 8 25-34 35 27 How long has it been since you last looked for work? 1 –2 months 3 –6 months 27 12 8 12 12 8 2 2 3 3 7 –12 months More than 1 year 7 54 9 71 10 70 6 90 6 88 71 29 66 34 38 62 16 84 79 92 222 234 Do you think you might look for a paid job one day? Yes No 91 9 Base: respondents not looking for work103 Similar patterns can be seen with regard to partner’s jobsearch. Overall, 23 per cent of non-working partners were looking for work: 11 per cent of the partners of male respondents and 41 per cent of partners of female respondents. Fewer partners were looking for work in Scheme B areas (20 per cent) than in Scheme A (26 per cent) or control areas (24 per cent). Only five per cent of the partners of respondents who were claiming IS when sampled were looking for work compared with 38 per cent of the partners of recipients of JSA. Respondents with health problems were less likely to have partners that were looking for work (19 per cent compared with 28 per cent of others), and older partners (particularly those aged 55 or over) were rarely actively seeking employment (13 per cent). The substantial proportion of the sample who were not looking for a job at time of first interview had important implications for any likely effect of ETU on getting these people into paid work, particularly as most people’s reasons for not seeking work were other than financial considerations. Many of the respondents sampled as IS, rather than JSA recipients, seem placed beyond the reach of the immediate incentive effect of ETU, and in some cases, out of the labour market altogether. 65 4.4.2 Jobsearch methods Table 4.24 illustrates the jobsearch methods used by respondents looking for work in the four weeks before interview. The most commonly used methods were looking at advertisements in newspapers and magazines (89 per cent) and in the Jobcentre (81 per cent). Compared with the 1996 survey, most methods were used by slightly fewer jobseekers in 1998; the exception being speaking to staff in Jobcentres about vacancies which had increased from 49 per cent to 55 per cent of respondents. Table 4.24 Jobsearch methods in four weeks before interview Multiple response percentages Unemployed survey 1996 Unemployed survey 1998 Asking friends and family Looking at adverts in Jobcentre 68 86 65 81 Attending Jobclub Attending Restart course 20 6 15 5 Speaking to staff at Jobcentre Speaking to staff at Careers Office 49 11 55 10 Contacting private employment agency Contacting employer directly 24 48 19 40 Looking at adverts in newspapers Looking at adverts in shop windows 90 55 89 48 Trying to find self-employed work Other 14 6 11 4 Base: respondents looking for work 1124 1202 Note: individuals could give more than one answer and so be counted several times. The majority of jobseekers (51 per cent) spent between two and five hours a week looking for a job and 59 per cent spent between £1 and £9 per week on jobsearch (Table 4.25). The mean number of job applications made in the four weeks before interview was five but this was heavily influenced by a few people who claimed to have made many applications; the median was two. Sixteen per cent of people had attended at least one interview in the previous four weeks. There was little difference between respondents in ETU areas and those in control areas. Focusing on the jobsearch methods of those who had found work did not identify any clearly successful strategies. It seemed that people who had managed to get a job were less likely to use many of the methods. On average, people still looking for work at first interview had used 4.4 methods (4.8 in 1996) while those who had found work had used three methods (3.9 in 1996). One fifth of those who had found work by time of interview said they never used the Jobcentre to look for work. On average they spent slightly less time looking for work than current jobseekers and spent no more money on jobsearch activities. 66 Table 4.25 Jobsearch activity by ETU type Column percentages Scheme A Scheme B Control All Time spent per week looking for jobs Less than 1 hour 2 – 5 hours 14 48 13 53 14 51 14 51 6 – 9 hours 10 – 19 hours 15 15 14 11 12 16 14 14 20 or more hours Don’t know 4 4 5 4 4 4 4 4 Money spent per week looking for jobs Nothing Less than £1 14 15 14 14 15 11 14 13 £1 - £4 £5 - £9 37 20 40 19 40 22 39 20 £10 or more Don’t know 10 4 9 4 8 4 9 4 Any interviews with employers for job vacancies in last 4 weeks? Yes No 18 82 16 84 13 87 16 84 2 2 3 2 441 374 391 1206 Median number of job applications in last 4 weeks Base: respondents looking for work 4.4.3 Work and wage expectations Type of work wanted Most people looking for work at first interview were seeking employment (69 per cent) while 10 per cent particularly wanted self-employment and 21 per cent were prepared to consider either. One in ten people specifically wanted to work less than 16 hours per week. Those aged 55 or over were more likely to want to work less than 16 hours a week (17 per cent) but there was no difference between men and women in their preference for work below 16 hours per week. Men were more likely though to specifically want work of more than 30 hours a week (63 per cent) than were women (52 per cent). One fifth of respondents were happy to consider any hours. More people in control areas particularly wanted full-time work (67 per cent) than in Scheme A (56 per cent), or Scheme B areas (61 per cent). Two-thirds of people looking for work wanted a permanent job (one that they could feel secure in for as long as they wanted). Another 15 per cent wanted at least one year’s security to the job. Eighty per cent believed it would be difficult to get the degree of security they wanted in a job. 67 Aspiration wages Respondents were first asked how much money they would need to be offered before a job would be worth taking (their target wages). They were then asked to name the lowest amount of take-home pay they would ever be willing to accept (their minimum wages). Typical target hourly wages were £4.00 per hour or £140.00 per week (median). Women had lower target earnings (£3.85 per hour), as did those aged under 35 (£3.75 per hour) and respondents living with their parents (£3.75). Above average target wages were mentioned by those with partners (£4.50), respondents aged 55 or over (£5.00), people in socioeconomic groups 1 to 3 (£5.00), people with mortgages (£4.86), and those with degrees (£5.35). Most respondents thought it would be difficult to find a job at those wages (36 per cent said quite difficult and 52 per cent said very difficult). Just nine per cent expected to get a job paying those wages. The minimum wages jobseekers were prepared to accept were only slightly lower on average than their target wages (£3.50 per hour or £130.00 per week median) (Table 4.26). They were slightly higher than the minimum wages quoted by the unemployed sample in 1996 but this probably reflects the different characteristics of the respondents in each survey. Again, women were prepared to accept lower wages than were men. Those aged under 25 gave lower wages (£3.00 per hour) as did those living with parents (£3.13 per hour). Above average wages were sought by those with partners (£3.75), respondents aged 55 or over (£3.95), people in socio-economic groups 1 to 3 (£4.00), people with mortgages (£4.00), and those with degrees (£4.29). Table 4.26 Median minimum wages for those looking for work at first interview Unemployed survey 1998 Men Women All Minimum acceptable weekly wages £140.00 £110.00 £130.00 Minimum acceptable hourly wages £3.51 £3.36 £3.50 Base: all respondents looking for work 1060 345 1441 Few people expected to be much better off if they got a job paying their minimum acceptable wage (14 per cent), and a similar proportion expected to be worse off (13 per cent). Most expected to be a little better off (40 per cent) or about the same as they were presently (27 per cent). Did ETU affect aspiration wages? There was no difference between ETU Scheme areas and control areas overall, but respondents in some areas (for example, Sunderland at £3.16 per hour) gave lower minimum wages than did those in other areas (such 68 as Southend at £4.25 per hour). Overall, there was no indication that ETU had suppressed aspiration wages for unemployed people in the individual Scheme A and B areas compared with their control areas (Table 4.27). However, the lack of difference in aspiration wages between respondents in pilot and control areas is perhaps not surprising considering that awareness of ETU was not high among this sample (see Chapter 6 for further details). Table 4.28 considers the target and minimum weekly wages of jobseekers in the pilot areas who knew about ETU1 compared with those who did not. People who knew about ETU had lower target and minimum wages and these differences were statistically significant. There was no significant difference in the number of hours they expected to work for these wages (the average for both groups was 37) which suggests that if ETU was influencing their aspiration wages it was not doing so by encouraging them to work shorter hours. Table 4.27 Median target and minimum wages by ETU area Target Target Minimum Minimum Base weekly hourly weekly hourly Newcastle (A) Sunderland (B) 137 114 £150.00 £150.00 £3.95 £3.75 £120.00 £120.00 £3.27 £3.16 Middlesborough (C) 142 £150.00 £3.75 £113.00 £3.29 Castleford (A) Doncaster (B) 127 105 £150.00 £150.00 £3.73 £4.00 £120.00 £120.00 £3.33 £3.33 Rotherham (C) 118 £150.00 £3.85 £100.00 £3.33 Southend (A) Bournemouth (B) 125 105 £200.00 £180.00 £5.00 £4.86 £160.00 £150.00 £4.25 £3.85 Southampton (C) 89 £150.00 £4.54 £140.00 £4.00 North Wales (A) Perth (B) 137 127 £160.00 £150.00 £4.05 £3.75 £136.50 £122.50 £3.50 £3.40 South Wales (C) 115 £150.00 £4.00 £130.00 £3.33 All 1441 £140.00 £4.00 £130.00 £3.50 1 See Chapter 6 for a full discussion on the questions relating to awareness of ETU. In this section, knowledge of ETU is based on people having been able to name ETU when a description of the benefit was put to them. These people appear to have had a more detailed knowledge of ETU than did those who said they had heard of the benefit but could not name it. Aspiration wages for this latter group (with less detailed knowledge) were also lower than for those who had not heard of the benefit at all but the difference was not statistically significant. 69 Table 4.28 Average expected and minimum weekly wages for jobseekers in pilot areas by awareness of ETU Unemployed survey 1998 Aware of ETU Not aware of ETU All £153.99 £150.00 £169.49 £160.00 £166.89 £160.00 £127.63 £120.00 £137.64 £130.00 £135.91 £130.00 164 813 977 Target weekly wages - Mean - Median Minimum acceptable weekly wages - Mean - Median Base: all respondents looking for work at first interview in pilot areas Note: awareness of ETU was determined by people being able to name ETU as the benefit described to them (see Chapter 6). Mean values are presented here along with median values as statistical tests were conducted to see if the difference in means was statistically significant. The difference in means between the two groups was statistically significant at 95 per cent level. Multivariate analysis found that the effect of awareness of ETU on aspiration wages was not statistically significant after controlling for a range of personal characteristics (area, gender, age, marital status and educational qualifications). However, these personal characteristics were also found to be related to the likelihood of a person being aware of the benefit (see Chapter 6 and Appendix B) and so it may be difficult to disentangle these effects. In summary, while there was no strong evidence that ETU suppressed aspiration wages for this sample it is possible that any effect cannot easily be measured in the context of low awareness of the benefit. Expected wages at second interview As discussed in Chapter 1, the National Minimum Wage was introduced in between the first and second interview for this sample and so it would be expected that jobseekers would have increased the level of wages they were looking for accordingly. This was indeed the case as the average lowest wages jobseekers would accept at second interview were £3.75 per hour. Again, women would accept lower wages (£3.60 per hour) than men (£3.75 per hour). There was no difference between ETU areas and control areas. However, it would be expected that jobseekers would increase the wage levels sought over time anyway because of increases in the cost of living. The average increase of 25 pence an hour between first and second interview is only an increase of seven per cent. Also, it is possible that those looking for the lowest wages more easily found work and so were not looking at second interview. Chapter 5 discusses this in more detail. 70 4.4.4 Aspirations for the future What is the most likely thing to happen to you over the next couple of years? All respondents were asked to choose the most likely thing they thought would happen to them over the next couple of years (Table 4.29). Just over half (54 per cent) thought they would be working more than 16 hours per week and 77 per cent of these thought they would no longer be claiming benefit. Five per cent saw themselves working for fewer hours than this and 47 per cent of these still expected to claim benefits. Twenty nine per cent thought they would remain unemployed and most (86 per cent) expected to still be claiming benefit. Twelve per cent of people thought they would be doing something else (63 per cent of these claiming benefit). Of these, 13 per cent saw themselves going into fulltime education, 27 per cent thought they would retire and three per cent thought they would be looking after home and family. More respondents in the 1998 survey thought they would be unemployed (29 per cent compared with 22 per cent in 1996) and claiming benefit (44 per cent compared with 36 per cent in 1996). Table 4.29 What is the most likely thing to happen to you over the next couple of years? Column percentages Unemployed survey 1996 Unemployed survey 1998 Working 16+ hours per week Working <16 hours per week Unemployed 60 4 22 54 5 29 Something else 13 12 Claiming benefit Not claiming benefit 36 58 44 50 Don’t know 6 6 1940 2120 Base: all respondents People in Scheme A areas were more likely to see themselves working in the near future (57 per cent) than in Scheme B or control areas (52 per cent) (Table 4.30). Despite the similar expectations between Scheme B and control area respondents towards work, more people living in Scheme B areas thought they would be claiming benefit in the future (48 per cent compared with 42 per cent in control areas). 71 Table 4.30 Most likely thing to happen over the next couple of years - by ETU type Column percentages Unemployed survey 1998 Scheme A Scheme B Control areas Working 16+ hours per week 57 52 52 Working <16 hours per week Unemployed 5 27 5 29 5 31 Something else 11 14 12 Claiming benefit Not claiming benefit 43 50 48 46 42 53 Don’t know 7 6 5 745 692 683 Base: all respondents Categorising people according to the benefit they were receiving when sampled, showed that the two types of respondents had very different expectations for the future (Table 4.31). Seventy per cent of JSA recipients expected to be working over the next couple of years and 60 per cent no longer expected to be claiming benefit. In contrast, just over a quarter of IS claimants thought they would be working and the majority (71 per cent) expected to remain on benefit. Table 4.31 Most likely thing to happen over the next couple of years - by benefit type at sampling Column percentages JSA IS All Working 16+ hours per week 65 22 54 Working <16 hours per week Unemployed 5 23 4 48 5 29 Something else 7 26 12 Claiming benefit Not claiming benefit 34 60 71 23 44 50 Don’t know 6 6 6 1586 599 2187 Base: all respondents Men were more likely to see themselves working in the near future (61 per cent) than were women (50 per cent). Similar proportions predicted unemployment though (28 per cent of men and 30 per cent of women). Differences were also apparent between single respondents and those with a partner (Table 4.32). Almost two-thirds of single people thought they would be working over the next few years in contrast to 43 per cent of respondents with a partner. Women remained less likely to think that they would be working but this was mainly because higher proportions thought they would be doing something else rather than being unemployed. 72 Table 4.32 Most likely thing to happen over the next couple of years - by gender and partnership status Cell percentages No partner Has partner Men Women All Men Women All Working 16+ hours per week Working <16 hours per week 66 2 50 8 61 4 43 3 30 10 37 6 Unemployed Something else 23 9 25 17 24 11 43 12 40 21 41 16 Claiming benefit 38 53 42 52 44 49 Not claiming benefit Don’t know 56 6 40 7 51 7 44 4 48 8 45 6 Base: all respondents 1062 397 1459 388 270 658 There was a stark divergence between younger and older respondents in their thoughts about the future (Figure 4.3). Eighty five per cent of under 25 year olds thought they would be working in the next couple of years and just eight per cent thought they would be unemployed. In contrast, over half of people aged 55 or older expected to be unemployed and only 23 per cent thought they would be working. Figure 4.3 Most likely situation over next couple of years by age group 4.5 Movements into work Section 4.4 illustrated that a substantial proportion of the sample were not looking for work at first interview (38 per cent of non-working respondents) and others were not optimistic about the possibility of working for 16 or more hours a week in the next few years (46 per cent). Section 4.2 has already shown that these expectations were accurate for many people as few respondents had moved into any form of work either by first (11 per cent) or second (18 per cent) interview. The small numbers of respondents in work at either interview limits what can be said about 73 movements into work in this chapter. However, more complex multivariate analysis in Chapter 5 does explore all exits from unemployment over the time period. 4.5.1 Looking for work at second interview At second interview, the work status of respondents followed a similar pattern as at first interview (Table 4.33). More people were in work but the proportion not looking for work was almost the same. Among those sampled as receiving JSA, only 11 per cent said they were not looking for work at second interview (around 18 months later), but among those sampled as claiming IS, more than half (53 per cent) were not looking for work. This adds further weight to the view that the two sample groups had very different long-term attachments to the labour market. Table 4.33 Work status at time of second interview by benefit type at sampling Column percentages JSA IS All In work 21 8 17 Waiting to start job Has looked for work in previous four weeks 1 54 * 12 1 43 Not looked for work in last four weeks but would like a job 13 27 17 Not looking for work 11 53 22 Base: all respondents at second interview 963 346 1309 The proportion looking for work had fallen between interviews (Table 4.33). Some of this group had moved into work (16 per cent) but a similar group were not working but were no longer looking for a job (18 per cent) (Table 4.34). Almost six in ten of those in work at first interview were in the same situation at second interview (58 per cent). Another 26 per cent were actively seeking work at second interview. The majority of those not looking for work at first interview were also not looking at second interview (86 per cent). Nine per cent were looking for a job and five per cent were in work. 74 Table 4.34 Work status at first and second interview Column percentages Work status at second interview Work status at first interview Not looking In work/waiting Looking but would Not looking to start job for work like job for work 16 66 10 5 Looking for work Not looking but would like job 58 26 7 Not looking for work 8 6 18 45 27 9 19 67 147 731 139 292 In work/waiting to start job Base 12 Base: all respondents at both interviews. Figures in bold refer to the percentages in each group at first interview who were in the same situation at second interview. 4.5.2 Workers at first interview Of the 10 per cent of respondents who were in paid employment by the time of interview, six out of ten had acquired a permanent job, 24 per cent had a temporary position and eight per cent were on a fixed term contract. Another eight per cent said they did not know how long their job would last. Thirteen per cent of respondents in ETU areas who were working at least 16 hours per week were claiming ETU (discussed further in Chapter 6). Younger people were more likely to have entered work (14 per cent of those aged under 25 but just eight per cent of people aged 45 or over), as were those with qualifications (14 per cent of respondents with academic and vocational qualifications compared with seven per cent of people without any). People with a mortgage (13 per cent) and those living with parents (12 per cent) were more commonly in work than were respondents living in rented accommodation (six per cent). There was little difference between ETU areas and control areas overall, but rates of work varied between individual areas, for example Southampton, Doncaster and Bournemouth had 15 per cent of respondents in work at interview compared with just six per cent in Southend and South Wales. Previous work experience was also associated with having moved into work by first interview. Those who were employed for 16 or more hours a week at first interview had spent an average of 39 per cent of their time since January 1993 (an average of 27 months) in this activity compared with 14 per cent (around 10 months) of those not working at first interview. Most of those working 16 hours or more hours a week were in fact working at least 30 hours (78 per cent) and median hours worked were 39 per week. Average wages were £3.53 per hour (Table 4.35), considerably lower than the average of £9.53 per hour earned by all employees in Great Britain in 1998 (New Earnings Survey). 75 4.5.3 Workers at second interview Sixteen per cent of respondents were in paid employment by the time of interview. Most of these (72 per cent) had acquired a permanent job, 17 per cent had a temporary position and seven per cent were on a fixed term contract. Another four per cent said they did not know how long their job would last. Almost one-quarter of respondents in ETU areas who were working at least 16 hours per week were claiming ETU (23 per cent) (discussed further in Chapter 6). Average wages were low among the group of workers but above minimum wage levels. Again, most of those working 16 or more hours a week were employed for 30 or more hours (70 per cent) and median weekly hours were 37 (Table 4.35). Hourly wage levels appeared lower in both Scheme A and Scheme B areas compared with control areas but small numbers of respondents in work at second interview makes this result unreliable. Chapter 5 discusses this in more detail. Table 4.35 Average (median) wages and hours worked for respondents employed at 16 or more hours per week Average Average Average hours worked net weekly net hourly Base per week pay pay First interview Men 84 40 £140.50 £3.51 Women Scheme A 31 37 28 40 £89.75 £132.00 £3.60 £3.33 Scheme B Control areas 42 36 37 38 £111.00 £130.00 £3.60 £3.53 All 115 39 £124.00 £3.53 Second interview Men 103 38 £140.00 £4.00 Women Scheme A 45 53 35 37 £110.00 £128.00 £3.59 £3.60 Scheme B Control areas 45 51 35 37 £121.00 £140.00 £3.78 £4.03 All 148 37 £130.00 £3.78 Note: bases are too small to provide wage information on people who were self-employed or working for fewer than 16 hours per week. As at first interview, younger people were more likely to have entered work (26 per cent of those aged under 25 but just six per cent of people aged 55 or over) (Table 4.36). This was also the case for single people (15 per cent compared with eight per cent of those with a partner) and those with qualifications (22 per cent of respondents with qualifications at A level or higher compared with nine per cent of people without any). Respondents with a mortgage (16 per cent) and those living with parents (19 per cent) were more commonly in work than were those living in rented accommodation (nine per cent). There was little difference between ETU areas and control areas overall, but rates of work varied slightly between individual areas. 76 Again, those not in work at second interview had very little recent work experience as they had spent, on average, just 13 per cent of their time since January 1993 (around 10 months) in paid work of 16 or more hours a week. The equivalent figure for those employed for 16 or more hours a week at second interview was 31 per cent (25 months). Table 4.36 Percentage in work at second interview Row percentages Not in Work Work Self- Base work 16+ hours <16 hours Gender Male employed 859 83 14 1 2 Female 449 81 11 7 1 Partnership status No partner 893 81 15 2 2 Partner 416 86 8 4 2 Age group Under 25 209 69 26 2 3 25 – 34 35 – 44 228 212 80 86 18 10 * 1 1 2 45 – 54 55+ 343 316 86 88 9 6 4 4 1 2 Owns outright Mortgage 113 138 82 73 11 16 6 7 1 4 Tenant With parents 558 402 88 78 9 19 3 1 * 2 Other/missing 98 90 5 2 3 Housing tenure Qualifications None 611 86 9 4 1 Vocational only GCSE D-G 232 84 10 3 3 or equivalent GCSE A-C 190 81 16 2 1 or equivalent GCE A level 179 74 21 2 3 or higher 97 76 22 1 1 Driving licence Yes 544 78 16 3 3 No 764 86 10 3 1 Continued 77 Table 4.36 Continued Row percentages Not in Work Work Self- Base work 16+ hours <16 hours Scheme A Scheme B 463 410 82 85 13 12 3 2 2 1 Control areas 436 81 13 5 1 Area Newcastle 127 81 13 3 3 Castleford Southend 119 96 80 83 13 12 5 2 2 3 North Wales Sunderland 121 132 85 89 12 9 0 2 3 0 Doncaster Bournemouth 119 58 82 72 13 21 1 3 3 4 Perth Middlesbrough 101 113 89 78 9 16 2 5 0 1 Southampton South Wales 95 121 77 87 12 12 10 0 1 1 Rotherham 107 83 13 3 1 Base: all respondents 1309 83 13 3 1 employed ETU type 4.6 Summary 4.6.1 Movements into work This survey comprised people whose benefit claim had commenced between six and 15 months before they were sampled. However, few respondents had recently been in contact with the labour market. Forty four per cent of the sample said they had not worked in the previous five years and another 14 per cent said they had never had a paid job. Many of the movements into work occurred in the short gap between sampling and first interview which is probably because the likelihood of leaving unemployment falls as the length of benefit claim increases (Smith et al, 2000). In the average gap of 70 days, seven per cent were working for 16 or more hours per week, three per cent were working less than 16 hours per week and one per cent were self-employed. Six per cent were undergoing some form of training or education and 63 per cent said they were unemployed and claiming benefit. One in ten people said they had been sick or disabled for more than six months at time of first interview. By second interview, an average of 10 months later, 13 per cent of respondents were working 16 or more hours a week, three per cent were working less than 16 hours and two per cent were unemployed at second interview. Just under half said they were unemployed and claiming benefit (49 per cent) and 15 per cent said they had long-term ill health. Ten per cent were undertaking training or education. 78 At both interviews, young people and those with educational qualifications were more likely to have moved into work. By second interview, 26 per cent of respondents aged under 25 were working 16 or more hours a week as were 22 per cent of those with qualifications at A level or higher. Return-to-work wages were low at around £3.78 per hour for a 37-hour week. Although few people had moved into work, some of those who had were claiming ETU (13 per cent at first interview and 23 per cent at second). 4.6.2 Looking for work Of the non-working respondents at first interview, 38 per cent had not looked for work in the previous four weeks. However, one-third of these said they would have liked a job if a suitable one were available. Sixty per cent of these said they were not currently looking for work because of health problems, 10 per cent had caring responsibilities, and 11 per cent were undertaking further training or education. The majority of people not looking for work were IS claimants when sampled as only 18 per cent of non-working respondents who had been in receipt of JSA at sampling said they were not looking for work when interviewed. Male respondents, single people and those with educational qualifications were more likely to be looking for work. In contrast, people living with a partner, those aged 55 or over and those who reported a long-standing illness or disability were less likely to be seeking employment. 4.6.3 Aspirations for the future Most people looking for work were seeking employment (69 per cent) while 10 per cent particularly wanted self-employment and 21 per cent were prepared to consider either. One in ten people specifically wanted to work less than 16 hours per week. The minimum wages jobseekers were prepared to accept were low and around the level of the National Minimum Wage. There was no difference between ETU Scheme areas and control areas overall, but respondents in some areas (for example, Sunderland) gave lower minimum wages than those in other areas (such as Southend). However, there was no strong evidence that ETU had suppressed aspiration wages for unemployed people in the individual Scheme A and B areas compared with their control areas. Just over half of respondents (54 per cent) thought they would be working more than 16 hours per week over the next few years and 77 per cent of these believed they would no longer be claiming benefit. Almost one in three thought they would remain unemployed (29 per cent). Respondents sampled as receiving JSA were more optimistic about working in the future (65 per cent) than those sampled as receiving IS (22 per cent). Single people and younger respondents were also more likely to think that they would be working over the next couple of years. 79 5 5.1 Introduction LABOUR MARKET OUTCOMES In this chapter, attention turns to econometric estimation of labour market outcomes for the unemployed sample. Such an approach is needed if one is to isolate the effect of specific individual or household characteristics on the outcomes of interest. It also allows testing of the statistical significance of any patterns that emerge. As such, the content of this chapter should be viewed as a complement to the preceding descriptive analysis. The three outcomes that are considered are: • the movement away from unemployment; • the wages earned by those moving into work; and • the expected wages among those out of work. These outcomes are clearly inter-related. The first outcome is of primary interest since it indicates the success of ETU in bringing people into work. However, the process by which this is achieved is also relevant and can be examined by considering the other outcomes. The employment effects of ETU should have operated by providing a supplement to income that allowed individuals to lower their reservation wage – the wage below which they will choose not to work. With a lower reservation wage, a greater number of possible jobs can be considered. If ETU did have an effect, one would expect those leaving unemployment in the pilot areas to have found jobs with lower wages than those found in the control areas. Thus, the second outcome listed above explores the extent to which this happened. For those who did not find work, one can examine whether ETU had an effect by considering their wage expectations. Once again, being in a pilot area should have reduced the expected wages relative to the control areas if ETU operated as intended. This is the third outcome variable considered. In the remainder of this chapter, each of these outcomes is considered in turn. The analysis follows a similar format to that used previously in this evaluation (Finlayson et al, 2000) and considers only those individuals who responded to both the 1998 and 1999 interviews. While sample attrition over these two years could have introduced possible bias, the results in Appendix A suggest that this was unlikely to be a significant problem. A brief overview of the sample is provided in Section 5.2. There are a number of detailed differences from the earlier research on the 1996 and 1997 samples in the analytical approaches adopted and these are highlighted as they appear. However, the technical estimation details are not covered in the narrative that follows, although the model results appear in Appendix C. 81 5.2 Labour market transitions In this section the movement away from unemployment is considered. A different approach is used from that followed when examining the 1996 unemployed sample. In the earlier work, only employment at the time of the 1997 follow-up survey was used to provide information on the labour supply effects of ETU. With the newer data, the movement into work is examined using survival analysis. This aims to identify those factors that influenced if and when an unemployed individual moved to an alternative economic status. By 1999, sufficient time had elapsed since the introduction of ETU that such an approach was possible. This different approach means that the results presented in this section are not intended to be directly comparable with those based on the earlier surveys. For example, whereas the earlier analysis presented findings on those factors associated with being inactive at both the 1996 survey and the follow-up 1997 survey, the analysis presented below focuses on the destination of the unemployed and therefore does not consider those individuals who were observed initially as being inactive. However, spells of unemployment that ended with a move to inactivity feature in the analysis. Another source of difference is in the definitions adopted. With the earlier survey, inactivity was defined by making use of a job search variable. The survival analysis uses ‘work history’ data in which inactivity is based on individuals’ reports of their employment status each month. Retrospective information on job search is unreliable and so was not used for this purpose. The precise composition of the groups considered is given below. 5.2.1 An overview of the sample There were 2187 respondents to the survey in 1998 of whom 60 per cent participated in the follow-up survey in 1999, giving a usable sample of 1309. The initial sample was taken from those with length of benefit claim of between six and 15 months at the point of sampling. Inevitably, there was some deviation from this target by the time of interview. For those interviewed in both 1998 and 1999, economic status is shown in Table 5.1. The majority (two-thirds) of respondents were unemployed and claiming benefits in 1998. This total is further boosted if one includes non-claimant unemployed. The other most significant categories are those who had moved into employment (mostly full-time) or self-employment and those who were permanently sick or disabled. Although the analysis that follows focuses on spells of unemployment and excludes spells of an alternative economic status, an individual who was not unemployed at the time of the 1998 interview may still feature in the analysis if a period of unemployment occurred after this point. Likewise, unemployed spells between the start of ETU and 1998 are included. 82 Table 5.1 Employment status of the 1999 re-interviewees Column percentages Employment status 1998 1999 Employee, 16 hrs. a week or more 6 13 Employee, less than 16 hrs. a week Self-employed 3 1 3 2 On a New Deal work placement On a government training or 3 2 education programme Unemployed and claiming benefits 2 64 1 49 Unemployed and not claiming benefit In full-time education 7 6 (not government training or education) Temporary sick/disabled 1 1 (less than 6 months) Permanently sick/disabled 1 2 (6 months or more) Looking after the home/children 9 1 15 3 Retired Other 1 1 3 1 1302 1308 Base Table 5.1 also indicates the extent to which employment status changed between the two interviews. Most significantly, there was a fall in the number of unemployed claimants, and this was mainly accounted for by equal-sized rises in employment and inactivity. It is revealing to investigate the nature of these changes in more detail. Table 5.2 cross-tabulates 1998 employment status with that of 1999. The numbers in the main section of the grid show the percentage moving from each 1998 category to each 1999 category. From this, one can see that the main exits from unemployed claimant status were to employment (mostly full-time) and inactivity in the form of permanent sickness. However, there were also some transitions from permanent sickness to claimant unemployment, although the smaller base size for the sick in 1998 means that the 14 per cent change translates into only 16 respondents in number. In fact, the permanently sick, as their naming suggests, were the least transitory group, along with the retired. The most transitory were New Deal work placements and government training and education programmes. This reflects the short-term nature of these interventions. None of those classified as temporary sick (less than six months) were similarly classified in 1999, for obvious reasons. 83 Table 5.2 Changes in employment status 1998-1999 for the 1999 re-interviewees Row percentages 1999 employment status 1998 employment status 1 2 3 4 5 6 7 8 9 10 11 12 13 Base FT employee (16+ hours) PT employee 1 51 1 4 1 0 25 8 1 0 5 1 4 0 85 (<16 hours) Self-employed 2 3 38 22 30 0 3 44 0 0 0 0 18 11 8 11 0 0 3 0 0 0 0 0 3 0 0 11 40 9 New Deal work placement 4 30 3 0 11 5 43 3 3 0 0 3 0 0 37 Government training/ education 5 23 0 0 4 8 62 4 0 0 0 0 0 0 26 Unemployed claimant 6 10 2 1 3 1 61 3 1 2 11 2 2 1 833 Unemployed non claimant 7 3 9 0 0 0 29 34 0 1 5 8 9 1 86 8 0 0 0 0 0 50 0 42 0 8 0 0 0 12 disabled 9 Permanently sick/ 0 0 0 0 0 47 7 0 0 47 0 0 0 15 disabled Looking after 10 4 0 2 0 1 14 3 0 2 70 3 2 0 113 home/children 11 Retired 12 0 0 8 0 0 0 0 0 0 0 15 6 8 12 0 0 8 0 8 6 46 6 0 71 8 0 13 17 Other 0 7 7 0 7 40 7 0 0 13 7 0 13 15 13 3 2 3 1 49 6 1 2 15 3 3 1 1301 FT education Temporary sick/ Total 13 Note: some categories have very small bases. Figures in bold indicate the percentage of people in each activity in 1998 who were in the same activity in 1999. 5.2.2 Modelling the move away from unemployment Analytical issues As mentioned above, survival analysis has been used to model the movement away from unemployment. If we were uninterested in the nature of the economic status that ended the unemployment spell, the problem could be approached using a logistic regression model. This is a standard approach that is routinely followed. However, since there is interest in whether the unemployment spell ends in employment or some other event, the problem becomes more complicated and the modelling approach has to account for multiple outcomes. To cope with this, the move away from unemployment is analysed using a multinomial logit model. This is a generalisation of the logit (or logistic regression model) which allows for more than one possible outcome. In fact, four employment states are identified: • unemployed: unemployed claimants and unemployed non-claimants; • employed: full-time (16+ hours) and part-time (less than 16 hours) employed and self-employed; • training: New Deal work placement, government training, education; • inactive: sick/disabled, looking after the home/children, retired, other. 84 Information was collected in the survey on all changes in employment status since 1993. From this, a monthly employment history was constructed and it is this that has been used to model the movements away from unemployment. Only employment status since October 1996 (the date of ETU introduction) has been considered. This results in a possible maximum of just over three years of employment records for those respondents interviewed towards the end of 1999. Information on all unemployment spells within this period has been included in the analysis. The modelling approach used allows for the possibility that an individual may have moved from unemployment to another employment status and then back to unemployment. By taking account of this, it is possible to examine whether survival in unemployment differs with subsequent unemployment spells. In constructing the employment histories, those spells of unemployment that had persisted for more than two years when first observed were excluded from the analysis. This is essentially an arbitrary cut-off point and was imposed to allow the focus of the analysis to be on those for whom unemployment was a relatively recent experience, although two years is, of course, a considerable claim period. Simple inspection of the transition rates into work indicates that concentrating on the less established claims increases the likelihood of observing a spell that ends in employment: only one tenth of those spells of more than two years’ duration ended in work compared with about one-third of other spells. In fact, the exits from unemployment in the final sample can be broken down as follows: 30 per cent found work, 14 per cent went into training and four per cent became economically inactive. The outstanding 52 per cent remained unemployed. On the basis of these exit rates, one might expect the results relating to moves into employment to be clearer than those relating to training and inactivity. As shown below, this is indeed the case, although there are additional reasons why this is unsurprising. The model performed reasonably well in terms of providing a range of significant and plausible results for the move into employment. However, the results for the other two categories were not as satisfactory. In particular, no variables were found to be significant in explaining the move from unemployment into inactivity. While this is disappointing, it is important to bear in mind that the significance of variables in the model is relative to the reference category (in this case the unemployed). Hence, the lack of significant variables influencing inactivity can be viewed as indicating that the determinants of inactivity were similar to those of unemployment. This is not an implausible interpretation. The detailed estimation results are presented in Appendix A. 85 An ETU effect? The main substantive interest is in the effect of ETU on the movement from unemployment. No such effect was evident. This was true for both Scheme A and Scheme B areas, and for all outcomes. Hence, the strong inference is that ETU had no effect on reducing unemployment by whatever means. While this is a fairly negative result, it is in keeping with the other findings reported so far on the effects of ETU. Specifically, research based on those in 1996 and 1997 showed little effect of the benefit either for those in work or those unemployed. This was matched with low awareness and limited take-up among those eligible. It is nevertheless interesting to consider the other results from the model. The first noteworthy point is that gender appeared to play no role in explaining movements into work or inactivity. This is an unusual finding since most labour market models divide along gender lines. However, the sample excludes those with children, so it is perhaps not surprising to see little difference between the sexes in terms of their movement away from work since any such difference is likely to be driven by fertility factors. Women were, however, less likely to exit unemployment to training. Age was found to have an important influence on movements into both work and training. In both cases, the effect was negative. This means that older unemployed people were less likely to find work and less likely to move into one of the categories representing training. The problems facing older workers in the labour market are well known. Similarly, training was more likely for the younger unemployed, particularly since this included participation in the New Deal for young people. Having a partner reduced the likelihood of finding work. However, this was offset when the partner was working to the extent that having a working partner increased the probability of moving into employment. This is a well-recognised phenomenon and is associated with the polarisation of households into those with two earners and those with no earners. Having a partner also exerted a strong negative effect on the likelihood of moving onto training. Those with vocational qualifications were more likely to move into training. They also appeared more likely to move into employment, although this was not statistically significant. Being able to drive increased the chances of finding work. There were strong effects associated with accommodation type. Those who owned their property outright were more likely to find work. This is possibly due to their lack of housing costs resulting in a readiness to accept lower wages. The reference category was those who rented but received full Housing Benefit when unemployed. People in this situation 86 may have had a disincentive to work if their HB was reduced as a result of their finding work. Those making a non-zero rent payment were more likely to work. Finally, two variables were included in the model to capture the change over time in the probability of leaving unemployment. The first related to the length of the unemployment spell. This showed a negative effect on finding work. This indicates that the longer one was unemployed, the more difficult it was to find work. Offsetting this, however, was the general tendency for individuals to find work as time passes. This was a tendency that existed independently of the labour market characteristics of the individual. If one is willing to interpret this as a macro effect, one would expect it to be positive since it reflects the generally improving economic conditions during the period covered by the analysis. Figure 5.1 summarizes the movement away from unemployment for an ‘average’ individual in the sample. This shows the probability of having moved to one of the alternative economic statuses as time passed. The graph covers the period from October 1996 to the end of 1999. This type of graph is often referred to as a ‘survival curve’ because it represents what proportion survive in the initial state (here, unemployment) as time progresses. This shows that that unemployment spells were most likely to be ended by finding work. However, this is a slow process with only 30 per cent finding work within two years. Figure 5.1 Movements away from unemployment, by destination 87 Allowing for characteristics to vary over time The analysis presented so far was based on all unemployment spells since the introduction of ETU in October 1996. As retrospective information was only available on a month-by-month basis for employment status, a necessary assumption was that characteristics as they existed at the time of the 1998 interview could be made to apply to all periods considered. This is fairly innocuous in the case of characteristics which will have remained fixed (such as gender) or which changed systematically (such as age) but is less defensible for other variables. In view of this, it is instructive to consider just those unemployment months since the 1998 interview. This also allows additional variables to be included in the model which were previously excluded on the grounds that they were likely to exhibit too much change over time. Re-estimating the model specification already discussed on this reduced timespan gave essentially similar results to those already discussed. This provides some assurance as to the robustness of the findings presented earlier. As a general comment, the level of significance of the findings was lower as one would expect since the estimates were based on a smaller number of observations. The effect of including a number of new variables was then investigated. The variables considered were: • the minimum acceptable wage in 1998; • awareness of ETU in 1998; • variables related to health in 1998. Given the reduced significance associated with the smaller sample, only the new variables are considered in the discussion that follows, although the full model results are given in Appendix C. Overall, 17 per cent of unemployment spells ended with employment, seven per cent with a move into training and three per cent with a move into inactivity. The results relating to the new variables are as follows. The minimum acceptable wage exerted a significantly negative effect on the probability of moving from unemployment into work. This variable can be interpreted as representing the reservation wage (although see below for further discussion of this point) and therefore its effect was as expected: those who were prepared to accept jobs with lower wages moved into employment more quickly. For movements into training or inactivity, it was insignificant. Awareness of ETU in 1998 appeared to play no independent role in determining exits from unemployment. This was true regardless of whether one considered exits to employment, training or inactivity. Two measures of awareness were tested and neither was found to have a significant influence on movements out of unemployment. The first 88 was based on those who said ‘yes’ to the question: ‘A new social security benefit was introduced in 1996 in some areas of the country that pays extra money to some people who work and who have no dependent children living with them. Have you heard of the introduction of this benefit?’ The second was more specific and was based upon those who said ‘yes’ to the question and could name the benefit as ETU.2 However, it is not easy to measure understanding and awareness of a benefit in a survey and these difficulties may be reflected in this result. Finally, a range of variables relating to the health of the respondent were included in the specification, but they did not prove to be significant in explaining exits from unemployment. In particular, one might expect them to have provided some explanation of moves into inactivity. However, a variable indicating whether there was anyone the respondent cared for ‘because they have a long-standing illness, disability or infirmity of any kind’ did approach statistical significance. This suggests that carers may have been more likely to move into inactivity. 5.3 Wages Analytical issues As with the analysis of the 1996-97 unemployed sample, a practical problem when analysing wages is that only a small number of respondents were in work at the time of interview. Considering those working at the time of the 1999 interview yields a sample of only 165, and this reduces to 144 once missing values on wages are accounted for. This restricts the range of analyses that can be carried out. However, unlike the 1996 and 1997 samples, there was no theoretical objection to boosting this sample by including all those working more than 16 hours a week at any time since the first (1998) sample since all jobs fall within the period of ETU. Following this approach increases the useable sample size to 292. This was not an option in the earlier analysis since, in this case, the first (1996) sample pre-dated the introduction of ETU and so to pool wages information across the two samples would have distorted the estimated ETU effect. As is standard in the estimation of wages models, the estimation procedure followed takes explicit account of the selection into the sub-sample observed to be working. This is necessary since this sub-sample is no longer a random sample of those unemployed in 1998. Rather, the subsample is made up of those people who were looking for work and for whom their reservation wage was exceeded by a wage offer that they have received. In view of their non-representativeness, estimation results based on this sub-sample cannot be extended to the sample as a whole. However, by accounting for the selection mechanism (i.e. the probability of working) in the model it is possible to correct for this and thereby generate general results. The results relating to this selection into 2 See Chapter 6 for further discussion on awareness of ETU among this sample. 89 employment are not discussed in this section, since the movements away from unemployment have been considered more fully in Section 5.2. The effect of ETU in this section was assumed to arise simply from living in one of the pilot areas. No account is taken of whether those in the pilot areas participated in ETU, or were aware of ETU. The only distinction is between living in a Scheme A area or a Scheme B area. This is a similar approach to that taken in the previous analysis and any (regression-adjusted) differences detected between the pilot and control areas incorporates both the direct effects of the benefit and any indirect effects that might have arisen. Previous findings Before presenting the results of the analysis, it is useful to briefly summarise the results of the previous analysis using the 1996-97 unemployed sample. While the overall finding was that there was no significant effect of ETU on wages, there were some indications that in the Scheme B areas there was a tendency for wages to be lower than in the control areas. Given the novelty of ETU at that point, the statistical significance of this effect was interpreted generously, since one might expect the benefit to have not had a detectable influence by that point. Rather, the findings were viewed as indicative. By the time of the 1998 survey, the benefit was more established in the pilot areas (although take-up remained low) and the results must be interpreted accordingly. Findings The results appear to confirm the suggestions of the previous research (see Appendix C for detailed results). People living in a pilot area had a lower level of wages, once other factors were taken into account. The difference was greater in the Scheme B areas than in the Scheme A areas, as one would expect given the more generous provision of the benefit in the former. Moreover, these differences were statistically significant. This finding appears robust. Removing those individuals for whom the predicted wage differed from the actual wage by the most (i.e. trimming the top and bottom one per cent from this distribution of errors) and reestimating did little to alter the substantive finding, and in fact increased both the measured ETU effect and its significance. These effects appeared large. The average worker in a Scheme B area was predicted to have a wage some 19 per cent beneath that of the average worker in the control area. The corresponding proportion for those in the Scheme A areas was 14 per cent. These differentials are similar to what would be expected from a superficial inspection of earnings within the three types of areas. The econometric evidence shows that the differences persisted (and, in fact, increased) after controlling for other factors. 90 Some of the other variables in the wage equation deserve mention. Being female had a negative influence on wages. While it would have been more revealing to model the wages of males and females separately, this was not possible given the small number of individuals (particularly women) observed to be in work. Age had a non-linear effect with older workers attracting higher wages until they reached their early forties, after which wages fell. It should be borne in mind that the restriction of the sample to those without children results in a bipolar age distribution with relatively few people (43 per cent) over the age of 25 and under the age of 50. Qualifications were important, with vocational and academic qualifications seemingly having an equivalent positive effect on wage levels. The remaining characteristics were of little statistical significance, although there appeared to have been some wage premium attached to those who in 1998 were craft workers. A final consideration is that of training. This provides some indication of the ‘quality’ of a job. Individuals may be more willing to accept lower wages if they are being provided with training. To investigate this, the provision of training was modelled. The results (see Appendix C) show this not to be the case. Those in the pilot areas were no more likely to receive training than were those in the control areas. 5.4 Expected wages The results of Section 5.3 suggest that ETU had a negative effect on wages for those leaving unemployment. The role of ETU was to supplement earnings to the extent that the individual was able to consider lower paid work than would otherwise be possible. The lower wages in the pilot areas points to some success on this front. To investigate the process further, one can focus specifically on expected wages, which we interpret as a proxy for the reservation wage of those seeking work. Reservation wages The focus now is on hourly wages. Since ETU eligibility relates to weekly wages, this was the more appropriate measure when considering the wages of those in work. However, the motivation for examining expected wages is the insight that it provides into the jobsearch process. The reservation wages are better observed as an hourly rate since this provides a more accurate measure of the price of labour. The practical reason for this is that the definition of how many hours there are in a working week varies across individuals. The basis for this further investigation is the respondents’ answers to the question: ‘what is the lowest weekly amount in take-home pay that you would ever accept?’ A supplementary question (‘how many hours each week do you think you would be working for that money?’) allows the hourly wage rate to be calculated. This derived wage will be referred to as the expected wage. 91 The expected wage provides an insight into the reservation wage. However, it is important to be aware of the limitations of such an insight. While the stated minimum acceptable wage may be similar to the reservation wage for those individuals with a good knowledge of their value in the labour market, it is likely to be less accurate for those people who are less informed. The similarity between expected and reservation wage is likely to be closer for those individuals who are more actively involved in jobsearch. This is because a characteristic of the jobsearch process is the revision of reservation wages in the light of increased knowledge of the labour market and the opportunities available. As an illustration of the extent to which the expected wage can differ from the reservation wage, of the 165 working at the time of the 1999 interview, 20 were working at a lower wage than they had claimed in the 1998 interview was the minimum they would accept. An ETU effect? On the face of it, there was little evidence that ETU had the anticipated effect on expected wages. Table 5.3 shows the expected wage across all types of area. If anything, those in the Scheme A areas appeared to have had a higher expected wage than those in the control areas. This is a similar result to that found in 1996-97. However, unlike 1996-97, there was a general increase of roughly six per cent in expected wages across the three types of area. This increase was common to both men and women when aggregated across the three areas. Although Table 5.3 shows a larger increase for women in Scheme A areas, small cell sizes mean that these figures must be interpreted with caution. This is a different finding from 1996-97 where wage expectations were shown to fall in the pilot areas relative to the control areas. However, these earlier years spanned the introduction of ETU, so the change over time would reflect the impact of programme introduction. This is not the case with the 1998-99 samples, since ETU would, in theory, have had an effect on expected wages in both years rather than just the latter year. 92 Table 5.3 Expected hourly wage Row percentages Type of area Expected hourly wage (£) Men £ Scheme A Women N £ N All £ Men Change 1998/99 Women All Men Women All N £ N £ N £ N % % % 3.60 210 3.43 69 3.56 279 3.78 185 3.87 55 3.80 240 4.8 11.4 6.3 Scheme B 3.49 Control areas 3.51 191 189 3.40 3.45 56 73 3.47 3.49 247 262 3.72 3.74 177 156 3.68 3.62 42 46 3.72 3.71 219 202 6.2 6.1 7.6 4.7 6.7 5.9 Analytical issues To proceed further requires multivariate techniques. Using regression analysis, one can control for characteristics that vary across individuals to allow sharper focus on the unique effect of ETU. The approach taken was to pool the 1998 and 1999 expected hourly wage and examine them jointly. Since the increase in expected wages was similar across the pilot and control areas, the inclusion of a simple dummy variable indicating whether expected wages relate to 1998 or 1999 can control for differences over time. Information on expected wages was only collected for those respondents who were out of work. To focus on the role of reservation wages in the jobsearch process, only the expected wages of those who were unemployed are considered. For those unemployed at the time of both interviews, expected wages in 1999 were used rather than those in 1998. This was essentially an arbitrary decision. Those individuals who did not provide details of their expected wages when unemployed were dropped from the analysis. As with the examination of actual wages among those who had found work, those who have not found work represent a non-random sub-sample of the population. Modelling techniques similar to those used in the analysis of wages are incorporated to allow for this. Specifically, the estimation of interest is preceded by an estimation of the probability of being observed in the sub-sample. These preliminary results are used to correct for selection into the sub-sample in the main model. Again, only the results of primary interest are discussed here. The estimation results are presented in Appendix C. The selection adjustment term was positive and very significant, indicating that a simple regression model would have provided estimates that were biased upwards. Findings The main result is that ETU appeared to have no effect on expected wages. If one is prepared to restrict the possible effect to being negative, the effect in Scheme A areas approached statistical significance but remains unconvincing, particularly since one would expect a more noticeable effect in the Scheme B areas. The other characteristics appear plausible. 93 The higher expected wage in 1999 than in 1998 was captured by the dummy variable indicating the year. Women expected wages that were, on average, 32 pence an hour lower than those of men. Partnership status did not play a role, nor did the employment status of the partner, where relevant. Older individuals expected higher wages but, as with actual wages, this relationship was non-linear: beyond the age of 45 years, expected wages began to drop. Those living with their parents were also likely to expect lower wages, by about 50 pence an hour. Those with qualifications at degree level expected wages some 70 pence an hour higher than average and, while vocational qualifications were not found to be significant in altering expected wages, being a driver was associated with an increase of 30 pence an hour. Finally, those who had spent a larger proportion of the time since 1993 in employment expected a lower wage. This only amounted to six pence an hour on average, but its statistical significance makes it an intriguing result. It is perhaps indicative of a greater realism on the part of those who were better acquainted with the labour market. The effects of the remaining characteristics (socioeconomic group and type of area) were not significant. 5.5 Summary and discussion The main findings of the multivariate analysis can be summarised as follows: • those living in an ETU pilot area were no more likely to enter work than those in the control areas; • those in the pilot areas who found work since their 1998 interview were more likely to earn lower wages than those in the control areas; • those with lower expected wages were more likely to find work, but expected wages were not influenced by living in an ETU area. Therefore, people with lower wage expectations were more likely to find work. As ETU seemed ineffective in reducing expected wages, it was unsurprising to find that ETU was similarly ineffective in helping the unemployed back to work. However, it is important to bear in mind the distinction made earlier between expected wages and reservation wages. The concept of expected wages is, to some extent, notional and the level of these expected wages may get revised during the jobsearch process, particularly if people become aware of in-work benefits while looking for work. One the other hand, the reservation wage (or minimum wage that someone will accept) exists as a fixed constraint at the point of job offer. As there may not be an exact match between expected and reservation wages, the result that ETU had no effect on expected wages cannot be interpreted as evidence that it had no effect on reservation wages, particularly in the context of low overall awareness of ETU (see Chapter 6). The lower levels of wages among those who had found work in the pilot areas might indicate some effect of ETU on the reservation wage. However, this effect, if it existed at all, was not translating into an increased rate of job entry. 94 6 6.1 Introduction THE EXPERIENCE OF EARNINGS TOP-UP Earnings Top-up had been available for around two years in the pilot areas by the time survey respondents were first interviewed. This chapter first discusses the knowledge and awareness of the benefit of people in ETU areas (Section 6.2). Next, the experiences of the small group of people who had received ETU are considered in Section 6.3. Finally, all respondents who had not received ETU were asked to think about whether they would want their earnings from low-paid work ‘topped up’ with benefits if such a scenario was possible (Section 6.4). By the time of second interview, the pilot of ETU was coming to the end of its three-year period. The chapter briefly considers whether experiences of ETU had changed in the year between interviews. It may have been possible that the first interview itself triggered respondents to think about a benefit that topped up earnings (see Sections 6.2 and 6.4) that may be reflected in the responses given at second interview. 6.2 Awareness of ETU All respondents living in areas where Earnings Top-up was available were asked the question: ‘A new social security benefit was introduced in 1996 in some areas of the country that pays extra money to some people who work and have no dependent children living with them. Have you heard of the introduction of this benefit?’ People who said that they had heard of the introduction of this benefit were then asked: ‘What is this benefit called?’ 6.2.1 Awareness of ETU at first interview Almost two years after its introduction, awareness of ETU was low among the sample as only 29 per cent of respondents at first interview said that they had heard of the introduction of the benefit. Just over half of these (16 per cent of all respondents) were able to name the benefit as ETU (Table 6.1). Thirty eight per cent of people who said they had heard of the introduction of the benefit did not know what it was called and nine per cent gave a different name from ETU. People who had been receiving JSA were more likely to name ETU (18 per cent) than were IS claimants (eight per cent). Area differences Overall, there was little difference in awareness between Scheme A and Scheme B areas, but respondents in some individual areas had greater levels of awareness than others. Recalled knowledge of the benefit was highest in the Castleford and Barnsley area (39 per cent) but only half of 95 these were able to name ETU. One-third of respondents in Sunderland had heard of the introduction of the benefit and one-quarter named ETU. The lowest level of awareness was in Southend where one-fifth of respondents said they had heard of the benefit and just seven per cent named ETU. Below average levels of awareness were also found in Perth and North Wales. Levels of awareness tended to follow the pattern of ETU awards, outlined in Chapter 1, in that areas with the most ETU claimants (Sunderland, Newcastle, and Doncaster) also tended to be the areas where survey respondents had the best knowledge of the benefit. Table 6.1 Awareness of ETU at first interview Cell percentages Knows benefit Base is called ETU Scheme A areas 765 29 14 Scheme B areas 717 30 17 Area Newcastle upon Tyne (A) 186 31 15 Sunderland (B) Barnsley, Castleford, Pontefract, 180 33 25 Wakefield and Dewsbury (A) Doncaster (B) 192 186 39 31 20 20 Southend (A) Bournemouth (B) 186 177 20 32 7 14 North Wales (A) Perth and Crief, Dumbarton, 201 24 12 Stirling (B) 174 23 9 Gender Men 1035 30 15 Women 444 28 17 Partnership status Has partner 1050 31 17 No Partner 432 25 11 Benefit type at sampling JSA 1066 33 18 IS 404 21 8 Work status at first interview In work at interview 149 36 26 Unemployed and claiming benefit 932 31 15 Unemployed but not claiming benefit 75 21 7 Training course Ill health/disability 92 168 31 21 22 8 Other 66 21 11 1482 29 16 All respondents in ETU areas 96 Has heard of benefit Work status People who were working when interviewed were more likely to have heard of ETU but some of these would have been claiming ETU. Excluding people in ETU jobs, 29 per cent of those in work of at least 16 hours a week said they had heard of the benefit and 16 per cent named ETU, similar to recalled knowledge among the unemployed group. Respondents who defined themselves at interview as unemployed and not claiming benefit or as sick or disabled had the lowest levels of awareness of ETU. Age group People aged 25 to 34 had the highest level of recalled knowledge: 36 per cent said they had heard of the benefit and 21 per cent could correctly name it (Figure 6.1). The same proportion of under 25s were able to name ETU compared with just eight per cent of respondents aged 55 or over. Figure 6.1 Awareness of ETU by age group 97 Figure 6.2 Awareness of ETU by household type Marital status Respondents with a partner were less likely than were single people to recall ETU. Less than a quarter of respondents who lived with only a partner (24 per cent) said they had heard of the benefit and just one in ten could name ETU. Around a third of people living alone or with parents said they had heard of the benefit and 18 per cent named ETU (Figure 6.2). Again, this fits with the pattern of ETU awards outlined in Chapter 1 where ETU claimants were typically single and aged under 25. Educational qualifications Recalled knowledge was greater in respondents with educational qualifications (Figure 6.3), particularly those with academic qualifications. It was also slightly higher in people living in households that had previously received Family Credit (35 per cent said they had heard of the benefit and 21 per cent could name ETU, compared with 28 per cent and 15 per cent respectively in other households). 98 Figure 6.3 Awareness of ETU by qualification type After controlling for age and area in a logistic regression model, qualifications and experience of Family Credit had no significant influence on the probability of being able to name ETU. Respondents with a partner were only about 60 per cent as likely to be aware of ETU as those without a partner and people who defined themselves as long term sick or disabled at interview were around half as likely to have heard of ETU as others. Respondents in paid work at time of interview (but not receiving ETU) were found to be 52 per cent more likely than others to be aware of ETU (Table B.1 in Appendix B). Source of knowledge of ETU Respondents who said they had heard of the introduction of the benefit were asked where they had heard of it. The most common answers related to official sources. One-fifth of people said it had been recommended at the Jobcentre or ES office and 32 per cent said it had been mentioned there (Table 6.2). Nine per cent said they had seen a publicity display in the Jobcentre. 99 Table 6.2 Where did you hear about the introduction of ETU? Cell percentages Source Percentage Official sources Mentioned at ES office/ Jobcentre Recommended at ES office/ Jobcentre 32 20 Publicity display in Jobcentre Letter from DSS 9 7 Leaflet through door Other 3 3 Other people Friends/neighbours Relatives 14 4 Employer Workmates 1 1 Citizens Advice Bureau * News Item Article in newspaper/magazine 4 Item on TV Item on radio 2 1 Advertisements TV adverts Newspaper/magazine adverts 7 5 Radio adverts Other advert 1 2 Base: respondents in ETU areas who had heard of the benefit 430 Note: individuals could give more than one answer and so be counted several times. * Less than 0.5 per cent. Fourteen per cent of people said they heard from friends and four per cent cited relatives. Few people recalled advertisements about ETU: five per cent had seen newspaper or magazine advertisements. Seven per cent thought they remembered television advertisements although ETU was not advertised on television and was only advertised on radio in one of the pilot areas (North Wales). Advertising for ETU was stopped in April 1997, just six months after the benefit was introduced, and so it is not surprising that few people recalled having heard about ETU this way. 6.2.2 Awareness of ETU at second interview 100 Awareness of ETU was higher at second interview than at first (Table 6.3). Forty one per cent of respondents living in ETU areas said they had heard of the introduction of the benefit compared with 29 per cent at first interview. People who could name the benefit as ETU comprised 24 per cent of the sample at second interview but 16 per cent at first. The same patterns in awareness were apparent as at first interview. Knowledge was higher in areas where the volume of ETU claims was highest. It was also higher among young people and those with academic qualifications. Among those sampled as receiving IS levels of awareness were below average. To some extent, the overall increase in awareness at second interview would be expected as the first interview informed respondents that the benefit was called Earnings Top-up after the awareness questions. Table 6.3 Awareness of ETU at second interview Cell percentages Knows benefit Base Has heard of benefit is called ETU 458 392 38 45 18 31 126 132 40 43 21 27 Barnsley, Castleford, Pontefract, Wakefield and Dewsbury (A) 119 46 28 Doncaster (B) Southend (A) 119 92 44 28 32 8 Bournemouth (B) North Wales (A) 58 121 59 36 50 14 Perth and Crief, Dumbarton, Stirling (B) 83 39 20 565 285 41 41 23 26 621 229 44 34 26 18 In work at interview Unemployed and 142 51 36 claiming benefit Unemployed but 417 42 23 not claiming benefit Ill health/disability 52 137 42 32 21 18 Other 102 37 21 Continued Scheme A areas Scheme B areas Area Newcastle upon Tyne (A) Sunderland (B) Gender Men Women Benefit type at sampling JSA IS Work status at first interview 101 Table 6.3 Continued Cell percentages Knows benefit Base Has heard of benefit is called ETU 18 – 24 25 – 34 141 159 42 44 31 29 35 – 44 45 – 54 132 220 48 39 25 20 55+ 197 36 18 Household type Lives alone 279 43 27 Lives with partner only Lives with parents, no partner 179 249 37 42 21 28 Other 143 41 16 Qualifications None 389 38 19 Vocational only Academic only 147 138 38 50 24 33 Academic and vocational 176 45 29 All respondents in ETU areas 850 41 24 Age group 6.3 Experiences of ETU 6.3.1 Experience of ETU at first interview This section examines the experiences of respondents who had claimed ETU. Unfortunately, small numbers of people receiving ETU at the time of either interview limits the amount that can be said about the experience of receiving ETU for those previously unemployed in this sample. Number of ETU claimants There was an average gap of 70 days between the sample being selected and when they were interviewed. At time of interview, 24 people (1.6 per cent of the sample in ETU areas) said they were receiving ETU. A later question showed that it was the partner of the respondent who was receiving ETU in two of these cases. Another 32 people said they had previously received ETU and 11 people said they had applied for ETU but had been turned down. Seven respondents and three partners said they had applied for ETU at the time of interview and were waiting to hear whether they would receive it. Another 27 people were identified as being eligible for ETU but not claiming at interview – 29 per cent of all those in work but not claiming ETU at first interview. Of the 22 respondents who were receiving ETU, 14 gave their economic status at interview as employed for 16 or more hours per week. The others had changed activity by the time of interview. These fourteen 102 people who said they were working at least 16 hours per week whilst receiving ETU comprised 13 per cent of all respondents working for 16 or more hours per week in ETU areas at time of interview. Sources of information Of the 67 people who had ever applied for ETU, 58 per cent had heard about ETU from an ES office or Jobcentre and 15 per cent had heard from friends or neighbours. Most people had claimed ETU straight away: 36 per cent said they claimed as soon as they heard about ETU, 24 per cent claimed as they had just got a new job and 24 per cent said they claimed because they were earning less money. Most people (73 per cent) managed to answer all the questions for the claim form. Financial considerations Of those who had received ETU at some time (56 people), 59 per cent had received the amount they expected, nine per cent received more than expected, 14 per cent received less and 18 per cent said they did not know what to expect. Just two people said they lost other benefits that they had expected to get because of ETU. The majority of these people (86 per cent) said that working and claiming ETU was a better life for them than not working and claiming IS/JSA. 6.3.2 Experience of ETU at second interview Number of ETU claimants There was an average gap of ten months between the sample being selected and time of second interview. This gives a total average period between sampling and second interview of 15 months. There were 30 current ETU claimants at second interview (3.5 per cent of people in ETU areas). Another 34 respondents said they had previously received ETU and 12 said they had applied for ETU and been turned down. Of the current claimants, 26 of the 30 had not been receiving ETU at first interview. There were few people in work identified as eligible for ETU but not claiming (18 respondents) but they did make up 20 per cent of those in work in ETU areas but not claiming ETU at second interview. Another 16 per cent (14 respondents) would have been eligible for Scheme B ETU but lived in a Scheme A area. Numbers are too small to calculate a take-up rate for ETU for this sample but 22 respondents of the 40 calculated as eligible for ETU at time of second interview were receiving it. Of the 30 respondents who were receiving ETU, 25 gave their economic status at interview as employed for 16 or more hours per week and one said they were self-employed. Three defined themselves as currently unemployed and claiming benefit and one said they were on a New Deal work placement. The 25 people who said they were working at least 16 103 hours per week while receiving ETU comprised 23 per cent of all people working for 16 or more hours per week in ETU areas at time of interview. Sources of information Of the 71 people who had ever applied for ETU, 72 per cent had heard about ETU from an ES office or Jobcentre and 17 per cent had heard from friends or neighbours. Most people had claimed ETU (89 per cent) straight away. There were three main reasons for claiming: 24 per cent said they claimed as soon as they heard about ETU, 45 per cent claimed as they had just got a new job and 21 per cent said they claimed because they were earning less money. Most people (76 per cent) managed to answer all the questions for the claim form. Financial considerations Half of those who had received ETU at some time had received the amount they expected, 17 per cent received more and 22 per cent less than expected. Eleven per cent said they did not know what to expect. Only four people said they lost other benefits that they had expected to get because of ETU. The majority of people (26 out of 30) currently claiming ETU said that working and claiming ETU was a better life for them than not working and claiming IS/JSA. 6.4 Attitudes towards an ETU benefit Towards the end of the interview, respondents not claiming ETU were presented with the scenario: ‘Imagine you were able to get your earnings topped up with benefits while working 16 hours a week or more.’ Respondents were asked to consider whether they would take a lower paid job than they wanted and get their earnings topped up or hold out for a higher paid job. Overall, 59 per cent of respondents not in work at first interview said they would take a lower paid job and get an Earnings Top-up (Table 6.4). At second interview, the proportion was exactly the same suggesting that views had not changed significantly in the year between interviews. Certain groups were more likely to accept a wage top-up than others. People who had been receiving JSA were more agreeable to the idea of a wage supplement (64 per cent) than were IS claimants (47 per cent). Women were more likely to accept a top-up (63 per cent) as were respondents aged 25 to 34 (66 per cent). People with a partner tended to be more reluctant to accept a top-up (53 per cent) as were those who said they had a long-standing illness or disability (54 per cent) and respondents aged 55 or over (50 per cent). There was no clear relationship between acceptance of a wage top-up and housing tenure, educational 104 level or awareness of ETU. Of course, these differences may be as much a judgement on taking paid work at all as they are of the value of accepting a wage top-up. People in Scheme A areas were slightly more likely to say they would accept a top-up (62 per cent) than those living in Scheme B areas (57 per cent) or control areas (58 per cent). People who were working at least 16 hours a week at interview were asked if they would apply for a similar benefit to ETU (that topped up their earnings) if one was available. Two-thirds of those currently working at least 16 hours per week (but not claiming ETU) said they would apply for such a benefit but 69 per cent of these said they would maintain their present working hours even if it meant they did not get a top-up. The majority of people (38 of 48) who were working less than 16 hours per week said they would increase their working hours to get an Earnings Top-up. Among the small group currently working, men were more likely to say they would apply for an Earnings Top-up (68 per cent) than were women (63 per cent). In contrast to non-working respondents, those with partners were more likely to say they would apply (69 per cent) than were others (65 per cent). People in Scheme B areas were also slightly more likely to say they would apply (69 per cent) than those in Scheme A areas (63 per cent) or control areas (66 per cent). Broadly, these views were unchanged at the second interview in 1999. 105 Table 6.4 Percentage accepting a wage top-up by sociodemographic characteristics Cell percentages Base Would accept wage top-up Benefit type at sampling JSA IS 1366 531 64 47 1316 579 57 63 18 – 24 25 – 34 340 327 63 66 35 – 44 45 – 54 327 506 62 58 55+ 392 50 Partnership status No partner 1310 62 Has partner 588 53 Gender Men Women Age group Long-standing illness or disability Yes 832 54 No 1066 63 ETU area ETU Scheme A area 666 62 ETU Scheme B area Control area 617 615 57 58 All 1895 59 Base: people not currently working or receiving ETU. 6.5 Summary At first interview, awareness of ETU among the unemployed sample was very low: only 16 per cent of respondents could name ETU. Considering that unemployed people were a main target group for ETU and that the sample had been recently claiming benefit for at least six months it would not be unreasonable to expect awareness to have been higher. Also, the benefit had been available in their local area for almost two years by the time of first interview and so was well established. Respondents in certain areas (like Sunderland, Newcastle and Doncaster) had higher levels of recalled knowledge and this seemed to follow the volume of ETU claims in the area. Single people aged under 35 and those with academic qualifications also had above average awareness of ETU. People most commonly said they had heard of ETU at the Jobcentre or benefit office. Almost six in ten respondents not currently receiving ETU found the 106 idea of a wage top-up acceptable if one was available. The remainder (41 per cent) preferred to hold out for a better paid job. Older respondents, those with a partner, and people with health problems were less likely to accept the idea of a wage top-up. Two-thirds of the people who had found work by the time of the interview said they would apply for a wage top-up if one was available but most (69 per cent) would not have changed their working hours to get the benefit. Few respondents had direct experience of claiming ETU. At first interview, two per cent were current recipients and a similar proportion said they had received ETU in the past. These proportions had risen slightly to around three per cent at second interview. 107 7 CONCLUSIONS This report has described a sample of people who were selected as being unemployed and claiming Jobseeker’s Allowance or Income Support for between 26 and 65 weeks in June 1998. None of the sample had children and so they represented a key target group for the ETU pilot. 7.1 First interview - summer 1998 In the gap between selection and first interview, a minority of respondents had found work, most of it fairly low paid. But they had found it at the same rate, and for the same money, in the two ETU pilot areas compared with the control areas. A few of those in the pilot areas had also claimed ETU when entering work, but too few to be counted as an influence on people’s rate of return to work. Of the remainder, many respondents seemed likely to experience continued difficulties getting and keeping work. Typically they were poorly educated and often had only little recent previous work experience, especially the women among them. Significant numbers reported health problems or caring responsibilities that restricted their participation in paid work. At first interview, the prospects for this sample seemed less encouraging even than those faced by a similar sample interviewed prior to the introduction of ETU in 1996. The 1998 sample were found to be more disadvantaged in terms of their qualifications, work experience and health, probably because of the greater proportion of older respondents interviewed in 1998. Despite the improved economic conditions over the two years, reductions in claimant unemployed totals over the two years separating the surveys had left the more disadvantaged behind. Their rates of jobsearch were low and poor health was actually the most common explanation given by the substantial proportion of the sample (23 per cent) that had not looked for work in the four weeks before interview and also said that they did not want a job. Most of these had been selected as receiving IS, rather than JSA, and this difference between recipients proved crucial. ETU was intended to help everyone who could work, not just to give an added boost to those receiving JSA, but many of the IS sample (more than one in four of all respondents) seemed to be losing touch with the labour market altogether. On the other hand, most of those who were looking for work were the kinds of people who would have qualified for ETU if they found it. The majority had previous experience in low-skill, low paid jobs and were seeking wages of around £3.50 per hour, though even this rate would have left single people on the margins of eligibility at much more than 30 hours work a week, particularly in Scheme A. 109 As found previously, a second crucial distinction lay between those paying or not paying for their accommodation. For people without significant housing costs even working 16 hours a week in a low paid job supplemented with ETU could have given a significant boost to their income. For example, people working 16 hours at £3.60 per hour would be between £25 and £40 per week better off claiming ETU than if they had continued to receive only JSA. But for people paying rent, working and claiming ETU usually incurred a substantial loss of Housing Benefit leaving them with little additional income as workers. In the 1998 sample, 44 per cent of respondents were tenants (most receiving HB) and another one in ten had a mortgage to pay. This was much higher than in the corresponding 1996 survey, reflecting the older age profile of the 1998 unemployed and the greater numbers no longer living with parents. As a result, the minimum wages they said they would accept were higher, which tended to place more of them beyond the reach of ETU. A serious obstacle to ETU helping this unemployed sample back into work was the lack of awareness of the benefit itself. At first interview, recalled knowledge was disappointingly low, considering the sample was comprised of people with recent experience of claiming benefit for at least six months and living in areas where ETU had been available for around two years. Just 29 per cent of respondents in the ETU areas said they had heard of the introduction of an in-work benefit for people without dependent children and only 16 per cent could name this benefit as ETU. Awareness was no higher among those who had got jobs, discounting the minority (one in eight of them) who had claimed ETU. Furthermore, not everyone liked the idea of a top-up to potential wages in work even when such a scenario was put to them. Two-fifths of people preferred to hold out for a job paying the minimum wage they felt they could live on, rather than accept a lower paid job with a top-up. There may be many reasons for this. Qualitative research on ETU indicates that some people felt that there should be no need for a benefit top-up as employers ought to pay a ‘decent wage’ in the first place (Vincent et al, 1999). The introduction of the National Minimum Wage can be seen, in part, as a reflection of this view and indeed it was enough to move many workers out of eligibility for ETU. Clearly, if some type of wage supplementation were to be introduced for workers without dependent children, it would need to allow for higher full-time wages than those tested in the ETU pilot, otherwise it would simply be subsidising part-time work. 110 7.2 Second interview summer 1999 There was little change in the situation of respondents between the interviews. As predicted, few had moved into employment of 16 or more hours a week by the time of second interview (13 per cent) and ETU appeared to have had no significant influence on movements into work. However, it is necessary to add a few cautions to this seemingly negative story. It is important to remember that this sample had been unemployed for between six and fifteen months when sampled and so was not representative of all unemployed people in the pilot areas at that time. It may be possible that ETU had a measurable effect on movements into work for people who had been unemployed for shorter periods of time. It is also possible that we were not able to measure any ETU effects in this survey. There were probably too few movements into work for us to be able to reliably capture any ETU effect unless it was particularly large. The aim of the evaluation was to compare the pilot areas with the control areas and to attribute any difference to ETU. But, for the size of the effect to be measurable there would need to be widespread knowledge of the benefit. Measuring awareness and understanding is not easy, but despite these difficulties, the evidence from this survey is that awareness was poor. As advertising for ETU was stopped in April 1997, just six months after the benefit was introduced, this is not surprising. ETU also appeared to have no influence on the wage expectations of those who remained unemployed. However, the concept of expected wages is, to some extent, notional and the level of expected wages may get revised during the job search process, particularly if people become aware of in-work benefits while looking for work. In contrast, the reservation wage is the minimum wage that someone will accept at the point of job offer. In the context of low levels of awareness of ETU, the result that ETU had no effect on expected wages should not be interpreted as evidence that it did not, or could not have, lowered reservation wages. 111 Department of Social Security Research Report NO 131 Earnings Top-up Evaluation: Effects on Unemployed People Part Two • Econometric Analysis Abigail McKnight 8 INTRODUCTION This section of the report contains the findings from an evaluation of the impact of Earnings Top-up on claims for unemployment benefit. Earnings Top-up (ETU), an in-work benefit available to low paid single people and couples without children, has been piloted in a number of areas around Britain for a period of three years (October 1996 to October 1999). A pilot of this scale provides the scope to collect and analyse a substantial amount of data and to look in detail at the impact of an inwork benefit in the short and longer term. This report forms a small part of an extensive evaluation of the ETU pilot covering the first effects (Finlayson et al, 2000; Vincent et al, 2000), take-up and the impact on wages and employment from an employee and employer perspective (Marsh et al, 2001; Lissenburgh et al 2001; Heaver et al, 2001; Vincent et al, 2001). The various strands of the evaluation have drawn from an extensive analysis of contrasts and similarities in labour market conditions across the ETU pilot areas (Green, 2001). The focus of this study is to assess the overall impact of ETU on unemployment in the ETU pilot areas and to estimate the impact of ETU on individual groups of unemployment benefit claimants. 8.1 The ETU pilot: aims and objectives Earnings Top-up, as its name suggests, is an in-work benefit designed to ‘top-up’ the earnings of low paid workers (employed or self-employed) and thereby increase their overall income. This benefit was made available to people without dependent children, whether or not they were part of a couple, whose circumstances met with various employment, income and savings criteria. Once eligibility was met and entitlement was established, ETU was paid at a fixed rate for a period of 26 weeks. At the end of 26 weeks individuals could reapply and have their situation reassessed. Employees could continue to reapply until the end of the pilot, but some restrictions applied for the self-employed3. Its design and operation broadly mimicked an in-work benefit available to people with dependent children, known as Family Credit before the introduction of the Working Families’ Tax Credit in October 1999. This meant that in areas where ETU was being piloted, nearly all individuals working more than 16 hours a week in low paid jobs qualified for an in-work earnings supplement. The main exceptions were full-time students and couples where the partner of the low paid worker was paid well enough to take their income over the entitlement threshold. 3 If self-employed workers had never earned more than £20 a week from selfemployment during the previous four awards they were not entitled to a further award. Those earning more than £20 a week from their business could continue to apply until the end of the pilot. 115 Earnings Top-up was designed to improve the work incentives and increase employment sustainability for people without dependent children who only had access to low paid work. The Department of Social Security (DSS, 1995) outlined two main objectives for ETU: • to improve the incentive for unemployed single people and couples without dependent children to take work of 16 hours or more a week, without worsening incentives for other; • to improve the incentives for those on low incomes to stay in work by raising their incomes relative to out-of-work support, without reducing their hours of work. This study forms part of the evaluation commissioned by the Department of Social Security and involved a consortium of research centres, Policy Studies Institute, Centre for Research in Social Policy, and the Institute for Employment Research, led by the Policy Studies Institute to assess how far Earnings Top-up achieved its main aims and objectives. The basic question to be addressed in this part is whether ETU led to a decrease in flows into unemployment and whether or not the presence of ETU increased flows out of unemployment. An examination of the experience of different skill groups is made to assess the differential impact of ETU and to test for the presence of substitution effects4. 8.2 Design of the pilot The pilot, which began in October 1996, ran for a three year period and involved eight areas and two different rates of benefit (referred to as Scheme A and Scheme B). Like Family Credit, ETU comprised of a basic credit, payable in full to people whose earnings plus the credit fell below a threshold. A withdrawal rate was set at 70 per cent (i.e. for every pound over the threshold, ETU was reduced by 70p). Credits and thresholds varied for couples, single people aged 18-24 years and single people aged 25 and over. Scheme B had a higher threshold than Scheme A for single people and a higher credit for couples and this is why Scheme B is referred to as the more ‘generous’ scheme. The eight areas chosen for the pilot were identified on the basis that they were characterised by high levels of unemployment and low wage employment. This was done to maximise the potential impact of ETU during the pilot. This means that areas were chosen and were not randomly selected for the pilot. The non-random selection of areas means that the results from the evaluation must be interpreted with this in mind. The areas were also chosen to represent four distinctly different types of labour market: urban areas, large towns, seaside areas and rural areas. A Scheme A and a Scheme B area was chosen in each of these labour market types. The Scheme A areas (piloting the less generous benefit) 4 116 Substitution, in this context, occurs when an ETU recipient is recruited (or retained) at the expense of an existing employee or another person looking for work. were roughly twice the size of the Scheme B areas in terms of potential ETU claimants. A comparison area was selected for each labour market type. In total there were eight pilot areas and four comparison areas. Table 8.1 lists the ETU pilot and comparison areas, showing how the areas fit into the classification scheme. Table 8.1 ETU pilot and comparison areas Scheme A Scheme B Comparison area Urban Newcastle-upon-Tyne Sunderland Middlesborough, Hartlepool & Stockton Large town Castleford, Wakefield Doncaster Rotherham and Worksop and Barnsley Seaside area Southend Bournemouth Southampton and the Isle of Wight Rural area North Wales Perth South Wales 117 9 METHODOLOGIES FOR SOCIAL PROGRAMME EVALUATION Large scale economic evaluations of government funded active labour market programmes are relatively new in the UK, which is in complete contrast to the United States where there is a long history of social experiments and evaluations. Hausman and Wise (1985) note that in 1985, $500 million had been spent on social programmes and evaluations in the previous ten year period. Since this time expenditure on social experiments is likely to have increased rather than declined. In fact, following the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, individual States have been given greater flexibility and autonomy with their welfare budget and this has led to an explosion in large and small scale social experiments. In the UK, since the election of the Labour government in 1997, a much larger share of social programme budgets has been spent on evaluation. All of the New Deal programmes and pilots have been, or are currently being, evaluated. A number of prominent methodologies have been employed over time to evaluate programmes. In the early days the most common form of evaluation involved comparing the experience of programme recipients before and after they received the ‘treatment’. For labour market programmes this usually involved comparing the experience of employment, unemployment or earnings. This type of analysis suffers from the well-known problem of sample selection bias. That is, in a voluntary programme, individuals most likely to benefit from the programme will choose to participate. Consequently, the measured impact of the programme will be exaggerated by examining voluntary participants only and the findings from this group cannot be used to estimate the impact of a programme on the wider population. In addition, there is a well documented ‘dip’ in the prospects (in terms of employment, earnings, income) of participants immediately prior to programme participation. Ashenfelter (1978) found that average earnings of individuals participating in government training programmes declined prior to entry into the programme. This result has been replicated elsewhere (see for example Heckman and Smith, 1999) and is often referred to as the ‘Ashenfelter dip’ or ‘pre-programme dip’. This ‘dip’ may be due to either the anticipation of participating in the programme or because most of the programmes are designed to identify and help those most in need. In many social experiments the potential participants have the choice between participating in the experiment – in the knowledge that they may end up in the treatment or control group – or not. This can lead to potential problems relating to selection bias. In some cases, such as a number of the welfare-to-work experiments in the US, individuals receiving welfare do not have a choice in whether or not they participate 119 in the trial but the local welfare office does. This means that even if randomisation is achieved after the population has been chosen this population may not constitute a random sample. Social experimentation, where individuals are randomly assigned to a programme in a similar way to that used in medical trials, has been very popular among programme designers and evaluators in the United States. In the UK social experimentation is rare and seems to face a stronger ethical opposition. There are, however, a few exceptions. In 1989 an evaluation of the Restart5 programme was based on random assignment of part of a cohort of unemployed people to the programme and a second sample to a control group on the basis of a National Insurance digit sequence (Dolton and O’Neill, 1996). In the early stages of the New Deal for Lone Parents prototype programme, lone parents with existing claims for Income Support were randomly assigned (on the basis of a digit in their National Insurance number) to the programme on a monthby-month basis (see McKnight, 2000). However, once assigned to the programme, lone parents could choose to participate or not as NDLP is a voluntary programme. There now appears to be a greater appetite for social experiments in the UK than in the past and so it seems likely that government programmes will be piloted in this way in the future. Social experimentation is fairly widely used in the US and advocates argue quite vehemently that results derived from non-experimental data should be treated at best with caution (LaLonde, 1986 and Friedlander and Robins, 1995). Meta-level analysis6 comparing the results derived from experimental and non-experimental data has shown wide discrepancies in results (Friedlander and Robins, op cit). This appears to be largely due to the inability to control for selection bias and control for the other factors influencing the experience of participants and nonparticipants. It has been argued that judicious choice of models and use of results from a pre-programme period to improve the model, reduce discrepancies between findings from experimental and non-experimental data (Heckman and Hotz, 1989). In recent times a number of highly respected commentators have highlighted the weaknesses in random experiments and the results they can provide (Heckman and Smith, 1995). The main objections are that participants in an experiment may not constitute a random sample due to problems associated with self-selection into the experiment and individuals 120 5 The Restart programme involved a compulsory interview with Restart counsellor after six continuous months of claiming Unemployment Benefit. The Restart counsellors concluded the interview with a recommended course of action designed to help the job seeker in their search for work. 6 Meta-level analysis involves secondary analysis of results attained across a range of previous estimates on a particular issue. dropping out (attrition) during the programme. Control group members may also change their behaviour in the knowledge that they are involved in a trial and are being monitored (often referred to as the ‘Hawthorne effect’). This means that they no longer represent the true counterfactual (i.e. what would have happened in the absence of the programme). The control group may also be able to find (and may even be helped to find) good substitutes for the programme and will, therefore, not represent the counterfactual situation associated with no treatment. Experimental data may be able to provide answers in terms of the average impact of the programme but are generally weak in terms of providing answers to questions on the distributional impact of the programme, what works best, on whom and how. Programme developers and policy makers need this type of information to help in programme design and delivery. Another weakness of experimental data is that it is impossible to estimate what the wider labour market consequences are. An active labour market programme may have knock-on effects. It is possible that a programme can improve the prospects of one group while simultaneously leading to a deterioration in the prospects of another group. This may occur through displacement or substitution effects. Substitution occurs when a programme recipient is recruited (or retained) at the expense of an existing employee or another person looking for work. Displacement occurs when the creation of additional jobs in a firm employing programme recipients leads to a reduction in employment in other firms. Some of this may be picked up through the difference in the prospects of the treatment and control groups but if other groups7 are affected then this will not be measured using data from microexperiments. Some programmes, particularly voluntary programmes, operate through increasing individuals’ awareness of education and training programmes, availability of job search assistance or an in-work benefit through the use of media campaigns and promotions. It is, therefore, not practical to pilot this type of programme through a randomised trial8. Evaluations using non-experimental data have often resorted to comparing the experience of participants with a comparison group of non-participants. The non-participants may be identified in alternative secondary data sources and may or may not live in the same area as the participants. Many evaluators are forced to use a ‘before and after’ comparison treating the introduction of a national programme as a type of ‘natural experiment’. This usually involves identifying a ‘best-fit’ comparison group (see Blundell (2000) for a review of evaluation techniques and some ideas on how the impact of the Working Families’ Tax Credit can be evaluated). 7 The control group usually comprises individuals who qualify for the treatment, but substitution and displacement may occur among individuals who would not be eligible for the programme. 8 Advancements in digital technology may mean that in the future advertising campaigns can be targeted at small groups. 121 The Earnings Top-up pilot fits into the category of a macroexperiment. In a macroexperiment the unit of treatment is a whole group or community. The advantage of macroexperiments over microexperiments is that they avoid a number of problems related to design, implementation and interpretation (Harris, 1985). These include participant self-selection, attrition, ethical constraints on individual randomisation. In addition, microexperiments cannot provide answers to questions on wider secondary effects, while macroexperiments mimic the way in which a policy option would be delivered if it were fully implemented. 122 10 DESIGN OF THE ETU EVALUATION The Earnings Top-up pilot and evaluation, running since October 1996, overlapped the previous as well as the current administration. The change in political environment naturally changes the policy environment. When the pilot was put in place in 1996 the Conservative government had in mind an extension to Family Credit to people without dependent children. In October 1999 the Labour government replaced Family Credit with the Working Families’ Tax Credit. If an earnings supplement were to be introduced today it would take the form of an employment tax credit rather than in the form in which Earnings Top-up was piloted.9 However, the basic principles of the benefit and the tax credit are the same and the results from the ETU pilot can provide a reasonable estimate of the likely impact of an employment tax credit for people without dependent children. The design of the ETU evaluation, which makes use of pilot and comparison areas to assess the counterfactual (what would have happened in the absence of ETU), is particularly important in this evaluation due to the timing of the pilot. ETU was introduced in the pilot areas in October 1996 and at the same time two major changes also took place which are known to have affected the claimant count (Sweeney and McMahon, 1998). Jobseeker’s Allowance (JSA), a new policy for administering unemployment benefit, replacing Unemployment Benefit and Income Support for unemployed job seekers, was introduced on 7 October 1996. JSA led to many changes in the rules governing entitlement to benefit (reducing the length of contributory benefit entitlement from 12 months to six months, providing additional help and advice in job search, introducing stricter eligibility checks). In addition, significant changes were made to the organisation of, and the computer systems that manage, the payment of unemployment related benefit. Sweeney and McMahon (1998) show that both these factors led to sharp falls in the claimant count between October 1996 and April 1997. While some of the fall can be ascribed to the new rules and regulation of unemployment benefit receipt, some has arisen due to administrative error. For the present study the information from the comparison areas can help to identify changes in the experience of unemployment that can be ascribed to ETU and those brought about as a result of other factors. The methodologies employed in the evaluation of social experiments are generally dictated by the design of the experiment, the availability of data 9 An example of such an employment tax credit is outlined in The modernisation of Britain’s Tax and Benefit System. Number Six (HM Treasury, March 2000). 123 or both. The evaluation of ETU is no exception. The pilot takes the form of a macroexperiment but the areas were not chosen on a random basis, this means that the evaluation techniques have to recognise this. If the ETU pilot had been designed as a microexperiment with random assignment then a simple comparison in outcome variables of interest between the ‘treatment’ group and the ‘control’ group could be made to estimate the impact of the treatment. This type of experiment would have been wholly inappropriate for the ETU pilot because for the pilot to replicate what would occur, were it to be introduced at a national level, information about the scheme must be widely available to individuals and employers. In a macroexperiment without random assignment it is necessary to ‘benchmark’ the outcome variables of interest by using information before the pilot began. This information can then be used to test for divergence in the outcome variables after the programme has been introduced. The technique is often referred to as estimating the difference-in-differences. The availability of a long run of data before the pilot can allow evaluators to develop a predictive model of the labour market and then use this model to forecast the outcome variables of interest after the programme. Any difference between the predicted values and the actual values provides an estimate of the impact of the programme. This type of approach was adopted by the evaluators of the New Deal for Young People (NDYP) (Anderton et al, 1999). They used information on outflow rates from unemployment for a period of around 10 years before the introduction of NDYP, in January 1998 in a number of Pathfinder areas and in April 1998 nationally, to forecast unemployment outflow rates for the first year of the programme. A comparison between predicted outflow rates from unemployment and observed outflow rates for different groups of claimants provided an estimate of the impact of NDYP. This type of approach could be adopted here but the availability of individual level unemployment data for the ETU pilot and comparison areas is limited to a period of 21 months prior to the introduction of ETU, which is not a long enough period to estimate a forecasting model. This technique was also more appropriate to the evaluation of NDYP were the client group could be precisely identified in the unemployment data series. Although changes in the area level experience of unemployment is of interest, one of the aims of ETU was to improve the work incentives for the client group without worsening incentives for others. To judge whether or not this has been achieved it is necessary to compare the experience of different groups of unemployment benefit claimants. A number of different methodologies to evaluate the impact of Earnings Top-up are adopted here. An examination of flows into unemployment and flows out of unemployment is made for all ETU pilot and comparison areas for a period of 21 months prior to the pilot and 27 months into the 124 pilot, to assess whether or not there has been a divergence in the experience of unemployment in the ETU pilot and comparison areas. Flows into and out of unemployment are also explored for different skill groups by partnership status, age groups and at a detailed area level. To estimate the impact of ETU a simple version of the difference-indifferences methodology is employed using information from the period before the pilot programme was in place along with information from comparison areas to test whether or not ETU was associated with changes in the inflow into unemployment. The technical detail of the methodology employed is outlined in Section 10.1. Readers who are not interested in the technical detail can omit this section. 10.1 Technical detail of evaluation methodology The ratio of inflows into unemployment (I) in the ETU Scheme A pilot areas and the comparison areas in the M months before ETU was introduced is calculated as follows: 1 Μ ΣΙ A i A/C UIM = i=m Μ ΣΙ C i i=m For the E months during the ETU pilot the following ratio is calculated: 2 Ε A ΣΙ i A/C UIE = i=e Ε C ΣΙ i i=e The impact of ETU on inflows into unemployment in the ETU Scheme A areas is estimated by calculating the difference in these two ratios: 3 ETUA ∆UI A/C A/C = UIE _ UIM The impact of ETU in the Scheme B areas is calculated using the same technique. This technique is used to estimate the impact of ETU on all unemployment benefit claimants and to estimate the impact of ETU on groups most likely to qualify for ETU; single males and single females by skill group. The analysis by skill group can identify whether or not ETU has had a greater impact on the unskilled group (at whom it is targeted) and whether there is evidence of substitution effects among other skill groups. 125 To assess whether or not ETU had an impact on outflows from unemployment separate from an impact on inflows it is necessary to calculate the ratio of outflows (O) to inflows (I) in the ETU areas relative to the comparison areas. If ETU has led to a decrease in inflows then outflows will be lower as a consequence. This is why it is necessary to normalise outflows on inflows thus providing a measure of net outflows rather than gross outflows. For ETU Scheme A areas the following calculation is made for the period of M months before the pilot: 4 Μ A Μ ΣΟ ΣΙ i A/C UOIM = i=m Μ C A i i=m Μ C ΣΟ ΣΙ i=m i=m i i For the E months of the ETU pilot the following ratio is calculated: 5 Ε A Ε ΣΟ ΣΙ i A/C UOIE = i=e Ε i=e C ΣΟ i=e A i i Ε ΣΙ C i i=e The impact of ETU on unemployment outflows in the ETU Scheme A areas is estimated by calculating the difference in these two ratios: 6 ETUA ∆UOI A/C A/C = UOIE _ UOIM The impact of ETU on unemployment outflows relative to inflows in the Scheme B areas is calculated in the same way. The impact of ETU on unemployment outflows is calculated for all unemployment benefit claimants, and for single males and single females by skill level, partnership status, age groups and at a detailed area level. It should be clearly stated at this point that the objective of this part of the evaluation of ETU is to assess the wider labour market impact of ETU not to assess the impact of ETU on an individual’s labour market experience. 126 11 DESCRIPTION OF THE DATA This study makes use of a specially commissioned dataset covering administrative records of all unemployment benefit claims from January 1995 to December 1998 in the ETU pilot and comparison areas. These data were extracted from the Joint Unemployment and Vacancy Operating System (JUVOS) and provided by the Office for National Statistics. Information contained in this special dataset includes the start and end date of all benefit claims, gender of claimant, marital status, type of claim, destination of claimant, sought occupation, usual occupation, date of birth, pilot and comparison area indicators10. The dataset covers a period of 21 months prior to the introduction of ETU and 27 months of the ETU pilot programme. The dataset, therefore, does not cover the complete period of the ETU pilot but avoids the introduction of the National Minimum Wage in April 1999 and the general decline in the ETU caseload throughout 1999 (the last new claims for ETU were accepted in September 1999) shown in Figure 11.1. For a detailed analysis of ETU claims and claimants see Marsh et al (2001). Figure 11.1 The Earnings Top-up caseload from October 1996 to January 2000 by claimant type 10 A data report providing more detailed information on this dataset and how it was constructed along with basic validation is available from the DSS on request, McKnight (2001). 127 The final dataset contained information on 1,846,553 unemployment benefit claims. Nearly three-quarters of these claims were made by men (73 per cent) over the sample period. Table 11.1 shows the age distribution of claimants at the start of their claim. Over one-third of claimants were aged 16-24 years and nearly two-thirds were under the age of 35 at the start of their claim. Table 11.1 Age of unemployment benefit claimants at start of claim 16-24 years 25-34 years 36% 26% 35-44 years 45-54 years 17% 15% 55+ years 6% Base (=100%) 1,846,553 The majority of unemployment benefit claimants were single (56 per cent) and a further 29 per cent were married. The rest were divorced (seven per cent), separated (three per cent), cohabiting (three per cent) or widowed (one per cent). Marital status was not known for two per cent of the sample. Figure 11.2 shows, for males, the number of unemployment benefit claimants by age at the start of their claim. The shading indicates the marital status of claimants. The largest category of claimants comprises those who start their claim at 18 years of age. Very few claimants begin a claim for unemployment benefit under the age of 18 because 16/17 year olds only qualify for unemployment benefit under special circumstances; 16/17 year olds must be classed as being in a vulnerable group (those forced to live away from their parents, couples with children, those released from custody or local authority care) or they may qualify under the severe hardship rules. The chart shows the gradual decline in unemployment benefit claim starts by age and a fall in the proportion of claimants who were single. Most young men under 30 starting a claim for unemployment benefit were single. 128 Figure 11.2 Number of male unemployment benefit claimants according to age at start of claim and marital status The chart for females (Figure 11.3) shows a very similar picture albeit scaled down by the lower number of claims for unemployment benefit made by women. Like young men, most young women making a claim for unemployment benefit were single. Figure 11.3 Number of female unemployment benefit claimants according to age at start of claim and marital status The reason for leaving unemployment benefit is shown in Table 11.2. Individuals leaving unemployment benefit were most likely to enter work (57 per cent). Just over one-fifth (21 per cent) left to unknown destinations (failed to attend, ceased claiming, not known), some of whom are likely to have entered work. Six per cent transferred to a government supported training scheme. Approximately 11 per cent of Unemployment Benefit claimants left Unemployment Benefit to claim another benefit, with five per cent claiming Sickness Benefit, two per cent Incapacity Benefit, one per cent Income Support and the rest claiming other benefits not specified. 129 Table 11.2 Destination of claimants leaving unemployment Destination % of all completed claims Found work 57% Failed to attend/ceased claiming/not known Transferred to government training scheme 21% 6% Claimed Sickness Benefit Claimed Incapacity Benefit 5% 2% Claimed Income Support Claimed another benefit 1% 3% Gone abroad Full time education 2% 1% Other 2% Base (=100%) 1,598,017 One of the great weaknesses of administrative data on unemployment benefit claims is the lack of information on claimants’ skills. This lack of information on human capital has hampered a number of previous analyses which have made use of JUVOS data. For the purposes of the current study a classification of claimants’ skills has been developed using information provided by claimants on their usual and sought occupation as part of their JobSeeker’s Agreement. Information on occupation has been used successfully elsewhere to proxy skill and education levels (see for example McKnight (2000) and Nickell et al (2000)). The assumption made here is that the type of occupation that the claimant is usually employed in or that they are seeking provides a fairly good proxy for the level of their skills. A four-fold classification of skill levels - high skilled, skilled, low skilled, unskilled - is developed using the following mapping of the Standard Occupational Classification (SOC90) minor groups detailed in Table 11.3. The skilled and low-skilled categories are further sub-divided into non-manual and manual to create a six-fold classification of skill. 130 Table 11.3 Definition of skill levels Minor Groups Skill level Major Group description (2 digit) High skilled Managers and administrator (excluding office managers and 10,11,12,15,19 manager/proprietors in agriculture and services). Professional occupations. 20-27,29 Office managers and managers/proprietors in services. 13,14,17 Associate professional and technical occupations. Buyers, brokers, sales reps. 30-39 70,71 Manual Managers/proprietors in agriculture. Craft and related occupations. 16 50-59 Non-manual Clerical and secretarial occupations. 40-47,49 Personal and protective service occupations. Sales occupations (except buyers, brokers, sales reps). 60-67,69 72,73,79 Plant and machine operatives. Other occupations in agriculture, forestry, fishing. 80-89 90 Other elementary occupations. 91-95,99 Non-manual Skilled Low skilled Manual Unskilled The skill composition of the unemployment benefit claimants is shown in Table 11.4. As one would expect there are very few high skilled individuals in the population of claimants, making up only seven per cent. The majority of claimants are either low skilled (44 per cent) (33 per cent low skilled non-manual) or unskilled (19 per cent). It is interesting to note the difference in the skill structure of male and female claimants, with a smaller share of females in the unskilled group but a greater share in the low skilled group. The low skilled non-manual group contains the three largest occupational areas of female employment (secretarial, personal services and sales). In the skilled group women are more likely to be skilled non-manual, while men are more likely to be skilled manual. Table 11.4 Distribution of skill among unemployment benefit claimants Skill level Males Females All claimants High skilled 7% 9% 7% Non-manual 9% 10% 9% Manual 27% 3% 21% Non-manual Manual 21% 14% 65% 4% 33% 11% Unskilled 23% 10% 19% Missing 1% 1% 1% 1,345,546 490,512 1,846,553 Skilled Low skilled Base (=100%) 131 Figure 11.4 shows the distribution of claimants by skill group and area type. The chart shows a similar skill structure of unemployment in the four area types with slightly more high skilled claimants in the rural areas, a slightly higher proportion of skilled manual claimants in the urban areas and a lower proportion of unskilled claimants in the seaside areas. Figure 11.4 Distribution of skill among unemployment benefit claimants by area type - claims starting before the start of the ETU pilot Figure 11.5 shows the distribution of skill among claimants in the pilot and comparison areas. This chart shows a similar distribution of skills among claimants in the Scheme A (ETU A), Scheme B (ETU B) and control areas (ETU C) before the introduction of ETU in October 1996. Figure 11.5 Distribution of skill among unemployment benefit claimants by Scheme A, Scheme B and comparison areas -claims starting before the ETU pilot 132 While the skill classification used here is not ideal because it relies on an occupational mapping and may pick up changes in demand for occupational areas, the categories are broad enough for this not to be a major issue. Levels of educational attainment on their own, if they were available, may well be a poor indicator of skill for this group because skill is a combination of educational attainment, training and work experience. Younger cohorts tend to have higher educational qualifications but lack work experience and training held by older cohorts. On balance, it is felt that the occupational mapping into skill groups provides a fairly good proxy for skill. Finally, Table 11.5 shows the distribution of unemployment benefit claimants across ETU pilot and comparison areas. It is not surprising that the urban areas and the large towns tend to have the largest claimant populations. This table highlights the fact that Scheme A areas tend to be roughly twice the size of Scheme B areas (the more generous scheme) in terms of unemployment benefit claimants. The difference in size was designed to achieve roughly equivalent ETU claimant populations. In terms of size, the comparison areas fall in between the Scheme A and Scheme B areas within an area type. Table 11.5 Distribution of unemployment benefit claimants across ETU pilot and comparison areas Area Scheme Claimants Newcastle-upon-Tyne Sunderland Urban (A) Urban (B) 15% 8% Middlesborough, Hartlepool & Stockton Castleford, Wakefield & Barnsley Urban (C) Large town (A) 13% 12% Doncaster Rotherham and Worksop Large town (B) Large town (C) 5% 8% Southend Bournemouth Seaside area (A) Seaside area (B) 9% 4% Southampton & the Isle of Wight North Wales Seaside area (C) Rural area (A) 7% 10% Perth South Wales Rural area (B) Rural area (C) 4% 5% Base (=100%) 1,846,553 133 12 FLOWS INTO AND OUT OF UNEMPLOYMENT IN THE ETU PILOT AND COMPARISON AREAS The analysis of the unemployment benefit data begins with a descriptive overview of unemployment inflows and unemployment outflows over the period January 1995 and December 1998. Figure 12.1 shows the monthly flows into unemployment by the ETU pilot and comparison areas. Two main features are evident from this chart. Firstly, unemployment inflows follow a seasonal pattern, with annual peaks at the end of the academic year and calendar year. Secondly, there has been a fall in inflows into unemployment from approximately the time that JobSeeker’s Allowance was introduced in October 1996. The fact that the decline is evident in ETU pilot and control areas suggests that this ‘step’ change in inflows was not associated with the ETU pilot. It is not possible to tell from visual inspection of this chart whether ETU has led to a relative decline in inflows to unemployment in the ETU pilot areas. Figure 12.1 Monthly inflows into unemployment in ETU Scheme A, Scheme B and comparison areas Figure 12.2 shows the monthly outflows from unemployment in the ETU pilot and comparison areas. Outflows from unemployment follow a much stronger seasonal pattern than inflows, with a large fall in outflows during December and peaks in September. Between January 1995 and November 1998 there has been a fall in outflows from unemployment tracking the fall in inflows and the decline in the stock of unemployed. 135 Figure 12.2 Monthly outflows from unemployment in ETU Scheme A, Scheme B and comparison areas The following sets of charts show the time series of unemployment inflows and unemployment outflows for the urban areas, large towns, seaside areas and rural areas by ETU pilot and comparison areas. In addition, the ratio of inflows to outflows is shown: when the line in these charts is above one then inflows are greater than outflows and the stock of unemployment is added to and, conversely, when the line is below one more individuals are leaving unemployment than joining and the stock is falling. Inflows into unemployment fall over this period in most of the ETU pilot and comparison areas. There is a fairly large decline in inflows in Rotherham and Worksop (the large town comparison area) after March 1998 relative to the large town Scheme A and Scheme B areas. It is not clear what led to this decline but further examination of the location of claimants suggest that it does not appear to be due to changes in geographical boundaries. Southend is particularly noteworthy in terms of the fall in unemployment inflows. Take-up of ETU was particularly low in Southend. It is thought that this was due to the fact that wages in Southend are higher than in the other areas and consequently fewer people were eligible to claim ETU (Green, 2001). The fall in unemployment inflows observed in Southend is extremely unlikely to be associated with ETU. There is much less change in inflows in the rural areas and fairly stable levels of inflows in Perth (B) and South Wales (C). Outflows follow the strong seasonal pattern in all the ETU pilot and comparison areas. Outflows also follow a general downward trend, although, Perth (B) and South Wales (C) show very little change over this period. Rotherham and Worksop (C) and Southend (A) show greater declines in inflows relative to the other area types shadowing the changes in inflows. 136 The peaks in inflows in September and the falls in outflows in December lead to a situation where unemployment inflows are greater than outflows at these times. Apart from the peaks, unemployment outflows generally exceed unemployment inflows over the sample period. There are no clear indications from these detailed charts of unemployment inflows and unemployment outflows that ETU has led to a large decrease in inflows or a large increase in outflows. 12.1 Flows into and out of unemployment - urban areas Figure 12.3 Flows into unemployment - urban areas Figure 12.4 Flows out of unemployment - urban areas 137 Figure 12.5 The ratio of inflows to outflows - urban areas 12.2 Flows into and out of unemployment - large towns 138 Figure 12.6 Flows into unemployment - large towns Figure 12.7 Flows out of unemployment - large towns Figure 12.8 The ratio of inflows to outflows - large towns 139 12.3 Flows into and out of unemployment - seaside areas Figure 12.9 Flows into unemployment - seaside areas Figure 12.10 Flows out of unemployment - seaside areas 140 Figure 12.11 The ratio of inflows to outflows - seaside areas 12.4 Flows into and out of unemployment - rural areas Figure 12.12 Flows into unemployment - rural areas 141 Figure 12.13 Flows out of unemployment - rural areas Figure 12.14 The ratio of inflows to outflows - rural areas 12.5 Flows into and out of unemployment - high skilled and unskilled claimants 142 The final chart in this section (Figure 12.15) maps the inflows and outflows according to claimants’ level of skill using the classification detailed in Chapter 11. Two of the four skill groups – high skilled and unskilled – are shown here to test the validity of the classification for the statistical work that follows and to highlight the different experience of unemployment across skill groups. It is immediately apparent that these two skill groups follow different seasonal patterns of unemployment. The unskilled group experience a dramatic fall in outflows from unemployment in December while the high skilled have a peak in inflows in July followed by a peak in outflows in September (shadowing the academic year). There appears to be no strong downward trend in inflows or outflows for the unskilled group even though overall unemployment rates fell as the economy picked up. In contrast, the seasonal pattern for the high skilled group dampens over the period and this is accompanied by a downward trend in inflows and outflows. Figure 12.15 Unemployment inflows and outflows for high skilled and unskilled claimants 143 13 STATISTICAL ANALYSIS OF UNEMPLOYMENT BENEFIT CLAIMS In this chapter a version of the difference-in-differences technique, outlined in Chapter 10, is used to estimate the impact of ETU on inflows into unemployment and outflows from unemployment. The differencein-differences technique is used to estimate the change in the outflow rates in the pilot areas relative to the comparison areas before and after the introduction of ETU. The validity of the technique rests on the rather strong assumption that any change in the relative experience of unemployment in the pilot versus the comparison areas is unchanged with the exception of the introduction of the pilot. There are clearly a number of factors that can change over time at the local level which can alter the local labour market. However, the detailed analysis of the ETU pilot and comparison areas (Green, 2001) conducted as part of the ETU evaluation suggest that, with the exception of Southend, there have been no major changes to the local areas that comprise this evaluation. 13.1 The impact of ETU on all unemployment benefit claimants Table 13.1 contains the results for inflows into unemployment. The relative flows into unemployment between the two pilot area types (Scheme A and Scheme B) and the comparison areas are ‘benchmarked’ in the period before the ETU pilot. In the Scheme A areas there were 44 per cent more inflows into unemployment relative to the comparison areas prior to the ETU pilot programme. In the period after ETU was introduced in the Scheme A areas, inflows fell relative to the comparison areas to 41 per cent. In the Scheme B areas a similar relative fall in flows into unemployment is recorded. These falls suggest that ETU has had a small positive effect in the reduction of flows into unemployment. The higher fall in Scheme A areas than in Scheme B areas is perhaps counterintuitive given that the Scheme B areas had the more generous version of ETU. It was shown earlier that one of the greatest reductions in inflows to unemployment in the Scheme A areas was observed in Southend. It was also noted that this reduction was much more likely to be due to the earlier and greater pick-up in the local economy than in the other seaside areas. The figures in brackets in Table 13.2 show the estimate of the difference-in-differences when the seaside areas are removed. This estimate is much more in line with expectations of a one per cent reduction in relative inflows in the Scheme A areas and a two per cent reduction in the Scheme B areas. 145 Table 13.1 Flows into unemployment in ETU pilot areas relative to comparison areas ETU A/ETU C ETU B/ETU C Jan 95-Oct 96 1.444 0.698 Nov 96-Nov 98 1.413 0.681 -0.031 (-0.007) -0.017 Difference Note: the figure in brackets excludes the seaside areas. ETU also aimed to increase the work incentives of low paid workers and thereby increase outflows from unemployment. Outflows are, of course, a function of inflows (and ETU was designed to reduce inflows) and therefore it is necessary to measure changes in outflows as a share of inflows. Table 13.2 shows that in the period after the introduction of ETU the pilot areas record a small increase in outflows as a share of inflows in the pilot areas relative to the comparison areas. This result suggests that ETU has had a small positive effect on increasing outflows from unemployment. The figures in brackets show the estimates for ETU areas when the seaside areas have been excluded. Table 13.2 Outflows as a share of inflows in ETU pilot areas relative to comparison areas Jan 95-Oct 96 Nov 96-Nov 98 ETU ETU ETU ETU A B C A/ETU C 1.025 1.040 1.028 1.030 1.026 1.024 0.999 1.016 1.003 1.006 0.017 (0.007) 0.004 Difference ETU B/ETU C Note: the figure in brackets excludes the seaside areas. 13.2 The impact of ETU on single unemployment benefit claimants There is not enough information in the JUVOS dataset to precisely identify potentially eligible ETU claimants. For example, there is no information on whether or not unemployment benefit claimants have dependent children. However, an analysis of single males and single females can provide an indication of the impact of ETU11. When the ETU caseload reached its peak in March 1999 at 24,503, 14 per cent of claimants were couples, 40 per cent were single people under 25 and 46 per cent were single people aged 25 and over (Marsh et al, 2001). The identification of unskilled claimants among this population of single people should provide a fairly good proxy for the ETU target population. Table 13.3 shows the relative flows into unemployment in ETU Scheme A areas relative to the comparison areas for single males by skill group before and after the introduction of ETU. Relative falls in inflows are 11 146 Single parents are more likely to claim Income Support than Jobseeker’s Allowance. recorded for unskilled single males and skilled single males, low skilled manual males and skilled manual males and an increase for low skilled non-manual males, skilled non-manual and high skilled males. When the seaside areas are excluded a reduction in inflows in ETU Scheme A areas is observed for unskilled and low skilled manual males. Table 13.3 Flows into unemployment in ETU A areas relative to ETU C areas by skill group - single males Unskilled Low skilled Skilled High skilled manual non-manual manual non-manual Jan 95-Oct 96 1.247 1.460 1.656 1.278 1.605 1.600 Nov 96-Nov 98 1.203 1.402 1.681 1.242 1.659 1.683 -0.045 (-0.073) -0.058 (-0.057) 0.025 (0.053) -0.036 (0.001) 0.055 (0.083) 0.084 (0.124) Difference Note: the figures in brackets exclude the seaside areas. Table 13.4 contains the results for ETU Scheme B areas. In ETU Scheme B areas unskilled and skilled non-manual single males record a relative fall in unemployment inflows and an increase for low skilled, skilled manual and high skilled males. Overall the results for single males suggest that ETU has been associated with a fall in inflows for the unskilled and with a greater decrease in Scheme B areas than Scheme A areas (although this no longer holds when the seaside areas are excluded) and these gains may well have been associated with a reduction in the fortunes of low skilled single males. It is not clear why the high skilled single males in the ETU Scheme A areas experience a relative increase in flows into unemployment but it seems very unlikely that this has anything to do with the ETU pilot. These individuals usually work in managerial or professional occupations. It was shown in Table 11.6 that they constitute a fairly small proportion of unemployment benefit claimants (seven per cent). Table 13.4 Flows into unemployment in ETU B areas relative to ETU C areas by skill group - single males Unskilled Low skilled Skilled High skilled manual non-manual manual non-manual Jan 95-Oct 96 0.709 0.645 0.740 0.587 0.770 0.753 Nov 96-Nov 98 0.646 0.687 0.780 0.604 0.743 0.748 Difference -0.063 0.042 0.040 0.017 -0.027 0.005 Outflows as a share of inflows (Table 13.5) appear to have increased in the ETU Scheme A and Scheme B areas for unskilled single males and skilled single males and, with the exception of low-skilled manual single males in Scheme A areas, a decrease in outflows for low skilled and high skilled single males. 147 Table 13.5 Outflows as a share of inflows by ETU area type and skill group - single males ETU area type ETU A/ETU C ETU B/ETU C Unskilled 0.024 (0.011) 0.012 Low skilled (manual) Low skilled (non-manual) 0.010 (0.019) 0.000 (-0.003) -0.028 -0.029 Skilled (manual) Skilled (non-manual) 0.023 (0.014) 0.026 (0.034) 0.010 0.060 High skilled -0.001 (-0.016) -0.008 Note: the figures in brackets exclude the seaside areas. Flows into unemployment for single females have fallen in Scheme A (Table 13.6) and Scheme B (Table 13.7) areas relative to the comparison areas for the unskilled, skilled and highly skilled. Small increases for low skilled females may indicate the presence of substitution effects. Excluding the seaside areas suggests that the effects in Scheme B areas are greater than in Scheme A areas. Overall the impact of ETU on inflows into claimant unemployment for single females is greater than for single males. Table 13.6 Flows into unemployment in ETU A areas relative to ETU C areas by skill group - single females Unskilled Low skilled Skilled High skilled manual non-manual manual non-manual Jan 95-Oct 96 Nov 96-Nov 98 1.534 1.357 1.495 1.634 1.469 1.477 1.675 1.519 1.933 1.781 1.749 1.657 Difference -0.176 0.139 0.007 -0.156 -0.152 -0.092 (-0.111) (0.109) (0.037) (-0.221) (-0.085) (-0.024) Note: the figures in brackets exclude the seaside areas. Table 13.7 Flows into unemployment in ETU B areas relative to ETU C areas by skill group - single females Unskilled Low skilled Skilled High skilled manual non-manual manual non-manual Jan 95-Oct 96 Nov 96-Nov 98 0.814 0.669 0.595 0.662 0.693 0.710 0.721 0.801 0.878 0.789 0.818 0.695 Difference -0.144 0.066 0.017 0.080 -0.089 -0.123 Increases in outflows for unskilled single females and skilled manual females are found in ETU Scheme A and Scheme B areas relative to the comparison areas (Table 13.8). A fall in relative outflows was recorded for low skilled, skilled non-manual in Scheme A areas and high skilled single females. The impact on single females is greater than for single males. 148 Table 13.8 Outflows as a share of inflows by ETU area type and skill group - single females ETU area type ETU A/ETU C ETU B/ETU C Unskilled 0.071 (0.059) 0.051 Low skilled (manual) Low skilled (non-manual) -0.037 (-0.025) 0.001 (-0.014) -0.105 -0.067 Skilled (manual) Skilled (non-manual) 0.080 (0.079) -0.014 (-0.020) 0.093 0.000 High skilled -0.005 (-0.009) -0.005 Note: the figures in brackets exclude the seaside areas. A fall in relative outflows from unemployment may not be a good thing if it is accompanied by an increase in flows into inactivity rather than an increase in the flows into work. While there is no reason to expect that ETU would lead to greater outflows into inactivity it is still worth checking whether or not the greater outflows were into non-work options. The destination of leavers from unemployment in ETU Scheme A areas, Scheme B areas and the comparison areas were compared, looking specifically at the proportion who are known to have entered work. Comparison of the pre-ETU period with the post-ETU period shows an overall decline in the proportion of claimants who leave unemployment and are known to have found work: from 58 per cent to 55 per cent. It was noted earlier that ETU was introduced in the pilot areas at the same time as the introduction of Jobseeker’s Allowance (JSA) in October 1996. It has been shown elsewhere (Sweeney and McMahon, 1998) that in the post JSA period a smaller proportion of claimants leaving unemployment benefit enter work. A comparison between destinations in the ETU pilot and comparison areas shows that the relative flows of claimants into work is unchanged between the pre-ETU and the post-ETU periods. 13.3 The variation in the impact of ETU on unskilled unemployment benefit claimants by age group and partnership status An interesting question is whether or not the impact of ETU varied for different age groups. There were two rates of credit for single people aged 18-24 and those 25 plus with the more generous payment to the older age group. It has also been suggested that an employment tax credit may be introduced for older rather than younger workers as younger workers tend to experience greater upward earnings mobility. There is also a form of credit now available for some older workers via the New Deal for 50 plus. From April 2000 Unemployment Benefit claimants aged 50 or over who have been claiming for at least six months qualify for a £60 a week credit (tax free) when they take a full-time job and £40 for a part-time job. The Credit is paid for up to 52 weeks providing the recipient stays in work. In this section the analysis concentrates on unskilled claimants and provides separate estimates for males and females and by partnership status. The definition of partnership status adopted identifies claimants who are single and a second group classified as nonsingle, which includes married, divorced, separated, cohabiting or 149 widowed. Individuals in the former group are more likely to qualify for ETU than those in the latter group as individuals in the latter group are more likely to have dependent children. Analysis of the estimated impact of ETU on unskilled single claimants by age at the start of their claim shows a reduction in inflows for males and females in all age groups (Table 13.9) with the exception of the 35-44 age group (for males in Scheme A areas there is a reduction in inflows when the seaside areas are removed). This may be due to substitution effects as the younger and older age groups are more likely to be low paid and therefore benefit more from ETU or this group, even though they are single, may have dependent children and therefore not qualify for ETU. The greatest reductions in inflows are found among the youngest age group (16-24 years) and the oldest age groups (this does not apply to females in Scheme A areas where a slightly different picture emerges). The negative impact on inflows is generally greater in the Scheme B areas (the more generous version of ETU) than the Scheme A areas when the seaside areas are excluded from the Scheme A areas. Table 13.9 Flows into unemployment for unskilled single males and females by age at start of claim Males Females ETU ETU ETU ETU Age group A/ETU C 16-24 years 25-34 years -0.059 (-0.069) -0.037 (-0.082) B/ETU C A/ETU C B/ETU C -0.078 -0.040 -0.297 (-0.120) -0.305 (-0.246) -0.210 -0.221 35-44 years 45-54 years 0.038 (-0.055) -0.021 (-0.076) -0.023 -0.055 0.193 (0.432) -0.157 (-0.176) 0.125 -0.503 55+ years -0.385 (-0.555) -0.060 Notes: the figures in brackets exclude the seaside areas. The figures for females aged 55+ have been excluded due to small sample size. For non-single males reductions in inflows are observed for all age groups with the exception of the oldest age group (55+) in the Scheme A areas when the seaside areas are removed (Table 13.10). In the Scheme B areas there are reductions for all ages with the exception of the 45-54 age group. The negative impact falls with age up until age 45. In the Scheme B areas the impact on inflows is greater for single males than for nonsingle males. A slightly different picture emerges for non-single females. In the Scheme A areas there is a reduction in inflows for the youngest age group but increases for all other age groups (whether or not the seaside areas are included) with the increase rising with age. In the Scheme B areas there are reductions in inflows for all age groups up to 35-44 years and an increase in the 45-54 age group. The greatest reduction in inflows is found in the 25-34 age group. The impact for single females is greater than for non-single females which is not surprising given the fact that some of the non-single females will not qualify for ETU due to the 150 presence of dependent children. There is also some evidence of substitution effects between single and non-single females. Table 13.10 Flows into unemployment for unskilled nonsingle males and females by age at start of claim Males Females ETU ETU ETU ETU Age group A/ETU C B/ETU C A/ETU C B/ETU C 16-24 years 25-34 years -0.214 (-0.293) -0.004 (-0.121) -0.071 -0.033 -0.049 (-0.106) 0.011 (0.009) -0.044 -0.129 35-44 years 45-54 years 0.057 (-0.056) 0.039 (-0.009) -0.015 0.007 0.060 (0.151) 0.272 (0.356) -0.032 0.031 55+ years 0.033 (0.051) -0.052 Notes: the figures in brackets exclude the seaside areas. The figures for females aged 55+ have been excluded due to small sample size. Table 13.11 shows a consistent picture of increases in outflows in the pilot areas after the introduction of ETU for single males and females in the youngest age group (16-24 years) with a greater impact in Scheme B areas where the more generous version was being piloted (when the seaside areas are removed from the Scheme A areas). Increases in outflows are also found for the 25-34 age group, with the exception of males in Scheme B areas, the 45-54 age group and the 55+ age group for males. There is evidence of substitution effects working against the 35-44 age group (this may be partly due to the fact that claimants in this age group are more likely to have dependent children even though they are single). Overall the effect on females is greater than the effect on males and, when the seaside areas are excluded from the Scheme A areas, the impact in Scheme B areas is greater than in Scheme A areas. Table 13.11 Outflows as a share of inflows for unskilled single males and females by age at start of claim Males Females ETU ETU ETU ETU Age group A/ETU C B/ETU C A/ETU C B/ETU C 16-24 years 0.017 (0.006) 0.017 0.113 (0.047) 0.094 25-34 years 35-44 years 0.030 (0.018) 0.018 (-0.012) -0.001 0.001 0.193 (0.109) 0.046 (-0.101) 0.189 -0.099 45-54 years 55+ years 0.067 (0.062) 0.153 (0.134) 0.010 0.176 0.315 (0.296) 0.241 Notes: the figures in brackets exclude the seaside areas. The figures for females aged 55+ have been excluded due to small sample size. For non-single males and females (Table 13.12) there are increases in outflows for the 16-24 and 25-34 age groups with the largest increases in the younger age group. The 35-44 age group for men and women and the 45-54 age group for men experience reductions in outflows in the 151 pilot areas. This may be due to substitution effects as these age groups are more likely to have dependent children and are less likely to be in low paid jobs prevalent among the younger age groups and the oldest age group for which there are increases in outflows. Table 13.12 Outflows as a share of inflows for unskilled nonsingle males and females by age at start of claim Males Females ETU ETU ETU ETU Age group A/ETU C 16-24 years 25-34 years 0.102 (0.112) 0.045 (0.051) B/ETU C A/ETU C B/ETU C 0.076 0.056 0.065 (0.077) 0.083 (0.031) 0.191 0.021 35-44 years 45-54 years -0.002 (0.000) -0.029 (-0.057) -0.007 -0.019 -0.018 (-0.064) 0.059 (0.036) -0.063 0.072 55+ years 0.079 (0.076) 0.090 Notes: the figures in brackets exclude the seaside areas. The figures for females aged 55+ have been excluded due to small sample size. The analysis of inflows and outflows for unskilled, unemployed benefit claimants by age group suggests that ETU has had a greater impact on single claimants than non-single claimants, females more than males and Scheme B areas more than Scheme A areas. The younger age groups and the older age groups experienced greater reductions in inflows and greater increases in outflows than the 35-44 age group. This age group experienced increases in inflows and decreases in outflows characteristic of a substitution effect. There is also some evidence of substitution of single claimants in favour of non-single claimants who were less likely to qualify for ETU. 13.4 The variation in the estimated impact of ETU on unskilled single male and female claimants by area type Finally we look at the areas in more detail to see if there is any local variation between area types. Table 13.13 shows the change in inflows for single unskilled males and females by area type and Table 13.14 shows the change in net outflows. Table 13.13 Flows into unemployment for unskilled single males and females by area type Males 152 Females ETU ETU ETU ETU Age group A/ETU C B/ETU C A/ETU C B/ETU C Urban Large town 0.013 0.033 -0.172 0.078 0.018 0.257 -0.117 0.056 Seaside Rural 0.012 -0.288 0.028 -0.111 -0.460 -0.675 -0.058 -0.512 Table 13.14 Outflows as a share of inflows for unskilled single males and females by area type Males Age group Females ETU ETU ETU ETU A/ETU C B/ETU C A/ETU C B/ETU C Urban 0.022 0.023 0.026 0.043 Large town Seaside -0.031 0.084 -0.025 0.070 -0.003 0.152 0.018 -0.062 Rural 0.041 -0.007 0.179 0.180 In the urban areas there is a large negative impact on inflows for single unskilled males and females in the Scheme B area (it has been noted elsewhere that this area - Sunderland - had the highest ETU take-up rate). However, there is a small positive increase in inflows in the Scheme A area. Outflows increase in the Scheme A and Scheme B areas and the greatest increase is found among unskilled single females in the Scheme B area. In the large towns increases in relative inflows in Scheme A and Scheme B areas are observed for both males and females. This is due to the relative decline in inflows and outflows in the comparison area (Rotherham and Worksop) shown in Chapter 12 (Figures 12.6 and 12.7). Consequently there is a decline in inflows in the Scheme A and Scheme B areas relative to the comparison area. However, there is an increase in net outflows for single unskilled females. Seaside areas show increases in inflows for unskilled single males in Scheme A and Scheme B areas but large decreases for unskilled single females. The large decline in the Scheme A areas is due to the growth in employment for residents of Southend rather than the result of ETU (take-up of ETU was very low in Southend). Although an increase in inflows is observed for single unskilled males in the Scheme A and Scheme B areas this is accompanied by increases in outflows relative to inflows. Single unskilled females in the Scheme A area have an increase in outflows (the ‘Southend effect’) but a decrease in outflows in the Scheme B area. The rural areas show the most consistent picture with the largest ‘ETU effect’. There are large reductions in inflows and, with the exception of the Scheme B area for single unskilled males, increases in outflows. The lower average wages in rural areas (Green, 2001) may be an explanation for why ETU appears to have had the greatest impact in rural areas. Overall the detailed area analyses for unskilled single men and women show some local variation. The findings are consistent with what we would expect given the findings coming out of other parts of the evaluation of ETU. The high take-up rate in Sunderland is reflected in the decreases in inflows and increases in outflows for single unskilled men and women. 153 The employment growth in and around Southend has led to large decreases in inflows (except for unskilled single males) and increases in outflows, which cannot all be ascribed to ETU. The decline in inflows and outflows in Rotherham and Worksop (the large town comparison area) after April 1998 results in increases in relative inflows and decreases in relative outflows in the ETU Scheme A and Scheme B areas. The rural areas show a fairly consistent picture of decreases in relative inflows in the ETU pilot areas and increases in relative outflows. 154 14 MOVEMENTS BETWEEN UNEMPLOYMENT BENEFIT AND EARNINGS TOP-UP In addition to the JUVOS database a special dataset was constructed from the administrative records of ETU claims identifying all claims made by individuals in the JUVOS dataset via their National Insurance number. Linking these two datasets allows us to identify transitions between Unemployment Benefit and Earnings Top-up. In total the ETU database identifies 70,716 claims for ETU made between 8 October 1996 and 23 November 199912. Although nearly three-quarters of unemployment benefit claims are made by men (see Chapter 11), 45 per cent of ETU claimants moving from unemployment were women, which may explain why ETU appeared to have a greater impact on female unemployment than male unemployment. Table 14.1 shows the total number of claims made by these men and women. Females were more likely to make more than one claim for ETU than males with more than half (55 per cent) making two or more claims for ETU compared with 49 per cent of males. Table 14.1 Total number of claims for ETU by gender Number of claims Males Females All 1 2 51% 25% 45% 25% 48% 25% 3 4 13% 7% 15% 8% 14% 8% 5 6 4% 1% 4% 2% 4% 2% Total (=100%) 32,076 38,640 70,716 Source: ETU administrative micro dataset 1996-1999 The average claim for ETU made by men in the JUVOS dataset was £28 per week, they worked for 30 hours per week and were paid £2.83 per hour. In contrast the average claim made by women in the JUVOS dataset was £24 per week, they worked 26 hours per week and were paid £3.22 per hour. A much higher proportion of the women making claims for ETU were single (91 per cent) compared with men (85 per cent)13. Overall the proportion of ETU claimants who were part of a couple was lower in the JUVOS dataset than among all ETU claimants (12 per cent compared with 14 per cent). This high proportion of single claimants among the JUVOS population justifies the earlier emphasis on single unemployment benefit claimants. 12 Note that the total number of claims for ETU could take place before or after the claims for Unemployment Benefit and may or may not be sequential. 13 A similar figure is found when we look at all claims rather than claimants. 155 In the ETU pilot areas 588,904 claims for unemployment benefit ended between October 1996 and November 1998. Of these claimants, 2.7 per cent (16,135) moved from unemployment benefit into work supplemented by ETU14. Sixty per cent went on to make a second claim for ETU (one year of supplemented employment), 37 per cent made a third claim for ETU (1.5 years of supplemented employment), 21 per cent made a fourth claim (two years of supplemented employment) for ETU and 10 per cent made five or more claims for ETU (at least 2.5 years of supplemented employment). Assuming that 55 per cent of claimants leaving unemployment benefit move directly into work, five per cent (323,897) of unemployment benefit claimants who found work had their income supplemented by ETU. If we restrict the analysis to single claimants, as single claimants were more likely to claim ETU, we find that five per cent of all claims for unemployment benefit ending in this period (October 1996-November 1998) involve a transition to ETU. Assuming that 55 per cent of claimants leaving unemployment benefit move directly into work, nine per cent of single unemployment benefit claimants who found work had their income supplemented by ETU. Figure 14.1 shows the number of claims for unemployment benefit made by single claimants that end in a claim for Earnings Top-up for each month between October 1996 and November 1998. The dips in December are associated with the small number of unemployment benefit claims that end in this month. Figure 14.1 Claims for unemployment benefit ending in a claim for ETU - single claimants 14 156 A claim for ETU starting within four weeks of a claim for Unemployment Benefit ending is considered to be a transition between Unemployment Benefit and ETU. It was shown in Chapter 13 that ETU appeared to have a differential effect on flows into unemployment benefit and flow off Unemployment Benefit by skill group. To look at this relationship in terms of actual claims for ETU the proportion of single unemployment benefit claimants moving between Unemployment Benefit to ETU is computed for different skill groups and by gender (Table 14.2). Table 14.2 Proportion of claims for unemployment benefit ending that involved a transition to ETU by gender and area type - single claimants Gender ETU areas Skill level Scheme A Scheme B Total Females Males High skilled Skilled non-manual 1.9 3.3 2.0 2.9 2.0 3.0 Skilled manual Low skilled non-manual 5.9 5.5 2.4 3.2 2.5 4.2 Low skilled manual Unskilled 6.3 8.7 3.2 2.7 3.5 3.4 Total 5.2 2.8 3.5 High skilled Skilled non-manual 4.2 7.3 3.8 6.5 3.9 6.7 Skilled manual Low skilled non-manual 13.6 12.5 5.3 7.8 5.6 10.0 Low skilled manual Unskilled 12.4 16.8 8.1 6.6 8.6 7.6 Total 11.7 6.7 8.0 Source: ETU administrative micro dataset 1996-1999; JUVOS micro data 1995-1998 A greater proportion of unemployment benefit claims ending in the Scheme B areas led to claims for ETU than in the Scheme A areas (eight per cent compared with 3.5 per cent). This was undoubtedly due to the greater generosity of ETU piloted in the Scheme B areas. In both areas females moving off unemployment benefit were more likely to move onto ETU than males. In the Scheme A areas 5.2 per cent of females compared with 2.8 per cent of males and in the Scheme B areas 11.7 per cent of females compared with 6.7 per cent of males moved onto ETU at the end of their claim for unemployment benefit. Not surprisingly, a higher proportion of claimants with lower levels of skill moved from unemployment benefit into a job with ETU than claimants with higher levels of skill. Females in the lowest skill group were four times as likely to move from unemployment benefit to ETU than females in the highest skill group (8.7 per cent compared with 1.9 per cent in the Scheme A areas and 16.8 per cent compared with 4.2 per cent in the Scheme B areas). Such a gradient was not evident for males in the ETU Scheme A areas, however, a gradient was observed among men in the Scheme B areas with the largest proportion of claims ending in a claim for ETU found among low skilled manual males (8.1 per cent). 157 The analysis of the transitions from unemployment benefit to Earnings Top-up has provided additional supporting evidence for the increases in flows off unemployment benefit and decreases in the flows onto unemployment benefit found in Chapter 13 that can be attributed to ETU. 158 15 CONCLUSIONS In this part of the evaluation of Earnings Top-up, an analysis of Unemployment Benefit claims before and after the introduction of ETU has been conducted. The results indicate that after the introduction of ETU, inflows to unemployment in the pilot areas relative to comparison areas fell. There also appears to be an increase in outflows as a share of inflows after the introduction of ETU in the pilot areas. It has been shown, with the aid of a skill classification of Unemployment Benefit claimants, that the period during the ETU pilot was associated with falls in inflows into unemployment in the pilot areas relative to the comparison areas for unskilled claimants in the younger and older age groups. Unskilled Unemployment Benefit claimants also appear to have benefited in terms of an increase in outflows relative to inflows. However, some of these benefits may have been at the expense of the low skilled group, couples and those aged 35-44. ETU appears to have had a greater impact among single women than men and individuals living in rural areas where low pay is prevalent. An analysis of transitions between Unemployment Benefit and claims for ETU supports these findings. A larger proportion of women than men completing spells of unemployment moved into a job supported by ETU. These transitions are greater in the ETU Scheme B areas (where the more generous version of ETU was being piloted) and for lower skilled benefit claimants. ETU claimants moving from unemployment go on to make a larger number of subsequent claims compared with all ETU claimants. The large number of subsequent claims may explain why the introduction of ETU has led to a reduction in flows into unemployment as well as flows out of unemployment. Overall the results suggest that an in-work benefit for single people and couples without dependent children can not only raise incomes of low paid workers but also reduce the harmful churning at the lower end of the labour market. Recent research (Gregg, 2000) has shown that the experience of unemployment early in an individual’s career (even after controlling for individual specific characteristics) is associated with poorer outcomes in later life. The benefits of ETU to the younger age group (16-24 years) suggest that any future employment tax credit could benefit this group in the short and longer term. 159 APPENDIX A A1 Sampling UNEMPLOYED SURVEY 1998 SAMPLING INFORMATION The sample for the 1998 Unemployed Survey was drawn from the Department of Social Security Departmental Central Index (DCI) on 30th June 1998 for all individuals that met the following criteria: • currently in receipt of either Jobseeker’s Allowance, Income Support, Training for Work Allowance, or any combination of these benefits. • aged 18 to 63 years inclusive; • claiming benefit for 26 to 65 weeks inclusive; • a residential address containing a postcode prefix for the eight ETU and four control areas. As people with dependent children would not be eligible for ETU, the sample was then checked against Child Benefit records to exclude individuals who were receiving Child Benefit. Cases were then sorted by ETU area and the appropriate number of cases selected to get a random sample of 750 cases per ETU/control area. Therefore, despite the different sizes of the local labour markets and rates of unemployment, the aim was to have equal number of respondents for each of the twelve ETU or Control areas. Excluding people aged under 18 and over 63 and those in receipt of Child Benefit was done at an earlier stage for the 1998 survey to try and reduce the number of doorstep checks by interviewers to establish eligibility for the survey. In 1996 more than half (54 per cent) of people contacted for the unemployed survey were not eligible to take part. These individuals had been selected from the DCI in May 1996 if they were in receipt of Unemployment Benefit, Income Support, Employment Training Allowance, or any combination of these benefits and had a residential address with an appropriate postcode. From then, individuals with a claim duration of between 26 and 65 weeks inclusive were selected and again, cases were then sorted by ETU area and the appropriate number of cases selected to get a random sample of 750 cases per ETU/control area. Interviewers then checked whether people were of the correct age and if they had children when they first made contact. This slight change in sampling procedure for the 1998 survey should not have affected the composition of the final sample interviewed. A2 Response rates Nine thousand people were initially selected and were sent a letter from the Department of Social Security giving them details of the survey and a telephone number to contact if they did not wish to take part in the study. Around 13 per cent of these people opted out of the study giving a final issued sample of 7,813. As the Child Benefit screen should have removed many non-eligible people from the sample, not all cases were 161 needed. The processed sample was 4,830 cases, drawn evenly from all 12 areas. Doorstep screening found that 922 people were ineligible to take part, usually because they had dependent children living with them. This was 19 per cent of people contacted, suggesting that the Child Benefit screen for this survey had reduced the number of ineligible people by half compared with the 1996 survey. This was as expected: most Child Benefit claims are in one person’s name (usually the mother) so the Child Benefit screen would not remove all people with children from the sample. Despite calling at least four times, interviewers were not able to contact 611 people. If these are included, the response rate is 67 per cent. However, some of these non-contacts would have been ineligible for the survey. Excluding non-contacts gives a response rate of 82 per cent. The true underlying response rate lies within this range. If the proportion of ineligible people was similar for those not contacted this would give an overall response rate of 69 per cent. Table A.1 Analysis of response rate – unemployed survey 1998/99 Base Percentage Issued sample 1998 Screened as ineligible 4830 922 100 19 Not contacted 1247 26 - Moved/untraceable - Ill/dead/away for survey period 516 91 11 2 - Late opt out - Not available after four or more calls 29 611 1 13 Contacted but not interviewed 574 12 Refused Other reason for non-interview 399 75 8 2 Achieved interviews 1998 2187 45 Response rate 1998 (upper limit) 2661 82 Response rate 1998 (lower limit) 3272 67 Refused to be contacted after first interview 192 Continued 162 Table A.1 Continued Base Percentage Issued sample 1999 1995 100 Not contacted 495 25 - Moved/untraceable - Ill/dead/away for survey period 197 19 10 1 - Late opt out - Not available after four or more calls 13 67 1 3 Not processed during fieldwork period 199 10 Contacted but not interviewed 191 10 Refused Other reason for non-interview 135 56 7 3 Achieved interviews 1999 1309 66 Response rate 1999 (upper limit) 1500 87 Response rate 1999 (lower limit) 1958 67 Note: Upper limit response rate excludes all those not contacted and so calculates the proportion of completed interviews as a proportion of the contacted eligible. Lower rate response rate includes those not available after 4+ calls, non-processed cases and, for 1999 survey, all those who refused after first interview. Both exclude those screened as ineligible for survey. The aim for 1999 was to re-interview all those interviewed in 1998. To encourage this, a £10 incentive was paid to participants at second interview. However, 192 respondents said they did not wish to be contacted again at the end of the first interview so these people were not part of the issued sample in 1999. Some respondents were not contacted because they were untraceable or unavailable during the survey period (12 per cent). Another three per cent were not available after four or more attempts by the interviewer. Unfortunately, 199 cases (10 per cent) were not processed in the fieldwork period so it is not known whether these people would have been contactable or not. The true response rate for 1999 lies somewhere between 67 and 87 per cent, depending on how the refusals after first interview and the non-contacts are treated. However, it is closer to the lower limit than the higher limit. A3 Response bias Overall, we have a second interview for 60 per cent of those who completed a first interview. Fortunately, the missing information appeared to be reasonably random as there was no obvious response bias. Tables A.2 and A.3 show the characteristics of respondents in 1998 by those who were interviewed in 1998 and at 1999. 163 Table A.2 Characteristics of respondents in 1998 at each interview Column percentages 1998 Interview 1999 Interview Gender Male Female 69 31 66 34 Age-group 18 – 24 25 – 34 35 – 44 45 – 54 55+ 19 18 16 24 21 18 16 17 28 21 Marital status Single Separated/widowed/divorced Partnered 48 23 29 47 22 31 Tenure Owner-occupier Lives with parents Tenant Other 18 30 44 8 20 31 44 5 Household type Lives alone Lives with partner only Lives with parents, no partner Other 36 21 24 19 31 23 25 21 Ethnic group White Other 99 1 99 1 Highest academic qualifications None GCSE D-G or equivalent GCSE A-C or equivalent GCE A level Degree or higher degree 65 13 15 4 4 65 14 14 4 3 Vocational qualifications Yes No 37 63 38 62 Driving licence Yes No 42 58 40 60 Economic status at first interview In work (16+ hours) Not in work 7 93 7 93 2187 1309 Base 164 The main few findings of interest were that slightly fewer of those who had been living alone in 1998 were interviewed in 1999, perhaps as they were more difficult to contact. There was also a slight shortfall in second interviews in Bournemouth. Table A.3 Sample characteristics in 1998 at each interview Column percentages 1998 Interview 1999 Interview Sample type JSA 72 74 IS 28 26 ETU scheme type Scheme A 35 35 Scheme B Control areas 33 32 31 33 Newcastle Castleford 8 9 10 9 Southend North Wales 9 9 7 9 Sunderland Doncaster 8 8 10 9 Bournemouth Scotland 8 8 5 8 Middlesbrough Southampton 9 7 9 7 South Wales Rotherham 8 9 9 8 2187 1309 Area Base 165 APPENDIX B RESULTS FROM THE LOGISTIC REGRESSION MODEL All respondents living in areas where ETU was available were asked the question: ‘A new social security benefit was introduced in 1996 in some areas of the country that pays extra money to some people who work and have no dependent children living with them. Have you heard of the introduction of this benefit?’ People who said that they had heard of the introduction of this benefit were then asked: ‘What is this benefit called?’ This logistic regression model estimates the probability of a respondent being able to correctly identify this benefit as ETU. Table B.1 Logistic regression model: probability of being able to name ETU Odds Standard 95 per cent ratio error P Area Newcastle Sunderland Doncaster Southend Bournemouth North Wales Perth and Stirling 3.64396 7.76827 4.59739 1.52951 3.24615 2.95367 2.03783 1.00846 1.95980 1.22878 0.54882 0.96009 0.84193 0.66943 0.000 0.000 0.000 0.232 0.000 0.000 0.030 2.12326 4.73784 2.72274 0.76186 1.81809 1.68939 1.07040 6.27247 12.73703 7.76279 3.07067 5.79594 5.16410 3.87962 Age group Age 18 – 24 Age 25 – 34 Age 35 – 44 Age 45 – 54 2.40697 2.27756 1.70326 1.97934 0.68152 0.65497 0.51717 0.54253 0.002 0.004 0.079 0.013 1.38184 1.29625 0.93935 1.15667 4.19259 4.00177 3.08841 3.38712 Marital status Has partner 0.62500 0.12064 0.011 0.40693 0.89206 Economic status at interview In work (not ETU job) 1.51795 Ill-health 0.54320 Other 0.89427 0.36049 0.17963 0.19256 0.079 0.065 0.604 0.95304 0.28410 0.58636 2.41771 1.03860 1.36378 Ever lived in a household that has received Family Credit Yes 1.18749 0.28977 0.481 0.73607 1.91575 Educational level Academic qualifications Vocational qualifications 0.75701 0.75060 1.79259 1.50961 1.16579 1.06448 0.25682 0.18974 0.486 0.726 confidence interval Reference group is single person aged 55 or over living in the Barnsley area (Scheme A) and who was unemployed at time of interview. They had no qualifications and no previous experience of Family Credit. 167 APPENDIX C RESULTS FROM THE MODELS IN CHAPTER 5 Table C.1 Modelling exits from unemployment since introduction of ETU (1) (2) (3) Employed Training Inactive Scheme A area 0.042 0.159 0.266 Scheme B area (0.26) -0.108 (0.66) -0.033 (0.61) 0.447 Female (0.63) 0.050 (0.13) -0.489 (1.02) 0.297 Age (0.32) -0.030 (1.86) -0.047 (0.83) 0.013 (4.05)** -0.419 (3.92)** -0.990 (0.81) 0.291 If partnered, partner employed (1.93) 0.602 (2.25)* 0.709 (0.71) -1.141 Qualifications: vocational only (1.93) 0.368 (1.15) 0.700 (1.07) -0.242 Qualifications: academic only (1.73) 0.221 (2.03)* 0.348 (0.46) -0.077 Qualifications: vocational and academic (1.14) 0.181 (1.08) 0.821 (0.15) 0.251 Full driving licence (0.98) 0.266 (2.83)** -0.145 (0.56) -0.183 Accommodation: owned (1.81) 0.815 (0.65) 0.881 (0.49) 0.119 (2.58)** 0.143 (1.72) 0.470 (0.19) 0.153 (0.54) (1.35) (0.27) 0.330 (1.49) 0.095 (0.30) -0.098 (0.19) Accommodation: rented (non-zero amount after rebate) 0.499 0.133 -0.501 Accommodation: mortgage (1.97)* 0.368 (0.33) 0.128 (0.75) -0.084 Accommodation: other (1.14) 0.736 (0.21) 0.024 (0.13) 0.184 Length of unemployment spell (2.19)* -0.011 (0.04) 0.001 (0.23) -0.005 Month (1.77) 0.031 (0.10) 0.032 (0.34) 0.030 (3.96)** -5.44 (2.68)** -5.82 (1.52) -8.65 (9.06)** (6.37)** (5.73)** 14466 14466 14466 Partnered Accommodation: lives with parents, rent-free Accommodation: lives with parents rent-free, paying rent Constant Base Absolute value of z-statistics in parentheses. * significant at 5% level; ** significant at 1% level 169 Table C.2 Modelling exits from unemployment since 1998 interview (1) (2) (3) Employed Training Inactive Scheme A area 0.134 -0.247 -0.158 Scheme B area (0.45) 0.102 (0.54) -0.644 (0.24) -0.059 Female (0.33) 0.306 (1.32) -0.833 (0.09) 0.351 Age (1.16) -0.015 (1.63) -0.059 (0.60) -0.027 Partnered (1.18) -0.299 (2.62)** -0.099 (0.94) -1.611 Qualifications: vocational only (0.93) 0.375 (0.19) 0.616 (1.45) -0.828 Qualifications: academic only (1.04) 0.443 (0.93) 0.468 (0.74) 0.056 Qualifications: vocational and academic (1.32) 0.187 (0.78) 1.011 (0.08) 0.257 Full driving licence (0.58) 0.380 (1.87) 0.091 (0.38) -0.519 Accommodation: owned (1.48) 1.210 (0.22) 0.802 (0.82) -41.169 Accommodation: lives with parents, rent-free (2.42)* 0.410 (0.92) 0.555 (0.00) 0.343 (0.96) (0.99) (0.42) 0.031 (0.08) -0.597 (1.02) -0.186 (0.26) Accommodation: rented (non-zero amount after rebate) 0.654 -1.094 0.097 Accommodation: mortgage (1.57) 0.433 (1.00) 0.650 (0.11) -0.243 Accommodation: other (0.82) 0.170 (0.72) 0.704 (0.20) -41.652 Length of unemployment spell (0.26) -0.008 (0.81) -0.004 (0.00) 0.009 Month (0.92) 0.061 (0.27) -0.044 (0.46) -0.029 Expected wage at time of 1998 interview (1.85) -0.288 (0.88) -0.226 (0.40) 0.299 Aware of ETU at time of 1998 interview (2.12)* -0.157 (1.07) 0.135 (1.15) -0.455 (0.51) (0.29) (0.65) -0.653 (1.22) -0.538 (0.69) 1.267 (1.71) -7.371 (2.95)** 1.196 (0.32) -3.338 (0.62) 4142 4142 4142 Accommodation: lives with parents rent-free, paying rent Caring for someone with long term illness at time of 1998 interview Constant Base Absolute value of z-statistics in parentheses. * significant at 5% level; ** significant at 1% level 170 Table C.3 Modelling wages Log of pay Scheme A area Scheme B area Female Partnered If partnered, partner employed Age Age squared Qualifications: vocational only Qualifications: academic only Qualifications: vocational and academic Full driving licence Receiving IS at time of 1998 interview Percentage of time employed since 1993 SEG: prof/man/tech (1-3) SEG: clerical SEG: craft SEG: pers serv/sales SEG: plant ops SEG: other Type of area: rural-isolated Type of area: rural-village Type of area: large town or city, suburban Type of area: large town or city, innercity Selection adjustment variable Constant Base -0.149 (1.97)* -0.288 (3.50)** -0.181 (2.46)* -0.078 (0.81) 0.178 (1.20) 0.045 (1.84) -0.001 (1.94) 0.346 (3.52)** 0.359 (3.73)** 0.239 (2.63)** 0.000 (0.00) 0.122 (0.89) 0.220 (1.03) 0.089 (0.66) -0.047 (0.41) 0.232 (1.85) -0.057 (0.63) 0.075 (0.74) 0.249 (1.77) -0.248 (0.97) -0.054 (0.64) -0.029 (0.40) 0.193 (1.12) -.004 (0.02) 3.848 (11.56)** 1272 Absolute z-statistics in parentheses. * significant at 5% level ** significant at 1% level 171 Table C.4 Modelling whether job provides training Training Scheme A area -0.093 Scheme B area (1.00) 0.029 Female (0.30) 0.010 Age (0.11) 0.014 Age squared (0.63) 0.000 Qualifications: vocational only (0.37) 0.009 Qualifications: academic only (0.07) 0.049 Qualifications: vocational and academic (0.40) -0.224 Full driving licence (2.18)* -0.164 Percentage of time employed since 1993 (1.83) -0.992 SEG: prof/man/tech (1-3) SEG: clerical SEG: craft SEG: pers serv/sales (6.92)** 0.142 (1.13) 0.369 (3.14)** 0.175 (1.43) 0.230 SEG: plant ops (3.16)** 0.260 SEG: other (2.35)* -0.006 Type of area: rural-isolated (0.06) -0.098 Type of area: rural-village (0.74) 0.118 Type of area: large town or city, suburban (1.40) 0.056 Type of area: large town or city, innercity (0.82) -0.198 Constant (1.09) 0.282 (0.71) Base Absolute value of z-statistics in parentheses. * significant at 5% level; ** significant at 1% level 172 1272 Table C.5 Modelling expected wages Expected hourly wage Scheme A area Scheme B area Dummy variable indicating year = 1999 Female Partnered If partnered, partner employed Age Age squared Living with parent(s) Long-standing illness, disability or infirmity Qualifications: vocational only Qualifications: academic only Qualifications: vocational and academic Full driving licence Percentage of time employed since 1993 SEG: prof/man/tech (1-3) SEG: clerical SEG: craft SEG: pers serv/sales SEG: plant ops SEG: other Type of area: rural-isolated Type of area: rural-village Type of area: large town or city, suburban Type of area: large town or city, innercity Selection adjustment variable Constant Base 0.131 (1.51) -0.008 (0.09) 0.246 (3.67)** -0.318 (3.46)** -0.010 (0.11) 0.023 (0.13) 0.079 (3.65)** -0.001 (3.26)** -0.485 (5.25)** -0.082 (1.04) 0.005 (0.05) 0.063 (0.58) 0.171 (1.66) 0.299 (3.77)** -0.320 (2.30)* -0.052 (0.33) 0.196 (1.38) 0.020 (0.14) 0.126 (1.24) -0.084 (0.77) 0.046 (0.31) 0.258 (0.89) -0.046 (0.51) -0.016 (0.23) 0.112 (0.66) 0.883 (10.14)** 1.692 (3.97)** 1059 Absolute z-statistics in parentheses. * significant at 5% level; ** significant at 1% level 173 C1 Technical details of approach used in modelling wages and expected wages Both wages and expected wages were estimated using a maximum likelihood selection model. This provides different estimates from those given by a two-step model, but is to be preferred since it takes full account of the correlation between the selection mechanism and the wage equation. In fact, the results showed the selection adjustment (l) to be insignificant for the wages model. Interestingly, the two-step estimates returned an insignificant ETU effect in this case. This is due to the adjustment to standard errors needed in the two-step approach; the point estimates were essentially identical under the two approaches. The exception is the point estimate of l which was larger in absolute terms under the two-step method: this estimate is used in the adjustment of standard errors. With expected wages, l was found to be very significant. In order to examine the robustness of the results to outliers, those individuals with modelled residuals in the top or bottom one per cent of the distribution were discarded and the model re-estimated. This did little to change the findings. 174 REFERENCES Part One Finlayson, L., Ford, R., Marsh, A., Smith, A. and White, M. (2000) The First Effects of Earnings Top-up, DSS Research Report No. 112, Leeds: CDS. Marsh, A., Finlayson, L., Ford, R., Green, A., White, M. and Callender, C. (1999) Low paid work in Britain, DSS Research Report No. 95, Leeds: CDS. Smith, A., Youngs, R., Ashworth, K., McKay, S., Walker, R. with Elias, P. and McKnight, A. (2000) Understanding the Impact of Jobseeker’s Allowance, DSS Research Report No. 111, Leeds: CDS. Sweeney, K. and McMahon, D. (1998) The effect of Jobseeker’s Allowance on the claimant count, Labour Market Trends, April 1998, pp 195 – 203. Vincent, J., Abbott, D., Heaver, C., Maguire, S., Miles, A. and Stafford, B. (2000) Piloting Change, DSS Research Report No. 113, Leeds: CDS. Part Two Anderton, B., Riley, R. and Young, G. (1999). ‘The New Deal for Young People: Early Findings from the Pathfinder Areas.’ Employment Service Research and Development Report ESR34. Ashenfelter, O. (1978). ‘Estimating the effect of training programs on earnings.’ Review of Economics and Statistics, Vol. 60, pp 47-57. Blundell, R. (2000). ‘Work incentives and ‘in-work’ benefit reforms: a review.’ Oxford Review of Economic Policy, Vol. 16, No. 1, pp 27-44. Department of Social Security (1995). ‘Piloting change in Social Security: helping people into work’, London: Department of Social Security. Dolton, P. and O’Neill, D. (1996). ‘Unemployment duration and the Restart effect: some experimental evidence.’ Economic Journal, 106 (March), pp387400. Finlayson, L., Ford, R., Marsh, A., Smith, A. and White M. (2000). The first effects of Earnings Top-up, DSS Research Report No. 112, Leeds: CDS. Friedlander, D. and Robins, P.K. (1995). ‘Evaluating program evaluations: new evidence on commonly used non-experimental methods.’ American Economic Review, Vol. 85, 4, pp 923-937. Green, A. E. (2001). Earnings Top-up Evaluation: Labour Market Conditions. Department of Social Security In-house Research Report No. 75. Gregg, P. (2000). ‘The impact of youth unemployment on adult unemploymen in NCDS. Department of Economics, University of Bristol Working paper, No. 00/495. Harris, J. E. (1985). ‘Macroexperiments versus microexperiments for health policy.’ In Hausman, J.A. and Wise, D.A. (eds). Social Experimentation, Chicago: University of Chicago Press. 175 Hausman, J.A. and Wise, D.A. (eds). Social Experimentation, Chicago: University of Chicago Press. Heaver, C., Roberts, S. and Vincent J. (2001). Earnings Top-up Evaluation: Qualitative Evidence, DSS Research Report No. 133, Leeds: CDS. Heckman, J. J. and Smith, J.A. (1995). ‘Assessing the case for social experiments.’ Journal of Economic Perspectives, Vol. 9, 2, pp 85-110. Heckman, J. J. and Smith, J.A. (1999). ‘The pre-program earnings dip and the determinants of participation in a social program: implications for simple program evaluation strategies.’ Economic Journal, Vol. 109, 457, pp 313-348. Heckman, J. J. and Hotz, V. J. (1989). ‘Choosing among alternative nonexperimental methods for estimating the impact of social programs: the case of manpower training.’ Journal of the American Statistical Association, Vol. 84, 408, pp 862-874. HM Treasury (2000). Tackling Poverty and Making Work Pay – Tax Credits for the 21st Century. The Modernisation of Britain’s Tax and Benefit System. Number Six. London: HM Treasury. LaLonde, R. J. (1986). ‘Evaluating the econometric evaluations of training programs with experimental data.’ American Economic Review, Vol. 76, 4, pp 604-620. McKnight, A. (2000). ‘Transitions off Income Support: Estimating the Impact of the New Deal for Lone Parents’ in Hasluck, McKnight and Elias Evaluation of the New Deal for Lone Parents. Early Lessons from the Phase One Prototype – Cost-benefit and Econometric Analysis, DSS Research Report No. 110, Leeds: CDS. McKnight, A. (2000). ‘Assessing the wider labour market impacts of Earnings Top-up using administrative data: technical report.’ Institute for Employment Research, University of Warwick, mimeo. Marsh, A., Stephenson, G., Dorsett, R. and Elias, P. (2001). Earnings Topup Evaluation: the Effects on Low Paid Workers, DSS Research Report No. 134, Leeds: CDS. Nickell, S., Jones, T. and Quintini, G. (1999). ‘A Picture of Job Insecurity Facing British Men.’ Centre for Economic Performance, London School of Economics, mimeo. Sweeney, K. and McMahon, D. (1998). ‘The effect of Jobseeker’s Allowance on the Claimant Count.’ Labour Market Trends, April 1998, Office for National Statistics. 176 OTHER RESEARCH REPORTS AVAILABLE: No. Title ISBN 1. Thirty Families: Their living standards in unemployment 0 11 761683 4 £6.65 2. Disability, Household Income & Expenditure 0 11 761755 5 £5.65 3. Housing Benefit Reviews 0 11 761821 7 £16.50 4. Social Security & Community Care: 0 11 761820 9 The case of the Invalid Care Allowance £9.70 5. The Attendance Allowance Medical Examination: Monitoring consumer views 0 11 761819 5 £5.50 6. Lone Parent Families in the UK 0 11 761868 3 £15.00 7. Incomes In and Out of Work 0 11 761910 8 £17.20 8. Working the Social Fund 0 11 761952 3 £9.00 9. Evaluating the Social Fund 0 11 761953 1 £22.00 10. Benefits Agency National Customer Survey 1991 0 11 761956 6 £16.00 11. Customer Perceptions of Resettlement Units 0 11 761976 6 £13.75 12. Survey of Admissions to London Resettlement Units 0 11 761977 9 £8.00 13. Researching the Disability Working Allowance Self Assessment Form 0 11 761834 9 £7.25 14. Child Support Unit National Client Survey 1992 0 11 762060 2 £15.00 15. Preparing for Council Tax Benefit 0 11 762061 0 £5.65 16. Contributions Agency Customer Satisfaction Survey 1992 0 11 762064 5 £18.00 17. Employers’ Choice of Pension Schemes: Report of a qualitative study 0 11 762073 4 £5.00 18. GPs and IVB: A qualitative study of the 0 11 762077 7 role of GPs in the award of Invalidity Benefit £12.00 19. Invalidity Benefit: A survey of recipients £10.75 0 11 762087 4 Price 177 178 20. Invalidity Benefit: A longitudinal survey of new recipients 0 11 762088 2 £19.95 21. Support for Children: A comparison of arrangements in fifteen countries 0 11 762089 0 £22.95 22. Pension Choices: A survey on personal pensions in comparison with other pension options 0 11 762091 2 £18.95 23. Crossing National Frontiers 0 11 762131 5 £17.75 24. Statutory Sick Pay 0 11 762147 1 £23.75 25. Lone Parents and Work 0 11 762147 X £12.95 26. The Effects of Benefit on Housing Decisions 0 11 762157 9 £18.50 27. Making a Claim for Disability Benefits 0 11 762162 5 £12.95 28. Contributions Agency Customer Satisfaction Survey 1993 0 11 762220 6 £20.00 29. Child Support Agency National Client Satisfaction Survey 1993 0 11 762224 9 £33.00 30. Lone Mothers 0 11 762228 1 £16.75 31. Educating Employers 0 11 762249 4 £8.50 32. Employers and Family Credit 0 11 762272 9 £13.50 33. Direct Payments from Income Support 0 11 762290 7 £16.50 34. Incomes and Living Standards of Older People 0 11 762299 0 £24.95 35. Choosing Advice on Benefits 0 11 762316 4 £13.95 36. First-time Customers 0 11 762317 2 £25.00 37. Contributions Agency National Client Satisfaction Survey 1994 0 11 762339 3 £21.00 38. Managing Money in Later Life 0 11 762340 7 £22.00 39. Child Support Agency National Client Satisfaction Survey 1994 0 11 762341 5 £35.00 40. Changes in Lone Parenthood 0 11 7632349 0 £20.00 41. Evaluation of Disability Living Allowance and Attendance Allowance 0 11 762351 2 £40.00 42. War Pensions Agency Customer Satisfaction Survey 1994 0 11 762358 X £18.00 43. Paying for Rented Housing 0 11 762370 9 £19.00 44. Resettlement Agency Customer Satisfaction Survey 1994 0 11 762371 7 £16.00 45. Changing Lives and the Role of Income Support 0 11 762405 5 £20.00 46. Social Assistance in OECD Countries: Synthesis Report 0 11 762407 1 £22.00 47. Social Assistance in OECD Countries: Country Report 0 11 762408 X £47.00 48. Leaving Family Credit 0 11 762411 X £18.00 49. Women and Pensions 0 11 762422 5 £35.00 50. Pensions and Divorce 0 11 762423 5 £25.00 51. Child Support Agency Client Satisfaction Survey 1995 0 11 762424 1 £22.00 52. Take Up of Second Adult Rebate 0 11 762390 3 £17.00 53. Moving off Income Support 0 11 762394 6 £26.00 54. Disability, Benefits and Employment 0 11 762398 9 £30.00 55. Housing Benefit and Service Charges 0 11 762399 7 £25.00 56. Confidentiality: The public view 0 11 762434 9 £25.00 57. Helping Disabled Workers 0 11 762440 3 £25.00 58. Employers’ Pension Provision 1994 0 11 762443 8 £30.00 59. Delivering Social Security: A cross– national study 0 11 762447 0 £35.00 60. A Comparative Study of Housing Allowances 0 11 762448 9 £26.00 61. Lone Parents, Work and Benefits 0 11 762450 0 £25.00 62. Unemployment and Jobseeking 0 11 762452 7 £30.00 63. Exploring Customer Satisfaction 0 11 762468 3 £20.00 64. Social Security Fraud: The role of penalties 0 11 762471 3 £30.00 65. Customer Contact with the Benefits Agency 0 11 762533 7 £30.00 66. Pension Scheme Inquiries and Disputes 0 11 762534 5 £30.00 67. Maternity Rights and Benefits in Britain 0 11 762536 1 £35.00 68. Claimants’ Perceptions of the Claim Process 0 11 762541 8 £23.00 69. Delivering Benefits to Unemployed People 0 11 762553 1 £27.00 179 180 70. Delivering Benefits to Unemployed 16–17 year olds 0 11 762557 4 £20.00 71. Stepping–Stones to Employment 0 11 762568 X £27.00 72. Dynamics of Retirement 0 11 762571 X £36.00 73. Unemployment and Jobseeking before Jobseeker’s Allowance 0 11 762576 0 £34.00 74. Customer views on Service Delivery in the Child Support Agency 0 11 762583 3 £27.00 75. Experiences of Occupational Pension Scheme Wind–Up 0 11 762584 1 £27.00 76. Recruiting Long–Term Unemployed People 0 11 762585 X £27.00 77. What Happens to Lone Parents 0 11 762598 3 £31.00 78. Lone Parents Lives 0 11 762598 1 £34.00 79. Moving into Work: Bridging Housing Costs 0 11 762599 X £33.00 80. Lone Parents on the Margins of Work 1 84123 000 6 £26.00 81. The Role of Pension Scheme Trustees 1 84123 001 4 £28.00 82. Pension Scheme Investment Policies 1 84123 002 2 £28.00 83. Pensions and Retirement Planning 1 84123 003 0 £28.00 84. Self–Employed People and National Insurance Contributions 1 84123 004 9 £28.00 85. Getting the Message Across 1 84123 052 9 £26.00 86. Leaving Incapacity Benefit 1 84123 087 1 £34.00 87. Unemployment and Jobseeking: Two Years On 1 84123 088 X £38.00 88. Attitudes to the Welfare State and the Response to Reform 1 84123 098 7 £36.00 89. New Deal for Lone Parents: Evaluation of Innovative Schemes 1 84123 101 0 £26.00 90. Modernising service delivery: The Lone Parent Prototype 1 84123 103 7 £26.00 91. Housing Benefit exceptional hardship payments 1 84123 104 5 £26.00 92. New Deal for Lone Parents: Learning from the Prototype Areas 1 84123 107 X £29.00 93. Housing Benefit and Supported Accommodation 1 84123 118 5 £31.50 94. Disability in Great Britain 1 84123 119 3 £35.00 95. Low paid work in Britain 1 84123 120 7 £37.00 96. Keeping in touch with the Labour Market 1 84123 126 6 £28.50 97. Housing Benefit and Council Tax Benefit delivery: Claimant experiences 1 84123 127 4 £24.00 98. Employers’ Pension Provision 1996 1 84123 138 X £31.50 99. Unemployment and jobseeking after the introduction of Jobseeker’s Allowance 1 84123 146 0 £33.00 100. Overcoming barriers: Older people and Income Support 1 84123 148 7 £29.00 101. Attitudes and aspirations of older people: A review of the literature 1 84123 144 4 £34.00 102. Attitudes and aspirations of older people: A qualitative study 1 84123 158 4 £29.00 103. Relying on the state, relying on each other 1 84123 163 0 £27.00 104. Modernising Service Delivery: The Integrated Services Prototype 1 84123 162 2 £27.00 105. Helping pensioners: Evaluation of the Income Support Pilots 1 84123 164 9 £30.00 106. New Deal for disabled people: Early implementation 1 84123 165 7 £39.50 107. Parents and employment: An analysis of low income families in the British Household Panel Survey 1 84123 167 3 £28.50 108. Evaluation of the New Deal for Lone Parents: Early lessons from the Phase One Prototype Synthesis Report 1 84123 187 8 £27.50 109. Evaluation of the New Deal for Lone Parents: Early lessons from the Phase One Prototype Findings of Surveys 1 84123 3190 8 £42.50 110. Evaluation of the New Deal for Lone Parents: Early lessons from the Phase One Prototype Cost-benefit and econometric analyses 1 84123 188 6 £29.50 111. Understanding the Impact of Jobseeker’s Allowance 1 84123 192 4 £37.50 112. The First Effects of Earnings Top-up 1 84123 193 2 £39.50 181 113. Piloting change: Interim Qualitative Findings from the Earnings Top-up Evaluation 1 84123 194 0 £28.50 114. Building Up Pension Rights 1 84123 195 9 £33.50 115. Prospects of part-time work: The impact of the Back to Work Bonus 1 84123 196 7 £29.00 116. Evaluating Jobseeker’s Allowance 1 84123 197 5 £16.00 117. Pensions and divorce: The 1998 Survey 1 84123 198 3 £36.00 118. Pensions and divorce: Exploring financial settlements 1 84123 199 1 £24.00 119. Local Authorities and Benefit Overpayments 1 84123 200 9 £26.50 120. Lifetime Experiences of Self-Employment 1 84123 218 1 £31.50 121. Evaluation of the Pension Power Power for you Helpline 1 84123 221 1 £28.50 122. Lone Parents and Personal Advisers: Roles and Relationships 1 84123 242 4 £29.00 123. Employers Pension Provision 1 84123 269 6 £35.00 124. The Changing Role of the Occupational Pension Scheme Trustee 1 84123 267 X £25.00 125. Saving and Borrowing 1 84123 277 7 £28.50 126. First Effects of ONE 1 84123 281 5 £38.50 127. Why not ONE? 1 84123 282 3 £25.00 1 84123 283 1 £34.00 1 84123 294 7 £26.00 1 84123 295 5 £33.00 Social Security Research Yearbook 1990–91 0 11 761747 4 £8.00 Social Security Research Yearbook 1991–92 0 11 761833 0 £12.00 Social Security Research Yearbook 1992–93 0 11 762150 1 £13.75 Social Security Research Yearbook 1993–94 0 11 762302 4 £16.50 Social Security Research Yearbook 1994–95 0 11 762362 8 £20.00 128. The British Lone Parent Cohort 1991 to 1998 129. Housing Benefits and the Appeals Service 130. Pensions 2000 (Attitudes to retirement planning) 182 Social Security Research Yearbook 1995–96 0 11 761446 2 £20.00 Social Security Research Yearbook 1996–97 0 11 762570 1 £27.00 Social Security Research Yearbook 1997–98 1 84123 086 3 £34.00 Social Security Research Yearbook 1998–99 1 84123 161 4 £30.00 Social Security Research Yearbook 1999–2000 1 84123 286 6 £27.50 Further information regarding the content of the above may be obtained from: Department of Social Security Attn. Keith Watson Social Research Branch Analytical Services Division 5 4-26 Adelphi 1–11 John Adam Street London WC2N 6HT 183