Living environment deprivation and crime odpm

New Deal for Communities Evaluation:

EEssttiim maattiinnggtthhee
IIm mppaaccttooffNNDDCCuussiinngg LLaabboouurrFFoorrcceeDDaattaa

Contents

CONTENTS ......................................................................................................................................................................3

INTRODUCTION ..........................................................................................................................................................10

PART 1 ............................................................................................................................................................................12

2.1THE EMPLOYMENT PENALTY OF LIVING IN A DEPRIVED AREA?.............................................................................39 2.1.1 Men.................................................................................................................................................................40 2.1.2 Women............................................................................................................................................................42 2.1.3 Low skilled Men .............................................................................................................................................44 2.1.4 Lone Parents (Mothers) .................................................................................................................................45 2.2POOLED CROSS-SECTIONAL ESTIMATES OF IMPACT OF LIVING IN AN NDCAREA .................................................46 2.3JOB ENTRY AND EXITS IN DEPRIVED AND NDCAREAS..........................................................................................50 2.3.1 Changes to Job Entry probabilities over time................................................................................................52 2.3.2 Changes to Job Exit probabilities over time ..................................................................................................54 2.3.4 Modelling Job Entry and Job Exits ................................................................................................................55 2.4IMPACTS OF NDC USING MATCHING ......................................................................................................................58

PART 3 ............................................................................................................................................................................61

APPENDIX 2...................................................................................................................................................................77

3

5

6

Over the decade of 1993 to 2003 there had been a rise in the proportion of people in the most deprived areas who also lived in areas classified as multi-cultural areas and as out of town housing. These were the most significant changes in geo-economic characteristics and ran counter the overall changes for all of England. Additionally, the population in the most deprived areas increasingly became comprised of lone parent and single person households and younger people, with growing populations aged 16-25, again counter to the national trend. The most deprived areas were also more likely to have higher proportions of recent movers. Ethnicity was also changing with large growth in the proportion of Asian and Black people living in the most deprived areas and NDC areas.

Trends in male employment showed that overall there had been an increase in employment rates, a decrease in unemployment and an increase in inactivity and that these trends were common across deprived and non-deprived areas but to different extents. The most deprived areas still had far above average unemployment and inactivity and below average employment rates, but this to some extent is tautologous because the definition of deprivation used includes direct and indirect measures of employment in the index. Trends in female employment showed increasing employment rates and decreasing unemployment across deprived and non-deprived areas. However, inactivity rates had fallen at the national level but had remained flat in the most deprived areas. Lone parents had greatly increased employment rates at national level but there was a less pronounced increase in the most deprived areas where the majority of lone parents reside.

The results of this analysis are provisional and, while our methods allow us to estimate outcomes and impacts of NDC, there are data and methodological constraints that mean that all such estimations of impact and outcomes are provisional. Overall, the evidence from these provisional findings does not show a strong relationship between NDC and reduced levels of worklessness but overall interpretation does not support a finding of a negative impact. In short, our findings on impact are provisional because it is both too soon to accurately measure outcomes and because of the difficulty in attributing potential effects that are in themselves likely to be smaller than those associated with direct DWP-type interventions operating in the same locations at the individual level.

In cross-sectional annual data models there was flowing found. Men were found to have a reduced employment penalty when living in areas where both Action Teams and NDC were operating together. But there was no discernable effect for men in any other NDC area. On the other hand, male employment penalties were seen to fall over the NDC period where the DWP Area Based Initiatives (ABIs) were in operation separately from NDC. This finding of significant difference for the DWP ABIs suggests that the underlying methodology can at least identify area-specific effects associated with area-specific programmes. Women were also found to have a reduced employment penalty when living in NDC areas that operated alongside Action Teams (ATs), but at the boundaries of being significantly different from the pre-programme period. Low-skilled men were also found to have a reduced employment penalty when living in NDC areas that operated alongside AT alone and alongside both AT and Employment Zones.

There are very good reason for our findings not matching those found by Noble et al in their parallel analysis of worklessness and NDC impacts (2005). The outcome measures are different – employment and job entry/exit compared to benefit exit. Nobel et al use a longer longitudinal sample as opposed to largely cross-sectional analysis and use administrative data not survey data.

8

New Deal for Communities is an Area Based Initiative operating in 39 locally defined deprived areas of England as part of the larger National Strategy for Neighbourhood Renewal overseen by the Office of the Deputy Prime Minister (ODPM). The aim of such interventions is to tackle multiple deprivation in the most deprived neighbourhoods in England.

This report presents the findings of additional analysis of NDC undertaken to complement the main evaluation of worklessness by the Social Disadvantage Research Centre (SDRC) for Phase I on the New Deal for Communities National Evaluation (Noble et al 2005). The analysis has been undertaken by the Centre for Analysis of Social Policy at the University of Bath.

10

“employment effect”, the difference in probability of being in paid employment that arises from NDC.

Our conclusions follow in Part 3 and it is important at this introductory stage to highlight one essential limitation in our analysis that will underpin such conclusions: we have no way of identifying those that have actually received a programme intervention (or “treatment” in evaluation language); we can only identify those who live in the areas where NDC operated and made such interventions available. This means we are potentially comparing a mixture of actual participants and their neighbours and peers who did not actually participate.

11

Our main source of data is the Labour Force Survey. In order to evaluate and isolate the impact of NDCs on employment we merged in other area based indicators of employment, deprivation and indicators of whether other area based initiatives were in operation. This Section of the report describes these data.

1.1.1 Survey Data

Postcode data are thus the crucial variables that underpin our ability to identify those living in NDC areas and to attach other geographic data that can assist in our analysis. At the heart of our analysis is a simple addition to LFS data that gives us a flag to show whether the respondent lived in the NDC area or not. There will be some measurement error over time using postcode rather than geographical boundaries, as postcodes are not be entirely consistent over time because of residential redevelopment. This means, in effect, that recently built houses with new postcodes will not exist in previous survey data and vice versa, old postcodes of property renovated or demolished will no longer appear in more recent surveys.

We use existing weights for all data. There is the potential for differential non-response in deprived areas and existing weights may not adequately account for this. There is also potential for such differential non-response to have grown over time, indeed the general picture of declining sample sizes in the NDC areas over the period 1993 to 2003 may reflect this. We have made no attempt to re-weight or to otherwise compensate for any such biases that result. As later discussion shows, we faced a number of theoretical and practical data measurement and estimation problems that took prominence in our methodology. Additionally, future extension of our approach is most likely to use the larger samples of the Local Labour Force Survey, which will address some of the specific problems of statistical reliability in smaller geographical areas.

13

Table 1 - Are Deprivation and New Deal for Communities Areas

14

Table 2 - Jobcentre Plus District Labour Market Clusters

The major drawback of these clusters is that they explain unemployment and job entry best and work less well for inactive groups such as lone parents and long-term sick and disabled people who make up the “workless” profiles of many deprived communities. The other drawback is that they operate at too large an area for many of those seeking work, especially in low-paid and part-time work. Job search in deprived communities is often limited to small geographical areas and constrained by public transport and other travel constraints4.

This second drawback of the Jobcentre Plus District Clusters is shared by our alternative geographical labour market indicator: characteristics at the Travel to Work Area (TTWA) level. TTWAs are derived from actual job travel data and thus the boundaries of TTWAs reflect long-distance travel to work by more affluent commuters. TTWAs can be huge, London for instance, or much smaller, in rural areas in particular. They are thus not consistently greater or smaller than the Jobcentre Plus District Clusters. We use TTWAs where appropriate and assess how far they can accurately capture and explain employment profiles as compared to the Jobcentre Plus District cluster variable. We do not use them together as they are highly collinear. The characteristics of TTWAs in 2001 that we use are job density and employment rates. In the majority of our models the TTWA level variables predicted better than the Jobcentre Plus variable and was preferred.

1.1.4 Definitions

The essential and common definitions to all our analysis are:

This section looks at overall trends in deprived and non-deprived areas of England between 1993 and 2003, and at NDC areas as a sub-set of deprived areas.

1.2.1 Area Change

_____________________________________________________________________

Text Box 1

Traditional Manufacturing (supergroup 2)

Built-up Manufacturing wardshad four per cent of the UK population in the 2001 Census, and 3.8 per cent in 1993 LFS in England, falling to 3.5 per cent in 2003. These wards are found in the north of England, and Bulwell in Nottingham, Halton Lea in Halton and South Bank in Redcar and Cleveland are typical wards in this group. The proportions of households with two or more cars, living in detached houses, in households with two adults and no children, and with a higher education qualification are lower than the national average. These wards have a greater proportion than the national average of people who are unemployed or long term unemployed, of lone parent households, of people who are separated or divorced, of household spaces which are terraced, of people who work in routine occupations and of households renting from the public sector.

households with one person, unemployment, public sector rental, people of a working age suffering from limiting long-term illness, people who are separated or divorced and flats.

Prospering Metropolitan (supergroup 4)

Multicultural Areas contained 3.1 per cent of the UK population in the 2001 Census and the LFS shows (Table 4) that 3.8 per cent of the English population lived in such areas in 1993, falling to 3.6 per cent in 2003. They are concentrated on the periphery of Greater London and Roxbourne in Harrow, St. Paul’s in Sandwell and Queen’s Park in Bedford are typical wards in this group. They have proportions far below the national average of detached homes, of households with two adults and no children, of women who are working part time, of households with two or more cars and of people who are aged 45 to 64. These wards have far higher than average population density and far higher than average number of rooms per household but also higher than average number of people per room. They have higher than average concentrations of people who are aged 0 to 4, women who look after the home, people who are unemployed, people who travel to work using public transport, terraced housing, people not born in the UK and people identifying as Indian, Pakistani or Bangladeshi.

Inner City Multicultural wards have 3.1 per cent of UK population living in them according to the 2001 Census, and Table 4 shows that 4.5 per cent of English LFS population lived in such wards in 1993 rising to 4.7 per cent in 2003. They are mainly concentrated in north east and south east London and have lower than average concentrations of households with two cars, detached houses, part-time women workers, people working in manufacturing, unpaid carers and people aged 45 to 64. These areas have higher than average population density, crowded housing, unemployment, separated and divorced people, single person households, flats, black people, people born outside the UK, people with higher education and public transport users. Typical wards are De Beauvoir and Victoria in Hackney and Gipsy Hill in Lambeth.

Commuter Suburbs contained 9.9 per cent of the UK population in the 2001 Census, and LFS data shows that 11.3 per cent of English population lived there in 1993 and that this had fallen to 10.5 per cent by 2003. They are concentrated in the South East of England. Silhill in Solihull, Mersey St.

Mary’s in Trafford and Uxbridge North in Hillingdon are the most typical wards in this group.

Accessible Countryside (supergroup 9)

Accessible Countryside wards contain 5.1 per cent of the UK population in the 2001 Census and have around five per cent of the LFS English population with a slight rise between 1993 and 2003. They are more scattered throughout England. They have lower than average percentages of people working in routine occupations and higher than average people aged 45 to 64, households with two adults and no children, households with two cars and higher than average proportions living in detached houses.

There has been remarkable change in the family composition of working age people in England between 1993 and 2003. Single childless people have risen from 23.5 per cent to 31.8 per cent, while couple families with no children have fallen from 33.1 per cent to 24.9 per cent. In families with children, lone parent families have risen from 7.9 per cent to 14.7 per cent while couples with children have fallen from 35.4 per cent to 28.4 per cent. These changes are more pronounced in the most deprived quintile of England. Most noticeably, lone parents, who were always over-represented in deprived areas, have become more so, rising from 13.1 per cent of families to 24.1 per cent, an eleven percentage point rise. The proportion of couple families with children has fallen faster than in England as a whole, from 32.3 per cent to 22.5 per cent. Changes in childless families are less pronounced. Childless couples were less likely to live in deprived areas in 1993, only 26 per cent as opposed to 33 per cent for all of England, and these have fallen by 8.8 percentage points, slightly widening the divergence from the national situation. On the other hand, single childless families, over-represented in 1993 compared to England as a whole, have risen from 28.5 per cent to 35.7 per cent in the most deprived quintile of areas.

How do NDC areas compare and how have they changed? Lone parents, originally a larger proportion of the population in NDC areas than in the wider quintile of deprived areas, have not increased proportionally faster than the bottom quintile. This means that, by 2003, 23.4 per cent of working age families living in NDC areas were headed by lone parents, much higher than the national average, of 14.7 per cent, but very similar to the most deprived quintile. The proportion of couples with children however have fallen faster in NDC areas than in the most deprived quintile and thus faster still when compared to England. In 1993 they were 35.1 per cent of all working age families and by 2003 they had almost proportionally halved to 18.3 per cent. NDC areas however have seen much greater growth in single person families; in 1993 they had proportions similar to the national average, at 31 per cent, but increased to 42.4 per cent by 2003. Childless couples have decreased proportionally at a faster rate than both the most deprived quintile and the English average falling from 19.4 per cent in 1993 to 15.6 per cent in 2003.