Investigating people-place effects in the UK using linked longitudinal survey and administrative data

Background

There is a long history of research and increasing interest in the role of place in shaping people’s economic and social life chances, in particular the literature investigating the existence and importance of so-called ‘neighbourhood effects’. Neighbourhood effects are commonly defined as impacts on individual-level outcomes that can be attributed to differences in the neighbourhood context, and which cannot be explained by past and present personal and family characteristics.

Scholars have suggested more than a dozen mechanisms through which neighbourhood effects affect objective wellbeing outcomes (e.g. education, employment status, occupation, income, health, and crime), and, more recently, subjective wellbeing outcomes (such as life satisfaction) have come into focus. Albeit, there is considerable disagreement between disciplines on whether neighbourhood effects exist and how important they are, the definition and measurement of neighbourhoods has been fairly unsystematic, and different methodological approaches have yielded different results. The prevailing view seems to be, however, that in the absence of real-world experimental or quasi-experimental evidence (for ethical or practical cost reasons), large-scale longitudinal panel studies augmented with longitudinal geocoded microdata at very immediate scales afford the best opportunities to identify causal effects as they help overcome identification issues relating to self-selection bias, unobserved heterogeneity, and reverse causation, which prevent any conclusions about genuinely causal effects.

Project aims and methods

The project will contribute to the most recent knowledge on the importance of place effects in several ways. It will provide new evidence for the UK on the presence of place effects and the relative contribution of these effects to individual wellbeing, compared to individuals’ characteristics and their family background. The research will use longitudinal microdata from Understanding Society: the United Kingdom Household Longitudinal Study (UKHLS) and its predecessor the British Household Panel Survey (BHPS), which is now incorporated in the UKHLS. Microdata from the UKHLS will be linked to external administrative data for very small geographies and multiple points in time, and at various scales of the neighbourhood. The project will implement sophisticated panel data regression models and causal inference techniques to better address the key identification issues which have hindered previous studies.

This project is one of three funded by the Nuffield Foundation that will be working closely together to better understand the drivers of individuals’ wellbeing.

Findings

Existing research

The research started with the team undertaking a systematic review of the existing neighbourhood effects literature from onwards 2002. The review focussed on six wellbeing outcomes (i.e., life satisfaction, self-reported health, social mobility, earnings, income and (un)employment) and on how a number of empirical challenges facing the neighbourhood effects research had been addressed. They conducted a detailed review of over 80 studies, yielding over 200 summaries of key results about neighbourhood effects on individual subjective and objective wellbeing. Three key findings stand out from the systematic review of the interdisciplinary literature:

1. There has been some progress in addressing selection bias in neighbourhood effects studies but progress is concentrated among studies that focus on objective wellbeing outcomes;
2. There still is a distinct tendency for quantitative papers to describe statistically significant associations between neighbourhood contexts and individual outcomes without testing (or specifying) the causal mechanisms that may lie behind observed neighbourhood effects, and this is more prevalent among studies that focus on objective well-being outcomes;
3. Studies used a wide range of spatial scales, with those looking at subjective outcomes focussing, on average, on more immediate and sociologically more meaningful units, and being somewhat more specific about the causal mechanisms involved.

The results are currently being written up for a critical review for the social science community. Preliminary findings from the research are available here.

Getting ready

In the second part of the project, the research team created a longitudinally harmonised dataset of small scale neighbourhood characteristics for linkage to individual and household level data from the Understanding Society study, a prerequisite for the planned longitudinal analyses in the next phase.

In longitudinal analysis, individuals and neighbourhoods are observed at multiple points in time and there are many sources of change that need to be disentangled to make appropriate policy recommendations: People change, places change, people change place (i.e., they move homes), and – so the working assumption in the neighbourhood effects research – places change people. The UK offers quite a rich suite of geocoded information available for linkage to survey data to allow us to capture relevant aspects of change, and there are a number of steps that analysts can take to account in the empirical modelling for the characteristics of people and places that do not change over time but which have not been measured or were measured with error. In November 2018, the project team trained a group of 15 analysts from the academic and policy research communities in how (not) to analyse neighbourhood data in Understanding Society. The overview presentation from the hands-on training course is available here.

A major undertaking in the project has been to create a longitudinally harmonised data set of UK Census 2001 and 2011 characteristics at the smallest census reporting unit (output areas) and bespoke aggregations thereof. Currently, the harmonised bespoke neighbourhood scales range from capturing the nearest 500 to 10,000 people. The general approach is similar to that adopted by other studies to create bespoke neighbourhoods based on administrative data. Instead of picking the nearest output areas based on as-the-crow-flies distance, however, the aggregation takes into account that groups of output areas that are part of the same Lower Super Output Area are regarded to resemble places that local people think of as a ‘neighbourhood’. The nearest neighbour output areas within the same LSOA are aggregated first and only when there are no further output areas in the same LSOA, the nearest neighbour from a neighbouring LSOA is picked.

New Evidence on Neighbourhood Effects

The second half of the two-year project is concerned with empirical modelling to gauge how much neighbourhood characteristics matter for wellbeing and how our conclusions may change in light of addressing the various identification issues. A briefing note prepared in November 2018 outlines the roadmap for the research and the appeal for researchers and data custodians to get in touch if they have additional data or applications has resulted in a number of promising new contacts and initiatives.

In March 2019, the project hosted an interdisciplinary workshop that offered researchers a forum to share their knowledge on state-of-the-art methods to address identification challenges in neighbourhood effects research, and to discuss existing evidence of neighbourhood effects for different neighbourhood boundaries and spatial scales, and on specific causal mechanisms. There were four presentations:

• Social scientists Gundi Knies (ISER University of Essex) and Patricia Melo (ISEG University of Lisbon) presented preliminary results from their research that investigates whether there is an effect of neighbourhood deprivation on four wellbeing outcomes: life satisfaction, mental health and physical health, and hourly wage.
• Epidemiologist Emily Murray (University College of London) presented an analysis of how neighbourhood deprivation experienced at different points over the lifecourse affects people’s body mass index (BMI) and how the relationships vary by selective migration and for different cohorts of people over time.
• Sociologist Heleen Janssen (TU Delft) presented a multiscale analysis of how living in an income deprived neighbourhood at age 16 affects individual earnings in ones’ 30s.
• Economist Florence Gofette-Nagot (University of Lyon) presented an overview of research and empirical findings on neighbourhood effects on employment in France, focussing on what different methods to address various identification challenges can tell us about the importance of neighbourhood effects.

The workshop was attended by researchers from across the UK and from multiple disciplinary backgrounds. The policy community and NGOs were also well represented, demonstrating the appetite for reliable empirical research on the importance of neighbourhood effects for wellbeing and how it may be positively affected by policy and practice delivered in local communities.

Later in the year, the project team refined the statistical analysis presented in the March workshop, focussing on life satisfaction and earnings. The effect of neighbourhood socioeconomic disadvantage is expected to be different for these two prominent outcomes. The effects on life satisfaction are predicted to be positive at the smallest scales of neighbourhood, e.g., due to social comparisons: ‘Keeping up with the Joneses’ is easier when the Joneses don’t live a life of plenty, all else considered. By contrast, neighbourhood effects on wages are predicted to be negative: ‘Keeping down with the Joneses’ is the norm when there is wide-spread joblessness and poverty in the employment area, e.g., due to lower-quality job information networks. Contrary to expectations, we did not find any evidence for differences in neighbourhood effects across different scales of neighbourhood nor across outcomes. While there is a negative association between neighbourhood deprivation and life satisfaction and earnings in the cross-sectional models, these are not robust once we control for unobserved factors that, so we argue, capture individual non-random selection into neighbourhoods. The research is published in Urban Studies journal.

The related results on neighbourhood effects on health-related quality of life, operationalised using the SF-12 mental health summary score and the SF-12 physical health summary score are presented in our project report. The results are interesting because we find a handful of instances of neighbourhood effects that are robust to controlling for unobserved individual and neighbourhood factors at some scales. More research is needed to explore which wellbeing outcomes are affected by neighbourhood socioeconomic disadvantage and at which neighbourhood scale.

The project report also includes the results from our literature review and we have published the review database so others may benefit from the information and could add to it.

The project is now completed but we will continue to post any updates on our project website on Researchgate.

Advisory Group

The project is one of three funded by the Nuffield Foundation that will be working together closely to better understand the drivers of individuals’ wellbeing. The other projects are Immigration and wellbeing and The impact of ethnic diversity on wellbeing and health. The projects share an advisory group.

Members of the Advisory Group

• Michaela Benzeval (University of Essex)
• Richard Harris (University of Bristol)
• Jonathan Portes (King’s College London)
• Ted Cantle (Institute of Community Cohesion)
• James Nazroo (University of Manchester)
• Maarten Van Ham (Delft University of Technology)
• Maria O’Beirne and Shayan Moftizadeh (Ministry of Housing, Communities and Local Government)
• Marley Morris (Institute for Public Policy Research)

Lead project partners

• Gundi Knies (University of Essex)
• Patricia Melo (University of Lisbon)
• Miles Hewstone and Miguel Ramos (University of Oxford)
• Matthew Bennett (University of Birmingham)
• Peter Howley (University of Leeds)
• Mirko Moro (University of Stirling)

The Nuffield Foundation is an independent charitable trust with a mission to advance social well-being. It funds research that informs social policy, primarily in Education, Welfare, and Justice. It also funds student programmes that provide opportunities for young people to develop skills in quantitative and qualitative methods. The Nuffield Foundation is the founder and co-funder of the Nuffield Council on Bioethics and the Ada Lovelace Institute. The Foundation has funded this project, but the views expressed are those of the authors and not necessarily the Foundation. Visit www.nuffieldfoundation.org