Utilising Synthetic Microdata to assess the spatial distribution of the inclusive economy

Presented at the British Society for Population Studies annual conference, Winchester on the 5th September 2022.

Abstract
A barrier to undertaking detailed spatial analysis of policy is the lack of high-resolution data on the composition and distribution of the population across key variables of interest. In this paper we utilise a spatially representative individual level ‘digital twin’ dataset of the UK population that has been built from a range of datasets (census, survey and administrative) using spatial microsimulation methods. We are concerned with assessing the spatial distribution of variables which contribute to how inclusive the economy is at a fine spatial resolution. These indicators include access to services, jobs and affordable housing which all vary spatially and temporally. We create an area-based classification from these variables and compare area types to understand how (un)equal the distribution of inclusivity is within a number of UK cities. We briefly outline the data used, methods, strengths and limitations of both the digital twin population and the indicators set before presenting results from the classification exercise. SIPHER’s inclusive economy indicators are designed for inclusion in models of relationships between policies for economic inclusion and health outcomes. The results will help policy makers by demonstrating where there is specific need across the inclusive economy domains.

Transcript

1. Utilising Synthetic Microdata to assess the spatial distribution of the inclusive economy Nik Lomax1, Ceri Hughes2, Ruth Lupton2 1School of Geography | University of Leeds | Alan Turing Institute n.m.lomax@leeds.ac.uk |Twitter @NikLomax 2University of Manchester BSPS| Winchester| 5 September 2022
2. The Synthetic Microdata
3. Spatial Microsimulation Sample or survey data Target or constraining data
4. Spatial Microsimulation Target or constraining data Sample or survey data
5. Creating a baseline population (spine) 2011 Census Full enumeration Spatially detailed Understanding Society survey data Representative sample Detailed attributes No spatial detail Mid-Year estimate and other constraint data Keeps constraints current
6. Simulated Annealing algorithm
7. https://bit.ly/3pYmuUs https://www.nature.com/articles/s415 97-022-01124-9
8. Allows for more spatially detailed analysis e.g. Employment rate (working-age population) NOMIS (Annual Population Survey) Synthetic population data Bolton
9. The Inclusive Economy
10. https://bit.ly/3AxkMhT
11. The inclusive growth (IG)/inclusive economy (IE) concept itself is both ill-defined and contested.
12. SIPHER is… concerned with economic inclusion rather than inclusive growth. The relationship between the extent and nature of inclusion on the one hand, and health and wellbeing outcomes on the other.
13. The consortium has developed a set of inclusive economy indicators. Meaningful to decision makers (capturing a recognisable, relevant aspect of inclusive economies) Number needs to be relatively small, adequately capturing the inclusive economy concept without being overly abundant and complex
14. Indicators - Economic outcomes Sub-domain Indicator decision 1 Participation in paid employment Percentage of working-age people who are employed 2 Involuntary exclusion from the labour market Share of working-age people who are long-term unemployed OR inactive due to ill health or disability 3 Wealth inequality Ratio of house price in most expensive area to least expensive area 4 Earnings inequality Ratio of weekly earnings between 80th and 20th percentiles 5 Poverty Percentage of children living in low income households 6 Decent pay Proportion of employees paid at or above the Living Wage (as defined by the Living Wage Foundation) 7 Job security/precarity – an aspect of job quality Share of employees in permanent work
15. Indicators – Wider Outcomes/Enablers Sub-domain Indicator decision 1 Whether people are gaining the skills and qualifications to enable economic participation and success Percentage of adults aged 20-49 with a Level 2 or higher NVQ qualification 2 Digital connectivity/inclusion Engagement with digital at LSOA level based on Internet User Classification (IUC) 3 Physical connectivity Public transport accessibility measure 4 Housing affordability Ratio of median house prices to median (workplace) earnings 5 Costs of Living Fuel poor households 6 Inclusion in decision-making Voter turnout in local elections
16. Data • Understanding Society – Wave 11 (aggregated to LSOA level via synthetic population) • Physical connectivity – Accessibility score from DFT (1-6) for each LSOA • Housing affordability - median LSOA house price data from ONS • All continuous variables • Variables scaled (subtract mean, divide by standard deviation) • Just Greater Manchester in first iteration
17. The aim of the K-means algorithm is to divide M points in N dimensions into K clusters so that the within-cluster sum of squares is minimized. Hartigan and Wong (1979) The K-means algorithm identifies k number of centroids, and then allocates every data point to the nearest cluster, while keeping the centroids as small as possible. https://towardsdatascience.com/understanding-k-means- clustering-in-machine-learning-6a6e67336aa1
22. More exclusion from labour market Below average employment More unequal earnings Less secure employment House prices are lower relative to earnings Low engagement with decision making C5 Least inclusive?
23. Unequal earnings Lower proportion in secure employment High house prices relative to earnings Affluent but not inclusive?
24. Most inclusive? Average?
25. Greater Manchester Most Inclusive Average Affluent not inclusive Exclusion from Labour Market Unequal earnings and not secure employment
26. Greater Manchester Most Inclusive Average Affluent not inclusive Exclusion from Labour Market Unequal earnings and not secure employment Healthy life expectancy
27. Next steps •Validate! (consult and correlate) •Incorporate additional variables (those identified, plus others?) •Extend to whole of Great Britain •Examine relationship with health outcomes
28. SIPHER is funded by: This work was supported by the UK Prevention Research Partnership (MR/S037578/1), which is funded by the British Heart Foundation, Cancer Research UK, Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Health and Social Care Research and Development Division (Welsh Government), Medical Research Council, National Institute for Health Research, Natural Environment Research Council, Public Health Agency (Northern Ireland), The Health Foundation and Wellcome.
29. Utilising Synthetic Microdata to assess the spatial distribution of the inclusive economy Nik Lomax1, Ceri Hughes2, Ruth Lupton2 1School of Geography | University of Leeds | Alan Turing Institute n.m.lomax@leeds.ac.uk |Twitter @NikLomax 2University of Manchester BSPS| Winchester| 5 September 2022