ONS progress update on Transforming Population and Migration

ONS progress update on Transforming Population and Migration Statistics System in England Wales Pete Jones Admin Data Census Project Office for National Statistics 1

Presentation Outline • Background to ONS research with admin data • Conceptual model for admin based statistical outputs • Estimating size of population • Wider population statistical system – including migration • Admin Based Income Statistics (ABIS) • Research plans 2018 and 2022 2

Background - ONS Admin Data Research for Census and Population Statistics • Beyond 2011 Programme (2011 -2014) • Census Transformation Programme • Introduce ONS Administrative Data Census • Replicate census outputs – “Research Outputs” • Aiming for a ‘parallel run’ in 2021 • Transforming the Population and Migration Statistics System • Admin data at the core of migration statistics: Autumn 2019 • Admin data at the core of population statistics: Spring 2020 • Need to ensure that the Statistical Design and Processing of 2021 Census is coherently aligned with the transformation of this system 3

Conceptual model Data sources Transformation process Statistical register Outputs Registers Census surveys Existing/ new surveys Linkage and transformation methods Other admin data Statistical Registers: • People • Addresses • Businesses • Activity Other data? * *eg. Big, commercial Quality measurement / assurance 4

Statistical Population Dataset (SPD) V 2. 0 – data source coverage English and Welsh School Census NHS Patient Register (PR) Included in SPD DWP and HMRC Customer Information System (CIS) Higher Education Statistics Agency (HESA) data Personal Demographic Service (PDS) ‘address moves’ – used to resolve address conflicts DWP benefits data – used to resolve address conflicts SPD population estimates 5

Comparison of Statistical Population Dataset V 2. 0 with the official midyear estimates by five-year age group and sex, England Wales, 2016 6

Coverage Surveys and Dual System Estimation 7

Integrating Admin Data Coverage Survey with Labour Market Survey Population Spine Address Register LMS Master Wave (200 k HH sample) Equal probability sampling for LFS waves and other ONS surveys Labour force survey wave 2 Opinions Coverage boost (150 k) ‘Boost’ optimises allocation of sample to areas with higher coverage error Expenditure Labour force survey wave 3 Annual Population Survey Labour force survey wave 4 Labour force survey wave 5 • • Under-coverage methodology established Over-coverage more complicated to design for - Dependent interviewing using admin data - Alternative sources of data to measure overcoverage - Model based approaches 8

The Population Statistics System Migration Stats Economic Stats system Census Pop Stats Social Stats 9

Wider Population Statistics System 10

Combining a stock based and flows based approach to produce population stocks and flows ONS Population Spine 2015 Apply rules Usually Resident Population 2015 Components of change derived from differencing 2015 and 2016 usually resident populations 1. Stocks Approach 2. Flows Approach 3. Evaluate differences between components of change and the usually resident population estimates they inform Hybrid Approach Model Components of change from admin data are added into the 2015 stock: • New to the population (births and immigrants) • Leaving the population (deaths and emigrants) • Special populations Triangulate the estimates Usually Resident Population 2016 derived from stocks and flows (hybrid) 11

Stocks: age-group source hierarchies (under development) First hierarchy Age Source Rationale Covers all E&W births 0 -4 year olds Birth registrations Third hierarchy Second hierarchy Some risk of outward migration since birth Source Rationale High chance of registering non-UK born children • Doing research to establish … for sources Personal Demographic hierarchy’ Service Will only qualify‘top with of evidence of active adult family eachmember age group at same address …. • 5 -15 year olds English and Welsh School Censuses State school pupils represent large proportion of population Child Benefit Extract Annual submissions reduce risk of over-coverage 16 -64 Will cover large proportion adult population not economically active with DWP/HMRC Our findings from PDSMWS work show evidence of lags in registration particularly for EU. Although do find presence Identify and remove short term of migrants on neither foreign students (using source (so PDS fills some longitudinal analysis / gaps) nationality) Periodic benefit payments on or near reference date Annual earnings exceeding income threshold HESA Evidence of low income longitudinally 65+ Likely to be pensions and health registrations … …. for children resident abroad Looking for evidence of economic activity CIS linked to: PAYE NBD SHBE Child Benefit Aim is to account for High take-up tomajority account for of population at top children not in state schools of hierarchy with minimal … over-coverage Child benefit unlikely to paid … …. 12 …

Sources for migration statistics Travel Events data – Exit Checks (Non EEA) Semaphore (All countries) Arr Dep VISA END DATE VISA START DATE VALID VISA PERIOD (visa length) First arrival Last departure (adapted from : Report on international migration data sources – July 2018) 13

Identifying Potential Migrants and Longitudinal Evidence 2013 LONGITUDINAL DATA - HESA LONGITUDINAL DATA - PAYE LONGITUDINAL DATA - BIDS LONGITUDINAL DATA - SC Potential Migrants 2015 MWS 2016 2017 HESA PR/PDS LONGITUDINAL DATA - PDS 2014 2018 14

Pre-publication slides ONS plans for income from admin data 2017 2018 Increased use of administrative data Direct estimates from administrative data to produce individual and household estimates Administrative outputs *Research Outputs Incorporate Child Benefit data Output quality Greater geographical detail Integrated sources LSOA Increased coverage of income components 2019 More detailed PAYE & Self Assessment data More developed income concept and definitions – net income and income deciles *Experimental Stats Impute Winter Fuel Payment and Christmas Bonus LSOA Incorporate Self Assessment and more detailed PAYE data OA? 15

Pre-publication slides Availability of administrative data on the components of income What’s missing from income: • Self-employment • Investments • Universal credit • Personal independence payments • Some other benefits e. g. Industrial injuries disablement benefit • Some non-UK pensions • Current transfers from nonprofit institutions and other households (e. g. child maintenance and parental contributions) • Anything which doesn’t go through the PAYE system (may be because it is below the tax threshold) 16

Pre-publication slides Admin Based Income Statistics • This year’s publication will be “Admin Based Income Statistics” (ABIS) • We are building on the Income Research Outputs previously published • This year the ABIS will be experimental statistics • Still based on the 2015/16 tax year • We have responded to user feedback to include: o Both individual and household income o Gross and net income estimates 17

Percentage of the population with PAYE and benefits income of £ 20, 000 or below by local authority Tax year 2013/14 Males and females, aged 16 and over 18 Data Source: PAYE employment and pension data and Tax Credits data from HM Revenue & Customs and benefits data from the Department for Work and Pensions

PAYE and benefits income distribution for males by age, England Wales Tax year 2013/14 Males, aged 16 and over £ 0. 00 £ 0. 01 to £ 5000. 00 £ 5000. 01 to £ 10000. 00 £ 10000. 01 to £ 15000. 00 £ 15000. 01 to £ 20000. 00 £ 20000. 01 to £ 30000. 00 £ 30000. 01 to £ 40000. 00 £ 40000. 01 £ 60000. 00 £ 60000. 01 and above No income information 16 -19 20 -24 25 -29 30 -34 35 -39 40 -44 45 -49 50 -54 55 -59 60 -64 65 -69 70 -74 75 -79 80 -84 85 -89 90+ 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Percentage of the population Data Source: PAYE employment and pension data and Tax Credits data from HM Revenue & Customs and benefits data from the Department for Work and Pensions 100% 19

ONS Pop Stats Transformation: 2018 to 2022 Ref year Decision point 2018 2019 Stock + flow estimates B Births B Deaths D D Admin Survey *Integrated with LMS Flow estimate B Adminbased stock? D Internal Migration I I International Migration 2020 Stock Flows Estimate? estimates I Int National stock estimates (PCS + Admin) Int IPS + Admin data + V small test PCS* (c 2 k) Admin at core of migration statistics 2021 Stock Estimate? IPS + Admin data + Small PCS* (c 50 k) Admin at core of population statistics Flow estimate B D Stock estimate Census + CCS (+ admin) 2022 (onwards) Stock + flow estimates B D I I Int IPS + Admin data + Small PCS* (c 50 k) “No surprises” PCS + Admin (+ Census) IPS + Admin data + PCS (c 350 k) 20

Questions? peter. jones@ons. gov. uk