ons presentation at rss south wales poverty & inequality stats event
TRANSCRIPT
Update on ONS data for poverty statistics and researchPoverty & Inequality in Wales: Statistics for Action28th Sept 2016
Richard [email protected] @richt2
Aims
• To update on some of the latest ONS poverty-related data and analysis developments
• To provide information on the ONS Data Collection Transformation Programme
• To outline plans for ONS’s household finance surveys and statistics
Relative income poverty in Wales
94/95
-96/97
95/96
-97/98
96/97
-98/99
97/98
-99/00
98/99
-00/01
99/00
-01/02
00/01
-02/03
01/02
-03/04
02/03
-04/05
03/04
-05/06
04/05
-06/07
05/06
-07/08
06/07
-08/09
07/08
-09/10
08/09
-10/11
09/10
-11/12
10/11
-12/13
11/12
-13/14
12/13
-14/15
0
5
10
15
20
25
% of individuals in households with equivalised income (BHC) less than 60% of median
Wales GB/UK
Source: DWP Households Below Average Income, 1994/95 - 2014/15
Small area estimation
• Model based estimates produced using a combination of:• survey data for indicators of interest• Area-specific auxiliary data (admin data and/or Census)
• At risk of poverty or social exclusion (AROPE) • Europe 2020 headline target• 3 components:
i. relative low income (equivalised disposable income below 60% of median) ii. severe material deprivation (enforced lack of 4+ out of 9 items)iii. low work intensity (adults in household worked less than 20% of potential in
previous year)
• Research to produce estimates at NUTS 2 level using SAE techniques
• Survey source: EU Statistics on Income & Living Conditions (EU-SILC)• Auxiliary data: 2011 Census, data on receipt of benefits (DWP) and energy
consumption (DECC/BEIS)
At risk of poverty or social exclusion
Surre
y, Eas
t and
Wes
t Suss
ex
Cumbria
North Y
orkshir
e
Cheshire
Easter
n Sco
tland
Devon
Herefor
dshir
e, W
orceste
rshire
& W
arwick
shire
East A
nglia
Kent
Linco
lnshir
e
East W
ales
Lanc
ashir
e
Outer Lon
don
South W
ester
n Sco
tland
East Y
orkshir
e and N
orthern
Lincoln
shire
Northum
berla
nd an
d Tyn
e and W
ear
South Y
orks
hire
West M
idlan
ds0%
5%
10%
15%
20%
25%
30%
35%
40%
At Risk of Poverty or Social Exclusion rate, UK NUTS 2 regions, 2013
• Development ongoing• Intention to make NUTS 2 level estimates of AROPE and 3
component indicators available annually from 2017 onwards
Source: ONS
MSOA level small area poverty estimates
• % of households with equivalised income below 60% of median (AHC)• Estimates at MSOA level• Experimental Statistics• Most recent data currently available 2007/08 (England & Wales)
• However… • 2011/12 estimates to be published 16 December 2016• 2013/14 estimates planned for Spring 2017• 2015/16 estimates planned for Autumn 2017
• Mailing list & further info: [email protected]
Small area poverty estimates
Source: ONS Small Area Poverty Estimates, 2007/08
Census Transformation Programme: Admin data research outputs on income
• Estimates of income distribution at local authority level for England & Wales
• Published 16 December 2016• Produced solely from administrative data sources
• Using PAYE data from HMRC and benefits data from DWP• Outputs of ongoing research – not Official Statistics
• Incomplete measure of gross annual income(i.e. before direct taxes) for individuals
• Limited or no data on certain income components e.g. self-employment, property/investment income
• Looking for feedback from users to inform development
ONS Data Collection Transformation Programme
Context – Environment for Transformation
• Wealth of information held by govt departments and other public bodies
• Legislation in place to secure access to (some) data (although limitations)
• Commercial and Big Data Sources – new opportunities?
Data Sources
• Digital Age – ‘digital by default’• New technology methods, systems, capabilities and
continuous developments• Risks with existing systems and methods
Technology
• more mobile, diverse population• Lifestyle changes• Less willing to engage with government ? Less willing
to participate in surveys• Expectation of digital methods but security, privacy,
concerns
Societal Changes
Goals - a future social statistics system will……
• Exploit the potential of non-survey data sources Wherever possible, replace survey collection with non-survey sources Use data from non-survey sources to improve survey design (e.g. precision, covariates) Use non-survey data to enhance and extend outputs (e.g. data on supplementary topics
or in development of model-based estimates)
• Maximise the take-up of online collection A redesigned survey portfolio taking into account availability of non-survey data and
online capability Implement online self-completion as the default mode of collection, where appropriate,
within mixed mode operation Implement a ‘new’ organisational structure and field collection model to deliver value for
money in supporting the future approach
• Implement systems to support the future statistical system Redevelop IT systems under a service oriented architecture approach exploiting
opportunities for re-use of Census Transformation Systems wherever possible Ensure that the business has available the capability, skills and tools to implement the
future statistical system
Vision for Future social statistics systemA
dmin
istr
ativ
e so
urce
s
DATA COLLECTION
Commercial Sources / Big
Data
User / output needs
Surv
ey
Sour
ces
Integrated social data
sources
Statistical Methods
Social statistics + Future Census?
Social statistics outputs
Registers
Plans for ONS’s household finance surveys and statistics
Current uses and outputsSurveys Uses / Outputs
LCF
SLC
•Effect of Taxes and Benefits (ETB), HDII, Nowcasting•Input into IGOTM•Household consumption data for National Accounts•Informs “basket of good” and weights for inflation indices•Estimates of food consumption and nutrient intake•EU Household Budget Survey
•Longitudinal EU-SILC (FRS provides X-sectional data)•Estimates of persistent at risk of poverty•Analysis of transitions in and out of employment / poverty
•Estimates of wealth and wealth inequality•Monitoring pensions up-take•Exposure to debt
WAS
Current survey designsLCF SLC WAS
Computer Assisted Personal Interviewing Postcode sectors / address selected from PAF (private HHs)
Clustered by postcode sectorStratified by Region
Implicit stratification by Census indicators (which differ across surveys)
5K HHs achieved(annual UK)
7K HHs achieved(annual – all waves UK)
20K HHs achieved each wave (10K annual)High – GBOver samples wealthy
Cross-sectional Longitudinal (follow up to FRS subsample)4 yearly rotational design - Individuals followed
Longitudinal: panel survey with annual boosts - Individuals followed
2 week diary for expenditure CATI (telephone) Keep in Touch Exercise (KITE) between waves
CATI KITE between waves
Current survey content – topic level
SLC LCF WAS
• Basic demographics and education • Tenure and accommodation, Mortgages• Economic status, occupation, industry, hours worked• Employment income• Benefits and tax credits (receipt and amounts)• Pensions• Income from property rental / pensions / assets• Health
• Childcare• Material Deprivation• EUSILC secondary
modules
• Detailed expenditure data
• Wealth – financial and physical assets
• financial planning• Savings and debt • Value of pensions
The drivers for change – Household Financial Surveys
Coherence: Responding to UK Statistics Authority monitoring review
Informing policy: Meeting needs of UK policy makers & IESS regulation
Transformation: Delivering ONS Data Collection Transformation programme
Efficiency: Minimising cost & burden of statistical production
• Large number of sources & outputs – difficult for users to know where to look• Outputs largely based on sources rather than themes
• Range of surveys makes responding quickly to changing policy needs more difficult• Difficulties in meeting user requirements on timeliness and regional data• 4-year longitudinal dataset considered insufficient
• Data collection relies on expensive face-to-face surveys with diminishing response rates•Survey based estimates prevent effective examination of top/bottom of distribution •Duplication of effort in data processing due to multiple sources & systems
Where we want to be
Core
(including labour,income, housing, material
deprivation, work
intensity etc)
Expenditure
Adm
in d
ata
Wealth
Other user needs (e.g. EU-SILC modules)
Dat
a av
aila
ble
long
itudi
nally
• Greater coherence / thematic approach• Joint analysis of income, consumption
and wealth possible• Best use of administrative data• High quality data for analysis of income
distributions (including top and bottom)• Responsive to user requirements• Precise regional estimates • Timely estimates• Regulatory requirements met• Make use of new technology and mixed
mode data collection• Reduced costs and respondent burden
Developments in 2016 and 2017
• Integration of the SLC and LCF - harmonised methods for sampling, collection (income data) and processing- Potential to improve sampling designs and therefore precision of UK and
regional estimates- Supports a larger sample for key survey estimate- Common methods for collection and data processing, drawing on best
practice
• Expansion of the Survey on Living Conditions (SLC) to a 6 wave longitudinal design- Larger sample for regional and longitudinal analysis- SLC will meet the full EU-SILC requirement
Developments in 2016 and 2017 (cont)
• Assessment of how administrative and other non-survey data could improve the surveys (initial focus on DWP, HMRC, VOA data – including income, tax and benefits data)- Could replace survey data, thus reducing questionnaire length. This
provides greater opportunity for online collection- Potential for use in sampling designs, to improve coverage and precision- Possible use in the editing, imputation, estimation processes to improve
data quality
• Responding to the LCF National Statistics Quality Review- Improvements to income data- Greater use of other data sources- Electronic data collection (diary)
Longer term development
• Incorporating administrative data into the statistical system
• Integration of wealth data into the data collection model
• Mixed mode collection