Enhancing Demographic Accounts for Informed Decision-Making

The presentation discusses the prototype model for demographic accounts to inform the 2023 National Statistician's recommendation on the traditional census. It delves into the limitations of the current system, the structure of demographic accounts, advantages and disadvantages, exploratory analysis, capacity-building, hybrid statistics system, and the need for more frequent updates and granular output to adapt to emerging challenges.

• Demographic accounts
• National Statistician
• Census
• Prototype model
• Data analysis

braddox Follow

Uploaded on Nov 26, 2024 | 0 Views

Download Presentation

Please find below an Image/Link to download the presentation.

The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.

Presentation Transcript

1. Demographic Accounts Prototype model to inform the 2023 National Statistician s recommendation on the traditional census Presentation to MARP December 6th2021 Louisa Blackwell, John Bryant, Duncan Elliott.

2. Contents Contents Limitations of the current system & emerging challenges Demographic Accounts structure including provisional reporting Advantages & disadvantages Exploratory analysis: national annual accounts for E&W Speeding up computation Data analysis Experimental series Maturity plans; further exploration of models Capacity-building and sustainability Appropriate long-term objectives

3. Hybrid statistics system: Demographic accounts in context Survey data SATELLITE COHORTS SATELLITE ACCOUNTS UK CEs Life expectancy Migrants Refugees International migration Fertility Internal migration Mortality Projections Admin data Etc etc Ageing Ethnicity SAE Households & families Covid & health Education Commercial data Benchmarking- population totals Demographic accounts- Longitudinal cohorts Cross-sectional characteristics Our best, coherent estimate of population and change 1% Longitudinal Study Population stock Births Deaths International migration Internal migration 100% Longitudinal Study Data and system models (pre- post-COVID) Multivariate attribution Cohort maintenance Coverage LMS Internal migration, syoa, sex Census, ABPE, syoa, sex Deaths, syoa, sex ABME, syoa, sex Births, sex Survey data Admin data Commercial data

4. Limitations of the existing system Limitations of the existing system 1. Manual, with substantial expert judgment Slow Difficult to replicate 2. Inflexible Difficult to change data sources or outputs 3. Critical dependence on Census Growing uncertainty and decennial rebasing Emerging challenges Emerging challenges Demand for more frequent updates Demand for more granular output Possible loss of traditional census Maximise use of administrative data Changes to input data Appearance of new data sources, eg cell phones

5. Structure Demographic account Demographic equivalent of national account Standardised format Consistency between stocks and flows System models Statistical models of regularities in birth, death, migration rates Formal representation of what an experienced analyst knows about demographic trends Data models Statistical models measurement, coverage errors Formal representation of what an experienced analyst knows about data quality

6. Demographic accounts Demographic accounts- - England and Wales England and Wales System models Post-COVID migration Pre-COVID population Post-COVID deaths Post-COVID population Pre-COVID deaths Pre-COVID births Pre-COVID migration Post-COVID births Births International migration Deaths Population stock Census, ABPE, syoa, sex Deaths, syoa, sex E&W migration, syoa, sex Data Births, sex Pre-COVID uncertainty Post-COVID uncertainty Pre-COVID error Post-COVID error Pre-COVID error Post-COVID error IPS CIs, syoa,sex Model CIs, syoa, sex Data models

7. Estimation windows for Dynamic Population Model Estimation windows for Dynamic Population Model Assumptions: 1) 15 month lag in availability of components (maximum lag from international migration?) 2)Key predictors are available monthly 2019 Demographic accounts Estimates 2020 2021 Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecJan Feb MarApr MayJun Jul Aug Sep Oct NovDec Feb-21 Mar-21 Apr-21 May-21 Jun-21 Jul-21 Aug-21 Sep-21 Oct-21 Nov-21 Dec-21 Monthly Reporting Best monthly provisional estimates from DPM based on incomplete predictors Finalised estimates from DPM using finalised predictors Forecasts based on historic time series and expert opinion

8. Advantages Automation Permits frequent updating, detailed outputs Transparency Replicable Permits experimentation Flexibility Easy to change inputs, outputs Uncertainty Formal measures of uncertainty Timeliness Uses early aggregate-level data

9. Disadvantages Computation Like much Bayesian computation, current methods are slow Current software not scalable to 340 Local Authorities Highly technical Need to think about communication with users Requires specialist statistical, computing skills at ONS

10. Frequency Monthly Annual Backseries Forecast Backseries Forecast Geography E&W Provisional Final Provisional LA Final

11. Exploring demographic accounts Using existing methods, software Account for all of England and Wales by single year of age (0 to 105+) and sex First proof of concept data models are relatively simple Single data sources currently used for entries and exits Census 2011, Patient Register and Statistical Population Dataset (formerly ABPE) version 2, where admin data sources assumed to have bias

22. Speeding up computation Current bottleneck: harmonising births, deaths, migration, population counts New approach 1. Split into cohorts 2. Apply efficient state-space methods within each cohort 3. Recombine Pilot: Regions of Greenland [show illustrative results] Working on software, methods for full-scale computation

24. Data analysis Demographic rates Describing and modelling trends in birth, death, migration rates Age by sex by Local Authority Measurement, coverage errors Identify systematic biases, errors Use results to inform system models and data models Drawing on previous work within ONS

25. Further exploration of models Model checking Tests of model, eg held-back data Sensitivity analysis, eg alternative priors Experimentation Effects of adding, subtracting data sources Simulation of coverage surveys

26. Capacity Capacity- -building and sustainability building and sustainability Bayesian Expert Group ISD WG Covid-19 Internal Migration sub-groups Quality dashboard Visualisation and near real-time feeds Jan- show & tell National Satellite accounts Eg migration, ageing, households & families, SAE Longitudinal Cohort feeds Triangulation Longitudinal Scientific Advisory Panel

27. Appropriate long-term objectives Monthly estimates What can current data sources support? Future data sources? Estimation without a traditional census How much can we expect from modelling plus administrative data? Can coverage surveys fill the gaps? Flexibility to support NS recommendation Evaluation of 2021 modelled estimates against the Census Can absorb survey-based new feeds for coverage adjustment Extension of the model Estimates by ethnic group, birthplace, households

28. Model structure Priors on the rates/means System models Rates/means True counts (unobserved) Arrays Datasets (observed) Data models Parameters of the data generating models

29. Frequency Monthly Annual Backseries Forecast Backseries Forecast Geography E&W Provisional 1 2 6 5 Final 13 14 10 9 Provisional 4 3 7 8 LA Final 11 15 12 16