In October we announced that we were going to postpone the 2017-based subnational population and household projections. Our previous blog described the reasons why we made that decision, and a short note describing the challenges that we had with the work was also published.
Since then we have been working to come up with a set of projections that deal with the issues that we had identified. This blog describes the changes we will be making to how the projections are produced, and our publication plans for 27 February.
Methodology improvement 1: migration
Previously we described one of the issues was the assumption of fixed migration in future years, which was giving highly unlikely projected population results given past trends. We have looked at a number of different ways to improve the migration component of the projections. With our technical advisory group we have changed the methodology as follows.
This will use migration rates. For internal inward migration this will be calculated as a rate relative to the rest of the UK population. For internal outward migration this will be calculated as a rate relative to the resident population of the local authority. Both of these mean that inward and outward internal migration is no longer constant and will depend on the size and the age structure of the population of the local authority and the rest of the UK.
This remains unchanged, that is, future years are held constant, on the basis of recent trends. This has not been changed because we found that changing the internal migration to a rate resolved the issues we had seen with the migration component. This is because most parts of Wales have far higher internal migration flows than international.
Methodology improvement 2: constraining the projections
Despite the changes to the migration component, we still found that the subnational population projections in total continued to show a different long-term trend to the national population projections (NPPs) published by the Office for National Statistics (ONS). As described previously, the national projections had, for the first time, shown a slight decline in Wales’ population over a 25 year period, yet our local authority projections, when combined, continued to show strong growth over the same period. Wales was the only nation that did not constrain its subnational projections to the NPPs. We considered with our technical advisory group a number of options:
- Constraining the total population only and not the components.
- Constraining the total population, the births and deaths components.
- Constraining the total population, births, deaths and internal migration.
It should be noted it is not possible using our software (POPGROUP) to constrain all the components (births, deaths, internal and international migration) as it requires one of those to be unconstrained to allow any adjustment to be made into that component. For example, should the total population be, say a 1,000 too high, an adjustment of 1,000 will be made to any component that is not constrained.
We assessed these options by looking at the components as well as the overall change by sex and age group and for the 22 local authorities. The second option provided the most plausible set of projections in our technical group’s view.
Due to the latest 2018-based NPPs showing a projected decrease in the population over the longer-term, it is likely that these next set of subnational population projections will be different to previous projections for most local authorities. These projections will reflect the latest demographic trends, which show decreasing fertility rates (resulting in fewer births) in Wales, and stalling life expectancy for example.
The methodological changes outlined above provide us with a plausible set of projections that, overall, give the same pattern as the NPPs. They also overcome the issues we had seen where implausible results were being produced for some age groups for some local authorities.
Changing the variants that are produced
As the new methodology constrains the projections to the NPPs, we can only produce variants that are also produced by the ONS, unlike before where we could produce our own (as the projections were not constrained). Therefore, following consultation with our technical advisory group, we are planning to initially publish high and low variants, and a zero migration variant. We will work with the ONS in future if we identify variants that we need that are currently not produced.
No changes to the household projections methodology
The household projections take the subnational population projections and apply household formation rates to them. Those rates are based on the 2001 and 2011 Census. We do not propose changing those rates at this point as results from the 2021 Census will be available in the next few years, which will allow us to review those formation rates.
Impact of the new methodology for the national parks
In order to best meet the needs of users, our plans were to publish the projections for the national parks alongside the local authority projections. However, due to the time it has taken to develop and ensure the local authority projections are robust, we are unable to publish national parks projections on 27 February. We want to have the same confidence in the constraining methodology for the national parks as for the local authorities – but there are multiple ways it could be undertaken, which we need to assess. We are therefore looking to publish these at the end of March.
What we plan to publish on 27 February
We will publish the local authority population and household projections. Alongside that, we will also be publishing the variants. Due to the amount of work it has taken, and the wide user demand for these projections, we will not be providing detailed commentary. Having discussed with our technical advisory group, it was felt that on balance having the data sooner would be of greater value to most users than having detailed commentary. We will, of course, still look to produce that detailed commentary, and we are aiming to do that at the end of March to coincide with the publication of the national parks projections.