Right now, the most significant change to the benefits system in a generation is under-way. Many of the benefits currently provided are going to be subsumed under Universal Credit (Income Support, Housing Benefit, Income based Jobseekers Allowance/ Employment Support Allowance), some will continue to be provided as separate benefits (Contribution based Jobseekers Allowance, Employment Support Allowance) and some will be removed altogether (Council Tax Benefit). We have put together an accompanying summary of the statistical output changes [PDF].
There’s lots of analysis out there about the likely impacts of the changes on particular groups – see Inside Government, BBC News- Business, CLES, Guardian Data Blog, the National Housing Federation, Guardian Society, Child Poverty Action Group and Joseph Rowntree Foundation for interesting briefings .
However, the benefit changes will also result in changes to availability of open data. In this blog, I look at the impact on data available to target resources and evaluate how local areas are changing over time.
What’ so important about the DWP benefit datasets?
The benefits datasets published by DWP are some of the best available measures of neighbourhood level deprivation. The published data gives an indication of the number of people who are unemployed (Jobseekers Allowance), out of work due to poor health (Employment Support Allowance and the older Incapacity Benefit), have a low income (Income Support, Working Tax Credit, Housing Benefit, and Pension Credit), have a disability or acute social care needs (Disability Living Allowance and Attendance Allowance), as well as children in workless and low income households.
These datasets make up a significant component of the gold standard deprivation measure – the Index of Multiple Deprivation (IMD), but unlike the IMD they have the additional benefit of being regularly updated (quarterly, and in the case of Jobseekers Allowance – monthly) enabling analysis of trends – how communities are changing over time, and the impact of events such as the recession on different types of area. The benefits datasets are also more straightforward to interpret then the complex scores in the IMD – they show the number and proportion of people in an area in receipt of a particular benefit – with changes over time being similarly straightforward to interpret.
The move to Universal Credit will affect the available data in a number of ways, which I’ve explored below.
1. Breaking the time series makes it harder to see how areas are changing over time
Perhaps the biggest issue in terms of data changes, is that it will become much harder to reliably identify how areas are changing over time. Where previously we were able to use the benefit data to track changes in unemployment, low income, poor health and other issues back to the beginning of the century, the change in benefits will effectively break the time series. So we will find it harder to use the benefit data in future to identify, for example, whether local areas have regained their ground since the start of the downturn in 2008.
An additional issue in the short-term is that changes are going to be phased-in at different speeds in different areas – with earlier pilots in some areas, with some groups and some benefits. So benefit changes may well reflect differences in Universal Credit implementation rather than meaningful differences between areas.
Also, new claimants will be moving onto Universal Credit before existing claimants. So, areas with high benefit turnover – such as areas with more seasonal or temporary employment – will see faster shifts of people onto Universal Credit than in areas with, say, more long-term unemployment or high levels of work limiting illness. Again, this will affect our ability to use the data to compare between areas and over time.
It is worth noting that the DWP are attempting to mitigate some of the impact here. The monthly claimant count figures released by Nomis will continue to be published, incorporating claimants receiving unemployment payments both via Universal Credit and Jobseekers Allowance. Also, DWP are publishing a “unified customer view” backdated to 2010 which provides a consistent measure of all statistical groups eligible for Universal Credit. Therefore we will continue to be able to compare unemployment levels as before and other benefit claimant rates back to 2010 (but not earlier).
2. The missing denominator
At present, the majority of DWP benefits are “people level” benefits i.e. they count the number of individuals receiving benefits. Universal Credit will be administered at household level, which means that the headline benefit measure will be the proportion of households receiving benefits rather than proportion of people (though DWP also plan to publish some individual measures as part of the publication).
In order to measure the proportion of households receiving benefits we need a reliable measure of the total number of households in a local area. Unfortunately, unlike estimates of the total population (published annually down to neighbourhood level as part of the Mid-Year Estimate series) there is no annual estimate at neighbourhood level of the total number of households.
So unless ONS start publishing local area household estimates, we will be reliant on Census 2011 data as the key denominator, which will become increasingly out of date. As a result, it will become harder to reliably compare between areas or over time.
3. More detailed breakdowns
Not all changes to the data are bad news – some will improve on the information we currently have. Importantly, DWP are intending to publish data for more breakdowns than currently available.
As well as the age and gender breakdowns currently available, DWP are proposing to make it possible to analyse the data by other factors such as ethnicity, asylum status, health, disability, education levels, relationships, tenure, and labour market status (employment, unemployed, inactive, self employed in part time employment etc). This will give us a richer picture of how different groups are faring on key deprivation, employment and health indicators.
4. Analysing flows on and off benefit
Under the current benefit system, people tend to move on and off benefits when their labour market circumstances change. However under the Universal Credit system, people are likely to remain within the system when circumstances change. For example, a person cycling frequently between work and unemployment will keep one Universal Credit claim throughout, merely changing the amounts paid to match the changes in status.
As a result the DWP will hold administrative records of the multiple statuses for a single claimant/ assessment unit, allowing a better analysis of changes in circumstances. This will give researchers a better source of benefit flow data, allowing a more detailed analysis of labour market transitions than is currently possible.
5. More frequent, more up-to-date data
Most DWP data is currently updated quarterly, but in the future this data will be updated monthly giving a broader range of snapshot figures for comparing change over time. Moreover the data is planned to be rather more up-to-date than at present – with a time-lag of three and a half months rather than the six months at present. We will therefore have a more up-to-date picture of benefit claimant rates than is currently available.
The bottom line
The welfare reforms are going to significantly change the data landscape. The changes bring opportunities for strengthening the evidence base and increasing our understanding of how local areas and groups are faring over time. However, the breaks in time series, and changes to eligibility criteria, mean that it will become more difficult in the short term to evaluate the impact of local programmes, and how different communities are responding to changes in the economy.