Skip to content

How a jump in Universal Credit statistics was overstated by 25% …

October 13, 2014

Update on 15th October 2014: The stats for September were released today – I have included an updated graph at the end of this blog.

Original Post: As the Universal Credit Project prepares for a nationwide roll-out for a small number of vanilla claimants and some couples, I explain a change to the way that the Department for Work and Pensions (DWP) reports caseload statistics. Counts are now made on the 2nd Thursday each month, meaning that the latest month-on-month increase in caseload was 25% overstated. There also remains a ‘Factor x’ downward revision to previous months’ statistics that remains unexplained…

Shrinking men

A case of ‘incredible shrinking statistics’?

tweetbutton shareonfacebook1

Keen-eyed Universal Credit observers should have raised their eyebrows at the most recent Universal Credit caseload statistics. These seemed to show a 48% increase in the number of people claiming Universal Credit between August and July (up from 7,460 to 11,070 claimants).

Official caseload

 Wow! A 48% jump in official statistics from July to August!

A 48% month-on-month acceleration from the pitifully low numbers of cases being previously handled would be welcome news from the beleaguered Universal Credit project. To meet the self-imposed deadline of 2017 rollout DWP staff had been told a year ago to expect “212,000 average monthly migrations” (HT Brian Glick, ComputerWeekly).

A Closer look…

In a footnote to the report attached to the press release, DWP mentions that the statistics are no longer compiled at the end of each month, but now on the 2nd Thursday of each month, and invites “users to assess the impact of this change”.

So here goes:

Revisions to the Universal Credit May statistics

Graph

Quite simply, the numbers have been revised down. In some months by quite a lot.

Why?  Well, it is due to two major effects:

1. Some 2nd Thursdays fall much later in some months (e.g. Thursday 14th August 2014) than other months (e.g. Thursday 8th May 2014)

2. Factor “X” (which I will return to at the end)

Working days per month

Vast swings in numbers of working days
between 2nd Thursdays each month

Month on month change

Large swings in the month-on-month 
working days in the DWP UC statistics

For the first time: a smoothed graph

So here I present a graph of ‘before’ and ‘after’, smoothing the official statistics by dividing by the number of actual working days by the average, so as present a comparable month-on-month graph:

Smoothed UC statistics

 The ‘X Factor’

There is also another consistent factor in the presentation of the statistics: a consistent revision each month downwards of the previous month’s statistics. In other words, each month the DWP is overstating the size of the claimants caseload by between 5% and 10%.

The DWP does not explain why this should be, but it did note on the last set of statistics that the recently trumpeted rise in August UC stats was based on “provisional caseload figures”…

Downward revision

Update 15th October 2014: The stats for September were released today – here is an updated graph.

Notes:
– I have used a (geeky) log scale. The target is about 10,000,000 claimants (which is obviously way off the top of the scale…
– On one hand, if exponential growth is achieved, then the 2017 target could be in sight…
– …on the other hand, if this month’s (adjusted) growth of 6,362 continues linearly, then it will take 120 years to roll-out

Smoothed UC statistics - incl. Sept

© Brian Wernham 2014 CC BY-NC-ND

Advertisements

From → Agile Governance

Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: