11 May 2014

Analysis of a Million Sanction Decisions - Part 2

Those of you who read my previous post where I tried to link the propensity of JCP staff to refer claimants for a sanction may recall that I didn't have access to the individual Jobcentre case-load figures. I've now located Statistics on Jobseeker's Allowance by Jobcentre district and by month from March 2008 to October 2012 and used the October 2012 figure for the claimant count at each Jobcentre and hence calculated the rate of sanction referrals for most Jobcentres. (Remember, a "referral" is JCP staff "reporting" the claimant for sanction - an "adverse decision" from the Decision Maker" is by no means certain) I say "most" Jobcentres because some have been renamed, closed or opened between the two sets of data. If I couldn't match them up, I took them out. That's why your favourite Jobcentre might be missing. There are other limitations

Here's some analysis - feel free to  download the data and do your own too.

1. To get an idea of the variation between Jobcentres, I did a scatter plot

This shows that highest sanction referral rates are found in some smaller Jobcentres. A few of the lowest rates are in this group too. Large Jobcentres are very middle of the road.

2. Suspecting something "unnatural" about this chart, I sorted by sanction referral rates and plotted a histogram

  • Slopes are pronounced at around 4% and 8%. This suggests that JCP is monitoring the statistics and paying closer attention to offices outside the 4-8% range. There might even be an internal understanding that 4-8% is OK
  • The lack of any offices with less than 2.5% might be an indication of a mandatory target to exceed this figure
  • Again, there are some high rate outliers.
See below for some high/low rate offices

Finally, for a different take on sanction geography, see Geographical Variations in JSA Sanctions and Disallowances by Dr David Webster. This has some interesting work on the correlation between sanctions and local unemployment rate. It's also where I found the link to claimant count data :-)

No comments: