Prisoners’ dilemma

Some August political froth going on about benefits being paid to prisoners who, on the whole, aren’t entitled to get them.  The combination of incompetent government and the least worthy to be supported makes good tabloid fare.  But it’s worth digging below the surface (though no need to dig very deep) to draw out two thoughts:

  • There is a short-term political cost to government in attempting to understand problems.  The story takes flight because there is now some quantification of the problem (albeit that something in the order of 0.0001% of benefit expenditure is at stake) .
  • The underlying problem is that this is a special case of the much more general cause of fraud and error:  a benefit claim may be entirely legitimate at the time it is made and assessed but become illegitimate because of changes to the customer’s circumstances.  Some of those changes could in principle be picked up by closer integration of existing data sets, some of them could only be discernible automatically by much more substantial changes. But there is a cost in both financial terms and in data sharing (and therefore potentially privacy) which needs to be taken into account.

A few examples:

  • Death:  Nobody makes a huge song and dance about it, but DWP tracks the probate of estates.  If there were income-related benefits in payment before death and evidence of undisclosed capital at probate, the benefits can be recovered from the estate.
  • Employment:  An obvious one, falling into two clear sub-categories: (i) legitimate, where until there is close to real time capturing and recording of employment status government data can never quite catch up; (ii) illegitimate where there isn’t any data to start with
  • Leaving the country:  Nobody currently keeps track of who is and isn’t in the country – but those records are to be created for border control purposes.  Is it sensible to match those records against benefit records?  Is there any sense of the cost/benefit equation in doing so, either in financial terms or more broadly?
  • Health status:  A much bigger issue for benefit entitlement than imprisonment.  There are undoubted efficiency improvements  which could come from data integration – but we are all rather sensitive about our health records (to say nothing of the way health professional use that sensitivity ferociously to guard their turf)
  • And now prisoners.

So the bigger question is what level of life-state surveillance  is appropriate to minimise leakage from the benefits system.  The optimal social balance – and even the optimal financial – balance may well not come from attempting to force leakage to zero.

And of course we won’t even be able to have those debates if any attempt to bring data to bear on the problem is used as evidence of malevolent inefficiency.