Monday, 1 June 2015

Social Impact Bond predictions

Well that escalated quickly. Social Impact Bonds (SIBs) went from an interesting new idea less than 5 weeks ago to actual government policy today. This background paper (pdf) is worth a read. While obviously positive about the concept the backgrounder also points to some pretty serious contract design issues which make SIBs undesirable for R&D projects, for things that might be canned for political reasons, and also for the following two categories:
  • status quo programmes, where it is difficult to attribute outcomes to individual programmes; and
  • new and unproven services, where it may be difficult to obtain investor ‘buy in’.
The process starts with the government identifying a measurable social outcome it is willing to pay for. Then a complex deal is struck involving at least three parties: the government, investors, and service providers. The investors back the service providers financially and the government pays both of these groups if and when the measurable social outcome is achieved. This is said to be good because the investors take the financial risk.

Here are a few predictions.
  1. The investors will generally make profits. This will be partly because they'll do their due diligence and try hard to only back winners. But they'll be helped by political aversion to pushing investors under the bus when things turn bad, as is currently happening with the Charter Schools experiment.
  2. Cherry picking will be endemic. The easy way to make money in this business is to restrict the scope of your services to people that are more likely to succeed. The government officials running the bonds will need to be very vigilant.
  3. No-one will examine the counterfactual. The situations where SIBs might work are also exactly the situations where sharper contracts could be written with existing agencies, which would also have cost-effectiveness and risk reduction benefits.

3 comments:

  1. Not hard to set a control group where there wouldn't be resource to roll out to full population anyway. Just need make sure they set it properly: random allocation across groups.

    Note too that government's letting Relationships Aotearoa fall over suggests they'll not run patsy assessments.

    My main worry in these here is avoiding oversight by multiple agencies: bureaucratic bogging down helped wreck the UK prison experiment.

    ReplyDelete
  2. Not hard to set a control group where there wouldn't be resource to roll out to full population anyway. Just need make sure they set it properly: random allocation across groups.

    Note too that government's letting Relationships Aotearoa fall over suggests they'll not run patsy assessments.

    My main worry in these here is avoiding oversight by multiple agencies: bureaucratic bogging down helped wreck the UK prison experiment.

    ReplyDelete
  3. Get daily ideas and methods for earning $1,000s per day FROM HOME for FREE.
    GET FREE ACCESS INSTANLY

    ReplyDelete