Monday, November 05, 2012

When is an incentive not an incentive?

A post from the local Labour Party about ATOS led me to have a look around. In an attempt to avoid legal matters I'm going to present a hypothetical situation that in theory could be applied to multiple situations.

Consider a government department that pays out money to claimants who present specific conditions. To determine if their claims are true the department outsources the checking to a private firm who employ experts to assess the claimants.

Both the department and private company state that there are no quotas involved regarding payment or bonuses. In theory therefore the company is free to assess fairly and honestly and not expect to be penalised if the results do not match the department's expectations.

That's the theory. Now I'll throw an extra into the mix - the very reason that the department outsourced these checks is because they believed that the previous system allowed too many "scroungers" in that is claimants falsely presenting the conditions that would allow them to gain monetary assistance from that department.

Given that extra consider a hypothetical situation in which the department's assumption was wrong.

The private company performs their task honestly and reports back that although there were a few scroungers (and there always will be) by far the previous assessments were accurate. The department now has two options

1. Accept the results.
2. Deny the results.

In the first instance the question is likely to be raised about how much money was spent on a private company to confirm the in-house results. The checks are likely to be returned in-house and the company not retained after the expiry of their contract.

In the second instance the government is assuming the company has failed in their checks and a new company will be brought in to do a 'proper' job'

Therefore if the company wishes to retain its contract there is an implied incentive for the results they present to adhere to the departments stated assumptions.

But wait it's not the company who do the reports they simply employ experts and compile their results. However as above so below. If the company, whom I've already ascertained wishes to retain its contract, hires two experts one who passes most claimants and one who fails most claimants which of the two is most likely to remain employed?

There is therefore an implied incentive on the experts to keep both themselves employed with the company and with the company being contracted by the department and therefore for them to adhere to the government's assumptions.

Considering it in reverse there is an incentive for the company not to check the experts' work provided it matches expectations and there is an incentive for the department not to check the company's work provided it matches expectations.

Worse yet even if independent experts were brought in to check they'd still be employed by the department and therefore they'd be exactly the same incentives on them to sign off on the work of the company.

So how can this be avoided? The flaw lies in the department's announcement of why it's outsourcing this work and what it expects the results to be. This is why I've tagged this with "Science" because that's one of the fundamental points of work carried out - you don't start it by stating what you expect the results to be because it's far too easy to nudge or cherry-picker the data to match your expectations.