Saturday 8 April 2017

Lack of knowledge or foresight can matter

Policy makers often don't know all the consequences of a policy they enact or don't think through all the implications of a policy change. And sometimes this really matters.

Megan McArdle at BloombergView discusses an example:
What happens when you suddenly offer parents generous family leave benefits, paid at the expense of the government? You can probably think of dozens of outcomes. But here’s one you might not have been expecting: people die.

That’s the finding of Benjamin Friedrich and Martin Hackmann, in a new working paper at the National Bureau of Economic Research. The culprit? Nurses, who skew female, provide a lot of vital health care, and made heavy use of Denmark’s new paid family leave benefit when it passed in 1994. Since the supply of nurses was limited, and their skills could not easily be replaced, hospital readmissions went up, and more troublingly, mortality spiked among elderly patients in nursing homes.

Advocates of paid parental leave are no doubt bristling at the implication that their favorite benefit might kill people. But that’s not quite the right implication to take away from this paper. What it really highlights is how difficult it is to know how a given policy will turn out. Had officials understood that in advance, they might have taken steps to mitigate the effects -- such as training extra nurses beforehand. The problem, in other words, wasn’t necessarily family leave policy, but the limited visibility policymakers have into the outcomes of their plans.

To see why, consider what the paper actually found. When parental leave came along, it reduced the supply of nurses. But that impact wasn’t felt evenly. In hospitals, where doctors make more of the medical decisions, it seems to have been costly to patient health. But in nursing homes, where nursing staff have more power over daily operations, it seems to have made a much bigger difference. Meanwhile, nursing assistants seem to have been little impacted by the change in leave policy; while they were also likely to make generous use of the leave, health-care facilities seem to have had little difficulty replacing them.
McArdle continues,
So too with something like parental leave. It isn’t enough to know how many workers you might have who might take advantage of the policy. You also need to know how vital their work is to the enterprise, and how readily the enterprise could adapt to their absence. What regulatory or structural barriers might exist to prevent you from rapidly training extra workers?

It may not be possible to know this in advance. Nursing administrators may find it easy to deal with temporary shortages by enticing retired nurses to temporarily return to the profession, or getting their existing staff to work overtime. It is only in the face of a sustained, large and widespread supply shock that those traditional resources will prove completely inadequate.

Moreover, even if we knew what the impact was likely to be, we probably don’t have exact estimates of the practical effects of suddenly slashing the supply of licensed professionals, like teachers and nurses, who will be hard to replace. I mean, we can probably hazard a guess that fewer teachers and nurses means sicker patients and kids who can’t read as well. But how big will the change actually be?
Thinking through the likely effects of a policy is important. Details matter. A given policy change can have very different consequences for different sectors of the economy. Just coming up with an idea that sounds nice or is politically convenient isn't enough. The law of unintended consequences will always get you if you don't think things through. And sometimes even if you do.

Thus when policy makers suffer from a lack of knowledge and/or a lack of incentives to investigate the effects of a policy change things can go very wrong. When uncertainty is high making evaluation of policy changes against pre-specified criteria for success required after a given time offers the opportunity to modify or rescind policies that are found not to be working.

No comments: