From a recent podcast interview with Tyson Yunkaporta
This post began as a comment on David Walker’s post on David Card’s Nobel Prize for his study which showed that at least in the situation he investigated a smallish rise in the minimum wage didn’t reduce employment. This was regarded as a huge result — still is judging by the Nobel Prize he got for his trouble.
What this makes me think of is how impoverished economic discourse is for its preoccupation with being a ‘science’. Of course that doesn’t mean it should make stuff up. In my opinion, economics should be like medicine. It should understand itself as helping us navigate the world with a view to improving it. That makes it, in Herbert Simon’s lingo a ‘science of the artificial‘.
Simon divides the disciplines into ‘sciences’ and ‘design’. My way of summarising this is to say that the sciences are about the universe. They are about what is. And design is about the multiverse which is to say the infinitude of ways the universe might come to be configured, and most particularly the sub-set of those universes that we think we might be able to bring into existence. Design is then about seeking to build the best world we think is possible.
Medicine and engineering are in the same boat. None of this means that they have any ‘get out of gaol cards’ as far as obeying the laws of nature is concerned or any other laws for that matter. They’re reality based. But the laws of nature are just one input, with other inputs being such things as:
- the things we value most and least
- what we have control/mastery or some influence over and
- what we might gain control over
This then situates or frames what we are doing much better. You’ll note that neither medicine, nor engineering (nor pretty much any other discipline) do anything as weird as separate their discipline into ‘normative’ and ‘positive’. That’s because it’s understood that the ‘positive’ exists for the purpose of the normative.
More to the point, in such disciplines one doesn’t have the preoccupation with these set piece factoids that one does in economics such as ‘do higher minimum wages depress employment’. The equivalent question in engineering might be “are bridges built of concrete more or less stable than bridges built of steel?”. If one is trying to raise the incomes of those on low incomes, what we know of this set piece about minimum wages is of course of some significance. But it will appear alongside other issues that are relevant. Other things that might be taken into account include:
- If one is in that part of the curve where there’s a tradeoff between increasing pay for those on low pay and perhaps reducing demand, then one can be more confident that the former effect dominates for small rises than large rises (assuming some agreed way of trading them off).
- Other means of support might complement minimum wage rises. This could be done in two ways — static benefits like earned income tax credits increasing earned income and complementary strategies in which government funding is made available for upskilling workers with phase ins of higher minimum wages.
- This might complement more general productivity enhancing measures to improve the labour market such as “Windows on Workplaces“, the proposal I took to the 2020
- Industry specific programs to support the gradual increase of employee and management skills.
These are just a few possibilities, and not very imaginative ones at that. For instance one might allow certain firms to pay lower wages in specific circumstances (if it looks like the alternatives are worse) but rule it out elsewhere and that itself might be wrapped up in various considerations to which that gives rise.
Even these examples are not giving you quite the flavour I’m trying to give you because they’re all government policies. But it’s noticeable that the proposition we’re dealing with “do minimum wage hikes cut employment” is one of those questions that is a set piece in an economics textbook or an economic paper.
As I commented a few years ago:
There’s a deep academic literature on questions like “does performance pay for teachers or school vouchers, or charter schools improve student outcomes”. But the answer to these kinds of questions is usually that “it depends”. As Deborah Johnston puts it in discussing aid to Africa “It is an over-simplified and erroneous question to ask, do cash transfers work?”
One could add the question “do lower class sizes improve performance”. I prefer this question as illustrating what I’m on about to the question of minimum wages because that question sits in among numerous alternative ways of improving performance. And many of those ways are not particularly amenable to central government direction. To improve performance, most of the good work needs to happen within schools. And it’s not ‘scientific’ work in the sense that I’ve used the term. The literature on class sizes, with its econometric methods, is good to know as a set of averages detected in various studies. But as you go about seeking to improve the performance in a district, or a suburb, a school or a class, that will give you a much more fertile context within which to make progress including by varying performance, and seeing the results and varying them again and so on.
I’ll conclude with the observation that I wish social scientists would think of their job not by analogy with natural scientists, but much more by analogy with a football coach. Of course academics don’t think that way because football coaches aren’t intellectually glorious like scientists are.
However they are, quite unambiguously, practitioners of design.
That is, they have to be reality based if they’re to be any good. They’re interested in stats and whatever might be measured to help them think about things, to help them measure not only the way the world is, but the way others manage to do better and worse, and to measure changes they seek to bring about. But they won’t be fanged up about the stats. They’ll understand that they can provide a useful bit of what I’ve called — without much success around here “ground truthing“, but they won’t mean much without being plausibly embedded in wider stories. They could commission econometric studies looking for correlations between the length of passes and success, but they’ll understand that that will probably produce lots of inconclusive and silly answers. They’ll understand that their stats need to be embedded in and subordinate to a whole prior world of judgement.
More generally, they’ll understand that their ‘scientific’ tasks — that is the extent to which they must be reality based — as just one aspect of a rich menu of things they must try to master. They must:
- imagine alternative futures,
- try to determine which ones to aim for and how.
Then they need to begin the practical aspects of their design practice which involves helping craft skill improvements that dovetail with the future they’re aiming at and endlessly testing their progress, battle hardening what progress has been made and trying to build it into business-as-usual so it can become the foundation for the next set of improvements. And they’ll be much more successful at all of this to the extent that they can involve others in as many of these tasks as they can.
That’s designing your favourite version of the multiverse of possibilities, which, to the extent appropriate will include scientifically investigating the universe as it is.