Anthony Evans has written a fine paper (pdf) proposing how economists might make themselves more useful. There’s much to like about this - not least his stress upon the need for economists to be imaginative and his claim that it is “simply anti-intellectual” to believe in economic forecasts. Curmudgeon that I am, though, I’d like to quibble on two points.
First, I fear Anthony has too much confidence in economists’ ability to build useful scenarios. The problem is that extreme events are often not captured by scenarios. For example, back in 2007 loads of economists had a disaster scenario. But these revolved around an unwinding of consumer debt, or a meltdown of hedge funds, or a dollar collapse triggered by global imbalances. Very few indeed had a remotely accurate credit crunch scenario.
And this is not an unusual failing. One of my old bosses used to work for an oil company in the early 80s. He was fond of saying how, in 1980, they had all sorts of long-term scenarios for an oil price of $100+pb, or $50pb, or whatever. But not one had oil dropping to $10 as it did in 1986.
Personally, I find simple volatilities more useful than scenarios. These told us in 2007 that there was a roughly one-in-six chance of equities dropping 20% or more over a 12-month period. This simple number was a good reminder of the likelihood of trouble. By contrast, even an accurate scenario is inviting dismissal: “yeah, but how likely is that?”
Secondly, there’s something economists can do that’s implicit in Anthony’s paper but should be explicit - namely, focus upon mechanisms.
We can group these into (at least) six categories. They are: standard neoclassical mechanisms - the Econ101 toolkit; the role of institutions; the effect of social norms; of bounded knowledge (but full rationality); of principal-agent issues; and the impact of limited rationality.
Good economic story-telling will often involve all six. Take for example, the question: what would happen if we abolished the minimum wage?
Ordinary neoclassical economics says wages will fall and this will price some workers into jobs. But the story doesn’t end there.
Putting institutions into our story will tell us that, thanks to the operation of the tax credit system, tax credits for some workers will increase as wages fall. So inequality might not increase much, but only because government spending does. A focus upon institutions would also make us ask whether monopsony models - which predict that employment might fall as wages fall - are relevant.
Then we can think about social norms. These might lead some workers to reject the offer of a low wage because the NMW established the norm of a fair wage. They might think: “hey, that’s less than the old minimum wage. It’s exploitative. Stuff your job” If so, we’d expect workers not sensitive to such norms of fairness to take the job instead. This might lead to increased employment of migrants, and less employment of indigenous workers - especially those with other options, such as students or housewives.
The comes the question of bounded knowledge, but full rationality. Put yourself in the position of a boss recruiting low-wage workers. He might not be able to measure their quality before he hires them - his knowledge is limited. So he might not want to cut wages, for fear of only being able to recruit terrible workers. In this case, abolition of the NMW might not reduce wages in the first place.
We can then add in principal-agent problems. As wages fall, workers whose effort can’t be easily directly monitored by bosses might start to shirk more, so productivity falls.
What might a firm do about this? This is where bounded rationality comes in. If bosses are rational, they might anticipate this extra shirking and decide not to reduce wages. But if they are irrational, they might be taken by surprise by the extra shirking. In such a case, productivity falls as wages fall, leaving firms no better off than with a minimum wage, and so no more able to expand.
Of course, good economics would flesh these points out further, and add some empirical evidence.
None of this is inconsistent with what Anthony says. Quite the opposite. This focus upon mechanisms requires economists to use imagination - to ask: “what would I do in that guy’s shoes?” - and to tell stories. Good economics is about more than cranking through models and generating numbers. All of which is what Anthony’s saying.
First, I fear Anthony has too much confidence in economists’ ability to build useful scenarios. The problem is that extreme events are often not captured by scenarios. For example, back in 2007 loads of economists had a disaster scenario. But these revolved around an unwinding of consumer debt, or a meltdown of hedge funds, or a dollar collapse triggered by global imbalances. Very few indeed had a remotely accurate credit crunch scenario.
And this is not an unusual failing. One of my old bosses used to work for an oil company in the early 80s. He was fond of saying how, in 1980, they had all sorts of long-term scenarios for an oil price of $100+pb, or $50pb, or whatever. But not one had oil dropping to $10 as it did in 1986.
Personally, I find simple volatilities more useful than scenarios. These told us in 2007 that there was a roughly one-in-six chance of equities dropping 20% or more over a 12-month period. This simple number was a good reminder of the likelihood of trouble. By contrast, even an accurate scenario is inviting dismissal: “yeah, but how likely is that?”
Secondly, there’s something economists can do that’s implicit in Anthony’s paper but should be explicit - namely, focus upon mechanisms.
We can group these into (at least) six categories. They are: standard neoclassical mechanisms - the Econ101 toolkit; the role of institutions; the effect of social norms; of bounded knowledge (but full rationality); of principal-agent issues; and the impact of limited rationality.
Good economic story-telling will often involve all six. Take for example, the question: what would happen if we abolished the minimum wage?
Ordinary neoclassical economics says wages will fall and this will price some workers into jobs. But the story doesn’t end there.
Putting institutions into our story will tell us that, thanks to the operation of the tax credit system, tax credits for some workers will increase as wages fall. So inequality might not increase much, but only because government spending does. A focus upon institutions would also make us ask whether monopsony models - which predict that employment might fall as wages fall - are relevant.
Then we can think about social norms. These might lead some workers to reject the offer of a low wage because the NMW established the norm of a fair wage. They might think: “hey, that’s less than the old minimum wage. It’s exploitative. Stuff your job” If so, we’d expect workers not sensitive to such norms of fairness to take the job instead. This might lead to increased employment of migrants, and less employment of indigenous workers - especially those with other options, such as students or housewives.
The comes the question of bounded knowledge, but full rationality. Put yourself in the position of a boss recruiting low-wage workers. He might not be able to measure their quality before he hires them - his knowledge is limited. So he might not want to cut wages, for fear of only being able to recruit terrible workers. In this case, abolition of the NMW might not reduce wages in the first place.
We can then add in principal-agent problems. As wages fall, workers whose effort can’t be easily directly monitored by bosses might start to shirk more, so productivity falls.
What might a firm do about this? This is where bounded rationality comes in. If bosses are rational, they might anticipate this extra shirking and decide not to reduce wages. But if they are irrational, they might be taken by surprise by the extra shirking. In such a case, productivity falls as wages fall, leaving firms no better off than with a minimum wage, and so no more able to expand.
Of course, good economics would flesh these points out further, and add some empirical evidence.
None of this is inconsistent with what Anthony says. Quite the opposite. This focus upon mechanisms requires economists to use imagination - to ask: “what would I do in that guy’s shoes?” - and to tell stories. Good economics is about more than cranking through models and generating numbers. All of which is what Anthony’s saying.
Thanks Chris, I'm chuffed that you found the paper of interest. Regarding your points:
I fear Anthony has too much confidence in economists’ ability to build useful scenarios
When I wrote the paper I was only thinking about academic economists, and I have very little confidence in their ability. As is probably clear I relied heavily on blog posts by Tyler Cowen, Jeff Hummel, Bryan Caplan, and these guys are the only ones I found who are even remotely asking the right questions.
I hadn't considered the extent to which non-academic economists can and do use scenarios. I'm not familiar with any good studies, but they must exist. Having said that, any decent scenario analysis will consider "black swan" events (and would have done so long before Taleb popularised the notion of Knightian uncertainty).
The whole point is to mock the "how likely is that" response because it's that type of thinking that makes people focus narrowly on what's "likely" and completely ignore the once-a-generation event that has a massive impact.
Regarding your other point, as you say it's not inconsistent with what I'm trying to say. I only wished more economic journalists used their imagination, rather than scurrying around for a predication.
Posted by: aje | September 01, 2009 at 04:49 PM
We can group these into (at least) six categories. They are: standard neoclassical mechanisms - the Econ101 toolkit; the role of institutions; the effect of social norms; of bounded knowledge (but full rationality); of principal-agent issues; and the impact of limited rationality.
Alternatively, they can try to grasp some very fundamental economics - as long as the CBs rule the roost and FRB is the rule, the same sorry scenatio will be repeated - for profit.
Posted by: jameshigham | September 01, 2009 at 07:36 PM
This sounds like a serious bout of 'whatifin'.
You can only really plan against the things that predictably have happened in the past. Speculating over what might happen in the future is seriously mad.
In dynamic systems, it is best to remove the things that predictably cause systems to malfunction. Targets and bonuses are two of those things as the Systems Thinker John Seddon has highlighted.
By removing the systems conditions linked to dysfunctional behaviour you can reduce some of the more extreme events and increase the healthiness of the financial institutions.
See The Systems Thinking Review for examples of targets and standards and the impact that they have in the public sector
http://www.thesystemsthinkingreview.co.uk/
Posted by: [email protected] | September 02, 2009 at 08:48 AM
"You can only really plan against the things that predictably have happened in the past. Speculating over what might happen in the future is seriously mad."
I don't even know what this means. Are you suggesting that it's impossible to act under uncertainty? That wondering about what might happen is "mad"? Bizarre...
Posted by: aje | September 02, 2009 at 10:51 AM
And a lot of it reflects a switch from bank deposits to securities; foreigners “other investments” in the UK, http://www.watchgy.com/ mostly bank deposits, fell by £143.2bn in Q1. And of course there’s no guarantee such buying will continue.
http://www.watchgy.com/tag-heuer-c-24.html
http://www.watchgy.com/rolex-submariner-c-8.html
Posted by: rolex datejust | December 27, 2009 at 04:40 PM