Chris Giles says that in predicting that a no-deal Brexit will cause house prices to fall 35%, Mark Carney’s has fallen “into the trap of giving his audience precise numbers at a time they are neither knowable nor helpful.” This poses the question: when is precision and detail useful, and when not?
In this case, Chris is right. A precise number gives the impression of knowledge which is in fact not available to us. I’d prefer that Carney’s warning was expressed as:
We have mechanisms which point to house prices falling because of Brexit: lower income expectations, plus heightened uncertainty (plus momentum effects after these have kicked in). These probably outweigh the positive effects upon house prices: increased optimism among Brexiters; and slightly lower interest rates. Although we can’t quantify how these mechanisms will play out, we can be moderately confident of the direction of change.
There’s another context when detail is unimportant – when we put it before general principles. Take, for example, McDonnell’s plan for increased worker ownership. The significant thing here – at this stage – is not the details, but the question: do the benefits of greater worker ownership outweigh the disbenefits of a slightly higher cost of capital caused by the dilution of owners’ current stakes?*
This is especially the case because the detail – at this stage – is malleable. The £500pa cap on the amount of dividends workers can get, or the 10% ceiling on their stake, can both change in light of evidence.
In fact, there’s another reason why the details of MCDonnell’s plan aren’t very important. It’s that our uncertainty about the size and net direction of their behavioural effects surely swamps any quibble about those details. There is, as Thomas Meyer wrote, sometimes a trade-off between truth and precision in economics. Precise numbers can distract us from what’s really important, such as the confidence intervals around those numbers, and the costs of being wrong. In the run-up to the financial crisis, banks had precise estimates of value at risk which proved to be useless. They and us would have been better served by less precision and more truth. To take another, example, media demands that parties’ spending plans be “fully costed” ignore the fact that there’s massive uncertainty about future government borrowing.
Keynes didn’t actually say “it is better to be roughly right than precisely wrong”, but he should have done.
This speaks to another context when precision is irrelevant – when it is mere noise. The FTSE 100 fell by precisely 35.24 points on Friday. For almost all practical purposes, however, this knowledge is useless. It tells us nothing worth knowing, and certainly nothing about what really matters for investors, which is where the market is heading. In fact, for the purposes of this question, almost all precise knowledge is useless – and in fact, worse than useless if relied upon for asset allocation.
So, when does detail matter? One is that it can be a guide to whether a policy is practical. McDonnell’s plan, whatever other demerits it might have, clears this hurdle. Brexiters’ plans for the Irish border, by contrast, don’t; they amount to little more than “technology, yeah!”
Detail also matters, of course, when it cannot change – when we’ve signed contracts and treaties. It’s reasonable to demand more detail of those proposing a trade deal with the EU than of those proposing a change in tax rates: a trade deal is a bastard to change, a tax rate less so.
It also matters as a way of telling us whether somebody has thought seriously about a problem or not. The lack of detailed proposals for Brexit, for example, is as Chris Grey says, a sign that Brexiters just haven’t confronted basic realities.
Finally, of course, there is some truth in the old saw, “the devil is in the detail.” I suspect, for example, that the appropriate amount of worker say in the running of any business depends upon the detail of how knowledge, and about what, they have relative to management. This will differ from firm to firm, and cannot be known to policy-makers. Hayek’s great insight, of course, is that sometimes (often) important detail cannot be known by a single person.
Sometimes, then, detail matters and sometimes it doesn’t. I fear, though, that the media doesn’t sufficiently appreciate this. Precision makes better headlines: “Carney warns of 35% slump in house prices” is a better headline than “Carney thinks house prices will probably fall by an unknowable amount.” And interviewers sometimes prefer cheap gotcha quizzing about unimportant detail than about points of principle (at least when they are interviewing John McDonnell rather than Brexiters).
There are two useful things we should bear in mind here.
One is number sense. Sometimes, what matters is not a precise number, but questions such as: how confident can we be about that? Where does that number come from? Is it a lot or a little? The other is to be on guard against the illusion of knowledge. Detail and precision sometimes give us mere overconfidence, not genuine useful knowledge. We should ask: why does this matter? When we do, we can see that precision is only sometimes helpful.
* The policy is about corporate governance, not about giving a short-term boost to workers’ incomes – a point some have missed.