Stumbling and Mumbling

On forecasting failures

Economists take a lot of stick for their inability to foresee crises and recessions. But I wonder: wouldn’t it be fairer to chastise politicians for their lack of foresight?

Two current events make me ask.

One is the scandal about Windrushers being deported.  This injustice was not just foreseeable but foreseen. Ministers were warned of it two years ago. It should be no surprise that a “hostile environment” and crude targets would hit law-abiding British citizens. As I said in 2016:

Just as “anti-terror” laws have been used to harass journalists and peaceful protestors, so immigration controls will hurt decent people. And for the same reason - because they are the softest targets.

The second is that confusion about the fate of the Irish border is holding up Brexit negotiations. Again, this should be no surprise: we were warned of the problem at the time of the referendum.

And yet in both cases, some ministers and commentators seem surprised. They remind me of the hotel guest who complained of the poor view to whom Basil Fawlty replied:

May I ask what you expected to see out of a Torquay hotel bedroom? Sydney Opera House perhaps, or the Hanging Gardens of Babylon, herds of wildebeest sweeping majestically?

 

In this context we should ask: what can people reasonably be expected to foresee and what not?

It might be that recessions are inherently unforecastable because they are emergent phenomena which arise from micro-level shocks which are sometimes amplified by network effects. Similarly crises result from falls in asset values. But the same “often negligible” (Keynes’ words) knowledge of the future that causes asset values to be precarious also stops us from predicting such falls.

Maybe, therefore, crises and recessions are not reliably forecastable.

But the political events I’m considering are not like these. Everybody should know that targets can be misused. And anybody who had thought about knew that a hard Brexit meant a physical border between Northern Ireland and the Republic.

This is not of course to say that all political problems are foreseeable and all economic ones not. All I’m asking is for us to consider in which contexts it is possible to have foresight and in which it isn’t. Only by answering this can people be properly held to account for not foreseeing events. Only if you have a theory of what is knowable and what isn’t can you fairly blame people for not knowing something.

Commitment & analysis

Is political campaigning compatible with the detached analysis required of journalism? I ask because of something David Aaronovitch tweeted about Owen Jones:

He is a tireless political activist in a partisan cause seeking very specific outcomes. Nothing wrong with that, but it doesn't sit at all with analysis. There's no distancing possible. Or contrary information.

This might be true. The danger in Owen’s campaigning is that he meets too many people who agree with him which reduces his self-doubt: they say you should never meet your heroes but you should never meet your admirers either.

On the other hand, though, doing so means he meets people outside the Westminster Bubble, and folk on the doorstep and on the streets who tell him he’s wrong.

I can’t say which mechanism dominates.

What’s more, we’ve no reason to suppose that centrists are any more detached and self-critical than active partisans. As my blog’s tagline says, there’s a difference between extremists and fanatics: it’s possible to be a fanatical centrist and a non-fanatical extremist.

The newspaper columnist who associates with like minds at work and at home is perhaps as prone to groupthink as Owen is. Associating with liberal centre-leftists and liberal centre-rightists does not expose you to a good range of opinion: “we are all different, they are all the same” is one of the more egregious cognitive biases. This might be why many centrists were shocked by Brexit and the rise of Corbynism.

What’s more, we all tend to cleave to our opinions once we have expressed them: the danger of doing this was one cease why I ceased to be a sell-side economist. Ego-involvement afflicts us all, left, right and centre.

It takes discipline and mental strength to question one’s own opinions and seek discomfirming evidence. I don’t see a good reason why these abilities are more likely to be found in the centre of the political spectrum than on the left.

In fact, perhaps the opposite.

We must always ask: if my ideas are so good, why are they unpopular?

We leftists have two replies to this.

One lies in economic conditions. Capitalist stagnation has made people more illiberal and intolerant. I fear that centrist commentators aren’t sufficiently awake to this. They don’t sufficiently appreciate that economic realities have changed since the 1990s, and this justifies policies which are very different from Blairism. One virtue of Marxism is that it forces us to look at the economic base. You change people’s minds by changing their living conditions. Centrists don’t seem to me to get this.

Secondly, we have a theory of ideology. Many leftists tend to blame the media. I think they exaggerate: the fact that people supported injustice before the emergence of the mass media suggests another mechanism is at work. That mechanism is ideology: capitalism emergently produces beliefs which sustain the system. Whilst many centrists rightly applaud and use the work of Daniel Kahneman, they fail to appreciate sufficiently that it can be seen in part as supportive of a Marxian theory of ideology.

One of the big challenges we face is how to combat this. I fear that wagging the finger and telling folk they’re wrong isn’t sufficient.

There’s something else. We leftists no longer see the transition to socialism as being achieved by protest or storming the Winter Palace. Instead, we regard it as a matter of interstitial transformation (pdf), or accelerationism – of helping to encourage and expand non-capitalistic forms of organization. For me, socialism is about empowering people in real senses. This is why many of us see Brexit as a genuine dilemma: the voice of the people, even if mistaken, counts for something. This is miles away from the technocratic top-down policies which centrists have often favoured.

I hope that this suffices as a sketch to suggest that political commitment and analysis are compatible. Of course, this isn’t to say that they always are in practice. But the notion of dispassionate centrists facing unreflective leftists is surely simplistic.

I know you’ll reply that my notions of leftism are idiosyncratic. But are they? Or is it just that they seem so because these ideas aren’t often heard in the mainstream media?

* I know. Centrists such as Larry Summers (who re-coined the phrase secular stagnation) and technocrats at the IMF, among other places, have written tons about stagnation and inequality. But this hasn’t percolated through to newspaper columnists.

The media closed shop

I suspect that Owen Jones gets even more abuse when he’s right than when he’s wrong. So it has been with his claim that national newspapers and broadcasters are “full of people who made it because of connections and/or personal background rather than merit.”

This is true. As he points out, there’s lots (pdf) of evidence (pdf) that columnists and top journalists are disproportionately from posh backgrounds. Of course, some of these have merit. But the fact is that background does matter.

Owen gives some reasons for this: the decline of local newspapers as a stepping stone for talented journalists from poor backgrounds; the reliance on internships which are (mostly) only accessible to those from rich families; and the sheer expense of living in London.

I want to add something, though. It’s that there are strong economic mechanisms whereby even well-meaning senior journalists might prefer to hire people from posh backgrounds.

They need people they can trust – to file reports on time, spell names right and avoid libel actions; they do not want to have to rewrite copy moments before deadline. And who are they most likely to trust? It’s people they know – not necessarily family and friends (though as Owen says that happens), but those who have been working as interns. That’s one bias to posh applicants. Failing that, they’ll hire those like themselves: we’re more inclined to trust people who look and sound like us (affinity fraudsters trade on this). Again, this creates a bias to the posh.

Now, there’s nothing unique to the media here. Like hires like in any occupation, simply because it’s a way of mitigating the principal-agent problem.

It’s for this reason that the fact that MI5 used to vet BBC employees matters even today. The men who were selected as “sound” in the 70s and 80s hired the senior people who work there today. And they are likely to have had a bias towards people like themselves - “sound” people. Path dependence thus generates a bias against subversives even if overt vetting has ended.

There’s something else – overconfidence. People from posh backgrounds are more likely to be confident and overconfident than those from poorer ones (the correlation isn’t one, but it’s greater than zero). We all know that overconfident people are more likely to be hired because hirers mistake overconfidence for actual ability.

And this might be perfectly sensible, as Jung Hoon Lee and Shyam Venkatesan point out in a new paper. Overconfident people, they say, are more likely to work hard because they believe doing so will pay off. And these are the sort of people you want. The posh boy who thinks he can become editor (not least because posh boys do) might well work hard in junior jobs in order to get there. As Arsene Wenger said, “if you do not believe you can do it then you have no chance at all.”

On the other hand, whereas some poor kids do have a hunger for success, others lose motivation when they realize they can pay the bills (I know: I’m one). This creates a double selection against such people: they might not get hired in the first place, and even if they do they might stop climbing the greasy pole. Posh people will then be disproportionately in top jobs because those from poorer backgrounds have dropped out: this is perhaps even more true in finance and law than in the media. And because like hires like, so this pattern perpetuates itself.

My point here is that journalism looks like a closed shop not just because bad people hire their mates, but because rational people make rational decisions. This is consistent with the fact that even “liberal” employers such as the Guardian and the art world are biased towards posh white people.

All this is to mostly agree with Owen. Journalism (of the sort he’s discussing) is disproportionately populated by posh people, and this massively distorts political debate. I fear, however, that it is very difficult to change this. Perhaps our best hope is simply that the mainstream media continues to decline.

Good bad theories

In 1945 George Orwell wrote an essay on good bad books – “the kind of book that has no literary pretensions but which remains readable when more serious productions have perished.” He gave as examples the Sherlock Holmes stories. Uncle Tom’s Cabin and Dracula. These are in contrast to bad good books, works of alleged literary merit which aren’t much cop: Orwell cited George Meredith but you can add other examples. We might add to this that there are good good books (Dostoyevsky) and bad bad ones (Dan Brown?)

I suspect that the same distinctions apply to economic theories.

A good bad theory is one that lacks solid microfoundations but which works (roughly) in an ad hoc way. This might be true of the better type of structural economic model and of those lead indicators which predict but don’t explain, such as the tendency for inverted yield curves to predict recessions.

A bad good theory is one that doesn’t necessarily have predictive power but whose failure actually helps to illuminate the real world, often by alerting us to the fact that particular assumptions behind the theory do not in fact hold. In this category I’d put the Miller-Modigliani theorem, the Coase theorem (hat tip Derida derider), the Mehra and Prescott equity premium puzzle (pdf) and the CAPM (pdf).  

Then we have good good theories – ones that work in practice as well as theory. The bog-standard supply and demand framework, transactions cost economics and theories of irreversible investment are all in this set, among others.

Bringing up the rear we have bad bad theories – ones that are wrong in practice but which, unlike bad good theories don’t much help us understand why. Marginal productivity theory (at least for higher-paid workers) belongs in this category. So too does the money multiplier, the LM (pdf) curve, and at least the simpler DSGE models (pdf).

Of course, you might disagree with my examples. And I’ve certainly left a lot out here. Where does Hecksher-Ohlin-Samuelson international trade theory fit? Or representative agent models (bad bad or bad good, I suspect) Or Marx’s tendency for the rate of profit to fall? Or emergence? Etc, etc, etc.

Even so, I suspect this schema might be more illuminating than the standard mainstream vs heterodox one. Debates about this too often consist of: “the mainstream’s rubbish”: “you’re attacking a straw man”: “no he isn’t.” Can we move on?

Facts vs hand-waving in economics

On Twitter this morning Jason Smith asked a good question. Is this, he asked, an “anonymous blog comment from a simpleton? ... Or analysis from a prominent financial economics professor?”:

A company has $100 in cash, and $100 profitable factory. It has two shares outstanding, each worth $100. The company uses the cash to buy back one share. Now it has one share outstanding, worth $100, and assets of one factory. The shareholders are no wealthier. They used to have $200 in stock. Now they have $100 in stock and $100 in cash. It’s a wash.

It could be from the professor, because this is a clear statement of the famous Miller-Modigliani theorem, that a firm’s value is independent of its capital structure.

Equally, though, it could be from a simpleton because we have abundant evidence from around the world that share buy-backs do in fact raise share prices (pdf). Cliff Asness has explained why:

First, repurchases might signal that management believes that shares are undervalued…Second, because interest payments are tax deductible, debt financed repurchases can be viewed as good news due to the resulting lower tax burden. Third, investors may feel as though it is better for management to return excess cash to shareholders, rather than chasing less economic “pet” projects. This kind of agency cost is often characterized as “empire building,” and avoiding it has long been viewed as one of the benefits of returning cash to shareholders.

In other words, the assumptions behind the Miller-Modigliani theorem do not hold: these include no taxes, no agency problem and full information.  

This is not to criticize them. I’ve always read their theorem (pdf)  not as saying “capital structure doesn’t matter”, but rather: “when you see a share price change in response to a change in capital structure, it’s because (at least one) explicitly-stated assumptions do not hold.”

There’s one of these assumptions I want to focus on. It’s that shareholders know as much about the firm as management. If this were the case, buy-backs or dividend cuts would not have any signalling merit. But of course they do: we were reminded of this this morning, when Debenhams share price fell after its dividend cut (though it has since recovered*).

Now, the assumption of full information – not just in this instance but in others – is sometimes justified as a necessary simplification. For me, this won’t do. Bounded knowledge is not some contingent accident we can assume away. It is the human condition**. Assuming that men are gods is no basis for analysing real human behaviour, which is what economics should be.

And this brings me to my beef with comment that Jason cites (and the many more in that vein). Economics, for me, is not about armchair theorizing. It should begin with the facts, and especially the big ones. The facts are that share buy-backs do usually matter, so thought experiments that say otherwise are wrong from the off. Similarly, the fact that wage inflation has been low for years (pdf) is much more significant than any theorizing about Phillips curves. And to take another example, the fact that most fund managers don't beat the market (pdf) in the long-run counts for vastly more than theorizing about the pros and cons of the efficient markets hypothesis. (Feel free to add other examples.)

For me, economics should above all be an empirical discipline. The extent to which it is – especially in the way undergraduates are taught – is a debate I’ll leave to others.

* As of this writing.

** Hayek was explicit about this, which is a big reason why he was not a mainstream neoclassical economist.

On thin predictions

Back in 2016 I wrote that “immigration controls will hurt decent people…because they are the softest targets.” Paul Cotterill interprets this as meaning that I saw the Windrush scandal coming. I think he’s being too kind, but this illustrates an under-appreciated point – that good predictions can be “thin” in the sense of containing little data.

I’ll give you two other examples from my day job.

One is that back in 2007 I pointed to record foreign buying of US shares as a warning that the All-share index would fall sharply over the following 12 months. Which it did.

Then last February I forecast that sterling would rise. Which it has – by 15% against the US$.

Now, I don’t say this to toot my own horn.  (There's a reason why these examples are ten years apart!) I do so to point to something these calls have in common. Both are lacking in description. The forecast that share prices would fall in 2008 makes no reference to the financial crisis. And my forecast for sterling ignored most of the things that are usually thought (perhaps wrongly!) to determine exchange rates such as growth or interest rates.  

 Instead, both these calls rested upon single facts and single theories – that foreign buying of US equities was a measure of irrational exuberance (it still is) ; and that exchange rates overshoot (pdf).

These are not isolated examples. A single simple fact – is the yield curve inverted or not? – has done a better (pdf) job of predicting recessions than the more data-intensive forecasts of macroeconomists, most of whom have missed (pdf) almost all recessions.

What this tells us is that sometimes we don’t need big models to predict things – be they micro-founded or not. One strong fact sometimes beats lots of weak ones*. In these cases, we have lead indicators, but not any detailed explanation.

In saying this I am echoing Jon Elster:

Sometimes we can explain without being able to predict, and sometimes predict without being able to explain. True, in many cases one and the same theory will enable us to do both, but I believe that in the social sciences this is the exception rather than the rule. (Nuts and Bolts for the Social Sciences, p8)

You might think that this contradicts the message of Tetlock and Gardner’s Superforecasting. They say that foxes (who know many things) make better forecasters than hedgehogs, who know one thing.

I’m not sure there’s a contradiction here. For one thing, many of the cases they consider are one-off political events – sometimes quite obscure ones – where success requires careful data-gathering. In the cases I’m considering, however, data is abundant and the trick is to distinguish between signal and noise, to decide what to ignore. Occasionally, one fact sends a stronger signal than many. Which corroborates Tetlock and Gardner’s point, that “in most cases, statistical algorithms beat subjective judgment.”

This is not, of course, to say that simple lead indicators always work. Obviously, they don’t. My point is merely that forecasting is not the same as modelling, nor the same as telling a good story.

* Certainly, rules of thumb often beat attempts to forecast particular events, but that's a different story.

The Windrush scandal: myths exposed

The appalling spectacle of Windrush generation immigrants being hounded and deported should help to dispel two great myths about politics.

Myth one is that racism is confined to the “white working class”. Of course, some of these have backward attitudes which they express crudely. But equally, very many or more have lived happily with and alongside Windrushers. The harassment of them has been authorized by nice posh people who probably wouldn’t dream of using “politically incorrect” language.

Racism is not just about words. It’s about actions. In particular, it’s about the actions of those in power. As I complained months ago, we hear too much about the racism of the white working class and too little about that of the ruling class.

Only a few days ago the BBC and other right-wing media were hyperventilating about the (yes, disgraceful) anti-semitism of a few idiots. They were doing so whilst ignoring a more harmful form of racism being conducted by those in office.

Myth two is the myth of perfectibility – the idea that it’s possible to conduct a perfect policy with no ill-effects.

It is the case that most voters want immigration controls. As with Brexit, Theresa May interpreted this as meaning that they wanted a harsh and extreme version thereof. Hence her creation of a “hostile environment” for migrants: Ian Dunt and Stephen Bush are right to note that the deportations of Windrushers are the natural effect of this.

But this is not what voters wanted. They don’t want Windrushers expelled, just as they don’t want cuts in the numbers of foreign students. When voters want immigration controls, they are thinking of unskilled migrants and criminals, not French doctors, Chinese students or elderly Brits.

Maybe in theory it’s possible to distinguish between “good” and “bad” migrants. And maybe some governments somewhere do have the ability to do so. But the UK does not have that ability. As Amber Rudd said, the Home Office “sometimes loses sight of the individual.” (She’ll be livid when she finds out who’s running the department).

In practice, then, perfect immigration controls are impossible. This means we have to choose which mistake we make. Do we have a “hostile environment” which will hurt decent people? Or do we have a more liberal regime which will bring in some people voters would rather exclude? I’d prefer the latter. As I wrote months ago:

if you give power to the state it’ll be misused, because the actually-existing state is a stupid bully. Just as “anti-terror” laws have been used to harass journalists and peaceful protestors, so immigration controls will hurt decent people. And for the same reason - because they are the softest targets.

The case for liberty is, in large part, that the state is not to be trusted with extensive powers.

It’s tempting to conclude with some bromide hoping that lessons will be learned. This, though, is too optimistic. Politicians and the media rarely learn deep lessons from evidence.

Syria: the knowledge problem

There’s a danger that the question of whether we should intervene in Syria is becoming a left-right issue. Not only is this false, it’s a means of (deliberately?) ignoring the basic issue – one that is much more general than merely the conjunctural question of Syria.

It’s false because there are many rightists who have doubted the case for military intervention, such as John Baron, Julian Lewis and, I gather, Kate Andrews on Question Time last night. I don’t think this is wholly because they are little Englanders who care only about British interests. It’s because of their stance towards a key general question in politics: how much can governments know?

The case against bombing Syria is not that we should support Russia or Assad or that a “political solution” can be found. It’s that we do not know enough about the country to be confident that intervention will work. Yes, the situation is awful. But it’s quite possible that bombing will make it worse.

Faced with uncertainty, we must err on the side of doing nothing: if in doubt, do nowt. The Brainard principle should apply to all policy, not just monetary policy.

To put this another way, bombing Syria is an irreversible decision. We cannot unbomb the country if we decide that we were wrong to do so. If we don’t bomb, though, we can bomb later if we decide that the case for doing so has become strong. Bog-standard economics tells us that when we have uncertainty and irreversible investment opportunities, we should wait and see. (Dan Davies, has, as usual been good on this).

This is one lesson of the Chilcot report. He wrote (par 863):

Ground truth is vital. Over‑optimistic assessments lead to bad decisions.

Can we really be confident we have sufficient “ground truth”?

The contrary case to all this has been put by Johnny Mercer:

On issues of sensitive intelligence and national security, the PM sees the whole picture, and we should not constrain her

This, though, begs the question: does the PM really have the whole picture?

From this perspective, it’s no surprise at all that some rightists oppose military action. Free marketers such as Kate think the government doesn’t know enough about the economy to intervene successfully. By the same token, they might doubt whether governments can know enough to intervene militarily in other countries. This is a consistent scepticism about what policy-makers can know. Hayek’s famous essay on the limits of knowledge doesn’t just apply to economics.

Much of New Labour had the opposite view. I suspect that Blair’s decision to go to war in Iraq was the product of his overall ideology – an overconfidence about what top-down leaders could know.

In this sense, the Syria question should not be a left-vs-right one at all. Free market rightists and market socialists like me agree that centralized knowledge is often insufficient and so are intervention-sceptics. Those who are more optimistic about state capacity disagree*.

The debate we should have – not just in the Syria context but more generally – is: how much can we know? But because many politicians and columnists have built careers upon being overconfident, this is a question they don’t want asked. As Upton Sinclair said, “It is difficult to get a man to understand something, when his salary depends upon his not understanding it".

* You might object that this is not Corbyn's motive for opposing intervention. Maybe not. But doing the right thing for the wrong reasons isn't entirely to be deplored. 

The Tyranny of Metrics: a review

Measurements can mislead us, especially when they are used as management targets. That’s the thesis of Jerry Muller’s The Tyranny of Metrics. He writes:

What can be measured is not always what is worth measuring; what gets measured may have no relationship to what we really want to know. The costs of measuring may be greater than the benefits. The things that get measured may draw effort away from the things we really care about. And measurement may provide us with distorted knowledge – knowledge that seems solid but is actually deceptive.

Muller provides lots of examples of this, mostly from the US. But you’ll all have examples of your own. In universities the Research Assessment Exercise (now the REF) contributed to increased administration costs and perhaps to the replicability crisis by incentivizing the publication of mediocre research. In schools, targets can encourage teaching to the test, endless revision and a focus upon the marginal student to the neglect of both the strongest and weakest. Waiting-time targets might distort clinical priorities. Immigration targets deter foreign students and lead to the harassment of people who have lived here for decades. Sales targets encourage workers to mis-sell financial products, cook the books, or increase risk by encouraging (pdf) “liars’ loans. And so on.

All this will be familiar, especially to those of you acquainted with the work of John Seddon – although Muller puts it all concisely and well: Campbell’s law, which says that measures will be misused, was proposed way back in 1979. And it’s going to remain a big issue thanks to the rise of big data and machine learning – which, I suspect, will be the over-hyped discovery of weak correlations that don’t hold out-of-sample.

The Tyranny of Metrics is not, however, a diatribe against targets. Muller points to the experience of some US hospitals to show that metrics can work. They do so, he says, when they are “based on collaboration and peer review”:

Measurements are more likely to be meaningful when they are developed from the bottom up, with input from teachers, nurses and the cop on the beat.

In other words, metrics can succeed when they are complements to knowledge: when they organize the tacit and dispersed professional judgements of people who know ground truth. Tim Harford makes a good point when he criticizes Muller for overlooking Paul Meehl’s finding (pdf) that simple statistics often out-perform professional judgment, but I don’t think Meehl was arguing that statistical models should be constructed and used in complete ignorance of professionals’ know-how. The latter informs the former, whilst the former in turn checks the latter.

Indeed, Muller’s evidence shows that metrics can fail when they are substitutes for knowledge – when top-down bosses try to supplant professional know-how.

This raises an under-appreciated point - that questions of what statistics tell us and whether target work are not merely technocratic ones. They are deeply political, because claims to power and wealth are founded upon claims of expertise. As Alasdair MacIntyre wrote:

Civil servants and managers alike justify themselves and their claims to authority, power and money by invoking their own competence as scientific managers…But is this true? Do we now possess that set of lawlike generalizations governing social behaviour of the possession of which Diderot and Condorcet dreamed? Are our bureaucratic rulers thereby justified or not? (After Virtue, p86-7).

The belief that people and organizations can be managed and controlled by simple measures imposed from above is one of the main foundations of bosses’ claim to power. In showing us that this belief is at least sometimes wrong, Muller has written a much more political book than many readers might think. And in doing so, he has also helped to advance the case for greater worker democracy.

Politics as risk management

There’s one point about politics which seems to me to be childishly obvious but which is under-appreciated, and which helps illuminate a number of policy issues.

It’s that we don’t live in a wholly knowable deterministic world in which setting policy is a matter of pulling the right levers with predictable effects. Instead, policy operates upon a distribution of uncertainties*. In some cases, the best policy is not so much one that achieves the best outcome but one which reduces the risk of bad outcomes – either the small risk of something really bad or the larger risk of moderately bad outcomes.

We should think of policy as risk management. Just as investors don’t try to maximize returns but also pay some heed to avoiding risk, so too should policy-makers.

As I say, this should be childishly obvious. But let’s apply it to some current issues.

 - Anti-semitism in the Labour party. It seems this is mostly a matter of a few (though too many) idiots. This is low-level stuff: idiots spouting shit is normal in politics. Tolerating it, however, carries a big risk. Anti-semitic words might help to normalize anti-semitic behaviour: attacks and harassment of Jews. Low-level anti-semitism might therefore escalate. It’s reasonable then to demand that Labour have zero-tolerance of anti-semitism not so much because current levels of it are a big problem, but rather as a means of reducing the risk of something worse developing.

 - Calls for military action in Syria. The problem here is that we know that military ventures carry huge risks. The rule “don’t start wars” might not be correct in every case. But it’s a good minimax rule of thumb. It’s a way of avoiding very heavy losses – of lives and money as well as political capital. Yes, interventionists are wholly right to say that inaction has consequences too. But it’s not clear to me that inaction is often as potentially catastrophic as action. This is especially the case because in foreign policy William Goldman’s maxim seems to apply: nobody knows anything. And Brainard’s principle – that policy should be cautious where there is uncertainty – should apply to all policy, not just monetary policy.

 - Cuts to police numbers. Amber Rudd claims there’s no evidence these have led to rising violent crime in London. But if crime were wholly insensitive to police numbers, we could abolish the police entirely without crime rising. That seems absurd. There must therefore be some point at which fewer police leads to more crime. And we have at least one piece of evidence here that we might be around that point. Stephen Machin and colleagues show that when police were redeployed from outer London to central London after the 7/7 terrorist attacks “crime fell significantly in central London relative to outer London.” That suggests police numbers do matter. Granted, we cannot prove that people have been killed as a result of police cuts. But it is surely plausible that the cuts have increased the risks some people face of suffering crime.

 - Fiscal policy. Again, this is a matter of risk management. As Simon has said, the problem with austerity is that it is risky. It increases our reliance upon uncertain monetary policy – policies which our low rate of growth show to have been inadequate. On the other hand, its advocates claim that austerity reduced the risk of a bond market crisis and rising borrowing costs. Personally, I think that risk was small and manageable by QE. But the point is that the debate is about risk management.

I could of course go on: one case against Brexit, for example, is that it increases uncertainties for (to me) no offsetting benefit. Conversely, the virtue of many policies aimed at increasing productivity – better education, stronger competition policy, better financing of start-ups and so on – is not so much that they will definitely work but that they are free hits: even if they don’t raise productivity they’ll do little harm and perhaps some good.  

Although the general principle here should be obvious to all adults (though you may well disagree with my applications of it) I fear it is under-appreciated in the media. Columnists are more often than not overconfident gobshites who fail to grasp the extent of uncertainty and boundedness of their knowledge: how often do they attach confidence intervals to their claims?

And imagine – if you can - the improbable event of a politician appearing on (say) the Today programme and saying “We cannot know the precise effects of this policy, but we hope it will at least avoid serious harm.” John Humphrys would give her a horrible time. Our dumbed-down media, with its demand for certainty where none is possible, mitigates against sensible politics. “We don’t know” is a phrase that should be used more often.

* I mean uncertainties in the Knightian sense: often, we cannot quantify probabilities or know their precise distribution.