September 5, 2016

A clear statement about migration

I wrote this for Politico Europe, but they weren’t interested after much editing about. Apparently there were too many charts.

A clear statement about migration, says Theresa May of the vote for Brexit. The last thing you’ll find in the data is clarity. Or migration.

There has been a wealth of efforts to understand Brexit through data. But the most telling statistic in most of them is the R^2 value, the measure of how well a regression line fits the data. The higher the R^2, the more of the spread in your data you’ve managed to explain. Famously, although there is a faint correlation between some measures of migration and the vote, the R^2 value is pathetic – the data set is nothing but outliers.

It gets worse. Some analysts tried to save migration as an explanation by looking at the change in foreign-born population, rather than its level. This chart from The Economist is the classic statement. Perhaps the voters were shocked and bewildered by the speed of change, rather than its content, or something like that. Or maybe it’s a soft racist argument like Jacques Chirac’s Le bruit et l’odeur speech.

The problem is, again, the R^2 – without a very few extreme outliers, mostly very conservative small towns in the Fens with significant numbers of migrant farm workers, there wouldn’t be any effect at all, as Jo Mitchell points out here. You’ll notice they didn’t quote an R^2.

It gets still worse, though. Those outliers are dramatic, but they disappear when we control for the size of their population (from here). Small populations exaggerate all percentage changes; they show extremely high rates of immigration precisely because they have so few immigrants, and even if they voted Leave by a big margin, they had little impact on the contest because they have so few voters. We can deal with this by plotting votes rather than percentages – I’ve plotted the net Leave lead, i.e. Leave minus Remain, giving us each local authority’s contribution to the overall result.

As you can see, the Fenland outliers have vanished and so has the correlation. It makes sense; nobody ever won a general election in South Holland and the Deepings, a constituency that has been Conservative since 1922. Instead, a clutch of populous, Leave-leaning but contested, urban but not metropolitan districts emerge as the key battlegrounds. Dudley, for example, contributed 61,666 net Leave votes.

Let’s try something else. One argument – classically put by Daniel Davies in Vox – is that the problem is migration, but it’s internal migration. Post-industrial northern towns and the run-down seaside are emptying out as the young seek opportunity in the big city. It’s an elegant argument, with all the more emotional force because both Dan and I did just that ourselves. Unfortunately, the data doesn’t stand it up.

In fact, it’s the other way round, the correlation is quite strong, and the R^2 is at least less bad. This seems baffling; London is actually more populous now than it’s ever been, and deeply Remainy. Also, Cornwall is hardly the land of opportunity. The explanation may be age structure – young people flock into the big cities, retirees go the other way – but we’re already trying to save our hypothesis by fitting stories to an unconvincing data set. It needs a lot of nuance.

What about total population growth? Sorry, but that’s even less helpful.

We could try some other approach. The Right is convinced it’s all about immigration. The Left is convinced it was a massive protest vote about austerity. This is hard to test because there is no official data on total government spending by locality. Without it, we’d have to build our own private hell of cost-allocation problems. The Centre for Cities managed to create a snapshot for 2013-2014, but austerity is all about change in the fiscal stance. Also, a Keynesian would object that the allocation of the government deficit is what counts, so we’d need tax revenue data as well. And, anyway, it doesn’t tell us much.

There is, however, data for spending by local governments. A large fraction of the UK austerity programme consists of cuts to the Department for Communities and Local Government’s financing to town councils, so this ought to be a useful proxy for total spending.

We don’t find much correlation with the level of spending. But we do with the change from 2012-2016. Austerity, defined as the reduction in local government spending power, predicts about 20% of the variation in the net vote for Leave.

Another variable that does seem to have some predictive power is pay. The short-term change in median gross weekly earnings doesn’t seem to matter, but their level does, quite strongly. In fact, it’s better than austerity as a predictor; it’s the best one I found, with R^2 of 0.23. I ran the same analysis, using the Annual Survey of Hours and Earnings data set, for the 25th, 75th, and 90th percentiles of earners, but I didn’t find anything interesting.

So, it looks like the immigration story is a bust. The internal migration one might be saved with a lot more nuance, and you know what they say about nuance? And a pretty direct leftwing story about austerity and poverty seems to work better than anything else. Until you actually look at poverty itself. The standard measure of poverty in Britain is the government’s Index of Multiple Deprivation, which sums up a gaggle of social evils as a handy score.

This is interesting because the IMD tells us what happened after the welfare state did its thing – it’s a measure of poverty and inequality after redistribution, while the ASHE is a measure of income as determined by the market, before taxes and transfer payments. And the IMD doesn’t seem to show any correlation at all.

Now this is interesting. A major economic and political orthodoxy throughout the world since the late 1980s has been that economic change, however jarring, is basically healthy because the winners can compensate the losers through the tax and transfer system. This doctrine was the source of legitimacy for the whole free-trade agenda – NAFTA in the States, Single Market completion and the Eurozone in Europe. And now it’s breaking down. Transfers don’t buy legitimacy, and maybe they never did.

For the UK, there’s an important and difficult problem here. The UK doesn’t do regional policy well, but redistribution of income between regions does happen to a very significant extent. Leaving aside the rows about the Barnett formula and Scottish oil, let’s just remember that 30% of all taxes paid in the UK are paid by Londoners, who make up 13% of the population. How much more are you going to ask them for?

Part of the problem is that redistribution must be done, but it must also be seen to be done, like justice. The UK, very unusually among federations, doesn’t really have an explicit political process to determine how government spending is divvied up. There’s no equivalent of a Länderfinanzausgleich. Therefore, it’s hardly surprising nobody thinks they have any control over it.

This also suggests another way economic unions fail. It’s a commonplace that the Eurozone is troubled because it lacks both a big discretionary budget and full labour mobility, unlike the United States or Germany. Therefore, bits of it can end up with an inappropriate real exchange rate and high unemployment for an indefinite period of time. But the UK doesn’t have any language barriers, and it does have a big federal budget. You could say the same for the US. However, bits of it still seem to end up stuck in a low-income equilibrium for decades.

Perhaps that internal migration hypothesis deserves another look? Perhaps, without explicit and forceful regional policy, some places just empty out? And does that remind anyone of Donald Trump?

August 23, 2016

Fear and Loathing on the Campaign Trail 1972, in 2016

There are some books I re-read regularly. Hunter Thompson’s Fear and Loathing on the Campaign Trail ’72 is one; I read it every time there’s a presidential election on.

In 2008, the first time around, it was a ghastly memento mori for the failure of a great movement campaign, and also a reassuring reminder of the consummate competence of the Obama campaign. In 2012, as the Obama campaign purred along to the win, it formed a sort of demented counterpoint of bad craziness, this time mirroring the weirdness of the Republicans.

This time? I got a gaggle of insights into how a great insurgent campaign fails. If the 2008 reading was about Obama, the current one is about Sanders, Trump, and Jeremy Corbyn. Obviously, Trump is perfect for Thompson’s style, but that’s not my point.

Thompson mentions, in the postmortem interviews near the end of the book, that McGovern did startlingly badly with black voters. This is telling because HST doesn’t talk much about that for the rest of the book. He talks about individuals, but he doesn’t talk about how McGovern tried to address blacks or didn’t, which is weird because one of the best things in the book is Thompson’s coverage of the practicalities of politics.

An obvious conclusion is that he didn’t write about it because it wasn’t there. The same has been said about Bernie Sanders in this classic blog post, and we probably need to talk about the Corbyn offering the two black ladies each others’ jobs.

Thompson also discusses how he thought McGovern was successfully addressing the working class during the primaries, but failing to do so during the general election. Again, he doesn’t want to talk about his own besetting case of the great hippie sin, massive condescension to workers and to the unions that represented them.

The rallies were great. The volunteers were all bright and young and handsome. The campaign did nothing better than campaigning to its own volunteers.

What Thompson didn’t and couldn’t know is that both the collegiate sneering and the Boss Tweed swagger would be swept away by the macroeconomic revolution of Reagan and Paul Volcker soon enough, of course. Thompson missed that this civil war in the Democrats would destroy both combatants, which is weird seeing as he classically diagnosed it in Fear and Loathing in Las Vegas. Perhaps he couldn’t accept his own role in it.

With regard to the Corbyn, it’s more the other way around. He addresses the RMT tolerably well. The problem is everyone else. If we’re lucky and Duncan Weldon is right, the macro shift that finishes him might be a swing to reflation across the G-7. If we’re lucky.

And, of course, what HST wanted was a campaign that specifically would destabilise and maybe finish the Democratic Party. He repeatedly says as much. Oddly enough, he didn’t much like the results. But that’s what happens if you decide to piss your side’s institutions up the wall because the rallies are great.

The specific failure mode is that if you don’t care about that stuff, you end up not caring about the skills and operational practices that underlie it. Everyone but HST thought the killer was competence; lacking it, the campaign never got close enough to put Nixon under pressure.

This year, the book is a painful story of a movement campaign that never escaped from its deeply hippyish and middle-class roots and that as a result risked losing the whole party.

August 22, 2016

Please stop looking for Labour voters in the Fens. They never existed

Chuka Umunna is apparently off to Boston, Lincolnshire to understand Brexit. He should save himself the journey. You can understand Boston really well from the simple fact that it has elected a Labour MP – indeed anything other than a Conservative – precisely once in history, immediately after the First World War. It’s really conservative. This should be simple. If any political party should worry about it, it’s the Tories. It’s their seat!

But there’s a weird kind of tourist industry in going there to look for the Very Real Concerns. John Harris seems to be there every other week. I suspect it may be something to do with the fact it’s really conservative big-farm country, and hence it’s easy to find copy-generating racists, and that it’s not actually that far from King’s Cross station, and therefore a reporting trip there is both practical and cheap.

Also, if you were to ask my dad the ex-immigration officer, you’d learn that immigrants certainly didn’t start showing up there one day in 2004; for many years before that, there was a specific carve-out of the rules for seasonal farm workers going there. Immigration there isn’t new; it’s probably as old as the engineering interventions that made it farm-able just after the civil war. A 70s Cambridge Marxist might have made something of that, but nobody imagines UKIP voters in Boston spend their spare time reading Christopher Hill.

I have a less snarky point, though. Boston voted for Brexit by the highest percentage in the country. This has a lot to do with the fact not many people live there. The Economist is very pleased with the insight that although it doesn’t have that many immigrants, the percentage increase is large. This has a lot to do with the fact that not many people live there and not that many immigrants choose to live there.

Consider the following chart. I’ve plotted the population change due to immigration against the Leave lead in votes. Referendums aren’t population weighted – it profits you not to win Skye by 90% but lose Haringey by 1%, in much the same way that piling up votes in safe seats doesn’t help you win a parliamentary election. Also, I’ve started the clock in 2004, the year VRC fans reckon it all went wrong.

Note that when you take into account how many votes they actually delivered, the Fenland outliers just vanish. They’re just not very important. Think about it; who ever won a general election fighting hyper-safe Tory seats in the Fens?

On the other hand, look at Dudley, which delivered 61,666 net Leave votes all on its alone-io, more than anywhere else in the country. Dudley has about four times the population of Boston, and delivered about three times the net Leave vote. We could look at Doncaster, or Wakefield, or Sandwell. Wakefield has the same level of immigration in that timeframe as Wandsworth, but it was about a third more Brexity. Lambeth had four times as many immigrants as Wandsworth, but it was just as Remainy.

Sandwell! That’s next door to Dudley. From a Labour point of view, we seriously failed in the West Midlands in 2015 and that was probably why we didn’t win. It looks like that’s why we failed to prevent Brexit. Whatever the man who wants “Decent funky house and no trash” – don’t we all – is up to, his time would be better used in Birmingham.

Referendums, like elections, are won where the net votes are. Labour isn’t going to take Boston or Wokingham, we probably shouldn’t care about that, and we shouldn’t trade winnable seats for forlorn hopes.

August 20, 2016

Wild speculation rides again!

Here’s a translation of the Israeli Defence Forces’ new strategic concept, from the Belfer Centre at Kennedy School of Government. From the section on “Characteristics of the Operating Environment”:

Increased threat of fire on the home front (volume, pace, accuracy,
size of the payload, survivability) and an attempt to create a strategic threat against national weak spots and the national economy. This is in addition to an ongoing endeavor by the enemy to assure the survival of its firepower through decentralization, camouflage, protection and the use of the civilian environment to provide it with a bargaining chip and “victory photographs.”

This July 2014 TYR Flashback refers.

July 28, 2016

A Brexit charts tip

Picking up on a tip from Jo Mitchell’s excellent post here, I replotted the data in the previous post using the Leave lead in votes, rather than percentage terms. This is, of course, appropriate because referendums aren’t counted in terms of parliamentary constituencies. Also, this means that tiny outlier constituencies – looking at you, South Holland and the Deepings – aren’t overrepresented relative to places like, well, Birmingham.

A couple of interesting things result. For a start, the R^2 value doubles to a whole 0.21! I win at data! Secondly, putting the Fenland outliers back in their box tells us something interesting and important, which is that austerity-battered northern and midland cities contributed hundreds of thousands of net Leave votes. Not only was winning in Boston, Lincolnshire probably impossible, it couldn’t possibly have helped in the light of relative failure in the West Midlands. There are lots of people in Birmingham. Labour’s Midlands problem is a thing.

Pulling London out of this doesn’t help much. R^2 nudges up by a hair. It further confirms, though, that you should stop obsessing about tiny places that voted weirdly, and worry about big ones that voted kind of normally but for the wrong side.

In that way, the Breferendum was far more like a parliamentary election than it seemed. Where it was decided, austerity does actually seem to have mattered quite a bit.

July 25, 2016

Austerity And Brexit

Ever wondered how austerity affected Brexit? Sure you did, but there’s a reason nobody made a nice chart yet. To answer the question, you’d need a breakdown of consolidated central, regional/national, and local government spending by some geography or other (parliamentary constituency, local authority, super-output area, whatever). That in turn means you’ve just constructed your own exquisite hell of cost-allocation accounting to deal with the UK’s numerous overlapping jurisdictions.

Something like this data set does exist – here – thanks to the Centre for Cities, who wanted it for this report – but sadly it was so much work that they haven’t either taken it backwards to create a time series or updated it. If you want to know the effects of austerity, your question compiles down to something like “what is the correlation between the vote for Brexit and the percentage change in the sum of real-terms Departmental Expenditure Limit and Annually Managed Expenditure between 2010 and 2016, for each specified geographical area?”, so this can’t help us.

But there is some data available. Since 2012 the Government has published an annual analysis of how much cash money local authorities have to spend, both in absolute terms and on a “per-dwelling” basis as a crude population weighting.

The current report, for 2016-2017 onwards, is here. This doesn’t cover the NHS, most social benefits, big infrastructure, or central government salaries, but it strikes me as a defensible hypothesis that the same councils who get clobbered by DCLG-managed austerity will also get it from the Treasury. Also, it only covers England. But it’s as close as we get to publishing an acknowledged financial settlement between bits of the country.

Here we go, then.

Greater austerity – i.e. a bigger cut in spending on a per-dwelling basis – is to the right. The Leave lead is just the percentage Leave vote minus the percentage Remain vote. The answer seems to be “Nobbut bugger all” – there is a correlation but it’s very, very weak and the R^2 is 0.12.

We can think about this by pulling out some data points with similar values. Uttlesford, which somehow managed to get a 17% uplift in its funding, was a tie; so was Birmingham, despite a -28% spanking. Great Yarmouth did even worse financially, -29%, and voted to Leave by 43 percentage points. Castle Point was even more Brexity than Great Yarmouth although it lost “only” 9.72% of its money – a little less than Kingston upon Thames, which went or rather stayed Remain by 20 points. Great Yarmouth’s financial suffering was rivalled only by Hackney’s, which set the record for voting Remain.

It’s almost typical of Brexit-related charts, really – R^2 values are terrible, outliers abound, and correlations, such as they are, are powered by weird special cases. Ashfield, Nottinghamshire got away with an -8.9% cut, but racked up the sort of Leave vote – 40+ points – you might expect in South Holland and The Deepings. Can it be…the influence of the great statesman who was its MP for 18 years?

Snark aside, I do think there is a serious point here. This was a remorselessly personal, emotional event rooted in life courses and irreducible choice. That’s why statistical aggregates and polling were poor predictors and why impressionistic journalism seemed to work so much better.

July 1, 2016

Something cheerful to take your mind off it

I spent Wobbly Weekend 2: The Wobbling, in part, going through the effects of a deceased relative. I guess it was the right weekend for it; it wasn’t going to depress me any more. Anyway, we found this envelope of briefing documents, dated 1963 at the earliest. A reminder, I think, that things could be a lot worse, although God knows how today’s political elite would have coped with the Cold War.

Context is here.

I didn’t know that the WRVS, at the time joined at the hip to the then Civil Defence system, set out to brief three million women, i.e. 20% or the titular one in five of the female population, with the standard Glasstone Effects of Nuclear Weapons spiel and basic advice (some of it not so basic) on protection against flash, blast, fire, and fallout.

But they did, and the strategy they pursued was pretty interesting – startlingly similar to a classic Alinskyite community-organising campaign, getting opinion leaders, serial volunteers, and people respected in the community (and my Auntie Peggy was all three if she was anything) to take the training course and then spread the message horizontally through their social networks. You wonder what the In campaign might have looked like if we’d adopted that strategy.

Also, I rather like the serif typography – very different to Calvert & Kinnear’s Transport/Rail Alphabet work, although the two are contemporaries. Obviously you wouldn’t expect women to go without serifs, not even in the event of nuclear war!

June 22, 2016

Leave: the biggest 419 e-mail ever

What worries me about the referendum is the trust. Every poll I can think of shows people on the Leave side claiming they trust no-one.

At the same time, Nigel Farage says smoking is just the best for your health and the official Vote Leave campaign issues maps suggesting Iraq and Syria are candidates for accession to the European Union, talks about the finances of the EU without counting the rebate Margaret Thatcher secured, and suggests we all need to be more like Albania.

I can see it – drive my cigarette boat over the Adriatic in a 500 horsepower blitz of seawash, slash some clown who lacked respect and toss his carcass in the harbour, dance like a wild peasant youth with a dark beauty – but I’m not seeing Michael Gove and Iain Duncan Smith keeping up.

There’s a reason why they come out with so much bullshit. There’s also a reason why they come out with so many outright lies. A couple of years back, people at Microsoft’s research lab were trying to work out why spam is so obvious. We’ve all wondered how anyone believes the Nigerian prince and his money are on the other end of the wire, or why anyone bothers trying to sell us drugs that are free on the NHS. The Microsoft researchers realised something important.

Even a leopard needs to sleep. Predators only have so many hours in the day, so they have to pick the prey that gives them the best chance of a feed. Spammers are looking for suckers, but you can’t pick those by looking at their e-mail address. The more time you waste on someone who’ll get wise, the less money you’re making and the more likely you are to get shopped to the police.

What you can do, though, is scare off all the other people. If you make your scam so excessive, so shameless, an outrageous camp striptease of a scam…well, you don’t need to worry about picking your marks any more, because the only people you’re talking to are the ones who weren’t paying attention. All the others have gone. It’s a kind of negative marketing – the point isn’t to get people in, it’s to put them off.

That’s where all this bullshit is coming from. The Leave campaign is the most fully realised exercise in negative marketing yet. The whole point is to show that they’re going to lie flat out to your face. It has to be obvious. They think you’re an idiot, a sucker, a bag of money on legs. You should see it as an insult to your intelligence.

In case you’ve wondered, there’s a reason why every Remain campaign message seems to come from people who define themselves by something they do, from scientists and trade unionists to midwives and skateboarders. We want you to think about what you do best, and act with the generosity of confidence. They want you baffled and scared by what you believe is your own mediocrity. Nothing else could put such a bunch of dullards in charge.

Let’s not be the country where 50% plus 1 vote chooses the Angus Steakhouse option.

June 10, 2016

The public rejects racism, but sadly you can’t say the same for bullshit

One thing the referendum campaign has cheered me up about, paradoxically, is the social acceptability of outright racism. One thing it’s profoundly depressed me about is the social acceptability of outright bullshit.

Consider the Leavers’ arguments about immigration.

If you’re not going to make some essentialist argument that foreigners are just bad – i.e. to come right out as a racist – you’re going to make some argument about population. There are too many people, pressure on public services, etc. But the UK population is growing quite strongly even without the contribution from net migration. Logically, if you believe net migration is a problem because population you should believe population growth is. Nobody on that side wants less population, nor do they have any plan to spend more on public services, develop cities outside London, or whatever.

So it’s only a problem if they’re foreigners? Isn’t that…a bit racist? Well, now we get the argument that we could have just as many immigrants, but from the Commonwealth rather than the European Union. First of all, if you believe this, I’ve got a bridge to sell you. Secondly, this makes so little sense. If you’re worried about too many people, or foreigners taking our jobs, why would Nigerians be any better than Italians?

The next dodge is the famous points system. The problem here is that once you set up a system where you get into the country if you have X points, you’ve implicitly committed to accepting anyone who makes the cut. If you believe that an Englishman has won first prize in the lottery of life, yadda yadda, you should also believe that it would be worth doing almost anything to rack up the points. All countries that have had a points system have done so in order to get more immigrants. Also, if you really are worried about immigrants from Europe, who are the two iconic figures of European immigration? The Polish plumber and the French engineer, both of whom would ace the shit out of any conceivable points system.

The appeal to points is interesting in its Michael Young, Rise of the Meritocracy quality. We’re going to get rid of the immigrants by setting them an exam! Because, as Young pointed out, privilege that is expressed by credentials you get by passing an exam is seen as justified, not least by the people who pass. Ironically, as the thing about exams is that you can pass them if you practise a lot, this promises to subvert the whole thing. And of course no generation was ever as trusting in exams as the people who want to leave the EU, who were also precisely the people Michael Young was worrying about. It’s as if the baby boomers want to check out with one final, epic act of credentialism, a giant collective A-level. Perhaps Young’s predictions finally came to pass, just with regard to nationality rather than class?

Anyway, what gets me about the whole rhetorical circus is that the people behind Vote Leave really, really believe at some level that Britain has a racist majority (note that John Mann MP, a big fan of unpopular-populism, has suddenly discovered Euroscepticism), but at the same time, they realise that everyone who has ever tried this has lost, horribly, and there’s probably a reason for that. That’s why they have to include the unlikely promise of lots more Pakistani immigrants, pretend to care about schools whose budget they slashed as education secretary, or outsource their prejudices to an exam paper.

Something has to fill the gap between the two beliefs, and that something is bullshit in the full Harry Frankfurter sense of the word – speech divorced from reality, to which it makes as much sense to say “truth” or “lies” as it does to say “green” or “capacitive”. The prejudice against that still needs work.

June 10, 2016

Yes, Michael Gove is an extremist and has been for years

Martin Kettle is all worked up to learn that Michael Gove is either a cynic or an extremist, now his extremism, or cynical pose of same, affects an issue he cares about.

Where was Kettle when Gove wrote a whole book about the secret rulers of the world’s scheme to hand Europe over to the Arabs? Didn’t he read his own newspaper’s detailed coverage of Gove picking a special adviser who turned out to be a eugenicist? Or its award-winning coverage of him using his wife’s GMail account for confidential government business in order to break the Freedom of Information Act? Did he not think someone who was a News International editor in the phone-hacking, gak-ridden imperial phase might be a bit of a cynical careerist bastard? What about when he brought the eugenics guy back to help run his referendum campaign? Didn’t he notice him arguing that the country needs to be more like Albania?

None of them affected Kettle’s holidays, I guess, so there you go. Personally I said back in 2006 or thereabouts that Michael Gove would take us into the next Iraq War, and it’s telling that David Cameron kept him a long, long way from anything related to foreign policy. Now, though, he thinks he’s found a way to perpetrate a similarly atrocious policy disaster without even being in the Cabinet.

I do hope, though, that the referendum might yet give the commentariat a Gary Larson moment on Vote Leave’s whole tin-pot triumvirate of tat, Gove, IDS, and Boris Johnson.

They’re arseholes! And what have we been licking? Sadly, Larson can also offer us the likely response.