Wednesday, March 20, 2013

Osborne unveils Plan D

Plus c’est la même chose, plus ça change…

The more George Osborne insists on sticking to his policy, the worse the results become.

He’s like a driver who refuses to reset his faulty satnav long after it becomes clear that the geography is nothing like what he’d expected. He’s still turning right in 200 metres, but the turn takes him into a ditch instead of the hoped-for motorway.

In 2010, we planned to borrow an extra £471 billion by 2015/16. Today I can announce that we will hit this target two years early.

He didn’t say that, of course. But it’s true.

So imagine what he would say about a Labour Chancellor who had presided over this:

At some point, the government has to take responsibility. Three years is past that point.

(Note on data: The public finance figures are becoming harder to unravel. The ones I’ve used above exclude the effects of various special factors such as the asset price facility and financial transactions. The result of this exclusion is to raise the borrowing and lower the debt in the most recent figures. See tables 4.36 and 4.37 of the OBR report.) 

Monday, March 18, 2013

The press regulation deal is both more and less than it seems

I wonder whether our political leaders know what they are doing.

Their agreement includes a draft Royal Charter establishing the new press regulation system. The Charter states that, once passed, it can only be altered by a two-thirds majority in both the Commons and the Lords.

This provision is being given legal force in an amendment to the Enterprise and Regulatory Reform Bill:

Where a body is established by Royal Charter after 1 March 2013 with functions relating to the carrying on of an industry, no recommendation may be made to Her Majesty in Council to amend the body’s Charter or dissolve the body unless any requirements included in the Charter on the date it is granted for Parliament to approve the amendment or dissolution have been met.

This amendment is deliberately general. As David Cameron, fearful of charges that he had caved in, explained:
This is not by any stretch of the imagination statutory regulation of the press. Nor is it even statutory recognition of either the independent press regulator or indeed the Royal Charter. It is a three line clause which applies to all Royal Charters of a particular nature from this point onwards.  
Colleagues may ask whether this no change clause could be used in future for a more aggressive approach to regulation of the press. Because it doesn’t mention press regulation and it doesn’t even mention this Royal Charter it is no more in danger of being used in this way than any other piece of legislation on the statute book.
But the objection he didn’t anticipate is the opposite one: couldn’t this no-change clause be used to set up other Royal Charters in defiance of the will of Parliament?

It looks like it could. The Privy Council – which is, in effect, the Cabinet – could create a Charter regulating some other industry, and include a no-change rule that required a two-thirds (or three-quarters, or nine-tenths) majority for its amendment or repeal.

But there’s a but.

The Enterprise and Regulatory Reform Bill, when passed, won’t itself be subject to the two-thirds rule. It could be amended by a simple parliamentary majority to remove the no-change clause. So a Cabinet that tried to bypass Parliament could be reined back in.

But there’s another but.

This also applies to the current Royal Charter. The no-change clause can be overturned by a simple majority, and then the new press regulation system can be scrapped in the ordinary way.

This feels like locking your valuables in a strong safe and then putting the key to the safe in a shoebox.

So, I wonder whether our political leaders know what they are doing.

Friday, March 08, 2013

Cameron opposes faster cuts. But why?

The hole in David Cameron’s economy speech is telling. Here it is:

Some say cut more and borrow less, others cut less and borrow more. Go faster. Go slower.  Cut taxes. Put them up. We need to cut through all this and tell people some plain truths. So let me speak frankly and do just that. There are some people who think we don’t have to take all these tough decisions to deal with our debts. They say that our focus on deficit reduction is damaging growth. And what we need to do is to spend more and borrow more. It’s as if they think there’s some magic money tree.

His argument against slower deficit reduction is pretty feeble. But his argument against faster deficit reduction is non-existent. After the brief initial mention of people who take this view, that’s it.

There’s a reason for this. Any argument he makes against faster public-spending cuts is an argument that Labour (or Vince Cable, or the Economist, or anyone else) can quote against his policy.

In exactly this way, the general points in his argument against slower cuts are used against him by people who are urging faster cuts. But clearly, he feels either that those people are politically insignificant or that he doesn’t have the strength to fight on two fronts.

Cameron, Osborne and Clegg have created an arbitrary timetable of cuts and then nailed themselves to it.

For instance, in 2013/14, the government plans to borrow £99 billion. Why is that the correct number? Why would £109bn or £119bn cause a market panic while £99bn is OK? Why would £89bn or £79bn stifle the recovery while £99bn is OK?

As Cable says, “nobody knows how the markets might respond” to a change of plan. “The balance of risks remains a matter of judgement.”

I’ll leave it to others to make the case for faster cuts. But the case for slower cuts is not the insanity he presents it as: “They say that by borrowing more they would miraculously end up borrowing less.”

The case can be summarised simply: more haste, less speed.

If you fire public-sector workers and scrap investment projects and cut benefits too quickly, the economy suffers. Cameron knows this, which is why he opposes faster cuts. By making less haste at cutting the deficit, he hopes to make more speed.

The argument for slower cuts is the same, only more so.

Nobody knows the optimum speed. But Cameron asks us to trust that the plan he worked out with Osborne and Clegg back in the summer of  2010, based on economic forecasts that reality has torn to shreds, managed to get it right.

Wednesday, March 06, 2013

The inevitability of bad predictions

Yesterday the Guardian published a piece of psephology by John Ross. He gets one thing right but almost everything else wrong.

I’ll start positive. The thing that Ross gets right is his main point: Conservative support has been in decline for a long time. He says that since 1931, the Conservative share of the vote has dropped by an average of 0.2% a year.

I agree. I’ve only looked back to 1945 (covering 18 general elections rather than Ross’s 20), but I also get an average 0.2% decline a year – with, of course, a lot of variation around this general trend.

But what Ross doesn’t mention is that the Labour vote has also declined, by an average of 0.2% a year (since 1945). Conversely, the Liberal/Lib Dem vote has risen over this period by an average 0.3% a year. You can see the rough picture from this chart:

These numbers do, though, depend on your starting-point. As Hopi Sen points out, 1931 was a stunning Conservative landslide; 1945 was a Labour one. If you start at 1974, after the first Liberal surge, the Labour trend is, on average, flat and the Lib rise is under 0.1% a year. If you start at 1983, the Libs are in slight decline. If you start at 1997, the Conservatives are on the up.

But for the sake of argument, let’s stick with the longer-term picture.

Things really go wrong when Ross looks to the future. This paragraph contains one of the highest concentrations of wrongness I’ve ever seen:

Taking these projections, if the Tories won the next election, they would get 34.6% of the vote, and if they lost they would get 30.3% of the vote. As there is no doubt at present that the Tories will lose, they will get 30.3% of the vote. As always there is a bit of statistical noise in any calculation, so 29.3% to 31.3% would be a reasonable range, but 30.3% is the central figure.

What he seems to be doing is separating elections that the Conservatives have won from ones they have lost, and then extrapolating the trends for both categories.

This is a logical flub. You don’t first ask whether the Conservatives will win and then go on to wonder what vote they’ll get. You first ask what vote they (and other parties) will get, and then use that to see who’ll win. Votes determine victories, not the other way round.

So, having established that there are only two possible Conservative vote shares in 2015, he then says “there is no doubt at present that the Tories will lose”. Might there be doubt in the future? I guess it’s doubtful. But the interesting thing here is that Ross already has the election result predicted without even using his system. Presumably he’s looking at opinion polls like the rest of us. So what’s the use of the system?

Then there’s the “bit of statistical noise”: he reckons his system’s predictions have a margin of error of plus or minus 1%. That’s a lot better than the 3% that a normal-sized opinion poll has. I wonder how he arrives at this number?

He doesn’t say, but if I wanted to arrive at such a number, I’d start with this chart:

The dots are actual vote shares and the lines are the overall average trend. The distances between the dots and the lines show how close the model has been in the past. You’ll notice that most of the dots are more than 1% away from the relevant lines.

In fact, the median error is 3.8% for the Conservative vote, 2.1% for the Labour vote and 3.6% for the Liberal/Lib Dem vote: half the time, the model was wrong by more than these amounts.

But bear in mind that this model is based on a very small sample of data: 18 election results since 1945 (or 20 for Ross since 1931; adding the extra two really won’t change the picture). So the confidence intervals of any conclusions we draw from it may well be large. And they are. The standard deviation of the error in the predicted Conservative vote is 4.1%, in the Labour vote 4.6% and in the Lib Dem vote 4.3%.

Assuming a normal distribution, two-thirds of observed results would be expected to fall within one standard deviation of the central result. For a typical 95% confidence interval, we need to go plus or minus two standard deviations: 8.4% for the Conservative vote, 9.2% for Labour, 8.6% for the Lib Dems.

So, my version of Ross’s model gives these central projections for 2015: Conservatives 34.3%, Labour 31.5%, Lib Dems 26.4%. But all I’d be confident in saying is that the Conservatives will get between 25.9% and 42.7%, Labour between 22.3% and 40.7%, and the Lib Dems between 17.8% and 35%.

Probably. Assuming that there’s a genuine phenomenon here that will continue in the future. And ignoring all polling evidence.

In conclusion: Yes, long-term trends are noteworthy. But let’s not read too much into them.