Thursday, February 09, 2012
It's all in the data
Posted by David Smith at 04:30 PM
Category: Thoughts and responses


Readers are probably already bored with the debate between me and David "Danny" Blanchflower, formerly of the Bank of England's monetary policy committee, but this is a final instalment, from me at least.

In his response, here, there is a Hallelujah moment. He has, he says, never said that the slowdown is entirely due to the coalition's strategy. "Fiscal tightening is just part of the explanation," he writes.

Exactly. Tax hikes and spending cuts reduce growth but so does the high-inflation squeeze on real incomes, the eurozone crisis and most importantly, as Rogoff and Reinhart have demonstrated, because recoveries after financial crises are usually weaker and more protacted than normal ones. So I look forward to hearing Danny spreading the blame for the slowdown, though I am not hopeful.

Danny also suggests I have "helped to establish the false narrative that Labour caused the crisis". Not so and, indeed, I have gone out of my way to stress the global nature of the crisis and that it would have been little different had the Tories been in charge.

But Labour was in charge and, as Ed Balls has conceded, has to accept part of the blame for its failure to regulate the financial system. It also has to accept part of the blame for the clumsy initial response to the crisis by the Bank of England and Treasury, which for a while made a difficult situation worse.

There would have been a lurch into large deficit whatever the state of the public finances before the crisis but the absolute size of the deficit was larger because of Labour's failure, under Gordon Brown, to stick to the golden rule. At the time, only people like Martin Weale of the National Institute were brave enough to point this out but now it is generally accepted, as in the recent Office for Budget Responsibility exercise. A pre-crisis deficit of 1% of GDP would have meant a smaller deficit when the crisis hit, and potentially more room for emergency Keynesian measures.

The big difference between us, I think, is not ideological but in use of the data. I examine the numbers, and report them. If there is good news among the bad, I report that. If it is bad, it gets reported as bad. You cannot hide bad news.

Was I being optimistic in reporting the January service-sector and manufacturing purchasing managers surveys as being good news? "UK economy shows renewed vigour in January" said Markit, here, which produces the data. Not my words, theirs.

Is it over-optimistic to say house prices were "resilient" at the end of 2011, when an average of the main house price indices shows them to have been flat over a year in which household real incomes recorded a record fall and unemployment rose? I don't think so.

As for GDP and the prospect of upward revisions in future, he should know that the Office for National Statistics has not begun to scratch the surface yet. Box 2.1 on p26 of this OBR report, here, shows the extent to which the initial data was revised up over many years, as it will be again. This, by the way, makes comparisons between this cycle and the 1930s facile.

As I say, it is important to look at the data. Danny is known as a labour market economist. Last year I criticised him for gloomily highlighting a "dramatic decline" in working hours in the second quarter of last year. Much of it, I pointed out, was due to the extra bank holiday for the royal wedding.

His response was to highlight the "monthly" figures for hours worked in March and May in the Labour Force Survey, which were also weak, and to say:
"Good try David. So how exactly would a bank holiday in April lower hours in March or May? Time to get your facts straight, mate, before coming after yours truly. The coalition is responsible for reductions in the demand for labour, hence the poor spending data."

This was, unfortunately, a second error, which a labour market economist should not make. The "monthly" figures on the ONS website for Labour Force Survey data are three-monthly averages, as the statisticians are happy to confirm. So "March", is the February-April average, "April" is the March-May average and "May" is the April-June average. They were all affected by the extra bank holiday. Boring but true. The facts were straight.