Policy beyond forecasts

The OBR has published a long report on why it got its GDP forecasts wrong. This is an exercise in missing the point.

We don't need a 99 page document to tell us why the forecasts went wrong. They went wrong because that's what forecasts do. The future is inherently unknowable, so forecasts will very often be wrong.

Getting a forecast wrong is no crime. What is deplorable is basing a policy upon something as unreliable as a forecast. Blaming economic forecasters often serves a political function. It deflects blame for bad policy away from those in power - be they politicians or bosses - and onto economists.

The question we should be asking is: can we base macroeconomic policy upon something more robust than a forecast?

In monetary policy, the answer could be yes: one purpose of the Taylor principle is to take forecasting out of policy. Granted, a forecast is implicit - inflation and the output gap matter because they are thought to predict future inflation - but the principle doesn't waste time on formal forecasting.

The analogue in fiscal policy is the automatic stabilizer; as GDP falls, so taxes fall and welfare spending rises, helping to moderate the recession. Such stabilizers, however, aren't sufficient to fully describe how fiscal policy should be set. Hence the debate between Keynesians (or Kaleckians!) and Austerians.  Insofar as this debate rests upon very fallible economic forecasts, it is, however, unsatisfactory.

Hence my question: could fiscal policy become more automatic, more state-dependent - as the Taylor principle envisages monetary policy - and less forecast-dependent?

In one sense no. There are awkward administrative issues in having state-dependent increases or decreases in infrastructure spending. The rule "build a new airport iff the output gap exceeds x%" is probably impractical. There are, though, three other possibilities:

1.Have a bigger government sector. As Dani Rodrik pointed out years ago, open economies tend to have bigger governments. This is because such economies are more vulnerable to unforeseeable risk, and having a bigger state sector is a way of offsetting those risks.

2. Have stronger automatic stabilizers - for example, higher marginal tax rates or higher benefit levels.

3. Have a system of job guarantees, in which the state (or local government) acts as an employer of last resort.

Now, my rightist readers will point out that there are drawbacks to all three of these solutions. They are right - though how much so I'm not sure. This is because there's a fundamental trade-off; policies that reduce risk tend also to blunt incentives.

Without such policies, however, there will be pressures on governments to use discretionary policy. But if this relies upon a sill "predict and control" mentality, it too has large costs.

Another thing: There is, if you'll allow the phrase, a third way here. Having active markets in GDP securities and other macro risks would allow those exposed to recession risk to buy insurance against it.This would reduce the need for governments to try (unsuccessfully) to prevent such events. However, I fear that Robert Shiller and I are the only two guys on the planet who think this might be a good idea.  

Comments

[miltonkeynes]

Aren't market monetarists with you on markets in GDP securities?

[From Arse To Elbow]

"Getting a forecast wrong is no crime". Michael Fish would no doubt agree.

[Nick Rowe]

miltonkeynes beat me to it. Add Scott Sumner to your list. But not just because it allows people to insure themselves against NGDP risk, but because it lets the monetary authority observe market NGDP expectations in real time. And stabilising NGDP expectations would be an important automatic stabiliser. Much better than an inherently backward-looking Taylor Rule, which only responds to two variables, and with an unavoidable lag.

[Nick Rowe]

Plus, the existence of a very small stock of NGDP perpetuities (paying an annual dividend of one trillionth of NGDP forever) would prevent "dynamic inefficiency" (where the nominal interest rate is forever less than the NGDP growth rate) because that would make the market price of a trill perpetuity infinite.

[Chris Purnell]

"I fear that Robert Shiller and I...."

And he ought to get a Nobel Prize, so you're in very good company.

[shtove]

Shiller: decent man with an unfortunate name.

Didn't he plug an ETF based on the US housing market, so that buyers could hedge purchase price risk?

[Frances Coppola]

Chris Purnell got there first. Shiller - evident Nobel material. And it looks like you can add Nick Rowe to the list of people who approve of GDP securities. You are definitely in good company. All you need now is someone to make a market in the things.

That OBR report is the closest I've ever seen a Government department get to admitting "Oops, we got it wrong".

[jobs in banks]

Thanks a lot for those tips!

[Ralph Musgrave]

I like Chris’s “flexible fiscal policy” idea, but I don’t agree with his claim that because infrastructure projects cannot be turned on and off at will, that therefore a more flexible fiscal policy is not possible. It would be perfectly feasible to tell schools, hospitals, the police, etc that the size of their budgets is liable to be changed by a small amount at a moment’s notice by central government. In a school for example, hiring or sacking a supply teacher is no big deal. And for the police, buying a new car or delaying its purchase is no big deal.

Second, I don’t see why (as Chris suggests) we need a BIGGER government sector for the above to work. The idea works with a big OR SMALL public sector.

[Zorblog]

What about automatic working-time reduction, partially compensated by unemployment benefits.
I believe that's more or less what they have in Germany, and one of the reason this country was not severely by the crisis (apart from the euro distortions, of course).
A big problem with recessions is the economic uncertainty they create because people fear losing their jobs (a notion that is absent from main economic models by the way). If working-time reduction is shared, the problem does not exist any longer. It's a micro-mechanism with a macro stabilizing effect.

[Nick Rowe]

Ralph: as someone who has worked doing university budgets, I can say that hiring one more or less supply teacher is a very big deal, *if you have to do it quickly*. Because you need to tell the students months in advance which courses will be offered. And you can't suddenly cancel the course halfway through the year. If our budget went up or down at a moment's notice, we would simply offset that by saving increases and dissaving decreases, which means it wouldn't work at all as a quick fiscal policy change. The government would give us more money, and we would just sit on it for a year. If the government gave us less money, we wouldn't cut actual spending until next year.

[Frances Coppola]

Ralph: what Nick said. Education budgets really can't have that degree of uncertainty. In fact I don't think any budget can be subject to sudden alteration without notice. It would be impossible to plan or manage.

[Frances Coppola]

Chris, I've now read the OBR report. I agree with you up to a point - they do spend far too much time explaining small forecast/actual differences that really aren't significant. But the growth forecast is very wrong and I do think they had to explain that. Not that they were able to satisfactorily, of course, without following the IMF into a "mea culpa" about multipliers. Their argument that it is due to persistently high inflation and weakening exports looks a bit thin to me.

[Ralph Musgrave]

Nick Rowe makes a good point: I overstated my case and chose a bad example (teachers) to illustrate it. Obviously money dished out with a view to getting fiscal stimulus will not necessarily be spent.

That problem actually occurred on a catastrophic scale in the U.S. in that most of the stimulus money channelled to states was hoarded – at least according to this source:

http://johnbtaylorsblog.blogspot.co.uk/2012/03/in-blog-post-yesterday-paul-krugman.html

On the other hand ALL FORMS OF STIMULUS (monetary and fiscal) involve much longer lags than we would like. So to re-state my point in a less silly way: “The fact that infrastructure projects cannot be turned on and off at will is not a reason to throw out the baby with the bathwater: i.e. avoid all forms of fiscal stimulus. Reason is that while the effects of non-infrastructure fiscal stimulus are uncertain, they can be turned on and off more quickly than infrastructure projects.”

Re teachers (appropriately qualified teachers, that is) they are a bad example in that it’s often not easy to get people with ideal qualifications in a hurry. So would it be possible to engineer an increase in employment using less well qualified people who would do relatively peripheral or unimportant tasks? I wrote a paper claiming that IS POSSIBLE. The paper actually gives theoretical backing to the British government’s “Work Programme” though I wrote the paper long before the Work Programme was implemented. See:

http://mpra.ub.uni-muenchen.de/19094/

[Frances Coppola]

No, Ralph, I don't agree. Creating jobs that can disappear when the political wind changes isn't increasing employment for the longer-term, which is what we really need.

People doing "peripheral" jobs should no more be subject to sudden changes that result in them losing their jobs without notice than people doing jobs that are considered "more important". They deserve better treatment than that.

The public sector divides itself into a protected "core" of largely well-qualified, highly articulate and well-paid people with secure jobs and - importantly - the support of powerful unions, and a growing "shanty town" of people doing part-time, casual and insecure work, often for very low pay. Your proposal would not touch the "core", which protects itself very well - but it would make the lives of the "shanty town" workers even more difficult. And the shanty town workers are not necessarily any less qualified than the core workers. They are just more vulnerable.