Did the Bank of England cause Brexit?

Suppose that by the mid-2000s, immigration from the EU (and the potential for additional immigration) had led to an important shift in the UK labour market. The possibility of bringing labour from overseas meant that old relationships between the tightness of labour market and wage increases no longer held.

You might think that was bad for workers, but that is not so. It would mean what economists call the natural rate of unemployment (or NAIRU) has fallen. Unemployment can be lower without leading to wage increases that threaten the inflation target, because workers fear that the employer can resort to finding much cheaper overseas labour. It reduces the power of workers in the labour market, but also leads to overall benefits. (This is just an example of the standard result that reducing monopoly power is socially beneficial.)

But it is only good news if the Bank of England recognises the change. If they do not, we get stagnant wage growth and unemployment higher than it need be. The obvious response is that the Bank will know there has been a change because wages will start falling faster than they would expect based on previous relationships. However that effect may be masked by the well documented employee and employer reluctance to actually cut nominal wages. Add in the shock of the financial crisis, and this change in the way the labour market works might well be missed.

Here is the big leap. Suppose the above had happened, and the Bank of England did not miss the change. Monetary policy would have been much more expansionary, bringing unemployment well below the 5% mark. Nominal wage growth would have been stronger, and a buoyant labour market would have generated a feel good factor among workers. With more vacancies and less unemployment, concerns about immigration would have begun to fade. The Brexit vote would still have been close, but would have gone the other way.

You may say how could monetary policy be more expansionary given how close we are to the Zero Lower Bound? If that was the case the Bank should have said they were out of ammunition, and placed responsibility with the government and austerity. But for the last two years at least, the Bank could have cut interest rates and has not. You could blame the relentless expectation in the media and financial sector that rates would increase, but the Bank should be able to rise above that.

Of course the Brexit blame game is easy to play when the vote was so tight. The most speculative aspect of this chain of thought is the initial premise about a shift in the NAIRU created by immigration potential. While the possibility makes sense, whether the data backs it up is much less clear. Yet there is some evidence of a structural shift in the UK labour market in the mid-2000s, as Paul Gregg and Steve Machin report.    

Is zero the new normal?

Larry Summers talk on secular stagnation has led to a burst of discussion on whether nominal interest rates may be at their Zero Lower Bound (ZLB) for longer than we might have thought. I would like to use this post to clarify a number of different ideas that may be involved here. Crucial to this discussion is the concept of the natural real interest rate (NRR). I will define this as the real interest rate that keeps inflation constant. (Sometimes economists define the NRR as the real interest rate that would occur if all prices were flexible, but I think that can be misleading when we are at the ZLB.) There are important issues about risk and different real rates that Tyler Cowen mentions which I will have to ignore.

Crucial to Summer’s argument is that our problems did not all start with the recession. However it may be worth just noting some arguments that the ZLB may be around for some time that do begin with the recession.

1) It takes a long time to adjust balance sheets

One way many economists think about the current recession is that it involves balance sheet adjustment: consumers and firms need to save to reduce their borrowing or increase their wealth. Ideally we would try and offset this by encouraging them to make this adjustment more slowly, or encouraging others to offset this, through negative real interest rates. If adjusting balance sheets takes a decade rather than five years, this may mean that the NRR is negative for much of this period, so we will be stuck at the ZLB for some time to come.

2) Fiscal policy

Tightening fiscal policy lowers the NRR. One of the unusual features of this recession relative to earlier downturns is fiscal austerity, and this will reduce the NRR.

3) Financial intermediation

There has been a lot of discussion about how recessions that result from a financial crisis may be longer lasting. In some countries there is a concern that banks are still carrying a large amount of bad loans, and that this may inhibit their lending for some time. Others worry that tighter regulation could have the same impact, although this is disputed. Just as a reduction in credit rationing can lead to a prolonged economic boom, an increase in rationing can have the opposite effect. There are also more complicated arguments involving a shortage of safe assets (like government debt) that can be used as collateral.

 4) Pessimistic expectations

The way the global economy eliminates deficient demand is not through price flexibility per se, but through movements in real interest rates. If the ZLB means that output is below the natural rate for some time, this could lead consumers and firms to revise down their estimate of what long run output will be. (We can see this happening already when some argue that productivity has permanently fallen as a result of the recession.) If these expectations are more pessimistic than those of policymakers, interest rates might have to be at the ZLB until these expectations are revised. (I discuss some related ideas here.)

All these stories generally start with the recession. However in the decade before the recession, the real interest rate associated with stable inflation appeared to be much lower than before: this is Bernanke’s savings glut. It seems clear that we have to take a global perspective on this, otherwise we end up chasing contradictory red herrings: worrying about the lack of investment in the US, and also worrying about overinvestment in China. The importance of thinking globally is emphasised in Daniel Alpert’s book, and by many others.

One straightforward possibility is that the long run equilibrium NRR has fallen. In standard macroeconomic models, this rate is usually related to impatience, population growth, and technical progress. We know that population growth has shown substantial declines in the developed world, and will show similar declines in the developing world, so that world population may eventually stabilise in around 100 years. Some have also argued that the rate of technical progress has already, or is about to, decline. So there are good reasons for believing that the long run equilibrium NRR has fallen. It might be possible to construct demographic arguments for the medium term NRR to be below the long run NRR, but I will leave that to those who know more about savings behaviour in China and its neighbours.

Another, but rather different, popular argument relates to inequality. It is often put very simply: as the rich have a lower propensity to consume out of income, shifts in the distribution of income towards the rich will tend to reduce aggregate consumption. For a more sophisticated discussion, see Interfluidity.

Often solutions to problems depend on a good diagnosis of why these problems have arisen, but I can think of two solutions that appear to be robust to any of the stories outlined above. The first, discussed by Ryan Avent and made fairly explicitly in recent remarks by Blanchard, is to raise the inflation rate. The second, which a great many economists would sign up to even if they swear they are not Keynesian, is a sustained increase in public investment. The advantage of the latter over the former is that the former has clear costs, whereas the latter could have clear benefits. Of course we may need to do both.

Let me finish with a point about government debt. A major reason why high government debt is a problem in the medium to long term is that - unless Ricardian Equivalence holds - it crowds out private capital. It does that by raising the NRR. Too much saving goes into buying government debt, so there is not enough to invest in private capital. Yet if the NRR is actually negative, and likely to stay very low for some time, and this is a problem because of the ZLB, then the fact that government debt is currently raising the NRR is useful. To put it another way, this means we have plenty of time to deal with the problem of government debt. Which is good, because all the analysis suggests that it is optimal to reduce debt slowly. In the short term, high public debt is helping, not hurting. (Similar arguments can be made in relation to unfunded social security schemes.) 

How not to run fiscal policy: more lessons from the Eurozone

In my last but one post I noted the case of European Commission estimates of the output gap as an example of what can happen if you do not allow for the asymmetry implied by firms’ reluctance to cut nominal wages. Their methodology implied that the natural rate in Spain had more than doubled in a decade, which seems nonsensical. However the economists at the Commission at least came to recognise the problem, and had proposed making some changes to get more reasonable numbers. As I will explain, the methodology they use is reasonable given the state of macroeconomic arts, which is why in that earlier post I used this as an example of a failure of macroeconomics generally, rather than economists at the Commission. In this post I want to note what happened next, which in contrast does seem to reflect badly on how the Commission works.

But before getting to that, a few background points. First why this matters. As part of the “two pack” (don’t ask), Eurozone economies will now have to submit their budgets for approval by the Commission. Approval will depend, among other things, on the Commissions calculations of the structural budget deficit, which is the deficit corrected for the cycle. Measuring this is difficult, because we do not observe the output gap, which itself depends on both the natural rate of unemployment and the underlying trend in productivity (technical progress), which we also do not observe.

You can say at this point why bother - just stick to looking at the actual deficit. That’s an overreaction. For most Eurozone countries we are pretty clear about the sign of the output gap, so making some adjustment should be better than doing nothing. To see the kind of stupidity that arises from just focusing on actual deficits, see the Netherlands.

In principle we can use information on what we do observe, like wage inflation, to make inferences about what the natural rate of unemployment is. So if wage inflation depends on the gap between actual unemployment and the natural rate (called the NAWRU by the Commission), we can switch things around to make this an equation telling us what the natural rate is, given observations on actual wage inflation.

There are three kinds of problem that arise in doing this. The first is that our estimates will be only be as good as the specification of the wage equation. If we leave important factors out of the specification of the wage equation (like a reluctance to cut the nominal wages of existing workers because of morale effects - see this paper by Eliaz and Spiegler for example), we will get our estimates of the natural rate wrong. The second is that some of the things that we are sure do determine wage inflation, like inflation expectations, may be difficult to measure. Paul Krugman discusses the possibility that inflation expectations in Spain might have become anchored here. Finally no equation is perfect, and if we do not allow for these inevitable errors we will get a ridiculously bumpy series for the natural rate. So we need to apply some kind of smoothing.

The way the Commission tackled these problems is described in detail here. The fact that they use a Kalman filter to deal with smoothing issues seems sensible to me: I’m quite fond of the Kalman filter, ever since I wrote a paper with Andrew Harvey, Brian Henry and Simon Peters that used it to estimate labour productivity back in 1986. [1] I suspect they are getting the implausible results for one of the other two reasons. But the key thing to take away is that there are no easy answers here, and this kind of problem requires quite specific macroeconomic expertise. Furthermore, the importance of particular issues may vary between countries, so country specific expertise should be helpful.

Following on from this last point, it is clear that this is not just an issue for Spain. As the Irish Fiscal Advisory Council notes here (p68), the Commission estimates for the natural rate in Ireland also look implausibly high. So it is not surprising that the Commission would want to adjust their methodology to give more plausible numbers. (Matthew Dalton looks at the implications of their methodology for the US here.)

This all matters. If the Commission underestimates the output gap because it overestimates the natural rate of unemployment, then it will overestimate the structural budget deficit, and the country concerned will come under considerable pressure to undertake further austerity. Readers will know why I think that will be a very costly outcome. (More on this from Cohen-Setton and Valla here.)

At least economists at the Commission have now recognised the problem, and changed their estimates. But as Matthew Dalton reports

“The change was approved by technical experts at a meeting last week and was expected to be supported at a Tuesday meeting of more senior officials in Brussels. But an article published in The Wall Street Journal about last week's decision generated concern in some national capitals about its effects on budget policies, an EU official said. The new methodology will be sent back to the expert committee for further discussions, in an effort to understand what its impact will be on all 28 EU countries, the official said.”

Reassuringly, Dalton adds that "The commission is fully on board with the new methodology," the official said. "We believe it is superior." But he also notes that the new methodology “was supposed to have been used by the commission in its next round of estimates of the structural deficit, to be published in November. Now that will have to wait, if it is approved at all.” As one of my commentators pointed out, this article in a leading German newspaper may have contributed to this official hesitation.

So the Commission will go on making estimates that it knows are overestimating structural budget deficits, because of ‘concern in some capitals’ about the implications of using better estimates! None of this does the Commission any good in terms of its competence to help determine national fiscal policy. (Of course nothing can top the incompetence recently shown by the US congress, but that is no excuse.) 

Luckily there is an obviously better way to proceed, even within the confines of the deeply flawed Fiscal Compact. Many Eurozone countries already have their own ‘fiscal councils’: independent bodies set up to provide scrutiny of national fiscal policy. It should be central to the mission of these bodies to estimate the output gap and structural deficit, as it is impossible to look at fiscal sustainability without doing so. So why not get estimates of the output gap from these institutions who will be able to take into account country specific factors, and use the academic expertise that exists in those countries to maximum effect. (Spain and Ireland are hardly short of good macroeconomists.) Unlike the governments of those countries, fiscal councils should not be prone to bias in producing these estimates. The Commission can play a coordinating role, getting experts from the national fiscal councils together to share ideas and expertise. This seems to me a clearly better way to proceed, unless of course your goal is to maximise the influence of the Commission.

[1] Harvey, A., Henry, S.G.B., Peters, S. and Wren-Lewis, S. (1986), Stochastic trends in dynamic regression models: an application to the employment output relationship, Economic Journal, vol 96 pp 975-985.