Winner of the New Statesman SPERI Prize in Political Economy 2016

Friday 14 September 2012

Denial, and Bernanke the Brave

Someone teased me about my recent post on Zero Lower Bound denial – can any disagreement on theory or evidence now be labelled by either side as denial? I admit that one of the negative aspects of blogging (for me at least) is that it can lead to an indecent attraction to neat catchphrases and rhetorical flourishes. However there was some method in my madness.

What I take as the characteristic of denial, whether it’s about climate change or macroeconomics, is that a strong position is taken not on the merits of the case, but because of a dislike of some  implication of accepting the proposition being denied. So climate change denial would come not from a consideration of the evidence, but distaste for the idea that it is an externality that requires government intervention to correct. Evidence that this form of denial exists comes from the clear correlation between climate change denial and a pro-market ideology (see Oreskes and Conway, 2010). In macroeconomics, demand denial may also come from dislike of the idea that state intervention (monetary policy) is required to ensure the aggregate economy stays on its efficient path, which is why those that keep suggesting our current problems are not about deficient demand also tend to come from the political right. We could call this bias or wishful thinking rather than denial, but the key point is that evidence on the issue appears to have no impact because the source of the belief lies elsewhere.

In the case of central banks, zero lower bound (ZLB) denial would be a belief that ‘unconventional’ monetary policy is almost as effective or reliable as conventional policy not on the merits of the case, but because the central bank must be seen to be in control. Now no one wants to hear a pilot tell passengers that they are no longer in control of the plane. However a better analogy here would be the pilot not telling the co-pilot, because fiscal policy can also be used to stabilise the economy. In the case of perhaps some economists, ZLB denial may be due to an aversion to the use of fiscal demand stabilisation, and hence the need to talk up the effectiveness of actual or potential monetary policy actions or regimes. But it was the central bank case that I really had in mind: hence my use of the quote from the Woodford paper about central bank wishful thinking.

To take the UK example, I have previously discussed (towards the end of this post) a speech made by George Osborne that stated that monetary policy was all you needed for stabilisation and appeared to ignore the ZLB, at the very moment that UK interest rates hit 0.5%. Now if I had been Governor of the Bank of England at the time, I would have said in private at least that the world had changed, and I was no longer sure I could achieve the Bank’s mandate. Perhaps this was said, and maybe one day we will find out. It was not the stance the Bank took in public.

Of course, my use of the term denial also signifies that I believe the idea that unconventional monetary policy is somehow on a par with fiscal policy in its reliability and effectiveness is fairly weak. I want to make two points here. First, I think it is important to distinguish between targets and instruments. My previous post asserted that nominal GDP targets in a world where unconventional monetary policy was ineffective would only reduce the impact of the ZLB, and not eliminate it. I think this is a useful case to consider because it helps clarify ideas: I still read people saying that nominal GDP targets in themselves would mean that large negative demand shocks would be less likely to take us to the ZLB, and I do not understand why this is.
Second, the key issue for me is not whether QE will or will not have some significant effect. My suspicion is that it can, and Woodford’s baseline case takes too idealised view of financial markets.[1] However our lack of knowledge about the extent of the deviation from Woodford’s idealised view mean that these effects appear of an order more uncertain than the impact of fiscal policy, which is why you have to use the latter, and why austerity during a recession is foolish.

Which brings me to the recent Bernanke decision. If the FOMC were ever in denial, they are no longer. What I think is significant about this decision is that it combines more QE with a suggestion that the Fed may at some point in the future allow inflation above target. I am often asked how we know that QE will be temporary – that central banks will claw back the reserves they have created when the economy recovers – and the easy answer is that they remain committed to their inflation targets. For central banks that have used QE, being seen to be totally committed to inflation targets is their way of showing that they are not permanently monetising government debt. Ironically I wonder whether QE has made central banks less willing to consider changes in the targeting regime, because doing so at the same time as printing lots of money would look too much like the prelude to significant monetisation. In that sense Bernanke and the FOMC have crossed an important psychological barrier, and hopefully the world will be better for it.

[1] See for example Gagnon on markets in general, and Julian Janssen in a direct comment, and David Glasner here in relation to the FOREX market – see also this evidence. On selling currency to engineer depreciation there is of course the obvious point that this would not help a global economy at the ZLB, and as a result would not be received well by other countries.


  1. Simon: "I still read people saying that nominal GDP targets in themselves would mean that large negative demand shocks would be less likely to take us to the ZLB, and I do not understand why this is."

    I do not understand why you do not understand that.

    1. An NGDP *level path* target would have the same sort of "automatic stabiliser" property that a price level path target would have, compared to an inflation target. If a negative demand shock hits, people will expect a temporarily higher than usual inflation rate, which lowers the actual real interest rate for any given nominal interest rate. This is a fairly standard New Keynesian idea

    2. Compared to a price level path target, an NGDP level path target should make that automatic stabiliser property more reliable. Because we do not understand the short run Phillips Curve very well, so we do not know how an increase in NGDP will be divided between an increase in P and and increase in Y. If the SRPC is very flat (or if people expect it to be very flat), the temporary increase in expected inflation following a negative shock under a price level path target might be very small, so the automatic stabiliser force of a price level path target might be very weak. But under those same circumstances of a very flat SRPC, an NGDP level path target would imply a big temporary increase in expected real growth, which would itself increase current demand and have a big automatic stabiliser effect.

    3. If the negative demand shock happens to coincide with an adverse SRAS shock (sometimes stuff happens) the automatic stabiliser forces of a price level path target might be very small, zero, or even negative. (Because the price level might actually rise above the target path when the two shocks hit at the same time.) But an NGDP level path target will still have an automatic stabiliser.

    4. The Wicksellian natural rate of interest is probably positively correlated with the natural real growth rate. An NGDP level path target would lead to a higher inflation rate when the natural rate of growth is low, which will tend to be when the natural rate of interest is also low. A higher inflation rate is like insurance that reduces the risk of hitting the ZLB. So an NGDP target gives you that extra insurance against hitting the ZLB exactly when you need it -- when the real growth rate is low and the natural rate is probably low too.

    Or did I totally misunderstand you there?

    1. Nick. There is a lot here, so lets stick to your point (1).

      I think we agree on how this automatic stabilisation works with price level rather than inflation control. What I find difficult to see is why this would (a) mean that interest rates did not hit the ZLB when they would under inflation targeting (b) imply no role for fiscal intervention. To keep things simple and short, let’s just compare price level and inflation targeting, and assume that there is a predictable, contemporaneous relationship between the (known) output gap and inflation, so we only need to talk about the latter.

      The example in my previous post supposed that the monetary authorities have full knowledge of the demand shock. Take first a demand shock that can be dealt with without hitting the ZLB. Then there is no difference between price or inflation targets: in both cases the bank completely neutralises the demand shock. Now increase the size of the demand shock. At some point the ZLB constraint bites, and that point is independent of the target. When it does bite, the price level target does better than the inflation target, but it does not neutralise the shock, so there is still something for fiscal policy to do. The target makes no difference to when the ZLB hits. Do you agree in this case?

      What I find difficult to then formulate is why a central bank, facing a large demand shock, will ever say to itself with an inflation target I need to cut rates to zero, but because there is a price level target I do not. Because to do the latter admits they will not be doing enough today. To put it another way, price level targeting has an automatic stabilising role when things go wrong, and why would the central bank plan for things to go wrong. Now there may be some case involving dynamics, or maybe partial and asymmetric information, but I have not been able to pin it down.

    2. Simon: OK. I'm with you. If the central bank observes the demand shock in real time, there is no difference between IT and PLPT. The ZLB imposes the same constraint in both cases.

      I'm assuming central banks don't observe the demand shock in real time, because there's a lag in getting the data, for example, so the bank makes (ex post) mistakes and doesn't succeed in keeping inflation exactly on target. So the demand shock hits, the bank doesn't see it until too late, and the inflation rate falls below target. Now the bank observes the shock, and people know the bank observes the shock. Under IT people expect 2% inflation from now on; under PLPT people expect higher than 2% inflation temporarily.

      Now if the demand shock is permanent, and permanently lowers the natural rate below the target rate of inflation, that temporary increase in expected inflation still won't be enough. So yes, I was (implicitly) assuming a demand shock that isn't permanent. The simplest case is where the demand shock lasts for 2 periods, and the central bank observes it with a 1 period lag.

      Or maybe the bank gets the data in real time, but there's a lag between the cut in interest rates and firms responding by increasing investment (some sort of time to start building story, because they need to decide what to build and where and hire the architects and draw up plans before they start digging).

      In other words, central banks should plan on making (ex post) mistakes.

  2. I am sorry to be asking this again, but there had been no response the previous time (from either Simon or anybody else). Is there any technical reason why central bank rates cannot be negative? Why is zero a lower bound?

    1. To make this more concrete, there has already been some discussion on this issue, e.g:

      Buiter, Willem H., 2009. "Negative nominal interest rates: Three ways to overcome the zero lower bound," The North American Journal of Economics and Finance

      While this web-page gives some possible reasons why negative rates may be disruptive, but no technical reasons per se:

      The ECB gives a more interesting explanation of the risks, but I do not find their arguments very clear

  3. Christos: because people would just hold currency and earn 0%, which is better than negative%.

    Simon: I collected my thoughts a bit better here:

  4. Nick: OK, people can and probably would hold currency but the question is banks - as the rate in question is that from the central bank to other banks, not from banks to people. Consider for example a loaning rate of 0% and a deposit rate of -1% for the central banK [1]. Banks can get free loans, but the only useful thing they can do with them is to invest them. If they keep the money, it will not matter if they have taken the loan or not.

    [1] maybe not for the overnight rate for transfers, which may have to be kept fairly close to zero to maintain liquidity.

  5. How can there be growth without the market correcting itself?

    Hopefully Bernake position will be made very difficult after the election and he will resign.

    We tried the Keynes logic for 5 years now and it has failed, time to change scenario. Increase Interest Rates, fight inflation and let the banks/reckless individuals fail. Then growth will return.

    I dont believe that any of the CB will reverse QE. This is a race to bottom and hopefully the elected parties will soon remove the independence of the Central Banks and ask for who is responsible for the change in remit both in Fed/BoE and ECB.

  6. I am not an economist so I can’t comment on Bernanke, but I’d like to comment on denial.

    Denial appears to be a key tool for most economists (and politicians). The problems we face are so complex that I’d suggest that no-one understands them fully, so it is always easy to find some aspect of any argument that has not been considered. Any view can be attacked as denial of some inconvenient truth. The most interesting aspect of this is that everyone can see that everyone else is in denial, but no-one can see it in himself/herself.

    When free-market economists propose austerity to solve our current problems, they are saying that debt is the most important problem, so it is easy to accuse them of denial over job growth. Austerity will lead to more austerity, so unemployment will increase rather than decrease. Keynesians are, rightly in my view, very vocal in accusing the free-marketeers of denial and a belief in ‘confidence fairies’.

    However, as far as I can see, the same thing is true, in reverse, about Keynesians. They say that jobs are the top priority. This means that they can be attacked on denial over debt increases. If part of the problem is that we have too much debt then how can the solution be more debt? Do Keynesians believe in ‘debt repayment fairies’? Prominent Keynesians never seems to answer this effectively.

    For example, do Keynesians have answers to the following questions?

    Should we set any limit to government debt? Any limit at all? Current levels? Ten times current levels? 100 times? 1,000 times?

    If there is no limit then why should we not just run deficits forever? Why do we need to repay government debt if unlimited debt has no consequence?

    On the other hand, if there is a limit, then what policies would Keynesians recommend as we approached that limit? The Keynesian argument is that only more government spending can cure our current problems. However, near the limit, such policies would see us breach the limit, so what would we do? Only austerity would prevent us from breaching the limit. However, if austerity would help in a far worse situation, then why won’t it help now? In this latter case, surely the two sides are arguing only about the size of the limit?

  7. Analysis program on Radio 4 last night discussed use of facts versus bias and denial. It concludes that we are also biased. Very relevant to current debates in economics between incompatible schools of thought - not just left versus right, but also different schools within Keynesian brand.


Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.