Winner of the New Statesman SPERI Prize in Political Economy 2016

Wednesday 26 June 2013

Government default, reserves and QE

For macroeconomists. Paul Krugman needed more coffee to get his head round the latest paper by Corsetti and Dedola. I had a similar feeling, and I have written this post to try and get my thoughts in order. So this comes with a health warning, which is that I may have failed.

This post is all about the distinction between central banks and governments, but to make those distinctions clear, its helpful to consolidate their budget constraints. Initially assume that this consolidated entity can finance its deficit by either issuing bonds or printing cash. It is therefore true that this entity need never default: if no one buys its bonds, it just issues cash. However that does not mean that it will never default. Suppose, as Corsetti and Dedola do, there is a direct and simple link between cash and inflation, while issuing debt has no impact on inflation. Avoiding default by printing cash therefore creates inflation, which is costly. It may be so costly that the government/central bank entity chooses to default, rather than bear those costs.

I do not think this idea is controversial, but failure to distinguish between ‘need’ and ‘choose’ can cause confusion. What may be controversial in the above is the simple quantity theory link between cash and inflation, but in fact we do not need anything so sharp. As long as printing cash is more likely to raise inflation than issuing debt, the proposition still holds. What the exact nature of the inflation link will influence, of course, is the likelihood the government will choose to default rather than create higher inflation.

This choice may also be influenced by the existence of independent central banks. If governments give control over printing money and inflation to a conservative central banker in the Rogoff sense, then this could decrease the likelihood of using inflation to avoid default, and therefore makes default more likely for a given fiscal position. This may also, of course, influence fiscal policy. Whether independent central banks have that much independence is of course debatable.

Now let’s add the possibility of multiple equilibria. The ‘good’ equilibria is the one where the interest paid on government debt reflects the ‘true’ probability of default i.e. it reflects the circumstances in which the consolidated government chooses to default. The ‘bad’ equilibrium is one where the rate of interest reflects a much higher probability of default. As De Grauwe has recently emphasised, because the market can in effect force default by not buying debt, this bad equilibrium is possible.

However as our consolidated government/central bank can print cash, does this rule out the bad equilibrium? The answer, according to Corsetti and Dedola, is not necessarily. The reason is straightforward. Printing cash is still costly, because it raises inflation. The bad equilibrium could still exist, because to prevent it would require creating an undesirable amount of inflation, which the consolidated government will not at the end of the day do, whatever it may say it will do.

However Corsetti and Dedola note that in practice, central banks are buying government debt not by printing cash, but by creating reserves. So they extend their model to include a third asset, bank reserves. In terms of the consolidated government, what are these reserves? In Corsetti and Dedola I think reserves are default free debt. They pay a risk free interest rate, but there is no chance of default, because (by some means) the central bank always promises to pay back the interest and capital with money. However, like government debt and unlike money, there is no link between the quantity of reserves and inflation.

Now I find it intuitive that with this additional asset, it is now possible to remove the bad equilibrium, as the authors show. The government/central bank can simply choose to swap any normal debt that the market will not buy with reserves, which the market will buy, because reserves are default free. The government/central bank is essentially ruling out its option to default. Why does the government/central bank ever issue bonds?  In the absence of the possibility of a bad equilibrium, it might prefer issuing bonds because it wants to keep the default option.

I have traditionally thought about Quantitative Easing as being equivalent to swapping debt for cash. For exactly the reasons Corsetti and Dedola outline in a world without reserves, to do this permanently would raise the price level, so as a result QE is strictly temporary. But now add reserves of the Corsetti and Dedola type. QE then involves a swap between two types of debt: one default free and one not. The consolidated government/central bank is reducing its option to default on debt, in order either to remove a bad equilibria, change the term structure, increase the supply of completely safe assets, reduce its interest bill, or something else.

Now what is crucial in this analysis is that issuing reserves, unlike issuing cash, has no implications for inflation. You might object that while that clearly seems realistic at the moment, this reflects the peculiar circumstances of the zero lower bound (ZLB), and in the longer term additional reserves would be inflationary because they would allow private banks to create more loans and deposits etc etc. However in this context a recent post from Paul De Grauwe and Yuemei Ji is interesting.

De Grauwe and Ji, among many other things, consider the situation in a which the central bank buys government debt with reserves, and the government defaults. It is then often suggested that the central bank loses the ability to control interest rates and inflation. De Grauwe and Ji argue that this “does not hold water” for two reasons: the central bank can reduce the money stock by either raising reserve requirements, or by issuing interest bearing bonds. So would a permanently higher stock of reserves necessarily imply higher inflation once the ZLB was over? Following De Grauwe and Ji the answer is no, because either private banks can be forced to hold more reserves and not create more deposits, or the central bank could exchange reserves for some other kind of central bank asset that was still default free but which had no knock on implications for the banking sector.

So to the extent that reserves are just default free debt, or can be turned into default free debt, QE does not need to be temporary even with unchanging inflation targets. Whether this is of any consequence I’m not sure, particularly as central banks do not want to make QE permanent, as recent events illustrate. Seeing reserves as default free debt may also not be terribly interesting for the UK or US right now as the perceived default risk of each country’s debt is pretty low anyway, and so the chances of a bad equilibrium emerging are equally low. But the implications for the Eurozone are clearly much more interesting.


  1. There are two "bad equilibriums": one where people fear default, and one where people fear inflation. Ideally a government would like to commit to not default, and at the same time commit not to inflate. But these commitments are at war with each other. An ironclad commitment not to default implies some risk of inflation; an ironclad commitment not to inflate implies some risk of default.

  2. "Seeing reserves as default free debt may also not be terribly interesting for the UK or US right now as the perceived default risk of each country’s debt is pretty low anyway, and so the chances of a bad equilibrium emerging are equally low. But the implications for the Eurozone are clearly much more interesting".

    So what should the ECB do then?

  3. I see the attractions of doing this by increasing reserves, but my concern is that there is no direct linkage between doing so and increasing aggregate demand. Banks don't generally give loans to people for spending money - they give them out for capital investments & major purchases, and I don't see a lot of regular citizens taking out a loan to buy a PS4 any time soon. If it's not putting more money into people's regular spending accounts, it's not going to greatly impact AD, and I thought that was the eventual goal. Or am I misinterpreting?

  4. Inflation and default are not fully comparable imho.
    Inflation can basically also be used for smaller scale corrections, basically what BoE is doing, default not it is always a big event with a lot of repurcussions.

    On the probabilities of the choice. Looks nearly impossible to do. Especially you would have to know what the government will decide. You simply donot know that. One of the big problems for markets now,they have to make an estimate of political decisions or events like elections and are clearly not able/not equipped to do that.
    It is not only choice. It default is usually a road that is gone. With and that is a real complication rather far up that road you get borrowing in other currencies (as the own simply is not accepted anymore in international trade). Foreign currency means simply default.
    So the probability is not only government choice but also lets call it market conditions of a particular moment. If a default can be avoided.
    Usually also inflation is already stiff as well at the latter stages. So even with a default you also have high inflation. Makes estimating the probabilities even harder.

    It looks like that there is a second bad equilibrium. Simply in general chances of default of sov debt are estimated too low. Fully understand that governments like low interest (especially when they are are heavily indebted), but it looks like in practice for 'normal' countries default estimates are generally too low.
    Not sure if this is relevant for the model. As long as it is accepted by markets (including the CB intervention), it will likely work.

    Will be hard to keep inflation and interest rates down if the country defaults. In theory it is possible. But everybody will be selling the bonds both government and private sector. A thing eg the BoJ missed (and that ws on a much smaller scale than a default. It thought that buying 70% of new issues would do the job but all existing bondholders also might leave. You can simply not tackle that. So interest rates in practice imho will always go up.
    And with that the currency is basically to go down (much more difficult to protect than up). Basically foreign sellers wanting USD for their bonds. And with normal countries, with a lot of imports, inflation as well. Will not be 1:1 but still a decent percentage.

    On QE. Inflation doesnot seem to be a problem at the moment. However the present inflation rate estimates by markets could (COULD not is) be considerably more volatile than normal. What I mean that some people still expect high inflation because of future money printing for excess debt. While others see deflation. The extremes are more common now than under normal circumstances.

    QE buys are not only from banks. It is hard to see how you can control from whom you buy. So it will be a mixed bunch. Including private persons and these might use it for anything. Anyway as basically everywhere now shows there isnot much room in increasing reserve requirements and similar stuff for other reasons. If you do that you run directly into another problem (deleveraging).
    Reversing is notoriously complicated, if you want to do it cleanly. Also there the markets might require limitations possibly at a time when for other reasons you might want to do a sell.
    All look simply very difficult to mange processes.

  5. My uni econ prof made it simple: Inflation (from printing money) is default in different clothes.

    While reserves may be default free debt for govt, they are not risk free. If it was, it would be a great way for govts to simply spend as much as they wish and simply park the debt in the central bank. I would challenge the authors to identify what risks (or constraints) exist for reserves and built that into their model.

  6. This seems like an extraordinarily roundabout way of getting to the conclusion that the MMT folks draw much more directly, from a head on critique of the quantity theory and the monetary multiplier--e.g., from a recognition that the broad supply of (credit) money is endogenous, because loans create deposits not v-v, and that excess reserves have no effect on money supply but rather simply force the short safe rate to zero (or to the support rate on those reserves).

    And note: the idea that 'printing cash', in contrast to emitting reserves, can have inflationary effects just seems like a very curious way of saying that fiscal policy, in contrast to monetary policy, can have inflationary effects? Helicopter drops would be fiscal policy--deficit spending (specifically a kind of tax rebate) by another name. Not that inflationary pressure from this channel should be any kind of worry right now, but the point is that monetarist notions are superfluous here. It would then be a straightforward Keynesian story about rising aggregate demand pressing upon capacity.

    Also: I thought it was well-understood in central banking circles that the purpose of QE is to change the term structure of rates by compressing the yield curve--that it is, in effect, a sort of half-hearted attempt to treat long rates as policy rates, without explicitly declaring a rate target. Surely, at least, that is the one effect that can be directly and unambiguously attributed to such asset purchases by the CB? (Whether such an effect is particularly useful as 'stimulus' is of course another question.)

    To theorize any other effects for QE would seem to require hauling out the expectations epicycles, at which point nearly any effect can be attributed to the policy--so long as these 'expectations' are assumed to be able to survive indefinitely, even when divorced from any economic fundamentals.

    Such expectations-based theories about QE's wonder-working abilities always seem to me to explain too much. What could ever falsify them? One can always say the dependent variable didn't budge because expectations shifted, or were insufficiently 'anchored' etc. Well sure, as one would expect when they have no basis in reality!

    1. Yes: I get the impression that Corsetti and Dedola have taken a simple idea (MMT) and made it incomprehensible.


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.