A thought-provoking post from Toby Nangle on this topic raises some difficult questions that go to the heart of monetary economics and policy, and take in important practical questions like what are official statistics for.
In his post, Toby recaps on the arguments of those who have become known as ‘inflation-truthers’. These are commentators, who, reasoning from versions of the quantity theory of money, [pq=mv], predicted that money expansions associated with quantitative easing might generate runaway inflation. When they didn’t see this inflation, they speculated that it was there, just that the authorities had conspired to manipulate public statistics so that it could not be seen.
Toby doesn’t try to rescue this conspiracy theory. It’s hard to disprove it, although you will find that almost every practising economist who has ever interacted with government or the ONS will dismiss it as absurd. Unfortunately, the experience of Argentina, where this does seem to be happening, gives this dark view life it does not deserve.
Instead what Toby does is to get at something that did happen, and might be the source of what is itching inflation truthers. And that is that quantiative easing lowered yields and thereby raised the cost of providing for a given lifestyle in the future by saving. In so far as it raised this cost, the argument is put that this is a kind of inflation, so perhaps this is the inflation that the inflation truthers were thinking would happen all along as a result of the money expansions. And, extrapolating the logic, perhaps, if it is a kind of cost, we should include it in a proper definition of inflation. Such a suggestion harks back to debates at the end of the 1990s and early 2000s when inflation targeters’ inflation was low and on target, but asset prices were booming. Proponents then wanted tighter monetary policy to curb the boom they thought would lead to a crash (and which eventually did, of course) and some thought that the way to do that was to redefine inflation so that it included the asset price inflation [short hop from implicit yield] that would make the target index overshoot.
The more I think about this, the more I conclude that there are not any definitive answers to the questions raised here. There are some points we can make about monetary theory, which the inflation truthers get wrong, but the theory is just that, and all the more a work in progress since we had to think about QE. And there are some practical and historical points to be made about inflation indices, but about which different people could reach different conclusions.
The first question raised is whether the fact that QE lowered yields proves the inflation truthers right all along about the fact that MV=PT would assert itself somehow. From theory, the answer is no. PT=MV is not so much a theory as a piece of book-keeping. Standard monetary theory that accounts for the existence of an effect of QE on yields tells us that the monetary expansion bit was of no consequence, except in so far as it might have signalled something about lower future interest rates. The bit that lowered yields was the bit that involved a twist of the maturity structure of government assets out there, replacing long dated securities with short-dated equivalents. The thought experiment of two ‘bits’ of QE is just that, but recognise that buying long dated securities with reserves – what the Bank of England actually did – is the same as doing a conventional open market operation, (reserves for bills), and then a twist (bills for long dated gilts). MPC emphasised the importance of the expansion of money in the early days of QE. And some of their educational literature still does. But this was just hopeful bluster, at least as far as the theory we have tells us.
Empirical evidence on the effectiveness of QE doesn’t refute this basic notion. Though it shows that money expansions lowered yields, it also shows that twists lowered them. And the fact that money expansions did lower them might have nothing to do with the ‘money leg’ of the transaction.
Note that the inflation-truthers’ contention, that PT=MV asserts itself somehow, wasn’t anyway true at the zero bound. As the economy approaches the zero bound, velocity [jargon for how much real balances people want] falls [real balance demand rises] which is not surprising, since economics 101 tells us that as the price of something falls we want more of it [the price being the nominal interest rate in this case].
The second question raised is whether the fall in yields constitutes a cost for those trying to provide for their savings. This is certainly true, other things equal. Lowering yields means that yields on savings are lower! Implying that you need more of them to leave you with the same sized pot as you had thought. But, as the BoE explained pretty well in its evidence to the House of Lords, other things are not equal. Absent QE and the lowering of yields, it’s plausible that real activity and the value of funds invested in private assets, would have been much lower. So the reduction in the yield does not tell the whole story for savers.
Even if we had established that the ‘cost of saving’ had risen, this would not mean that QE was a bad idea either. If it had the effect of redistributing funds from savers to borrowers, then one might have anticipated that this would raise aggregate demand, since the latter we think have a higher propensity to consume out of income than the former.
This then leads us to the question of whether this ‘cost’, if it was a cost, should be included in the inflation index. This is impossible to answer definitively. In a free country, anyone can construct whatever weighted averages of things they want! So whether adding other things to what one conventionally thinks of as inflation is worthwhile depends on why we are doing it.
Conventional practice is to define inflation to mean the change in the amount of money needed to buy the same basket of consumption goods: or, if this basket changes, and, to dig down to the theoretical fundamentals, to generate the same amount of utility that such goods provide.
This definition has certain theoretical purposes. For example, in macroeconomic models with flexible prices, it is this definition that provides the instantaneous link between the level of the money supply [requiring its own definition!] and the level of prices, referred to by the P in PT=MV. And it is the definition which allows us to connect the rate of growth of money and the rate of growth of prices. Using this definition, we can also develop reasoning about long run optimal monetary policy [the optimal rate of growth of money, or level of nominal interest rates]. In this flexible price world, that policy is to induce negative inflation equal to the real rate, so that the return on holding money is the same as holding other risk free assets. Thus maximising the social benefits to holding money. You could still figure out optimal monetary policy if you added to this definition a weighted sum of the real rate embedded in a long security and Brazilian rainfall, but it would make life slightly harder. A more economical way to put this is that we define inflation to be the thing that monetary policy can choose, and the thing that can only be determined by monetary policy, in the long run.
In models of macroeconomies with sticky prices – like those used by all the major central banks – the same concept of inflation produces a refined steer about the appropriate long run monetary policy, and also a guide to short run, cyclical monetary policy. The refined steer about long-term average levels of interest rates is that they need to be a bit higher than in worlds without sticky prices. The same force in flexible price models works to pull optimal policy to try to equalise the return on monetary and other assets by generating deflation. But this force is offset somewhat by the fact that changing prices invalidates the price choices of those who can’t readjust, so the optimal rate of inflation becomes a weighted sum of the Friedman Rule and zero. And thus average nominal interest rates are higher. The steer from these models about short run monetary policy is that we should try to keep inflation – conventionally defined – as close as possible to this long run level, traded off against other short run objectives too, like controlling real activity. Once again, there could be no particular theoretical objection to defining ‘inflation’ to include Brazilian rainfall. It wouldn’t affect the setting of interest rates in this model world. But it would make communication of policy tricky in the real one.
That said, if you wanted to record the cost of those trying to provide for a certain level of utility in the future in terms of resources set aside today, you could adjust the inflation rate for a change in real returns. But those returns are not something that conventional monetary policy can do much about in the long run [except via reducing nominal uncertainty in the economy]. This ‘cost’ is something that the ‘fiscal’ bit of monetary policy [the twist bit of QE] should pay attention to. But, weighed against the other effects of QE, it might well still prove to be the right thing to do.