Over at the ASI blog, Tom Clougherty ask a curious question: "What is inflation?" Tom's concern is whether inflation is "really" a matter of rising consumer prices, or of a rising money supply. Others ask what the "real" inflation rate is, talking about how particular prices are rising much faster than official inflation rates suggest.
Well, to "inflate" something is to blow air into it causing it to expand (from the Latin "flare", to blow, since you asked). So, red-cheeked and exhausted, many of us have inflated balloons for children's birthday parties. More generally, we metaphorically see things as "inflated" when they expand – especially when they expand more than they ought to. So, for example, I might have an inflated sense of my own importance.
Thus there is no universal answer to the question "What is real inflation?" All we can respond is: "Well, what sort of inflation are you interested in – the inflation of my balloon, my ego, a particular index of the money supply, the true underlying money supply, a policy index for consumer prices, the true underlying average cost of living?"
The useful way to unpack Tom's question is presumably something like "What form of inflation ought we most to care about?" His claim is that we ought to care most about changes in the money stock, because the overall consumer price level might change for "real" reasons of the balance of supply and demand. Now on the face of it that's a bit of an odd claim. After all, if the price level for consumption goods rises but the price level of the assets I own is unchanged and/or the total value of my cash holdings is unchanged, then I have become poorer in the sense that I cannot use my assets and cash to purchase as much consumption as before. Why would it be a mistake for me to care about that?
Presumably Tom's claim would be (quite correctly) that sometimes the efficient thing is for those that hold assets and cash to become poorer relative to those that trade consumption goods and have labour/machines to produce further such consumption goods. Crudely put, changes in the overall price level are a mechanism whereby the economy changes the relative wealth of past savers and future workers. But just because that might be efficient doesn't mean I shouldn't care about it!
Tom's proposal is that what we ought to care about, instead, is the money supply. If we were to do that, it would not be because the money supply has any intrinsic value of its own. Rather, it would be because controlling the money supply would be a useful policy instrument allowing us to produce the efficient overall result, particularly (for the purposes of this discussion) in respect of the efficient level of change in the level of consumer prices. But of course we cannot directly observe the money supply any more than we can observe consumer prices. All we can do is to construct surveys that enable us to estimate these things – no, worse than that: all we can actually do is to construct surveys that enable us to estimate proxies for these things (a "proxy" stands in the place of something else, functioning enough like it to be useful for our purpose).
Then, of course, there is the issue that apart from not being able to measure either the money supply or the price level, we cannot control either of these things directly. What we aim to do is control other things – such as the central bank interest rate, and/or the monetary base (which we can control directly if we choose to do so), and/or the capital requirements of banks, and/or the level of government deficits, and/or public sector pay, and/or the prices of particular goods – and so on from a spectrum of policy instruments over which we have fairly direct control and might usefully affect what we are interested in to instruments that, although policy could control them, it ought not to try to do so.
So now things are tricky. We can't measure either the money supply or the price level precisely, and if we could measure either of them we could not control them directly. All we can actually do is to measure proxies and then take them into account in some way in setting policy.
You might think: "OK. Well, we can't control them directly, but we can set targets for them. So the key question is whether we should set targets for the money supply or the price level." But of course we don't set targets for the "price level" if by that you mean a price level that measures the overall cost of living. No-one sensible does that. To try to do so would create all kinds of problems, because many of the instruments we use to try to implement policy would themselves change the price level. The best known effect (though by no means the only one) is that when interest rates rise, which typically happens when one is trying to get price inflation down, the initial effect is that price inflation rises because interest rate rises raise mortgage costs. That is why the Bank of England's inflation target used to be the "RPIX" measure – where the "RPI" was the estimate of the change in the overall cost of living and the "X" meant we excluded the cost of mortgage interest payments. These days we use a different index as our policy index – the "consumer prices index" (CPI). Now that still isn't a measure of the overall cost of living (contrary to just plain wrong government claims that it is such – it only includes a bit more than 70% of the goods and services that enter into the RPI). CPI is a policy index.
There is a further point to understand here. Tom's argument depends on the idea that the price level might legitimately and efficiently change. But even the Bank of England policy index (the CPI) is not targeted in levels, but in changes. That is to say, the Bank of England's task is not to control how the level of CPI changes over time. It is to control how it changes from one year to the next. These are quite different tasks. If I am set a task of controlling the price level that would be a matter of saying that prices should be permitted to rise at, say, an average of 2% per year. So if the index were 100 in year one I would aim for it to be 102 in year two, 104.04 in year three, and so on. But if I am targeting a change in the price level (an annual inflation rate), then what has gone before doesn't matter. So I might begin with 100 and a targeted annual inflation rate of 2%. But if the actual inflation rate between years one and two is 3%, so the price-level is now 103, that becomes the basis for the next year's 2% inflation target, meaning that I try to get the price-level to be 105.06, not 104.04. This isn't just a theoretical point. The Bank of England has had inflation above target for most of the past year. It is likely to see inflation go above 5% later in the year, perhaps 6% next year. As an inflation targeter, the Bank of England isn't supposed to "recover" any of that overshoot later.
So now let's reflect upon where we are. Tom says that policy shouldn't aim to control the price level. But policy doesn't target it, in four ways. First, it can't observe the price level. Second, it couldn't control it directly even if it could observe it. Thirdly, our targets don't even involve our best proxies for the overall price level but, rather, and quite deliberately, price indices that we know exclude all kinds of important consumer prices. Fourthly, even in respect of the indices that we do target, we don't attempt to control their level but, rather, the change from one year to the next.
Furthermore, even for Tom's preferred money supply concepts, we wouldn't be able either to observe them directly or to control them even if we could observe them. And Tom would still face the decision as to whether he was trying to control the level of his obervable and somewhat influenceable money supply measures or their change.
One more thing. When we have targets for our policy indices, they are not rigid. There are zones of discretion around them. The target is supposed to place some constraints upon policy, but not to make it mechanical – as if we could mechanically calculate what to do to achieve our desired policy index inflation target and then always did that. When we have a price index as our policy target, there is nothing to stop us from taking money supply index measures into account when exercising our discretion. Indeed, if Tom is right (as he is) in thinking that money supply measures are important indicators, then we ought to take account of changes to the money supply. (Indeed, had the Bank of England taken more account of such measures in 2005-7 we might not be in the mess we are now.) Trying to affect the money supply would be an important mechanism for achieving our desired price inflation or price-level outcome. Likewise, if we had any sensible money supply target, the exercise of our discretion would be influenced by our observations about changes in the price level.
So the key question is not whether we control the price level or the money supply. We do neither. The key question is whether we are better to
a) have a target (or goal of some sort) specified in respect of either the level of or changes in some policy index proxy of consumer prices, and then take proper account of money supply indicators in exercising discretion constrained by our prices target;
b) have a target (or goal of some sort) specified in respect of either the level of or changes in some policy index proxy of the money supply, and then take proper account of consumer prices indicators in exercising discretion constrained by our money supply target.
I haven't tried to answer that point here. Regular readers will know that I favour a price-level target (partly for the reasons Tom sets out). But you can read about that in Chapter 1 here.
In this post I have tried simply to make the following points:
- There is no general answer to the question "What is inflation?" There is inflation in all kinds of things – balloons, egos, money supply, cost of living. The thing that comes closest to a genuine question here is "What form of inflation should we care about most?"
- We do care about inflation in the consumer prices level, because it changes the relative value of accumulated assets (particularly monetary/financial assets) and labour.
- We cannot observe or directly control either the price level or the money supply. We can only measure proxies for them and then attempt to indirectly influence policy indices relevant to them.
- The key issue is thus not whether one controls the price level or the money supply. It is whether it is better to target some imperfect and imperfectly controllable measure of prices, taking account of the money supply, or to target some imperfect and imperfectly controllable measure of the money supply, taking account of prices.