Wednesday, July 17, 2013
"Asset price inflation" is not inflation
Situation 1: Imagine there is a new craze for meatloaf. It's a fad. Meatloaf restaurants pop up all over America, from Park Avenue to Lubbock, Texas. As a result of this increased demand, the price of beef goes up.
Situation 2: Imagine everyone in the country magically agrees to change the definition of "one dollar" to be what used to be called "fifty cents". The price of everything would immediately gets multiplied by two. That includes beef.
The first of these situations reflects an increase in the degree to which people are willing to give up other useful stuff - back massages, cars, etc. - for beef. The second does not. Hence, we would like to differentiate between those two situations. Traditionally, economists do this by calculating a "price index", which is an average price level of all the useful stuff that people buy. In the first case, the price index will go up only a little, or maybe even not at all, as a result of the increase in desire for beef that led to beef's price rise. This, economists call a "relative price change". But in the second case, the price index goes up a lot as a result of the change in the unit of account. This, economists call "inflation".
There are some subtle points here, concerning what would ultimately happen as a result of these two situations, and how we would measure these changes if we were gathering data. But you get the basic point. Prices can change because of changes in the value of one thing relative to another, or they can change because of changes in the unit of account. We want to distinguish between these things. And "inflation" is the word we have chosen to refer to the latter kind of change, not to the former.
This is why I don't like the term "asset price inflation". This is a term that some people use to refer to increases in asset prices. They use the word "inflation" for two reasons:
1. They think that asset prices change in response to monetary policy. The price index is also widely believed to change in response to monetary policy. Hence, they see an analogy.
2. They think that asset prices, such as the price of used houses, should be included in the Consumer Price Index (currently only "housing services", not house prices themselves, are included in the CPI).
I could argue with either or both of these. But I am not going to. Because the reason I dislike the term "asset price inflation" has nothing to do with either of these points.
Suppose we did include house prices directly in the CPI. They would still be only a small part of the CPI. Suppose house prices were 10% - a hefty chunk - of the "market basket" that is used to calculate the CPI. And suppose house prices went up by 10% while the rest of the price index remained unchanged. That would increase inflation by only 1 percentage point.
Most of the increase in house prices (9/10 of the increase) would not represent inflation. It would represent a rise in the relative price of houses. It would mean that people are willing to (implicitly) give up a larger quantity of back rubs, iPads, cars, and meatloaf in order to obtain a house. It would not indicate a change in the unit of account. The unit of account would remain relatively constant, changing by only one percent, even as house prices rose very substantially.
So to point to an increase in the price of houses and to call that "asset price inflation" is just as misleading as it would be to point to an increase in beef prices and call that "beef inflation". There's only one kind of inflation, and that is inflation itself. You have to add everything in before you can tell me how much inflation there is.
Now at this point you may be thinking, "OK, I see your point, but so what? House prices are important, so central banks should pay attention to relative changes in house prices. So this is all just semantics. Call it inflation or call it whatever, rising house prices still means the Fed needs to tighten."
But here's the thing: We know inflation - the real inflation, i.e. changes in the price index - is bad if it goes too high or too low. We know that, whether the ideal rate of inflation is 1% or 6%, it almost certainly isn't 15% or -4%. Within certain bounds, price stability is good. With asset prices, we have no such guide. If houses go up in value by 15% in a year, are we sure we want to push them back down? Our recent experience with a housing bubble probably convinced a lot of people that the answer is "yes", but remember that housing price rises don't always indicate a bubble. And on the flip side, suppose house prices went down by 4% in a year. Would we want the Fed to always push that back up?
What I'm saying is: Even if the Fed really ought to respond to asset prices, the way in which is should respond is not very similar to the way in which it should respond to inflation.
So even the loose, casual analogy between inflation and asset price changes is flawed. They aren't even similar phenomena. They don't generally have similar impacts on the real economy. And they don't demand similar responses from the Fed. So let's stop using the misleading term "asset price inflation"! Let us strike it from our lexicons!!