I heard a great talk the other day on how difficult it is to measure the owner-occupied housing component of CPI. This is not merely a technical point, as it makes up about 20 percent of the index.
Before the early 1980s, the owner-occupied housing component was measured as the user cost of housing: that is, the value of the housing stock multiplied by an appropriate rate of return plus depreciation and maintenance net of expected appreciation. This measure had the problem of being exceedingly volatile, in contrast to rents, which are not. So beginning in the 1980s, households were simply asked what they thought was the level of rent that their house would command. So how many of you out there know what your house would rent for, and by how much it would change each year? I thought so.
Turvey's point is that we have a major component of the CPI that is not directly observed in any market, and that this creates some serious issues as we try to figure out real incomes. He suggested looking at the cost of new houses, but this of course has the problem that (1) new houses are not very representative of the stock and (2) are in very different locations from the general stock. The problem remains in search of a solution.
No comments:
Post a Comment