I recently explained why I thought price level estimates may not be as useful as the ideal price index is in economic theory. Although theoretical treatments of the price level do consider quality-adjustments that are necessary, a lot of technological innovation may better be thought of as a reduction in the price of some new good from infinity rather than an improvement in the general quality of an already-existing item. Three problems immediately strike me as relevant:
(1) As an example of how revolutionary innovation presents a conceptual problem, consider that the leap from horse-carriages to automobiles could be included in the quality-adjusted price for transportation, albeit imprecisely. On the other hand, the Internet allowed us to do things that were previously impossible; by the mid-1990s the price of a consumer email had fallen from approximately infinity a decade ago to the cost of a computer and dial-up service. Similarly, the advent of modern home appliances provided a higher quality substitute for full-time homemakers at a lower price than the unmeasured opportunity cost of having women work at home.
(2) In addition to conceptual problems such as the above, quality adjustments are too crude and may give a false sense of empirical precision. A so-called hedonic adjustment for laptops might regress the price against memory and display resolution to recognize innovation in computer technology that may not be evident in the price. But in practice these adjustments can’t retain relevance for more than five or six years; for example the relationship people have with their personal laptop has changed immeasurably over the past 20 years.
(3) There’s no good way to properly adjust for durable goods. The Bureau of Labor Statistics presents this as a dilemma because economists think the price of durables increases more slowly than frequently-consumed goods. A more conceptually-intractable issue is that as the economic life of goods increases the durable purchase is depreciated over a longer time; a special version of quality improvements that aren’t measurable.
Each of these pose a big problem because they affect some goods and not others. For example, the most durable purchases, like houses and cars, have sufficiently developed rental markets allowing economists to discover imputed prices. Laptops and phones may last longer now, though it would be hard to observe. This results in an inconsistency that corrects itself, resulting in a directional bias. Let’s say laptop prices increased by 20 percent over 5 years, quality remained similar, but usable life increased from 5 to 15 years. A personal expenditures based index would initially note an increase in price level as people purchased the more expensive laptop. However, as they replace laptops at a much lower rate than before, the relevant measured expenditure would indicate a decline in the price level. After some time, the measured price level of the two laptops may even out, but there is a problem because so long as we know that there is rapid innovation within an industry, the presented price level estimate presents a directional bias.
Year to year changes in the price level may not have some of these problems, but they are also irrelevant when they are as low as they are. It may be true that things that already exist have a tendency to slowly increase in price over time; but it’s not clear to me why that’s relevant for long-term monetary policy decisions, since the measure doesn’t seem to tell you much that walking around a few retail stores does not.
Measures of inflation seem more useful to me as a benchmark price for common consumer items, than as some estimate of a theoretically-relevant price level that is used to inform monetary policy choices or other important economic policy.