March 08, 2006

On Incomparability

One of the things I've been thinking about recently is the difficulty in comparing and/or equating different products so that you can derive meaningful metrics about the category. To some extent this is a reflection of a larger ontological problem of deciding which objects are meaningfully the same, but it has special relevance in an economic world defined by technological progress.

While this may seem somewhat esoteric, I would argue that it's incredibly important to understand this phenomenon because it systematically makes us overestimate inflation, underestimate standard of living increases, and, in general, be more pessimistic than warranted about Progress (with a capital 'P').

This difficulty is particularly problematic when trying to measure the price of something over time or space. As a simple, mundane example, take the price of gas over time. While we pretend that there is some consistent thing called "gasoline" and that it's price is denominated in something consistent called US dollars, there are three things that make it difficult to compare the prices over time:

  1. The distinction between nominal dollars and real dollars. This is the most obvious, and most serious people writing about prices over time will use some kind of constant dollars. The leading ways of doing this (chained or unchained CPI, GDP deflator, etc.) are all valiant attempts to address the Incomparability Problem, but all have serious flaws (that I'll discuss more below). But under any scenario, a dollar 50 years ago is not worth the same as a dollar today.
  2. The gas itself has changed. While gas seems like a nice, simple commodity, handily measured in concrete units like gallons, the term hides the fact that gas has changed dramatically over time. Tetra-ethyl lead content, octane level, ethanol and MBTE admixture, volatility – all these make the gas you buy today very different from 30 years ago. For instance, maximum allowed volatility is lower today in order to reduce emissions during refueling (resulting in a "better" product for some definition of better.) How does/should this impact your analysis of price change over time?
  3. Consumers don't really care about gallons of gas, they buy "passenger miles", or maybe even "cargo miles". As with many products, the utility that consumers get out of the product is coupled to a complex network of enabling technology that itself changes over time. The fuel efficiency of engines has gone up dramatically over the last several decades allowing cars to go further (or haul more) for the same gallon of gas. If my car today can go 29.1 miles per gallon vs. 18.8 MPG in 1977, how does/should that factor into the "real" price of gas?
As you might imagine, the case gets infinitely harder when you move away from a commodity like gasoline. Additionally, the complexity increases as you move from the "same" product over time, to the "same" product in different places, cultures, or regulatory regimes. For example:
  1. Medical care – rather than across time, medical care illustrates the problem across regulatory regimes. Is cancer treatment, for instance, the "same" when you have to wait months for treatment vs. a couple of weeks? While the treatment itself may be the same, factors around the provision of the treatment make the former a qualitatively different product than the latter, particular if it means the difference between success or failure. If we pretend they are the "same" we will be confused why people pay more for the latter (in the US) than the former (in Canada).
  2. Frozen food – one effect of recently having a baby has been a sizable uptick in the amount of frozen food that Julia and I eat. I have to say that I have been truly astounded at the quality of the frozen food available at the grocery store – specifically the change in quality from when I was a child and last ate my share of "TV dinners". Now, these new frozen foods probably cost more (even adjusting for inflation) than the old ones, but how does one compare the price of excellent frozen Indian dishes, vegetarian Mexican food, health food, etc. with the Hungry Man™ and other meager selection we had in the early eighties? The situation is replicated in every category of food. How many more options does the butcher store at the supermarket now offer? Do you even remember when fruits and vegetables had seasons, out of which you were out of luck? If we pay twice as much for avocados, but we can get good ones in February, how do we count that?
  3. Housing – while often lamented as a component of sprawl, it's uncontroversial that the average American home has gotten significantly larger over the last several decades. Meanwhile, housing costs as a share of family income have stayed relatively flat or decreased (driven primarily by lower nominal mortgage interest rates and higher family income), even with the current bubble. Can we call wages stagnant if the same percentage of them buys more house each year?
  4. Consumer electronics – the fast-paced world of high-tech gadgets is obviously problematic. An ever larger portion of consumers' income goes to computers, MP3 players, digital cameras, televisions, cell phones, etc. By any measure of delivered features (gigahertz, megapixels, gigabytes, diagonal viewing area, etc.) there has been massive deflation in these product areas. Five years ago, the iPod didn't exist. Now, more than 41 million of them have been sold, with 14 million in the first quarter of 2006 alone! Sure, I paid a heck of a lot more for "mobile music device" (four times as much?) than when I bought my first WalkManŽ, but is that really relevant?

The list goes on and on. In virtually every category, the product has gotten dramatically better either functionally or aesthetically – in ways the consumer can appreciate directly. And in cases where it hasn't, it's often due to significant new regulations that either impart a diffuse benefit (e.g. make the product safer, less polluting, etc.) or protect the spoils of rent-seekers. George Reisman (via Nick Szabo) has more on these aspects.

So, why is this relevant? Why do I think it's important to remind ourselves of these facts? Because it's directly relevant when we consider "real" wages, standards of living, measures of growth, and even inequality. Are American workers "getting a raw deal" with "flat wages"? Should Paul Krugman be as exercised as he is? Is American GDP growth actually much higher than the dollar figures show because of the hidden improvement of things? Are we therefore pulling farther ahead of the rest of the world than one would think? Or are these improvements leaking across borders to lift (some of) the rest of the world up faster too?

These are all important questions that don't have easy answers. But too often they are ignored or glossed over in discussions. But the effect on true standards of living are increasing – and increasing at an accelerating rate. And the danger of that is that, blind to what is really happening, we may make policy choices that curtail these improvements in the future.

Posted by richard at March 8, 2006 09:15 PM


Ugh! So what do you suggest? I think I'll just be satisfied with the Economist's Big Mac index.

Posted by: Michael Weiksner at March 10, 2006 01:26 AM

Great article! My response is here . Generally I agree that incomparability makes weighted price indexes highly problematical, but give some reasons to believe the current U.S. CPI underestimates rather than overestimates dollar inflation.

Posted by: nick szabo at March 20, 2006 06:51 PM

Good site, good blog, thank

Posted by: Devid at April 29, 2006 10:03 PM