In the conclusion of my interview with James Bessen we talk about the difficulty of measuring software-driven economic growth, a topic I’ve written about before.
Timothy B. Lee: How should we think about the value that consumers get from the rapid technological changes you’ve described?
James Bessen: Cowen argues that innovation today isn’t the same quality as it was in his grandmother’s day. I think you have to be very careful about that because while it’s true that innovation today tends to be qualitatively different in a couple of ways, that doesn’t necessarily mean it’s any less significant.
The innovations of a hundred years ago were a lot about the mass production of standardized goods. You think about the automobile or electrical appliances, these were things that affected most people and so you can look at an invention from that time, like the automobile was something that most people eventually used. Electrification was also something most people used. It affected most lives. That was because these were standardized products produced for a mass market.
In contrast, information technology is about meeting custom needs. It allows things to be tailored.
What’s an example of that?
One of the things that the desktop publishing revolution did was that it allowed tailored advertising. Software allowed A&P to target advertisements very finely. They could work from a database, track the items, automatically modify the flyers that were going out to each neighborhood—geared to the demographics of those neighborhoods and the particular things they were selling in those stores. It all worked in a very efficient way. It would’ve been possible to do that before, but it would’ve been way too costly.
This is an example of flexible manufacturing, which is also useful in terms of producing goods that are tailored to peoples’ needs.
Today’s super markets carry 50 times as many items as the grocery store of 80 years ago. That’s made possible by flexible manufacturing, computerized logistics, inventory control. Supermarkets have computerized systems for keeping track of what’s being sold at the register, what’s being shipped from the warehouse, and so forth.
Last time I was in the supermarket, I counted and I think there were 12 kinds of apples, 10 different tomatoes, etc. Which I’m guessing would not have been true in the 1970s.
I remember when I was in college, I went to dinner at a friend’s house. They were Italian, and we had to go down to the North End in Boston to get Italian sausage with fennel seeds. Now you can find that all over the country in all sorts of neighborhoods.
Obviously somebody is buying these things. None of these things affect most people like the automobile did. But most people are benefitting from some of these things. So it’s a qualitatively different technology. It’s much harder to measure its impact, with all of these products. It’s very hard to judge what the different qualities of those 12 types of apples, or 20 types of olive oil, or whatever it is.
This must make it hard for the statisticians at the Beureau of Labor Statistics to compute inflation rates.
It’s an impossible problem to measure those things. Quality change is difficult enough to measure for something like an automobile or a computer. Here you’re talking about two orders of magnitude more products. So in a sense, innovation and technological change have a very different feel today than they did in Grandma’s day, but that doesn’t mean that they’re anything less.
This rings true. At the same time, it seems compatible with much of Cowen’s argument. Improved supply chain efficiencies seem broadly comparable to, say, gains in energy efficiency. It’s welfare-enhancing, but not necessarily GDP-enhancing. I think this is a more fundamental difference than something useful merely being difficult to measure.
It’s welfare-enhancing, but not necessarily GDP-enhancing.
It’s not entirely clear what Cowen’s argument is, but I think the existence of a lot of welfare-but-not-GDP-enhancing technologies would undermine it. I take Bessen’s point (and mine in earlier writings) to be that the apparent stagnation of GDP largely reflects a measurement problem rather than an actual decline in the rate at which consumer welfare is improving. Energy efficiency gains that didn’t show up in GDP for some reason seem like it would be a point in favor of this hypothesis.
Well, I imagine that if we delve far enough into this rabbit hole we end up in the kind of “how do you define utility?” arguments that seem to make Will Wilkinson’s arguments addressing inequality ultimately unsatisfying to me. The gist of my objections boil down to my sense that recent growth gains seem to have come to the non-rich disproportionately in the form of better consumption goods and luxury experiences; stuff that’s nice to have, but non-essential, and definitely not anything that could be called an asset. Better software on phones, more fruits for sale in the grocery store, more impressive diagnostic tests at the doctor’s office.
But these gains haven’t come at all in the form of wage growth, and in the meantime various household costs have gone up substantially. So the gains in welfare have a bit of a bread and circus feel to them. Which is more than a little funny, given that Bread and Circus was one of the first chains to bring that sort of improved fruit and vegetable selection to the neighborhood I grew up in.
These gains haven’t come at all in the form of wage growth.
Real wages, as measured by the BLS, are a fraction. Nominal wages are the numerator, and the CPI is in the denominator. If my spending on apples rises by 10%, the BLS needs to figure out whether that reflects me shifting my spending to more “up-market” products, or whether the grocery store just raised its prices across the board. I take Bessen’s hypothesis to be that the BLS is mis-identifying the former for the latter, and as a consequence the CPI is rising faster than it ought to be.
That may or may not be right, but the point is that this isn’t a separate subject from whether wages are rising. The BLS’s claim that real wages haven’t risen reflects a huge number of implicit value judgments about the relative quality of various products. If it’s true that increased spending reflects paying for quality improvements, then wages have been rising and the government isn’t interpreting the situation correctly.