Empirical Price Theory

It's about the world, not the model

We talk a lot about UCLA price theory. But what is UCLA econometrics? According to Merton Miller “You go to Armen Alchian and you ask, ‘Armen, is this number about right?’ And Armen says, ‘Yeah, that sounds right.’ So you use that number.”

While I love that joke, it makes it seem like price theory, such as that from UCLA is anti-econometrics or anti-empirical. Plus in this price theory newsletter, we tend to focus on the theory in price theory. Let us never forget one of the things that made price theory unique from mathematical economics of the 1970s and 1980s is that it was heavily empirical.

Data and, more generally, many types of empirical work are intimately intertwined with price theory and always have been. Gary Becker’s book on human capital is subtitled “a theoretical and empirical analysis.” Kevin Murphy, probably the leading active figure within price theory, did much of his early work on empirical labor economics. We gotta take the data seriously.

Here are three types of empirical work that mesh well with price theory, qua theory, that we have emphasized in this newsletter.

Empirical Supply and Demand

Price theory only has a few tools. (All hammers, the critic snickers.) And the most used tool is obviously supply and demand. With that, I’d call the most common form of empirical work within price theory “empirical supply and demand.” I’m not great at coming up without snappy taglines.

By “empirical supply and demand,” I mean a broad class of empirical papers, all driven by the tool of supply and demand. This is in contrast to the old idea of “measurement without theory” or the modern fixation on causal identification, divorced from economic theory.

One simple example of this approach that I love is a speech by Kevin Murphy on inequality. It is not at all technical or the fanciest analysis but I think it illustrates the power of empirical work with the aid of supply and demand He lays out the basic forces of supply and demand that are driving growing inequality. Also, see this newsletter which draws on Murphy’s work on inequality.

If that speech is too fast and loose for you, check out "Changes in Relative Wages, 1963-1987: Supply and Demand Factors" by Murphy and Larry Katz, who you empiricists may have heard of. The back and forth between data and theory is central and that even comes out in the structure of the paper. They show some basic data on wages over time, they go to the simple theory, and then they go back to further interrogate the data. Much of the great work in economic history follow this approach as well.

It’s not about testing a theory. It’s not about causal identification divorced from theory. It’s the blend of theory and data that’s powerful.

Measuring Transaction Costs

The UCLA school of price theory was known for its emphasis on the theory of transaction costs. But these economists also did empirical work, including something that I don’t see anyone doing anymore: measuring transaction costs.

For example, Demsetz in 1968 backed out transaction costs from bid-ask spreads on the NYSE. The theory and connection to data are quite clever, especially for the time. Wallis and North try to measure transaction costs for the whole economy.

One reason I find this approach so interesting and well-suited for a price theorist is that it’s basically inconceivable without theory. You can never directly measure “transaction costs” in the same way you can measure “wages”. You need the theory to tell you what to measure, especially since it is almost always going to be an indirect approach. Demsetz measured bid-ask spreads as a type of shadow of transaction costs, not transaction costs themselves. In this way, it is like the measurement of human capital. You would never go out and try to measure human capital if you followed the measurement without theory approach. You wouldn’t even think of it. (To be fair, none of us would have thought about it without the geniuses of people like Becker but that’s a different point.)

Case Studies

If measuring transaction costs has fallen out of favor, case studies have really fallen out of favor. Serious empirical economists do causal identification. Case studies are for the plebs (or the sociologists).

I’ll argue case studies can be an extremely powerful tool when combined with economic theory. I have in mind papers like Demsetz’s “Towards a Theory of Property Rights”, which is mostly remembered for its theoretical angle (it’s even in the title) but it also included a short case study applying the theory to Native Americans and the fur trade.

A more in-depth case study is in Steven Cheung’s “The Fable of the Bees: An Economic Investigation.” In that paper, Cheung shows how beekeepers and farmers were able to work together to overcome the “market failure” that many economists thought was inevitable. Again, this paper highlights the interplay of theory and evidence. The standard theory said this was impossible. Evidence corrected the theory.

Yet, it wasn’t measurement without theory. No one without a theory (if we posit that is possible) would think much about the bee market. The market is interesting because of the theory

I’d often rather see a new theory with a simple case study showing how the theory maps to the world than just an abstract theory alone. I even say this as someone who mostly writes papers without case studies and loves abstract theory. The case study adds a lot.

I admit that I’m no expert on empirical work by any stretch of the imagination. Yet I’m struck by how when one looks at the giants of price theory, basically all of them were doing serious empirical work in their time, even if some of it looks simple today. If you’re a serious empirical economist in 2021, I hope you’ll see price theory as a complement and not a substitute for what you are doing already.

Share


P.S. I just published my first op-ed (expect more to come). It’s at Techdirt on how the FTC will drive away top economists. As a serious empirical economist myself (see 👆), I would never selectively read the evidence to support my argument, but….. it seems my prediction may already be coming true.