The basic thesis today is that the specification of alternative consumption baskets for measuring inflation is a good historical/narrative/rhetorical method, and should be taken up as such by folks with policy in view. “Inflation” is always only ever deeply socially situated, and mostly works as a point of social and political contestation. Rather than fighting over which is the capital-T True account of inflation, we should instead build out a profusion of accounts of inflation all originating from different constituencies with clear and justifiable methodological choices. From there, we can maybe start to triangulate towards what’s going on in the economy in the discourse.
The title is a little goofy, and a play on a core idea in the work of Harold Bloom: that the only interpretation of a poem is another poem. Unlike Derrida, Bloom says that this dynamic is only true for “Strong Poets” (something meant to be read as Famous White Bigwigs, the kinds of people David Foster Wallace would deem important). In Derrida, the only interpretation of anything is another interpretation, bonus points if you can take apart another interpretation and use its bits for yours. It only matters where it comes from insofar as that fact of origin is an unavoidable aspect of the interpretation itself. The argument I’m trying to make today maybe comes close to “the only way to beat a model is with another model,” but I’m using literary figures to make sure everyone understands that I’m trying to situate things within rhetorical rather than predictive space. Ultimately, the interpretation of a measurement of inflation is only ever going to be a different measurement of inflation.
Tooze has this great book, “Statistics and the German State,” the argument of which is that for there to really “be” a macroeconomy that is susceptible to management, there have to be legible ways to abstract from it. As I’ve said in other posts:
“What we call the economy is really just a bunch of objects and activities out there in the world. We cut these objects and activities up in a particular way that we think of as “economic” in order to direct them to particular ends, or to understand why some things happen instead of others. The trouble is, measurement is fundamentally theory-laden. That is to say, when you decide to start measuring the economy, you first have an idea of what it is you’re measuring, in order to pick it out from the stream of consciousness as a particular thing.”
This problem holds 1000x for governments looking to intervene in their economies. How can you know that the way you’re cutting these things up is the best way to create the economy you want? Activists also want to use the governments language when lobbying one way or another, so the choices the government makes in terms of aggregation and theoretical specification have huge sway on the discourse at large.
Lately, people have been arguing over one of those ways, “inflation.” The way it shows up in the discourse is an absolute mess, because “inflation” corresponds most closely to the folk idea that there is a definite, clear, and isolable amount of stuff a dollar buys. Where you are, when it is, what else is going on: doesn’t matter. There has to be an answer to the question “how much a dollar cost.”
The problem is – as everyone clever has been pointing out since the beginning of the pandemic, and really back to the 1930s – there’s no “final” measurement of inflation that meaningfully holds for all people in all locations. Pretty much everyone buys different things in different amounts every year. Worse, things are usually priced differently in different places!
So to get something that you can get a handle on, you have to spin up a consumption basket. The idea is that, if you track movements in the price of a handful of things that people usually use their money to buy, you can get an estimate of “inflation” in the large and in the main. The Consumer Price Index (CPI) does this in the US by looking at movements in the price of a basket of goods in different cities, and weighting them accordingly. They even have Secret Shoppers who go in and check the price on a gallon of milk, or a bacon egg and cheese, or a lawnmower. Honestly seems like a fun job. The Personal Consumption Expenditure (PCE) measure of inflation is a little different, because it relies on statistics collected about how much people actually spend by category, rather than what stores price goods at, and aggregates and weights differently. Everyone nowadays looks at these two most of the time, and a lot of arguments wind up being over the correct window of CPI to use, or whether Core PCE is better than plain old PCE because it strips out certain noisy serieses.
The trouble is, it’s hard to really make these measures mean anything to anyone except inflation cranks and a couple hundred people at the Federal Reserve. No normal person, in their real life, feels a 15 basis point move in CPI or PCE. People who make their living modeling, expecting, and trading these moves do, but that is really not a large part of the economy.
So a natural response to this state of affairs is to say “hey, these measures are pretty arbitrary, and don’t say anything about any particular social story, except in EXTREME circumstances (i.e. if you get a CPI-like measure registering +20% or more for a couple years and people start yelling “Weimar” in a crowded theater), so they must be fake or wrong.” You get normal people trusting histrionic outbursts on cable tv about a 15bp move, or Shadowstats, Bitcoin, and goldbug types saying that inflation has actually been One Million. These arguments are always pretty funny, because you get charts of M1 or M2 with rage comics overlaid on them and silver plotted on a second axis, just ridiculous.
The other alternative – and the one I’d like to argue for here – is to take a deconstructive stance. The inflation measures that exist today are composites of serieses of prices that you can get ahold of, or at least get ahold of the movements of the deltas in. If you can get these, you can use an inductive social approach to figure out what a useful consumption basket for an alternative inflation measure might be, and then actually calculate that measure using the existing data!
Rather than arguing that there exists a single correct measure of inflation, and that we could calculate it “if only we ______,” there is ample room to deconstruct existing measures and reconstruct them around a constituency you represent, or social story you want to tell. As long as things are priced differently from one another, differently from themselves in the past, differently in different places, and purchased in different amounts, the “value of a dollar” is always going to be locally situated. The idea that the only valid levels of analysis are the individual and the full aggregate is a little silly to me.
Someone who has done this pretty effectively – even though I don’t agree with his politics at all – is Oren Cass. A few years ago he pulled together something he calls the “Cost of Thriving Index.” This is a measure of inflation built using a consumption basket that more or less represents what Cass sees as a baseline for a white middle class family: owning a home, college tuition, new car every now and again, health insurance. What he shows in the report is that if you use this social type as the inductive starting point for your measure of inflation, the story is actually dramatically different than the story told by CPI or PCE. When you bring movement in median wages among that social type into the story, things get even more dire. His big point is that you used to be able to get to the standard of living he uses as a baseline using around 6 months of the median wage, and now it takes around 13 months of the median wage per year. Since there are only 12 months in a year, this is, ah, a problem.
A house and car and health insurance is something that people can naturally connect to, and may actually judge their economic prospects in terms of. It tells a story about the experience of a specific constituency using the method of “constructing a representative consumption basket.” Building a measurement out of a weighted sample of a wide array of consumer products – milk, rental cars, haircuts, and musical instruments, to take a random sample – tells a less coherent social story. Worse, these are all pretty elastic goods: if beef is expensive when I go grocery shopping, I just get pork instead. Lord knows I have ditched even pork sometimes recently because of all the wildness in price from Asian Swine Flu in China, I’m not paying $3/lb for pork shoulder lol.
The important thing to remember though, is that Cass’s measurement is not “more correct” than CPI or PCE or any other inflation measure: it is the same data used to tell a specific story about a specific social group that is parseable for political and rhetorical purposes. There certainly exist the kinds of families of four that the exercise is built around, but there certainly exist a multitude of other kinds of folks in different geographies with different aspirations as well. Please please please do not read me as endorsing Cass’s program.
I have frankly, for a long time, wanted to use this as a way to do something like “Marxian Type of Guy Studies,” and look at something like, “given a consumption basket that represents a Type of Guy (a starter pack meme for example), its cost over the years, and prevailing median wage over where people with that consumption basket and social structure lived, how many people get to be that type of guy?” I think the classic example would be the always-maligned “New York Hipster.” Their consumption basket has changed from the 50s to the 70s to the 00s to today, at the same time as wages, asset values, and a bunch of other things have changed. The specific social signifiers and structures have changed as well, but it’s possible to ground all this in good empirical data gathering! You would want to think about the rent on a terrible apartment, the cost of a pair of jeans, whatever particular weird social signifiers people are up in arms about at the time. I don’t know, how much did Shea Stadium or Palisades pay for rent, and how different was that from what CBGB paid in the 70s or what the Apollo paid in the 30s?
Competing aggregates of “inflation” that are socially-specific have the potential to give a far better window into the history and stakes of certain present and future approaches to policy. They force you – by constructing your own basket and justifying the inclusion of specific goods and specific geographies – to address things bottom up. Then the story contained in the aggregate has an effective field and a clear range of impact. Inflation isn’t a process that happens “in aggregate.” It’s a measurement of the outcome of billions of pricing decisions made every day by legions of unrelated folks, not a single decision somewhere that radiates out to every individual part of the economy. If a whole bunch of price decisions are trending the same way in, say, San Francisco, that may mean something for people in San Francisco but not Omaha.
At the same time, working bottom-up helps inoculate against the idea that the optimal response to inflation is the curtailment of demand. If the price of housing is rising somewhere, the best answer is probably to build more housing, not make everyone so poor that they can’t afford to keep bidding the price up.
It is also just good political-epistemic practice to construct differently-situated differential accounts of the same theoretical “process” when that process has weight and power in the discourse. Instead of subsuming everything under arguments about the “correct” value for one hyper-aggregated and “objective” measure, you can paint a more finely-grained picture of competing stakes among competing social groups. If you have to actually defend why movements in the price of good X matter, you can’t help but make a rhetorical and political argument for its inclusion in the index. If you go farther and construct your index, you can make it solve actual rhetorical problems.
Ultimately, I just want people to go out and dig into the detail tables in the inflation data, and make their own trackers. At the very least, for organizations to recognize that this data is available and represents a good way of ordering the economic universe. We’re all good postmodernists now, and sometimes there are better ways of approaching the problem than waiting with bated breath for the next CPI print.