Inflation and the Substitution Effect

I sometimes pick up my groceries at Iceland (the supermarket, not the country). When I was there today, I noticed a bunch of frozen pizzas which were typically on sale at £1 now on sale at 2 for £2.50. I thus bought zero of them, opting for a substitute (that turned out to be pretty good!). This is one example of a shortcoming of the generally accepted way of measuring inflation (via a consumer price index).

Inflation refers to an increase in general prices of goods and services (in numerical terms). With the devaluation of the pound after the Brexit vote, I’ve experienced imported inflation (because a pound buys fewer dollars/yen etc., and overseas suppliers of goods want to be paid in their currency).

Of course, one problem is how one measures “general prices of goods and services”. Typically, inflation is measured via changes in a consumer price index. This is an aggregate of the prices of a supposedly representative basket of goods and services. The basket is reviewed periodically, as some goods become more (or less) frequently consumed. For example, in the UK, this is managed by the Office for National Statistics. The 2016 basket is listed here, and changes included the addition of video game downloads and removal of nightclub entry.

Of course, individuals’ spending patterns can be very different from the “average”. One can imagine a personal inflation rate that could be very different from CPI. For example, I don’t drive, so a significant increase in the price of cars might not affect me very much. (There could be some knock-on effects, e.g. if the cost of minicab rides increases.) Similarly, an increase in the cost of recreational boats (which are part of the basket) would have little impact on me. Conversely, I consume a fair amount of potatoes, so a cost increase there could have an outsize impact on me…

It probably won’t, actually. If potatoes become expensive, I will eat fewer potatoes and more rice or noodles.

Something similar actually happened after the Brexit vote. I used to enjoy an occasional instant ramen treat from the Japan Centre in London. However, with the pound falling almost 20 percent against the yen in the wake of Brexit (from 160 to about 130), prices were increased significantly. In many cases, this exceeded the 25% cost increase, perhaps because they wanted to avoid further increases if the pound fell further. GBPJPY is now about 149, but the GBP prices haven’t moved, possibly for the same reason. In any case, I’ve switched to eating more of other forms of carbohydrates.

Note that the above doesn’t always work as goods actually need to have a reasonable substitute. Medical services, (possibly imputed) rent and education come to mind. There are opportunities for geographic arbitrage here (e.g. travelling to Brazil as a medical tourist, living in an RV or attending university in continental Europe respectively), but these tend to have higher barriers to entry.

This is the key intuition behind chained CPI; it aims to account for people consuming different goods and services as prices change. This is currently being considered in the US as part of Trump’s tax reform (and would, in the future, save the government money by reducing tax bracket adjustments – chained CPI would be lower than CPI).

Of course, computing chained CPI is a hard problem. For example, there was a parasite problem in salmon farms in early 2017, causing supply to fall. (Again, this was another instance where my consumption habits changed.) Is trout an acceptable substitute? Cod? Chicken? Tofu? Any kind of food? I’d answer yes to the first and kind-of to the second. For the last three, that would probably only be the case in emergency circumstances. There were similar problems with an iceberg lettuce shortage in early 2017, and yet people continued to buy them in spite of prices almost tripling.

Furthermore, one typically only gets to witness substitution effects after the fact. If the price of lettuce triples, it’s difficult to predict how many people would switch or partially switch to substitutes. In practice, statistical offices usually come up with a preliminary estimate and then revise it when they have more data available.

As an individual who has responded to the substitution effect quite a few times, especially since I moved to the UK, I think chained CPI makes sense. For me at least, the substitution effect is real and powerful. It hasn’t received the friendliest of receptions in the US, at least in part for political reasons (it hits the poor and elderly hard, for various reasons). It does make sense to have a different metric to use for benefits or the state pension, perhaps more tailored to the relevant populations (should video game downloads, generally speaking, really go into an elderly CPI?). However, in general the chained CPI seems more accurate and I see no reason not to use it.

Leave a Reply