One of the common complaints about the (CPI) is that it doesn’t reflect the way inflation feels to the individual. Much of the reason that is so is the fact that we have built-in cognitive biases; a simple example of this is that we tend to encode price increases as ‘inflation’ while price decreases are recorded as ‘good shopping.’ Accordingly, prices which oscillate get recorded in our brains as ‘inflation’, even if the net movement is not much.
My favorite example of this is gasoline. Ask anyone, any time, if they think gasoline prices are higher or lower than they were (say) four years ago and they will say ‘higher’ about 90% of the time. Right now, for example, we know gasoline is at an incredibly high level – $4.101 was the national average at the end of July. That’s compared to the level four years ago, at the end of July 2022, when it was … $4.212. And wildly higher than it was in 2008, when the July 2008 AAA price was … $3.898. In other words, for the last 14 years gasoline has risen at a compounded average annual rate of 0.36% per year, well below the inflation rate.
Sometimes it’s up, looking back over some fixed period; sometimes down. But over the last decade and a half, gasoline basically hasn’t moved anywhere (and that’s including the fact that gasoline taxes have risen).
But if the Bureau of Labor Statistics were to announce that gasoline was basically unchanged since 2008, some people would lose their minds with screams of conspiracy theory. It’s not, unless the American Automobile Association is in on the conspiracy too.
A more fair critique (but not to the extent that conspiracy theorists would have it) of the CPI is that the index is a creature of a government agency and represents standardized weighting and collection methods that may or may not always fairly represent what a particular consumer faces with respect to price increases. This problem is acute in the United States, where a vast geography and very diverse systems of local rules and regulations can directly affect the national average even if it only directly impacts a constrained region. So, when Mayor Mamdani freezes rents in New York City, it will impact national rent inflation a little even though rents are only frozen for a relatively small number of Americans.[1]
And that leads to the fairest complaint at all: if the government engages in price-fixing, does that magically cause inflation to vanish? Of course not! Although rents may be frozen for some units in New York, costs for landlords continue to rise. You can’t wave a magic wand and make inflation go away, but unfortunately you can wave a magic wand and affect the measurement. I wrote about this last year in “Mamdani’s Effect on the CPI”,[2] which in turn points you to a podcast I did a couple of years ago on price fixing. The bottom line is that fixing prices does not change inflation, but it changes inflation measurement because the inflation gets displaced into poorer quality. Quality is hard to adjust for, and whenever the BLS tried people screech about ‘hedonic adjustment’ even though poorer quality would of course increase inflation by making up for the part that is missed in the price level itself.
Anyway, there aren’t many easy solutions for manipulative mayors.
The salient point in that article on the rent freeze is that landlords’ costs are increasing. The methodology for estimating rent inflation is based on estimating landlords’ cost pressures, with the assumption being that the return on investment for a landlord doesn’t drastically change over the years when you average across many markets.
The model has done an excellent job of not getting sucked into the ‘persistent deflation’ story apparent in various measures of rent changes that focus on apartments that are actually turning over, such as the Apartment List index.
These data aren’t false; they just measure something very particular and that is how much competition there is among landlords to let out vacant units. That doesn’t tell us a lot about what landlords are charging for currently-occupied units – and, more to the point, it can’t be the case that rents fall persistently when landlord costs are rising persistently.
But – it is very important to realize that measurement of inflation is not the same as inflation. This isn’t unique to inflation, of course. The monthly Payrolls number depends on the sample and the response, and it is well known that since it doesn’t capture new businesses or closing businesses it is inherently inaccurate. It’s true of virtually all economic data, in fact: the number is a measurement at a point in time, based on a specific methodology and calculation, and so only suggests the underlying metric being measured. The difference is that a massive notional amount of securities are indexed to the CPI!
The Mamdani effect on rents only amounts to probably 5bps/year on the CPI rate. But if the Socialist wave gains momentum (a hard-left Senate candidate won the Democrat nomination yesterday in Michigan!), we need to be cognizant of two things. First, Socialism has a wonderful record of causing an increase in actual inflation when we include the historically-unblemished record of declining standards of living. (Not only that, but eventually inflation happens anyway. As a reminder of that, note that for the second year in a row Obamacare premiums are going to rise by double digits, surprising absolutely no one who understands economics). But second, Socialist policies would indeed have a tendency to cause measured inflation to decline. You think the Fed is in a pickle now? How about when measured inflation is 1%, but adjusted for quality it’s 6%!
***
[1] Fortunately, freezing rents in NYC, and so artificially changing the measured inflation in New York, won’t result in freezing Owners’ Equivalent Rent in the same area. Although OER is based off the primary rent survey, the BLS recognizes that rent-controlled units are not market rents and so using those in the sample for OER would be misleading. One does wonder how the BLS will calculate OER for New York if Mamdani freezes all rents.
[2] Note one error in that article is that I assumed OER would also be affected. As noted above, it won’t be.