The idea that slow marginal rent inflation would be followed by a steep deceleration in the measures of average rents in the official price data was helpful. However, it seems now to be well past its use-by date. So, from here on in, we should probably stop dismissing those government rent data as mere laggards and replacing them with our sense of how marginal rent inflation is tracking. The main reason is that the apparent gap between marginal and average rent inflation has narrowed a lot. So now, there is probably more to lose from the slippage between how marginal and average rents are measured than there is to gain from the presumed lead provided by marginal rents.
I have been edging in this direction for a while, mostly because the official measures of average rents have indeed decelerated steeply, just as the earlier hypothesis (which was hardly some brilliant insight) predicted. And most recently, the various private sector proxies of marginal rents have begun to hook higher. For example, the 3-month rate of change in the seasonally adjusted version of Zillow’s Observed Rent Index (ZORI) has accelerated 200 basis points, 3 ½%, since January. I mentioned that in an earlier note, although I fear I may have misreported it as the 12-month change. And late last week, Cotality reported that the 12-month rate of change in their Single Family Rent Index (SFRI) has picked up to 1.8%. The chart below shows that headline inflation rate split between attached and detached, but it is actually easier to read than the Cotality depiction at the top of their report. To me, that chart has too many lines.
Marginal rents have perked up, at least a bit...
I do not know what the short-term rates of change in the Cotality are, but their report mentions that sequential rent growth has recently returned to “seasonal patterns,” which I think means that it is running at a normal pace in seasonally adjusted terms. I prefer to monitor these measures of marginal rents, rather than forecast them. But my own bias is to take this normalization idea seriously because I do not see a strong case for rent growth to remain durably depressed. It appears to have fallen below normal in the wake of a wave of multifamily completions, and it may well have since renormalized. The government measures of average rents have also renormalized recently, although from the opposite direction.
Given that the lead-lag relationship here now seems to be less important, differences in how marginal and average rents are measures loom a bit larger in relative terms. For example, the BLS’s so-called “experimental” measure of marginal rents very clearly leads their experimental measure of average rents, which is helpful from the perspective of method, because it tells us to watch the marginal when there is a large gap. However, both the experimental marginal and average rent metrics use an hedonic adjustment that is slightly different from that used in the official government statistics. So, for example, the BLS’s experiment measure of the average is very tightly correlated with the official measure of the average when measured as inflation rates. But there is a gap there that averages about 50 basis points over time, which pollutes the marginal measures as well. When the leading marginal measure is running 500 basis points below the lagging official measure, quibbling over such precision misses the point. But that is no longer the case. So, I will quietly return to just taking the government measures roughly at face value as a measure of what is actually going on.
Changing the subject slightly away from method and toward speculation, it does seem as though the idea that the CPI will now be biased downward relative to the PCE Price Index because of the higher weight it assigns to rents is now less compelling. If the Trump Administration were to fiddle with the CPI to make inflation look lower, that might mess up the TIPS market. But that is separate. The rents issue is now a lesser reason to expect the CPI to track a bit low relative to “true” inflation.
Some might put this whole discussion much more briefly. This issue is just far less important than it was. That too would be fair.
… although this is the main source of the convergence
(between average and marginal)
Data are actual to July and effectively actual to August because the inference from data in the CPI release is very tight.
