With Krugman Humiliated, This Is What Goldman Thinks True Rent Inflation Is
Last week, in the aftermath of the flaming red-hot CPI print, we quoted Academy Securities chief strategist Peter Tchir who said that "Rent Inflation Seems Crazy", referring to something we first pointed out long ago, in the summer of 2021, when we first made the observation that - even as the Fed was confident inflation would be "transitory" - the central bank (and the CPI report itself) was making a huge mistake in that they were looking at lagging rent, shelter and OER inflation prints. In fact, as we said last week, the "the time to panic about soaring rent was a year ago... as we did here, but not the Fed." For a more detailed analysis, read our article from Sept 2021 titled "What Rental Hyperinflation Looks Like: "Soaring Prices. Competition. Desperation" - it pretty much explains everything.
The time to panic about soaring rent was a year ago... as we did here, but not the Fed.
— zerohedge (@zerohedge) October 13, 2022
Now BLS has just caught up to reality 6-9 months ago. Meanwhile real life rents are finally fading https://t.co/sJtLFdfvND pic.twitter.com/sjJqpCB6Gt
We then added that just as the CPI - and the Fed - were wrong in basing their assessment on 6-9 month delayed shelter inflation data in the summer of 2021 and were underestimating inflation, they are doing the same mistake now, just it's mirror image: they are looking at delayed soaring rental inflation and overestimating CPI, when in reality OER and rents have already peaked and in many cases are sliding fast as far more accurate real-time rent indicators such as Zillow and Apartment List demonstrate.
Not long after, others starting noticing what we first said last September, including JPMorgan...
... distinguished democrat economists (at Harvard of course) such as Jason Furman who apparently was unaware of the lag in the CPI's shelter metric...
Some of us did pay attention to this in 2021https://t.co/iVCU0RFtap https://t.co/ug2cjf6ueN
— zerohedge (@zerohedge) October 15, 2022
... but it wasn't until top democrat economist #1 Larry Summers mocked the most "special" Nobel prize winning economist of all, (and top democrat economist #2) Paul Krugman, that the topic of lagged shelter inflation really hit the mainstream...
.@paulkrugman would be more convincing if he had focused on the housing distortion when it was holding overall inflation down in 2021. He would be more credible if he acknowledged that new rental price inflation of his cited measure (Zillow) is still running at 6 percent. 3/N pic.twitter.com/iRRvdeLR8k
— Lawrence H. Summers (@LHSummers) October 17, 2022
... about a year after we first explained how and why the Fed is making a terrible "transitory inflation" mistake by focusing on the lagged data and ignoring real-time.
So with everyone now talking about government shelter inflation data, real-time rent data from Zillow and Apartment List, it was only natural that overnight none other than Goldman would publish a note titled "Shelter Inflation: The Catch-Up Effect" (available to professional subs), in which the bank tries to calculate what the true rent/OER measure is (which is critical due to its weight in the CPI basket, and thus setting the Fed's reaction function).
The Goldman paper starts off logically enough, by laying out the reported burst in the rent and OER measures of CPI:
Monthly shelter inflation—a term that we define here as the rent and owners’ equivalent rent components of the CPI, excluding lodging away from home, which is included in the official definition—surged to a new high in September. The latest jump was particularly eye-catching because it appeared at odds with more encouraging signs of deceleration in timely web-based alternative rent measures.
Goldman then goes on to explain how alternative rent measures differ in scope and definition from the official CPI measure:
The most important difference is that the most popular alternative measures from companies like Zillow only capture rents on new leases for units that are turning over because that is the only rent information visible to a rental website. In contrast, the CPI measure covers rents on both new leases and continuing leases for existing tenants. The latter usually adjust more gradually, both because landlords are reluctant to alienate and lose tenants by raising their rent too abruptly and because the cost of moving reduces the risk that tenants will move out when market rents decline more quickly than their own. This difference is especially important at the moment because market rents on new leases grew very quickly over the last two years and opened a substantial gap with rents on continuing leases, as we will show below.
While we have traditionally used Apartment List and Zillow as they contain more than enough data to make a conclusion, Goldman expands on this, and in the table below summarizes nine rent measures from rental websites, REITs, and industry sources. The table shows that these measures reached a much higher peak growth rate than the CPI measure, have grown more cumulatively over the course of the pandemic, and have decelerated this year, in contrast to the acceleration in the CPI measure.
What to do with this data? Well, a new paper by economists at the Bureau of Labor Statistics and the Cleveland Fed introduced a new rent index derived from a subset of the CPI sample limited to new leases. Unlike the official CPI shelter measure, this New-Tenant Repeat Rent Index (NTRR) is conceptually comparable to the alternative rent measures. Like those measures, it also reached a very high peak year-on-year growth rate, has risen by more than the CPI shelter index during the pandemic, and has decelerated since the start of this year.
Next, Goldman redoes an analysis we have put together countless times (without having access to the bank's resources or brainpower) and in the next chart, the bank shows "the monthly or quarterly annualized growth rates of the higher-quality alternative rent series that we have access to. These measures have slowed to an annualized growth rate of 3% in recent months, equivalent to roughly 3½% in CPI terms after adjusting for quality decline as rental units age."
Here, Goldman's own model projects a 3% increase in new lease inflation in 2023, a sharp deceleration from the 2022 growth rate "as the vacancy rate rises on a jump in new supply and the labor market deteriorates somewhat, but the higher cost of financing owner-occupied housing continues to spill over positively to the rental market."
That said, while new lease inflation is set to tumble, Goldman notes that continuing lease rent growth is likely to run hotter as these units catch up to new lease market rates:
Using the NTRR Index and the All-Tenant Repeat Rent Index (ATRR) reported in the BLS and Fed paper, we estimate an implied Continuing-Tenant Rent Index, shown as the grey line in Exhibit 7. Our series implies that catch-up boosted continuing lease rent growth to an 8¼% annualized pace in 2022Q2—driving the strength in CPI shelter inflation, which lags the ATRR by a quarter because changes in rent are backdated to when they took effect in the latter—and an 8½% pace in 2022 Q3, and that these units are still priced 3% below the going market rate, on average.
To estimate the speed of catch-up, Goldman then compares the size of the gap between new lease and continuing lease rent levels to the growth of the Continuing-Tenant Rent Index relative to trend. This approach shown in Exhibit 8 suggests that catch-up has closed roughly 20% of the gap per quarter during the pandemic, although as the economists caution, "it is hard to know—the pace appears to have been much faster in 2022Q3, but in contrast there was little evidence that levels tended to converge at all before the pandemic." This, they say, makes the near-term outlook for rent inflation, which depends heavily on the speed of catch-up, quite murky.
Putting the above together, and based on its forecast of new lease rent growth and the best guess of the speed of catch-up, Goldman expects continuing lease rents to rise 6% in 2023. And since rent growth is likely to be faster on continuing leases than on new leases, it should also be faster in the CPI than in alternative measures, as shown below.
Specifically, Goldman's new lease and continuing lease rent growth forecasts imply that CPI shelter inflation will rise from 6.8% year-on-year currently to peak at 7½% next spring before gradually decelerating to 5.9% at end-2023. This revised
shelter inflation forecast implies that core PCE inflation will stand at 2.9% at end-2023 (vs. 2.7% previously) and core CPI inflation at 3.2% at end-2023 (vs. 2.9% previously). This was picked up by none other than the Fed's mouthpiece Nick Timiraos earlier today:
But overall shelter inflation could continue to run higher because rents on renewed leases could continue to outpace rents on new leases due to "catch-up" effects.
— Nick Timiraos (@NickTimiraos) October 18, 2022
Goldman: "Our estimate of the speed of catch-up suggests that continuing lease rents will rise 6% in 2023." pic.twitter.com/XR5GwPxNyN
Finally, the bank also considers a downside scenario and an upside scenario. If new lease rents are instead flat next year - as they likely will be in a full-blown recession - the analysis implies that CPI shelter inflation would nevertheless still come in at 5% at end-2023. If instead the BLS new rent index fully closes the gap with the alternative data, which would also imply a wider gap between new and continuing lease rents, then Goldman expects 7% CPI shelter inflation at end-2023.
The bottom line from Goldman's economists is that its analysis implies that - in almost any case that does not involve a recession, which of course is laughable since even Bloomberg's own model now forecasts a 100% probability of recession in one year - CPI shelter inflation is likely to be high in 2023. On the other hand, in a recession, all bets are off, and both new leases and continuing will tumble.
Ironically, despite finally looking at rent inflation correctly, and finally disaggregating the lag effect, Goldman once again reaches the wrong conclusion, namely that inflation will come in hotter rather than cooler when using real-time market indicators, to wit:
Shelter inflation has an outsized weight in every measure of the underlying inflation trend that Fed officials watch, especially CPI-based measures, as shown in Exhibit 10. This is likely to keep these measures—core, trimmed mean, median, and sticky price— far above target next year, even if new lease rent growth remains at roughly the pre-pandemic rate and in fact even if it runs at 0% in 2023.
As for the implications for monetary policy, Goldman predictably takes the false conclusion and extends it to what the Fed should do:
While the FOMC has signaled an intention to hold the funds rate steady once it reaches a sufficiently restrictive level of 4.5-4.75%, we see some risk that ongoing firmness in these inflation measures could make hiking further than currently planned the path of least resistance. For that reason, we have noted in recent months that we see the risks to our peak funds rate forecast as tilted to the upside. However, substantial disagreement about the correct way to measure shelter inflation argues for looking at inflation measures that put less weight on shelter inflation, not more, when the decision is of greater consequence.
Goldman says this "ongoing firmness" in shelter inflation measures could make the Fed more likely to raise rates beyond the current terminal rate projection of 4.6%.
— Nick Timiraos (@NickTimiraos) October 18, 2022
"For that reason ... we see the risks to our peak funds rate forecasts as tilted to the upside."
Hilariously, since Goldman's explicit conclusion refutes Krugman's view, the NYT pet economist was most displeased, as the following thread reveals.
If you think that these lagged effects will in practice lead to further Fed tightening, well, I hope they're smarter than that 5/
— Paul Krugman (@paulkrugman) October 18, 2022
All of that, is of course, irrelevant, because what Goldman is actually trying to say is that since it still refuses to make a recession its base case, one should disregard its shelter inflation forecasts, and thus "put less weight" on shelter inflation as a whole. In reality, if and when the bank finally does change its base case to recession some time in the net few weeks, most likely after the midterm elections, watch as the bank's next shelter inflation forecast sees price growth tumbles to zero - and then sinking below it - as soon as the first quarter.
The full Goldman rent inflation note is available to pro subs here.