E-piphany

In the last couple of CPI reports, two trends have started to become apparent. One is that inflation is broadening, in that more categories are seeing accelerating inflation. The second is that shelter, the largest and slowest-moving subcategory, has been decelerating.

Is there an inflationary process going on, then? Or are we in a disinflationary or even deflationary period? The answer to this question depends a little bit on what we think the inflation process is, and a little bit on figuring out what is really happening in housing.

I originally wrote this as one long piece, but decided rather to split it into two. So today I’m going to talk about why I think the shelter component of CPI is probably not about to enter into an inflation-dampening slide; on Monday or Tuesday I’ll discuss the inflation process itself and the broadening of inflation.

I have long said that if you’re going to predict core deflation, you’d better have a reason to expect deflation in housing. And if you’re going to predict accelerating core inflation, you’d better have a reason to expect acceleration in housing. Shelter is the 800-pound elephant in CPI; it moves slowly and ponderously and if it sits in one place, you’re not going to move the broad index very far. And the story for the better part of two decades in inflation is largely this: shelter inflation has been consistent and in the 2%-4% range; everything else in core has been fairly consistent but lower.[1]

The exception, of course, was in 2009-10, when the housing implosion dragged down rents in a short but deep dip. But if you’ll look at the following chart, which splits core into Shelter and ex-Shelter components, you’ll see that nothing else really looked like deflation was in the offing. Indeed, if you take out shelter, core inflation barely slowed in the aftermath of the Global Financial Crisis (Keynesians, take note).

To get a real inflationary outcome, then, we’ll need two things. One of these is that shelter needs to not decline, but rather accelerate. And the other is that non-shelter inflation needs to pick up. None of this needs to happen tomorrow, but it needs to happen. So, when Keynesians see Shelter declining markedly, they become gleeful because it sure doesn’t look like the enormous deficit spending and explosive money growth is doing anything to cause inflation.

Not so fast! Let’s take a deeper look first. The three main components of Shelter are (a) Owner’s Equivalent Rent, which is ~73% of Shelter; (b) Rent of Primary Residence – that is, apartment renters – which is ~24% of Shelter; and (c) Lodging Away from Home, which is about 2.8% of Shelter. (The difference is rounding and a little bit of “Tenants’ and household insurance.”) The chart below shows the y/y changes in these three pieces, unweighted.

Clearly, Lodging Away from Home provides a ton of volatility, and that massive decline – and not much of a bounce – has deducted about a third of a percent from the year/year growth rate of the broad Shelter CPI index. Lodging Away from Home is clearly and directly impacted by a lack of travel, so to get a better idea of the general shelter trends let’s strip that away. The two main categories, then:

Not surprisingly, Primary Rents and OER track each other very well. This makes some people very angry. Some mathematically-challenged inflation-conspiracy “alternative stats” sites claim that the Bureau of Labor Statistics changed to the rental-equivalence method of computing shelter costs to “hide” the real inflation, which they say is really 5-6% per year higher than what is reported. This is obvious nonsense, and I hate to always have to debunk it because it’s so easy to do. The reason that the US uses rental-equivalence is that you can think of a house as providing two things to the owner: an investment, and a place to live. It’s important to separate these two functions of a house, and so the BLS reasons that the “place to live” part is in competition with other ways you can get a place to live without the “investment” part: in other words, a rental. This gets confusing at times when home values are rising faster or slower than rents. This happens because although the “owner’s equivalent rent” part of a home is in competition with “rent of primary residence” – and so these two series move together as the chart above shows – the investment value of the house has no analog if you’re a renter. So when home prices and rents diverge, it is best to think of the divergence being caused by the investment portion.

Incidentally, you can understand this equivalence by reflecting on the fact that these two situations have equivalent financial outcomes: (1) live in your owned home and (2) rent your owned home to someone and use that income to rent someone else’s owned home. In both cases you are living in a house, and get investment returns related to home prices. But in the second one you’re clearly paying rent, and you also own a home that earns you the change in home prices plus a stream of rental payments, even though the two rental streams in this case (the one you’re paying, and the one you’re receiving from your tenant) happen to cancel. That’s essentially what is happening, then, in (1) and it’s the reason the BLS looks at housing in that way.

So, how much difference does it make over time to use rents, rather than home prices, to measure the cost of “shelter services”? The chart below shows the median price of an existing home, compared with OER. Note that prior to the bubble era, rents and prices moved pretty close in lock-step…through the late 1990s, there was no difference – certainly not 5-6% per year! And even with the wild swings induced by the ‘environment of abundant liquidity’, median home prices have risen ~4.2% per year since 1982, compared with ~3.2% for OER. One whole percent per year, which means the dastardly BLS has understated Shelter costs by 0.7% per year, and overall CPI by roughly 0.2% per year! Oh, and as lately as 2013 the difference was exactly zero percent per year. This seems like a conspiracy theory in search of a conspiracy.

Back to the story we came here to discuss. Why are rents decelerating, and is there reason to think that will continue?

In some parts of the country, notably in dense metropolitan areas where taxes and violence are rising, rents have not only slowed their rise but are actually in decline as renters can easily pick up and rent elsewhere. But elsewhere, rents are showing normal trends. The following picture, sourced from https://www.apartmentlist.com/research/national-rent-data, illustrates this phenomenon. New York, as President Trump recently noted, really is becoming a ghost town and rents are falling briskly in L.A., San Francisco, Boston, Washington, Minneapolis, Seattle, and Chicago and more mildly in Portland, Denver, Chicago, Houston, Dallas, and Philly. But they’re rising in lots of other places.

Also, rents in the CPI aren’t ‘asking rents.’ They are based on a survey of landlords, who report the rents they have actually received in the latest month plus any amounts they expect to receive in arrears. So if someone has paid rent to the landlord, that rent is included…but someone who is behind on the rent isn’t treated a zero. Instead, the landlord reports what amount of rent he/she expects to eventually receive. At times like the one we are currently living in, this can matter. While the government grants early in the year actually kept incomes surprisingly robust despite the lock-downs, those streams of money are drying up and with it, the potentially ability of some renters to make the rent. If landlords give a grace period to the renter, that doesn’t show up as a rent decline. But if the renter sneaks out the back window like George Thorogood,[2] it will. Nevertheless, at the moment the number of renters making full or partial payments is not too far below par, according to the NMHC Rent Tracker. However, if after a “blue wave” election the Congress declared a “rental holiday,” there would be a massive decline in the BLS rental figures. I give that a very low probability of happening, but it’s a risk to be aware of.

Overall, then, rents are softening mostly in the big cities and this is being reflected in Owners’ Equivalent Rent as well (for a good if wonky explanation of how the BLS measures Owners’ Equivalent Rent, see this publication on the BLS website). But how confident are we that this will continue? Notably, exactly the opposite thing is happening in home prices. Just this week, the National Association of Realtors reported that the median price paid in Existing Home Sales rose 15.4% over last year – the fastest rise since 2005.

That’s remarkable, and I must say unexpected. Some will say that this is a sign of a wave of people moving to the suburbs; maybe that’s true. But while 10-20% per year home price appreciation going forward seems unlikely, it is also unlikely that home prices are going to fall very much since the inventory of homes available for sale is near the lowest level in many, many years; historically, low inventories lead home price appreciation by about 12 months.

Although rents are decelerating, at least looking backwards, I find it very hard to believe that can be sustained. Apartment rental and home ownership are, after all, substitutes to some degree, and while they sometimes diverge for a time rents almost always eventually follow prices. In the Global Financial Crisis, for example, the drop and eventual recovery in home prices led OER by about 18 months, which is why it was pretty easy at the time to know that October 2010 was going to be the low in core CPI, plus or minus a month or two.

Now, lagging home prices is one way to get rents, and as you can see it doesn’t really suggest that a collapse in OER is imminent. But at Enduring Investments we also use an income-based model since one might reasonably suspect that when people’s incomes decline they are less likely to be willing and able to pay high primary rents, and over time if the amount that people spend on shelter is reasonably stable then the rise in nominal incomes is likely to parallel the rise in nominal rents. And although it doesn’t fit the contours of rents as well as one based on home prices, the implication of that model is very interesting, thanks to the big jump in incomes due to government transfer payments.

If you look at home buyer traffic, and what’s happening to home prices, it doesn’t seem a big stretch to suggest that at least some of that is resulting from the income replacement schemes that have left some people actually better off than before the crisis. The income model actually suggests we could see a decent acceleration in rents in the latter half of 2021.

When we put our various models together, we get a more stable picture that suggests the recent dip in rents is overdone and not likely to be the start of a significant slide lower (again, assuming no rental holiday!); in fact, the 800-pound gorilla might begin moving north soon.

That’s all for today on housing. On Monday or Tuesday, I’ll have a much shorter piece discussing the broadening of price pressures, and how that factors into the outlook. Thanks for reading. Be sure to stop by Enduring Investments if you would like to start a dialogue on this topic and how we can help you manage inflation risks.

---

[1] And for the better part of this last decade, I’ve been writing periodic updates about housing inflation, which you can find on my blog in the “Housing” category at https://inflationguy.blog/category/economy/housing/. What’s amazing is how many times I say “recently, there has been some alarm about OER” or “some people have been saying housing inflation is about to head lower.” Seems a recurring theme!

[2] “So I go in my room, pack up my things and I go/ I slip on out the back door, down the streets I go/She a-howlin’ about the front rent/She’ll be lucky to get any back rent/She ain’t gonna get none of it.” – One Bourbon, One Scotch, One Beer as sung by George Thorogood & The Destroyers.