E-piphany

In terms of explicit protection against inflation, it is hard to beat a bond that pays cash flows that are expressly linked to the price level. However, there are many other asset classes that respond to changes in the price level and which, therefore, have at least some value as an inflation hedge. One asset class that many investors own quite a lot of, and is therefore important to look at, is real estate.

Residential Real Estate

There is both good news and bad news about residential real estate when it comes to being an inflation hedge. The good news is that your home can be a hedge even while it is also serving as a home, so this is one case in which your money really is “working for you!” Also, real estate has been a reliable hedge for inflation over very long periods of time, and home price “bubbles,” on both the positive and negative sides, have been generally small and rare: in the Great Depression, real estate departed from its long-term trend by about 40%, and in the bubble preceding the Global Financial Crisis real estate departed by nearly 50%. Other than those two periods, however, real estate has generally been within 20% of “fair value” for a century and a quarter at least.

The bad news is that residential real estate does not, and cannot, provide great performance relative to inflation. Over a long period of time, residential real estate will provide an after-inflation return of 0.25% – 0.50%. The chart below (source: Robert J Shiller, Irrational Exuberance, data through 2021Q1 available at http://www.irrationalexuberance.com) shows that real home prices until these latest bubbles (only the first of which, as of this writing, really looks like a bubble: see this blog post for an argument about why the second spike is less worrisome…so far) have grown about 0.25% since the late 19th century. But we have more information than that. As Dr. Shiller has pointed out elsewhere, we have reliable home resale records in old Amsterdam going back more than 500 years and so we know that in at least one extant case home prices haven’t significantly outpaced the price level.

But we can make a stronger statement than that. Measured properly, the resale price for any given home or set of homes cannot rise much faster than the price level interminably. Why? Think about the example of the Dutch homes which are 500 years old and which are worth roughly the same amount today as they were worth 500 years ago, in terms of the general price level. That is, you have to exchange roughly the same amount of goods and services today as you did 500 years ago in order to buy a house. But let’s suppose that, instead of returning approximately zero in real terms, that those houses had appreciated at a mere 1% per year over inflation. In that case, one of those Dutch houses today would cost about 144 times as much as they do. That is because 1%, compounded for 500 years, is a really, really big number: 14,377%, in fact.

That is too big of a number to be even remotely plausible. If home prices had risen even close to that much, no one could afford a home.

In fact, this general statement is true of any good or service. The price of any good or service, measured over a very long period of time, cannot rise very much differently from the general rate of inflation, because even a small difference measured over a long time causes that good or service to take over the entire consumption basket. The reason that some financial assets, like equities or bonds or arguably farmland, can outpace inflation by more is that they are productive assets. But a house is not productive – its only output is as a place to live, and that “dividend” cannot be stored and is fully consumed every year as it is produced.

So why do home prices rise at all in real terms? Some of this is probably measurement error. Even if you track the same house throughout time, it isn’t the same house in 2021 that it was in 1921. The homeowner has likely added indoor plumbing and electricity, for example, or extra rooms, or a better driveway. On the flip side, and more persistently, homes age – if you let that 1921 house sit and made no changes, the roof would have long ago fallen in. Economists try very hard to account for these “hedonic” adjustments that would otherwise make home prices seem to decline, once you adjust for the steady deterioration of existing homes.[1] But it’s probably most fair to say that home prices tend to keep up with inflation, and maybe provide a small increment over inflation, but it is hard to get rich just owning your own home.

That being said, most homeowners own their homes on a leveraged basis because they have borrowed via a fixed-rate mortgage. Moreover, the interest cost of that mortgage is – still as of this writing – tax advantaged because it can be deducted against income by those who itemize on their tax returns. Is that a reliable route to wealth?

Let us suppose that the cost of a fixed-rate mortgage today is 5% and that a home is 80% financed (borrowing, say, $160,000 of a $200,000 home) with $500/month in property taxes and insurance. If this homeowner has a 33% tax burden and home prices are expected to rise at the general level of expected inflation, then the homeowner/investor breaks even – ignoring the value of the home as a place to live – if this is true:

Rise in home equity = after-tax mortgage cost

Home value today * expected inflation = (1-tax rate) * 5% * financed amount + taxes + insurance

$200,000 * expected inflation = (1-0.33) * (5% * $160,000 + $500 * 12)

$200,000 * expected inflation = $9,380

Expected inflation = $9,380 / $200,000

Expected inflation = 4.69%

So, an indebted homeowner in this case wins as long as inflation is expected to rise at 4.69% or better. This makes sense, because while the homeowner has a leveraged investment in the home the cost of that leverage is not negligible. Without taxes and insurance, the breakeven inflation rate at which the house pays for itself would only be 1.69%, but most homeowners pay taxes and insurance! Owning a home is not a road to immediate wealth.

Owning a home is, however, an excellent forced-savings plan, and it is still an inflation hedge. That is, to the extent that a homeowner has a fixed-rate mortgage, accelerating inflation still leaves the homeowner better-off. Moreover, as time passes the deal gets better and better. In the example above, if the mortgage has been paid off then the breakeven inflation rate is only 2.01% – so, if the Fed successfully targets 2% inflation, then whatever you are paying in insurance and property taxes, after those are used to reduce your federal tax bill, essentially is added to your home’s value. It’s a pity that those gains might also be taxed when you sell the home.

On the plus side, you won’t pay rent so you will essentially have a fully-owned, solid inflation hedge and free living quarters. A road to wealth? Probably not, but not a bad deal.

Measuring Housing Inflation – Why the BLS Method Makes Sense

The foregoing discussion involves the value of real estate, and in particular residential real estate, as an asset. But many of the most animated discussions about the measurement of inflation itself revolves around how we should measure the cost of housing as a consumption item, and many of the serious misunderstandings about the CPI have root in the question of why the BLS changed its method of accounting for housing inflation in the early 1980s.[2]

Feel free to skip this section if you are only interested in the investment question and not the measurement question.

Here is the issue: as we discussed above, when you buy a home you are getting two things that are intimately bundled together. First, and most important, you are getting a place to live. Second, you are getting an investment. If we don’t separate these two things, then when home prices are soaring the “cost” of living in a home is less than zero because the rise in the value of the home will outweigh whatever it cost you to live there. And in housing busts, your cost of shelter will skyrocket. Does that make sense? Let’s consider two equivalent situations (and we ignore taxes and other frictions for the purposes of this illustration). First, let’s suppose you live in a home that you own, like the situation we just described.

And let’s compare that to the following situation. Let’s suppose that instead of living in the home you own, you rented it to someone else for five years, and meanwhile you took that rent and rented another place for your own family to live in during those five years.

Note that you are in the same position you were in before: you have lived somewhere for five years, and you have had an investment that seasoned for five years. In this case, though, you can easily see two sources of return/value. First, you have a rental cash flow, that equals your cost of consuming shelter; second, you have an investment in a home where your wealth goes up and down with home prices. The only difference is that you have unbundled the housing transaction.

That is what the Bureau of Labor Statistics (BLS) essentially does now when they measure owner-occupied housing, and that method has been basically unchanged for nearly four decades. There are some countries that treat housing differently, but in my view this is the most sensible approach. If you are interested in the various things that the BLS has considered, and has implemented in the past, and an explanation of why they prefer one method over another, read “Treatment of Owner-Occupied Housing in the CPI” by Poole, Ptacek, and Verbrugge. But, in a nutshell, the BLS says: whether you live in a home or in an apartment, you are paying rent and that is the “consumption” cost of housing that you cannot avoid. When you pay rent, you pay it to the owner of the asset (the property). The only difference when you own the home or the apartment is that the owner of the asset is you. You don’t actually write yourself a check, because that would be stupid, but that’s essentially what you’re doing.

In the “inflation breakeven” example above, I said you would have “free living quarters.” That’s one way to look at it, but the other way to look at it is to say that the breakeven on the asset is reduced by the amount of the rent you would otherwise have had to pay. Your own personal accounting determines how you look at the transaction, but the BLS says you’re paying rent to yourself…and if you weren’t, you would have to be paying it to someone else so either way, we should account for the economic reality.

Other Types of Real Estate – Comment

In the first section of this letter, I told a mild fib.

What I wrote was “…a house is not productive – its only output is as a place to live, and that “dividend” cannot be stored and is fully consumed every year as it is produced.”

As I’ve just illustrated, that isn’t quite true because in a sense you are paying rent to yourself. The value of the house clearly shouldn’t be different if you’re paying rent to yourself, or if you let the house to someone else who pays you rent. If the value was higher in the latter case, everyone would buy houses and rent them to other people, and then live in a house themselves that they rented from someone else. No one would want to own an “unproductive” house that was merely for their own use! Obviously, that is absurd – and so we can see that a house is “productive” real estate in the same sense as rental property is (however, the tax consequences may be very different, and we are abstracting from that here. In many cases there may indeed be locations where rental property is simply so tax-advantaged that no one buys a home to live in, or vice-versa).

In fact, almost all real estate is “productive” in the sense that it produces value in real time, and that product is consumed in some way. If it were not so, then valuing real estate would be very difficult since we naturally value cash flows – real or nominal – rather than flows of services.

Which type of real estate, then, is the “best” investment? In a frictionless economy, it wouldn’t make any difference what use the land was being put to. If farmland was more valuable than timber, then investors would buy forests and turn them into farms – and vice-versa. In this sort of economy, where real estate use was completely flexible, we could argue that all real estate should have essentially the same expected rate of return. The only value difference would come from location, but since location doesn’t change for a given piece of real estate the expected returns would be uniform across all locations and types of real estate.

In reality, there are substantial impediments to changing land use – zoning ordinances, for example – and different tax consequences for different uses of land. It is plausible, then, that timberland might have a higher expected return because paper use is skyrocketing but we can only grow more trees just so fast, or a certain parcel of commercial real estate may have value because the area around it is being re-zoned or is anticipated to grow more rapidly in the future for other reasons. But in each of these cases, the source of excess returns is not the real estate, but rather the restrictions on transitioning a particular piece of real estate into something else.

As a result, we cannot say that farmland, or timberland, or commercial real estate, or residential real estate offers inherently better returns at any given time. It depends on how the details of land ownership, taxation, and use are changing and are expected to change in the future. The underlying real estate, though, is a simple real asset and should not dramatically appreciate, in real terms, over long periods of time.

---

[1] So you see: “hedonic” adjustment isn’t some crafty, secretive way that the BLS pushed the CPI readings lower than they should be. In this case, hedonic adjustment pushes the CPI higher than it would otherwise be. And, in fact, this effect is roughly equal in size and opposite in sign to those other, nasty, hedonic adjustments you read about from hyperventilating non-experts in inflation…so that hedonic adjustments in aggregate cause no significant change in CPI over time.

[2] One conspiracy-laden website claims that the “real” CPI calculated using pre-1982 methods would be dramatically higher. Not only is this wrong, and demonstrably bad math, it ignores the question of how housing should be accounted for in the consumption basket.