I’ve written before about the ways homeownership in the United States is heavily subsidized by the federal government, at the expense of current and future taxpayers:
- preferential tax treatment of capital gains on primary residences, with $250,000 or $500,000 (depending on filing status) of a home’s appreciated value being completely tax free;
- the tax deductibility of mortgage interest (on the first $750,000 of a home’s value for new mortgages);
- a federally backed system of securitization which ensures liquidity for mortgage backed securities and encourages banks to issue mortgages while taking on virtually no risk themselves;
- and the exclusion from taxable income of “imputed rent,” the amount of value a homeowner receives by occupying a dwelling they also own, instead of paying (taxable) rent to someone else.
Different policy makers, journalists, and think tanks focus on different elements of this system, but I don’t want to litigate any one piece of this policy puzzle. I’m merely pointing out that these policies create an enormous federal tax and regulatory subsidy for owner-occupied housing of all kinds.
What is missed in criticism (much of which I agree with) of this system is that it exists in order to encourage Americans to make a totally idiosyncratic bet, not on the value of land in the United States, or the trajectory of residential housing prices in the United States, but on the value of a specific parcel of land, structure, or condominium unit.
No matter how sure you are in a stock pick, commodity bet, or options strategy, you wouldn’t put 80-100% of your net worth into one of them without a significant amount of downside protection. Since we decided homeownership was an important goal of American economic policy, we decided to create an enormous subsidy to encourage people to make what would be, under any other circumstances, an extremely unfavorable bet.
Comparing idiosyncratic bets on real estate
I started to wonder: if you wanted to make an idiosyncratic bet on US real estate, and have somewhere to live, would you be better off buying a house and living rent-free in it, or buying the Vanguard Real Estate Index Fund (VGSIX) and paying your rent with distributions from the fund?
Both vehicles should provide access to the US real estate market and a stream of income. The mutual fund provides access to a diversified portfolio of US real estate investment companies and periodic dividend distributions, while the single-property option is a concentrated bet on a particular parcel and income in the form of imputed rent you’re (not) paying yourself.
To answer this question, first I pulled the price and distribution data of VGSIX back to January, 1997, and compared it to the Median Sales Price of Houses Sold for the United States (MSPUS) data available through the St. Louis Federal Reserve’s FRED project. Since the median sales price in January, 1997, was $145,000, I used that as the starting value of the mutual fund investment as well.
The two numbers we’re interested in are capital appreciation (the market value of the mutual fund or the house) and income distributions (the annual income received by owning the asset). Once you have those numbers, you can slice and dice them in a variety of interesting ways.
Why a real estate mutual fund?
Before I get into the numbers, you might be asking, why would you want to get your rent from a real estate mutual fund, instead of a diversified portfolio of stocks and bonds? Isn’t that an awfully concentrated bet on a particular sector, and wouldn’t it be better to diversify?
The answer is yes, which is precisely the point of this analysis: if you think a large sector-specific bet on real estate is too much concentration in your investment portfolio, you should be even more skeptical about a bet not just on the real estate sector, but on a particular unit in a particular building on a particular plot of land.
VGSIX and median home prices do track each other over long time periods
The first question we can ask is simple: over the entire time period, does the Vanguard Real Estate Index Fund actually provide access to the same asset class as individual homeownership?
An investment of $145,000 in the median US home in 1997 would be worth $305,125 in 2016, while the same investment in VGSIX would be worth roughly $309,474 (the average of the starting and ending balance in 2016). The maximum deviation was in 2009, when REIT prices bottomed out, while the median home price was more resilient.
VGSIX is much more volatile than median home prices
During the 20-year period I looked at, VGSIX experienced 3 minor and 2 major decreases in market value:
- Between January and December 2002, between January and December 2013, and between January 2015 and December 2015, VGSIX dropped up to 3%.
- Between January 1998 and December 1999, the investment in VGSIX dropped 22%.
- And between January 2007 and December 2008, VGSIX dropped over 52%
Meanwhile, the median home price only fell once, between 2007 and 2009, when it dropped from $244,950 to $215,650, a decline of 12%.
VGSIX distributions fall more often, but by less
Remember, the point of this comparison is to look at the possibility of using real estate distributions to pay rent. That means the volatility of mutual fund distributions matters more than the volatility of the fund’s price. During the 20 years I looked at, VGSIX distributions fell in 8 years. However, the peak-to-trough drawdowns were relatively modest, except during the global financial crisis.
- Between 1998 and 1999, distributions fell 4% before recovering in 2000 to above their 1998 level;
- Between 2000 and 2002, distributions fell 7%;
- Between 2005 and 2010, distributions fell a total of 48%, with year-on-year drops between 3% and 35%.
The risk of the strategy, then, comes from experiencing a large decline in distributions after anchoring your expectations to a particular value. Given that VGSIX has experienced a peak-to-trough fall in distributions of 48% in just the last 20 years, you should be prepared to withstand a drawdown of at least that much in the money you have available to pay rent.
Case study: what will the median home price buy you today?
In 2017 VGSIX paid out $1.13 per share in distributions (including dividends and return of capital).
In the 4th quarter of 2016, the median home price was $310,900, which would have bought 11,305 shares of VGSIX, distributing (in 2017) $12,775, or about $1,065 per month.
Were that to be reduced by 48% to just 59 cents per share, however, you would only left with $556 per month to pay your landlord. Of course, renters also have the luxury of following prices down, so if radical cuts to dividend distributions reduce your spending power, as a renter you’d have the option of moving to a more affordable unit or location, or renegotiating your rent. Likewise as distributions increased you’d have more money available to move to a more expensive location.
Taxes, liquidity, distribution, leverage, and timing
There are some limitations and nuances to this kind of analysis, so let’s take a quick look at them, if for no other reason than to abbreviate the arguments in the comments section.
First of all, the issue of taxes. The United States has a fairly curious system of taxation whereby the owners, rather than the occupants, of real estate pay taxes on it. I call this curious because it means a homeowner has to personally cut a check to the city or state every year, while a renter usually has no idea what portion of their rent is going to their landlord as income and which portion is going to pay property taxes. In states with limits on property tax increases, two tenants paying the same amount in rent may have totally different allocations of that rent between their landlord and the state, depending on how long the landlord has owned the property.
Meanwhile, a tenant paying rent with mutual fund distributions pays capital gains taxes, decreasing the amount of taxable distributions that can be spent on rent, while an owner-occupant receives the imputed rent of the property tax-free. This is an important nuance to be aware of, but is too dependent on local tax policy for me to provide any general insight. I suspect the value of tax-free imputed rent is somewhat higher than any potential benefit to a tenant of avoiding property taxes, but that’s an empirical question I don’t know the answer to.
Second, the issue of liquidity cuts strongly in favor of the mutual fund owner. I used market values in this analysis but the term “market value” means something very different in the two cases: in the case of the mutual fund, it’s the actual amount you would receive for selling your shares on any day the stock markets are open. In the case of the median home, it’s the price that home sold for after days, weeks, or months sitting on the market and before paying fees to one or more real estate brokers.
Third, that brings me to the question of the distribution of home prices. I used the median home price in each year, which is the price above which and below which 50% of homes sold at. That is not, however, the price of the same house, because the distribution of property values shifts over time around the country. While the median home price in 1997 was $145,000, and the median home price in 2016 was $305,125, the median house in the first case might be in Illinois and in the second in Arizona. VGSIX did a good job of tracking median home prices over the 20-year period, but your particular home is virtually certain to deviate from the median by more than VGSIX did — either by appreciating more than the median, or failing to keep up with it.
Fourth, leverage is another area where the homebuyer has a key advantage: due to the federal system of subsidies, you can buy $145,000 in housing for just $29,000. It’s true you could also use leverage to buy VGSIX (or, more easily, VNQ, the exchange-traded version of the fund), but you’d find yourself paying higher, non-deductible interest rates, and be subject to margin calls should the fund’s value drop enough to leave you underwater. By contrast, as long as you keep making your mortgage payments, you can stay in an underwater home for as long as you’d like.
Finally, there’s the issue of timing. The key feature of the mutual fund strategy is that you can move without selling. Just take your rent budget and spend it somewhere else. Homeownership means that in order to liberate the imputed rent you’ve been using to live on, you have to sell your home entirely, at whatever price you’re able to get, and then make the decision whether to rent or buy all over again.
In the absence of the enormously expensive regime of subsidies provided to owner-occupied housing, I believe the financial advantages of mutual fund investing would swamp those of homeownership:
- if imputed rent were taxed as ordinary income the way other rents are, then homeowners would have to more carefully consider if they’re getting as much value from their homes as a potential renter would;
- if mortgage interest were not tax deductible, after-tax mortgage interest rates would be more closely aligned with rates charged on other kinds of secured loans, making leveraged housing purchases less attractive compared to other kinds of debt;
- if a federally-backed system of securitization didn’t exist, banks would be less willing to make mortgage loans to marginal buyers, requiring shorter terms, variable interest rates, or higher down payments;
- if capital gains on residences were taxed the same as gains on other capital assets, there would be less incentive to use housing as a form of tax-advantaged savings account.
Without those benefits, the idea of making a leveraged bet on residential real estate in a particular time and place would make as much sense as making a leveraged bet on the price of Apple stock, pork bellies, or bitcoin.
However, given the existence of those benefits, the picture becomes much murkier and almost completely contingent on the specific buyer, location, and property.