• riccardo

About investing in houses

Updated: Oct 6, 2021

#investing #savings #realEstate #decisionMaking #riskManagement



 

A couple of years ago I wrote a broad overview on the price of houses and their correlation with inflation and interest rates. The objective was to understand how houses perform as long-term investment compared to other saving alternatives. Last week I went back looking at that post and I was not satisfied with my original analysis; I thought about writing a new one. While working on this new version I realized that a couple of concepts from the analysis could help us evaluate our current situation in 2020, it will be covered at the end. I somehow speculate here on concepts and events because I think it is a topic particularly suiting “what-if” reasoning, which makes of this post something closer to an essay. I would encourage the reader to go through the writing by following the main concepts rather than trying to get any single number during the first reading. The final takeaways will likely be able to connect the ideas expressed throughout the post.


I will open with a long-term view covering the last 60 years, and follow with a short-term one focusing on specific moments. How quantitative people might know, while the average improves by using longer periods, volatility (deviation) requires a higher sampling frequency. Simply said, we can get a better sense of the possible swings by looking at what happens in the middle.


Underlying scenario


The value of houses, and assets in general, are arguably affected by inflation and interest rates. Without having any forecasting claims on the last two parameters, we will try to understand how they evolved in time and how that correlates with the price of houses. One common way to look at inflation and rates is the following: when the former rises above specific thresholds, the latter are raised as well [by central banks] to slow inflation down before it reaches excessive levels. Higher rates cannot only slow down inflation, they can also push down the price of assets like houses and stocks. With higher rates, investors would require cheaper price-tags looking for higher returns (e.g. same rent collected on a house requires a cheaper acquisition price to yield a higher Return On the Investment). That higher required ROI would follow the comparable higher rates investors could alternatively get by simply purchasing new 10-yrs treasuries bearing higher coupons (i.e. paying higher rates). We could use a simple picture to summarize this speculation (Figure 1).


Figure 1

It might be interesting to add to this discussion some indicator of the level of debt outstanding over time. The interested reader can refer the Federal Reserve as a good online resource for that.


We can now look at a snapshot of inflation and interest rates in the last few decades. Government bonds (10-year US treasuries) are used as proxy for interest rates because of the long-term characteristic and the availability of the data.


Figure 2

As we can see in Figure 2, inflation and rates raised across the 70s and peaked in the early 80s, possibly as a response to inflation reaching its maximum during the late 70s. As inflation decreases then, rates are lowered as well.


House numbers


After we went over the underlying scenario constituted by inflation and interest rates, now we will add the price of houses in time. We will use US values for coherence with the US treasuries mentioned above.


Table 1 - sources at the end of the post

Figure 3

Numbers in Table 1 are used to build Figure 4 below showing nominal prices (what an investor would be asked to pay back then) and the inflation-adjusted values in 2020 (historical nominal prices translated into 2010s decade’s dollar-terms) – 2010s’ values are not exactly the same because there are 5 years of compounded inflation from the middle of the decade.


Figure 4

What Figure 4 tells us is that a house bought during the 50s for $11.9k should be thought during the 2010s as something purchased for about $106k, therefore potentially yielding a +122% real return “above-inflation” if sold 60 years later for about $235k. Those numbers imply a pre-tax real compounded annual return of houses of about 1.34% - that is, if we started with $106k in 1950 and we compounded that at 1.34% per year, we would end up with $235k in 60 years.


We can compare that real 1.34% return realized by houses in 60 years with the one of the US stock market (e.g. S&P500), arguably a comparable alternative for savings. The compounding annual average return of the market is almost 7% in the last 60 years - see the historical prices of the ticker ^GSPC on Yahoo finance. Assuming about 2-3 percentage points in costs and fees, the return decreases to about 5%. Through Table 1 it is possible to calculate the average compounding inflation across 60 years being 3.7%, which subtracted from 5% (approximate method) would yield 1.3% real average compounding return of the market - comparable to the 1.34% of houses. Readers should realize that, as we have neglected taxes in all calculations, there could be other hidden costs not considered here in any way (e.g. maintenance and illiquidity for houses, holding fees for financial products following the stock-market).


In terms of capital gain, houses seem to provide a comparable good investment across long periods, we want to focus now on specific decades. This view might be of more interest to possible savers, and it could also allow us to apply these thoughts to our current years. We will build a picture of the 70s, 80s, and 90s, which as we saw above were periods with big swings in inflation and rates. If we want to understand how a house bought during the 70s would perform if sold during the 80s and 90s, we need two different inflation-adjusted values: during the 80s in 80s’ dollar terms and during the 90s in 90s’ dollar terms. Table 2 below presents a few possible scenarios (calculated through data in Table 1); we can focus directly on the last three rows with the final inflation-adjusted results.


Table 2: Nominal and Inflation-adjusted returns of houses
Table 3: Nominal appreciation of houses

Anything purchased during the 60s and 70s, and sold during the 80s and 90s obtains a good return in real terms. That was in big part due to the exceptional nominal appreciation of the 80s (119%, table 3), possibly fueled by decreasing rates over the decade, and able to more than offset the inflation of the period. A sale during the 70s would not allow for a good real return (0.4%) because of the rising inflation reaching the highest level across all decades considered here, and eating into the nominal appreciation (65%). Finally, houses bought during the 80s when the nominal appreciation was pushing prices up – in conjunction and possibly due to decreasing rates - do not allow for a good real return if sold during the 90s (-0.5%).


We can once again compare those results with the stock market. Table 4 below presents the nominal and the inflation-adjusted results computed as before.


Table 4: Nominal and Inflation-adjusted returns of the S&P500

The positive real returns (inflation-adjusted) of the stock-market are more concentrated toward the 90s. Both the magnitude of positive and negative returns is bigger than the one of houses, suggesting higher volatility of the stock market.


Operating return


Anybody who went through the evaluation of an investment, in general, knows that the final value is made of an initial investment and a final sale constituting the capital gain, and an additional return from the cash-flows that the asset returns in time during the investment. So far in our discussion, we have considered only the capital-gain contribution, which means we focused on price-tags going up or down. We have not considered the actual operating return of houses [and stocks] as assets. To conduct a complete analysis of returns we should include cash-flows in time from rent and similar [and stock earnings and dividends]. Focusing on our current years, we can picture a reference of 3% per year in lease [and possibly in stock dividends]. That number could represent a house asking $850/month in rent, or about $10k/year, and costing about $300k [and a stock paying out about 3% per year of its selling price in dividends]. This is a crucial component of the final return, and more importantly, it is maybe the number we can use to understand where we are now with the housing market.


We can now put everything together, but before summarizing everything in our takeaways, an additional note:


It is important to remember that we had a focus on US prices and markets; while that could be extrapolated to other countries, especially western ones, differences may apply. Moreover, considerations on volatility are obviously excluding the 2008-crash involving both houses and stocks. This writing does not want to deal explicitly with that singularity, however, excluding it completely could falsify this writing since those kinds of moments can make or break any investment endeavor. We will connect therefore to that argument toward the end.


Takeaways


Some general results that we could extrapolate from the numbers above:

  1. Across the last 60 years investing in houses seems to return a pre-tax compounded annual real return of about 1%, similarly to the stock-market - we already mentioned some possible additional expenses excluded here

  2. Houses show milder volatility than the stock market

  3. It could be interesting to speculate further whether lower volatility relates to better shielding provided by houses against inflation, and in general, against the volatility of rates (the 70s and 80s)

From our results above on capital-gain, it seems houses do have in general some flavor of an “all-weather” investment over the long term. That could imply that houses do not provide an exceptional return, but a robust one able to preserve somehow the value of our money in many situations. However, the real returns of Table 4 show that there are still periods like the 80s having strong nominal appreciation (higher purchasing price-tags) which make it tough to get a good return from houses. That is particularly important in our current days where we might still have some extra appreciation of the 2008 bubble - in part possibly sustained by the extremely low rates keeping asset prices high. A comparison in time between the return from cash-flows like leases and rents (return = annualized cash-flows from rent / purchasing price), and additional considerations on possible differences in inflation and rates, could help us evaluate the current situation. If we were evaluating an investment right now, we could compare that 3% above with the values of the past - we should research a bit the possible rent in the past for our type of house and location and divide it by the historical nominal prices from Table 1, or the exact one if we can find it). Nominal terms are fine because we are computing percentages not needing to be actualized. Adding then a second comparison of how much that 3% constitutes a premium over inflation and rates now and then, we could finally get a better understanding of the current market and, most importantly, our exposure to possible events. Please note, we are not implying here any forecasting nor timing of the market (arguably impossible), this is about trying to answer the question: “what kind of deal and exposure am I getting right now by buying this house?”.


While the core of the analysis above gave us a general understanding of the possible robustness of the capital-gain component of houses, the last view on the operating return could help us understand whether we are entering the investment well positioned to capture that. Simply stated, how much we preserve of that robustness might be about not overpaying at the entrance into the investment. Much of the measure of the over-payment might be inside that value we assumed above to be 3%.



 

This post does not constitute in any way investment advice


 

Some resources used for the data of the post.

Inflation:

Prices of houses (note: I did not use directly this inflation-adjusted values but calculated myself to keep the calculation coherent with a unique methodology):


 

Main image's tag: