In Part 1 of the blog, we discussed the following:
The financial media is all over the place with a housing bubble call: a bubble is coming, a bubble is here, a bubble has come and gone. Consensus remains elusive.
Why current housing prices do not constitute a bubble, despite their pervasive narrative.
In Part 2 of this blog, we would like to focus on two important follow-ons:
Why bubbles happen and the effects post-bubble PTSD.
Why fundamentals justify most of the price appreciation in housing to date and why, looking forward, further gains should accrue.
Why bubbles happen and the effects post-bubble PTSD
Perhaps the foremost expert on financial bubbles is Charles Kindleberger via his bible on financial dislocations: Manias, Panics, and Crashes: A History of Financial Crises. In assessing the various bubbles and subsequent calamities from over the last few centuries, he notes similar patterns:
Excesses begin as a novel idea takes hold. Some examples from our recent include the idea of “clicks and eyeballs” as well as the assertion that “housing prices have never gone down en masse across the entire country.”
Credit and foreign capital inflows expand. This creates additional demand for the desired product relative to its supply. This begets even further expansions in both categories on the back of the newly-created money. Of course, as the quantity demanded rises faster than the supply, prices rise.
Capitalists jump into the market. Seeing the newly-formed demand for the product, capitalists take advantage of the price hikes, both on the demand side and on the supply side:
On the demand side, people purchase either with the intent of reselling at a higher price or bring forward their own demand curves because purchasing it today would be more cost-effective than waiting to buy it when the product is truly needed and prices are even higher.
On the supply side, manufacturers see an opportunity to accrue super-normal profits and take on capital to build more production facilities with the intent of satisfying the increased demand.
Prices continue to ascend because demand increases consistently to stay ahead of supply increases. It is at this point that a “new paradigm” takes hold, many justifications and rationalizations for the price increases coming to pass. These beget even further expansions in credit and eventual price increases as those inside the bubble on both the demand side and the supply side are blissfully unaware of their peril.
Demand eventually dries up because much of the demand was not organic demand but rather self-fulfilling demand banking on higher prices to beget further higher prices. This comes from combination of reduced credit and/or no more “greater fools” left in the pool of potential buyers. Additionally, market insiders, recognizing the current prices left value in the dust a while ago, begin to take their balls and go home, exacerbating demand and supply imbalances. At this point, the price ascent runs into an air pocket, morphing into a sharp decline as a swath of sellers, fueled by expansion in excess capacity, are greeted with a dearth of buyers.
There is a rush for the exits as the rest of the market finally catches onto the fraud that was the “new paradigm.” If the asset is more liquid (e.g., stocks), much the fall comes imminently, with some residual fall afterward. If the asset is less liquid (e.g., real estate), the fall takes much more time. Regardless of which, the feedback loop we saw on the way up (i.e., rising prices beget more demand, which beget even further price increases) reverses into an analogous dynamic on the way down: falling prices beget more desire to sell, begetting additional price depreciation.
With the benefit of hindsight, we can see that the internet bubble and housing bubble followed this exact script. So, what is the distinction between what occurred then and what is occurring now? There are two main issues: post-bubble PTSD and fundamentals.
Post-bubble PTSD happens after EVERY bubble. In essence, those burned by the bubble are constantly on guard for a repeat of the bubble, which effectively prevents the bubble from happening. This is why bubbles do not happen with regularity, and when bubbles reoccur, they are either with different assets or in a different geography. Some of the most famous bubbles such as Tulip mania and the South Sea and Mississippi bubbles occurred 300-400 years ago. Of more recent vintage, we have noteworthy examples (in chronological order): 1920s U.S. stock market bubble; late 1970s precious metals bubble; late 1980s Japanese economy, stock, and real estate bubble; late 1990s internet stock bubble; and the mid-2000s U.S. housing bubble.
As the historical pattern shows with each boom and subsequent bust, investors shun the prior area, whether that be asset class, geography, or both. They develop a bunker mentality for the asset that prompted the crisis; we all know the dynamic: once bitten, twice shy. Of course, the rational thing to do is to increase risk after crises, yet behaviorally it is such a hard thing to do. In fact, investors often go so far as to go in the opposite direction by betting against the crisis well after the crisis has ended. Think about all the investors who invested in hedge funds and tail-risk strategies well after the Global Financial Crisis ended (and in some cases continue to do so), which is the reverse of the optimal investment strategy. The lasting effects of the most recent crisis persistently cause investors to erroneously fight the last war rather than following the historical template of the next war, which ironically causes them to miss out on what generally end up being tremendous investment opportunities. We call this post-bubble PTSD, which is why it is so hard for a bubble to be re-blown.
A perfect example is the plethora of bubble calls for the technology sector in the early-to-mid-2010s:
Investors were still so traumatized from the internet stock bust on the heels of the 1990s bubble that the obvious call was that another tech bubble had developed. However, we can see in retrospect how that call went: the 2010s was a technology-dominated decade, causing many apprehensive investors to underweight the juggernaut.
We believe the current housing market and the underlying mentality is analogous. A sector that blew up less than 15 years ago begins to find its legs, and “pundits” are ready to make the “obvious” call that a new bubble is forming, especially because of a relatively short period in which the housing market was extremely hot. Thus, the Pavlovian call for a bubble owing to recency bias, a call which we referenced in Part 1 of the blog and which we are likely to see ad nauseum over the coming years as the housing market continues its ascent.
Why fundamentals justify most of the price appreciation to date and why, looking forward, further gains should accrue
This can be distilled primarily to one statistic, with two important secondary statistics: Housing starts, demographics, and mortgages.
Reduced Supply of Houses. In the wake of the overbuilding that occurred during the housing bubble, housing starts plunged (Figure 1).
Of course, this was necessary, as a housing surplus developed during both the ascent of the housing bubble (houses were being built for both owners who were destined not to be able to remain in them as well as speculators trying to make a quick buck) and during the decline (as the foreclosure glut mounted and then went through the process). A plunge in housing starts was just what the doctor ordered to clear the glut while builders and speculators who were left holding the proverbial bag cleared their inventory.
However, as with all pendulums, the housing glut swung too far, becoming a housing deficit as millennials who had theretofore been shacking with their parents were now on the prowl for housing as household formation took off. So, while the coronavirus exacerbated this dynamic with a mad scramble for housing, the demand-supply imbalance was already in place. This in not unlike what we saw in the oil market as oil prices plunged – rigs were taken offline as excess capacity became apparent in the wake of plunging oil demand.
Most people are unaware of exactly how few homes have been built in this country over the last decade or so. To put the figures from Figure 1 in perspective, from March 1977 through November 1978, we averaged ~1.5 million single-family housing starts per month when the U.S. population was around 220 million. In December 2020, the U.S. population was 330 million, yet the “bubble” of housing starts in the wake of the coronavirus scramble was merely 1.3 million - 50% more people yet fewer housing starts! And to take things to a real extreme, housing starts in March 2009 were just 353,000, which compares to 1.3 million built 50 years earlier when the U.S. population of 175 million was merely half the March 2009 population. Put differently, at its most extreme, 75% fewer homes were being built for twice the population!
Figure 1. 60+ Years of U.S. Housing Starts Show the Last Decade Plus have defied historical precedent
Source: St. Louis Federal Reserve
Increased Demand for Houses. Not only do we have reduced supply, but we have additional demand drivers. Importantly, these demand drivers are not ephemeral like the coronavirus bump. Rather U.S. demographics indicate that there is a strong pent-up need for housing. Figure 2 shows historical and projected age brackets. Ages are grouped as follows:
Green: Ages with likely strong demand for housing
Yellow: Ages with modest demand for housing or the potential beginning to scale out
Red: Ages with potential for selling housing
Black: Ages with little to no direct effect on the housing market
As we progress from the earlier dates to the later dates, we can see a growing number of age brackets with higher housing demand growing as a percentage of the U.S. population. While many assumed that millennials scarred by the Global Financial Crisis would live in apartments or with their parents ad infinitum, 2020 data suggest the process of millennial homebuying is well underway. And if what we have seen of late is indicative of the future trend (excluding the effect of the coronavirus bounce), we are going to need a lot more houses in the country, a spigot that will take many years to turn on in its full effect.
Moreover, as the home-buying groups grow, we are not seeing a large growth in the Baby Boomer cohort, which would likely be sellers (i.e., the groups in red). While it’s possible that some inventory may come onto the market late in the 2020s and into the 2030s, this is by no means a guarantee and would likely take time to play out. So, the idea that there is going to be a huge flood of supply to hit the market imminently is highly unlikely.
Combining the effects of these two dynamics, simple supply and demand dynamics would dictate that housing prices should not only be rising now but also set for further gains in the future merely on the lower supply generated over the last 12-13 years.
Figure 2. The future of the housing market has strong demographic tailwinds, both present and future.
Source: U.S. Census Bureau
Differences in Mortgages. A key cog in the wheel that drove the housing bubble 15 years ago was the poor quality of mortgages. Option ARMs, NINJA loans, and other assorted subprime dreck led to an artificial increase in demand for housing. Today, things are much different. We would point to three components of today’s mortgages that make things different now than they were 15 years ago:
Mortgage quality is far superior, as shown in Figure 3. During the climax and denouement of the housing bubble, people with poor credit were originating more mortgages than people with strong credit. Now, however, the gap between high credit and low credit mortgages is higher than it has been in the last two decades.
Mortgage servicing is far less onerous to households, as shown in Figure 4. Far less disposable income is required to make mortgage payments, with current levels almost 40% lower than they were in the fourth quarter of 2007 (i.e., down from 13.2% to 8.2%). Of course, some of this is due to lower interest rates, but that doesn’t change the fact that servicing mortgages is easier than it has been in the forty years of data, despite the fact that housing prices are higher than they were at that time.Another important facet of this is that even if rates were to move higher, many homeowners have locked into long term payments at their current low interest rates. This is a strong contrast with 2007, when many originated interest-only loans that were due to reset as rates went higher.
Of course, none of this precludes this situation from eventually becoming a bubble. Although, history suggests that a repeat bubble will not be the case, we must acknowledge that is a possibility. However, the data point to no bubble at this time and nothing imminent.
Figure 3. Mortgage quality is far superior to that of 15 years ago
Source: New York Fed Consumer Credit Panel
Figure 4. Mortgage servicing is far less onerous to households