Introduction
.
I am super interested in the direction of the housing market. Are you? You should be.
.
Housing normally contributes to a significant portion of one’s net worth. The Federal Reserve reported that housing assets were over $25 trillion in 1Q 2018, or about a quarter of overall assets, but it is even a higher proportion of median net worth (42% in 2002 according to the US Census Bureau). The NAHB notes that housing makes up about 17-18% of GDP through residential investment and consumption of housing services (rents, utility payments, etc.). However, since housing is volatile (see figures 1 and 2), it may have an even greater impact on GDP growth. When people purchase a new home or existing residence, they buy associated goods (furniture, lawn mowers, etc.) and employ homebuilders and people who make the air conditioners and other whatnot for the home. These wage-earners then spend their income (on hamburgers, cars, etc.) and influence the economy. Plus, changes in housing wealth correlates with changes in overall consumption (see this education piece by the Federal Reserve Bank of San Francisco).
.
Personally, when I purchased my home in mid-2004, I complained to my wife weekly about being ripped off. I stood in the cul de sac and told all of my neighbors how we were all taken and the housing market would fall. The next year, I visited a parade of homes to check on prices. I was happy to see that prices continued to rise as that gave me more of a cushion for the eventual market collapse. Then, my complaints about what I paid were only once per month. In the summer of 2006, I met my realtor who facilitated the sale of my home at the farmer’s market on the Wisconsin Capital Square. We have one of the best farmer’s markets, but the conversation turned sour even after eating my sweet maple cream whoopy pie. My realtor is a really nice guy, but he was caught up, like about everyone else, in the trend and we disagreed about how housing could fall on a national basis and it was a bubble. Then the crash came. My only solace is that I could say, “I told you so,” and this makes a nice story to tell my students. I also timed the recession with my mortgage refinancing. I had a low teaser rate that reset five years later to a variable rate. I expected a recession and lower fixed rates at the reset date, and I got that call right as well and refinanced at a lower fixed rate.
.
I live 86 miles from work. I plan on selling my home, to move closer to work, before the next recession if it comes prior to my son finishing high school. I hope housing prices fall post my sale since I plan to rent for a year and buy my next home for a good deal (to make up for my bad deal in 2004). Although, if you look at figure 2, I may be disappointed because housing prices for existing homes do not always fall in a recession. Still, the downtrend in prices in a recession is lower with each cycle, perhaps because inflation rates have declined since the 1970s. Thus, I watch housing closely, and recent weakness caught my attention, and it should interest you as well.
.
.
The purpose of this story, other than I like telling it, is that you should care about housing and listen to me as I have a keen interest in it and got the last call on the bubble and ensuing correction correct.
.
Data over the last few months suggests there are some cracks in the foundation of the housing market. Specifically, new home prices fell (figure 3); although, this is a volatile series and the six-month average gain is still positive (2.3%). Second, sales of existing homes have been down for several recent months (figure 4). This could be a cause for alarm, but this could also just be normal fluctuations.
.
Only during recessions do new home prices fall on a sustained basis (figure 5); however, there are plenty of periods when the one-month annual change is negative outside of recessions (figure 6). The same can be said for existing home sales that sometimes decline outside of recessions (figure 7). Plus, annual growth in housing starts (shipments) and existing home prices is still well in the positive range (figure 8). Both events – the decline in new home prices and existing home sales – are a little scary as sustained falling or weak price growth and home sales happens in every recession (figures 1-2).
.
.
The Good and Bad News
.
While there are cracks in the foundation, it may not be a sign of a bigger housing problem (GOOD NEWS). Or, yet again, maybe it is (BAD NEWS). I present both sides of the issue below.
.
(GOOD NEWS)
.
Some people claim the reason for slumping home sales is that home inventories have been declining for several years and there just are not enough homes on the market. Figure 9 shows that inventory and sales are inversely related, so I am not sure if I “buy” this theory of the rationale for low home sales (and purchases).
.
(BAD NEWS)
.
Inventories are finally rising (up 4% over twelve months through June), and inventories are negatively related to housing prices (figure 10). This makes sense if the seller and buyers are smart. Inventories rise when the seller can get a good deal, but sales could slump if the buyer is unable to afford the price (prices are back to all-time highs) (figures 11 and 12).
.
(GOOD NEWS)
.
Yet, maybe home sales can rise further. After-all, housing starts are still significantly below their past peak (figure 11) and so are existing home sales (figure 12). Although, remember that the recent past high represented a bubble.
.
(BAD NEWS)
.
I estimate replacement demand for new housing units to be about 700,000 (figure 13), and we are currently running at over 1 million.
.
There are roughly 325 million people in the US and population grows at about 0.70% to 1.00%. Let’s assume 0.75% growth as recent data suggests 0.70% as the US population is aging and growth has slowed. If there are 2.5 dwellers per household (this has been declining over time) and 65% own a single-family home, we have a need for 634,000 new single-family homes each year. Add a little for replacement demand (natural disaster and older homes that are abandoned) and we get to a demand of about 700,000.
.
.
(GOOD NEWS)
.
My number (700,000) is probably not too far off for long-term demand, but the short-term is being boosted by some upward forces.
.
Actual household formations are running a little higher than my estimate of just under 1 million (figure 14). The latest number jumped to 1.2 million (twelve-month average). Unemployment is very low which means young people will be leaving their parents’ basements in greater numbers. According to Pew Research Center, in 2017, about 31.9% of the adult population lived in a shared household, up from 27.4% before the financial crisis. This means that there is possible significant pent up demand for new households.- Home ownership rates fell significantly since 2004, but it could be rising. I assume 65% of households own a home, a little higher than today (figure 14), but every percent raises the number of new homes needed by 1.3 million (1% multiplied by 130 million households) and the past peak was 69%..
.
.
(BAD NEWS)
.
However, rising housing prices may price many young people out of a home. Figure 16 shows that housing affordability is falling, and it leads housing sales (six-month average) by eight months (leads starts by 10 months). Plus, interest rates are rising, which is negatively correlated with existing home sales with a six-month lead (figure 17) (it also correlates negatively with housing starts with the same lead).
.
(GOOD NEWS)
.
While affordability is declining, the financial obligation ratio (FOR) is still low. The FOR measures payments for mortgage, consumer, and other obligations (rents, auto, insurance, and property taxes) as a percent of disposable personal income. It peaked at over 18% in 2007, but it is now under 16% (figure 18). It is positively related to bankruptcy filings (figure 18), which is negatively related to housing starts (figure 19). FOR is still healthy, but it is moving in the wrong direction.
.
.
(BAD NEWS)
.
The current FOR is ok, but this is significantly due to interest rates being low and partly because personal income has recovered with the economy. The liabilities to GDP and to assets ratios have been rising since the late 1970s (figure 20) as interest rates declined (figure 21). Both ratios are down from peaks, but they are still not low versus the long-term averages. Consumers have cleaned up some liabilities and GDP and assets prices have fully recovered from crisis lows. Liabilities to assets is down more than liabilities to GDP, but assets are inflated at the later stages of the cycle (see Positioning the Cycle and Tactical Allocation Over the Cycle) and due to quantitative easing. Thank goodness for lower inflation and lower interest rates or the FOR would be higher! Uh oh, rates are rising.
.
.
(GOOD NEWS)
.
Rates would have to rise a lot to bring the FOR back to pre-crisis levels. Let’s do the math. We can back out the rate using the mortgage service ratio (MSR) and current rate on 30-year mortgages. I realize all mortgages are not 30-year fixed, but please allow me this simplification as these are just rough guesses.
.
The MSR is about 4.4% right now. If rates rise 1% to 5.8% (about 21% of current rates (4.8%)), this would make this ratio rise by the same amount (21% of 4.4% is 0.9%). The MSR is part of the FOR, so this would raise the FOR to 16.7% (up from 15.8%). The peak was 18.1% in December 2017.
.
To get to 18.1%, 30-year mortgage rates would have to rise 54%, from 4.8% to 7.3%, assuming all else is equal. However, all else will not be equal as in this scenario we can assume the cost of other debt, rents, etc. would also rise. Thus, to get back to crisis levels of obligations as a percent of disposable income, mortgage rates may have to increase to 6%+ (mid-point of 4.8% and 7.3%). I doubt anyone is forecasting this and I am not as well, but it is an outlier event that one should consider.
.
(GOOD NEWS, BUT ALSO BAD NEWS)
.
Consumer confidence is high and changes in the level is correlated with housing starts (figure 22) and existing home sales. However, the rate of growth is slowing, and confidence is back to nearly all-time highs (figure 23).
.
.
What Does the Market Think?
.
Housing stocks agree with me – there are cracks in the foundation of housing. The industry has underperformed the S&P 500 by 21% from highs reached at the end of 2017 and by 2.5% over the last twelve months through July (blue line in figures 24-9).
.
Housing stocks lead new home prices by 12 months (figure 24), and given the underperformance, housing investors are not pleased with the direction of the recent numbers. They also lead affordability by six months (figure 25), and affordability is down. They have an inverse correlation with inventories, which are finally up (4%) on an annual basis through June (figure 26). Slowing housing starts and home sales have not helped (figures 27 and 28), and neither has rising interest rates (figure 29).
.
Each of these six variables are pointing in the wrong direction for housing, and at least three of them (affordability, inventories, and interest rates) are probably not set to change direction. Perhaps new home prices, starts, and existing home sales could turn up, but as discussed earlier, this is far from certain.
.
.
Maybe the stocks have priced in a slowdown (not a recession) as the P/E is only 12.2 on trailing 12-month earnings (source: FactSet). Still, earnings can quickly drop to negative during recessions (figure 31) and the stocks get clobbered (figure 30).
.
.
Let’s assume current earnings are about this cycle’s high. Earnings are still well below the past peak, but so are housing starts. Let’s also assume that earnings drop to zero during the next recession, which is being kind since earnings dropped well below zero in the last. If the peak this time is lower than the last, the trough may also be higher. We can then conclude that normalized earnings are about half of the current level, which then means the P/E on normalized earnings is about 24 (two times 12.2 or 12.2 divided 50% of the current earnings).
.
Performing similar analysis for the S&P 500, but assuming earnings fall perhaps 30% versus a roughly 50% drop during the last recession, yields a current normalized P/E of roughly 28 (the S&P 500 is trading at 19.4 on trailing 12-month earnings which needs to be divided by 0.70 or 70% to determine the normalized P/E).
.
Therefore, housing trades at a discount to the market; it trades at a 24 normalized P/E versus 28 for the S&P 500. This means that investors are not entirely optimistic about housing. On the other hand, just a four-point discount on a mid-twenty P/E for a highly cyclical industry is probably not extreme at this later stage of the cycle (see Positioning the Cycle) and makes the industry perhaps too risky for my more cautious positioning right now (see Tactical Allocation Over the Cycle).