2020 was a surprisingly solid year for housing in light of the pandemic and the havoc that it wreaked in other parts of the economy. December data will not be reported until later this month. I project that single-family housing starts will total 971,700 units in 2020, up 9.5% over 2019 and new residential sales will be 804,000, up 17.9%. That is pretty remarkable, considering that single-family starts and sales were down double-digits in March and April.
The catalyst for housing’s plunge, its subsequent rebound and its continued strong performance has been the pandemic. Some of the strength represents a shift of sales from spring to summer by those delayed their purchases due to the pandemic. Many urban apartment dwellers have also sought to move to suburban single-family homes to reduce the risk of catching COVID-19 and obtain more space and creature comforts as they work from home.
The demographics are favorable for such a switch: the ranks of 25-29 year-olds and 30-34 year-olds – the prime home buying cohorts – have increased in recent years. The pandemic may therefore be pulling housing demand forward, as some buyers have acted earlier than originally planned (perhaps with the help of some down payment assistance from their parents). Still, housing has remained stronger for longer than I would have anticipated, since by now most people have probably adjusted to life under COVID-19.
The strength in housing sales is evident in both the national housing data and in the recent financial reports of publicly-traded homebuilders. This is occurring despite other economic and housing-related data that is not supportive of strength in housing.
At the very least, it is hard to conclude that this year’s strength in housing is sustainable. I will present all of (what I believe are) the relevant data points in the remainder of this report.
The recent strength in new home sales is evident in the chart below. Sales started the year off at a seasonally-adjusted annual rate (SAAR) of 774,000 units in January, the highest since July 2007. From there, it was straight downhill until sales bottomed at a SAAR of 570,000 units in April. They then soared to a peak of 979,000 units in July, but have since been gradually losing ground. The deceleration seemed to gather steam in November, with sales down 11% from October to a SAAR of 841,000 units, due perhaps in part to a second wave of COVID-19 infections that is now sweeping the country. Even so, at the November pace, sales were up 21% year-over-year.
Although housing sales turned positive in May, they continued to gather steam in June and July. Perhaps this caught some builders flat-footed. In October, the NAHB’s chief economist, David Dietz, noted that the gap between new home sales and single-family starts had reached an all-time high. That gap has since receded a bit as the pace of sales has slowed.
Mr. Dietz removes from his calculation the starts of houses that are not for sale. The chart on the previous page, which plots the SAAR of single-family homes under construction against the SAAR of single-family starts (at SAARs), shows that construction activity has only begun to ramp up over the past few months, even though sales have been on the rise since May.
Builders have apparently had trouble lifting production over the past several months to meet the rising demand. Early on, construction activity was hampered by the pandemic and protocols put in place to protect construction workers. Later, there were announcements of supply shortages – for labor, lumber and most recently, finished lots – that have hampered production increases and deliveries. Although sales are up sharply, single-family housing completions have been flat (on a rolling 12 month basis) for nearly an entire year, since before the pandemic began.
A deeper look at the new home sales data also highlights some of the difficulty that builders are having keeping up with sales. The months supply of homes available for sale has dropped precipitously in recent months, as show in the chart on below.
In addition, completed (or speculative) homes as a percent of total homes available for sale fell to 15% in November, about half of the average of 30% over the past 35 years. Builders have fewer spec homes available for sale. The percentage of available-for-sale homes that are under construction has risen proportionately. It seems that builders are struggling to keep pace.
Despite flat single-family housing completions, construction spending is on the rise. Yesterday, the Census Bureau reported that unadjusted monthly spending on new single-family production increased 1.9% in November from October. That marked the sixth consecutive monthly increase. Year-to-date, new home construction spending is up 6.4%. While some of the increase may represent dollars spent on pricier homes, the year-to-date increase in construction spending is above the 1.6% year-to-date increase in the average price of the average home sold. Thus, unit output is rising, so builders should begin catching up in the months ahead.
Some Economic and Demographic Data is Out of Sync with the Strength in Housing. The key drivers of housing demand have traditionally been employment and consumer confidence. Both are not supportive of the strength in housing that we have seen since the onset of the pandemic. Some key demographic data also present a conflicting picture of recent developments in the housing market.
The chart on the previous page shows the dramatic loss in employment and increase in the unemployment rate since the start of the pandemic. Year-to-date through November, the economy has lost a net 9.3 million jobs according to the payrolls survey and 9.1 million jobs according to the households survey. The official unemployment rate was 6.7% in November, down sharply from 14.7% in April, but nearly twice the 3.5% rate of a year ago.
In its November employment report, the Bureau of Labor Statistics reported that 14.8 million people were unable to work because their employer closed or lost business due to the pandemic. It estimates that 3.7 million people, or roughly 2.4% of the labor force, have permanently lost their jobs. Even with a speedy and effective rollout of coronavirus vaccines, it will likely take time to recover those permanent job losses. The Philly Fed forecasters predict that the unemployment rate will not fall below 6% until 2023.
It goes without saying that most people without jobs cannot buy homes. Some have argued that the majority of those who have lost their jobs are low wage earners who would not qualify to buy a home anyway. However, the BLS statistics indicate that for those aged 25 years and over who have earned a bachelor’s degree or higher, job losses since February number 1.8 million or roughly 20% of total job losses and their unemployment rate has more than doubled from 1.9% to 4.2%. There have also been 2.5 million job losses among those aged 25 to 34 years (regardless of education), which is considered the prime home buying cohort. All other things being equal, it is difficult to explain why job losses of that magnitude have not had a demonstrably negative impact on the demand for housing.
Besides the apparent headwind from job losses, consumer confidence remains well below pre-pandemic levels, as shown in the chart below:
Consumer confidence, as measured by the Conference Board, fell 47 points or 35% in just two months from February 2020 to the April low. Since then, it has gyrated, rising sharply one month only to fall back in response to setbacks in the economy and the pandemic. The recent declines in November and December, which have brought the index back near the April low, seem to have been caused mostly by the ascending second wave of the pandemic.
According to the survey, present conditions for consumers have improved modestly over the past nine months but expectations about future conditions have remained low. The percentage of respondents expecting business conditions to improve in the next six months has declined from 42% in May and June to 29%, while the percent anticipating that they will worsen has remained steady at around 22%. These figures are not supportive of robust housing demand.
Low Mortgage Rates Have Helped Propel Housing Demand. Alternatively, the Federal Reserve’s accommodative monetary policies have pushed both Treasury yields and mortgage rates lower, providing significant support to housing demand. Mortgage rates have been drifting downward since late 2018 and have recently been hovering at historic lows.
In the final week of 2020, the Federal Home Loan Mortgage survey reported average rates of 2.67% for 30-year fixed rate mortgages, 2.17% for the 15-year fixed rate mortgages and a 2.71 initial rate for 5-year/1-year adjustable rate mortgage. Some of the 2020 surge in housing demand is undoubtedly due to the desire of buyers to lock in low rates. Low mortgage rates also help marginal buyers qualify for a mortgage loan (or buy a higher priced house with more options). I calculate that the 100 bp drop in mortgage rates that has occurred over the past year gives buyers about 13.5% more spending power (excluding their down payments).
A Step-Up in House Price Appreciation Reduces Some of This Extra Buying Power. Various sources suggest that house prices have increased at a faster rate in 2020, especially during the fourth quarter. The year-over-year average rate of appreciation has been hovering around 8%. Before the pandemic, house prices had been rising at 4%-6%.
The S&P Corelogic Case-Shiller U.S. National House Price Index, which uses a repeat sales methodology to record the actual price increase on each specific sale, increased 7.0% on a seasonally-adjusted basis and 8.4% on an unadjusted basis in October vs. a year ago. The Federal Housing Finance Agency’s Purchase Only House Price Index, which also employs a repeat sales methodology, rose 7.8% in 20Q3 vs. a year ago.
In its most recent report on existing home sales, the National Association of Realtors pegged the year-over-year increase in the median sales price at 14.6% and the increase in the average price at 11.3%. However, this does not take into account differences due to changes in the mix of sales (i.e. the average size and location of each sale).
House prices have been rising at a rate that is more than twice the rate of inflation for the past decade. The increase has been fueled by the steady decline in mortgage rates. A substantial portion of the benefit from falling mortgage rates has therefore accrued to sellers (i.e. owners) and intermediaries. Buyers have realized some benefit, but the potential benefit from increasing affordability has been limited. In my mind, this significant, multi-year increase in real house prices is a sign of dysfunction in the housing market that limits the potential increase in homeownership and also the subsequent spending power (i.e. disposable income after the mortgage payment) of home buyers.
Mixed Readings on Homeownership and Household Formations. By definition, in order to add one house to the housing stock (and avoid a vacancy), there should be a corresponding increase of one household. Thus, household formations should correlate perfectly with housing demand (supported by employment and consumer confidence).
With that in mind, the most recent estimate of households by the U.S. Census Bureau is a headscratcher. Last month, the Bureau, through its Current Population Survey, estimated that there were 128.45 million households in the U.S. in 2020, a decline of 128,000 from the 128.58 million households estimated for 2019. This is the first recorded annual decline in households in the post-World War II era.
In arriving at this estimate, the Census Bureau notes that the survey’s nonresponse bias (i.e. the proportion of individuals and households that did not respond to the Bureau’s outreach) was especially significant this year. It also said that typical response patterns changed by education, Hispanic origin, citizenship and nativity, requiring it to make probability weighted adjustments that may have affected the accuracy of the published results.
Even so, the general thrust of the 2020 household estimate – that households in total decreased slightly in 2020 – is consistent with what I would expect to see as a result of the economic consequences of the pandemic. It may be appropriate to wonder how any houses could be built, if household formations were near zero; but it is also likely in any case that the unusual strength that we have seen in housing represents a shift of renters to buyers, and thus an increase in the rate of homeownership.
Indeed, the Census Bureau’s homeownership report appears to confirm this theory, even though the timing of the reported increase in homeownership does not match the rebound in new orders. According to the Bureau, the rate of homeownership in the U.S. jumped from 65.3% in the 20Q1 to 67.9% in 20Q2 and then eased to 67.4% in 20Q3. The 20Q2 increase is attributable to a big shift of properties previously classified as vacant to owner-occupied and also from renters becoming owners during the early months of the pandemic.
Here too, however, the Bureau notes that its field workers were only able to conduct about half of the in-person interviews that they typically conduct in both 20Q2 and 20Q3. Such contact may be especially important today because of the significant growth in the single-family rental market that has occurred in recent years.
Thus, this year’s sharp increase in homeownership preceded the sharp rebound in new home sales. Furthermore, a substantial portion of the rebound in new home sales has yet to be delivered, so those buyers do not yet show up as owners.
With such a strong shift from renters to owners, there should also have been an increase in rental vacancies. However, the rental vacancy rate was reported to have declined to 5.7% in 20Q2, before bouncing back to 6.4% in 20Q3, in line with its average level over the past several years.
Data collection has undoubtedly become more challenging during the pandemic. The difficulty of reconciling the 20Q2 data from the homeownership report with general developments in the economy and the reported decline in household formations with housing completion rates raises questions (in my mind at least) about the accuracy of other reports coming out of the Census Bureau in 2020.
2020 Housing Statistics and Forecast for 2021. In my February 2020 housing market update report, I noted that housing started the year quite strong. Extending that strength, I anticipated a solid year for the new homes market and for housing in general. Although the economy was much weaker than I anticipated due to the pandemic, the housing market will end 2020 with better performance than I predicted in February. New and existing home sales and prices will end the year up meaningfully. My projection for single-family housing starts will be essentially on target.
Selected Single-Family Housing Statistics: 2017-2021F
Sources: U.S. Bureau of Economic Analysis, U.S. Commerce Dept., U.S. Census Bureau, U.S. Dept. of Housing and Urban Development, National Association of Realtors, S&P Corelogic Case-Shiller and Lark Research projections.
For 2021, assuming that economy continues to recover from the April pandemic lows and grows at a pace at least in line with consensus forecasts and also assuming no further negative shocks from the pandemic, national politics or geopolitical incidents, I anticipate that housing sales will increase only slightly and house prices will rise at a more moderate pace in 2021. The tepid outlook for sales reflects an anticipated pause in demand, following last year’s surge that pulled a significant amount of demand forward. 2021 should also be a year when the housing industry catches up on its backlog; so housing completions (and homebuilder deliveries) should rise at low double-digit rates.
Sustainable growth in new housing beyond 2021 almost certainly requires a return to near-full employment and a gracious exit by the Federal Reserve from its unprecedented accommodative monetary policies that threads the needle between stoking inflation by keeping interest rates too low for too long and throttling the economy by raising them too fast. If that happens, consumer confidence should rebound at least to pre-COVID levels and personal incomes should rise at a pace that supports modest growth in the broader economy.
Results of Publicly-Traded Builders Positioned Mirror the National Data in 2020. After slowing in 20Q1 and 20Q2 with the onset of the pandemic, the average year-over year percent change in 20Q3 new orders for 11 publicly-traded builders that I follow surged to 47.4%, a post-2008 financial crisis high. That surge was in line with the increase in national new home sales of 44.6% for 20Q3. All of the 11 builders reported double-digit increases ranging from 16% to 81%.
As noted earlier in this report, the growth in unit closings has lagged new orders, due to various supply constraints mostly associated with the pandemic. In 20Q3, these 11 builders posted an average increase of only 7% in unit closings. A few builders reported declines in closings, while others had modest increases and several had solid double-digit gains. The growth in closings is expected to increase over the next few quarters; but it is not clear at this time how long it will take for the builders to catch up with their backlogs, which were up 38.3% in units and 40.2% in dollar value in the quarter over the prior year.
With high backlogs going into 2021, builders are expected to post earnings gains averaging more than 30%, according to consensus estimates, about twice the gain expected for 2020 earnings. Based upon 2020 year-end closing share prices, the 11 builders are valued at 10.4 times expected 2020 earnings and only 7.9 times projected 2021 earnings. These low multiples, which are well below market averages, imply that investors have some doubt about whether the builders can achieve 2021 consensus earnings projections. This undoubtedly is influenced at least in part by the slow ramp in unit deliveries that we saw in 2020 and also to uncertainties about the ultimate sustainable levels and growth of net new orders.
2020 was a very good year for homebuilding stocks, despite the gyrations caused by the pandemic. My index of share prices of the eleven publicly-traded homebuilders advanced 21.2% in 2020, better than the gains of 16.3% in the S&P 500 and 18.4% in the Russell 2000. This gain occurred even with the 67% decline posted by the Index over the four-week period from February 21 to March 20. The stocks recovered quickly, earning back all of their February-March losses by mid-August and going on to close higher by year end.
Yet, the group has underperformed the broader market averages since late November, posting a modest loss, even though the S&P 500 and Russell 2000 continued to log gains. The relative underperformance is probably the result of the modest slowing of new home sales (as reported by the Census Bureau) over the past two months, some profit-taking going into the end of the year, and perhaps some uncertainty about what 2021 will bring, given the divergence between the amazing strength in new orders vs. the ongoing weak fundamentals of the broader economy.
So far, all of the homebuilders’ gains have come because of the surge in their order books. In order for their shares to sustain those gains (or continue to move higher), the builders will have to realize the potential revenues and earnings that are now reflected in their backlogs and demonstrate that some of the strong order growth realized in 2020 will carry into 2021 and beyond with improving economic fundamentals.
January 6, 2021
Stephen P. Percoco
16 W. Elizabeth Avenue, Suite 4
Linden, New Jersey 07036