U.S. Housing
US Housing
With somewhat of a quiet day in the U.S. markets yesterday, I thought I would use this opportunity to address a topic a number of you have asked about recently. These are my thoughts, somewhat unstructured, at trying to bring a more objective and balanced view using some of the available data.
U.S. Housing
Real estate related headlines are currently dominated by the question of where home prices are going, and the resulting opinions give a wide range of possibilities, with the vast majority seemingly engendering fear calling for an epic 2008 style crash.
Measures that are often emphasized to highlight the potential downside risks include:
Median home sale price up 36.5% in 2022 Q2 from 2019 Q2
Existing home sales down 5.9% sequentially in July from June
Elevated measures of median home price to median income
Various macro fears including the looming possibility of a recession, inflation concerns and a struggling stock market
30-year fixed-rate mortgage rose to a weekly average of 7.11% on 10 Oct 22.
However, there are just as many reasons that fundamentally explain the home price rise and provide arguments for why home prices in most markets could hold or even continue to rise, albeit at a slower rate:
Wages have risen by 16.1% nationwide in 2022 Q1 since 2019 Q1, and 6.7% year over year.
Employment has now surpassed pre-COVID levels
Remote work has shifted personal preferences and may have caused half of the rise in home prices, according to a working paper by the National Bureau of Economic Research.
New single family home permits were down 11.7% in July 2022 on a seasonally adjusted year-over-year basis, while multifamily unit permits were up 26.2% year over year.
NAR Report : “`We're witnessing a housing recession in terms of declining home sales and home building,´ Yun added. `However, it's not a recession in home prices. Inventory remains tight and prices continue to rise nationally with nearly 40% of homes still commanding the full list price.´"
Millennials continue to buy their first homes as they now comprise the 26-41 age range in 2022 with a generational cohort roughly 10% (6 to 7 million people) larger than their Gen X predecessors, providing a large source of home buying demand.
In terms of median home price to median income ratios, whilst home prices have risen substantially over the past 3 years, home buying is still within reach for many potential home buyers (within the actual home buying income deciles) in most markets, and broad national measures using median home prices and median incomes are flawed – a have a “deep dive” on this, reach out if you would like it.
This holds true even if 30-year mortgage rates pass the 7.11% mark.
Nevertheless, many potential homebuyers have surely seemed to move to the sidelines to watch the current volatility settle out. However, there are also signs that home buying interest overall hasn’t lessened. According to a UBS Consumer Labs Research report, “This year search interest increased +1.6% for existing homes (EH) and +22.1% for new homes (NH) from the end of May'22 through Aug'22”.
What Drives Homes Prices?
With that background, we’re back to the question: where are home prices going and when? Simply looking at a nationwide aggregate requires the consideration of many variables and leads Zillow to forecast a year over year increase of 1.4% while Moody’s sees a 0% up to a -5% decline.
However, the complexity lessens when the question is approached from the MSA level where a few key local market metrics can paint a clear picture of what to expect. Each local market has a unique story that will vary from the nationwide average. One is able to use historic data to set baseline expectations with Zillow’s ZHVI metric providing reliable approximations for the typical home price in each major market going back to 2008.
I have found that from 2010 to 2019 one variable was the dominant explanatory variable for Y/Y home price change in an MSA: the change in the number of employed people over the previous two years within that market. For example, if you wanted to know how home prices were going to change in a given city from 2015 to 2016, you would want to look at the increase or decrease in the number of people with jobs in that city in 2015 relative to 2013. This one leading indicator predicted 64% of the home price growth variance in markets with over 400,000 workers in a simple linear regression model using data from 2010 - 2019. This size threshold covers markets roughly the size of Tulsa, Oklahoma and larger.
Having one variable that covers 64% of the total variance is huge. After accounting for this 2-year employment growth, other variables such as changes in home supply, demographics, wage growth, etc. all had relevant, but small, impacts when predicting year over year price change. We know that considering all the variables available in a complex, non-linear, multi-variable, model will yield a more complete prediction. However, it’s often beneficial to avoid creating a black box when you have one predictive variable this strong. Focusing effort instead on understanding that one variable while filling in the remaining gaps with expert knowledge is often a better alternative.
In the regression above, I only checked in on the data once per year. An easy way to improve the model would be to also consider rolling 1-year and 90-day employment changes and update the forecasts more frequently. For example, in the 2015-2016 hypothetical, if you were in a given market in early 2016 and your expectations of 2016 home prices were set by strong employment changes from 2013 to 2015, but you then observed the employment situation drastically worsen, you would logically lower your expectations. So, while rolling 2-year employment change is the single most important predictor, when you build out the model and augment it with rolling 1-year and 90-day measures, model predictability improves to ~70%.
While I do expect that wages, housing supply, demographics and other similar variables have an impact on prices over longer time horizons, it’s very clearly all about workers in the short term.