Housing: Part 349 - Homeownership rates

The Census Bureau recently published the 2019 first quarter numbers on the housing stock.  Homeownership rates had bottomed out in 2016 at 62.9%.  That was one quarter which was probably an anomaly.  Generally, the bottom appears to have been about 63.5%.

It had generally risen since then, up to 64.8% last quarter.  However, this quarter, it moved back down to 64.2%.

I have been watching this number because there have been mixed signals on the housing market.  As the analysts at AEI point out, by some measures, mortgage standards have been easing.  However, according to the New York Fed, there hasn't been any loosening to pre-crisis standards in terms of originations by FICO score.

At the same time, the low rate of building has been levelling out along with prices and resales in some markets.  It seems unlikely to me that homeownership can continue to recover without easing in terms of borrower quality.  My interpretation of this mix of data is that the various factors that are causing a shortage of housing supply are pushing up housing costs, which leads to the use of riskier mortgage terms, and that part of the problem is that the constraints on lending to financially marginal households who would have been buyers in previous generations is one factor that is causing the shortage.  Looser lending would help pull up prices in low tier markets, back to price points that make new building profitable.  That's what needs to happen to lower housing costs in general.

However, my hypothesis would need to be reconsidered if homeownership continued to rise while borrower-based lending standards remained tight.

This quarter is an interesting number, because it presents the possibility that my point of view is correct.  On a noisy measure like homeownership, it is hard to tell what is noise and what is signal until some time has passed.  It could be that the 62.9% number and the 64.8% number were just noise, and that homeownership bottomed out at about 63.5% and only very slightly rose to about 64.2% where it will remain.  If it is still in this range in a year, that is plausibly the case.  If this quarter turns out to be the outlier, and homeownership is up to 65% next year, then perhaps a recovery is possible in homeownership without expanding lending to more marginal borrowers.

Time will tell.


On High Tier vs. Low Tier Prices

I want to discuss tier price levels for a moment.  There has been some recent recovery in low tier prices vs. high tier prices, which appears to lend credence to the idea that lending is loosening substantially, in spite of the FICO score data.  There are a couple of caveats to note here.

First, there is a bias in the way some analysts use Case-Shiller indexes.  As with so many factors on this topic, it is purely a function of priors.  It isn't a bias at all if the conclusion that credit markets were the main cause of rising low-tier prices before the crisis is already taken to be true when the data is analyzed.  However, it does appear to be a bias if you question that conclusion.

Case-Shiller has 20 city-specific indexes, which includes all 5 Closed Access cities.  Be careful looking at analysis of low- versus high- tier prices that uses those indexes. If the data is heavily populated with Closed Access data, it will not be indicative of national markets.  Low-tier vs. high-tier prices act differently in those cities than in other cities.  I go into that a little bit here.  Or, better yet, buy Shut Out to read about it(using code 4S18MERC30 for a discount).

Furthermore, even in national stats there is a bias here.  The problem is the extreme nature of the walloping we handed to low tier housing markets.

Imagine a city where high tier markets bottomed out at a 20% decline and low tier markets bottomed out at a 50% decline.  As a proportion of the peak price, that's a 30% additional decline in the low tier.  In spite of conventional wisdom to the contrary, in most cities, that wasn't undoing anything.  Low tier prices hadn't risen significantly higher than high tier prices had.

The bottom came around 2012.  Now, if someone uses 2012 as a baseline, they may find that high tier prices have recovered by 25% since then, while low tier prices had recovered by a whopping 40%.  If one treats the 2012 market as the benchmark, it would seem that low tier prices are 15% overvalued.  But, if we treat the pre-crisis level as the benchmark, then there has been no catch up.
  High Tier 80% x 1.25% = 100%
  Low Tier  50% x 1.4% = 70%

In other words, low tier prices are still 30% undervalued compared to high tier prices, and there has been no catchup at all.  Because conventional wisdom has been so blind to the fundamental causes of the crisis, this sort of bias is very common among academic papers, policy papers, and general journalism.  It's fascinating how an innocent shift in priors can create these self-fulfilling biases in the analysis.  There are clues about these biases though, once your frame of reference moves to a place where useful questions aren't obscured by priors.  On this issue, for instance, it would seem to be a mystery why low tier building rates are still dead if prices are 15% inflated.  But, it isn't really a mystery at all if those prices remain relatively low.

Keep in mind, this is a hypothetical example, although it is a realistic number for many cities.  There has been some low-tier recovery. Maybe the post 2012 appreciation rates have been more like 50% compared to 20%.  But, in this example, for instance, to completely re-attain previous norms, low tier prices would have to double from their lows while high tier prices only rise 25%.  Even though this is arithmetically the case, it would be quite a difficult point of view to sell to someone who is convinced that excess lending or speculative buying is the ever-present monster under the bed that needs to be thwarted.  As Russ Roberts of EconTalk fame frequently laments, data isn't as powerful as we would like it to be in these debates, because the issues are just too complex and too bound up by differing premises.  Isn't it striking?  There is a stew of ever evolving priors informed by data, informed by priors, etc. etc. which leads us to a point where it isn't possible to come to agreement about whether low tier prices are overvalued at  a 100% appreciation level vs. 20%.  How do we ever come to know anything?