Housing: Part 214 - More Color on the real reason for price appreciation in "credit constrained" zip codes.

I have done some previous posts on the reason why low priced homes increased in value by more than high priced homes, in some cities. (Here's a post.)  This had little or nothing to do with credit.  It had to do with tax benefits.



idiosyncraticwhisk.blogspot.com  2017

Today, I will go into a little more detail to show how clear this pattern is.



idiosyncraticwhisk.blogspot.com  2017

I compared Seattle and Los Angeles in the previous post on this topic.  Both cities had high expectations for future rent inflation at the height of the bubble in 2006.  So, Price/Rent ratios at the top of the market were in the high 20s in both cities - far above the national norm.  The difference between these cities is that rents aren't as high in Seattle as they are in Los Angeles, so home prices in absolute terms in Seattle are still somewhat lower.  There are few zip codes in Seattle where the median home is worth more than $500,000.  But, most of Los Angeles is above that price threshold.

This pattern of Price/Rent ratios exists in every MSA.  Given that, and comparing White House tax numbers with BEA imputed rental income numbers (income tax benefits amount to about 25% of net rental yields after consumption of capital, and most of those benefits are claimed by high income households) we can infer that, at the low end of the market, dominated by landlord owners and by households with little in tax obligations, tax benefits are negligible.  At the high end of the market, the marginal tax benefit must top out at something well above 30% of the value of net rental yield.

In 2006, low end Price/Rent levels appear to generally run about 30%-40% below P/R at the high end, across cities.  Every city has a different peak Price/Rent level, depending on rent inflation expectations, property tax rates, etc.  But, they all have this pattern.  (Low end prices are currently more depressed than normal because of unusual credit constraints.)

But, it appears that at something around $500,000, the marginal tax benefit and the average tax benefit of any further increases in home value are about the same.  Above that level, P/R flattens out.  So, for cities where home prices rise above that level, this positive feedback loop in rising prices ceases.  It isn't that low priced homes in Closed Access cities were rising unusually during the bubble.  It's that their high end homes weren't rising as sharply as they would normally rise in a similar environment because their Price/Rent ratios were no longer expanding.

idiosyncraticwhisk.blogspot.com 2017

Here, we can see relative price levels, by price quintile, in Seattle and LA.  None of the quintiles reached that price level in Seattle, so there was no difference in price appreciation between quintiles.  In LA, all quintiles ended up generally above that price level, so each quintile moving up is more exposed to this cap in P/R inflation, and as you move up quintiles, price appreciation becomes more muted.

idiosyncraticwhisk.blogspot.com 2017

And, we can further confirm this pattern by looking at Riverside and San Francisco.  By the end of the boom, the top quintiles in Riverside were just pushing above that $500,000 level.  So, in Riverside, price appreciation in the top two quintiles is moderated, but the rest of the quintiles are still moving together.  You can see a little bit of that pattern in Miami and Washington, DC, also.



idiosyncraticwhisk.blogspot.com 2017

In San Francisco, the top two quintiles began the period at or above that $500,000 range, and the rest of the quintiles moved above that range by the end of the boom.  So, in San Francisco, we see the opposite pattern of Riverside.  In San Francisco, the bottom three quintiles have different rates of appreciation, but the top two quintiles are similar to one another, because both of those quintiles started the period at peak P/R.

This pattern is quite regular.

Notice also what we see here.  In the cities where low end prices


idiosyncraticwhisk.blogspot.com 2017

appreciated more, those prices collapsed early in the bust, along with the general price decline.  It was natural to assume that this was a natural retraction of unsustainable demand.  But, this was actually simply a reversal of this pattern, because now home prices were declining, so that the tax effect was creating a negative feedback loop in lower priced zip codes as they fell back below that $500,000 range.



idiosyncraticwhisk.blogspot.com 2017

And, we can confirm this, too, with a graph from another city - Phoenix.  Is there anyone who will doubt that Phoenix was the poster child of a bubble, with free-flowing credit?  But, Phoenix never pushed above that range.  The top quintile of zip codes just barely reached above $500,000 at the peak.  So, there was little difference among quintiles in Phoenix at the peak.  And, until late 2008, there was little difference in the rate of decline.

So, there are two distinct reasons for the exceptional decline in low priced homes during the bust.  The initial decline in 2007 and early 2008 was specific to the Closed Access cities and was simply an unwinding of this tax effect.  Again, it isn't so much that the low tier homes were falling faster in those cities, as it is that the high tier homes had more moderate price shifts because they didn't have this pro-cyclical feedback.  Notice how level the high quintile prices were in San Francisco and LA during the bust compared to the other cities listed here.


idiosyncraticwhisk.blogspot.com  2017

Then, there is a second wave of low tier price collapse after late 2008.  This is especially noticeable in Phoenix and Seattle.  We see this pattern in cities like St. Louis and Chicago, also.  This is the period where federal public policy was devastating to the lower tier market.  Fannie, Freddie, and Dodd-Frank have effectively shut down lending in the lower tier owner-occupier markets.  I have been attributing most of the lower tier collapse to the GSEs, with a late assist from Dodd-Frank.  But, looking at this next graph, and recognizing that the early collapse in first quintile prices in the Closed Access cities (and generally the excess gains in the Contagion and other cities, due to Riverside, Miami, and Washington) was due to the reversal of this tax effect, and not really due to credit contraction (relative to higher tiers of the market), Dodd-Frank appears to be a much stronger part of the late credit influence than I originally appreciated.

Notice that the lower tier collapse isn't as noticeable in the Closed Access cities as it is in the other cities during this later period.  That is because these effects are now mitigating.  All cities are affected by the severe constraints in low tier mortgage markets.  But, in the Closed Access cities, as prices rise again, the positive feedback of P/R would be pushing those low priced homes up at a faster rate if it wasn't fighting those constraints.


I find that even I tend to make demand-side inferences that I eventually have to backtrack on when I realize that markets are more efficient and intrinsic value is more important than I gave them credit for.  When I originally questioned the credit-supply explanation for this effect, I inferred from the pattern that it was the migration flows into and out of the Closed Access cities that created this effect.  I inferred that the marginal home buyers had higher incomes than previous owners, so that they could capture more of these benefits, and that had something to do with the sharper rates of appreciation in lower priced neighborhoods.

But, I was wrong.  Intrinsic value rules the day, in the end.  In the aggregate, housing markets appear to be even more efficient than I tended to assume.  There is ample inter-tier substitution throughout local housing markets.  In hindsight, it is implausible that some portions of a metropolitan market would have unsustainably high valuations because of demand-side factors.  (By this, I mean factors such as credit supply.  Real estate will always be dominated by local factors that are related to changings rent levels due to localized market shifts in amenities, safety, etc.)  In the aggregate, where those local valuation factors tend to average out, we can see that tax benefits are priced into home values quite systematically and the potency of credit supply as an explanation appears to be limited to the extreme case, where severe constraints prevent prices from rising to their intrinsic values.  (Or, stated from a different framing, the retrenchment of the owner-occupier market farther up into mid-tier markets pulls prices in those markets down to the landlord level, where owner-occupier tax benefits are not available.  So, the slope of the P/R relationship is now steeper.)

I suppose if I have changed my mind before, I might end up changing it again.  But, so far, the changes induced by my exposure to the data have been changes toward more respect for macro-efficiency.