In sort of the same vein as Scott Sumner's frequent complaint that it seems like in 2007 everyone suddenly forgot all the macroeconomics that they had learned over the previous 50 years, it seems to me that some basic finance has been forgotten, too.
Given a certain market rate of return on highly liquid at-risk assets, investors seem to be able to earn excess returns over time for holding assets with similar risks, but with less access to ownership or less liquid markets. Price is the inverse of yield. So, small cap stocks, assets which cannot be easily purchased on liquid markets, small private businesses, etc., tend to sell for lower prices (higher implicit expected returns). This is why back-testing might suggest higher risk-adjusted returns in some alternative asset class, and then when the financial sector responds by creating a bunch of ETFs and other mechanisms that provide access to that asset class, the returns disappoint. Once the asset class becomes liquid, it loses its excess returns. This is a common problem in finance. Return anomalies frequently disappear after they are exploited.
So, traditional models of finance would predict that, once all factors are accounted for, owner-occupied housing should have a higher rate of return than investments with similar risks, because there are obstacles to ownership. As those obstacles are removed, required returns will decline to normal levels, providing a temporary price boost, and leading to future returns more in line with similar assets.
Most neighborhoods were never credit constrained, so taking the general context of housing markets as a given, those neighborhoods would reflect equilibrium prices without an additional liquidity discount. In neighborhoods that are credit constrained, we might expect lower prices. And as those constraints are lifted, as they might have been in the 2000s when new mortgage products were widely marketed, we might expect those prices to rise to levels similar to the neighborhoods that were never constrained.
There is no reason to believe that the new price levels will be above a reasonable level. That is the case, even if the removal of credit constraints causes low priced homes (presumably more credit constrained) to rise more than high priced homes. Our presumption should be that access to credit makes the market more efficient, not less.
|Zip code level rents in 1998 and 2006 are inferred|
from MSA median rent changes for years before 2010
Instead, Dallas shows little change at all. Los Angeles, which has a pattern similar to the other Closed Access cities, has rising Price/Rent at the top end of the market. This naturally pulls prices higher, pulling all P/R levels up along with it. The slope of the relationship doesn't change much. It's just that all zip codes kind of climb the P/R ladder as prices rise in general. In both Dallas and Los Angeles, P/R stops rising around $400,000 or so, suggesting that this is related to the ability to capture tax benefits.
When I create the same graph with incomes on the x-axis, however, we do see a pattern that suggests credit expansion. In low income neighborhoods, the P/R ratio did rise slightly more than in the high income neighborhoods, in Dallas. In Los Angeles, it rose substantially more. By 2006, the slope of relative P/R by zip code income had flattened substantially.
In San Francisco, where all zip codes had reached that $400,000+ range where Price/Rent seems to flatten out, the P/R trendline completely flattened out by 2006, even going slightly negative.
In Boston, where top end valuations didn't move that much during the boom, the rising P/R levels at lower incomes show up clearly, too.
I think what we are seeing here is the difference between the extensive margin (new buyers) and the intensive margin (spending among existing buyers). I have shown that there was much less correlation between rising homeownership, non-conforming mortgage growth, and price increases than it seemed. Most of the rise in ownership came before the unusual rise in prices, and the rise in non-conforming mortgages was actually associated with a sharp decline in homeownership. Homeownership peaked in early 2004, just when the big jump in non-conforming loans was getting started. By 2006, and especially by 2007 when prices began to really tumble, homeownership was on its downward path.
Yet, clearly, when looking at zip codes by income, there was a strong rise in prices, negatively correlated with incomes. All of this together suggests the bulk of the expansion was at the intensive margin - higher prices and more debt even though ownership didn't seem to be expanding in these areas.
But, even here, the data balks. There isn't evidence in the broader survey data, such as the Survey of Consumer Finances, of an unusual rise in debt levels of households in the mid and lower income quintiles. And, the most extreme peculiarity about the cities where low income zip codes had the largest price increases is that, during that time, low income households were moving away from those cities by the hundreds of thousands. Such a strange juxtaposition. A massive outflow of households from the places where home prices were rising the most. Where was the demand coming from?
It seems clear that, to some extent, changes in population composition were important. But, in the end, I think it may be a distraction to think of this through a supply and demand framework. What caused what? Who was buying, and how much money did they have?
As I mentioned in the previous couple of posts, intrinsic value rules. And, this is where these Price/Rent to Price Level graphs come in. What caused the rise in prices? Supply constraints? Migration patterns? Expansion of credit access? These all probably played a part. But, when we note this relationship between P/R and Price, we don't actually need a special explanation for why low priced homes increased by more than high priced homes. Anything that leads to rising prices will trigger this positive feedback effect. Whatever the root causes of rising prices, those causes simply became strong enough in the Closed Access cities to make this effect noticeable.
Because what started me down this path was noticing the lack of evidence for mortgage growth and homeownership among low income households during the boom, I have tended to be more amenable to the passive credit school (the school of thought that believes credit expansion was more of an effect than a cause of the housing "bubble"). But, I think that is really the wrong question to ask. And, since it is the wrong question to ask, it leads both passive and active credit proponents to the wrong conclusions. They mostly disregard rent as an input to price, which seems defensible because prices were so volatile while rents tend to change slowly. So, prices, almost by definition, reflect irrational expectations. (If you disregard the sole source of income for a financial asset, by what basis could you even construct a rational model?) Then, they end up just arguing about whether it was mostly low income buyers, flush with new credit, who were irrational, or whether everyone was irrational.
Intrinsic value rules. What actually happened was that the market was ruthlessly rational in spite of us. Intrinsic value pushed these prices up that P/R ladder as the various influences on rising prices in the Closed Access cities became increasingly extreme. The only way we could have prevented this from happening would have been to implement extreme economic dislocation and financial repression. Oh, look! That's what we did!
If we pull the fetters off the mortgage market without fixing the Closed Access supply problem (or, alternatively, getting rid of the corporate income tax* or the owner-occupier tax benefits) low priced homes would naturally start climbing that ladder again as those markets heal. This would finally begin to heal the balance sheets of working class homeowners who have survived all that we have imposed on them. And, I don't see how that will happen without leading to a new chorus for macro-prudential regulations or monetary suffocation to shut it down.
* The largest owner-occupier tax benefit is the non-taxability of imputed rent. It seems to me that the most feasible way to eliminate this relative benefit would be to eliminate the corporate income tax.