Friday, October 30, 2015

Housing, A Series: Part 76 - There are two Americas.

There are two Americas, and there is only one practical way to merge them.  Houses in California and New York.

I have been pointing out how migration patterns in the US are backwards.  Net migration is from high income to low income cities.  I like the North, Wallis, and Weingast idea of open access social orders vs. limited access social orders as a way of understanding free and wealthy societies.  There are a lot of discussions these days about two Americas, the 1% and the 99%, about economic rents and monopoly profits.  I think these discussions are a response to something real, but it is not tax policy or negotiating power at the heart of this problem.  It's limited housing.

In some ways, I think there is something to the usual discussion.  Factors like unionization which were more prominent when incomes appeared to be less variable do have something to do with this.  As I discussed in the previous two posts, when persistent geographical competitive advantages emerge, firms with excess profits are captured, and some variety of methods of extracting those profits will inevitably arise.  Unionization and housing constraints are two forms of these methods.

The difference between these two periods really does have to do with the decline in unions, because unions were probably a particularly tempered way of extracting rents.  Today, the housing constraints that are creating economic rents in our large cities are operating in a context where highly skilled and intensively networking individuals can leverage their incomes practically without limit in the technology and financial fields.  So, it is true that unions in a semi-skilled mass production context put a cap on income inequality while today rent-seeking among tech and finance high performers is leading to a particular rise in the small number of earners at the top of the income distribution.

But, I think we need to be careful about one error in thinking about this distinction.  Both eras had rent seeking.  Both kinds tended to capture excess income for a select group by limiting access to opportunities.  The solution isn't to re-implement the former source of rent seeking.  The solution is to remove the source of economic rents.  This is clear when we look at Washington, DC, which has become the metro area with the highest incomes.  Those high incomes are not related to housing constrictions, but they are related to unionization, since the public sector is the one area where unionization continues to play a strong role.

Instead of trying to be a different version of a limited access society, let's strive again to be an open access society.  This is not as easy an idea to sell as it should be, because open access means letting builders build, letting lenders lend, and letting high income workers work.  But, doesn't it seem like they are the problem?  Of course, our punitive instincts at this point have made things even worse.  The nationwide housing depression that is the result of the collapse of our mortgage market has pushed housing expenses above their comfortable range across the country.  Now if you can manage to qualify for a mortgage, you can access economic rents from anywhere, and some of that additional 5% of incomes that is being extracted for housing can be yours.  Limited access, macroprudence edition!

The two Americas are a largely geographical phenomenon.  Here are several graphs that drive this point home.  Outside of New York City, Los Angeles, San Diego, San Jose, and San Francisco, America is a picture of classical economics.  Housing supply is elastic - in other words, housing markets are relatively efficient, so that if some opportunities for rising income arise, households can move there, and homes will generally be constructed for them at a typical cost.  Populations flow to opportunities, and costs and incomes are pushed competitively until the marginal worker is ambivalent about their set of relative opportunities.

All of these graphs are scatterplots of the 20 largest metro areas.  The x-axis measures net domestic migration and the y-axes are various measures of housing affordability.  The dark red dots are the NY/California problem cities and the US average and the orange dots are the other cities.

By the way, this is all from Zillow.  What a valuable courtesy it is that Zillow makes all of this data available.

1: Rent Affordability

Across the country, households spent a comfortable level of income on housing (rent or imputed rent) of about 25% until the banks met the business ends of our angry pitchforks in 2008.  That has risen to about 30% with the collapse of mortgage banking.  Even with that dislocation, the open-access portion of the country appears to be moving in equilibrium.  But, if you want in to the high income opportunity zones, you're going to have to pay up.

2: Rent

Here is a similar graph, based on median rent.  The cost of tech sector opportunity is about $20,000 per year in rent.  By the way, just taking an unweighted average among the problem cities, both rent and income tend to be about $15,000 above the national average.  This suggests that at the median most of the economic rents are going to real estate owners, and that the higher incomes are a mirage that mostly reflects the cost of entry.

These cities tend to have higher income variance than usual, especially at the top of the income distribution.  This makes sense, if we think through the supply and demand chart from yesterday's post that I have re-posted above.  Low income households will be at the margin of sustainability in these cities.  But, at very high income levels, the real demand for housing will eventually be elastic enough that those households will be able to manage housing expenditures more like people would in a city without a housing constriction.  In other words, when a household's income is high enough, they will naturally decrease their housing expenditures to the lowest level they can live with until their housing expenditures reach that 20-25% comfort range.  This causes de facto income inequality to be higher than measured inequality, because the higher a household's income is, the less of it they will have to pay to landlords.  This leads to high demand for high end housing and high income in-migration, because the value of living in these cities is much higher, in individualized real terms, for high income households than for low income households.

3: Price / Rent

We see the same pattern in price/rent levels.  Across the open access portion of the country, there is no relationship between Price/Rent and migration.  Populations and opportunities equilibrate.  Price/rent levels during the boom moved up somewhat because of the natural effect of real long term interest rates on home values.  But, again, we can see here that the limited access cities create a continual ratcheting up of housing costs and incomes.  The expectation of continuing limited access creates a premium for real estate owners.

One way to think of this is that real estate prices internalize expectations about the future and the value of limited access rents.  The unfortunate news is that we have already paid the full price for the error of our limited access policies and we can never recapture those losses.  That is because past real estate owners already captured the gains from all the expected future limited access rents - in the form of higher rents and in the form of higher Price/Rent ratios that reflected those expectations.  If we manage to overturn these longstanding policies, the value of urban real estate should plummet.  But, at this point, that is simply a transfer from existing real estate owners to new real estate buyers.  It is still worth doing, because it means less of our output will be diverted to unproductive rent collections and it means that costs in our most innovative sectors will be lower when global competition comes to challenge them.  But, the married professionals that cashed out their $5 million homes in LA and retired to mountain villas in the Rockies aren't giving it back.

4: Home Prices

The double whammy of higher rents and higher Price/Rent ratios means that home prices are especially high in these cities.  Again, the rest of the country sees no systematic relationship between migration and home prices.

A similar pattern comes through if we look at home prices  as a proportion of income.  Price to income in California is about twice the rest of the country.

Lastly, here is a chart comparing Home Price/Income ratios for NY, LA, and SF, compared to the national level.  The national level includes these cities, so it was distorted higher by these cities during the boom.  And yet, at the height of the boom national price/rent levels were just reaching the bottom of the range of price/rent levels in these cities.

There was a small inflationary effect on home prices from falling long term real interest rates, but as with all of these measures, this is largely the picture of localized supply issues, not national demand factors.

Also, I think it is worth thinking about our perceptions and the problem of mental benchmarking.  Think of the difference between how we reacted to what we thought were demand-side excesses in the national housing market and how we have reacted to the housing market in New York and California that for decades has operated at higher valuation multiples than anything the rest of the country saw during the boom.  Our sense of outrage is highly dependent on our presumption of who is to blame.  Our presumption was wrong here.  But, the satisfaction we get from public, shared outrage is not so dependent on diligence - in fact it is diminished by it.

Thursday, October 29, 2015

Housing, A Series: Part 75 - Interpreting Housing Supply and Demand

Yesterday, I discussed the odd US trend of net migration moving to places with lower incomes.  This is logically the equivalent of reversing the flow of migrants over the Mexican border.  This doesn't happen in functional places.  If we have another decade with the mortgage markets we have had since the recession, maybe it will get to that - Americans sneaking across the border to take jobs at maquiladoras in order to lower our living expenses.  That'd show those predatory lenders we don't need 'em.

Here is one graph from yesterday, showing the unfortunate correlation between high incomes and out-migration.

20 Largest metro areas
Notice that the average for the US is below the trendline.  In general, there is a slow migration from rural to urban (or, at least, suburban).  There are a lot of areas, many of them rural or small cities with very low housing costs and negative migration flows that don't show up on a graph of the top 20 metro areas.  If we remove the housing constrained cities from this graph and add in those rural areas, we would see the typical pattern with a pretty functional housing supply response.  We would see a slightly upwardly sloped relationship between incomes and migration.

from Zillow
As a proportion of income, rent (or imputed rent) has generally run at about 25% since the development of federal housing programs and modern mortgage financing democratized home ownership, lowering barriers to ownership and increasing demand.  This has not been the case since 2008.  After federal policies drove the housing market into the ground in 2006, households had to begin consolidating their housing consumption.  When incomes eventually dropped in 2008 and 2009, housing supply was already stretched, so the average housing budget remained level when incomes fell.  We have continued to have depression level homebuilding since then, which has led to a resurgence of rent inflation, as households bid up rents on the dwindling relative stock of housing.  As this chart demonstrates, the nationwide housing costs (in terms of rent) are now similar to what they were in parts of California in the 1980s and early 1990s.

What the long term nominal and real housing expenditure data from the BEA and metro area time series of rent affordability suggest is that households have a natural resting housing budget of about 20% to 25% of income.  Where housing can be built and where there isn't an adverse negative pressure on incomes, housing expenditures tend toward this level.  Now, keep in mind, this doesn't mean that homes in all areas converge to a similar price point.  The typical household in downtown Chicago and the typical household in suburban Chicago could both be spending 25% of their income on rent, but they would be living in very different units.  The downtown household has chosen to substitute location for size.

There are situations, then, where we might see rent inflation, but the typical amount of spending by the households living in those homes as a proportion of income remains low.  In a place like Phoenix, where a well-regarded section of the city may have grown up around a neighborhood, the homes in that neighborhood may be worth more than they were 30 years ago, but generally this will mean that lower income households are migrating to other parts of the city and the households in that neighborhood will reflect a higher income level over time.

Or, in a slightly different context, in a place like Washington, DC, where incomes have grown immensely, we might also see high rent inflation but rents that remain very affordable.  In this context, prized locations will be bid up by the more wealthy population, and households might be trading down on factors like space in order to capture a better location, like the household in Chicago, but they will tend to only bid housing up to that comfortable level of around 25% of income or so.

When I first began looking at the supply issue, I was including Washington, DC with the problem cities, because it had rent inflation as bad as anywhere.  But, over time, I have realized that this was incorrect.  The Washington MSA has been able to maintain decent population growth and rent is still affordable.  These issues are sort of like solving a 3-variable problem with 3 equations.  Each factor is a clue.

Washington is like an extreme version of those old-economy cities in their primes.  There are economic rents available, but the lack of access that creates them doesn't come from housing; it comes from other sources.  For Washington, it is public sector unions, patronage, the in-migration of high income workers in regulatory and lobbying sectors, the lack of competitive pressures in federal payrolls, etc.  As with the old-economy cities, those factors create economic rents so that new workers don't necessarily have a way to capture a proportional amount of the city's excess income, so even though housing supply is relatively elastic, there is not a natural inflow of population that pushes incomes down to more normal levels.

In Washington, DC, housing supply could be somewhat inelastic.  To the extent that there are supply constraints, they probably aren't significantly limiting population growth.  If Washington, D.C. added new housing supply, it would probably lead to expanded real housing expenditures per household.  In other words, typical households would tend to spend the same amount on rent, but the added housing stock would allow them to move upmarket.  Those new housing units would either be in prime locations that allowed more households to gain location value or they would be larger.

Looking back at the top graph, Washington is an outlier.  It doesn't belong with the open access cities or with the housing constrained cities.  It is an example of a city with non-housing limited access.  In a future post where I graph the city migration data relative to housing affordability indicators, Washington will not be an outlier.  It looks much more like the rest of the country in terms of affordability.

Back to our supply and demand chart, even where supply is relatively unconstrained, it is possible for rents to rise because of rising incomes or home values.  We can start to see the strain on households in some cities where rent expenses start to rise above that comfortable zone.  In these cities, housing becomes constrained enough that households have reached the bottom limit of their real housing expenditures, as opposed to their nominal spending.  In other words, if housing expenses rise to 30% of income in these cities, the typical household is unwilling to move to a cheaper unit to reduce spending.  This is a transitional category.  In California, Riverside and Sacramento probably fit here.  They have growing populations, so they are capable of building, but the inflow of households escaping the high cost coastal cities is stretching their ability to build.

These cities may create some financial stress, but not necessarily enough to force households to migrate away.  In some cities, like Boston, there may even be a bit of an income response as firms and workers adjust to the supply and demand of labor in the slightly high cost context.

If we move all the way to the left panel of the supply and demand contexts where housing supply is very inelastic, this describes coastal California and New York City.  In these cities, typical households have moved downmarket as far as they are willing to go and their budgets are still hitting their upper limits.  Here housing has become the central determinant of who will live in the city.  Marginal households are at the top of the range of feasible housing expenses.  Any increase in local incomes will cause rents to be bid up.  In our graph, we might imagine an increase in incomes as a stretching of the scale of the y-axis and moving the green demand line upward.  In a city with elastic supply (at the right end of the graph) this would lead to real home investments.  But, since housing quantities in these cities are not flexible, the new income will mostly go to rent inflation, bidding up existing stock.  This is similar to what happens in the transitional cities.  But, in these cities, that rise in rents pushes households at the margin above their budget constraints, and rising rents lead to out-migration.

These cities tend to have high incomes, high rent inflation, high rent expenses as a proportion of income, and migration outflows.  The outflows happen first because the housing supply constraint simply doesn't allow enough housing to accommodate naturally expanding populations.  But, also, high income households move into the city to capture the economic rents going to that city's workers, and bid up the stagnant housing stock.  On the margin, this moves a household above the sustainable level of housing expenses, and some lower income household will need to out-migrate to a lower cost city.

In this context, regardless of elasticities and pricing power in the labor market for most income levels, wages simply have to be raised to the level that allows for the minimum housing at the top of the sustainable range of housing expenses.

Rent expenses in San Francisco have nearly doubled as a proportion of income, just since 2000.  Out-migration was very high even before the recession at lower rents.  The combination of local housing constraints and national mortgage constraints must be causing some sharp financial suffering.  I suspect that if the mortgage market ever begins to expand again, there will be very high out-migration from Silicon Valley.

As for the local supply issue, these cities are so bad that local residents probably don't associate new building with lower rents any more, which just makes the local political problem worse.  At the margin, rent inflation is forcing households out of the city.  So, marginal new housing will not lower rents substantially.  But it will help keep them from rising more.  This will lead, on the margin, to some household who would have had to leave the city for financial reasons, but now can stay.  That is a lot harder to notice than a 5% drop in rents.  One can imagine this leading to angry reactions about how "trickle down" economics is based on lies.

Tomorrow, graphs....

Wednesday, October 28, 2015

Housing, A Series: Part 74 - Fleeing prosperity

The past twenty years have been defined by the intersection of two competing developments: (1) the revolutionary rise of the value of cities in human networking, innovation, and productivity, and (2) an inability - after decades of growth, after being designed with tremendous foresight centuries ago for this very purpose - of those cities to allow housing expansion and population growth.

Here is the graph from yesterday's post that shows a tendency for rent to rise relative to other costs during economic expansions and for incomes in the constrained cities to rise in general.

Here is a graph of population growth by decade of several metro area core counties.  LA and San Jose (Santa Clara Co.) saw high growth until the 1970s, and since then growth rates of all the problem cities have converged to a 3%-5% range.  For San Francisco and Manhattan, this is actually an improvement over long term trends.  The population of San Francisco only grew by about 10% from 1950 to 2010.  Manhattan had 2.3 million residents in 1910, which was down to 1.6 million in 2010.  We might expect this sort of depopulation in the rust belt.  But, Manhattan has among the highest incomes and the highest housing costs in the country.

There was some population growth in the 1990s in the other boroughs, but that dissipated by the 2000s.  Generally, the very long term lack of any population growth in New York City and very little in San Francisco suggests that this problem will not be solved by expanding housing in those cities.  Generations have governed under these policies.

The blue lines represent Atlanta, Dallas, Houston, and Phoenix - four large metro areas that do not appear to impose broadly dysfunctional housing constraints.  Those cities have median incomes between the national level of $54,000 and about $61,000.  And population growth tends to be stable and stronger than the national rate of growth.  These cities portray the classical expectation.  Incomes may be slightly above average, along with housing expenses, but the ultimate result is an expansion of housing and an inflow of workers, holding down both rents and incomes.

Los Angeles may be approaching the growth level of the other problem cities, but, at least in absolute numbers, it has continued to accommodate growth.  It is not as pure an example of the problem as Silicon Valley and New York City.  It does have very high rents and home prices, but, it doesn't have particularly high incomes.  It's at about $61,000, similar to Dallas and lower than the other high cost cities.  It could be that housing constraints have been high enough to push rents above the comfortable level we see in unconstrained cities, but not so far as to lead to the sharp population shifts and limited access rents to wage-earners.  Or the main factors could be differences in their population from other sources, with LA taking in immigrants from Mexico while Silicon Valley is more tied to the tech. industry and innovations drawing from Stanford and Cal Berkeley.

Here is a graph of net domestic migration from 2000 to 2007, for selected metro areas, by core and outlying counties.  We can see the general trend toward the suburbs across cities.  At first glance, this looks promising for New York, but such a small portion of the metro area's population is outside the core counties, its outlying growth is not significant.

There are several striking issues visible here.  The two cities with the most out-migration are San Francisco and New York City.  They had more out-migration than Detroit, both with regard to their core areas, which are the country's primary sources of economic opportunity and innovation, and with regard to the entire metropolitan area.  San Francisco is so bad even the suburbs had net domestic out-migration.  Chicago looks similar to Detroit here.  It does not have particularly high average incomes or housing expenses, so while it isn't necessarily in a class with Detroit, its challenges are not related to housing constrictions.

Boston also shows up here with a pattern of domestic out-migration, and I find generally that Boston follows the same pattern as Silicon Valley and New York City, but on a smaller scale and with somewhat less of an affordability problem.

That graph measured net domestic migration from 2000 to 2007.  The next graph is only for the period from 2000 to 2003.  This is a short period, but the rates of migration for the cities included in these Census reports correlate highly with the 2000-2007 numbers, and for this period the Census bureau has a report that covers net domestic migration for the top 20 metro areas.

These numbers also correlate somewhat with broader decadal population changes.  Net domestic migration tends to produce tighter relationships with incomes, housing costs, etc. than the more broad decadal population growth rate.  Partly this might be unfair to the coastal cities because there may be some persistent flow of immigrants into those cities that becomes domestic out-migration as the immigrant populations assimilate.  But, I think the reason the relationship is more clear is because the status value created by the supply constraints bring wealthy foreign tenants and the economic rents caused by housing constraints draw high skilled foreign labor to those cities, which puts more pressure on domestic legacy populations who have to move away due to cost issues.  The difference between Silicon Valley and Detroit, for instance, in this regard is fairly obvious, and the tensions it is creating in Silicon Valley and Manhattan are palpable.  So, I think the net domestic migration measure is the more meaningful one.

Even there, I think there are distortions that cause a distortion in these graphs.  Among the cities with less population growth than average, there are two groups - new economy cities with housing constraints, which can be identified by their high housing costs, and old economy cities without housing constraints, which do not have particularly high housing costs.  The old economy cities still tend to have high incomes, though.  The classical description of what is going on here would be that populations flow to where incomes are higher.  Why aren't they flowing to these rust belt cities, then?

These cities represent the previous generation of new-economy centers.  Housing wasn't particularly common as a limit to broad job market access.  But, in those cities, there were excess profits created by geographic advantages, the legacy of large capital expenses, and the sort of path dependent competitive advantages that accumulate as competition removes less competitive firms.  In those areas, instead of housing, tactics like unionization, higher minimum wages, and local regulatory and tax burdens limited access by outside labor and extracted some of the excess profits of the geographically captured firms for broad local benefits and political nest feathering.

Few are moving to the old-economy cities because of high income employment opportunities, and they don't feel like high income cities.  I think the high average incomes there are probably high mostly due to compositional effects.  Whereas high income workers are moving into places like New York City, it is low income workers who are moving out of the old-economy cities because of lack of opportunities.  Competitive frictions mean that some of the legacy industry still remains, and still provides some excess income to those who are fortunate enough to still be accruing the economic rents that have driven so many others out of town.

I think this is a danger for the new-economy cities.  Will they be the Detroits of 2075?  Can Seoul or Hong Kong end up capturing the cutting edge of the technological niches of the future?  The high costs of Silicon Valley may cause that to happen sooner rather than later.  And, as we see in the old-economy cities, the unraveling of an economy whose members have taken rents as a birthright is not particularly pleasant.  Some of the geographical advantages that originally gave Detroit a competitive edge for automaking certainly remain.  But, they are just not as large as they used to be.  The problem is that rents are sticky, but competitive advantage is not.  The housing problems in Silicon Valley and New York City are having a negative effect on all of us now, but those cities could eventually see consequences much worse than a costly housing market - rot and decline.

The way those graphs should look is that the regression line should be flat or upward sloping.  This is clearly the case in aspirational countries like China.  In China there are massive inflows into the cities, where incomes are much higher.  As the housing stock strains to keep up with the inflow, costs rise.  Eventually, costs equilibrate with incomes.  The less friction there is in the ability to expand housing, the less incomes and costs in the city will tend to rise - population rises instead.

The difference between housing constrained cities and growing cities that aren't housing constrained is stark.  In housing constrained cities, locals complain of poor people getting pushed out of their neighborhoods and rich people ruining the character of the city.  In non-housing constrained growing cities, locals complain of poor immigrants and minorities moving in and ruining the character of the city.  In housing constrained cities, the locals try to block outsiders from housing, and in non-housing constrained cities they try to block outsiders from the job market.

I will have more on this topic in the posts following this one.  Outside these problem cities, the US matches the classical expectation.

Tuesday, October 27, 2015

Housing, A Series: Part 73 - A Fable, with charts and tables

There once was a land called Opportunity.  It was an egalitarian place.  There were 150 million households that lived there.  They each earned $50,000 per year, spent $12,500 a year in rent, and lived in homes worth $150,000 each.  One quarter of the citizens were landlords, and maintained 4 houses, earning the same $50,000 per year as the other citizens.

It was a mousetrap based economy.  The capital city, New Califork, was located near the mines that contained the purest iron for their springs.  So, 30 million families lived in the sprawling metropolis.  It was such a vibrant place, full of ideas, that each year, New Califork's factories churned out mousetraps that could catch 10% more mice than the traps from the year before.

Opportunity had gone on like this for a while, with families across the land enjoying the abundance of more and more captured mice each year.  But, then one day, a terrible cyclone hit New Califork and picked up 15 million homes, like the storm from the Wizard of Oz.  When the storm had passed, 15 million homes had been scattered across the country.  Cleaning up after the storm was a long process, and for some time they were too focused on recovering from the storm to think about rebuilding, and they put a moratorium on replacement homes in New Califork until things got back to normal.

For the most part, they made due.  But, the largest problem was in New Califork.  The city still possessed all of its high quality iron, so there was more iron than the remaining workers could use.  Eventually, all the best scattered workers decided to try to move back to New Califork.  They started to offer the landlords more rent to move into homes in New Califork, and sometimes landlords would agree, sending some other unfortunate family out to live in the countryside.

Eventually, they managed to realign all of their activities so that they managed to maintain their mousetrap output.  But, the workers who had braved their way back to New Califork knew that they had made it because they were best mousetrap makers, so they demanded $100,000 annual salaries.  Their bosses complied.  Other mousetrap manufacturers were not able to move in to compete with them anymore, so the mousetrap factories that increased their wages to get the best workers could pass their costs on through higher prices.  Of course, the workers had promised the landlords higher rents in order to get back to New Califork, so they now spent $25,000 on rent.  The landlords didn't see any foreseeable end to this state of affairs, so the houses in New Califork now had a value of $450,000.  Eventually New Califork was divided between the new $100,000 workers and the old $50,000 workers, though, of course, rents had risen for all of them.

With so many fewer workers and less competition in New Califork, the mousetraps started only improving by about 2% per year.  And, what other changes did Opportunity see:

Opportunity after the storm is much like the US, and New Califork is a rough approximation in scale and levels of New York City, Los Angeles, San Francisco and San Jose.  The largest effect of the storm, over time, will be the lower productivity caused by the lack of workers in New Califork - the loss of rising quality of life that Opportunists will never know they missed, like the robotic, self-setting mousetrap that was never invented because its creator ended up working in the cheese farms in Opportunity's countryside instead of moving to New Califork with her brilliant idea.

Among the immediate effects are an extra 6% of inflation due to rising incomes and rents in New Califork.*

But, the two most significant effects are, (1) a large jump in the incomes of the top earners and (2) a decent jump in the value of homes.  The effect on high incomes is probably worse in the US, because New Califork has no income inequality, but, in the US, these cities tend to have the most positively skewed income distribution.

I could have added firms to the mix in Opportunity.  The effect on firms would have been similar to the effect on workers.  Firms located in New Califork would have tended to earn more income, so that there would be more variance in firm profits after the storm.  Workers with higher wages would have worked for firms with higher profits.  We also see this in the US.

The housing value may be unexpected.  Housing demand tends to be somewhat inelastic.  That is why rents in the constrained cities tend to represent a larger portion of incomes.  Normally, this wouldn't matter much, because in the long run, housing supply should be very elastic.  New homes will be built at roughly the same cost as existing homes, regardless of how much house people demand.  But, the problem in New Califork, and in metropolitan US, is that locals prevent new housing from being built.  This creates the odd result that by building those houses in less valuable locations all over the country instead of in our best cities, we actually raise the face value of the housing stock.  When the supply of homes is reduced, the rise in prices (in terms of rent) is steeper than the fall in quantity, so total dollar value increases.

Opportunity held a vote.  They were concerned about their newfound inequality and their lack of mousetrap growth.  A lot of people said that the low growth was caused by the inequality itself, because regular families didn't have as much money to spend on mousetraps anymore.  Should they allow new houses in New Califork again, or should they implement a tax and redistribution regime and an anti-inflation program?  They voted overwhelmingly for the tax and redistribution regime.  The backers made very good points.  They said that with all of the new income inequality and high inflation, nobody would be able to afford the houses they might have built.  Not only had inflation driven consumer prices up by more than 6%, but home prices, which the government of New Califork sneakily doesn't include in inflation figures, were up 20%.  So, regular families couldn't afford much of anything, let alone houses.  Maybe they would be able to after some income redistribution and a little less inflation.  They also noticed that some firms were starting to capture persistently higher profits.  Nobody was able to pin down the reason, but economists know that high profits can come from what is called "rent seeking" - a sort of unearned income that can come from government favors.  It was a suspicious development, and confirmed for the Opportunists that the new taxes were being taken from people who didn't deserve it anyway.  It was the only solution that made sense, if you think about it.

Of course, the New Califork landlords were all for building new homes.  They were making money hand over fist, and they were all greedily thinking of all the additional rent they could make with more houses.  But, with income inequality such a big problem already, obviously the last thing Opportunity needed was for the richest Opportunists to have even higher incomes.  The fact that some landlords were vocal about their support for new homes in New Califork probably drove more support to redistribution, if anything.  When people are suffering, it really is unseemly for those who have so much to be pushing their weight around for self-serving public policy.  And those middle class $50,000 workers still hanging on in New Califork said, "You mean, those landlords who have been forcing all of my friends and neighbors out of town so they can rent their houses to these rich people moving in?  They want to build more houses so they can move more rich people in and drive up rents even more?  Uh. No."

I visited New Califork again, some time later.  But, you know, after that storm, it just seemed like they were running to stand still.  After that, it just seemed like when they did have some good times, all the income went to those high income workers, and everything - especially houses - would just get more expensive.  The last I heard, they were so disturbed by the amount of money families were spending on housing that they put a stop to building across the entire country.

*  The problem in the US is worse than this, because in New Califork, families are still staying in the original units.  In the US, in 2005, the typical family moving to one of these cities not only raised their rent from about 25% to 33% of their income, but even to settle with that large cost increase, they have to downsize sharply.  So, our rent inflation is higher than Opportunity's.

Monday, October 26, 2015

Housing, A Series: Part 72 - Bankers have a numeraire problem

Following up on yesterday's post, I think there is an interesting problem regarding mortgage financing and duration matching, and it ends up relating to the problem of the credit crisis and bailouts.

I can't remember the link, but recently I saw a nice piece about the very long term performance of bonds.  One of the interesting parts of the piece was a graph of short and long term bond yields.  In the 19th century, when we used various monetary regimes that tended to be tied to gold, etc., short term yields were volatile, and long term yields tended to be lower than short term yields.  This is because there tended to be currency shocks and bank panics, but over the long term currency value was tethered to a commodity, so there was significant mean reversion.

But, with a fiat currency, there is much less mean reversion.  So, I think it is more useful to think of long term bond yield spreads not so much as reflecting maturity risk or inflation risk, but instead as numeraire risk.  The numeraire we use is the dollar.  Since there is only one source for dollars, by law, then long term bonds really have Federal Reserve risk.

This is where banks are unavoidably intertwined with the federal government.  There are different kinds of risks that damage them with any sort of numeraire instability, whether it is deflationary or inflationary.

The risk that reared its head in this crisis was somewhat unique.  We informally implemented a set of monetary and regulatory policies that were based on the presumption that the collateral behind bank assets needed to be devalued.  There was never a policy statement that said as much, but we thought we knew prices were too high, the Federal Reserve pushed interest rates to levels that usually lead to economic contraction, and as home prices fell, everyone looked at each other and nodded.  Even now, talking about his new book, Ben Bernanke reviews that period of time and says, "Well, again, we were aware of the fact that house prices were very high. And we thought it quite possible that they would correct at some point."  And everyone nods, and says what a shame it was that the banks did this to us.

In 2006 and 2007, the Fed mostly pushed real estate values down by aggressively selling treasuries and reducing the present value of future rent payments with high real interest rates.  Eventually, demand was so hampered that even expected future nominal rent cash flows declined.  And, by the end of 2008, access to mortgage credit had been so undermined that home prices ceased to even reflect basic interest rate or rent expectations.  This is still the case.

So, while there are significant governance problems with all of the discretionary crisis-mode provisions that were made, if I think about even a firm like AIG, especially after the shareholders had been wiped out, I'm not sure there was a more appropriate place to set the risk of what remained than on the backs of taxpayers.  AIG basically took a large, unhedged position on the idea that the federal government wouldn't lose its mind with regard to managing the numeraire on financial contracts.  That clearly wasn't a good risk to take.  But, given that we have little choice about what numeraire to use, it seems reasonable to me that we should match the consequences of numeraire risk to the source of that risk.

Really, this is the reason Fannie and Freddie exist.  The problem isn't so much that it is hard to match maturities between mortgages and houses.  The problem is that homes are real assets (their value grows with inflation) and it would be very difficult to create a mortgage contract based on real terms.  Actually, some of the negative amortization mortgages that were being used late in the boom would address this problem, but they aren't exactly politically popular.  Since it is difficult to create mortgages with real terms, what banks have is a numeraire matching problem.  Even the problem of prepayments, which is a major factor in mortgage valuations, is a numeraire problem.  Most of the potential fluctuations in rates that lead to prepayments and refinancing are related to persistent inflation variation.

So, a default crisis that is bad enough for Fannie and Freddie to fail is probably due to numeraire shocks.  This is why I think, as long as we have a discretionary fiat currency regime, we probably need institutions like Fannie and Freddie.  And, if we need Fannie and Freddie, they probably should just be public institutions.

This is one of many problems that NGDP level targeting would solve.  It would create mean reversion in yields and currency values and would reduce numeraire risk.  I think it would lead to a less steep yield curve spread, which should lead to more long term investments.  But, I think if it worked well and became permanent and trusted, it would remove the numeraire problem in the mortgage market.  There wouldn't be as much of a problem regarding the matching of nominal mortgages with real homes.  There might still be some reasons for securitization, but there could be more liquid and safe private markets, and many of these governance problems in finance would be decreased.

Sunday, October 25, 2015

Housing, A Series: Part 71 - Pushing Back against the Moral Hazard Meme

Scott Sumner has a post up at EconLog this weekend about the timeline of the financial crisis.  He brings up the too big to fail / moral hazard issue.  This is something others like Russ Roberts have focused on.

I would like to begin by saying that, in principle, I am completely on board with this issue.  A regime that includes informal and implied subsidies for risky behavior is dangerous.  We would expect that regime to lead to more crises and for there to be indecision and power brokering during those crises.  It's not an ideal set of policies to build a financial sector around.  I don't really have a problem with the historical examples of this issue.

But, on the other hand, my understanding is that the idea of a lender-of-last resort role for central banks is a longstanding legitimate function.  This would especially be the case if the crisis was a liquidity crisis that the central bank is charged with avoiding.  In some ways, there has been an unfortunate standard of using sloppy language to identify all of the Fed machinations toward the end of the crisis as "bailouts", and this has created a climate where legitimate stabilization operations have been tainted with negative semantics.

But, I want to take this even further.  Scott has pushed back against the idea that there was a "bubble".  I have been digging into the evidence all year, which I think confirms his position.  Very little of the rise in home prices (and, the accompanying mortgage growth) requires any demand-side explanation.  The banks weren't systematically pushing mortgages on households outside the traditional population of homeowners, and those households weren't moving into homes with rental values any higher than usual.  The boom was a reaction to supply constraints in favored locations.  The banks weren't the cause, they weren't taking on excessive risks, and the bust was not inevitable.

That last part is key.  Practically all analysis of the crisis begins by begging the question - the bust was inevitable.  And, even though Scott's EconLog post is reasonable, as far as it goes, I think he gives in too much to that notion.

This crisis was not a good example of the problems of moral hazard or of a fragile banking system.  In fact, I think some evidence points to quite the opposite, and instead, we should be surprised by how devastating and persistent Fed policies had to be in order to create broad economic dislocation.

By 2007, there was a strange mood in this country that housing speculators and bankers needed to be taught a lesson.  We assumed that home prices must be based on overly optimistic projections about the risk and reward of home ownership, so watching nominal home prices fall by a proportion unprecedented in modern times was not only acceptable, it was required.  These speculators needed to learn that unprecedented declines in price happen, and asset prices need to reflect these black swans.  But housing wasn't in a demand bubble, so a tremendous amount of damage had to be inflicted on the American economy for home prices to collapse the way they did.

About 1 million units per year has been a post WW-II floor that we generally only hit briefly before recovering - even in the 1950s and 1960s when US population was about half today's level.  By the time Bear Stearns failed in March 2008, housing starts were falling below a 1 million unit rate.  National home prices were down 7% and home prices in the major coastal cities were down 16%.

By September, when Lehman Brothers failed, housing starts were falling below a rate of 800,000 units.  (They would remain below 1 million units for 5 more years.)  Home prices were down 13% nationwide and 25% in the major coastal cities.

Even then, it isn't clear that economic problems would have expanded so broadly, even if the Fed let Lehman enter bankruptcy, if they had simply supported the nominal economy (as Scott suggests they could have done with Continental Illinois in 1984).  But the day after Lehman's failure, the Fed famously took an unnecessary discretionary hawkish position followed by a questionable new policy of paying interest on reserves that they raised briefly to over 1%.  Lehman was not the critical factor in the subsequent collapse.  If they had managed a buyout of Lehman, but had followed it with those same hawkish interest rate policies, it seems probable to me that the outcome would have been very similar to what we experienced.

It took a 13% dive in banks' most stable and extensive base of collateral (25% in the most active real estate markets) to create the financial crisis.  Call me crazy, but if you had told me in 2006 that home prices would need to collapse by that much in order to create a financial breakdown, I would have responded, "Really?  You think our banks are that stable and well managed?"

As I have said before, if home prices had fallen by 5% and a financial panic would have ensued, then, yes, that would be evidence that our financial system had been fragile.  But that is not what happened.  And, given what did happen, the only reason we accept the fragility story is because we assume that a 13% drop in home prices was inevitable.  (This eventually grew to 26% nationally and 34% in the major cities.)

Do I really have to beat back moral hazard arguments to say that it is well within the Fed's mandate to stop home prices from falling 34%, 26%, or even 13%?  Really?  Do we really need to worry about banks being too reckless in the future because they will be blasé about national double digit declines in real estate values?

Look, in the broad scheme of possibilities, we don't have to base our real estate markets on nominal mortgage contracts.  But we do.  And, as long as we do, banks will have some highly leveraged real assets on their balance sheets.  It would not be realistic to expect them to manage their balance sheets to withstand national 25% declines in real estate values without expecting some nasty system-wide repercussions.  In the previous century, nobody would have even suggested that we should demand that level of safety.

I support the idea of NGDP level targeting, or wage level targeting, as Scott lays it out.  Isn't it funny that nobody worries about the moral hazard of having a policy where aggregate wages never decline by a penny?  I think the main advantage of NGDP level targeting may be that it indirectly leads to stability in capital incomes that we cannot support directly.  In a world where even just implementing stabilizing monetary expansion in the middle of a deflationary shock is labeled by many as a Wall Street bailout and where we are willing to watch millions of middle class households lose their homes without screaming for monetary expansion because the banks had it coming, its clearly not popular to support Bourgeois concerns.  Because of this, I don't think we can fault the Fed too much for the crisis.  If they were blamed for bailouts for their efforts after September 2008, imagine the outrage that would have been aimed at them if they had dared stabilize the economy in 2007.

I would add that securitization had risen until the early 1990's and had levelled off since then.  Securitization was not at the center of the housing boom.  In fact, during the height of the boom, banks were increasing their share of held mortgages outstanding.  Securitization only began expanding again in 2006 and after because the liquidity crisis was hitting bank balance sheets.  So, even the moral hazard issue created by securitization itself is less central to the recent crisis than it is generally made out to be.

And, many small banks failed in the crisis.  Does too-big-to-fail even have that much to do with what happened?

The moral hazard issue is a real issue, but it is the wrong lesson to take from the crisis, and the treatment of Bear Stearns and Lehman Brothers are taken to be larger factors in the collapse than they deserve to be, in my humble opinion.

Friday, October 23, 2015

Housing, A Series: Part 70 - The cognitive bias at the heart of our hobbled economy

Maybe human beings will always be uncomfortable with voluntary exchange.  There is a pretense in political economy that there are free market ideologues who ignore inhumane power imbalances and there are those to oppose markets in the name of egalitarianism.  This is wrong on both counts.  Markets can be one of the most dependable threats to entrenched power.  And, it isn't the power imbalance itself that tends to make us feel uncomfortable about markets; it's the inscrutability of that power that bothers us.  We don't tend to be bothered by power imbalances that are part of a stable legacy or process.

So, we are more comfortable when pre-selected market makers pay millions of dollars for the right to manage a 25 cent spread on traded equities than we are with high frequency traders who spend millions of dollars for the ability to capture a 3 cent spread with unpredictable, volatile trading activity.

I might go so far as to say that the recent crisis was a self-inflicted wound created by this bias.  New readers may consider that to be a dumb thing to say.  I've spent the better part of this year discovering the evidence for the "self-inflicted" part.  The funny thing about the bias part is that the evidence for it seems so outrageous and clear, I wouldn't know how to convince someone who isn't convinced already.

I have posted this graph before, which shows the proportion of mortgages outstanding by type of holder.  It shows a few things that either most people seem to believe in error or that they just don't know.

First, there was a significant shift to securitized mortgages (Ginnie Mae, private pools, and other agencies) from the late 1960s until the early 1990s.  Then, the securitized share of mortgages leveled off.  Home prices were, if anything, fairly low at the point when the relative use of securitized mortgages leveled off.  There was no relationship between the Price/Rent ratio of homes and the growth of securitization, and there was little change in the share of securitized mortgages during any of the boom years from the mid-1990s to the height of the boom.  The share of these pools was 57% in 1995 when rent inflation began to rise, it peaked at 62% by 2002 before the steepest moves in home prices, and then declined back to 59% at the end of 2005 when housing starts and home prices peaked.

Within this group, there was a shift to private pools, much of which were subprime.  But, as we can see in the graph, there was a gradual shift from Ginnie Mae to private pools from about 1990 to 2003.  These are both pools for borrowers who can't generally qualify for conventional loans.  Ginnie Mae covers the credit risk of those loans by making borrowers purchase mortgage insurance while the private pools mimic the math of insurance by charging borrowers a premium rate and paying out that premium to the riskier tranches of the securitization.  Much of that shift happened while real home prices were falling in the 1990s and homeownership rates were level.  It continued at about the same pace in the late 1990s when homeownership began to rise while prices remained stable, and continued still until the end of 2003 when homeownership and home prices were both rising.  In fact, during that period, the decline of Ginnie Mae accelerated and private pools were not capturing all of the shift away from Ginnie Mae.

After 2003, the GSE's began to decline as a portion of the market also.  It was during this period that private pools shot from about 10% to about 20% of the market, until the private pool market collapsed in 2007.  This period was not associated with a rise in homeownership, and included the last period of sharply rising prices followed by two years of flat prices.

In short, private pools were mostly an alternative to Ginnie Mae, which had always made loans with many of the characteristics of subprime loans, and there is no obvious correlation between their increased use and any particular behavior in the housing market.  Of course, if you have read the previous 40 parts or so of the series, you already know that demand-side factors can't be responsible for more than a very small fraction of home price movements at the time, so this idea shouldn't be surprising.

I've taken too long to get to my point.  The point is that our bias about these things is so strong that just having that little sliver of mortgages outstanding shift from a public process to a private process caused everyone to have a freak out.  Ginnie Mae helps "make affordable housing a reality for millions of low- and moderate-income households across America by channeling global capital into the nation's housing markets....What Ginnie Mae does is guarantee investors the timely payment of principal and interest on MBS backed by federally insured or guaranteed loans."  What a great program, right?  Low income households have to pay a little more for Ginnie Mae loans, but they get access to the world of homeownership.  And taxpayers make good on any defaults so that investors can rest easy.  Great idea.  This is exactly what private MBS were doing, except the investors were taking the default risk instead of taxpayers.  But, we know one is benign and the other is predatory, right?  That's just a thing that we know about them.  That makes them different from each other.  And the outcome proves it, right?

Now, it would need to be admitted that the profit motive can be insatiable, so the problem with the private version of this process is that it will expand and suck in victims until the system is unsustainable.  That's what greedy financiers do, m'I right?  That would definitely have to be admitted if it were true.  During the period where private pools suddenly jumped from 10% to 20% of the mortgage market, homeownership rates peaked and began to fall, and all the net new homeownership was in the top 40% of households, by income.

We need to sit on this piece of information for a second.  As I have sifted through the data on this issue, time and time again I have been floored by the gulf between what we think we know and what the data says is true.  This data is from the Survey of Consumer Finances that the Fed conducts every three years.  Nobody here is lying on a mortgage application in order to qualify for their loan.  They just happen to have conducted the survey in 2004 and 2007, which nicely bookends the period where private pools doubled from 10% to 20% of outstanding mortgages.  Homeownership among the bottom 60% of households declined during this period.  The proportion of mortgage debt held by the bottom 80% of households declined during this period.  The majority of net new mortgages during this period was from these private pools.

Here are a couple more graphs.  The average owner's home, in terms of rental value, was declining during this period, while renters' rent was rising.  The systemic movement of households into too much house simply did. not. happen.  Finally, here is a graph (in 2013 dollars) of the marginal new mortgage debt outstanding, by income.  There was some increase in debt among the bottom 40% of households before 2004.  This was before the sharp upturn in private mortgage pools.  During the sharp upturn in private mortgages, even in terms of absolute levels, almost all of the new outstanding mortgages were to households in the top 40% of incomes.

This isn't just a rounding error or a slight difference in scale.  When you look at these graphs, your priors should tell you to expect a huge, unmistakable bulge in low income borrowing, pointing to the defining story of our time.  Not only is it not there.  The data tells the opposite story.  The defining story of our time is fiction.  The best villain stories frequently are.

And, let's think about the criticism of these private pools and the various securities they supported from an investor perspective.  They are criticized both for (1) predatory lending that saddled low income borrowers with rates that were unsustainably high and (2) unrealistic assumptions about systemic risks that led ratings agencies to apply too optimistic ratings to the securities and investors to demand too little of a rate premium for them..........

As with many opinions about the housing bubble and the recession, these ideas achieve a sort of zen level of incoherence.  They manage to be wrong while also contradicting each other.  The first part (ignoring the fact that it can't be true if the second part is true) is wrong because borrowers always had the option of going through Ginnie Mae, and in the aggregate these pools were clearly outcompeting the terms of Ginnie Mae loans.

On the second part, if this was true, the consequence wouldn't be a total collapse of the MBS market followed by massive defaults of borrowers.  The consequence would have been that MBS buyers would have demanded higher premiums and borrowers might have transitioned back into Ginnie Mae loans.  This wouldn't stop investors from offering to purchase new securities with higher premiums.  But investors didn't just adjust their risk spreads.  Instead, capital just dried up.

This year, Ginnie Mae reduced their mortgage insurance premiums, and earlier this year the President announced (HT: Nick Timiraos) that the lower premium would keep more money in the pockets of new homebuyers.  Of course, since this effectively acts as a decrease in the real interest rate on the mortgage those households are using, in an efficient market we would expect there to be some shift from other loan types into Ginnie Mae/FHA loans and some increase in home prices.  According to the article, this is exactly what happened.

Now, in this case I would not have been surprised if there hadn't been a price reaction, because home prices since the crisis don't seem to have been efficient.  But, a commenter on Nick Timiraos' tweet posted this interesting graph on the total interest rate (with insurance premium) over time on FHA/Ginnie Mae loans.  Public policy has been procyclical on banking matters, so a large premium increase was implemented in 2008 after the crisis, and even now the premium remains high.  This put the effective interest rate on FHA loans at levels not seen since the 1980s.  It could well be that low home prices on the types of homes dominated by these loans are have been in equilibrium with the interest rate level implied by FHA loans.  In fact, what we would expect to see in that case is a market dominated by all cash buyers, because they would require a lower implied yield than buyers who only had access to these high rate loans.

For the intrepid reader, I tentatively suggest reading the transcript of the President's speech.  It is a litany of how activist government programs, non profits, and the toughness of the President's affiliates has brought back the housing market, and how lenders are predators, how the President is clamping down on them, how they did this to us.

There are predators and profiteers.  Then, there are politicians, their supporters, and non-profits, who cause recovery, who make things happen.  Then you get jobs and production as a result of that work, which, you know, the President is so generous, he isn't even going to begrudge some profiteer if they benefit from the hard work he did along the way to save our housing market.

I'm not that interested in critiquing the speech.  It's a religious speech.  The sole purpose of that speech is a relentless lining up of sinners and saints, insiders and outsiders, doers and takers.  What is politics, after all?  I'm interested in how these cognitive biases are manicured and intensified.  In community, human beings are willing to believe really insane things without evidence when our minds move into affiliation mode.  A signal to move into that mode is language that is broad and absurd.  You don't argue about how many goats need to be sacrificed in order to ensure a good harvest on factual grounds.  The absurdity of the conceit prevents it.  Our minds are built to be taken in by this language.  When we hear language with generalized absurdities, we switch from practical, sane skepticism to a question of whether the speaker is one of us or one of them.

Now, don't get me wrong, the FHA needs to reduce their premiums even farther.  But, think about the President bragging, to applause, about lowering interest rates to get low income families into homes.  There is one while lie there, that that money stays in the buyers' pocket as opposed to going to the seller.  But, can you imagine the President getting applause for giving a speech for exactly the same loan with exactly the same terms, but through private markets?  That's supposedly one of the reasons we had the crisis, right? Those villains were lined up as the cause of the crisis by 2001, if not sooner.  We applaud for the benign public servants.  We aren't going to applaud for the people who did this to us, and who we knew all along were going to do this to us.  "Why aren't there more bankers in jail, Mr. President?"

Ignoring all the evidence I have gathered that undermines the demand story of the boom, what we have here is a mortgage program set by public policy, with effective interest rates moved up and down somewhat arbitrarily.  No wonder private markets that can fine tune the assumption of risk and adjust spreads daily were outcompeting Ginnie Mae.  It would be very difficult, in practical terms, to justify setting mortgage spreads this way,  But, we feel safer with it.

Imagine if we set, say, gas prices this way.  About 1/3 of the time, we would be hoarding or waiting in line for gas rations, until the pricing committee happened to meet and guess the correct price.  I wonder how much these high mortgage insurance premiums have been holding back both home prices and mortgage expansion.

PS.  The Federal Reserve is in a funny position here.  It is public and, so, benign.  And it also has the impossible task of trying to use a committee to guess the price of something important.  But, it is, for some reason, tainted by its connection to financial villains in a way that Ginnie Mae isn't.  So, while nobody can agree whether the Fed is loose or tight, and nobody can agree whether low rates are due to tight money or loose money.  Just about everyone can agree that, whatever the policy is, the Fed clearly set it to favor Wall Street over Main Street.  High or low inflation, in combination with either high or low interest rates, all favor Wall Street.  We really do live in times where the common man cannot catch a break.

I was happy to see Bernanke push back against this today, "It's ironic that the same people who criticise the Fed for helping the rich also criticise the Fed for hurting savers," Bernanke told the FT's Martin Wolf. "And those two things are inconsistent. But what's the alternative? Should the Fed not try to support a recovery?"

The sad fact is that, unwittingly, these errors of judgment do lead us not to support recovery.  I am afraid, though, that I must come to the conclusion that Ben Bernanke himself was drawn into this error, and played a role, not just in having a slow recovery, but in needing a recovery at all.

Thursday, October 22, 2015

Housing, A Series: Part 69 - Migration seems to mostly be between metro areas, not within them

Recently, I compared the Zillow Home Price Index (ZVHI) to the S&P/Case-Shiller National Index to comfirm what I had found in the BEA/Federal Reserve data.  Since Case-Shiller tracks individual homes, it was overstated in the 2000s as a national price index because households were purchasing marginal new homes that were less valuable as they migrated out of high cost cities.  The difference between Zillow and Case-Shiller did seem to confirm that question.

This is a graph from that post.  As an aside, I have added two additional series here.  These are estimated ZVHI indexes for the country with several of the problem cities removed.  In previous posts, I have compared Price/Rent ratios of the C-S 10 cities with the rest of the country, but that understates the effect of those cities, in a way, because in those cities both rent and Price/Rent ratios were rising.  This graph of prices gives a sense of the full effect these few cities have on national data.

The last version, with San Diego, Boston, and Washington, DC removed, is only slightly different than the version that only removes NY, LA, and San Francisco.  Those three cities really dominate the issue.

Housing outside those six cities, at its peak, was up 64% from 1996 levels, and in 2015 is up 50%.  Over 19 years, that is an average annual growth rate of 2.2%.  This represents 85% of the country.  It is crazy enough to attribute peculiar behavior in a single market to national credit or monetary policy.  We have been attributing peculiar behavior of 15% of households in a single market to national monetary and credit policies.

Now, we can compare city specific data between Zillow and Case-Shiller.  If the data for individual cities has the same drift, then either (1) the drift is coming from some other source unrelated to migration patterns and changes in the housing stock or (2) at least some of the migration is happening within metropolitan areas, from central cities to suburbs.  If there is no drift, then this would appear to confirm that the drift we see in national data can be attributed to migration out of high cost metro areas into low cost metro areas.

And, the data very clearly lacks a long term drift within individual metro areas.  A few of the cities have slight deviations during the 2006-2010 period, which seem to mostly be related to how dislocations from the crisis, such as foreclosure sales, affected measurement.  These devations generally disappeared after the crisis.  There is no city with a drift over time, which suggests that the drift has to do with migration between cities.

Zillow also attributes the drift to their more comprehensive coverage in low-cost rural areas.  Making some very broad estimates, it looks like each of these factors might contribute about equally to the drift over time in the measures - about half due to the concentration of Case-Shiller in higher priced markets and about half due to basis shift from migration out of higher priced markets.  We can say that, however much the drift between Case-Shiller and either Zillow or the BEA/Fed is related to basis shift as households moved to lower value homes, this shift appears to have happened mostly between cities, not within them.

On the individual city graphs, I have kept the scale the same so that it is easy to see the large differences in metro area home values, and also how much of those differences have developed over the past 20 years.

New York City doesn't look that pricey here, but it really is composed of three parts.  (1) Manhattan, which is similar to San Francisco and San Jose, (2) Brooklyn, Queens, and the eastern suburbs, which are similar to Oakland, Los Angeles, and San Diego, and (3) other suburbs, generally to the north and west, which are more similar to Detroit or Baltimore.  To the extent that I can find enough data to get down to the county level, I hope to get more flavor about the different parts of the puzzle.