Monday, December 31, 2018

Yield Curve Watch

It looks like the market expectation is that this is the cyclical high point for the Fed Funds Rate.  It will be interesting to see if the FOMC insists on any more hikes.

In the meantime, the yield curve has become quite inverted.  Here is a chart of Eurodollar futures, which I like because it has a longer duration than Fed Funds futures.  The higher line is the yield curve on November 8, at the high point.  Even then, it was slightly inverted.  But, since then, even though near-term Fed Funds expectations have fallen, the yield curve in the 2-3 year range has fallen more.

Here, you can see how, at these low rates, there is a natural upward slope to the yield curve because the zero lower bound creates asymmetry in the expected yields on longer durations.  If you are using 10 year treasuries or some other longer term yield to estimate the yield curve, then you are getting a false signal.

I would say that, at this point, barring an unlikely additional bump in long term interest rates, the question is only how hard the landing will be, and that depends on how quickly the Fed reverses course.  It would be prudent if at the January meeting they pulled back 25 or 50 basis points, but that doesn't appear to even be in the set of potential options.  That would be the only chance at getting the "normalization" to 5%+ that I hear people talking about in long term interest rates.

It seems like the prudent position to take here is to maintain defensive positions until interest rates head back toward zero, and be ready to transition to equity at some point after the Fed starts to chase the natural rate down.  There could be a lot of noise between now and then, but it seems likely that in a year or so, equities will be available at prices near or below today's level and Treasury yields will be lower. (These are poorly informed opinions.  Do not use this blog for investment advice, etc. etc.)

Friday, December 14, 2018

Housing: Part 338 - Price/Rent ratios

One of the key ideas that fuels conventional wisdom about the financial crisis and the housing boom is that Price/Rent ratios (or, relatedly real home prices) shot way up outside the norm during the boom.  This seemed to be proof that credit markets were fueling an unsustainable price boom.

One of the key discoveries I made was that, oddly, even though rent is in the denominator of price/rent, it has such a strong effect on price that when rents rise, the price/rent ratio rises even more, and likewise, real home prices would rise even more.   2018   (income on log scale)
I think I have posted some version of this graph before.  But, before, I have just shown r-squared values.  This version of the graph shows 1991, 2007, and 2018.  And, in addition to the r-squared values, I looked at the p-values.  I was surprised at how small the p-values are.  And, these are not weighted by MSA size, which I suspect would lead to even higher r-squared and lower p-values, because very large MSAs populate the far end of the regression.

The p-values are:

1991: .162 (not significant)
2007: 5.5 x 10-16(nearly zero)
2018: 2.5 x 10-35 (nearly zero)

And confidence levels are pretty tight.  The coefficient, at the 95% confidence level is (these are on a natural log scale, so this is the expected change in Price/Rent for each doubling of rent):

1991: -0.7 to 2.0
2007: 6.5 to 10.4
2018:  5.6 to 7.3

Interestingly, if I regress Price/Rent against the median income of each MSA, or against the median price, the relationship is very strong for every year.  I have written previously about how, within MSAs, there is a strong systematic relationship between Price/Rent and all three measures (rent, price, and income).  Within MSAs, each doubling in price is associated with a Price/Rent increase of about 3.  Between MSAs, each doubling of price is associated with an increase of 4 1/2 to 6 1/2.  Possibly, similar influences are at work, and the steeper relationship between MSAs is created because between MSAs, there could be an added systematic factor - expected rent inflation.

For each doubling of MSA median income, the 95% confidence range of the coefficient for Price/Rent is:

1991: 5.4 to 8.8
2007: 9.3 to 14.4
2018:  6.9 to 9.8   2018   (income on log scale)
Those coefficients are huge.  The median US Price/Rent in those years was 10.7, 14.7, and 12.3.  So, doubling the median MSA income is associated with a change in Price/Rent that is nearly as high as the national median Price/Rent.  A log-linear relationship would mean that the median home price in a city with a median household income of about $20,000 would be $0.  Actually, look around some cities today, like Cleveland, and it isn't too far off that.   2018   (income on log scale)
So, there has always been a strong relationship between income and Price/Rent both within and between MSAs, probably for similar reasons, such as that higher priced homes make better tax shelters, are more likely to be owner-occupied, have less credit constrained buyers, etc.  Incidentally, this is one reason why it has been really bad to block households from mortgage access because of low incomes, etc.  Homes in low-income neighborhoods are cheap.  It's the rare asset class where investors of lesser means have a natural advantage for getting higher yields.

But, the most interesting thing about this is the difference between the income effect and the rent effect.  I have concentrated previously on how during the boom (and since) rent has become more and more a factor in home prices at the MSA level, not less important.  So that rising Price/Rent levels were not actually a good signal of a bubble.

But, here, we can see that the reason that rent did become a more important signal was because rent and expected increases in rent, started to correlate with income, because of the Closed Access problem.  So, yes, rent has become increasingly important, but here, we can confirm that rent has become increasingly important only as a side effect of MSA income becoming more important and becoming rationed through rent.

Wednesday, December 12, 2018

Housing: Part 337 - Shelter inflation

This isn't anything earth shattering, but as I was updating this month's CPI numbers, I realized that I had never attempted to quantify the portion of shelter inflation that has been directly attributable to the five "official" Closed Access cities.

The first graph here is just a comparison of various annual inflation rates:
  • Grey line: Core CPI excluding Shelther
  • Black line: Core CPI
  • Green line: Non-Closed Access Shelter Inflation
  • Blue line: US Shelter Inflation
  • Red line: Closed Access Shelter Inflation
The main point to gather here is that, except for the foreclosure crisis, for the past 20 years or so, Closed Access rent inflation is pretty consistently in the 4% to 5% range.  During the housing boom, homes needed to be built in other locations, and the pressure pushing households into those homes from the Closed Access cities was continued demand for Closed Access homes.  That kept Closed Access rent inflation high, and the housing boom was facilitating the movement out of the Closed Access cities to further accommodate that demand.

As I have pointed out before, the top of the "bubble", in 2005, was the only point in the past 20 years where both shelter inflation and non-shelter inflation were both at approximately the 2% inflation target.  That was actually the closest we have been to a neutral monetary policy and residential investment level both at the same time.  As shown here, the decline in rent inflation at that point was entirely from non-Closed Access areas.  Then, the Fed raised rates to cut down residential investment, and non-Closed Access rent inflation moved back up.

During the recovery, the limit to building has been due to mortgage suppression, so it is nationwide, so rent inflation has been high everywhere - nearly as high in non-Closed Access areas as in Closed Access.

The next graph is a stacked graph.  Looking at the first graph, the gray and black lines are the same - core CPI without shelter and with shelter.  This shows how much of the gap is caused by non-Closed Access rent (gray to green) and how much is due to Closed Access rent (green to black).  The last graph is the three measures stacked again, but in reverse order.  First, the portion of US core CPI inflation that is due to Closed Access rent (red), then the portion due to non-Closed Access rent (red to green), then the portion caused by all other core inflation (green to black).

PS: One oddity is that, for non-shelter core inflation, the recession and immediate post-recession years are the only time that the measure was persistently near the target.

PPS: To clarify the stacked graphs, if non-shelter core inflation is 2% and shelter inflation is also 2%, then shelter inflation is shown as having no effect on core inflation.  The graph is showing how much of the gap between non-shelter core inflation and total core inflation is due to shelter inflation.

November 2018 CPI

Things continue to move sideways, not providing a strong new signal in either direction.  The next two months will be interesting, because core CPI excluding shelter last December and January totaled about 0.6% (not annualized).  Unless there is a similar statistical jump this year, by the end of January, core CPI will be back under 2% and core CPI excluding shelter will be back down close to 1.0%.  Potentially that could affect sentiment about future rate hikes.

For now, core CPI is 2.2%, Shelter CPI is 3.2%, and core CPI excluding Shelter is 1.5%.

Wednesday, December 5, 2018

Housing: Part 336 - Incomes and the Housing Market

Long-time readers have probably seen some version of this a number of times, but I have been poking around in the awesome Zillow data, and I don't think I have quite done this before.  I have posted individual cities before, but here, I have run regressions of MSA income against rent, prices, and various combinations of these measures.  I am trying to get a systematic time series representation of the importance of income on the housing market.  Here I have used the largest 64 MSAs.

In cross-sectional regressions against MSA median household income, from the 1990s to 2005, income became a much stronger predictor of both MSA median rents and MSA median Price/Rent.  It remains as strong a predictor today as it was in 2005.

Part of what has happened is that income has become a more important factor in MSA housing markets, and part of what has happened is that variance in incomes among MSAs has increased over time.

In the following graphs, the blue line is the US median.  The red line is the expected level for a city with median household income 1 standard deviation above the US median.  The green line is the expected level for a city with median household income 1 standard deviation below the US median.

There is a graph showing rents over time, price/income over time, and mortgage affordability over time.  This isn't news to any readers here, but:

1) The bubble wasn't driven by low-income markets.  Mortgage affordability was steady in low-income cities from 1995 to 2005 while it shot up nearly 50% in high income cities.

2) Whatever is causing housing starts to top out now, it sure as heck isn't high mortgage rates.  Mortgage affordability in low-income cities is well below any pre-crisis level.

The thing about low mortgage rates is that a low interest environment actually has some redistributive qualities.  Think of the housing market.  Home prices are somewhat sensitive to long term real interest rates.  So, when rates are low, people with wealth must pony up larger sums to purchase a home.  But borrowers shouldn't really care so much about the price.  If they can borrow cheaply, their liabilities and assets get matched up, and they can take out a mortgage with low payments and start to accumulate equity.  (Obviously, buyers must be careful about purchasing homes in low rate environments if they may need to sell the home soon when rates are higher, etc.)  But, this redistribution can't really happen if mortgage rates are low because there are obstacles to lending that correlate with socioeconomic status.

Tuesday, December 4, 2018

Discounted Pre-Orders for "Shut Out"

"Shut Out: How a Housing Shortage Caused the Great Recession and Crippled Our Economy" is now available for pre-order.  It will be ready to ship in January.

Great news: Enter this code on the Rowman & Littlefield site for a 30% discount: 4S18MERC30

If you know anyone who might be interested in the book, this is a good chance to get it at a better price: $28 instead of $40.

Monday, December 3, 2018

Yield Curve Update

I have written previously about the yield curve.  It appears to me that as interest rates get lower, there is an option value embedded in long term rates because of the zero lower bound.  That means that it is harder for the curve to invert at lower rates.

I suspect this comes from my "Upside down CAPM" way of thinking.  There is a relatively stable expected return on at-risk assets like corporate equity, and fixed income is a way to trade off some of those expected returns in exchange for cash flow certainty.  So, a real 10 year yield of 1% is really a payment of about 6% subtracted from the expected real yield on corporate equities of 7%.  Low real rates are a sign of risk aversion.  They are not stimulative.  It seems that others view them as stimulative.  They are wrong.  And, this gives them a false signal about the yield curve.  It makes it look like an inverted yield curve is less dangerous at lower interest rates, because the low rates are seen as stimulative.  But, an inverted curve at low rates is actually more dangerous, not less dangerous.

Here is a graph of the yield curve slope, my adjusted slope, and forward changes in the unemployment rate.

We have been treading right along the edge of "adjusted" inversion since 2016.  It seems to me that at this point in the recovery, the long term interest rate is a simple and important signal.  If the Fed can keep the yield curve spread between 0% and 1% (or, if my claim that an adjustment is necessary is accurate, then the spread now should be between about 0.75% and 1.75%), then that seems like a great first step in thinking about monetary policy through an interest rate lens.

My main concern is that if my adjustment is accurate, a positive yield curve of 0.5% or so is actually equivalent to an inversion, and even people on the lookout for an inversion won't notice it until it is too late.  The expected December rate hike puts us into inversion territory, in that case.  I have been early to this worry, and was surprised by rising long term interest rates, so you may want to take this with a grain of salt.  But, it seems like something worth watching.  If the unadjusted yield curve inverts, it seems unlikely that the Fed will accommodate nearly quickly or strongly enough.

Friday, November 30, 2018

Housing: Part 335 - Homebuyers are hedgers, not speculators

I did get a chance to look at the paper I wrote about in yesterday's post.  They do present reasons for why credit conditions were looser in 2005 than the raw SLOOS survey numbers would suggest, and they have other measures of credit markets that suggest a more symmetrical measure of credit conditions before and after the bubble and bust.  They do not show any regressions that I see that only include the boom time, which is the source of my dis-satisfaction.  But, there are probably some correlations in the paper that would still be statistically significant in the pre-2006 data.

So, I stand by my initial reaction, though I suspect the authors would have some responses that would require more detailed critiques than I offered in the post.

In any case, upon looking at the paper, I realized that there was another chart that offers some food for thought.  This is from the University of Michigan's Survey of Consumers.
One of my reactions to papers like this is that there is an extreme case of publication bias on these issues.  At some point, if there are 1,000 papers published on the question of whether credit was an important causal factor in changing home prices and 5 papers on whether supply constraints were, then the consensus is destined to settle on a conclusion that credit was the important causal element.  It's sort of a meta-level exercise in p-hacking.

Another area where the rhetorical presumptions lead to the conclusions is the choice of questions to ask in consumer surveys.  You can choose to survey home buyers or home sellers.  And you can choose to ask them whether they think rents are going up or whether they think prices are going up.  Without changing the actual beliefs of the respondents, the choice of questions and the set of responders can create a deterministic conclusion.

For instance, home buyers may bid prices up because they are seeking a rent hedge, but if surveyors only ask them if they think prices are going up, not if rents are going up, then those buyers will appear to be speculators rather than hedgers.

And, that is what is interesting about this U of M data.  It includes a question about expected home prices, and expected rising prices are never an important factor for potential buyers (the red line).  Furthermore, there is no relationship between whether home prices are seen as low (green line) and whether prices are expected to rise.  If anything, when potential buyers think it is a buyer's market because prices are low, they tend to expect prices to remain low.

In other words, potential buyers are clearly hedgers, not speculators.  They don't see low prices as an opportunity to capture capital gains.  They see low prices as an alternative to renting.  So many analyses of the housing market ignore rental value and treat the market as a purely cyclical and speculative activity.  Highly respected analysts and economists sometimes talk about housing as if the value of the investment is entirely a product of capital gains rather than rental income value.  In reality, in most locations, in real terms, rental income value is the overwhelming source of value for homeowners.  Actual households seem to understand that, even if only subconsciously.

Thursday, November 29, 2018

Housing: Part 334 - Credit supply and the housing bubble.

Tyler Cowen links to a new paper today, with this note: "Credit conditions really did matter for the housing bubble." (HT: Tyler)

I haven't looked at the paper yet, but I have looked at a set of slides, here.

My basic point of view here is:

1) Of course credit conditions matter.  This is standard finance.  Credit provides liquidity, and less liquid securities sell at a discount.  But, this is an asymmetric relationship in standard finance.  Liquidity doesn't lead to over-priced assets.  It just leads to asset prices that reflect the market rate of return with a lower liquidity discount.  One reason that homes are a good investment for many households is that liquidity is very constrained.  Transactions costs are high and they must be purchased as a whole, not piecemeal.  Returns on homeownership are highly correlated with the length of tenure, where these costs can be amortized over longer periods.  Developments that reduce the costs associated with those problems should increase home prices.

2) The outcome of the housing bubble and bust matches standard financial expectations.  Prices during the boom were as sensitive to long term real interest rates as we should expect them to be, highly sensitive to local rent inflation trends that were the result of a supply shortage, and sensitive to credit supply where the supply shortage had pushed prices high enough to create obstacles to conventional funding.  Credit supply is an ingredient here, but it is secondary to supply constraints.

3) The problem with analysis of the housing bubble and the financial crisis is that the notion that there was an unsustainable bubble that was destined to collapse was canonized before it was established empirically.  So, evidence that explains the bust is taken as evidence that explains the boom, and vice versa.  But, if the bust was not inevitable, then correlations during the bust don't tell us anything about the boom.  This goes back to points 1 and 2.  The bust is certainly explained largely by a negative credit shock, but this is an asymmetric relationship.  From that, it doesn't necessarily follow that a boom had been created by a positive credit shock.

If I get a chance to see the full paper, I will be happy to retract my comments here.  But, these slides associated with the paper do not appear to avoid these issues.

Here is a graph of credit standards from the slides.

Not only is the relationship between liquidity and yields or prices asymmetrical, but in this particular case, the scale of the negative shock was far greater than the scale of any other shift in lending standards.  The relationship between credit standards and home prices from 2006-2010 will dominate any statistical analysis here.

So, given my priors, what I would like to see from an analysis like this is the relationship for the period up to 2005 or 2006 and the relationship for the period after 2006 or 2007.

Here is a table of results from the slides.  They run regressions from 1991-2017, 2005-2013, and 2007-2017.  Elsewhere, they use 2000-2010.  This is unsatisfying.  There is a clear trend break to a negative shock that starts in 2006.  There is no analysis of the relationship during the boom that doesn't include that period.  For someone who looks at this with the standard presumption that the boom and bust are necessarily related, this might seem like more evidence that a bubble was largely due to loose credit.  I would like to see the regression from 1991-2005.

Here is a chart comparing the one year change in real home prices to the trend in credit standards.  The asymmetrical relationship is clear here.  I have not precisely replicated the regressions shown in the slide.  I have simply done regressions of the two measures shown in my chart.  For the periods analyzed in the slide, I find similar, strong correlations as the authors do over the periods they use.  For the period from 1991-2005, I find no correlation.

When I see the paper, I will update regarding whether this is addressed there.  In the meantime, this seems like another paper that found that collapsing credit markets were highly correlated with the housing bust and concluded that loose credit caused the boom...which is a shame, because the conclusion that does clearly follow from this data - that a negative credit shock led to a housing bust and a financial crisis - is the conclusion that should be motivating current public policy and retrospectives about the crisis.

Tuesday, November 27, 2018

Housing: Part 333 - David Beckworth interviews Robert Kaplan

David Beckworth recently interviewed Robert Kaplan from the Dallas Federal Reserve Bank (transcript).  They discussed many interesting things regarding monetary policy.  There were a couple of items that I thought might be interesting to get into here.

Here is one spot:

Robert Kaplan: ...The nominal GDP targeting has a lot of appeal in that it takes into account inflation. It takes into account growth. The other thing is we are a very highly leveraged country. It's nominal GDP that services our debt.
David Beckworth: That's right.
Robert Kaplan: In other words, you need to generate nominal GDP to service the debt. There are some challenges though with this approach and others, which I actually would like to see us debate.
What's an example? How to explain nominal GDP targeting, in that there's a catch‑up mechanism in nominal GDP targeting and a lot of other aspects that I think are not going to be easy to communicate. The good news about the current framework is it's relatively straightforward to communicate.

This seems true, on the surface, but I think the more important point is that, in a way, NGDP targeting really wouldn't require communication.  How can I say that?  Well, what I'm thinking of is the countless conversations today about whether the Phillips Curve is useful, whether inflation trends will reverse or accelerate, whether expanding credit is feeding "overheating", etc.  Think of the millions of hours of debate and analysis that go into developing or forecasting Federal Reserve policy choices and their consequences.  The problem with the current dual mandate is that there is too much communication, and all the communication we could muster will never lead to consensus or certainty about near term economic activity.

With a functional nominal GDP targeting regime, there would be little to communicate.  And, what a relief that would be!

The following excerpt is more to the point of the focus of this blog - credit markets and the financial crisis.  As David points out, even this conversation would be less salient in an NGDP targeting world.  Management and regulation of credit markets wouldn't be so important if it wasn't an important ingredient in sudden negative NGDP shocks.  Kaplan's response to that notion is a window into the problem of seeing the housing bubble as a result of excess credit rather than a shortage of housing supply.

Robert Kaplan: If you look at the household sector in this country, the household sector was extremely leveraged. Meaning if you took household debt divided by gross domestic product for the households, there was a very high degree of leverage.
The reason we didn't notice it is if you looked at household debt relative to asset values, it actually didn't look excessive, back to home prices. What the housing crisis exposed is a lot of households were dramatically over‑leveraged, but they were comforted by the fact that there were easy mortgage conditions and home prices were very high.
Obviously, I don't need to remind people when the housing sector collapsed, all of a sudden, the household sector, it was clear, were very highly leveraged. They've spent the last eight or nine years deleveraging.
I think one of the lessons also, which relates to mortgage availability and so on, was we've got to watch the health of the household sector. Even with that, the aggressiveness on mortgage offerings were probably the tip of the iceberg.
It's all the securitizations upon securitizations upon securitizations of those mortgage obligations which magnified those excesses. If we didn't have all the securitizations on top of this aggressive mortgage lending, it still would have been painful, but it wouldn't have been anywhere near as painful as what ultimately happened.
David Beckworth: This goes back to the point you made earlier about nominal GDP targeting. Again, in a different world, a counterfactual world where we did have a nominal GDP level targeted, this would have made that crash a whole lot nicer or less severe.
Robert Kaplan: Truthfully, I wasn't at the Fed. I've been at the Fed only three years. I actually probably have a slightly different take. I think there's a number of things we do at the Fed. One of them is monetary policy, but another big one is macroprudential policy.
I think if you don't have good macroprudential policy, it's very difficult to run a sensible...It makes monetary policy harder. I think we need to do both. You could debate, and I've been part of those debates, to question monetary policy leading up to the crisis, approaches for monetary policy.
I think if you don't have good macroprudential policy for, again, stress testing, monitoring of the non‑bank financials, I think it makes it very hard to avoid instability.
David Beckworth: That's a fair point. If you did have those imbalances build up, let's say, for the sake of argument, you did have that leverage, I think the point you made earlier is that a nominal income target, a nominal GDP target that would make the unwinding of that leverage much more manageable. Is that fair?
Robert Kaplan: Listen, what I've learned is if the household sector gets over‑leveraged, you've got to accept it's going to take a number of years for households to deleverage. They're not like companies, who can sell assets, raise equity, restructure, restructure their debt. Households can't do that.
I think the trick is a little bit of prevention. I think we want to get into a situation where we monitor the household sector more carefully and try to take steps to maybe moderate excessive debt growth at the household sector relative to income. 

Kaplan's comments reflect what I think is considered an uncontroversial set of stipulations:
  • Excessive credit led to home prices and household debt that were bloated.
  • When home prices collapsed, households were left with the excessive debt.
  • Deleveraging from that debt slowed down the recovery.
The solutions to these stipulated risks are:
  • Prevent household debt from rising.
  • Prevent excessive use of multi-level securitizations and financial derivatives.
First, I'll point out a bit of a contradiction here.  Multi-level securitizations and credit default swaps on those securitizations were developed in order to create securitizations that didn't require new mortgages.  High household debt and excessive complex securitizations and derivatives are substitutes, not complements.  They didn't additively lead to a more acute crisis.  In fact, the rise of complex securitizations and mortgage-based derivatives came from having more savers looking for safe assets than there were investors taking the primary risk positions on either securitizations or home equity, itself.  The reason complex securitizations were profitable for their underwriters was because investors were willing to pay a premium for securities with lower expected risk.

I have discussed this many times, so I won't go into it here again in more detail, but this is an important, if subtle, correction to the credit-fueled bubble narrative.  Synthetic CDOs, CDO-squareds, etc. were the first stage of the bust, and they came about because the core cause of the bubble was a lack of housing supply, but the bubble was addressed as if it was due to a lack of fear.  Investors in the CDO AAA-securities were risk-averse.

Regarding the other points, what if high home prices are generally due to an urban supply shortage, and rising mortgage levels are a side-effect of that problem?  Then, what will happen as a result of the proposed solutions?
  • Home prices will remain somewhat elevated because of high rents.
  • Since credit is a side effect of high prices, there will be natural pressures pushing up demand for household debt.
  • To reduce that demand for household debt, taxes or non-price constraints will need to be implemented to reduce the quantity of household debt.
  • In order to keep household debt at a normal level as a percentage of income, debt will have to be held low as a percentage of home values and/or homeownership will have to be lowered.
  • Regulatory obstacles to home ownership will raise the yield on home equity - to some extent through lower prices and to some extent through higher rents.

So, the policy that seems like the prudent policy for the Federal Reserve to follow is a policy that will create high yields for a set of households who meet regulatory approval and that will create high costs for households who do not meet regulatory approval.  Over the past several years, this has been the case.  Using BEA data on housing value added and Fed data on mortgage and real estate values, the past few years have been unique in providing real returns on home equity that are higher than nominal yields on mortgages outstanding.

And, it is highly likely that regulatory approval will fall sharply along socio-economic status lines.

I am not arguing here that high debt levels are not systemically destabilizing.  I am not arguing that we shouldn't be concerned about them.  I am simply pointing out that the only realistic way to enforce this macroprudential policy is to enforce higher-than-market returns for select Americans while limiting access to those returns.  To be honest about that means being clear-eyed about the cause of high levels of household debt.

Or, to put this another way, there are many sources of value in an economy.  A marketable college degree creates value, in the form of human capital, but it is difficult to have liquid markets in human capital.  So, there isn't a where you can see the current market value of college graduates and their current market wage.

Yet, in a way, housing sort of serves as a substitute for the market in human capital.  If a banker feels confident enough in your earning ability, she will allow you to take out a mortgage to commit to transferring some of the high wages you can earn to future payments.  The potential to foreclose on the house serves as a financial tool that facilitates this trade in human capital.  The banker serves as an intermediary, using the liquidity of the mortgage market and the stability of the housing market to facilitate trading activity in the human capital market.

That is what was happening before the crisis.  In most places, the mortgage and housing markets have developed to the point that more than 80% of households can complete that trade at some point in their lives.  This is a testament to the development of human capital (broad access to above-subsistence wages) and of real estate and mortgage markets.  But, our economy was hamstrung by a political limit to urbanization, which created a dichotomy: places that were exclusive and places that weren't.

That exclusivity is rationed through housing, and by happenstance there is a liquid market that measures the value of that exclusion.  There is a for houses.  Before the crisis, this trade in human capital and housing was still functioning, but in the Closed Access cities, this meant that only those with a large excess of human capital could engage in that trade.  They had to transfer a large stake in their future earnings over to the existing real estate owners to claim their place in exclusive labor markets.  In order to fully accrue the full potential of their human capital, they had to pay the toll to access the markets where wages were highest.

Home prices reflected the value of that exclusion, and homes traded at a value at reflected their claim on that earning power.  Certainly, the existence of these credit markets facilitated the market that revealed those values.

By focusing on credit as the cause of high prices, these transactions between human capital and the housing stock have been hobbled.  The undiscounted total value of future rents on properties has not been reduced.  "Macroprudential" management on mortgage markets has just added a significant premium to the discount rate that is applied to those future rental incomes.  This has lowered home prices in Closed Access markets from where they would have been, and it certainly has reduced household debt from where it would be in this Closed Access context.  But, because this is a misdiagnosis of the problem, where its effect has been the worst has been to block access to low tier housing markets in cities across the country that were never out of whack.  (I touched on this in the previous post.)

Macroprudential management has effectively been a step backwards to a less sophisticated economy, where access to ownership of real property requires a pre-existing stockpile of wealth, and those who have wealth earn higher returns on it.

Wednesday, November 21, 2018

Housing: Part 332 - The problem in a picture

I came upon some old data recently that I thought was worth sharing.  Sorry, this isn't updated past 2014 data, but the story hasn't changed that much since then.  Maybe real estate values have recovered another 10% or so, compared to personal income.

Here is the problem.  There are two housing markets in the US.  A closed one and an open one.  The closed market gives you access to the best economic opportunities in NYC, LA, Boston, and San Francisco (Closed Access cities).  It's limited to about 50 million people.  You want in, you gotta pay.

Sources: BEA and Zillow
Here is a graph of total real estate value as a percentage of total personal income.  In 1998, in the Closed Access cities and in the rest of the country, the ratio was about 2:1.

Then, as we entered the post-industrial economic era, competition for access to the Closed Access cities pushed real estate there up to close to 350% of income.  In the rest of the country (which includes the Contagion cities, Seattle, Washington, etc.), it didn't break 250%.  Even that increase can be effectively explained with low long term real interest rates.

By 2014, in the Closed Access cities, it was 248%, while the rest of the country was down to 166%.  Now, this is with very low long term real interest rates, so, if anything, it should be above 250% even outside the Closed Access cities.

Keep this in mind when you read countless articles complaining about affordability.  Homes are more affordable than ever, really, for owners.  It's just that some homes have a premium attached to them that is unrelated to the value of shelter.  Imagine how backwards economic and monetary policy is right now, that it is not unusual to hear people call for or accept contractionary monetary or fiscal policy because home prices are getting too high again, and macroprudential policy is called for.

Also, consider the countless articles and conversations that complain about how we bailed out the banks but left regular households hung out to dry.  You know what really killed those households?  Maybe it was the fact that they lost wealth, on average, that amounted to nearly a year's income.  When real estate value outside the Closed Access cities collapsed from 236% or incomes in 2006 to 157% in 2012, how many of these moral crusaders were demanding more monetary support because home values had clearly fallen too low?

Don't get me wrong.  It isn't the job of the Fed or the government to prop up home prices.  But, it is their job to allow markets to function.  The "bailouts" were only a very poor substitute for reasonable federal macroprudential and monetary policy.  But, any reasonable policy would not have led to such drops in real estate values, especially after 2007.  How many bailout critics would have supported those policies?  That would have created moral hazard.  Right?  Because everyone knew that homes were too expensive.

One more thing about that graph.  It shows less recovery than some other measures of price/income do.  I think the main reason is that normally, price/income is based on the price of the median home compared to the median income.  Since we have been in a decade-long housing depression, the aggregate value of real estate has risen less than the value of individual properties.  This is an important part of what is happening, but it is difficult to understand it with "bubble" thinking.  Bubble thinking presumes that more building is triggered by money and credit, so that more building equals rising values.  That has it backwards.  The red line rose much higher than the blue line precisely because that relationship is very strongly in the opposite direction.

Closed Access real estate rose in value so much because there are not as many new Closed Access homes.  And, even on a national level over time, aggregate real estate value has little to do with the rate of building.  The reason that real estate value has declined along with lower rates of building since 2007 is that credit and monetary policy pushed home values well below the value that could trigger new building in many markets.  The decline in value led to lower rates of building, not the other way around.

The way to reduce things like median price/income levels so that homes become affordable again is to build many, many new homes.  That will have very little effect on the aggregate value of real estate.  In fact, if we do it well enough, it will reduce the aggregate value of real estate.  But, it will be hard to trigger new supply until credit is loosened enough and prices rise enough that more new construction can be justified.

There seem to be many macroeconomic issues that have this strange, contradictory type of causation.  In this case, rising prices cause more building, but more building causes declining prices.  Clearly, more building could cause prices to decline so sharply that more building would cause total value to decline, even after adding new real supply to the housing stock.

Housing prices need to rise so housing prices can fall.

Monday, November 19, 2018

Housing: Part 331 - More on Mortgages and Homeownership

Here are a couple more graphs on mortgages and homeownership.  The first one is from the Survey of Consumer Finances, which is conducted every three years.

From 2004-2007, homeownership declined somewhat, but mortgaged homeownership increased.

The second graph has the number of mortgage accounts from the New York Fed and the number of owner-occupied homes from the Census.  Owner-occupiers topped out around the end of 2005 when housing starts peaked.  But, oddly the number of mortgage accounts shot up in 2006 and 2007.

It appears that there were three factors at work in 2006 and 2007:

1) An increase in the number of unmortgaged owners selling their homes and transitioning to renting while a declining but somewhat stable flow of first time buyers that were naturally leveraged continued.

2) Unmortgaged owners - mostly older households - taking on new mortgages.  (Most homeowners under 55 already have a mortgage.)

3) Increasing investor activity.

It is interesting that the sharp increase in mortgages in 2006-2007 is not associated with rising ownership or rising prices, and it is associated with sharply falling housing starts.  For all of those reasons, I think it has been incorrect for so many people to treat the late rise in mortgages, which performed terribly, as if they were responsible for the housing bubble.  They had nothing to do with it, and if anything, they were propping up a housing market that would have otherwise been in unnecessarily deep decline.

In my feistier moments, I wonder if this was actually a sign of a need for liquidity.  Interest rates were at their cyclical peak.  These weren't mortgages taken out at low rates.  Currency growth was very low at the time.  This was expensive debt taken out when cash was relatively scarce.

Would it be too crazy of me to say that there was already a liquidity crunch, and that the only reason nominal GDP growth was still limping along at rates that were only marginally recessionary was because households sitting on recent real estate gains tapped those properties for cash?  Were those households, on net, speculating, or were they getting cash wherever they could get it, and currency from the Federal Reserve wasn't where they could get it?

Maybe I'm wading into waters that are over my head here.  Please tell me if I am.  But, it seems to me that causation could go either way here.  Mortgage debt could rise in a search or liquidity, or the Federal Reserve could contract the growth in the money supply as a way to counter excess liquidity coming from a speculative bubble.  Wouldn't one clue about the direction of that causality be the direction of housing starts.  If the causal trigger was a flood of debt into hot housing markets, then housing starts would be rising.  If the causal trigger was a lack of liquidity, housing starts would be collapsing.  It seems like the evidence is pretty clearly stacked against the idea that the Fed needed to be counteracting the rising mortgage levels.  Housing starts and currency growth were both contracting.

In the narrative that treats everything as excess, each step along the way is just one more facet of the bubble.  So, the mortgages originated in 2006 and 2007 were just the last gasp of that process - a continuation of the excesses that preceded them.  It seems perfectly reasonable to say, "They did this to us.  They caused this to happen.  They created the bubble, and the bust was inevitable."

But, what if the "bubble" was primarily the result of a supply shortage?  There were still trillions of dollars of home equity to be harvested, even if those trillions weren't unsustainable paper profits created by a credit bubble.  So, that wealth was available to tap for liquidity as nominal economic activity contracted.

Instead of saying, "They did this to us." we should say, "They delayed the tragedy we imposed on ourselves, but we would not relent, so the tragedy happened eventually anyway."  Those borrowers in 2006-2007 might have saved us.

Rentiers - Closed Access real estate owners who were capturing monopoly profits - were claiming an outsized portion of new production.  In order to use their property values to claim that production they either had to sell them (to a new owner that was likely more leveraged) or they had to take out debt that was collateralized by them.  This accounted for more than 100% of new personal consumption expenditures during the boom.  Eventually, the collapse of sentiment and, eventually, property values, in real estate, caused that debt-funded consumption to collapse, and there was little monetary accommodation until the end of 2008.  Nominal consumption collapsed until that happened.

The borrowers and investors were maintaining the growth of the nominal economy until they couldn't anymore.  This is a good example of how fundamental our priors and presumptions are about what happened.  Priors that say debt is unsustainable, lender and speculator driven, and reckless, would lead to a conclusion that collapse would bring discipline.  That bad things are good.  But, changing those priors, recognizing that debt-funded consumption was the product of a deep inequity in our economy, that it was sustainable and natural as long as that inequity remains, leads to a conclusion that what Americans needed was relief.  Discipline was abuse.

The homeowners with growing equity are the monopolists that need to be tamed.  But, in 2006-2007, the borrowers were the only thing keeping us afloat.

Friday, November 16, 2018

Housing: Part 330 - Mortgage Originations by Age

The New York Fed now reports mortgage originations by age in its quarterly Household Debt and Credit report.  After the 3rd quarter of 2007, mortgage originations for households under 50 years old dropped by nearly half, and the number has remained stable at that level since then.  Households above 50 years of age have continued originating mortgages at about the same rate as before.

This chart really helps highlight the generational aspect of the housing bust.

Here, I have created a chart that shows the scale of the decline for each age group.  There was a bump in originations in 2003 due to tactical refinancing, so for this ratio, I used 1Q 2004 - 3Q 2007 and 4Q 2007 - present.

Keep in mind, homeownership rates were not increasing from 1Q 2004 to 3Q 2007.  The denominator for this ratio is a rate of homebuying in a static ownership context.

This is an example of how relentlessly overzealous the rhetoric about housing is.

One example of how rhetoric is overzealous is how homebuyers' expectations or hopes are always described in terms of their price expectations.  This creates an image of speculation.  But, rent is highly correlated with price, both in terms of the level and in terms of the rate of change.  So, hopeful homebuyers could just as accurately be described as having high expectations about rising rents.  This paints an image of hedging.  In fact, in my experience, homeowners and homebuyers more often fit the image of hedgers than of speculators.  Even when they express their expectations in terms of price, they are generally thinking about costs.

Analysts and researchers, whether from the credit supply school, the passive credit school, or any other school of thought regarding the housing boom, invariably impose the speculative image on homebuyers.  When academics ask homebuyers what their price expectations are, their choice of question is the key ingredient in the conclusion they will reach.  The answers they receive would correlate highly with the answers to the question of what rent expectations homebuyers have.  Do academics ever ask homebuyers about their rent expectations?

The age data highlights another area where housing rhetoric is overzealous.  Homeownership rates were pretty close to 64% for years, then they increased from 1994 to 2004 to about 69%, then they fell back, and are now lower than 64%.  Firstly, that flat trend at 64% before 1994 was misleading.  Homeownership was being balanced by two opposing forces.  By age, homeownership was declining, but at the same time, baby boomers were aging into age groups that generally own their homes.  The flat ownership rate was a mirage.  After 1994, the shift change was mostly among younger families, and it was simply a return to the homeownership rates of the late 1970s and early 1980s.  Among working age households, no age group was rising into unprecedented territory for homeownership.

But, many people believe that, essentially, 64% of Americans were qualified to own homes, then we recklessly made mortgages to 5% of Americans who had no business owning homes, and finally, after 2007, we came to our senses when those Americans started defaulting on their mortgages, and homeownership dropped back down to the reasonable level.

That description bears little resemblance to anything that happened.

But, passing around homeownership charts with a y-axis from 60% to 70%, that isn't disaggregated by age, creates another image of overstated excess.  It is more accurate to think of homeownership in terms of age.  The more reasonable way to think of it is that 80% of Americans will eventually own their homes.  In 1993, those households might have been making their first purchase at, say, 33.  And, by 2004, they were making that first purchase at 32.  The shift really is that small.  For instance, for 35 to 44 year olds, in 1994, the homeownership rate was 65% and by 2004 it was 69%.  In any post-WW II housing market, all of those marginal new homeowners were going to be homeowners by the time they were in their 50s in any case, and the shift required in the age of first time buyers to change the ownership rate by 4% is nothing.

The shift in imagery is enormous here.  Rather than beating the bushes for unsuspecting new suckers to take on unsustainable mortgages, mortgage lenders were lending to (1) more households in their 50s and 60s who had ownership rates that never shifted significantly over decades and to (2) households who were generally going to become homeowners anyway, but they were making those mortgages now somewhat earlier - in a shift, in the aggregate, that literally could be measured in months.  It amounts to practically nothing.

What has happened since then doesn't amount to nothing.  Ownership among all working-age age groups is well below previous HUD era lows.

Wednesday, November 14, 2018

October 2018 CPI

Shelter inflation had another moderate month, bringing the trailing 12 month shelter inflation rate down a little.

Core CPI still running just over 2% and core CPI excluding shelter still running at about 1.4%.  Still in tight but not problematic territory.

Saturday, November 10, 2018

Housing: Part 329 - Construction Hiring

I happened upon an interesting post by Jason Smith at "Information Transfer Economics".  He suggests that a crackdown on immigration in 2006 was a causal factor in the housing bust and recession.  This echoes speculation from Scott Sumner.

I think there is something to what they both are saying.  There definitely has been a downshift in immigration, and as Smith points out, there was some anti-immigrant legislation passed in 2006.  But, I don't think there is much to it.  Smith tries to argue that this is a causal factor in the housing bust.  But, he's stretching quite a bit to come to that conclusion.  Here's the case he tries to build:
Again, this is speculative. However it is not implausible that the anti-immigrant sentiment of the mid-2000s ended the "housing bubble". Employers continued to look for workers in construction, but suddenly were unable to hire as many starting in 2006 due to declining immigration
That it "is not implausible" puts it on par with many other proposed causes of the crisis, and as with most of the others, that is probably the most you can say for it.  Smith points to construction-specific JOLTS data.

When Mexican immigration slows in 2006, he notes that hires decline while openings remain strong.  But, those shifts are pretty minor.  There is some early decline in construction employment, relative to other employment but, as I have shown, the shifts in construction employment were mostly very late.  Here is JOLTS data for total employment:

So, there is a small dip in construction hiring in 2006, and there is an earlier decline in construction employment growth in 2007 compared to total employment.  But, here we can see the late factors too.  First, the spike in layoffs in construction come after the September 2008 crisis.  And, hiring and quits decline in late 2008 and don't really recover.  Those measures for total employment are back to highs, but in construction, they remain basically where they were in 2008.

Even though Mexican immigration declined along with the bust, domestic migration also declined in the Contagion cities.  In Phoenix, for instance, migration from both the Closed Access cities and the rest of the country peaked in 2005 and continued to fall sharply through the crisis.  Domestic outmigration from Phoenix was increasing at the same time.  It seems more plausible that all of those trends in migration are due to declining sentiment, the end of the "inferior good" boom of Closed Access households moving away to lower costs, declining employment growth and production in general, the inability of Phoenix to meet housing demand during the migration event, and the eventual collapse in sentiment and demand that reduced the Closed Access tactical outmigration.  As with domestic migration, international migration seems like more an outcome than a cause here, though it may have played some small role.

Furthermore, housing starts were declining, and by 2006, homebuilders were facing many cancellations, leaving empty homes to sell.  It seems unlikely that by mid-2006 labor constraints in construction were the driving force in the collapsing markets.

One thing the decline in immigration might be a factor in explaining is how 12 month employment growth in Phoenix could have gone from 6% in 2005 down to 1.4% by August 2007, yet the unemployment rate dropped from 3.9% to 3.1% during that time.

Oddly, Smith is basically making a supply side argument here - that a lack of construction labor triggered the end of the housing boom, and even agrees that the debt crisis was more of an outcome than a cause of the contraction.  But, he dismisses my supply side explanation out of hand.  Scott's immigration story is more of a demand side story, that there are millions of households who would have needed homes today if immigration had continued at previous levels.  And, again, while that is a reasonable inference, it depends on the notion that the decline in construction activity was due to an oversupply of homes.  But, the decline in activity has been due to mortgage suppression.  There are many households who are "overconsuming" housing today because they live in homes they would not qualify to buy, and they would likely downsize if they had to be tenants in today's housing regime.  Certainly, if a few million additional immigrants had come here, there would be added pressure on rents, and it would have had some effect on construction markets.  The question is, given mortgage markets as they exist today, how much would that added demand just put more pressure on rent inflation and how much would it trigger new supply.  I suspect it would have done more of the former than the latter.  Especially in low tier markets, prices are still below replacement cost, so many markets, especially in entry level housing, need quite a bit more rent inflation before the price ceiling for new supply (discounted value of future rents) moves back above the price floor (construction costs). Mortgage suppression has created this outcome by raising the discount rate, lowering the value of future rents.

Wednesday, November 7, 2018

Housing: Part 328 - Bank Capital and the Crisis

John Cochrane has an interesting post up today about the role of bank capital in the financial crisis.  He is referencing some recent work from Laurence Kotlikoff.
Larry puts it all together nicely by starting with the 2011 Financial Crisis Inquiry Commission report:
"There was an explosion in risky subprime lending and securitization, an unsustainable rise in housing prices, widespread reports of egregious and predatory lending practices, dramatic increases in household mortgage debt, and exponential growth in financial firms’ trading activities, unregulated derivatives, and short-term “repo” lending markets, among many other red flags. Yet there was pervasive permissiveness; little meaningful action was taken to quell the threats in a timely manner. "
Larry then takes apart each of these non-culprits, as below.

In my view, the understanding that the crisis was a run, that without a run there would have been no crisis, somewhat like the 2000 tech stock bust, and that lots and lots more capital is the only real answer, has emerged slowly over the last 10 years. Larry's essay is good for putting all the others to rest.

The point of this is to suggest stronger capital requirements for banks, and I basically agree with all of that.  In an age where there are money market funds, securitized mortgage securities, fintech, etc., is there a reason to subsidize and support a banking system built around a mismatch between assets and liabilities?  I don't think so.  There are a number of potential ways to change that system, and people with much more expertise than me will debate what the best way is.

The two-cents I will add here is simply that, in order to get to the conclusion that a systemically unstable banking system was the cause of the crisis, Kotlikoff and Cochrane dismiss many of the same supposed causes that I have also dismissed.  They have already basically come to the same conclusions I have about the causes of the crisis, but their focus is on bank capital rather than on what caused the stresses on bank capital.

Eighty percent of my job is done here, I think.  I would only ask them to take one step back and to consider that if so many of the supposed causes of the financial crisis are not particularly compelling, then maybe the stresses on the banking system were not inevitable.

Sure, given the stresses that the banks ended up taking on, a better banking system would have responded better.  But, those stresses should have never happened.  Both can be true.  It can be true that those stresses revealed weaknesses in the banking system, and it can be true that reasonable attempts at broad stabilization in 2007 and early 2008 would have prevented those stresses from ever developing.

I fear that for those who are advocating for a more stable banking regime, the idea that a fragile regime was a root cause of the crisis is a powerful point to promote, and that it would feel like making a rhetorical compromise to agree that the crisis could and should have been averted, even with the banking regime we had.  Yet, they already have come to conclude that the evidence underlying the presumption of inevitability is weak.  It will be interesting to see how they respond to a new narrative.

Monday, November 5, 2018

Dodge City and Demographic Shifts

Recently, Dodge City, Kansas was in the news because they are moving the one remaining polling place to a location outside of the city.  This is seen as an attempt to suppress the Hispanic vote.

I took a few minutes to look up the demographics of Dodge City, and I was surprised by what I saw.  Dodge City is about 65% non-white.  Manhattan and Lawrence, the locations of Kansas State and Kansas University, are 21% and 23% non-white.

Here is a breakdown of Dodge City, by age:

Over a 20 year+ span (from 75 year olds to 54 year olds), the proportion of the age group that is white declined from 94% to 38%.  They were largely replaced by Hispanics.  That is a massive generational shift.  Post-war Dodge City residents saw their children pick up and leave, en masse.  They were replaced by Hispanic families, so the stress of this gets played out to a large extent through racial politics.  But, if they hadn't been replaced, and the city had just died, the sea change would have been no less jarring.

As unfortunate as some of the political reactions to these changes are, this sort of thing doesn't happen without some social upheaval.  The transition to an information-based economy is no less epochal than the shift from agriculture to manufacturing was.

The Luddites saw that machines were replacing skilled workers.  The engine of progress was destroying their way of life, and they lashed out at it.  Maybe there is a parallel here to the current political realignment that is happening with regard to intellectuals and the Republican base.  The information age is driven by human capital rather than physical capital.  Today's Luddites are also raging against the engine of progress that is destroying their way of life.  Today, that is education and specialized skills.  Education might have been a boon for their children, but when those kids left Dodge City and didn't return, what did the rise of human capital do for Dodge City?


Thursday, October 25, 2018

Housing: Part 327 - More on Adam Ozimek's regressions and the wrong presumptions

I looked at some more data regarding the previous post that I think might help fill out my point.

To review, Ozimek finds that cities that have the strongest recoveries in building are the places where prices are also rising the most. So, he concludes that supply constraints aren't the cause of rising prices.  And, furthermore, the places with less building are places that had bigger bubbles.  He writes:
Instead of being correlated with rapid price growth, weak permitting is correlated with how big the housing bubble was in a metro area. This is clear in the data, where the permitting recovery is higher in areas that had smaller peak-to-trough declines in house prices during the bubble bursting.

My counternarrative is that we didn't have a bubble in housing supply.  The lack of urban supply has always been the problem, and was the core cause of the housing bubble.  And, I argue, that the main factor driving housing markets since 2007 has been extreme credit tightening that was imposed because of the incorrect presumptions about what caused the bubble.

Note that both narratives could explain the fact that cities with the deepest price collapses also have seen weaker supply recovery.  That could happen because they had oversupply or because the credit bust hit them harder.  Ozimek's finding only seems conclusive because my counternarrative is currently considered inconceivable and so nobody feels required to disprove it.

Ozimek follows the convention here when he says "weak permitting is correlated with how big the housing bubble was in a metro area".  This presumption is so deeply and broadly held, he can be forgiven.  But, he hasn't shown that at all.  He has shown that weak permitting is correlated with how big the housing bust was, and he is assuming that the bust was inevitable.  This is question begging, and it is universally practiced on this topic.  In fact, it is the reason we had a crisis in the first place, because as the crisis developed, the universal reaction was, "Well, this was inevitable."  But, it wasn't inevitable.  And, we can review the data to test the assumption.  Nobody else has done that because the assumption was canonized before the crisis even happened!

Here, I am using data that doesn't exactly match Ozimek's.  I am using some data I have previously downloaded on the largest 20 MSAs, going up to 2015.  But, it gives the same basic results as Ozimek's does, so I hope you will forgive me for having saved myself the time of gathering more extensive data.

As I mentioned in the previous post, supply constraints are still central, because the cities with the most recovery in permits, by Ozimek's measure, are the cities with the weakest building markets.  They have recovered to pre-crisis levels because their pre-crisis levels were extremely low.  Here is a scatterplot comparing the long-term annual permitting rate (permits/capita) to the ratio of the permitting rate in 2014-2015 / 2002-2003.  (In the previous post there is a similar graph with MSA income on the y-axis. It looks very similar, because building has recovered in cities with high incomes and constrained housing.)

Among the top 20 MSAs, there are two groups.  The Closed Access cities are the cities where economic potential is sharply limited by housing access, so they are building at rates similar to boom levels, and slightly higher than the 2002-2003 building levels.

Among other cities, there is no pattern.  Basically, building rates everywhere else have been cut in half (with variance) whether they built 0.005 homes per capita annually before the bust or 0.015 homes.

This is because a credit shock was imposed on the country, and it doesn't affect the Closed Access cities so much because they still permit fewer homes than any other city, and money is flowing into those markets in a desperate bidding war for access to prosperity.  That was our problem in 2000, 2005, 2015, and today.

But, let's back up a step.  What about Ozimek's comment that price declines during the bust are a signal of the size of the housing bubble?  We can test that with this data.  There is some relationship between rising prices during the boom and declining prices during the bust because of the brief price spike in the Contagion cities.  But that's about it.

I have created a measure of housing permits issued in 2004-2005 compared to 2002-2003.  This is a measure of how much extra building happened during the peak bubble years.  A regression between that measure and the MSA price change from 2005 to 2012 shows an insignificant relationship.

What about longer-term building?  A regression between the price change from 2005 and 2012 also has an insignificant relationship with building rates from 1996 to 2005.

We can also compare the decline in permits and the decline in prices.  The decline in permits from 2005 to 2012 is highly correlated with the decline in prices over that period.  Both quantities and prices declined together.  But, there is no relationship between declining permits from 2005 to the 2012 trough or to 2015 and price increases from 1999 to 2005.

The only way Ozimek can connect the price collapse to the bubble is by assuming it.  But, we don't have to assume it.  We can directly measure it.  And there is no evidence in this data that either high prices or high quantities before the crisis are associated with declining permitting rates.

Looking at all of these measures, the regression model that provides the best correlation with permitting rates in 2014-2015 compared to 2004-2005 has three variables that are all statistically significant.  In order of significance (t-stat in parentheses).

  • (4.80) The decline in MSA home prices from 2005 to 2012.  For each 1% (log) price decline, there was a decline of about 1.6% (log) of permits in 2014-2015 compared to 2002-2003.

  • (3.61) The increase in building permits in 2004-2005 compared to 2002-2003.  For each 1% (log) increase in the rate of permits during the boom, there has been a 1.3% (log) increase in permitting rates in 2014-2015.  That is not a typo.  The relationship is positive and significant.  This mirrors the construction numbers I recently reviewed.  MSAs that had increased rates of building at the peak of the boom are MSAs that have continued to have strong local markets.  (There is also a positive, but not significant relationship between permit growth in 2004-2005 and price appreciation from 2012-2015.)

  • (-2.60) The long term rate of building from 1996-2005.  For each additional permit/100 residents annually from 1996 to 2005 a city issued, the ratio of permits in 2014-2015/2002-2003 is 39% (log) lower.  But, as we see in the graph above, this really consists of the Closed Access cities, where permits have recovered, and the rest of the country, where permits are at half the 2002-2003 level and there isn't much relationship between previous building rates and the current rate.
There was no housing bubble.  There was a housing supply bust in the Closed Access cities that continues to this day.  In a misguided attempt to lower housing prices in the face of that housing supply bust, we have done nothing to change the dynamic of the Closed Access cities, where rates of building continue to peak at very low levels while prices rise.  And, the rest of the country is saddled with a credit bust that has decimated home prices, rates of building, and longstanding natural patterns of migration.  The cities that were harmed were not cities that were engaged in a random bout of speculation.  They were cities whose generous building and growth traditions were nonetheless overcome by a wave of housing refugees until we killed the market with macro-level bloodletting.

But, this account was assumed out of the story before the story was even written.  Practically every piece of analysis that has been written about the crisis has been based on a reading of the evidence that depends on assumptions that were wrong.

PS: MSA Income is an important factor I should have looked at before.  Income is related to the other variables in two important ways.  First, migration pre-crisis was strongly tilted toward cities with lower incomes.  That is the perversity at the heart of my research.  So, as I explained above, permitting rates are dominated by the Closed Access cities where few permits are issued in cities with high incomes, and that explains why the long term rate of building is a significant variable.  Once I add income, that variable becomes insignificant.

Income also relates to the decline of home prices after the crisis.  Since the credit bust was imposed based on borrower characteristics, buyers with low incomes were hardest hit, and thus, cities with low incomes were hardest hit.  This isn't as strong as the pre-crisis migration relationship, but it is strong enough that adding income to the regression with permit recovery also makes the 2005-2012 price decline insignificant.

A regression of permit recovery from 2003 to 2015 against just two variables - Income and the increase in building permits from 2002-2003 to 2004-2005 has an r-squared of 0.69.  Income is highly significant (t-stat = 5.68) and the level of MSA income 1% (log) higher is correlated with a log 2% increase in permit recovery.  For each 1% (log) increase in the rate of building in 2004-2005, permit recovery has been 1.0% stronger (t-stat = 2.77).  Again, this is the opposite sign of the bubble story.  More building in 2004-2005 is correlated with more building in 2015.

PPS: Oops.  Sorry.  I had an error in the excel worksheet.  (The error doesn't effect the regression described above, but the price collapse does have a correlation with building permit recovery.)  Income is actually highly correlated with the price collapse from 2005-2012.  An MSA with income 1% higher was correlated with 0.6% less of a price collapse.  Even with that relationship, the price collapse is correlated with permit recovery.  So, a regression with three variables - Income, the increase in permits from 2002-2003 to 2004-2005, and the price collapse residuals after accounting for the effect of income - has an r-squared of 0.79, and the significance of both income and permitting rates are increased compared to the two variable regression.  The coefficients are:
Income: 2.08
2004-2005 permitting rate: 1.32
Price Collapse: 1.02

So, stronger permit recovery is related to higher income, growth in the rate of building in 2004-2005, and less of a price collapse from 2005-2012.  All of this supports the conclusion that there was not a housing supply bubble but there was a credit bust imposed on low income borrowers after 2005.