Housing : Part 310 - The premise determines the conclusion, a continuing series

Here is an interesting symposium at the NBER on the financial crisis (HT: MR).  Previously, I have written about how the crisis and its presumed causes were predetermined.  When the question is asked, "What caused the financial crisis?"  The answer always comes in the form of "This is what caused the housing bubble."  The inevitability of the crisis is canonized.  It doesn't even need to be asserted.  This can be seen throughout the slides that are provided at the NBER link.

A set of slides from Nicola Gennaioli and Andrei Shleifer discusses the difficulty of seeing bubbles and preventing them from blowing up.  It includes this graph, which all reasonable people are supposed to agree is part of the "the banks did this to us" story, where banks got all leveraged up with irrational exuberance and short-term greediness.

Can I suggest that this seems a bit underwhelming?  I mean, there are legitimate debates to be had about the most systemically safe ways to fund investment banks, but I think if you showed this graph to anyone that didn't have priors that there was a massive financial crisis caused by risk-taking, nobody would look at this and say, "This is clearly the picture of a financial system ready to blow up in 2007."

Morgan Stanley is the only bank shown here that had leverage in 2007 that was significantly higher than previous levels.  Maybe you could argue that leverage had been too high for the entire decade shown on the graph.  But, then this is just axiomatic.  It's a plausible condition that is lying in wait to explain any crisis.  Really, in that case, you could remove the y-axis, or change the numbers to half or to double the numbers shown here, and the argument wouldn't fundamentally change.  I mean, if Morgan Stanley had been leveraged 20 to 1 or even 10 to 1, and a financial crisis struck, it's not like economists would all look at this graph, with that different scale, and say, "Well, leverage clearly didn't cause this crisis.  Now, if they had been leveraged 30 to 1, then leverage would be important."

No. Leverage is a plausible cause of financial crises, and so any level of leverage, in hindsight, can be called out as the cause of the crisis.  The premise is overwhelmingly the source of the conclusion.  And, certainly leverage is a plausible cause of financial crises.  That's what makes it such a compelling culprit that the premise itself seems sufficient to reach a conclusion.

Here's another slide from that deck.  Here, referring to Lehman and what appear to be optimistic expectations in 2005, they say, "Analysts at Lehman Brothers understood the consequences of home price declines. However, they severely underestimated the probability and magnitude of these declines."

Again, this is hardly new ground.  This is consensus stuff.  But look at those scenarios.  There is nothing wrong with them.  There is a 50% chance of home prices rising by 5% per year, and a 5% chance of a shock to home prices worse than anything we have seen since the Great Depression.

And, who is to say that those probabilities are wrong?  Who is to say that if we could relive the 2000s a hundred more times that 95 of those times would turn out just fine?  Oh, and by the way, this scenario analysis would be pessimistic if it was applied to Canada, Australia, or the UK over the same time period.  We do have several versions of economies entering 2006 with very high home prices, and the evidence suggests that having a generation-defining housing bust is highly unusual.

This is such a deep and ironic example of how the premise that a severe contraction was necessary actually caused the crisis, and then served as its own confirmation when that crisis happened.  This error of looking back at scenario analyses and judging it based on a single outcome only seems reasonable because the premise that the crisis was inevitable is so strongly held.  (And, I don't mean to single out these authors.  This is the consensus treatment.)

This forecast was made in the summer of 2005.  From August 2005 to August 2008, the national Case-Shiller price index dropped by about 7%.  That part of their worst case scenario was actually too pessimistic.  It was their expectation of stability after that which was too optimistic.  From August 2008 to the end of 2011, prices fell another 14%.  And, it was during that later period where nine out of ten of the mortgage defaults happened.

Now, I'm not going to spend paragraphs here walking through the entire timeline again.  Surely we can all agree that by the end of 2008, public policy itself is implicated in the eventual outcomes.  Public policy can even be implicated in the declining prices before August 2008.  But, the irony here is so deep.  What was the overwhelming reason for holding back on stabilizing policies throughout that time?  It was that we had to let prices drop to avoid moral hazard.  To impose discipline.  They had done this to us because of their optimism, greed, and riskiness, and they needed to learn a lesson.

It's fitting that Lehman failed in September 2008, right when the first three years of that pessimistic scenario ended.  Their pessimistic scenario covered the outcomes that had occurred up to then.  In September 2008, the Treasury took over Fannie and Freddie and cut off lending to entry level borrowers, creating a late collapse in low tier home markets that nobody seems to have noticed (because the premise accepted, even demanded, collapse) and the Fed implemented disastrously tight monetary decisions by holding the target rate at 2% and then implementing interest on reserves that sucked hundreds of billions of dollars out of the economy.

I see slides in these programs bemoaning the role of pro-cyclical financial markets in creating a boom and bust, but I don't see much about public demands for pro-cyclical regulatory and monetary regimes.  There is no doubt that the Fed and the Treasury could have avoided the post-2008 price collapse with earlier and more accommodative actions.  The premise was that contraction was necessary.  The premise was the reason we allowed or insisted on instability.  And the premise is why that subsequent instability can be blamed on the market that we imposed the premise on.

Another example of the strength of the premise, from the same set of slides is a reference to the work of Case, Shiller, and Thompson, who surveyed homebuyers for several years, and found that their long-term expectations for home price appreciation are unrealistically high.  This has been blamed for fueling the crisis.  The Shiller real housing chart that was so popular during the boom is referenced, which I have addressed before.  That chart is based on national average numbers, which completely erases the localized nature of the housing supply problem that caused the bubble.  Treating the housing bubble as a national phenomenon helps to feed the false presumption about its cause, because it is a lot easier to blame the bubble on national excesses if it is a national phenomenon.

Along this vein, the panelists reference the survey work of Case, Shiller, and Thompson, and note that during the years from 2003 to 2008, the average long term annual gains homebuyers expected in four different counties were:
11.6% Alameda County (San Francisco)
8.1% Middlesex County (Boston)
9.5% Milwaukee County (Milwaukee)
13.2% Orange County (Los Angeles)

They note "Forecasts were roughly in line with extremely rapid home price growth witnessed prior to the surveys but were way off from future realized growth."  Treating the bubble as if it was a national phenomenon and treating the bust as if it was inevitable means that we can ascribe (false) meaning to this result.  But, here is a graph of the median home price in each metro area (from Zillow).  These cities have very different stories.  Nothing in Milwaukee was outside of historical norms.  As with most of the country, prices were somewhat buoyant in 2004 and 2005, but that is understandable given the low long term real interest rates of the time.

So, how much of the "bubble" is explained by these expectations?  If Milwaukee buyers had high expectations but home prices were about $200,000, then did the expectation of 11.6% price appreciation explain $700,000 homes in San Francisco?  It seems more likely that there is some bias in the response to this question that has little effect on prices.  Let's say there is some effect.  Maybe 15%?  Maybe without these high expectations, San Francisco home prices would have only been $600,000 at the peak instead of $700,000.  What if home prices in San Francisco had stopped at $600,000.  Would we then have looked at the housing data and said, "Oh, expectations can't explain that.  Now, if homes were selling for $700,000, then we might be looking at a bubble, because then San Francisco prices would be 15% too high, and that would be a reason to suspect these biases in expectations."?  No.

Since the premise that demand, unmoored from rational value, exists prior to the analysis, this bias in buyer expectations can explain everything from $200,000 homes in Milwaukee to $700,000 homes in San Francisco, and everything in between.  And, when the "inevitable" bust comes, those high expectations will be sitting there, ready to fill in the narrative.  The reason it seemed like there was a bubble was that home prices in Boston, LA, and San Francisco were double or triple the price of homes in Milwaukee.  But, the false premises about its cause led us to watch the median home price in Milwaukee decline by 15% over the next five years - an incredible loss by any historical standard - and consider that reasonable, even though there was never a reason for homes in Milwaukee to lose a penny of value.

Another presentation by Aikman, Bridges, Kashyap and Siegert asks "Would macroprudential regulation have prevented the last crisis?"  But macroprudential regulation caused the crisis.  In their presentation, the first step to achieving macroprudence is identifying the buildup of risks in the economy.  The first item in their list of examples of challenges to achieving this is the recognition of a housing bubble.  While many of the tasks of achieving macroprudential stability are difficult and were not done well, according to the presenters, this first step was achieved, because the Federal Reserve noted correctly in 2005 that home prices were overvalued by 20%.

But, that was the problem.  Home prices didn't need to fall by 20%. As the housing market started to collapse, the Fed signaled that if home prices did fall by 10% or 20%, it wasn't going to do anything to counteract it.  That was a "correction".  The initial drops in housing starts were enough to buffer the sharp drop in demand that naturally followed.  But, when housing starts fell as far as they could, ratings agencies started to forecast unprecedented declines in prices, and the Fed continued to see instability as a necessary medicine for enforcing discipline and avoiding moral hazard, prices collapsed.  The more they collapsed, the more that the false premise led us to demand discipline and to rail against moral hazard.

Step 4 in their action plan is to "Take action to reduce the build-up in household debt".  The macroprudential action here, surely, should be local, since the rise in these balances was local.  And, the clampdown on lending to borrowers with low incomes and low credit scores, which seems like the obvious macroprudential response, has killed low tier markets, and it has nothing to do with what happened during the boom.  All of the rise in debt payments that were over 40% of income was among households with high incomes, because those are the households bidding up home prices in the Closed Access cities.

I don't see anything in these slides that seems to acknowledge the importance of supply constraints in rising debt levels.  The entire discussion happens within the premise that credit supply is the cause of both the boom and bust.

Another presentation also discusses leverage and over-reliance on short-term borrowing in the financial sector.  Here is a chart from that presentation:

I would point out here that most of the increase in home prices had happened by the time short term repo financing began to rise above the level of long term financing.  By late 2005, the Fed had raised the short term rate to nearly 5%, and the yield curve was inverted.  Banks weren't saving on interest expense when they increased their reliance on short term financing.  This wasn't a matter of "borrowing short and lending long" and pocketing the difference, while creating an externality of systematic risk.

It is certainly useful to consider ways in which a financial system can be more resilient, but these discussions are like a group of doctors standing around a patient who is repeatedly hitting his head with a mallet, and discussing the importance of avoiding headaches by staying hydrated.  Staying hydrated is important!  This is true!  But, it isn't the problem at hand.