Revisiting the Housing Bubble (part one)

Our recent economic primer has us re-thinking a previous housing bubble article. There was nothing wrong with the earlier article, but the graphs we used were not as dramatic as one that Yale economist Robert J. Shiller created back in 2006. The graph below was pulled from a New York Times article (in 2006) but other sites that have reported on this graph include, Business Insider (in 2009), Glenn Beck (in 2009) and an independent Flikr post that also attempts to predict the future (it’s worth a look).

Wikipedia defines an economic bubble as “trade in high volumes at prices that are considerably at variance with intrinsic values”. And it’s clear that Dr. Shiller’s graph shows trade at considerable variance with intrinsic values. (The graph shows only prices and not volumes, but we’ll assume the volumes were high.) Even though the peak for the current boom is crazy-high, you might be tempted to down play it, after all the curve goes back to the 1890s so some inflation in housing prices is to be expected. Unfortunately Dr. Shiller considered that and this curve is adjusted for inflation. The current boom saw housing prices at an all time high by a considerable amount! (Note, I suspect Dr. Shiller used a standard value of 100 as a generic reference value. If you’re confused by what “a house sold in 1890 for $100,000 inflation adjusted to today’s dollars” means try thinking of it in percent. Eg, a house in 1890 sold for 100% of its value, in 1920 for 66% of its value and in 2006 it sold for 199% of its value.)

Where did this money come from?

Part of the money came from world governments in a manner entirely analogous to our Popsicle stick economy. When governmental banks lower interest rates for home ownership (for example) this has the effect adding money to the economy. However, others believe (podcast) that most of the money came from world-wide investors seeking safe returns on their investments when government bond markets were yielding sluggish returns (fast forward the above podcast to around the 10-minute mark and start listening). Couple this with the supposedly safe financial wizardry of collateralized debt obligations and world-wide money poured into the housing market (listen at about the 30 minute mark on the podcast).

At this point, there was a lot more money running through the housing market than normal. But this money was stuck in the housing market, unable to directly flow into say the auto market or the agriculture market because you needed to buy or refinance a house to get access to this money. The Popsicle stick economy provides some insight here if you think carefully about it. With a lot more money in the housing market you’d expect to see housing prices rise AND people willing to pay these higher prices. Dr. Shiller’s graph shows this. His chart is of home values not home prices, people sold and other people bought houses at those prices.

Side Effects

Dr. Shiller’s chart is much more dramatic than our earlier chart showing the increase in the construction labor force during this time. Dr Shiller’s chart also tackles head-on the definition of a economic bubble, it tracks the variance between price and intrinsic value. Yet it still went largely unnoticed by the wider media in 2006 (we’ll address this issue in a future article). One reason may be because it was so unbelievable. However, bubbles have consequences and their effects should be able to be seen in a variety of manners. One such manner was our earlier construction labor force chart, however another chart, almost as dramatic as Dr. Shiller’s chart tracks the housing supply.

In a boom it is easy to imagine that builders build more houses than normal, and possibly more than the market needs. In the latest boom it is also conceivable that more people qualified to borrow money to buy a home than ever before (see the above podcast). These factors should show up in the housing supply. One measure of the housing supply is the “months of housing supply”. It’s a number that reflects both the sales rate of houses and the number of houses on the market. Basically, “how long will it take to sell all the houses on the market at current sales rate?” The data for the graph above comes from the US Census Bureau. The shaded areas match up with those in Robert Shiller’s graph. Notice that housing supply was relatively unchanged during each boom time. Just after each boom ended the supply of houses increased. (Although the 1980s boom showed some swings it largely followed the same pattern of lower inventory during the boom than immediately before or after.) The increase in supply immediately following a boom may be due to several factors:

  • the building of houses in progress finished after the boom ended and could not easily find buyers.
  • overpriced houses (post-boom prices) took longer to sell.
  • people who couldn’t afford their mortgages had their houses repossessed by the bank.

It is interesting to note that visually the long term average months of housing supply appears to be 6 months. Yet during the current boom that number appeared to be 4 months. This seems to corroborate the perspective in the above podcast that people who otherwise wouldn’t have qualified to borrow money to buy a home before the current boom were able to borrow money. Such an effect is not apparent in Dr. Shiller’s graph.

It is also encouraging to see that the current inventory levels are near this “long term average”, hopefully the housing market doesn’t have that much further to fall before bottoming out.


Dr. Shiller directly addressed the housing boom question by graphing price and intrinsic values over a 120 year span. This resulted in a very dramatic graph. But bubbles have consequences and those consequences should also be quite visible. The housing supply numbers over the past 40+ years shows similarly dramatic changes as you’d expect. Our earlier graph of construction sector employment shows another sign of a truly large bubble, but it wasn’t as dramatic as the 2 charts shown here. However when analyzing bubbles it would be great to view all the related data in one place. Seeing multiple independent signs of a bubble may help convince people that a bubble is occurring before it’s too late.


Footnote Dr. Shiller’s graph shows the 80s boom time occurring between 1984 and 1989. This paper (page 26) from Harvard shows the start date as 1982. For the graph above I’ve split the difference and started it in 1983. As a side note, lest you think Harvard is infallible, that same paper used very bizarre almost laughable graphics (figure 1 page 47). Tufte would not approve.

1 comment for “Revisiting the Housing Bubble (part one)

  1. December 12, 2015 at 1:20 pm

    There has been much hand-wringing about the fact that housing sratts is well below the 1 million + level before the recession. But to focus on that and ignore the gains being made in housing is is disservice.Sure, we have what I like to call of a recovery (that is, a long, flat period following a steep recession), but it is improvement nonetheless.Do we have a long way to go in this recovery? Yes. But just as the tiny acorn turns into the mighty oak, or a newborn calf turns into a monster bull, so will the tentative first steps if this recovery turn into something generally sustainable (depending on what happens over the next six-eight quarters, there may be bumps on the road). You begin moving a mountain by carrying away small stones. Well, these are the small stones we are carrying away.Those of you who were/are/want a V-shaped recovery will be disappointed. But to then deny that no recovery is occurring simply because it is not as fast as you would like is incorrect.This will not be a barn-burner of a recovery. We will not return to the pre-recession level of construction for a very long time. It will be, at times, a tenuous climb. But things are improving.By the way, this is also good news for the overall economy. The US economy has not entered into a serious period of economic decline as long as housing construction was growing year-over-year.

Leave a Reply

Your email address will not be published. Required fields are marked *