Has the U.S. housing bubble begun to reinflate?
In the past several months, there has been a lot of speculation to that effect, but so far, no one other than David Stockman has really come out and committed to an affirmative answer. And even Stockman didn't specify when such a new bubble in the U.S. housing market might actually have begun.
But what really sparked our interest in this topic today is the unexpected strength in the number of initial unemployment insurance claims being filed during the last several weeks, which along with the strength of the construction industry cited in the latest employment situation report, suggests that the U.S. housing industry is finally showing signs of robust growth, at least as measured by rising sale prices for homes.
Unfortunately, the apparently robust growth of housing prices in the last several months is suggestive of something other than fundamental factors at work. Fortunately, we developed an early detection method that might be used to confirm if a bubble is present in the housing market and if so, to identify specifically when it began. So, we're going to revisit the data once more to see just what might be brewing under the surface of the U.S. housing sector.
In doing that, we're going to push the envelope with our methods, as we'll be tapping new sources of data for median new home sale prices and median household incomes, in which these data items are reported monthly.
Let's get to work. Our first chart reveals the trailing twelve month average of the median sale prices of new homes sold each month in the United States from January 1963 through January 2013, as reported by the U.S. Census Bureau. The first data point spans the 12 months from January 1963 through December 1963, the second data point spans the 12 months from February 1963 through January 1964, et cetera.
In preparing this chart, we calculated the trailing twelve month average for median new home sale prices to account for the well-known effect of seasonality in housing sale data.
In looking at the chart, certain things stand out with respect to the apparently steady long term trends that are otherwise evident in the chart. Going from left-to-right, the first unusual thing we see is the small upward bump that begins around December 1986 and ends about four years later, as a new steady upward trend takes hold. Continuing to the right, we get to the 800-lb gorilla that represents the inflation and deflation phases of the U.S. housing bubble in the form of the large lump that appears to begin around December 2003 and appears to end around December 2008. We then see a steady trend resume in the two years that follow, which is followed by what appears to be a new spike upward. Could that be a new bubble forming as so many people are speculating just based on house prices alone?
The truth is that you can't really tell from this chart. It may be, or it may not be. For example, what about that four year long small lump from 1987 through 1990? Isn't that a bubble, if only a small one, too? How come we haven't heard about any of that in the economic history books?
The reason for that analytical vagueness is that housing prices are not really a function of time, although they are often treated as if they are.
In reality, housing prices are a very strong function of income. Although other factors can and do affect them, their prices are primarily determined by the household income of those who live in them. What's more, housing prices are very linear functions of income - if you look at housing expenditures by income level, you'll find that it follows a very straight trajectory.
That linear characteristic also applies over time. Here, for example, we would expect to see house prices follow a steady upward trend as household incomes steadily rise over time. If we see deviations from that basic pattern, that tells us that something other than income is affecting house prices, which is what makes our analytical methods so effective.
Today, we'll be doing that with Sentier Research's monthly median household income data, for which we thank Doug Short for converting into nominal (non-inflation adjusted) form, which saves us the hassle of having to match the different inflation-adjustment scales used by the U.S. Census Bureau and Sentier Research.
The downside to using Sentier Research's data is that it only goes back to January 2000. To get around that limitation, we'll also be presenting the U.S. Census Bureau's annually-reported median household income data, which goes back to 1967, and which we'll use as the backdrop for establishing the long-term trends evident in the U.S. housing market.
As we did with the monthly median new home price data, we'll be calculating the trailing twelve month average for these figures as well, so they have had the same adjustment, providing as much as an apples-to-apples basis for drawing conclusions from what we find. Our initial result is presented below:
In this chart, we're able to determine that there have been two major long-term steady trends. The first ran from 1970 through 1986, as median new home sale prices were consistently about four times (4.07X) the value of the median household income.
This trend ended when the Tax Reform Act of 1986 made it more desirable to have a large mortgage when the tax deductibility of other kinds of consumer debt was eliminated. Enacted into law on 22 October 1986, median new home prices began increasing significantly after November 1986, rising rapidly in 1987 before settling onto a new steady, long-term trajectory with respect to median household income, in which median new home sale prices averaged about 3.6X the amount of median household income. It turns out that the dip at the end of the "small lump bubble" is really the result of the recession that accompanied the Persian Gulf War following Iraq's invasion of Kuwait in 1990, which depressed housing prices along with incomes at the time.
That new trend continued through 2000, until the onset of the U.S. Housing Bubble in December 2001.
Here, after the Dot-Com Stock Market Bubble peaked as a monthly average in August 2000, large amounts of money began flowing out of the U.S. stock market. It was slow at first, as the market declined by less than 10% through March 2001, but that quickly changed as the deflation phase of the Dot-Com Bubble became much more volatile as the U.S. economy went through a period of recession.
With stock prices swinging by 10%-20% of its peak value in any given month through October 2001, many stock market investors either took their losses or pocketed their gains from the Dot-Com Bubble and exited the market. That money didn't sit around idly, as much of it went into the U.S. housing market instead during that time, which enjoyed growth despite the recession throughout 2001 as a result. The recession ended in November 2001, just as interest rate cuts by the Federal Reserve helped pull mortgage rates to their lowest level in more than a generation. November 2001 marks the true launching point for the U.S. Housing Bubble.
Afterward, housing prices began skyrocketing month after month as the U.S. Federal Reserve compensated for both the recession and the 11 September 2001 terrorist attacks by holding interest rates at levels far lower than economic conditions would warrant for a sustained period of time. Our next chart focuses more closely on the U.S. housing bubble years:
U.S. housing prices continued their rapid ascent through September 2005, before beginning to decelerate on their upward trajectory as the U.S. housing bubble neared its peak, as the Fed's series of quarter point interest rate increases finally boosted them to levels that actual economic conditions warranted. The peak came on March 2007, after which median new home sale prices held level through October 2007. The deflation phase of the U.S. housing bubble then began in the following months, as the U.S. entered into deep recession.
The trailing twelve month average of median new home sale prices then bottomed in December 2009 before beginning to recover and rise in 2010. However, median household income continued to fall for another year, and it was not until December 2010 that a new steady, upward trend began to form in the U.S. housing market as median household incomes began to rise once again.
The new period of order in the U.S. housing market saw median new home sale prices stabilize at roughly 3.34X the value of median household income, which is fairly consistent with the other long-term periods of relative order in the U.S. housing market.
That period of order came to an end after July 2012. Beginning in August 2012, something else other than household income has begun affecting the median sale prices of new homes in the United States. Through January 2013, median new home sale prices are growing at a rate that is consistent with what we observed during the initial inflation phase of the U.S. housing bubble following the end of the U.S. recession in November 2001.
We therefore conclude that the U.S. housing bubble has effectively reignited, with a new inflation phase having taken hold since July 2012.
The question that remains to be answered is "why?" We'll take that question on in upcoming posts.
Sentier Research. Table 1. Household Income Trends: January 2000 to January 2013 (in January 2013 $$). [Excel Spreadsheet with Nominal Median Household Incomes courtesy of Doug Short]. Accessed 13 March 2013.
U.S. Census Bureau. Median and Average Sales Prices of New Homes Sold in the United States. [Excel Spreadsheet]. Accessed 13 March 2013.
U.S. Census Bureau. Income, Poverty, and Health Insurance in the United States: 2011. Current Population Survey. Annual Social and Economic Supplement (ASEC). Table H-5. Race and Hispanic Origin of Householder -- Households by Median and Mean Income. [Excel Spreadsheet]. 12 September 2012. Accessed 13 March 2013.