Imagine that your financial advisor called you up one day and said:
“Great news…your investment portfolio gained 1% in January which is an annualized return of 12%. However, we have to subtract .05% from that return because historically January’s return has only been 0.95% since 1950. This brings our seasonally adjusted return to 11.4%.”
Of course, after the SEC pays a visit to the advisor to correct his performance reporting measures, the simple reality is that “what you see is what you get.”
While this example may seem a little farfetched – this is exactly what happens with a variety of economic reports that are released by various government agencies and member organisation/lobby groups. The reasoning for such data manipulations is not a nefarious scheme; but rather an attempt to smooth what is normally very volatile data. This is particularly the case with housing related data. As an example the chart below shows the data released by the Census Bureau for housing starts on both a non-seasonally and seasonally adjusted basis.
As you can see there is an extreme amount of volatility in the non-seasonally adjusted data. The Census Bureau takes the reported monthly housing starts data and annualizes it. Therefore, if 10,000 homes were started in January it is reported as 120,000 on an annualized basis. Then a seasonal adjustment factor is added to account for seasonal weather and demand patterns. For example let’s take a look at the housing start data that was just released for December of 2012.
The headlines read that “Housing starts surged by 12.1% in December proving that the housing recovery is back.” In reality the numbers were as follows:
- December starts: 61,500 (down 2.8% from November)
- Annualized December starts: 738,000
- Reported seasonally adjusted December starts: 954,000 (Up 12.1% from November)
- Seasonal adjustment to December starts: +216,000
Historically, the data smoothing methodology was “close enough” and the variations were, more or less, worked out over time. However, in the current economic environment, the seasonal adjustment process may be overstating that actual activity that is occurring within the underlying economy. With housing currently making a very small contribution to overall economic activity, just slightly more than 2.5% as shown in the chart below, the difference between the “real” economic impact of 61,500 homes being started nationwide versus 954,000, of which 216,000 were a mathematical seasonal adjustment, can be quite dramatic.
However, as in our financial advisor analogy, when it comes to the impact of the “housing recovery” on the economy “what we see is what we get” and nothing else. Therefore, in the quest to determine what the actual contribution to the overall economy that housing will provide, we need to look at the full process of housing on an actual basis.
The Housing Process Activity Index (HAPI)
The housing process begins with the permit to build a home which leads to the start of the construction process, the completion and the sell to an end buyer. The reason for looking at all four components is that many permits that are filed do not result in a start, many starts do not lead to completions and there are many completions that remain unsold for quite some time.
The Housing Process Activity Index (HAPI) takes into account all four variables. However, instead of utilising seasonal adjustment factors and annualizing the monthly activity, as with the Census Bureau, I use a 12-month average of the actual monthly activity to smooth the data. The chart below shows the HAPI as compared to its individual HAPI subcomponents.
This tells us quite a different story than what the media is currently reporting. On average over the last twelve months there were:
- 67,000 permits for new privately owned housing
- 65,000 housing starts each month
- 54,300 completions
- 30,900 sales
What this tells is that out of the 67,000 permits, on average, that were pulled each month to build a home only 30,900 actually wound up being completed and sold. This is a very different picture than the recent months seasonally adjusted data that showed:
- 903,000 permits
- 954,000 starts
- 686,000 completions
- 377,000 sales (as of November)
It is important to note that I am NOT contesting the manner in which the Census Bureau reports its housing data. I am, however, suggesting that the method used in the current economic environment may be overstating the actual activity that is occurring. By smoothing the non-seasonally adjusted data using a monthly average we see a very different perspective to the data. The HAPI begins to potentially answer the questions of why housing remains a low economic contributor and why construction employment is still mired at very low levels.
While housing has improved somewhat from the post-crisis lows it is far weaker than the majority of headlines actually suggest. Housing inventory has declined sharply from its peaks that have primarily been driven by speculative all-cash investors turning lower priced homes, and buying homes in bulk from banks, to be turned into rentals.
Furthermore, a large portion of the “housing start” story has been in the multi-family apartment space. As shown in the chart below the percentage of apartment starts, 5-units or more, is currently at some of the highest levels on record and has surpassed the peak seen in the last recession.
Currently, there are still more than 25% of homeowners underwater which limits their ability to move, refinance or sell their homes. However, as prices rise, there are two issues that begin to attack the housing story: 1) As prices reach levels where underwater homeowners can sell they will likely do so out of a psychology need to escape the “trap,” which will bring a large supply of homes back onto the market, and; 2) rising prices will eventually erode the profitability of buying homes for rentals which will bring the speculative frenzy that has been the driver of the recent recovery to a halt.
The housing recovery is ultimately a story of the “real” unemployment situation which still shows that roughly a quarter of the home buying cohort are unemployed and living at home with their parents. The remaining members of the home buying, household formation, contingent are employed but at lower ends of the pay scale and are choosing to rent due to budgetary considerations. Also, we should not discount the psychology of home ownership has dramatically changed since the crash as many of the “millennials” saw the financial damage their parents suffered and are opting out of taking such a perceived risk.
As I stated recently the optimism over the housing recovery has gotten well ahead of the underlying fundamentals. The overarching problem is that the housing market that is almost exclusively dependent on the continued push to artificially suppress interest rates combined with massive amounts of direct stimulus, and incentives, to bailout current homeowners and banks. This intervention is causing an artificial supply suppression which is likely to create a backlash in the future as the current supply/demand conditions are unsustainable.
While the belief was that the Government, and Fed’s, interventions would ignite the housing market creating an self-perpetuating recovery in the economy – it did not turn out that way. Today, these repeated intrusions are having a diminished rate of return and the risk now is that interest rates rise shutting potential homebuyers out of the market. It is likely that in 2013 housing will begin to stabilise at historically low levels and the economic contribution will remain fairly weak. The downside risk to that view is the impact of higher taxes, stagnant wage growth, re-defaults of the 6-million modifications and workouts, elevated defaults of underwater homeowners and a slowdown of speculative investment due to reduced profit margins. While many hopes have been pinned on the 2012 stimulus fuelled, China investing, and supply deprived housing recovery as “the” driver of economic growth in 2013 – the data suggest that may be quite a bit of wishful thinking.