The status of the housing sector has been a big question mark regarding the economic outlook. Some have suggested that housing is at or near bottom. One notable Fed official, Vice Chairman Donald Kohn, recently gave his opinion in a speech on January 8, 2007.
"Tentative signs have begun to emerge that the housing market may be stabilizing. Home sales appear to have flattened out since midyear, mortgage applications have been increasing, and consumers' perceptions of homebuying conditions, as reported in the Michigan survey, have improved. Nonetheless, even if the demand for housing is leveling off, housing activity may not yet have found a floor, given the sizable overhang of unsold houses. In my own judgment, housing starts may be not very far from their trough, but the risks around this outlook still are largely to the downside. . . . Building permits decreased substantially again in November, and inventories of unsold homes have only started to edge lower."
Vice Chairman Kohn cites a number of standard housing data series for his opinion that housing is stabilizing but "may not yet have found a floor." How do the data stack up against this view and what should we expect for near-term conditions?
Pending homes sales' free fall has slowed
A key housing data series cited by many as an early indicator of housing activity is the National Association of Realtors' pending home sales index (PHS). This series is based on contract activity for existing home sales nationally. This index slipped 0.5 percent in November but may be leveling off. If the PHS index begins a modest uptrend, what would that tell us about existing home sales? The chart below compares pending home sales with existing home sales. Based on historical averages, changes in pending home sales lead existing home sales by about two months. Based on averages, if pending home sales began to rise in December, we likely would not see the start of an uptrend in existing home sales until February. These data are a little volatile, but on average that should be the pattern.
Housing inventories are improving but remain a problem
Sales and expectations of sales drive actual construction. Homebuilders generally anticipate demand to some degree - some housing construction is "spec" (or speculative) while some is built to order. When spec housing gets too far ahead of sales, unsold inventories build up and homebuilders slow or reduce the pace of construction. As seen below, the ratio of unsold new homes to sales rose sharply from a recent low of 3.5 months supply in August 2003 to a recent high of 7.2 months in July 2006. Housing starts have fallen more sharply than sales and supply has fallen to 6.3 months for November 2006. Single-family starts fell 35 percent from January 2006 (the recent high) to October 2006 (the recent low) while over the same period existing single-family home sales declined 14 percent. Based on past data, it probably will take two or three additional months of declines in the stocks-to-sales ratio before starts begin an uptrend.
Permits lead starts?
What are the housing permit data telling us about upcoming trends in housing starts? Actually, this is a "trick" question. Contrary to casual opinion, monthly housing permit data have no leading information regarding starts. As seen below housing permits and starts move together very closely other than monthly noise. Certainly, in permit-issuing places (where permits are required to build), one must obtain a permit before beginning construction. However, permits are expensive and builders rarely obtain a permit prior to immediately beginning construction. Monies spent on permits are best left in the bank collecting interest until the last minute. Or if one borrows the money, one still should wait as late as possible to reduce interest being accrued. Except for unusual circumstances, changes in permits are usually also reflected in the same month's starts data. Starts usually occur within one month of obtaining a permit.
As an aside, why are starts generally higher than permits? If not all permits are used, then should not permits be higher than starts? The reason starts are higher than permits (levels) is because not all jurisdictions (counties, cities, etc.) require permits to start home construction. However, the percentage of starts in jurisdictions that do not require permits has been declining.
Starts lead actual construction
A no-brainer regarding statistical relationships in construction data is that starts lead actual construction activity or outlays. For housing, it generally takes four to six months to finish a house from start to completion. The table below shows the allocation percentages used by the Census Bureau in allocating the value of starts over the life of the project. These numbers are based on research on construction outlay patterns. For example as seen in the first part of the table, for starts that occur in January, 15.8 percent of the project's value of outlays occur in January, 21.7 percent occur in February, 21.0 percent occur in March, etc. Most of a residential project is completed within four to six months. These are average figures - some are completed faster and some are completed slower.
Monthly Progress Patterns for Private New Single
Family Residential Buildings by Month of Start
(Percent of the value of units started monthly)
Source: U.S. Bureau of the Census, "Construction Methodology" for value of construction put in place. Note: Month of start is first month of activity.
In the chart below, we see that starts were flat from early 2005 and into early 2006 except for some volatility in late 2005 and early 2006. Essentially there was a flat trend starting in early 2005. But residential construction outlays continued on an uptrend through all of 2005 and only began to decline in early 2006. This reflects the lagged relationship between starts and actual spending. If starts began to rebound in December, it would likely be mid-2007 before an uptrend in residential outlays begins. Since the residential construction outlays numbers are the primary input in the residential investment component of GDP, we can expect negative numbers for residential investment for another couple of quarters unless some weather-related effects temporarily boost starts and outlays for several months.
Employment lags construction spending
Hiring is an expensive process - and so is laying off workers. Any change in a company's workforce requires spending time and money. So, hiring and firing occurs only after changes in labor demand are firmly entrenched. In construction, this means that more workers are hired only after it is clearly apparent that there has been an upturn in construction activity. It is easier to have workers put in more hours until there is greater certainty that the need for the workers is going to last. As shown in the chart below, changes in residential construction workers lag the changes in starts by several months. Given the recent and sharp drop in single-family housing starts, it is likely that we will see a decline in residential construction workers well into 2007. Single-family housing starts have fallen 35 percent from their recent peak back in January 2006 whereas residential construction employment has slipped only 2 percent since its recent peak in September 2006.
The bottom line
Indeed, there are signs that housing is beginning to level off. Mortgage activity is starting to nudge up, mortgage rates are lower than several months ago, and pending home sales are no longer in a free fall. But Vice Chairman Kohn is correct - if past patterns hold true - that we have not seen the bottom in any of the major indicators for housing. We will see a rise in pending home sales before we see an uptrend in home sales. Housing starts will continue downward until the overhang in unsold home is worked down to an acceptable level. Construction outlays will tend to be negative until starts pick back up. And it is likely to be well into 2007 before we see a rebound in residential construction employment. The bottom line is that there are indications that housing is leveling off but the lagged effects of the decline will continue for a number of months.
![[Econoday]](../../../../../../../../../images/logo.gif){kind=logo}