Housing starts fell to an annual rate of 458,000 in April, the slowest pace on record dating back to the 1950s. However, all of the drop was due to multiple-unit structures, which are very volatile from month to month. Starts for single-family homes increased 2.8% in April and are now up two months in a row. For the reasons set out below, we believe the recent turn in single-family housing starts is the first sign of a major recovery in home building that will add substantially to the pace of real GDP growth in the next few years.

Housing Inventories and Home Building

Normally, residential builders should start about 1.6 million homes per year. This combines two key factors: population growth and net scrappage.¹ The 1.6 million figure also includes the impact of mobile homes (which reduce the need for regular housing starts) and homes that are started but never finished.

In the first four months of 2009, builders started homes at a 511,000 annual rate, about one-third the long term trend. For the time being, such a slow pace of housing starts is understandable. Builders constructed way too many homes earlier this decade, resulting in an excess inventory of more than four million units in 2006. At present, the excess inventory is still about two million too many homes. Given the slow pace of home building, however, the excess supply of homes is falling rapidly. If we normally need to start 1.6 million homes per year but are only starting 511,000, the excess inventory will fall about 90,000 per month. At this rate, the excess inventory of homes will be fully corrected in two years.

The problem with this is that if inventories are reduced to normal levels with housing starts still near a 500,000 annual pace, then we would eventually end up with a massive shortage of housing. One scenario is that forward-looking home builders soon begin gradually increasing housing starts, so that by the time inventories return to normal, home building is close to the long-term trend of 1.6 million units per year. For example, if housing starts increase at a consistent 37% annual rate for the next four years, then in early 2013 inventories will be back to normal levels and at the same time housing starts will be at the long-term trend, neither too high nor too low.

In the meantime, the increase in home building will contribute about one percentage point per year to the growth rate of real GDP. However, this will not happen as soon as overall housing starts begin to move upward. It will take about six months to go from a turnaround in starts to a turnaround in overall home building, including all phases of the construction process.

New Home Sales

Normally, about 80% of the homes that Americans build (and buy) are single-family homes. If housing starts are usually 1.6 million, that means 1.28 million should be single-family units. However, not all of the 1.28 million homes will count as new home sales, because about 325,000 per year are built on land owned by the homeowner. Subtracting these suggests the official new home sales number should average about 950,000 per year.

In the first quarter of 2009 new home sales averaged 348,000, well below the long-term trend. Given the overhang of excess inventory, it is understandable that new home sales should suffer. Why buy a new home when there are so many foreclosed properties for sale as well as other properties being aggressively sold by homeowners who are short their mortgage balances? In fact, new home sales typically average 17% of all home sales (existing plus new homes). Lately, the share has been already down to 8%. It would be difficult for new home sales, which have dropped substantially faster than existing home sales, to continue falling. Given the weakness in sales in the past couple of years, the ten-year sales average is already down to 938,000.

Home Prices

Every three months the Federal Reserve publishes data on the market value of residential real estate. Along with its quarterly GDP numbers, the Commerce Department calculates the rental value of homes. We use these figures to generate a price-to-rent ratio for the housing market. As the next chart shows, the P/R ratio generally hovers right around 15.

As an example, if a home rents for $1500 per month (or $18,000 per year), the owner should be able to sell the property for $270,000 ($18,000 times 15). At the peak of the housing frenzy, the P/R ratio hit 22, which means a home renting for $1500 per month could be sold for about $400,000. At the end of the first quarter, the P/R ratio was back down to 16.1 and so the corresponding home price was down to $290,000, or only about 7.5% above fair value.

Assuming very modest gains to rents, home prices will be back to a fair value P/R ratio of 15 by late 2009. It is plausible that the remaining excess inventories in the housing market will temporarily drive the P/R ratio below fair value in early 2010.

However, much of the price declines still in the pipeline will be clustered in the states with the largest overhangs of inventory, including California, Arizona, Nevada, and Florida. Many other areas around the country are likely at or near fair value already and prepared to show price gains as the broader economy recovers.

¹ Net scrappage is the number of homes destroyed by fire/disaster plus knock downs, minus the conversion of other types of buildings (such as old warehouses) into homes.

May 26, 2009

First Trust Economic Commentary

Brian S. Wesbury, Chief Economist

Robert Stein, Senior Economist