by Calculated Risk on 3/17/2009 04:25:00 PM

Update: Please don't confuse a bottom in single family housing starts with a bottom in house prices! See next post: Housing: Two Bottoms.

The title to this post would have been laughable in 2008 or 2007, but as I noted in Looking for the Sun, there is a reasonable chance housing starts will bottom sometime this year - so I suppose it is not too early to start looking.

A few key points:

Single-family starts may still fall further. Although I've started looking for the bottom for housing starts, this seems a little early in the year.

Focus on single-family starts. Ignore total housing starts. Multi-family starts are very volatile. Single family starts increased by 1.1% in February - not much.

One month does not make a trend, and housing data is revised significantly. The slight increase in February might just disappear.

Permits for single-family housing units jumped to 373,000 SAAR (starts were 357,000 in February). This suggests starts will probably be higher in March.

Supply is still too high. Months of supply was at an all time record high 13.3 months in January.

Click on graph for larger image in new window.

From the Census Bureau: "The seasonally adjusted estimate of new houses for sale at the end of January was 342,000. This represents a supply of 13.3 months at the current sales rate."

But the increase in Months of Supply has been driven by the denominator (sales), even though the numerator (inventory) has been falling steadily. note: Months of supply = inventory / sales.

The second graph shows the level of hard inventory for new homes (completed plus under construction). With starts below sales, hard inventory has been falling for some time.

Unless sales fall further, the months of supply should start to decline even with the current level of starts.

And this brings up a key point:

It is incorrect to directly compare monthly housing starts to monthly new home sales. The monthly housing starts report from the Census Bureau includes apartments, owner built units and condos that are not included in the new home sales report.

However it is possible to compare "Single Family Starts, Built for Sale" to New Home sales on a quarterly basis. The quarterly report shows there were 65,000 single family starts, built for sale, in Q4 2008 and that is less than the 82,000 new homes sold for the same period. This data is Not Seasonally Adjusted (NSA). This suggests homebuilders were selling more homes than they are starting – but not by much.

However starts have fallen much further in Q1 (almost 25% from Q4) although sales have fallen too (we only have January data for sales so far).

Note: new home sales are reported when contracts are signed, so it is appropriate to compare sales to starts (as opposed to completions), although this isn’t perfect because homebuilders have recently been stuck with “unintentional spec homes” because of the high cancellation rates.

This graph provides a quarterly comparison of housing starts and new home sales. In 2005, and most of 2006, starts were higher than sales, and inventories of new homes rose sharply. For the last several quarters, starts have been below sales – and new home inventories have been falling - but it continues to be a race to the bottom between starts and sales.

Of course, unless sales stabilize soon, starts might have to fall further.

The housing market is still very weak. The NAHB homebuilder index was near the record low this month. Housing starts are just off the record lows of last month. New home sales were at a record low in January. And sales and starts are nominal numbers - not adjusted for changes in population!

Do not expect a sharp rebound in housing starts. Even if this is the bottom in starts (or close to the bottom), there is still too much inventory - especially distressed existing home inventory - for a meaningful rebound in housing starts.

Impact on the economy: If this is near the bottom for housing starts, then the drag on GDP from Residential Investment (RI) will subside (new construction is usually the largest component of RI). As completions fall to the level of starts, the drag on residential construction employment will also ease. And finally, in the temporal order of a business cycle recovery, usually RI has to bottom first ...