“the last shoes to drop, have dropped” Recession

New home sales: despite the noisy sharp increase in August, the last pre-recession metrics are firmly negative

 – by New Deal democrat

Have I mentioned before that new home sales, while perhaps the most leading of all the housing data, suffer from being very noisy and heavily revised? Yes, I think I have, just about every month. And this month’s report is a good example of why.

Let me start with prices this month. Last month it was reported that the median price for new houses sold was $403,800. This month that number was revised down -2.2% to $395,100, and this month itself jumped $18,400 from there to $413,500, a 4.7% increase! This number is not seasonally adjusted, so the best way to look at it is YoY (red in the graph below, left scale), which was higher by 1.9% this month:

The three-month average takes out most of the volatility, and so measured prices are down -2.8% YoY, and the declining trend in the absolute number (blue) is intact as well.

But the volatility in the price metric is chicken feed compared with sales (blue in the graph below), which jumped just over 20% (!) from an upwardly revised 664,000 last month to 800,000 seasonally adjusted and annualized this month, a 3+ year high:

I’ll come back to houses for sale (red) in a moment. But while mortgage rates did decline to 10-month lows in August (and more since):

That hardly suggests a complete turnaround in the market. For example, mortgage rates were lower last October, and there was no comparable jump in sales. So, to return to the theme of “noisy and heavily revised,” let’s see what next month’s revisions are before we make a champagne toast, because it is perfectly likely that this is an outlier.

Now let’s return to the number of houses for sale (red in the graph above), which declined another -7,000 from a downwardly revised July to 490,000 annualized, a declined of -2.8% from peak. This is nearly certain confirmation that this metric, which along with employees in residential construction is one of the last to turn in the cycle, has decisively turned down.

Here is the long-term historical YoY look at new houses sold (blue) and new houses for sale (red):

It is easy to see that the former (blue) always turns first. Further, it almost always has been negative for a year or more before a recession occurs. The latter (red) – the number of houses for sale – typically turns negative YoY close in time to when a recession actually begins.

Now here is the post-pandemic look:

This month for the first time in a year, the number of houses sold zoomed higher to 15.4% YoY, while the number of houses for sale decelerated to a paltry 4.0% higher YoY. It could easily turn negative by about the end of this year.

In sum: housing is recessionary, and the last shoes to drop, have dropped.