Yesterday, I examined the effects of housing inventory trends on market prices and more specifically, the effects of the price of new listings hitting the market each week.
As soon I finished that article (well, while I still working on it…), I started thinking about the subset of housing inventory that enters the market each week – the new weekly new listings – sellers that put their house on the market for the first time. From yesterday’s post, the median price of new listings are showing definite trends depending on the market. Throughout 2009, this change in the housing inventory mix appears to have an effect on the market median ask price. Using San Diego as an example:
As the price of new listings (orange curve) steadily increased throughout the early Spring in San Diego, there was a fast rise in the market median ask price (black curve) – you can see a faster rise in prices early in the Spring than later in the Summer. As the Summer progressed, the price of new listings leveled off and started to decline in August. It’s at this time that the slope of the black curve representing median market ask prices) starts to flatten. (If you don’t believe the math I’m inferring here, you can estimate this effect quickly… Look at market ask price changes from 4/29 to 5/19 – median price rose more than $25k, but from 8/7 to 8/27 the market ask price rose by far less than $25k.)
This effect is even more obvious in Houston:
The “change in the mix” argument has some real merit, or so I think. But it doesn’t seem to be enough to explain the all of the changes in the market median price. In San Diego, new listings account for about 10% of the overall active inventory on average (and about the same in other markets), though accounting for more and more of the active inventory on a percentage basis each week:
There has to be an additional influence to affect median market ask prices is so rapidly. That’s when I thought that I should consider more than just the rolling inventory of homes available for sale, but also price changes of the existing inventory. It’d be nice to know what percentage of on-market homes have experienced price reductions. As luck would have it, we happen to calculate this stat every week here at Altos Research. (It’s the “Percent Price Decreased” stat in Altos-speak.)
Adding this stat into market analysis yields some additional perspective. Check out these San Diego, Phoenix, and Houston, overlaying the price of new listings and the Percent Price Decreased.
In San Diego, the higher new listing prices were coupled with fewer price reductions of the housing inventory already on the market – a definite sign of market strength this Spring. Notice how the rate of change in the Percentage Price Decrease stat slows simultaneously with the drop in new listing prices in August. That’s market efficiency for you. New sellers are indicating a weakening market, and more sellers may begin dropping their price this Fall. In either case, add this with the stabilization of inventory and their an indication of potential weakness for the San Diego market in the very near future.
Phoenix, which was grouped with San Diego in yesterday’s post, is showing nearly exactly the same effect for Percentage Price Decreased:
And Houston? (a market with consistently rising inventories this Spring)
In Houston, more and more on-market sellers experienced price reductions starting in May – about two months before new sellers entering the market started doing so at lower levels that the week before. (Don’t let the scaling on the right axis fool you – the Percentage Price Decrease is only up 1% since May, but it is clearly increasing and this is a 90-day rolling average so the weekly numbers are more severe on the up side.
It’s difficult to see the causality right now – do price decreases of the active housing inventory cause new listings to enter the market at lower levels? Or do new sellers that enter the market at lower prices cause the existing sellers to drop their ask prices? ( Or potentially neither – perhaps it’s that the two stats are coincidentally correlated, though I find that unlikely.)
Taking a national composite view of these housing market stats, the Altos 20-City Composite which includes the identical markets as the S&P-Case/Shiller 20-market index shows similar trending:
The price of new listings leads the downward turn in the overall market median ask price, and the existing sellers on the market appear to coincide with this turn with more and more sellers beginning to drop their prices in July – the same time that new sellers began entering the market at lower price levels:
After this initial analysis, my conjecture is that the effects are simultaneous because sellers are using the same local information to reach their ask prices. Sellers, whether they be on the market already or newly listing their homes, take a look at market activity each week and make an ask price decision. For new sellers, their decision is where to set their initial ask price. For existing sellers, the decision is whether or not to decrease their current ask price. Both decisions are highly emotional for sellers with a good bit of idiosyncratic information based on their individual situations.
In either case, housing market sellers are providing a way to project market-clearing prices that won’t be reported for several months down the road through traditional channels.
I’d like to dig deeper to run correlations and examine longer time series, which I’ll get around to, but thought I’d share some of my data geekdom.