What the Median Does Not Tell You

[powayseller]

The media’s reporting of the median is a great disservice. While the median is mathematically correct, it does NOT tell us the value of a particular home.

The median tells us about the distribution of homes sold. Furthermore, since it is reported year-over-year, the distribution of homes sold must be changed for one entire year before the median informs us of this fact. In San Diego, the median peaked in November 2005. I expected the media to report that the median had dropped every month since then. But they did not. Only in June 2006, when the median was below the June 2005 price, did the NAR and media reveal to us that the median had dropped. Because of this year-over-year reporting, the median lags the price changes in the real estate market by 1 year or more.

If you make an ordered list of all homes sold, the median is the price of the home in the middle.

Home prices can be dropping for one year, while the median is still going up. This fact puzzled me. As I sold my own home in a softening market, and my realtor told me numerous stories of sellers getting below summer 2005 sales prices, I started wondering how the median could still be rising.

My questions led me to Bob Casagrand, a realtor who writes a monthly detailed report on the San Diego housing market by digging inside the numbers.

Bob Casagrand taught me about the hidden stories in the median this spring, and I have been writing about it on this blog ever since. Mainly, the median is skewed by a change in the mix of homes sold; if you sell more expensive homes, it goes up; if you sell more condos and fewer luxury homes, it goes down.

I came across this quote from ASU’s website, which explains what we see now is typical. The figures are for the Phoenix, AZ market.

“As the housing market slows, the entry level sector basically disappears, while the move- up markets garner a larger share of the remaining activity.
Thus, it is not unusual to see median prices to be fairly stable or even increase as the move-up market works to be satisfied. For July 2006, 14 percent of all recorded sales were for homes priced from $125,000 to $199,999, 45 percent for $200,000 to $299,999 and 38 percent for homes priced over $300,000.” – Phoenix market, from ASU website

The San Diego market started changing in the summer of 2004. (See CAR website – median price has been flat since then.) Inventory had stayed flat at 3,000 up to March/April 04, increased to 6,000 by June, and then to 12,000 by September. As inventory rose, homes started lingering longer on the market, and prices stopped rising. Yet the median kept going up.

Inventory probably was rising because sales were falling. Higher prices and interest rates were squeezing out the first time buyer. Sales in the under $500K market decreased first, skewing the distribution of homes sold upwards, since fewer of the sales were in the entry level category.

With rising inventory, comes flat and then slowly declining prices. Days on market increased, and many sellers could not sell their homes to move up to the next tier of home. My neighbors took their entry level home off the market this spring, cancelling out their plans to move to a mid-level home in Temecula.

Thus, the ripple effect moved on. As the entry level buyers were reduced, sellers of those homes could not trade up to the mid-level homes. So after a few quarters of reduced sales at the low end, the middle tier was affected. By fall of 2005, the mid end was weakening too. That’s the time I had my own home on the market, and both offers I received were 5% below the comps we were using from just a few months before.

However, the + $1 mil home market was still hot. Those buyers had not yet felt the ripple effect.

This year, the high end made up a bigger percentage of sold homes. Last year, 8.5% of homes sold were above $ 1mil. This year, it was 10% of homes sold. Thus, the median kept going up, because more of the homes sold were the expensive homes.

But this doesn’t tell us that each individual home was flat or declining since the summer of 2004. They also don’t tell us that sellers were starting to pay closing costs and pay for more of the buyer’s repair requests, effectively reducing the sales price by 1% more. Hidden in the price are builder incentives, and paying realtor commissions, for total value often amounting to 5% or more.

So let’s use some numbers as examples, to see how the median can go up when each home’s value is actually falling. Below is a made-up list of homes listed in both 2004 and 2006 with its sale price ($0 if not sold). For simplicity, each 2006 home is sold at 10% less than the 2004 price:
2004 2006
$400K $ 0
$400K $ 0
$400K $ 360K
$670K $ 603K
$700K $ 630K
$700K $ 630K
$800K $ 720K
$900K $ 810K
$1mil $ 910K
$1 mil $ 910K
$1.5mil $1.35mil
median 700K $ 720K

So you can see that the lower end is not selling as much, prices are down 10% for each home, but the median is up 3%.

It is my understanding (no one has told me this next part), that this will work in reverse on the way down. Eventually, home prices will be back to a level where investors can be cash flow positive on rentals. They will start buying up the lower end homes, increasing the number of cheaper homes sold, and making the median move down further. The media, still stuck on the misleading median, will report that the median made a bigger dip, while in reality, demand is picking up, inventory is decreasing, and prices are rising.

After a quarter or two of increased activity at the low end, the mid level price range wil pick up, and finally, the high end. By the time this all occurs, at least one year will have passed to see a change in the month to month median. An increase in the year to year median, which is what the NAR, CAR, and Dataquick and the newspapers report, will be lagging by one more year.

So John Q Public will not know officially about the turnaround in the market for about 1-2 years after it has happened.

A more accurate indicator for tracking price changes is months inventory. It is the inventory divided by the current months’ sales. Also check HAI (housing affordability index), a NAR quotient combining prices, income, and interest rates. A quotient of 100 means that 100% of the median wage earner can afford the median priced home. In San Diego, were are around 9%. That means only 9% of San Diegans can afford the median priced home.

A more accurate indicator is the OFHEO house price index, or the new Case-Shiller index. Rich Tosano uses these in his charts. These indices track the price of the SAME house. You will see right away when the prices change. I hope the media will start reporting more accurate data.

Schahrzad Berkland

[carlislematthew]

So John Q Public will not know officially about the turnaround in the market for about 1-2 years after it has happened.

I can’t disagree with any of your analysis and commentary, but I’m not so sure about the particular quote above.

During the last couple of years, the NAR had an incentive to rely on the misleading median figure, as it kept the market and the speculation going as prices looked to be rising still. However, if/when your theory pans out (which I think it will, BTW) then the NAR may suddenly have a revelation! “Wait”, they’ll say, “the median is misleading, and this is why!”.

It will be in their interests to be able to demonstrate to the public that things are picking up and that “it’s a good time to buy”. The media just echos what they say anyway, and so I see no reason why they won’t do so this time. They’re in real-estate after all – they’re the EXPERTS!

I’m wondering if anyone who remembers the last SoCal real estate pop could help out here? I’m assuming that the recovery will probably be similar. What did the NAR say at the time? Was the median still going down while the recovery was actually in the early stages of the recovery?

[SD Realtor]

PS that was an EXCELLENT post.

[Peace]

Thank you again, Powayseller, for taking your time to educate us on this point. Great post and great illustration.

I think a similar phenomenon is what keeps the median income looking positive. Even IF salaries are increasing, the demands on incomes in the guise of benefits and other inescapable cost of living increases results in real wage loss. And the astronomical difference between the upper incomes vs what used to be the middle incomes is really skewing the median.

And this relates back to the housing affordability index. A $50,000+ salary a few years back was quite a bit different in actual net/disposable income than it does today.

[powayseller]

Interesting point. I remember Thornberg talked about rising benefit costs of employers were taking away from wage increases. I had not considered that wages were also reduced on the employee’s side by higher insurance co-pays. As far as cost of living, that is reflected by using *real* per capita income. I believe per capita income is adjusted by inflation. Of course, I would argue that inflation is actually much higher than the offical CPI, so wages are indeed falling even if the *real* number is the same.

[carlislematthew]

I would argue that inflation is actually much higher than the offical CPI, so wages are indeed falling even if the *real* number is the same.

I agree, especially if you’re referring to the “core” CPI. Sure, energy prices are volatile, but over the last few years the trend has been relentlessly upwards.

[Author]

Posts