Here are the 90-day moving averages of the median list and sold prices for detached homes in SD:
The 90-day median list-price-per-sf on this graph has blown through the roof – why?
It is purely from sellers’ greed, or is the data skewed?
The 90-day moving average of list prices could be influenced by a surge of higher-priced homes coming to market, or a combination of more homes listing at higher prices. The local MLS doesn’t allow median prices to be computed on samples of more than 500 listings, so let’s just check the last three months’ in NSDCC.
Here are the number of homes listed each month in NSDCC, and their median cost-per-sf:
| Listed <$800K | JAN | FEB | MAR |
| # listed | |||
| Median $/sf |
| Listed >$800K | JAN | FEB | MAR |
| # listed | |||
| Median $/sf |
The rapid increase in the median list prices was caused primarily by a surge of more-expensive homes coming to market (a 35% month-over-month increase of $800,000+ listings in NSDCC), which would cause the median to rise quickly. But how many people will look at the graph above and think sellers have gone crazy – I know I did!
Outstanding detective work, JtR!
You win one free internet for today.
On a serious note, are we to infer that “Sold price/sq ft” is also being skewed by those higher price listings?
I don’t know, but before I check, I’ll guess that the higher-priced homes are trending downward, inhibiting the overall average.
I was wrong – a big surge of NSDCC detached closings in March, but the under-$800,000s were flat, price-wise. The higher-priced median $/sf popped nicely:
Under $800,000:
Jan – 179 sales, $275/sf
Feb – 189 sales, $270/sf
Mar – 247 sales, $275/sf
Over $800,000:
Jan – 73 sales, $378/sf
Feb – 98 sales, $350/sf
Mar – 132 sales, $403/sf
Wouldn’t the price per sq. ft. come down if more larger homes are listed ?
Because the ppsf gets smaller and smaller as the square footage goes down.
Sorry. I mean –
Because the ppsf gets smaller and smaller as the square footage goes UP.
I just did a redfin search of sold SFRs in 92130 in the last six months.
1750 sq. ft to 2500 sq. ft: $337/sf
2500 sq. ft to 3500 sq. ft: $303/sf
Intersting how when price goes up, the ppsf goes up, but when the sq. footage goes up, the ppsf goes down.
Not sure what to think of this but it’s interesting.
Your theory is accurate in a perfect world. But the larger homes come with bigger lots, more extras, and best locations which combine to skew the $/sf higher.
Example – detached closings 1Q12:
Carmel Valley – avg. 2,200sf, avg. $311/sf (143 sales)
RSF – avg. 5,423sf, avg. $411/sf (49 sales)
Del Mar – avg. 2,623sf, avg. $536/sf (39 sales)
RSF home sizes are more than twice as big as CV, yet the avg. $/sf is higher due to the other goodies included. Then it really goes bonkers when you step into Del Mar, literally right next door to CV – yet look at the differences.
Those are the 1Q12 average $/sf stats above, here are medians:
CV – $311/sf
RSF – $378/sf
DM – $472/sf
I like to unpack statistics like this for errrors. Here, the local paper posts house sales figures by zip code, moving from county to county over a two month (or so period). But, depending on the month, and how densely populated the particular area is, a zip code might have only 2 or 3 sales in a month, and those sales can be completely different types of houses. Literally one showed a 200% drop in house price in a zip code. Why? The year before 2 houses had sold in a month, one owned by high net worth person. The next year, one sale, in a much smaller area. Ha.
Wouldn’t the huge gap be significantly narrowed if the surge of new higher priced inventory were initially priced correctly? I could sell a hundred new Porches today, if the price was right.
Interesting, Jim.
I was wondering if Redfin takes the average of the average or takes the sum of the costs divided by the sum of the total square footage. That would make a difference, too.
i.e. if you have one 4000 sq. ft house for $2M ($500/sq. ft) and a 2000 sq. ft house for $500K ($250/sq. ft), would the average be
$250 + $500 / 2 = $375
or
$2.5M / 6000 sq. ft = 416
@sdduuuude; awesome question. . .
As a person with a business that deals in averages and standard deviations, your question is perfect. I have found many of our competitors doing it the wrong way (your second example) and SQL makes it very easy to do it it incorrectly as well (again your second example). You have to intentionally calculate the $ per square foot for each house first, then average them. Easy enough to do in SQL, or any other database, but you have to think first.
Your second example is the $ per square foot if you bought everything.
Your first example is the average $ per square foot paid; which is the correct metric to use.
I really hope that Redfin, and the such, are doing it correctly.