Medians are for Dummies

[powayseller]

Medians are for dummies. They are a 18 month lagging indicator of the market. They do not tell the whole story. Enough said.

[an]

Agree, but it’s really sad that the majority of the public use those data to make their decision to buy and sell.

[powayseller]

It really is…Reporters in bubble towns really need a realtor license and MLS access, so they can do their own research. This is too important. Instead, they rely on Dataquick.

OTOH, if I were a seller, I would be very grateful of the optimistic slant given by the median and the U-T article. I was sweating bullets last January, while the headlines were still screaming “Real Estate is UP Again”. If I were a seller and the U-T was publishing facts, and giving us the real story inside the median, I wouldn’t be able to sleep. I’d be too worried that my buyers would catch on to the ponzi scheme and bail…

[Bugs]

I had an editor tell me once that the reason that content sometimes seems so lopsided is because it’s very common for the side that’s selling to be more vocal and more outgoing than the side that’s hedging. Squeeky wheel gets the headline, I suppose.

Just wait until the tide of public opinion swings the other way. We’ll be overwhelmed with the stories of the victims and the media will be looking to point the finger of blame at the various players.

[Peace]

Could someone please give me a crash course about how these figures are 18 months old? I know I could probably find this info out for myself but there is so much to read and so little time. Thanks

[powayseller]

Prices have been falling for a long time. My own experience is that my house would be very competitively priced at $830K in October 2005. This is not a overvalued listing, but a realistic price based on the last comp sale just 3 months prior. My realtor had sold the comp, and knew how it compared to mine. Important point is it was a realistic pricing. I had lots of traffic, so I was priced well. One buyer walked when I said I would not consider offers under $780K. WE had range pricing $780 – 830K. I had 2 offers at 780. Took the 2nd offer. So I took a 5% drop. Meanwhile, median was still going up.

Look at the bubble blogger links, for more stories of prices dropping. Most sellers are getting 2005 or 2004 prices. Those who bought in 2004 or 2005 and are selling, are bringing money to closing.

Yet, today’s U-T shows SFR is up 2% from last year.

First, this is a distortion of the inside story. Each individual home is down to 2004 or 2005 prices, but the rich people are still buying, skewing the distribution of homes sold. So while it is mathematically correct that the median is rising it DOES not mean that each house is worth 2% more than last year.

Second, the median was rising for the summer of 2005 and then dropped. It has been dropping since last fall, but the median is reported as year-over-year. So you don’t see the median reported as falling until it has been negative for ONE year.

This is so different from the way other economic data is reported. CPI, unemployment, and many other indicators, are reported month-to-month. I think a month-month indicator, along with yearly, is useful.

A more useful gauge of the housing prices is the data that Rich uses, the OFHEO data. It gives the price of the SAME house sold since the database started several decades ago. It tracks the price of the SAME house, not the change in the distribution of homes sold.

The OFHEO will show declining prices, but it is published many months late. It takes the government so long to collect the data.

Leading indicators are months supply, which incorporates inventory plus demand, and second is HAI.

By 12-18 months lagging, I mean that it will be the end of 2006 or middle of 2007 before the median for SFR homes is down 5%, the amount that my home was down in December 2005. So while home prices were down by 5% last winter, the public won’t be informed of this fact until 18 months later.

The DataQuick and NAR people are not interested in giving us accurate data, only in promoting real estate. We have all been misled. The only reason I know so much is because I have too much time on my hands and some good and smart friends who are realtors and figured this stuff out.

[sdrealtor]

I agree, the median price data is virtually useless

[North County Jim]

Most everyone here agrees the median is useless. There’s a lot of support for the idea that the mix of homes sold is skewing the median.

The problem is noone (at least to my knowledge) has yet presented data showing this to be the case. It seems to be all anecdotal.

Shouldn’t there be a relatively easy way of identifying this trend? Overall sales by zip code perhaps?

[North County Jim]

So I took a 5% drop. Meanwhile, median was still going up.

I don’t mean to pick on you PS but this illustrates the problem with using one data point to justify a conclusion. Did you necessarily have to take the 5% drop? How long were you on the market?

You were, it seems, an emotional seller who still hasn’t quite resolved whether you accepted the right offer. Could the comparable that you based the $830k number on been attributable to an emotional buyer?

What I’m getting at is the human element introduces a lot of noise into the data. Would two people selling the same house at the same time sell at the same price? Of course not.

[kiki]

“There’s a lot of support for the idea that the mix of homes sold is skewing the median. The problem is noone (at least to my knowledge) has yet presented data showing this to be the case. It seems to be all anecdotal.”

Jim the realtor published in his website data from the zips he track
************
Number of Sales, First Half of 2006
Number of Closed Sales, January 1 – June 30

2004 2005 2006 % chg. from 2004

$400K or less 324 61 63 -81%

$400K – $700K 1,065 1,185 859 -19%

$700K – $1M 332 396 318 -4%

$1M+ 97 115 84 -13%

Totals 1,818 1,757 1,324 -27%

(He removed 23 closings from 2006 of new homes from tract builders that wouldn’t have listed in the MLS in 2004)
************************

in these zip codes the sales of houses less than $700K have dropped by 34% while sales above $700K only dropped 6%. what i read from this is tht rich people are still being able to pay but the entry market is in fact suffering. i added the first two price ranges together because i think that even in 2005 the less than $400 were actually in the high 400s so they might be the only houses that have seen a litte appreciation – there might be fewer buyers but still this prices is attractive to generate a small price bid.

[Bugs]

The problem with comparing one year’s median to the another is that it’s based on the assumption that the same mix of units sold during both periods – this leads to the assumption that the typical home increased or decreased roughly in line with the averages. As long as the mix of units DOES stay constant from one period to the next this assumption is reasonable. It’s when the different market segments are acting in different ways that the assumption proves unreasonable.

The way I analyze it is by parsing the data into different categories in the same zip areas. Older homes tend to sell for less than newer homes and smaller homes tend to sell for less than bigger homes. This current cycle started in the late 1990s so I use that as a dividing line; and I use 3,000 SqFt as the divider in both age ranges. In effect, this puts the data into one of 4 categories. When you do this you can very clearly see the different distribution of data for this year vs. last year – we have more newer and larger homes selling this year, which has propped up the averages for the zip area.

There’s probably a dozen different ways to do this type of analysis, I just chose the parameters that were quick and easy.

[powayseller]

Jim, a few months ago, I kept reading on this blog that individual home prices were dropping, yet the median was still going up. I remembered my house sold for less than it could have last summer. So I started asking everyone I knew, “How come prices are dropping but the median is still going up?” I was given the answer by Bob C, who got it by analyzing the MLS sales.

The high end was 8.5% of sales last year, but is 10% of sales this year. Not a big difference, but enough to skew the median up.

My house would have sold for the higher price last summer. I explained the reason in another post yesterday.

I’m happy with my price, but not my husband. He wanted to hold out for a better offer.

[North County Jim]

Thanks. I feel better now.

[speedingpullet]

Actually, I sort of understand why the median was used – it can tell you quite a lot of things – but only in conjunction with the minimum and maximum data points, and number of data points used.

Sure, it will churn out a number, but if you can’t compare that number to the highest price (max) versus lowest price (min), then you have no idea whether the prices are skewed towards the bottom (ie a few very low prices pulling the median down) or positively skewed (more prices at the top pulling it up). Having the number of data points also allows you to calculate the other two ‘averages’ – mode and mean – which will give you a much clearer picture of whats going on over time.

Stuff like this is a bit of a bugbear for me – I spent many years of my life teaching Basic Math and Statistics at community college – unfortunately many people (and I don’t include the posters on this forum 🙂 ) don’t know the differences in the three averages, so just take a number (any number) as Gospel.

[powayseller]

Rich made some comments along this topic on the main page today. He explained that the Case Shiller index is a better indicator of price movement. For his own analysis, he used the OFHEO data, which tracks the sale of THE SAME home over time. This cannot be used for new homes, and the OFHEO data lags by many months, bec. the gov’t seems to take forever in compiling their data.

If you want to know how prices are changing, use Case Shiller index or OFHEO index. Forget the median.

[Author]

Posts