A Harder Look at Real Estate Statistics

The South Tahoe Association of Realtors (STAoR) staff went to a C.A.R. conference last week. Thus, they didn’t publish May market statistics until today. Honestly, I got impatient a few days ago and started mining our MLS for data. I had no idea what I was in for!

Two days, six phone calls, and five hours of data mining later, I have just a few nuggets for you. I’ll give you the biggest nugget first: The monthly median sold price that STAoR presents to us isn’t actually the median for that specific month. It is the median for the past twelve months. To clarify, STAoR gave us a median sales price of $428,000 for April 2008. I searched our MLS and found the following:

Median sales price for homes sold 4/1/08 to 4/30/08 – $379,000
Median sales price for homes sold 4/30/07 to 4/30/08 – $428,000

In fact, here’s a good bit of the data that I overturned:

Month & Year Listings at Month’s End Escrows at Month’s End Homes Sold Median $ Over Last 12 Mths. Median Sale $
May 2008

414

47

31

$425,000

$415,000

Apr 2008

372

44

21

$428,000

$379,000

Mar 2008

361

40

34

$425,000

$387,500

Feb 2008

374

52

16

$435,750

$412,500

Jan 2008

366

33

16

$449,900

$428,000

Dec 2007

382

28

34

$449,900

$502,000

Nov 2007

403

42

27

$445,000

$460,000

Oct 2007

353

15

39

$450,000

$425,000

Sept 2007

442

30

30

$450,000

$398,250

Aug 2007

538

45

38

$464,000

$421,500

July 2007

557

40

25

$463,000

$435,000

June 2007

552

33

35

$465,000

$440,000

May 2007

522

47

32

$465,300

$460,000

Apr 2007

423

32

28

$465,150

$392,500

Mar 2007

359

37

30

$475,000

$507,000

Feb 2007

327

25

25

$474,250

$492,500

Jan 2007

331

37

21

$474,500

$395,750

Dec 2006

339

29

40

$476,000

$497,500

Nov 2006

362

50

44

$474,500

$483,500

Oct 2006

411

55

53

$465,000

$450,000

Sept 2006

463

63

31

$474,500

$442,000

Aug 2006

548

46

36

$475,000

$427,500

July 2006

571

49

37

$485,000

$475,000

June 2006

532

51

37

$485,000

$480,000

May 2006

424

50

35

$485,000

$465,000

Apr 2006

302

50

32

$489,000

$504,000

Mar 2006

264

39

38

$489,000

$485,140

Feb 2006

262

51

21

$485,000

$489,000

Jan 2006

256

32

20

$482,000

$508,750

Dec 2005

198

n/a

31

$475,000

$450,000

Nov 2005

282

42

53

$475,000

$455,000

Oct 2005

300

69

55

$465,000

$485,000

Sept 2005

313

93

63

$455,000

$495,000

Aug 2005

270

96

79

$446,000

$510,000

July 2005

209

110

74

$430,000

$484,500

June 2005

188

113

65

$425,000

$456,000

May 2005

172

96

57

$410,000

$495,000

Apr 2005

116

91

72

$399,000

$488,500

Mar 2005

88

83

50

$390,000

$440,000

Feb 2005

72

86

34

$390,000

$449,500

Jan 2005

85

63

47

$383,250

$405,000

Before I continue, I’ll relay a couple things told to me by STAoR staff over the last two days. First, they are not the only ones who present data in this manner. They say that many boards in small markets do the same thing. There is good reason for this. It smoothes data and eliminates variance. With few homes being sold monthly in small markets like South Lake Tahoe, there can be spikes or depressions in the data that don’t accurately reflect the market. For instance, let’s say sixteen homes will be sold in June 2008 (extremely unlikely, but it happened in January and February). If nine of these homes are closed at over one million dollars, the median will be over one million, which is more than double the actual median value of South Lake Tahoe homes. STAoR eliminates variance by incorporating a larger data set.

They aren’t truly presenting a median sales price, however. Their statistic is more like the 50-day moving average that stock market analysts use. In fact, I call their stat a 12-month moving median. There are two problems with it: 1) It does not reflect what actually happened in a month; and 2) its data centers on a date half a year in the past. Check out the table. The 12-month moving median peaks in March and April of 2006. However, the market really peaked seven months earlier in August 2005. If someone was basing a decision to sell a few years ago on this set of data, they would have been seven months late and more than a few dollars short! In the same breath, when the market bottoms out, the 12-month moving median won’t reflect this until half a year later. It is in cases like this that the actual median is much more helpful in decision making.

I’ll be posting more reflections on my data mining in the next few days.

Posted on June 11, 2008 at 4:08 pm
Drew Kondo | Category: Statistics | Tagged , , ,

Leave a Reply

Your email address will not be published. Required fields are marked *