You said “So if the mix trends more towards middle or upper end homes your aggregate is going to go up more. ” and I’m not sure this is the case. If the index tracked the median sales price, then you would be correct. Here, Case-Shiller tracks repeat sales. So, even if the mix trends towards higher price ranges, sales in those higher ranges could still be trending down.

I’d be interested to hear his explaination.

Regardless of why, it still doesn’t sit well that the entire MSA goes up by 18% while each tier improves by much less.

I highly doubt that there is anything wrong with the Case Shiller index. I think it may be in the way you are interpreting the breakdown by tiers.

The aggregate does calculate out to 18.3% YOY. And what is causing the disperaity you see is probably the mix of housing between the tiers. If you look at the Aggregate index in 09 its closer to bottom tier index. Probably showing that most of the homes sold in 09 were low cost homes.

In 2010 they are a bit (although very small) closer to the middle tier, signaling that more mid tier homes have been selling. This might explain why the aggregate figure goes up.

The aggregate is just the summation of all the homes. So if the mix trends more towards middle or upper end homes your aggregate is going to go up more.

Trust me, I definitely believe that there are some in the real estate field that would like to put a rosy lens on real estate trends. But I don’t believe that Schiller is one of them.

Or simply not having much of an impact on the aggregate because there have only been a handful of 1M+ sales.

Therein lies the problem. Case Shiller’s definition of “high end” is much lower than what you and I (and many other bloggers) consider high end. Like it or not, the 500-700K segment of homes is much bigger than the 6-700K segment and the 1M segment combined. What we are seeing is gains on the low part of the high end are carrying the day and the stagnation, to possibly slight downturn you and I see in our definition of “high end” is going undetected.

I only focus on the tiered data myself. If you look at long-term charts, though, the Aggregate does seem to sit quite well inside the tiers data – usually near the middle tier. So whatever wackiness they have in weighting for the average, seems to come out in the wash in the past…

One thing about it passing your sniff test is that the SF MSA is an awfully large place, and the best neighborhoods are only a small piece of the pie – so what happens in large communities with a lot of turnover like Antioch is much more important than a small town like Alamo.

So I think many of us watch ‘good’ neighborhoods which are pretty far off the SF MSA average. And although we know what’s happening in ‘bad’ neighborhoods, we might weight, in our minds, two cities equally (or even unweight the ‘bad’ neighborhood), whereas C-S will be using sales volume as the base-weight – so 100 sales in Hercules matter 10 times as much as 10 sales in Orinda.

To me, the value in C-S is saying what the larger market is up to in the SF MSA, and then you have to think ‘how far can my special neighborhood deviate from this average?’ ‘how much more (or less) are people willing to pay to live here vs. Antioch, or the middle of Oakland?’. Because if prices fell 40% in a ‘bad’ neighborhood that’s not too far away, that WILL affect the good neighborhoods, despite dreaming homeowner’s passionate denials that they are too special.

So there are always going to be differences between local observation and C-S aggregated numbers, but they ought to be explainable.

By the way, I recommend using the Not-Seasonally Adjusted data. I want to know the seasonality – I’m smart enough to understand it and not misinterpret it, and all Seasonal-Adjustments are prone to error, especially during market volatilty.

Love your site – more East Bay detailed commentary is ambrosia!

In 2010, the weighting metrics are favoring higher-priced homes as “better” comps…but why?”

One thing that’s worth noting is that the tiers are based on the first sale, not the subsequent sale. So if a house last sold in 1995, you’re using that price to set the tiers. As an addition factor, Case-Shiller weights recent sales more heavily because even though Case-Shiller removes obvious renovations, it misses a few, and it’s more likely that an older sale has been renovated.

If you combine these two factors, a 2005 first sale would likely be at a higher price than a 1995 first sale, AND the 2005 first sale would be more likely to be in a higher tier than the 1995 first sale. Then, in the weighting process, the 2005 first sale would be weighted higher for being more recent. When you combine these two factors, you probably account for why the higher tiers are exerting more influence right now.

As to the lower tiers exerting more influence before, it’s worth noting that in late 2008 and much of 2009, not much of the high end was moving very much. So lower tier sales in outerlying areas exerted undue influence on the index.

Besides, if this was rampant at the low end, you would expect the lower tier to show appreciation above 18%.

This still doesn’t explain how the aggregate total is so much more than any of the tiers that make it up.