Chart challenge: San Francisco mimics Japanese land value decline (sort of)

Accepting our recent challenge to “show us a legitimate graph of home values for San Francisco, or the Bay Area, or California, or the Nation, hell you might as well make it the world for that matter, that looks like [the declining chart on this post]”, Paul G steps up to the plate with this chart and explanation below:

You did open the chart challenge up to the world so I thought I’d submit this one. I can’t upload images to your web-site so I’m emailing it to you. Feel free to post, if you want.

The green line represents the cost of residential land in six major urban centers in Japan since 1955. The blue line represents San Francisco prices using the Case-Shiller index, but, in order to make the peaks coincide, I’ve moved all SF values back 15 years.

No, I don’t think San Francisco prices are going to follow the same trajectory down for many reasons (inflation, the 30-year run-up in Tokyo compared with the 10-year run-up here) but it’s still an interesting graph.



Thanks for the graph! Very interesting to say the least.

Chart Fun: Declining Real Estate Values [theFrontSteps]

San Francisco or New York City, if you had to choose [theFrontSteps, Battle Royale]

12 thoughts on “Chart challenge: San Francisco mimics Japanese land value decline (sort of)

  1. Let’s not forget though that the Case Shiller Index is utter BS when it comes to the actual SF market. From <a title=”San Francisco Real Estate Blog” href=”Case Shiller Index” target=”_blank”>a recent blog post:

    …the thing that makes me think the numbers are somewhat useless [in a market like San Francisco], is that (as pointed out) the index only tracks single-family homes (not condominiums which represent half the transactions in San Francisco), is imperfect in factoring out changes in property values due to improvements versus actual market appreciation, and includes San Francisco, San Mateo, Marin, Contra Costa, and Alameda in the “San Francisco” index… which in my mind is their greatest error.

    But the graphs do look remarkably similar…. eery.

  2. Luba, I tend to disagree. The market for SFHs and Condo’s are not all that different and will always follow a similar-related trend in price. That is to say that if SFH’s in SF take a 5% hit there would likely be some similar / related hit to the condo market. It’s not like most cities where the stock of condo’s is sifnificantly cheaper than the housing stock. The average price PSF between SFH’s and Condo’s is fairly close making the tradeoffs between the two pretty easy.

    Who is this Paul person and why don’t we hear from him more often with these great, highly entertaining data points. Thanks Paul!!! Post more.

  3. “The green line represents the cost of residential land in six major urban centers in Japan since 1955.”

    Is that a valid comparison between “land” and CS? I don’t know this subject enough to make an argument. Someone please comment.

  4. By the way, a graph is however you make it to look.

    Paul aligned the top. However, take a look at the start of the blue line. That’s 2X where the green line is.

    A better comparison is to adjust the line proportionally to align the starting point…when you do that, the whole blue line will be reduced by half. So it will peak at about 100.

  5. John,

    I used land prices because that’s the only index available for Japanese real estate. It’s missing the cost of construction, of course. If a house price index existed, I think it would rise and fall more slowly than the land price index (because construction costs are more stable over time). In other words, it would make the bubble in Japan seem less severe than the bubble in SF.

    You’re proposing rescaling using the 1969 value (1984 in SF) as the base. It’s true that would show a comparatively bigger peak in Japan. But if we picked 1981 as the base (1996 in SF), the peak would actually be higher in SF i.e. the market rose by more in SF between 1996 and 2006 than it did in Japan between 1981 and its peak in 1991. In reality, I simply used the actual index values (which are both based on a value of 100 in 2000) and graphed all of them.

  6. PaulG,

    The problem with using the peak for alignment is that you can make everything like that graph, and the graph loses the meaning.

  7. It’s an interesting graph. However, I think matching graphs by peak is like back-testing your stock investment theories — you can create a strategy by using back testing and tweaking controls to make the graphs match up. But it has no bearing on the future performance of the strategy…

    Personally, it may be more telling to understand how the japanese real estate bubble came to be, and why it crashed. I was told by a friend who lived there the bubble was clearly a speculative bubble, and the crash came to be as the banks became reluctant to lend (just like now in the US).

    By the way, a friend of mine just bought a 3 bedroom house in Chiba city (45 min train ride into Tokyo), and his interest on the loan is 1.3%!!!!

  8. Alex asked for a non-fake graph that showed prices going down over a generation. I showed him one: house prices in Japan in 2008 are where they were in 1985. (In real terms, they’re where they were in 1981).

    For kicks, I overlay the San Francisco graph to show that the rise in prices here over the last decade was comparable to the rise in Japan in the ten years before their bubble burst. I lined up the dates of the peaks because that’s the simplest way to compare. I didn’t adjust the values (or “tweak the controls”) so I don’t think it distorts the data at all. The data is what it is.

  9. Our economy showed a .9% gain apparently, I read yesterday. I wonder what a chart for the accompanying Japanese economy looked like. Because I have read repeatedly that Japanese R.E. values went hand in hand with economic collapse.

    That said, yeah, Paul G. Alex asked and you delivered.

  10. PaulG,

    Of course you adjusted the values, when you adjusted the peak. The values are just index numbers. They are not absolute values. So, if you use 100 for year 2000 and 200 for 2008, it is no different from using 50 for year 200 and 100 for 2008.

    The point is, I can align everything this way. For example, there was a peak for the blue line at year 2000…or 1985 after you shift by 15 years. I bet if you make that the peak, the two lines match too.

  11. I was in Japan in from 1991 to 2005. Property prices went down double digits almost every year during that time. My in-laws saved up their entire lives for a retirement property which they paid USD 350,000. It is now worth about USD 140,000. This will happen in the U.S……. believe me.

  12. I have a question for Dave Jones (or anyone else who might know the answer to this question). In 1993 or so in Japan, after prices fell 20% or so, what was the word on the street at the time in terms of the future of real estate prices and the economy generally? Were people talking about the ‘recovery’, and how the worst is behind us and how real estate is near the bottom, etc., as seems to be the consensus here?

    On the brighter side – what about Japan as a real estate investment now? Can American citizens own land in Japan? Real estate as an investment in the US at the moment is a shaky proposition. But surely, after twenty years of decline, Japan has come close to a bottom.

Leave a Reply