Less house for your money….
One of the problems with the maps I’ve been generating is that when one looks at income one has to take into account cost of living. Unfortunately I’m only finding state-level maps, not county-level ones. I’m sure there’re out there, but the US Census has a lot of data files handy if you are persistent and know where to look, so I’m going to poke around and do my own calculations. The Census has home values for 2005-2007, the peak bubble years, as well as median household income for the early 2000s. Naturally one expects the housing value and income to track each other. That is, areas with higher median incomes should probably have higher median home values. But, some areas will have lower incomes vis-a-vis the value of the housing stock, and others higher. Below is a scatterplot which shows median home value vs. median household income (log-transformed). The outlier is New York county, Manhattan.
Since Manhattan is such an outlier I dropped it out of the data set, and did what I did previously, I calculated the deviation from the trendline of each point, above and below. Those counties above the trendline exhibited greater housing values in relation to median income, while those below were the inverse. The map below is shaded blue for those above the trendline, and red for those below. Shading is proportional to deviation from the trendline.
Things that jump out at me:
1) 2-4 years ago the Washington D.C. area seemed to have relatively affordable housing stock in relation to median income.
2) The West coast did not. Urban areas such as New York or Boston exhibited a decoupling between income and home values, but much of their hinterland remained relatively affordable. The same does not apply to the West.
3) The “resort counties” in the Intermontane West are not a surprise.
4) Interestingly, unlike the Northeast the South seems to be characterized by large metropolitan areas such as Atlana, Houston and Dallas remaining affordable, while the hinterlands are not nearly as underpriced as one might expect. I think the issue here is simply very low income in the rural South, so that any upward pressure on home values is not matched by increased wages.
Bottom 10 below trendline
Loudoun, Virginia -$136661.4
Fort Bend, Texas -$116864.8
Goochland, Virginia -$116020.9
Prince William, Virginia -$115841.1
Stafford, Virginia -$109999.9
Calvert, Maryland -$109217.2
Campbell, Wyoming -$109136.2
Powhatan, Virginia -$106159.3
Howard, Maryland -$105386.7
King George, Virginia -$100267.0
Top 10 above trendline
Summit, Colorado $128227.5
Blaine, Idaho $129168.4
Santa Barbara, California $134643.1
Eagle, Colorado $171239.4
Teton, Wyoming $181064.5
Santa Clara, California $190223.8
Santa Cruz, California $195552.2
San Francisco, California $206232.6
San Mateo, California $218667.7
Marin, California $255949.7
If I put New York county back into the data, it is $822057.7 above the trendline.
Steves’ point: above trend line reflects scarce buildable land, due to geography, land-use restrictions or water availability relative to population growth. Below trend line reflects ample buildable land relative to population growth.
The median Manhattanite makes $66,000, but lives in a million dollar home? Wow! (I’m assuming these are natural logs.) One thing I noticed about the graph is that the slope is much greater than one. As people’s income increase, they spend a higher proportion of their income on living somewhere where home prices are high. In economists’ terms, living in an expensive city is a luxury good, not a necessity.
One thing this could be compared to is median income vs. home price within a county. Does the same relationship hold? Within a single county, do richer people pay a higher proportion of their income for housing? If so, that would mean people just liked to live in expensive houses. If not, that would mean that rich people want to live in the particular places that have high home prices (San Francisco, Aspen, Naples)
I think that Steve Sailer’s theory has limited explanatory value. It can explain why the West Coast is uniformly overvalued, but does not apply to undervalued cities like Philadelphia (riverside) and overvalued ones like Salt Lake City.
chemdude, i’ll try and look at microdata at some point….
NYC might reveal the following:
1) Lots of people live in rent controlled or subsidized housing.
2) Many owners are not claiming residency in NYC or the US, and claim residency in low tax countries even though they actually live and work (usually for themselves) in Manhattan.
Salt Lake City is wedged in between mountains to east and west, and lakes to north and south.
Philadelphia is a famously sprawling city, surrounded by flat country, and its river is narrow and easily bridged for further dispersion
Razib: thanks–and well done…this must have taken you awhile
this must have taken you awhile
i wrote some R code to process a lot of this, so not THAT long.
Patrick hits the nail on the head.
Also: Who owns the homes, and how much do apartments affect things? What is a home?
1: What is a home? Apartments don’t count, but what about condos or ‘owned’ rooms in buildings.
2: How does the rate of ownership correlate? If the price changes are due to cutting off poorer people at different rates, is the data meaningful?
3: Who owns the homes? Do many large corporations buy condos in NYC as a perq for upper management who likely don’t ‘reside’ in NYC for tax purposes? Also, the UN is in NYC, and that means crazy embassy housing should push things up.
4: Self employed says a lot. People are likely charging the exhorbinate cost of housing to their company for tax reasons. In a place like Manhattan, I’d expect to see a chasm between real wealth and IRS-style ‘income’
Good info, but more important is what you get for the $.
In Fort Bend county, the $116864.8 buys a 2000 square foot 3-4 bedroom house in a nice neighborhood. $250,000 buys a 4000 square foot 5 bedroom 4 bath with an office, and media room, and a large upstairs kids playroom. Steve might call it affordable large family formation. Good school districts, lots of Asians. Even cut rate engineers/programmers from Asia working as contractors for chemical/energy companies and oil field services companies can afford a nice 3000 square foot house in Fort Bend. It is a lot easier to have another kid when you don’t have to pay for private schools to keep your kids safe and you have four or five bedrooms. Fort Bend is also very large and the western half is rural and small town which brings down the median price.
Probably every county on the list has some unique features.
New York City is, I think, unique among American cities. Most other major cities have thrown off satellite office parks in their suburbs, in which many of the best-paying jobs are available. This greatly expands the total area in which residences within a reasonable commute of well-paying jobs can be built,
But in New York City, almost all the best-paying jobs, along with culture and city amenities, are still concentrated in a very small area, lower Manhattan. This restricts the area within which people who work at these jobs can live and still have a reasonable commute.
Blame it on the last ice age. When the glaciers came down they cut such deep rivers on either side of Manhattan, the East River and the Hudson, that these streams could not be bridged until the age of steel, the latter 19th Century. By that time, New York City was nearly 300 years old. It had already built up a critical mass of wealth, finance, and other delights that no matter how intolerable the city sometimes became, the really ambitious never wanted to desert its best districts.
Fort Bend county has the highest proportion of Asians of any non-coastal-state county in the country.
Note that Riverside County annd San Bernadino, CA , are shaded slightly red. There is plenty of room in California, all the way to Death Valley.
You might regress the differences (+/- from trend) against population density, or changes in population or rate of economic growth.
Combined_090307.pdf
Related item: Robert Gordon has a very good paper on increases in income inequality (not nearly as much as thought, when properly calculated) at http://faculty-web.at.northwestern.edu/economics/gordon/SFChallenge_
You might regress the differences (+/- from trend) against population density, or changes in population or rate of economic growth.
yeah, good idea. i’m assembling a data frame day by day, so i appreciate the comments, though i might not get to doing the regression in the short-term.
Grumble. As usual in all such maps, Hawai’i is left out. Are we chopped liver or what?
As a SWAG, I’d guess we’d be pretty blue. Either that or I have a really bad perspective on incomes out here (not entirely unlikely, since I’m currently job-hunting and unemployed).
hawaii & alaska are not part of the output for the “map” function i’m using in R.
I’m pretty sure the big reason for why Manhattan is such an outlier is that a much smaller portion of people who live in Manhattan own their homes/apartments/whatever as opposed to renting them. That is, even if you make double the median income in NY, you probably will never be able to afford to buy your own place.
Another historical factor is that a lot of Manhattan neighborhoods used to be pretty crappy with very low prices. So, people bought units in the East Village and Chelsea and the Lower East Side in the 80s for prices you couldn’t find now in small towns in Kansas. But, when crime fell and the neighborhoods cleaned up, people found themselves with units they would never be able to afford now. As these people move out, though, the only people who will be able to afford to buy their units from them will be investment banker and lawyer types, along with a few trust funders and foreigners.
One thing that’s going on is that the market has priced in declining demand in cold-weather central states and increasing demand in Southern states. Minnesota is very red because it has a very productive economy right now, but a lot of locals are looking to cash in their chips and move somewhere warmer, so the price of homes is low.
Median income isn’t directly comparable with median home price since many income earners aren’t home owners. It would be interesting to composite this with price-to-rent ratios and see whether the same patterns hold.
How come no one’s commented on just how relatively unaffordable some of those chunks of the Deep South appear to be? Take a look at some of those counties in Mississippi and Alabama tinging blue. That speaks to some pretty deep poverty when housing prices are relatively unaffordable in those locations. I guess the anti-union crowds have done their work well.
Linkmeister: the numbers for Hawai’i (well, Honolulu, a county after all) today are:
10.9 vs 13.26 ($56311 (wikipedia) vs $575000 (my memory of recent honolulu real-estate pubs [i'm in the process of house hunting])
you’ll note if you put that on the scatter plot that it falls a good 2.4 logs above the trend, in line with New York, if one follows the slope of the rest of the population. the model median price would be 133k, which is only about 2x above the *pre*-bubble values in 1999. we are basically quite screwed in hawaii, as I’m sure you know.
but, my impression is that this analysis is leaving out expensive counties, or at least areas, in the west coast — Honolulu was for a while only the 3rd or 4th most expensive city, but I see almost nothing about the 13.0 level (440k).
aloha