Earnings, property taxes, house prices and foreclosures in New York state

While Googling around, I stumbled across the American version of the Taxpayers' Alliance, who call themselves the Tax Foundation. They've put together some detailed data on the above topic, which they published under the heading New Census Data Shows Where Property Taxes Hit Homeowners Hardest. Inevitably, they seem to quite like sales taxes.

The figures for median property tax bills, house prices and incomes by state are here and by county are here.

Their use of the word "hits" suggests that property taxes make homeowners (taking current and future homeowners as a group; transfers between the two net off to nothing) worse off. The point is that actually they don't. Another useful resource is Realtytrac, which gives us county-by-county figures for the number of foreclosures.

Taking New York state, the figures for the thirty-six counties for which data is available give us the following charts. I've given the coefficient of correlation in parentheses at the top of each chart:

1. Housing is a normal good. Unsurprisingly, house prices in areas with high wages is higher. That's mainly because of Ricardo's Law (high wages push up house prices) and also because high earners have more money to spend, so they can buy nicer houses or houses in 'nicer' areas:

2. Ricardo's law is the main factor. Higher earners have more money to spend, so you'd expect them to spend their extra net income on all sorts of stuff - cars, schools, holidays, housing. You wouldn't expect them to spend their entire extra income on higher housing costs.

But median households do. If you knock off 40% for income tax etc from median earnings and then deduct the cost of renting or buying ([median house price x 6%] + property tax) from that, median household net income for recent arrivals/purchasers is completely flat. That net income of about $24,000 is what median households need to spend on the 'basic minimum' and its only the surplus which goes into rent.

Which is what you'd expect: a median person can't make himself wealthier by moving to an area where he can expect higher earnings for doing the same job; he has to pay over that extra as an 'entry fee' to the landlord/vendor:

3. We can work that backwards. If you take the post-tax median income for each county and knock off the 'basic minimum' of $24,000, that gives you the amount which the median household is prepared to spend on rent/mortgage + property tax. So if you then deduct the property tax and multiply the result by 17 (i.e. discount it backwards at 6%), you get an expected median house price, which is a very good guide to actual median house prices:


4. So we know that property taxes push down house prices; for every $1 tax, the house price is $17 lower than it otherwise would be. Homeowners aren't "hit" by this tax at all; the seller's loss is the buyer's gain.

Further, it is quite clear that higher property taxes lead to more rational decision making by potential home buyers;
- they are less likely to be blinded by offers of interest- or repayment-free initial periods;
- land prices are less inflated so lower mortgages means that households are less affected by interest rate increases;
- a high property tax (median $9,044 in Nassau County) certainly focuses the attention.

As this money goes into county coffers, and government spending (good, bad or indifferent) is a large part of the economy (whether we like it or not), stable tax revenues mean that the local economy is less prone to economic downturns. If taxes are only raised on incomes, down swings become immediately self-reinforcing, but as households have clearly budgeted for the property tax, it's the next layer of spending after the 'basic minimum', revenues from such a tax are barely affected by a fall in 'above the line' incomes.

So we'd expect there to be a negative correlation between LVT levels and e.g. foreclosures. We can work out the implied Land Value Tax rate (property tax divided by total rental value minus $4,000 annual buildings maintenance costs) and then plot foreclosure rates against it.

There is a clear negative correlation between the two: the five counties with the highest foreclosure rates have very low implied LVT rates. Three of the four counties with implied LVT rates of 100% have negligible foreclosure rates. The coefficient of correlation is only -0.51 which is not particularly high, but better than nothing:

5. As we have seen from Chart 1, land rents are a kind of privately collected tax, being 100% of post-tax median income minus a "basic minimum" annual household spending of $24,000. In other words, relative to incomes, rents are highly progressive. As property taxes in most US states are proportional to home values, what this means is that property taxes are also inherently very progressive to incomes: in counties with median earnings $50,000 median property taxes are about ¢2,000 (4% of income); and in the three areas with median earnings of $100,000, property taxes are $9,000 (9% of income).

So that deals with the "Killer Argument Against LVT, Not" that higher earners would flee areas with high LVT. Simple fact is, they don't. Part of what they are prepared to pay for is being in a 'nice' area, i.e. an area with lots of other high earners - the higher property tax filters out lower earners.

We observe the same effect with private schools: although the children are no cleverer than at state schools and the teachers not better, it's only keen/pushy parents who send their children to private schools, so their children will work harder/behave better, leading to a virtuous circle; those schools then get a good reputation, so the next cohort of parents are prepared to pay extra to send their children there etc. So what the other parents are paying for is to send their children to private school with your children and vice versa: