Response to Edwin Mills
on Land Value Taxation

Mason Gaffney

[Reprinted from a Land-Theory online discussion, 29 April 2002]

1. p.1, Mills asserts with certitude that land rents are 5-7% of GDP (sic) today, and have been the same for 120 years.

  • A. He offers no support (but see #2).
  • B. He later makes as his major point that no one can know what land rents really are. How, then, does he know? He apparently does not see the contradiction.
  • C. He does not comment on studies by others showing rents to be much higher.
  • D. The relevant datum would be NET Domestic Product, not Gross.
  • E. He later writes, p.5, that land value is 15% of real estate value (again with no support). With a little figuring, if land rent is 7% of GDP, and therefore, say, 9% of NDP, and land rent is 15% of real estate income (as he implies), then real estate income would be over half of NDP. He does not seek to reconcile that with the general showing of data sources that it is much less.
  • F. He does not define "real estate," so we don't know if he includes corporate-owned industrial and mining lands, eleemosynary holdings, or what.

2. He cites unpublished "technical study" held by the publisher, and "available on request"; but gives no specific references. It later seems this "technical" study is really a theoretical one, based on some indefensible assumptions, discussed below.

3. He misquotes or misreads H. George on population, saying George wrote that rising population would hold down wage rates. George wrote the opposite, most explicitly. George wrote that the misuse and underuse of good land, forcing recourse to worse land, and overpricing good land, is what holds down wages.

This lets Mills say, p.2, that the ONLY (sic) reason for switching to LVT is to raise the capital-intensity of land use. George's reason was to make jobs, and raise the marginal productivity (hence wage rates) of labor.

4. Mills issues several dubious statements ex cathedra, on his own authority.

  • A. P.1, real labor earnings have risen greatly since 1900.
  • B. The causes of the rise are limitations on population growth (!), more human capital, and a larger stock of capital.
  • C. No one could understand the above without neoclassical economics (!)

As to "A", one should consider:

  • i. How much of the rise was due to the practical implementation of Georgist taxation - heavy reliance on the property tax, a tendency to overassess the land element, and an income tax, before 1941, that spared wages and focused on land income?
  • ii. It is widely believed that real wage rates peaked in the 1980s, even before taxes.
  • iii. It is doubtful if the CPI adequately reflects the higher cost of housing.
  • iv. Taxes are increasing focused on wages and salaries.
  • v. If higher wages are due to more human capital, the process of absorbing human capital - i.e. deferred entry into the labor force - should be debited from nominal wages.

5. p.1, col. 3, Mills says he will focus on whether a land rent tax is less distorting than "other taxes for which it might be substituted." He departs immediately from that, however, and discusses only the property tax on buildings - ignoring sales, payroll, excise, and income taxes. In fact, California after 1978 underwent a massive substitution of sales and income taxes for land and building taxes, but mostly for land taxes. That substitution could be reversed, and without incurring any of the problems that Mills poses about separating land from building values.

6. Pp. 2-3, Mills gives a brief glimpse of the theoretical model of his "technical analysis." It includes the assumption that "halving land while doubling labor" would "leave output unchanged." This is alleged to tell us "what the numbers show." Apparently these numbers are hypothetical, spun off from this model.

I would have no faith in a model with such a bizarre content, nor would I blame it on Paul Douglas who, by the way, supported LVT strongly when he, as a U.S. Senator, headed a housing commission under President LBJ.

Inconsistently, on p.5, Mills touts the "hedonic" approach to valuation, where the effects of improvements are "non-linear."

7. P.4, Mills alleges that a tax on buildings, by suppressing building, lowers a city's demand for land overall. Here he forgets entirely about how the tax induces substitution of land for capital, increasing land requirements. He never mentions timing of replacement and the tax-inducement of derelict buildings, and their negative neighborhood effects, that force demand outwards. Not a word about urban sprawl.

Land values are marked by continuity in space. To the extent the market is free to determine where new buildings go up (free of building taxes), the city will be compact, as builders add strength to strength. The market, unbiased by distorting taxes, combats urban sprawl and lowers the aggregate demand for land. Many critics of LVT have faulted it for promoting centralization, by letting the market work without bias. Now comes Mills with the opposite complaint - at least, at this point. But read on.

In his last column on p.5, however, Mills avows that to exempt buildings will lead to more centralization and higher ratios of buildings/land - i.e. to lower a city's overall demand for land. Here he is making a different attack on LVT -- too much street congestion -- and he reverses his analysis to suit the attack of the moment. This kind of self-contradiction is vexing to the reader, and makes one wonder how objective this work is.

8. Mills avers it is impossible to put a value on land, if it is improved.

  • A. P.3, "most taxable metro land is already developed." A few aerial photographs dispel that notion. Parking lots, e.g., are ubiquitous in CBDs. (Wonder what he means by "metropolitan land"?)
  • B. P.3, He rejects sales before tear-downs as being land sales ("they simply do not provide ANY basis for assessing land values"), for what look like captious reasons.
  • C. He rejects evidence gleaned by city purchases and sales of land. He raises the irrelevant bogey of the failed "Urban Renewal" program of the LBJ years, but that was not designed to garner data for assessment purposes, but rather to placate a few big city mayors (like Presidential hopeful Henry Maier) who wanted to evict poor blacks. At the same time, the suburb of Whitefish Bay, WI, had its own homegrown program of buying old houses and reselling land at the market, quietly amassing enough data to value the entire suburb.

9. P.3, it is much harder to value land alone than land with structures - he says, from his own authority without any support or rationale.

10. P.3, he alleges that Vancouver uses only "rules of thumb" to value land separately from buildings.

  • A. His source for this is an article on Pittsburgh, by Oates and Schwab. I am not aware that they studied Vancouver. As for Pittsburgh, it is true that its assessments were very bad, but there is no evidence to blame Vancouver for Pittsburgh's faults.
  • B. I have assessment data from Vancouver and environs ("The Lower Mainland"), by neighborhoods, showing that the fraction of residential real estate value that is land value varies from a low of 40% in some less desirable suburbs up to over 80% in the Point Grey and University Endowment Land districts in the southwest city. What rule of thumb does Mills allege that that follows? He does not say.
  • C. The officer who was in charge of assessments for the whole Province, Ted Gwartney, is available for Mills to consult (he now lives in Bridgeport, CT).

11. He overlooks the many cities around the world that base taxes on land without improvements; and the healthy condition of those cities. Among them are Sydney, Brisbane, Adelaide, and perhaps half the municipalities of Australia and New Zealand; Hong Kong; Taipeh; Johannesburg, and dozens of other cities in South Africa; Copenhagen; etc.

12. On p.4, col. 1, unless the taxing body can estimate both land and improvements both before and after the tax shift to LVT, "no improvement in resource allocation can result."

That is a pretty violent non-sequitur. Wm. Vickrey, an enthusiastic advocate of LVT, often remarked that its incentive effects do not depend at all on the accuracy of valuations; they only depend on the tax remaining unchanged when improvements are made. Perhaps he overstated the matter to make a point and a little stir, but he was a Nobel Laureate, and worth heeding.

13. P.5, Direct controls over land use have replaced the market, so there is no point in improving the market, anyway, says Mills. There are many reasons to improve the market, but the short answer to that foolish comment is, "Well, Prof. Mills, since it doesn't matter to you, let's do it my way."

14. p.5, Untaxing buildings would increase jobs in the center, and call for new infrastructure.

  • A. It surely would make mass transit more viable, cutting transportation costs.
  • B. It would abate urban sprawl, which so inflates infrastructure costs.
  • C. It would attract residences as well as jobs to the center, and reduce most trip lengths, thus lowering overall linkage costs.

15. On p.1, Mills misuses John Whitakers's article on Walker and George.

  • A. Mills has Walker and George in an "emotional" battle, like "romantic rivals," "that blocks intelligent exchange among thinkers." On this he cites Whitaker's 1997 article. That is not Whitaker's emphasis at all; he has them both contributing to forwarding intelligent thought in economics.
  • B. Perhaps Mills has in mind the semi-popular debate between Walker and George on the U.S. Census' unwarranted conclusion that farm sizes were becoming more equal, because the mean size had fallen. This was spirited on both sides. However, Whitaker omits it from his article (an omission I do not understand).
  • C. As for forwarding intelligent thought, George in this debate advised Walker he needed something like the Lorenz Curve (a term not then yet invented). Walker's views had been quite primitive. In 1900, however, the Census of Agric. adopted George's format.
  • D. As for "emotion," Walker was clearly carried away with contempt, led with his chin, and made a fool of himself. Later, in his text, he wrote of the land tax that "I will not insult my readers by discussing a proposal so steeped in infamy." Now, THAT is emotion! George seemed relatively calm.

16. On p.1, Mills crowns Walker as "the greatest 19th Century American economist." Not any modest "in my opinion," or "perhaps," but simply "the greatest," on Mills' authority, ex cathedra.

  • A. Perhaps Mills is not aware of Susan Carter and Richard Sutch, 1995. "Fixing the Facts: Editing of the 1880 U.S. Census of Occupations ... " NBER Historical Paper 74. Carter and Sutch conclude that Walker (then-Director of the U.S. Census) doctored 1880 Census returns to understate the incidence of child labor, "which was far greater in 1880 ... than has been previously thought." (pp.28-29). Why the doctoring? Their main burden is to show that doctoring occurred, but "Perhaps he was afraid of igniting social protest."
  • B. Perhaps Mills has not read the journal debate between George and Walker (reprinted as an Appendix to most editions of George's Social Problems). Walker's bumbling with means and measures of dispersion is not in the manner of "the greatest." His motive, aside from sheer negligence and overcommitment, may well have been the same as what Carter and Sutch suggest above.

In summary, this is a most unsatisfactory paper, and does not begin to meet its burden of showing that land taxation is not practical.