Wednesday, January 04, 2023

A Problem with Georgism

Henry George and his modern followers propose an LVT, a land value tax, a tax on the site value of land equal to what the rent on the land would be in the absence of any improvements. From an economic standpoint, the chief attraction of the LVT is that, since it is taxing something in perfectly inelastic supply, taxing it does not lead to any inefficient economic decisions. The site value does not depend on any decisions made by its owner, so a tax on it does not distort his decisions, unlike a tax on income or produced goods.

I believe I have discovered an error in that argument, an inefficient incentive produced by a Georgist LVT. Suppose there are two adjacent plots of land, one suitable for apartment buildings, one for a shopping center. Further assume that, due to administrative diseconomies of scale, the two plots will be worth more with different owners, would be owned by different people if there were no LVT.

I buy and develop one, you buy and develop the other. My conveniently nearby shopping center raises the rent you can charge for your apartments hence the site value of the land under them, so your LVT goes up. Your nearby apartments raise the rent I can charge tenants of my shopping center, hence the site value of the land under it, so my LVT goes up.

If, however, I bought both plots and developed them the increase in income from both plots would be income from improvements — so the LVT would not go up. Hence it may be in my interest to buy both even though they would be more productive, tax aside, with separate owners. 

More generally, the LVT is an incentive for inefficient integration of land holdings, makes a company town more profitable than a bunch of separately owned plots even if, in the absence of an LVT, it would be less profitable.

28 comments:

Anonymous said...

I am no expert, but the Georgism LVT has always confused me. I thought it was based on the value of the bare land and intended to be so unrelated to improvements that it would be the same $/acre everywhere in a jurisdiction. Thus a federal LVT would have the same LVT per acre in Nevada's Great Basin as in Manhattan. There are 2.3 billion acres in the US, meaning the LVT for the feds would be $2000/acre. That is incredibly cheap for Manhattan and probably equally expensive for the Great Basin.

But NYC would have its own LVT. 300 sq miles is 200,000 acres, and its budget is $100B; that's $500,000/acre.

If I understand your article correctly, you say the mall land base value goes up because the apartments nearby make the land more value, and vice versa. Doesn't this assume one parcel is developed before the other? Why are not both parcels more valuable once both are developed? And if my understanding of your article is correct, then how can the LVT be based on bare value?

Yes, I have always been this confused on what the Georgism tax is :-)

David Friedman said...

The LVT is supposed to be on the value of that particular piece of land. Its value can be increased by things happening around it — land in a city is more valuable, even bare, than similar land in a wilderness. But that's unearned value so far as the owner of that plot of land is concerned since he didn't make the city grow up next to it — that was the result of things other land owners did with their land. On the other hand the increase in the value of farm land due to someone digging out the boulders is an improvement and doesn't get taxed away, since if it was there would be no incentive to dig out the boulders.

In my example, the LVT of my land is based on what it would be worth if bare — which is increased by the existence of apartment buildings next to it full of potential customers for the shopping center I am going to build. The rent I end up getting is the sum of the land rent for the bare land and the return on my investment in building a shopping center. The land rent is taxed away, the return on my investment I get to make it in my interest to make the investment.

Arqiduka said...

IIRC Rothbard also adds that a hard LVT would also negate any incentive to discover efficient uses for undeveloped plots, to which your critique may be a partial and unintended answer. Maybe the medicine is worse than the illness though.

Anonymous said...

If the "bare value" is dependent on other land nearby, then wouldn't the mall parcel and the apartment parcel end up with both having increased value once both are built up? If one is built first, its bare value doesn't increase, but the undeveloped one does, and that seems in line with the Georgism philosophy of encouraging development. But as soon as the second parcel is developed, now the first one's bare value increases too, and they are both raised, although not necessarily equally.

Anonymous said...

I think this is more of a problem of assessing the bare value of the land, rather than the abstract concept of LVT. As the above post points out, building both the mall and apartment should each raise the value of the land under the other building, regardless of whether the owners are the same or different.

Maybe there should be a different proxy for the land value, rather than accounting statements. Perhaps the height of the buildings on the land. If you're going to build taller buildings, it is probably because the land under them is valuable enough to go to the extra trouble. So, consider a tax assessment based on the height of the buildings near yours. The height of your building is excluded, to avoid disincentives. If there are lots of open fields nearby, the land value is probably pretty low. If there are skyscrapers, the land value is high.

I guess you'd need a different system for agriculture/mines.

I suppose there could be some incentives to waste in this assessment method as well. Maybe instead of building a factory/office park on the edge of town, I build it further away in order to avoid the tax.

Anonymous said...

Well I'll be! I looked again at the Wikipedia article on Georgism, followed a link to Land Value Tax, and there it was:

"The land's occupants benefit from improvements surrounding a site."

That sure seems bonkers to me. Someone adds a lane to a freeway and every parcel within 50 miles has to be re-appraised. A franchise chain goes bankrupt, every parcel "nearby" has to be re-appraised. A new popular restaurant replaces the bankrupt chain store, re-appraise everything again. True, you'd only have to do this once every 5-10 years, but still, expecting uninterested burrocrats to properly handle these kind of subject appraisals is a recipe for corruption.

I once came up with a "voluntary" property tax. Everyone self-assesses their property. To prevent everyone low-balling these values, you can't sell or claim insurance for any higher value. To enforce honesty, any buyer who could prove he paid more would be refunded the excess from the seller, and the seller would pay a matching fine. The full sale price would remain as the self-assessed value.

The main appeal to me was eliminating any need for intrusive burrocracies needing to know too much or auditing every seller and buyer.

Ricardo Cruz said...

Anonymous, to avoid manual re-appraisals, some people have proposed interesting schemes using machine learning. I believe these series of articles by Scott Alexander covers that. In any case, the conundrum mentioned by Prof Friedman is really interesting.

SB said...

When I first heard about Georgism, the pitchman (addressing an environmentalist audience) claimed that land in the center of a city was so valuable that it could be used in only two ways: developed intensely to generate a lot of revenue, or preserved by a government as a park (because the government doesn't pay taxes). The latter didn't make sense to me, since the government would be forgoing a lot of taxes by preserving it as a park.

But yes, it was always intended that land's value, and therefore tax, is affected by what's around it. Which is true under "ordinary" non-Georgian tax systems too; Anonymous's point that "every parcel nearby has to be re-appraised" is already true in non-Georgian tax systems.

The difference is that a parcel's value, and therefore tax, is not affected by development on that parcel. Since "parcel" isn't defined a priori, it seems that one could cheat on the system by consolidating parcels before developing them, which is basically what David described. There might even be a perverse incentive to not develop a parcel until you can acquire all the land whose value it would affect.

Anonymous said...

Agreed that the core issue here seems to be in how the value gets appraised - in practice whoever assesses the tax has to figure out some protocol for appraisal. Perhaps this generates a class of LVT "cheats" due to gaming the appraisal system.

David describes a scenario in which proximity to another development raises the bare value, but you could imagine the opposite as well. Perhaps I arrange through a shell company to build a landfill next door to my apartment complex, if I know ahead of time that the assessment protocol will result in a lower LVT for the apartments. If I can determine a way to construct said landfill (perhaps so that it's sequestered or walled off enough) that I'm confident it won't actually negatively impact the rents I can get, it may be in my best interest to build a useless landfill just to lower my LVT.

David Friedman said...

@Anonymous:
“building both the mall and apartment should each raise the value of the land under the other building, regardless of whether the owners are the same or different.”

We want people to have an incentive to do things that raise the value of their land. The tax is for increases in the value of my land that someone else is responsible for. If I develop both plots the increase in the value of each due to the other is an increase in value that I produced, like the increase in value of land from digging out the boulders.

Different Georgists have different ideas about how to measure site value, but they agree on what you are trying to measure. My point is that there is an inefficiency even if you correctly measure what they want to measure.

The self-assessed property tax is an old idea. The standard version is that, having stated a value for your property, you are obliged to sell at that price to anyone willing to pay it. A real world example of the principle is a claiming race, a horse race where the entry requirement is being willing to sell your horse at the fixed price for that race. It’s a way of making sure all the horses are about equally good. You could enter a much better horse and win the race — but you would lose the horse when someone bought it at your understated price.

David Friedman said...

I should probably mention that before putting up this post I sent my argument to a friend who is the smartest Georgist I know. His response was that he was aware of the problem but most Georgists were not.

Unknown said...

@David Friedman (I'm the anonymous poster who suggested building height for assessment.)

A way of looking at the Georgist tax is that improving land has positive externalities to nearby landowners, so the government might as well be the institution to claim that income. From this point of view, your apartment complex and shopping mall are joining together to internalise an externality, and this is causing losses due to inefficient management. If the government weren't claiming the externality as a tax payment, this incentive for inefficiency would still remain. We could be talking about a lakeside resort and a fertiliser factory that have to be under the same management, because the resort will have difficulty monitoring and enforcing pollution agreements on a separate company.

This inefficiency only looks like a deadweight loss due to taxation because of two peculiar features of your example. First, the claimant of the externality is the government, through a tax, and, second, the externality is locked in once the buildings are constructed. The tax system is what prevents the buildings from being sold to different owners once they are built.

If we really wanted to incentivise people to do things that raise the value of the land, then my land value taxes should be paid, not to the government, but to my neighbours based on how their property increases the value of mine. Alternatively, we could empower the government to negotiate with developers to build closer to the efficient level. A developer might only want to build 9 story apartments, but, when considering the externality, the efficient outcome would be 10 stories. So, the government, which is now the sole entity collecting this externality, bargains with the developer to pay some of the cost of the 10th story. In my opinion, this would be a public choice nightmare in practice.

Anonymous said...

@David -- I know about the claiming idea, and don't like the incentives. It would be useful for eminent domain, if I believed in eminent domain!

Enforcing the self-assessed value by limiting court claims is enough, I think, without trying to tie it to selling prices and rewarding snitches. Almost all moveable property, like vehicles, art, jewelry, will be fairly self-assessed so criminal charges are for realistic values. Buildings would generally be for just the structure itself, since there is no need to insure the land's market value unless you are polluting it or it is eroding coast land. Anyone who tries to insure property for over its self-assessed value will run afoul of insurance companies which know you can't sue them for not paying that excess value. Or so it seems to me.

The appraisals seem the biggest Georgist problem to me. Current appraisals look at comparable properties and is only a problem for custom houses with nothing similar nearby. Georgist appraisals can't be that simplified, they have to factor in way too many subjective values, and seem to be begging for corruption.

But I am no expert, have never thought these matters through very deeply, and read your blog and others to learn.

Arqiduka said...

RE the challenge of appraisal, I think this is overestimated. Any sizable sales database could allow the application of a simple multiple regression model predicting price as a sum of improvement and unimproved value. The former would be informed by type of property (residential, commercial etc), square footage and a few dozen other variables, with the latter informed by linear distance from center, closest transport hub, ZIP, neighbourhood demographics and a few others.

I'm sure real estate agencies have such models now, and think nothing of the issues involved.

It doesn't need to be perfect, just needs to account for a large enough % of price variation.

I personally disagree with the wisdom of taxing unimproved value only, but if you want to do that appraisal wouldn't be the bottleneck.

Unknown said...

I think that the problem raised here is largely in administration. At an extreme, you could remove this distortion entirely by charging a national flat fee per Acre of land (that this would lead to some land possibly having a negative price doesn't matter too much to the theory). More reasonably, region based LVT would also eliminate the issue.

Nuño Sempere said...

Wohoho, this reminds me of Shapley values (). Something along the lines that: So LVT is approximating the counterfactual value of development as a difference between developed in under-developed land. But this runs into predictable problems arising from use of counterfactuals (). You could solve this using Shapley value of development instead. But you don't have access to the Shapley values, because they require multiple counterfactual worlds to calculate, to which you don't have access.

Nuño Sempere said...

Links in the above comment:
- https://en.wikipedia.org/wiki/Shapley_value
- https://forum.effectivealtruism.org/posts/XHZJ9i7QBtAJZ6byW/shapley-values-better-than-counterfactuals

Nuño said...

I believe that the analogies with the Shapley value <> counterfactual value problem is valid, and that one should be able to translate the examples in the article I linked above to the Georgism case.

Some other thoughts:
- You can't incentivize people with SVs because then otherwise they have an incentive to not actually do the work
- Fun case: Imagine if someone invents and patents a technology that increases the value of the land (e.g., barbed wire, nitrogenated fertilizers, etc.). How does this change the LVT? What happens if the invention is kept secret? What happens if the biggest landowner buys the exclusive rights to the invention?

Sam Harsimony said...

Paul Birch raised a similar objection in his thoughtful critique of Georgism here:

https://web.archive.org/web/20120119025941/http://www.paulbirch.net/CritiqueOfGeorgism.html

Finn Hambly said...

Pasting my responses that I left on Nuño's tweets about this.

I think this problem assumes that land gets segmented according to the land’s ownership, rather than in a more principled/consistent/ownership-agnostic way (which would both be better and more tractable).

If the two bits of land are evaluated as separate in both scenarios, I’m struggling to intuit how a lower LVT makes up for the loss in productivity of the apartment building (A) and shopping centre (B).

If the inefficient monopolist bought both plots, improvement A might be (counterfactually) 10% less productive, resulting in land B being less valuable (=> lower LVT).

But if you assume land is ~50% of site B’s total value, and your LVT is set to 80%, then you’ve got to see a (counterfactual) drop of at least 40% in land value for this inefficiency to be saving you any money. And if your land value has dropped by 40%, that almost certainly means your income is much lower too! I don’t see how this strategy makes sense (or if this problem would ever happen).

(Please fix my reasoning if I've got something wrong here)

Anonymous said...

From a different point of view ....

I gather that Georgism justifies the LVT partly because land is limited, unlike manufactured products.

How do skyscrapers affect this? 100 story apartment buildings in effect add land. Even multi-story malls multiply the amount of retail space available. Multi-story parking lots, same story. Few homes are more than a couple of stories, and multi-story residential housing was common when Georgism was developed.

gwern said...

Has this been a problem in any of the Georgist communities or any jurisdictions which heavily rely on land taxes or Georgist-like value capture mechanisms to date (eg Arden or Singapore or Hong Kong or Japanese train stations), or is this argument purely theoretical?

T. M. Adams said...

I think you have a flaw with respect to assessment/valuation here. Land value is solely a function of demand, not the thing you did with it.

If you build a skyscraper in the middle of the field in Nebraska, with absolutely zero infrastructure or other amenities, the surrounding land would not increase in value and might well decrease.

The reason why your land value goes up due to other development is because it creases the demand for that location by providing new goods and services. But the critical part is that demand went up.

Now, with that in mind, it doesn't matter if your land is adjacent or in combined into a single lot or is separate lots. The demand goes up due to development somewhere, then your LVT goes up.

It is true that, for example, if you build a business in your town that it will increase the value of your home. But in practice we're talking about at minimum dozens of economic actors, and usually more like millions, and the contribution to your own land values is pretty marginal.

Given that with an LVT the holding cost is pretty high for idle land, and the fact that any one person only has so much time and mental energy, I'm not sure that they'd sit around accumulating a bunch of land so that they can attempt to internalize all the value of their own improvements. Even though I demonstrated that's pretty unlikely anyway, since LVT increases as a function of demand for the lot, not as a function of the nearby capital investment.

Now, there is the relatively unique case of something like Disney World, which was the effort of a single economic actor building a whole community from scratch in otherwise marginal land. But one, we don't have many examples of that, and secondly, this sort of arrangement sees DW acting as its own sort of government, so internalizing the value of all of its improvements makes sense. Its a rent-funded community in much the same way Arden is.


Mikewind Dale (Michael Makovi) said...

I think that the LTV assumes that the land is taxed based on the most valuable use the land can possibly be used for.

If the land is most valuably used by 2 separate owners rather than 1 integrated owner, then the LTV is assessed on the value of the land assuming 2 separate owners, even if in reality, it's owned by 1 owner.

Similarly, the construction of a new business next door to your own wouldn't affect your LTV because the new business was already the optimal use of the land, so it was already included in your LTV before it was even built or established.

In other words, the value of the land is not something determined by what is actually being done with it, but what could be done with it.

The obvious problem, of course, is that this assumes omniscience. It's the problem of economic calculation all over again. No central planner can know this. Only a market with entrepreneurship and market prices can competitively discover the best uses of things - and even then, it's only a gradual tendency.

And a less obvious problem is this: what if the most valuable use of the land requires various complements that don't exist yet? Are people to be taxed based on the value their land will someday when, when currently nonexistent complements have come into existence? Cf. Mises's argument that even if central planners could calculate the Walrasian equilibrium, that wouldn't tell us how to get there. Even if someone in the year 1400 or 1500 had known that Manhattan island was made for New York City, that wouldn't have helped them build the city and populate it.

Justin Merritt said...

As long as the value of land changes with improvements, it is in effect an income tax.

David Friedman said...

@Justin:
The tax is supposed to be based on the site value of the land, its value net of any improvements.

Anonymous said...

As others have said this isn't a problem with LVT, merely a potential for an error in how land value is assessed, but in theory ownership should be a non-factor and in both cases the value of land is increased by the relative proximity of the new improvements.

Let's expand the example for a way this could be assessed in the trivial case. Imagine there is a 3rd adjacent plot on the other side of the shopping centre. This plot is zoned for residential but is currently unimproved land.

We can easily see the value of this unimproved land and conclude that the land under the apartment buildings is similarly valued and likely has increased in value from the presence of the new shopping centre. In non-trivial scenarios we can look at the change in prices for improved land as well and estimate the changes in land value through some index of land values in the nearby area.

It's probably worth pointing out that even improvements to a single house could slightly increase the land value under that house (by making the area marginally more desirable). So you don't even need two plots to create a scenario where your improvements do result in a tax increase, indirectly.

Prakash said...

Hi David

Isn't the real question to compare the dead weight of other taxation modalities or is your argument a purely "first best" comparison i.e. comparison with an ideal?

The way my thinking around this is, you're correct in the most basic sense. But once you step into the real world, the real constraint faced by entrepreneurs, you're probably not going to get the same kind of entrepreneurs in both cases.