There's actually a lot of similarity between this auction-based valuation and assessment-based valuation. An assessment-based system is always disciplined to sales, which *are* auctions when the seller is trying to get the fair market value for their land. The difference is that an assessment system doesn't try as hard to get the most relevant information — it just accepts whatever sale data is available from the neighborhood, and interpolates across time and space. For a plot of land that hasn't been sold in a long time you have to look at nearby plots of land that have sold more recently to inform the valuation. But by encouraging regular auctions with the compounding increase in the tax, this interpolation is unnecessary, and the assessed value can simply be based on the most recent valuation for that exact plot of land. It's easier to understand (no complex assessment models), requires less labor (no model maintenance or dispute resolution), and handles edge cases perfectly (valuation is based on the exact plot of land and its unique location and features). Very compelling!
Nit: It doesn't seem necessary to require all transfers of ownership to go through the auction; only that revaluation of the tax happens only with a public auction. If someone wants to transfer land to a family member without revaluation, they'd easily sidestep the auction-transfer requirement by keeping themselves as the owner on paper. It's the compounding increase in the tax that encourages regular valuations.
Sam, Thank you for this healthy mental exercise on a Thursday morning. I didn't follow the tax part: when do people start paying this land value tax? Only when the new owner takes possession of the land, at the price she bid? If no land is sold, how is it to be valued?
If you haven't seen it, take a look at 'Radical Markets' by Posner and Weyl who propose a similar land auction model, starting with a self assessment for tax purposes - with the proviso that anyone can buy your land once you have declared your value... set the value too low, and risk losing your home!
There's a lot to think through with this proposal, I'm glad you're enjoying it!
Yes, the tax is only paid by the winning bidder. The amount they pay annually is some small fraction of their winning bid (say 5%). As to when the tax is actually collected, I guess I didn't specify whether it should be immediately or if it should be collected a year later. I propose we just follow whatever a local jurisdiction does for property taxes.
By "no land is sold" I assume you mean that nobody places a bid in the public auction for a particular piece of land? I realize I didn't specify that either. In that case, I think the owner should revoke their ownership and the land is publicly available for anyone who wants to claim ownership. I could see either letting someone claim the land for free or holding an auction once somebody wants to claim it. Does that answer your question?
On the book Radical Markets, yes the Harberger tax that they propose has some similarities to this idea (something I think I should highlight in the next post)! Basically, if you were to set the tax growth rate inordinately high, people would reassess their land value annually in a re-buy auction. That looks similar to the self-assessment in Harberger taxes (though without the risk of someone buying your home)
Wow, this is amazing! Having the seller decide what to count as land and what to count as improvements is super cool. Love that it automatically handles any uniqueness about any particular plot of land.
Buying the land but not the property takes some head wrapping. I wonder if there are edge cases where properties do end up having to be taken down by the seller and those edge cases create a sentiment of unfairness. Maybe the Nash equil holds up.
Btw what’s the issue with using total sale values and doing interpolation and extrapolation (or even neural networks) to estimate values of land? Maybe that ends up being gamed over time
Yeah I think there will be cases where the buyer already knows they want to tear down the house and makes the seller do it. The seller still gets compensated most of the price of the teardown in the bid, but I think this situation could create resentment.
Extrapolating from total sale values would probably work pretty well, but I would put that closer to the "land value assessment" end of the spectrum (though requiring public sales is new). Presumably you could use machine learning, assessors, or prediction markets to help sort out what fraction of the total sale price is land value.
The first question is about the theory. My understanding is that Georgian theory is based on Richardo’s theory of land rent, which was developed for an agrarian setting. So I get that some land has an intrinsic value because it can grow corn. In urban settings it’s supposed to be about location, so that one is paying for access to subways, shopping, etc. So is the basis for a SFH land valuation about views or local amenities? It seems it would be difficult to determine next most efficient use, but aesthetic or convenience preferences are quite subjective. Maybe a measurement issue, but more importantly, what IS land value such that it could be identified and taxed?
Second: If SFH property values increase, rents and amenities stay the same, that increase is due to a subjective expectation of rising house values, such that gains will be made in the resale. How are such speculative gains related to land value? (Where I live, this is the real driver of unaffordable housing.) These seem like different beasts.
It's hard to say what the true value of a piece of land is, particularly because of how subjective it is. However, with Georgism we only really want to tax "unearned" value, not subjective value.
I think of "unearned value" as the rent an owner could collect while doing absolutely nothing of value. In other words, the rent they could collect by simply auctioning off the rights to use an empty plot of land (for example, a family might rent the land and build a SFH on it and continue to pay rent to use the land). This unearned value doesn't come from the landlords efforts so can be safely taxed without creating a distortion (in theory). Though nothing requires us to tax this unearned rent, it's a convenient way to fund public goods.
Even though this unearned value definition is more precise, estimating land value is still quite hard!
Second question: property values are kinda sorta the sum of pure land value and pure house value (other amendments matter too, there's also come synergies here). When overall property values rise, that is typically attributed to rising land value. It would be strange if houses themselves suddenly got more valuable across an entire city, especially since buildings typically depreciate in value over time. This growth in land value is another sort of unearned value (since it's unlikely that any one owner could raise land values across a city).
If you don't let the taxes grow along with the growing land value, then the owner gets to keep the unearned value (which can be acceptable). Though you have to be careful not to have taxes grow faster than land values.
Bad idea, which stems from the initial error of using selling prices as the basis of land value taxes.
The basic value of land is its gross annual value, which is its market rental value PLUS all property taxes currently payable. Selling prices are a derived value, being the capitalisation of that part of the gross value that is retained by the owner, plus an uncertain amount of hope value. It is an unstable value, dependent on, among other factors, interest rates, expectations of future interest rates, and the real estate cycle.
It lays LVT open to the valid objection that it is a tax on a value which cannot be realised. It also promotes general misunderstanding about the very nature of land economics.
This issue has been widely discussed, at least in Britain, where property taxes have traditionally and quite rightly been based on rental values.
Even though selling prices are derived from the thing we want to tax (annual ground rent), they bear a close relationship! Setting the tax rate appropriately should approximate taxing annual ground rent.
Its true that the landowner may not be able to realize the full value of the land, which is why it's important to set the tax rate to robustly under-tax land value (this is one of the topics in the follow-on post). Bidders should be rewarded for price discovery, and landowners should get some value from their land and local amenities.
The relationship between rent and price is far from close. It varies by several % about a mean of 5% (20 years' purchase). On top of that are factored in expectations of rental growth, development potential, inflation, and how the real estate cycle will develop. Worse still, the tax leads to a reduction in the selling price by am amount corresponding to the capitalisation of the tax payable. The LVT erodes its own base. It is like sitting on the branch of a tree and sawing oneself off.
Rental values are in most situations readily ascertained and are of their nature up to date There are well established and reliable methods of extracting the land value component of the rental values of developed land. British practice is, or was until recently, to value land and buildings separately, using rental figures.
The use of selling prices for LVT is conceptually flawed, stirs up justified opposition and sets the system up for failure.
The aim and purpose of LVT is to collect land rent for public revenue. Taxation of land rental values is transparent and people can understand what is happening.
The LVT movement could usefully give up the idea of taxation based on selling prices and promote a switch to rental value assessments where LVT is already applied. LVT based on selling prices is an important reason why it makes so little progress.
Note that land values will get reassessed regularly under this system, so values should adapt to changes in interest rates, rental growth etc. Overall, it looks like cap rates are relatively stable:
I don't expect these kinds of fluctuations to seriously change the efficiency of the proposal, so long as the tax rate is set appropriately. For example, setting the tax rate at 4% would ensure that owners get some surplus (and cap rates have stayed above this level for last 10 years). Though fluctuations can cut into the size of the surplus land rent that the owner receives.
> The LVT erodes its own base
True! This is part of the goal of land value taxes, to reduce the value of holding empty land. A good thread on this:
That thread also speaks favorably about land rent taxes.
I'm still agnostic about auctions, land value assessment, land rent assessment, and other forms of land valuation. I think they have different tradeoffs. What hasn't been covered here is how these valuation systems deal with other types of economic land, which I'll cover in the next post.
Capitalisation rates become volatile towards the end of the real estate cycle, eg in the early 1970s, late 1980s and the late 2000s. The chart in your reference diagram records just a part of the current cycle.
Values should be frequent. Annual review of rental values is feasible.
One aim of LVT is to reduce the price of land. If the tax is based on land prices, it destroys its own base. 4% tax on prices would make land worthless and that would put an end to LVT.
I be interested to see the existing literature on the elasticity of property and land taxes.
For example, Texas charges a 1.81% property tax. If land value is 70% of property value, then a 4% LVT would cost the same as a 2.8% property tax. Would moving to a 2.8% property tax destroy land values in Texas?
I would be surprised if it did, but I'm not that familiar with this area.
All taxes ultimately come out of land rental value. In conjunction with other taxes, the British business property tax has turned large tracts of the country into an economic desert.
The aim should be to collect 90% of gross rental value. To achieve this, rental values must be measured regularly and reasonably accurately, and collated in a database. Auctions based on selling prices would not achieve this.
LVT doesn't actually result in zero land values. The land value has to be nonzero if the tax is nonzero. The value becomes an indicator of what the rental value of the land is worth. E.g. take an LVT of 200%. I can only bid $5000 for a plot of land if I am willing to pay $10,000/year in taxes to have it. High LVT rates effectively become taxes on the land rent. https://peoplesrent.substack.com/p/investigating-land-value-tax-rates
Land price is the capitalisation of the annual rental value that is retained by the owner. Land price drops by about 20 times the annual land tax that has to be paid. If LVT assessments are based on land prices, then the tax cuts into its own base. This sets the system up for failure from the outset.
Yes, it is ultimately rents that should be captured from the land, and rents are a great way to inform tax assessment. You bring up good reasons why land value is itself harmful — it's speculative, and buyers assume a lot of risk in the case the land value decreases and they wind up underwater on their loan. One could argue that taxing land value out of existence is the point of the LVT. It's a pigouvian tax on the harmful effects of land value. But it is indeed weird to base the land tax on something that gets driven close to zero. Totally agree with you. Taxing rents themselves is much easier to think about! You might like this post about thinking with the Land Rent Tax instead of the Land Value Tax: https://peoplesrent.substack.com/p/the-land-rent-tax-and-real-time-georgism
This subject is discussed in several articles on the web site of the Land Value Tax Campaign, which is unfortunately now defunct. The web site will be transferred intact to another location with the same URL.
Appreciate the effort, but given that land and improvements are already separately valued in most states using long-established appraisal procedures, I think this a solution in search of a problem.
I probably am not understanding well enough -- maybe you covered this and I missed it?
1. If peeling the land off from the improvements means I have to publicly auction, then only be able to sell the improvements to the new land buyer, wouldn't it be better to sell them together? That way I can sell to whomever I want, without public auction.
2. If the only person I could sell the improvements to is the new land buyer, wouldn't the new land buyer have monopsony power such that they could get the improvements for almost free? (Or make me pay them, since I'm on the hook for removing the improvements if we can't reach an "agreement"?)
Good intuition, you're one step ahead of me! I'm going to look at this in more detail in the next post, but here's the gist:
Both the buyer and the seller have an incentive to keep things out of the land auction. For the buyer, this is because the dollars spent on a bid get taxed. The buyer responds to taxes by lowering their bid. Because the taxes lower their bid, the seller also wants to keep things out of the land auction, since they get less money than they otherwise would.
Moving to point #2, I should clarify that the original owner can sell the improvements to anyone, not just the auction winner. There's two key cases:
A. For items that are easy to take off the land (e.g. a potted plant or something) the original owner is better off selling the item on the market (or keeping it for themselves, or selling to the auction winnner at market price).
B. For items that are hard to take off the land (e.g. a house not designed to be moveable) the buyer does have a sort of monopsony power. Appendix 1 tries to model this and I find that the buyer can use that power to get the house for a little cheaper, but not for free!
There's actually a lot of similarity between this auction-based valuation and assessment-based valuation. An assessment-based system is always disciplined to sales, which *are* auctions when the seller is trying to get the fair market value for their land. The difference is that an assessment system doesn't try as hard to get the most relevant information — it just accepts whatever sale data is available from the neighborhood, and interpolates across time and space. For a plot of land that hasn't been sold in a long time you have to look at nearby plots of land that have sold more recently to inform the valuation. But by encouraging regular auctions with the compounding increase in the tax, this interpolation is unnecessary, and the assessed value can simply be based on the most recent valuation for that exact plot of land. It's easier to understand (no complex assessment models), requires less labor (no model maintenance or dispute resolution), and handles edge cases perfectly (valuation is based on the exact plot of land and its unique location and features). Very compelling!
Nit: It doesn't seem necessary to require all transfers of ownership to go through the auction; only that revaluation of the tax happens only with a public auction. If someone wants to transfer land to a family member without revaluation, they'd easily sidestep the auction-transfer requirement by keeping themselves as the owner on paper. It's the compounding increase in the tax that encourages regular valuations.
Sam, Thank you for this healthy mental exercise on a Thursday morning. I didn't follow the tax part: when do people start paying this land value tax? Only when the new owner takes possession of the land, at the price she bid? If no land is sold, how is it to be valued?
If you haven't seen it, take a look at 'Radical Markets' by Posner and Weyl who propose a similar land auction model, starting with a self assessment for tax purposes - with the proviso that anyone can buy your land once you have declared your value... set the value too low, and risk losing your home!
All the best
Andrew
There's a lot to think through with this proposal, I'm glad you're enjoying it!
Yes, the tax is only paid by the winning bidder. The amount they pay annually is some small fraction of their winning bid (say 5%). As to when the tax is actually collected, I guess I didn't specify whether it should be immediately or if it should be collected a year later. I propose we just follow whatever a local jurisdiction does for property taxes.
By "no land is sold" I assume you mean that nobody places a bid in the public auction for a particular piece of land? I realize I didn't specify that either. In that case, I think the owner should revoke their ownership and the land is publicly available for anyone who wants to claim ownership. I could see either letting someone claim the land for free or holding an auction once somebody wants to claim it. Does that answer your question?
On the book Radical Markets, yes the Harberger tax that they propose has some similarities to this idea (something I think I should highlight in the next post)! Basically, if you were to set the tax growth rate inordinately high, people would reassess their land value annually in a re-buy auction. That looks similar to the self-assessment in Harberger taxes (though without the risk of someone buying your home)
Wow, this is amazing! Having the seller decide what to count as land and what to count as improvements is super cool. Love that it automatically handles any uniqueness about any particular plot of land.
Fair play for thinking this through in detail!
Buying the land but not the property takes some head wrapping. I wonder if there are edge cases where properties do end up having to be taken down by the seller and those edge cases create a sentiment of unfairness. Maybe the Nash equil holds up.
Btw what’s the issue with using total sale values and doing interpolation and extrapolation (or even neural networks) to estimate values of land? Maybe that ends up being gamed over time
Yeah I think there will be cases where the buyer already knows they want to tear down the house and makes the seller do it. The seller still gets compensated most of the price of the teardown in the bid, but I think this situation could create resentment.
Extrapolating from total sale values would probably work pretty well, but I would put that closer to the "land value assessment" end of the spectrum (though requiring public sales is new). Presumably you could use machine learning, assessors, or prediction markets to help sort out what fraction of the total sale price is land value.
https://www.radicalxchange.org/concepts/plural-property/
Yes this system bears some similarities to Harberger taxation, perhaps I should touch on that in the next post.
I realized I forgot to give a shoutout to some other bid-based proposals which you might find interesting:
https://schalkenbach.org/wp-content/uploads/Decoupling-Land-and-Improvement-Values-PDF-Version.pdf
https://economicsdesign.com/featured-insights/deterring-metaverse-speculators-increasing-liquidity-3/
Two questions:
The first question is about the theory. My understanding is that Georgian theory is based on Richardo’s theory of land rent, which was developed for an agrarian setting. So I get that some land has an intrinsic value because it can grow corn. In urban settings it’s supposed to be about location, so that one is paying for access to subways, shopping, etc. So is the basis for a SFH land valuation about views or local amenities? It seems it would be difficult to determine next most efficient use, but aesthetic or convenience preferences are quite subjective. Maybe a measurement issue, but more importantly, what IS land value such that it could be identified and taxed?
Second: If SFH property values increase, rents and amenities stay the same, that increase is due to a subjective expectation of rising house values, such that gains will be made in the resale. How are such speculative gains related to land value? (Where I live, this is the real driver of unaffordable housing.) These seem like different beasts.
It's hard to say what the true value of a piece of land is, particularly because of how subjective it is. However, with Georgism we only really want to tax "unearned" value, not subjective value.
I think of "unearned value" as the rent an owner could collect while doing absolutely nothing of value. In other words, the rent they could collect by simply auctioning off the rights to use an empty plot of land (for example, a family might rent the land and build a SFH on it and continue to pay rent to use the land). This unearned value doesn't come from the landlords efforts so can be safely taxed without creating a distortion (in theory). Though nothing requires us to tax this unearned rent, it's a convenient way to fund public goods.
Even though this unearned value definition is more precise, estimating land value is still quite hard!
Second question: property values are kinda sorta the sum of pure land value and pure house value (other amendments matter too, there's also come synergies here). When overall property values rise, that is typically attributed to rising land value. It would be strange if houses themselves suddenly got more valuable across an entire city, especially since buildings typically depreciate in value over time. This growth in land value is another sort of unearned value (since it's unlikely that any one owner could raise land values across a city).
If you don't let the taxes grow along with the growing land value, then the owner gets to keep the unearned value (which can be acceptable). Though you have to be careful not to have taxes grow faster than land values.
Does that answer your questions?
Bad idea, which stems from the initial error of using selling prices as the basis of land value taxes.
The basic value of land is its gross annual value, which is its market rental value PLUS all property taxes currently payable. Selling prices are a derived value, being the capitalisation of that part of the gross value that is retained by the owner, plus an uncertain amount of hope value. It is an unstable value, dependent on, among other factors, interest rates, expectations of future interest rates, and the real estate cycle.
It lays LVT open to the valid objection that it is a tax on a value which cannot be realised. It also promotes general misunderstanding about the very nature of land economics.
This issue has been widely discussed, at least in Britain, where property taxes have traditionally and quite rightly been based on rental values.
Even though selling prices are derived from the thing we want to tax (annual ground rent), they bear a close relationship! Setting the tax rate appropriately should approximate taxing annual ground rent.
Its true that the landowner may not be able to realize the full value of the land, which is why it's important to set the tax rate to robustly under-tax land value (this is one of the topics in the follow-on post). Bidders should be rewarded for price discovery, and landowners should get some value from their land and local amenities.
The relationship between rent and price is far from close. It varies by several % about a mean of 5% (20 years' purchase). On top of that are factored in expectations of rental growth, development potential, inflation, and how the real estate cycle will develop. Worse still, the tax leads to a reduction in the selling price by am amount corresponding to the capitalisation of the tax payable. The LVT erodes its own base. It is like sitting on the branch of a tree and sawing oneself off.
Rental values are in most situations readily ascertained and are of their nature up to date There are well established and reliable methods of extracting the land value component of the rental values of developed land. British practice is, or was until recently, to value land and buildings separately, using rental figures.
The use of selling prices for LVT is conceptually flawed, stirs up justified opposition and sets the system up for failure.
The aim and purpose of LVT is to collect land rent for public revenue. Taxation of land rental values is transparent and people can understand what is happening.
The LVT movement could usefully give up the idea of taxation based on selling prices and promote a switch to rental value assessments where LVT is already applied. LVT based on selling prices is an important reason why it makes so little progress.
Note that land values will get reassessed regularly under this system, so values should adapt to changes in interest rates, rental growth etc. Overall, it looks like cap rates are relatively stable:
https://www.statista.com/statistics/245008/us-commercial-property-cap-rates/
I don't expect these kinds of fluctuations to seriously change the efficiency of the proposal, so long as the tax rate is set appropriately. For example, setting the tax rate at 4% would ensure that owners get some surplus (and cap rates have stayed above this level for last 10 years). Though fluctuations can cut into the size of the surplus land rent that the owner receives.
> The LVT erodes its own base
True! This is part of the goal of land value taxes, to reduce the value of holding empty land. A good thread on this:
https://twitter.com/GeorgistSteve/status/1654267040700719104
That thread also speaks favorably about land rent taxes.
I'm still agnostic about auctions, land value assessment, land rent assessment, and other forms of land valuation. I think they have different tradeoffs. What hasn't been covered here is how these valuation systems deal with other types of economic land, which I'll cover in the next post.
Capitalisation rates become volatile towards the end of the real estate cycle, eg in the early 1970s, late 1980s and the late 2000s. The chart in your reference diagram records just a part of the current cycle.
Values should be frequent. Annual review of rental values is feasible.
One aim of LVT is to reduce the price of land. If the tax is based on land prices, it destroys its own base. 4% tax on prices would make land worthless and that would put an end to LVT.
I be interested to see the existing literature on the elasticity of property and land taxes.
For example, Texas charges a 1.81% property tax. If land value is 70% of property value, then a 4% LVT would cost the same as a 2.8% property tax. Would moving to a 2.8% property tax destroy land values in Texas?
I would be surprised if it did, but I'm not that familiar with this area.
All taxes ultimately come out of land rental value. In conjunction with other taxes, the British business property tax has turned large tracts of the country into an economic desert.
The aim should be to collect 90% of gross rental value. To achieve this, rental values must be measured regularly and reasonably accurately, and collated in a database. Auctions based on selling prices would not achieve this.
LVT doesn't actually result in zero land values. The land value has to be nonzero if the tax is nonzero. The value becomes an indicator of what the rental value of the land is worth. E.g. take an LVT of 200%. I can only bid $5000 for a plot of land if I am willing to pay $10,000/year in taxes to have it. High LVT rates effectively become taxes on the land rent. https://peoplesrent.substack.com/p/investigating-land-value-tax-rates
Land price is the capitalisation of the annual rental value that is retained by the owner. Land price drops by about 20 times the annual land tax that has to be paid. If LVT assessments are based on land prices, then the tax cuts into its own base. This sets the system up for failure from the outset.
Yes, it is ultimately rents that should be captured from the land, and rents are a great way to inform tax assessment. You bring up good reasons why land value is itself harmful — it's speculative, and buyers assume a lot of risk in the case the land value decreases and they wind up underwater on their loan. One could argue that taxing land value out of existence is the point of the LVT. It's a pigouvian tax on the harmful effects of land value. But it is indeed weird to base the land tax on something that gets driven close to zero. Totally agree with you. Taxing rents themselves is much easier to think about! You might like this post about thinking with the Land Rent Tax instead of the Land Value Tax: https://peoplesrent.substack.com/p/the-land-rent-tax-and-real-time-georgism
This subject is discussed in several articles on the web site of the Land Value Tax Campaign, which is unfortunately now defunct. The web site will be transferred intact to another location with the same URL.
https://www.landvaluetax.org
Land values based on selling prices seems to be a USA thing. It sets LVT up from failure from the outset.
Appreciate the effort, but given that land and improvements are already separately valued in most states using long-established appraisal procedures, I think this a solution in search of a problem.
I see your point, but it's at least possible that this procedure is more efficient than current appraisal procedures!
More importantly, physical land is only one type of economic land that auctions can be used for. In the next post, I'll talk about that a little bit.
I probably am not understanding well enough -- maybe you covered this and I missed it?
1. If peeling the land off from the improvements means I have to publicly auction, then only be able to sell the improvements to the new land buyer, wouldn't it be better to sell them together? That way I can sell to whomever I want, without public auction.
2. If the only person I could sell the improvements to is the new land buyer, wouldn't the new land buyer have monopsony power such that they could get the improvements for almost free? (Or make me pay them, since I'm on the hook for removing the improvements if we can't reach an "agreement"?)
Good intuition, you're one step ahead of me! I'm going to look at this in more detail in the next post, but here's the gist:
Both the buyer and the seller have an incentive to keep things out of the land auction. For the buyer, this is because the dollars spent on a bid get taxed. The buyer responds to taxes by lowering their bid. Because the taxes lower their bid, the seller also wants to keep things out of the land auction, since they get less money than they otherwise would.
Moving to point #2, I should clarify that the original owner can sell the improvements to anyone, not just the auction winner. There's two key cases:
A. For items that are easy to take off the land (e.g. a potted plant or something) the original owner is better off selling the item on the market (or keeping it for themselves, or selling to the auction winnner at market price).
B. For items that are hard to take off the land (e.g. a house not designed to be moveable) the buyer does have a sort of monopsony power. Appendix 1 tries to model this and I find that the buyer can use that power to get the house for a little cheaper, but not for free!