does a land value tax have zero or negative deadweight loss?

i was recently reading the wikipedia article on optimal taxation theory, when i came across an assertion that land value taxes (LVT) have not zero deadweight loss, but actually negative deadweight loss.

Economic theory suggests that a pure land value tax which succeeds in avoiding taxation of improvements could actually have a negative deadweight loss (positive externality), due to productivity gains arising from efficient land use.

one of the citations was to the economist nicolaus tideman, so i reached out to him. he proposed the following example.

You own a downtown lot that is used for parking. Your income is $250,000 per year, and there is no land tax, so at a discount factor of 10%, your lot is worth $2.5 million for parking. But someone would give you $10 million as a building site. That is the market value of the land. You expect that price to rise by 10% per year, so you understand your full income to be $1.25 million. [that’s the 250k$ you make from one more year of parking revenues, plus the 1m$ appreciation from 10m$ to 11m$.] If you sell, your income from the sale price will be $1 million per year. Best to hold.

Now implement a land tax at a rate of 10% of the value of the land. The value of the land falls to $5 million, with taxes of $500,000 per year. Your 10% price gain, in this example, just offsets your taxes, so your net profit (parking plus price gain minus taxes) is $250,000 per year. But if you sell the land for $5 million, then, at the assumed discount rate of 10%, your income is $500,000. Now, because of the land tax removing half the income from speculating, it is best to sell.

but the problem with this example is that in order for the lot to be worth 10m$ to the prospective buyer, it has to be develop-able into an asset that generates a 10% return. but if it could do that, then you as the current owner would already have an incentive to develop it, rather than making merely a meager 2.5% return as parking.

winner’s curse

another relevant concept is called the winner’s curse. but whether the winner’s curse applies to land value taxes centers on the notion of what the winner’s curse actually means. in my understanding, there are two essential factors that determine variability in bids for an asset:

1. different levels of ability for the bidder to productively utilize the asset. (i.e. renoit can afford to pay more for scarce oil paint than i can, because he can turn it into a painting worth millions of dollars, whereas i have zero artistic ability.)
2. different levels of optimism about the inherent market value of the asset.

because the first factor essentially cancels out in the market, this leaves the second factor as the dominant determinant of bid variability, meaning the winner will be “cursed” with lower returns, by definition. so the winner’s curse doesn’t inherently hold any great implications for land value taxes.

that said, here’s my own re-framing of this scenario, which i believe offers a more interesting “true conundrum”.

You own a parcel worth 2.5M to you, but worth 10M to someone else. so it would seem to make sense to sell for profit. but you expect a future positive externality receipt (e.g. a gold meteor lands on your property, you strike oil, Google builds new offices nearby, etc.) with an NPV of 1M (an NPV analogous to the more complicated expectation of 10% growth in your example). So really, that other party should pay you 11M (the 10M they think it’s worth to develop, plus the price of that meteor that’s going to land on your property). But they don’t know/believe that 1M, so they’re only willing to pay you 10M right now. So you’re just going to wait until the meteor lands next year, and then sell it to them for the full 11M. But there’s economic loss here because the property went undeveloped for that full year.

now, in principle, you could just develop the property right now, and also pocket the externality. but maybe you don’t have access to the capital and/or skills to do so. in which case, this seems like a legitimate case of lost productive capacity. but, as i continue, there’s still a simple way out of this conundrum.

So then the question is, who’s right? If the prospective buyer is really confident they’re better at predicting the future than you, then they can just “short” the property. I.e. you sell the land for a future payment of 2f — e, where f is the actual future value and e is your expected future value (add in standard 0.3% to 3% shorting fees of course).

tideman agrees with this in theory, but suggests that such an esoteric financial instrument might not exist in practice. i would tend to agree with that. and, ironically, this may be the kind of arena in which blockchain-based “smart contracts” could actually be useful, because they could facilitate such esoteric financial instruments.