As I said, 100% is the aim, if not a practical possibility. The effect on the amount people are prepared to pay to buy a piece of land is the same. Zero, or close to it.
Vacancies act to regulate the market. This being one instrument to set rents, thus maximise the income from land. So at any one time there will be a range of rental values collected (whether by private landowners or the state) from 90% to over 100%.
As it happens, with the current plethora of data we have now on rents, selling prices and vacancies its very easy to tax 90-95% of land rents. With dedicated computer modelling easy to get better than this should we wish. Not a problem either way.
As for valuations, housing (which is by far the biggest part of agg land values) is the most straightforward, simply due to the fact homes can be put into classifications and compared like for like between locations.
Without any fancy computer modelling, this would capture > 90% of housing land rents. Of course, if the aim was to collect as much land rent as possible for tax revenue, it would be worth investing in computer modelling which would get very close to 99%.
Commercial property is a little more difficult, but the same principles apply.
As a thought experiment. Imagine you owned all the land in which ever country you lived in and wanted to maximise your income from it. I think you could do better than 2/3rds given the amounts at stake.
The aim is to collect 100% of lands rental value, which drops its selling price to zero. So people would assess the value of land as zero if they had to pay a full LVT.
By definition any tax that disincentivizes a human input, like discovering new land and uses for it, is not a land value tax. Many people causally assume that a land tax must somehow be bad or inefficient. They then concoct a tax whose incidence falls on land, labour and capital, call it a land tax, then proceed to beat up a strawman.
Let's take Bryan Caplan for example. In his regime, landowners are required to self assess their land's value for the purposes of taxation, but must sell their land to anyone who puts in a higher assessment. Caplan then stipulates that the landowners valuation is made public knowledge and then proceeds to prove how such a regime disincetivizes search and discovery. Which of course it does, as that stipulation is in effect giving away the landowners labour and capital used in search and discovery for free. Thus any such tax regime is grossly inefficient. Made efficient simply by not making the valuations public knowledge.
Caplan isn't a stupid man, so I can only assume that like many other economists who make similar errors, he has done so out of ideological zeal or blindness. His citation of Idi Amin as an example of the possible perils of land taxation is bizarre.
The assumption that the suppliers of capital will be disincentivized to supply it unless they get to keep the uplift in value of something they didn't is also bizarre and without any justification. Charles Hooper erroneously states that land values are "created". Goods and services are created by human inputs. Land values only arise by shifts in demand for its use. Causation is not creation.
Lets take the discovery of the internal combustion engine as an example. Before then crude oil seeping up from the ground was a toxic pollutant, making land worthless. After the ICE had been perfected and the motor car industry took off, land with oil underneath it became very valuable, The suppliers of motor cars not only shifted demand for something they supplied, but also for something they didn't. They didn't need the uplift in land values to incentivise them to supply cars.
The supply of capital and the timing of development are not affected by a land tax defined as a tax on the opportunity cost of leaving land undeveloped (even though it might be developed). Thus the taxation of a plot of land does not change in response to how that plot is developed, only how it could be optimally developed and the demand for exclusive use of it.
It should therefore go without saying that while a land tax is perfectly efficient in theory, how it is applied in practice needs consideration and tailoring to its application. Residential land is by far the most valuable in totality and also the easiest to value due to the large amount of data available, making the separation of building and location values by a computer algorithm easy. Commercial property is a bit trickier and may need individual assessments. Natural resources, due to search costs, best handled by licencing via a bidding process. There are plenty of examples around the world of this working successfully in practice. If governments try to charge more than land's rental value, that will lead to vacancy and lower revenues. So the system is self-regulating. The state becoming defacto landlord over all land and natural resources, motivated by revenue maximisation.
We set a precedent for the confiscation of peoples assets when we abolished slavery. Righting an injustice is never wrong, even if it imposes costs on the beneficiaries of injustice. Not equally sharing the returns from that which god/nature supplies for free, is unjust, leading to equality issues, such as housing and the need to tax output. LVT would be another step towards the realisation that fairness and efficiency are two sides of the same coin. Requiring landowners compensate those they exclude via the tax system is, in principle, exactly the same as paying compensation for any other good, service or labour received.
No discussion of LVT is complete without wheeling out Poor Widows in Mansions. In essence, we are told that we should suffer chronic inequality issues like housing and a shrunken economy in perpetuity, to protect one generation of wealthy retirees. This is completely immoral. Besides which, roll up and deferment can be offered, acting as an avoidable inheritance tax. Retirees, banks, pension companies etc having to invest in productive assets rather than land will improve efficiency, not reduce it.
One thing I do agree with Matthew Barnett with is Georgist tendency to exaggerate the effects of a LVT on ending land speculation. For a start, land speculation is good, as arbitrage puts it into the hands of people able to use it best. There is already an opportunity cost to not putting land to its best use and a land tax doesn't change that. By lowering its selling price a LVT might help those with less access to large amounts of capital. A LVT certainly would allow the market to rationalise existing housing, reducing costs. But I think the effect on individual land owners to put their land to better use is weak. However, the effect on the state and its policies on land use are likely to be profound, on the assumption that it will want to maximise revenues i.e the efficient use of land maximises its aggregated value.
Ultimately all policy changes are based on costs vs benefits. The quantifiable benefits of a fiscal shift to land taxes are as follows (I live in the UK where the ONS estimates land as 60% of total wealth, which would yield initial tax revenue of around £350-£400bn per annum. However as all taxes are incident upon land, their reduction would substantially increase rental values as would the growth from the alleviation of deadweight costs.)
1 The tax ends equality issues, like housing, overnight. All else equal, the aggregated selling price of land measures housing affordabilty/misallocation issues. A LVT drops that to zero by transferring incomes to those that find housing most unaffordable now, with a concomitant effect on average rents and houseprices, reducing those by 2/3rds (in the UK). If affordability is measured by the ratio of discretionary incomes to prices for typical working households, a shift to LVT improves that by a factor >3. That's what the solution looks like. Any other measure, such as supply-side, rent controls, social housing etc are very poor mitigation policies in comparison. A LVT allows the market to rationalise existing stock, reducing costs.
By allowing for radical reductions of taxes on output, the land taxes bring marginal locations back into production, re-balancing spatial investment and thus reducing costs.
2 A land tax raises enough revenue to radically reduce taxes on output and radically simplify the tax system as a whole. The deadweight costs of taxes in a country like the UK is 30-50% GDP. A LVT on its own would likely reduce that by a big chunk. In the US a recent paper estimated growth between 15-25%.
A land tax would not only radically reduce houseprices, but eliminate and boom-bust cycles due to their volatility. The US is estimated to have lost $2trn from its GDP because of the GFC, The UK fared even worse due to its reliance on the banking sector (GDP per capita has flatlined for 15 years).
Taken together, this represents a quantifiable, transformational win-win. Against that we have unquantifed assertions about LVTs negative effects.
Furthermore, there are unquantifiable benefits of LVT. Like how it would change incentives for the state to improve its land use policies and facilitate its accumulation for development. But most importantly, an LVT combined with a Citizens Dividend (UBI) makes it clear to all of us who owes what to whom and why. Socialism can be characterised as a set of polices designed to mitigate economic injustice. This neither solves them and imposes high economic costs. It also divides society into a false choice between higher growth vs more equality. This is politically and socially corrosive. The "Georgist" regime seeks to end socialism, thus uniting society. And that's priceless, in my opinion.