A Simple Way to Tax Land
Introduction
A land value tax (LVT) is an old idea that has recently attracted renewed attention. Unlike most other taxes, a LVT creates essentially no distortions, given that land is in fixed supply. A LVT differs from a real estate property tax, which applies to both land and structures.
It has been suggested that a LVT is difficult in practice because it is hard to measure returns to land. I will show, instead, that it is in fact possible to implement a reasonable LVT fairly easily. All we need is the price a property was sold for, the length of time it was owned, and a record of past investments in structures. This LVT does not require any estimates of land values, which are costly and to some degree arbitrary.
A Simple LVT
A real estate property typically consists of both land and structures, and we would like to tax the land but not the structures on it. I assume first that properties consist solely of land; later I discuss a way to separate the value of land from structures.
Let L(t) denote the value of a plot of land at time t, where t counts years. The land is bought at time t=0 and sold at time t=T. We typically observe L(0) and L(T), but not L(t) for 0<t<T.
There are two challenges. To begin, especially when people live in their own homes, we do not observe the value of the “dividends”, i.e. the housing services the property provides. Second, even if we did observe these dividends, taxing the final capital gain leads to well known incentive problems. For properties that have increased significantly in value, the benefit of holding on to the property outweighs the benefit of selling. This is known as a “lock-in” effect. Vice versa, for properties that have fallen in value, the benefit of selling outweighs the benefit of holding on.
Is it nevertheless possible to apply a tax to the returns to this land?
Consider that there is an opportunity cost when buying a plot of land. For instance, the buyer could have instead invested the money in safe assets. Given that the buyer chose to purchase land instead, the benefit they receive must be at least as large as this alternative.
Let us assume that safe assets yield a per-period fixed return r>0. (We could let r vary over time, but this adds mathematical complexity while not affecting the main results.) Thus, at time t, with land worth L(t), the opportunity cost is r L(t). Let θ denote the tax rate. The tax liability B(T) is then
Hypothetically, the lock-in effect would be mitigated by taxing returns on a continuous basis. But, as mentioned above, we don’t observe L(t), and there are many ways prices could have moved from L(0) to L(T).
Constant Growth?
Suppose we assumed that land increased in value at a constant rate: L(t) = L(0) exp(gt), where g is the annualized growth rate of the land’s value.
The tax liability then takes the following simple expression:
But, while this may seem reasonable, it could still lead to the lock-in effects described above.
Holding-Period Neutral Tax
In contrast, with a tax that is holding period neutral (HPN), the owner will be indifferent between holding on to the land or investing the proceeds in a safe asset. Alan Auerbach showed that the following tax satisfies holding period neutrality:
where
and
is the pre-tax return to land (in contrast with Auerbach I assume that other forms of capital are not taxed, as they arguably shouldn’t be). The HPN tax thus assumes that owners earned a constant pre-tax return r*.
The HPN tax simplifies to
which, if desired, can be closely approximated by a piecewise linear function.
Let us compare this HPN tax system with the constant growth tax system. Figure 1 shows an example with L(0)=1, r=2%, θ=20%, and T=10. We see that the HPN system taxes gains more heavily. This makes sense: the HPN system assumes that owners began with a property worth more than it actually was. Vice versa, the HPN system is gentler on losses, since it assumes that the property was initially worth less than it in fact was.
Dividends
What about the “dividends”, i.e. the services that a plot of land provides? These are usually not observed, especially in the case of owner-occupied housing. If they were observed, the HPN tax would simply treat them as a sale. That is, each dividend D(t) would trigger a time-t tax of
With financial securities (where dividends are observed), it would be important to levy this tax. Otherwise, there would be a strong incentive toward issuing tax-free dividends. With land, however, dividends are not a choice, and so a tax based solely on land values would not create this distortion.
There is, however, one issue to consider, given that land dividends are not a fixed fraction of land prices. Imagine, for instance, a world with no uncertainty and two plots of land that would have the same price if taxes were zero. The first plot has fixed dividends, but the second has constant dividend growth. Although returns are equal, the second plot will end up with a higher price, and therefore a higher tax burden. However, since both pre- and post-tax returns should be equal, the plot with dividend growth will trade at a discount relative to the plot without dividend growth. The key is that land is in fixed supply, and so differential taxation rates should not lead to inefficient resource allocations.
Structures
So far, we have assumed that properties consist solely of land. In reality, properties typically consist of both land and structures such as houses and office buildings. We do not want to tax these structures. To ensure this, we will need to know the value of a property’s structures, at least when it is sold.
How to do this?
One method is to continually hire professional appraisers. But it would be easier to look at investments in structures: we can estimate the value of any asset by looking at past investments and applying a depreciation schedule. Property owners will need to report any investments in structures as well as depreciation. There is a strong incentive for property owners to report these investments, as a higher valuation of structures will mean that these will constitute a greater fraction of the final selling price (and so the value of land will come out lower). This will also make a property more attractive to prospective buyers.
It would be easy for the government to verify that owners report depreciation correctly, since it has a record of past investments. Moreover, the IRS already has clear rules defining home improvements and depreciation.
In addition people who sell their homes today are allowed to adjust the buying price (“basis”) upward by adding the cost of any improvements made.
In short, accounting for the distinction between the value of land and structures does not require adding anything conceptually new to the tax code.
Final Thoughts
It would not be difficult to switch to this LVT. The main transition cost would be a one-time determination of the value of existing structures. In fact, this is not strictly necessary: the government could decide to set the value of existing structures to zero, so that for tax purposes all properties consist solely of land. This would not hurt the incentive to invest in future structures. Implementation could be done at the local, state and/or federal level. Ideally, this LVT replaces all other real estate taxes, such as property taxes, transfer taxes, taxes on rental income, and taxes on capital gains.
