A Fair Price for a Resource That Is Not Free

QuestionThe oil comparison, and the price

A state that owns oil can charge for it and share the proceeds equally, and nobody pays a new tax. Two questions follow for any other resource: what is a fair price when the asset is not free, and does the same structure work for something like data?

The method is the one used across this book: say exactly what was done, show how each number was produced, and mark what is assumed. Every figure below is invented.

Section 1Rent: the surplus that is left

Economic rent is the surplus value of a resource after every cost, including a normal return on the capital invested, has been paid. In symbols, with revenue R, operating cost C, capital K and required return r:

Rent = R − C − r · K

Take two invented firms, each with revenue 100, capital 200 and a required return of 10%, so the return on capital is 20 for each. Firm A has operating cost 60, so its rent is 100 − 60 − 20 = 20. Firm B, a marginal one, has operating cost 78, so its rent is 100 − 78 − 20 = 2.

This bounds a fair price. The floor is that the operator keeps its costs and a normal return, or it will not invest. The ceiling is the whole rent. A price is a share s of the rent with 0 < s < 1; where inside that range it falls is a judgment about values, which arithmetic cannot settle.

Section 2A levy on revenue, or a share of rent

A levy of 5% of revenue costs each firm 5, whatever its margin. For Firm A that is 5 out of a rent of 20, or 25% of its rent, and A stays viable with 15 left above its required return. For Firm B it is 5 out of a rent of 2, or 250%: B is left with 2 − 5 = −3 below its required return and would stop investing. A levy on revenue taxes a project whether or not it is profitable.

A 40% share of rent, by contrast, costs A 8 and B 0.8. Each keeps something above its required return (12 and 1.2), so both stay in.

The share also cannot reach the ceiling. If the firm cuts its cost by 1, it keeps 1 − s of the saving. At s = 40% that is 0.6; at s = 100% it is 0, so the firm has no reason to cut cost at all, because the saving goes to the owner. This is why practice sets the share well below the whole rent.

Section 3The dividend, and who pays

Equal share. If the take T is paid out equally to N people, each gets T / N. With an invented T of 1,000,000,000 and N of 20,000,000 people that is 50 each: a real payment but a small one, because a collective value is spread thin.

Share by value. If instead each person is paid in proportion to the value of their own data, three people with values 10, 30 and 60 and a take of 300 receive 30, 90 and 180, where an equal share would give 100 each. The top person receives six times the lowest, so a payment tied to individual value reproduces the differences it started from.

Incidence. Suppose a fraction θ of the levy is passed on to the same people through higher prices or lost services. With θ = 0.5 and a take of 300, the firm bears 150 and the users bear 150, so the net transfer from firm to users is (1 − θ) · 300 = 150. At θ = 1 it is zero: the users pay their own dividend. What θ really is depends on the firm’s market power and on how much of the levy it can pass on, which is an empirical question this page does not answer.

Section 4Where data differs from oil

• It is a new tax, not a royalty. Oil rent comes from a resource the state owns. The data proposals below are taxes on company income, so the feature that nobody pays a new tax does not carry over.
• There is no auction and no spot price, so the rent has to be estimated: the value a firm earns from the data, less the cost of collecting and processing it, less a normal return. This is the author’s reasoning, not a finding from the sources.
• The firm may be abroad. A country can tax a well on its soil; the firms that monetise its residents’ data may be elsewhere. Also the author’s reasoning.
• It is not used up. Data does not deplete the way oil does, so the question of what is left for later generations changes, though revenue depends on how platforms’ business models evolve.

Two real proposals show the gross-revenue design. New York Senate bill S6102 (introduced May 2019) would have taxed corporations at 5% of gross income earned from data about New York residents, put the proceeds into a state data fund, and paid a dividend to taxpayers. A companion bill, Assembly A09112, was introduced in 2020. The Senate bill is listed as dead as of January 2020. By Section 2, a 5% levy on gross income has the flaw shown there: it takes the same amount from a profitable firm and a marginal one.

LimitsWhat this does not show

• Any real firm, price or payment. Every figure is invented.
• The correct share of rent. Real regimes differ, and the choice is political. A published range of government take for oil wells (about 31% to 83% in one summary) was not traced to its source here, so it is not used.
• That the Alberta oil-sands royalty is optimal. One review of resource-rent capture calls it the closest of those studied to a rent-based design; this page only reports that, and has not read the full study.
• The legal or constitutional status of any data levy, or whether a dividend would change behaviour.
• The pass-through fraction θ. It is set by assumption, not measured.