On Progressive Taxation
Why it's better to tax consumption rather than income from a liberal perspective
Very short summary: This essay provides an account in favor of a progressive consumption tax, in light of the efficiency and fairness issues that affect the more common progressive income tax. I argue that the progressive consumption tax not only avoids the standard incentive problem but also responds to Hayek’s critique of the unfairness of progressive taxation. The progressive consumption tax is the most promising way to maintain a progressive tax system from a liberal perspective.
A large majority of countries in the world have a progressive income tax (PIT), making the PIT probably the most popular form of taxation.[1] While the degree of progressiveness of the income tax varies widely across countries and is very often a point of contention, the principle itself, that income should be taxed at a marginally increasing rate, is nowadays largely taken for granted. Many view it as the most convenient way to allocate resources to public spending while, in the meantime, reducing economic inequalities.
Economists tend to be less enthusiastic. The classical objection is in terms of incentive-compatibility. The more progressive the income tax, the more it disincentivizes individuals to work more to receive additional income, as well as to invest in resources and skill development to increase their future income. Well-known results in optimal taxation theory indicate that the optimal marginal rates depend on the price-elasticity of labor supply (optimal marginal rates decrease with the elasticity) and (as far as labor income is concerned) on the distribution of ability in the population (in terms of efficiency, an increase in a marginal tax rate is preferable if it mostly affects people inframarginally, i.e., people who are not in the highest part of the income distribution). More controversially, theoretical work suggests that marginal rates should decrease at the top of the income distribution —where and how much depend, again, on the distribution of ability.[2] The bottom line is that from an economic perspective, the case for a strongly progressive income tax is weaker than generally assumed because of the efficiency losses the PIT causes.
In The Constitution of Liberty, Friedrich Hayek develops a different argument against the progressiveness of the income tax.[3] Rather than pointing out the effect on incentives, Hayek’s argument targets the unfairness of taxing income at increasing marginal rates. For instance, consider the case of two lawyers who collaborate on a given case. Let’s assume that they provide the same service and are equally remunerated by their client. However, one of the two lawyers earns significantly more than the other during the year. With a PIT, the post-tax income effectively received by the two lawyers will differ, despite the fact that, by assumption, they produce exactly the same service to the same person. As Hayek observes:[4]
“This means that progressive taxation necessarily offends against what is probably the only universally recognized principle of economic justice, that of ‘equal pay for equal work.’ If what each of two lawyers will be allowed to retain from his fees for conducting exactly the same kind of case as the other depends on his earnings during the year —they will in fact often derive very different gains from similar efforts.”
It’s not exactly clear whether the principle “equal pay for equal work” really is “the only universally recognized principle of economic justice.” This principle is in competition with alternative ones, such as “to each according to one’s needs” or “to each according to one’s merit.” There are strong (and probably decisive) arguments against these latter two, and a good case can be made to defend the claim that, prima facie, two identical productive work should be remunerated the same way. But even this principle is not as universal as Hayek assumes. One way to make it more compelling is to reformulate it as follows: “equal pay for work that has equal value as subjectively perceived by those who benefit.” So let’s assume that this is really what Hayek meant by his argument against PIT.[5]
Now, let’s assume that we nonetheless don’t want to depart from the intuition that there is something commendable in having a progressive tax system. We may still want that, in one way or another, people who are wealthier contribute more to the provision of public goods and other public spending than those who are poorer. One possibility is to combine a (progressive or flat) income tax with a tax on wealth. Current debates in France over the so-called Zucman tax reflect this approach. Proponents of the Zucman, starting with Gabriel Zucman himself, argue that the French tax system is not progressive for the very high incomes (the top 0.0002%) because of many loopholes. In a Le Monde op-ed, Zucman, Olivier Blanchard, and Jean Pisani-Ferry point out that while a French person pays on average half of their income in various sorts of taxes (a very high rate, indeed), the wealthiest pay on average only 27%. This number is actually controversial (see, for instance, this response or this one), but I’ll set aside this point. The Zucman tax would make the very rich pay a tax of 2% on their total wealth, unless they already pay the equivalent through the other taxes.
The Senate just rejected the Zucman tax, so it has few chances to be implemented in France. This is not a bad thing. It’s clearly not a solution to the chronic public deficit problem —even less so once we account for the negative incentive effect it would generate. As I said in my previous essay, in the French case, it reveals a singular lack of imagination in a country where tax rates are already very high. Finally, even if we accept the contention that the French tax system is not progressive at the very top of the distribution (again, this may be disputed), it concerns a few thousand households. For the rest of us (including very, very rich people), the system still preserves progressiveness.
“The Banker and His Wife,” Marinus van Reymerswaele (c. 1539)
If you value progressiveness and are convinced that neither a PIT nor a wealth tax is a good way to achieve it, there is still a third way to proceed: a progressive consumption tax (PCT). In Europe, consumption is already taxed through the value-added tax (VAT) that consumers pay (with different rates depending on the country and the kind of good) when they buy a good or a service. A common objection to the VAT is that, because it is proportional (it is basically a flat tax) and because poorer households consume a bigger share of their income than the richer ones, it is, in practice, regressive: the more you earn, the less you pay in proportion of your income. To that extent, making the consumption tax progressive would counteract the implied regressiveness of the VAT.
Depending on the tax scheme adopted, the consumption tax could also be made genuinely progressive, i.e., ensuring that the higher your income, the more you pay both absolutely and relatively speaking. You would obtain this if you combine a relatively high tax-free threshold with steeply increasing marginal rates. Households whose total yearly consumption is below some threshold X would not pay any consumption tax at all.[6] Those consuming more than X would start to pay for any consumption amount above X, and the marginal rate would increase quickly. Moreover, contrary to the PIT, there is no theoretical limit for the higher marginal rate. In principle, you can tax very high levels of consumption above 100% (which obviously doesn’t mean that you should do it).
By experience, when you expose the idea of the PCT, most people react negatively. So let’s quickly answer the objection that is often made first. The PCT would be a direct tax, exactly as the income tax, contrary to the VAT. To proceed, the tax office basically needs the same information as the one required to levy the income tax. You only need to know each household’s (i) yearly income and (ii) yearly savings. This kind of information is already accessible to tax offices in most countries. A more relevant question is how a PCT would be implemented at the beginning, in combination with a flat or progressive income tax. Over the long run, the PCT should probably be considered as a substitute for the PIT —there’s indeed no point in combining the two. A transition period may nonetheless be needed.
However, let’s focus on the more substantive and relevant aspects. The PCT has the virtue of being mostly immune to the incentive problem that affects the PIT. Individuals would not pay tax on their income, but only on how they use it. Therefore, there is no incentive to work less or (if different streams of income are taxed differently) to disguise a source of income for another (one of the loopholes that may produce regressiveness at very high levels of income). This considerably simplifies the issue of efficiency that is so delicate in the case of the PIT.
Of course, a PCT would have an effect on how individuals use their income. It would encourage increased savings. Today, the opportunity cost of consumption is approximately equal to the interest rates at which your savings can be securely remunerated. A PCT would mechanically increase the opportunity cost of consumption, at least above some amount. Over the long run, increased savings are a good thing, especially in countries with economic imbalances. A PCT would be far more effective than putting tariffs in a country like the U.S., where saving rates are low and consumption is basically funded by foreign capital, for instance. In other countries like France, where saving rates are already relatively high but savings are not supporting enough entrepreneurial (and thus risky) activities, it should probably be combined with policies channeling them toward more productive uses. However, we don’t want to impede the poorest households’ ability to satisfy their needs, which justifies a relatively high tax-free threshold.
The economist Robert Frank has been one of the most vocal proponents of the PCT over the last couple of decades.[7] Frank’s argument for the PCT is mostly made in the context of his discussion of positional goods. According to Frank, a significant amount of consumption is motivated by conspicuous reasons. A lot of what people consume follows from invidious social comparisons: they buy new and bigger cars or fancier new phones not because they need them, but because they want to emulate others, or even “beat them.” Conspicuous consumption has a social cost that corresponds to the amount of resources that people allocate to goods that are not intrinsically useful but only for the purpose of social comparisons. Taxing or even forbidding conspicuous consumption would be impossible, on top of being probably immoral. However, a PCT offers an indirect solution to this problem. If we assume that households allocate resources to conspicuous consumption at least in proportion to their income (and we may plausibly assume that this is actually more than proportional), then richer households will be more than proportionally (because of the progressiveness of the tax) taxed. Plausibly, for very high levels of consumption, most expenses are tied to positional considerations. A PCT guarantees that any additional dollar or euro used at such levels will induce a high level of tax.
In a previous essay on the topic, I provided a different argument in favor of the PCT. Some economists have been suggesting the use of non-linear pricing in the context of the environmental crisis. The idea is that for goods whose consumption (or production) has negative environmental effects (e.g., flights, electricity), individual prices would go up with the quantity consumed. In principle, at least, non-linear pricing may spontaneously emerge in a free market economy. However, experience suggests that this is unlikely. It would then be up to the state to make it mandatory in some domains of the economy. I will not repeat what I said in my previous essay about this proposal, but just note one point: a PCT would basically achieve the same end without involving complex administrative battles, opening the door to regulation capture.
However, it can be objected that there is a major difference: while non-linear pricing schemes would target specific goods and activities, a PCT would indiscriminately apply to all goods and activities. This is related to a broader critique that some (starting with my wife!) have been voicing to me: the PCT is basically unfair because it will punish people when they decide to increase their spending for things that we would like to encourage or at least not discourage: going to the theater, buying books, paying for one’s children’s education, and so on. This critique has an intuitive appeal that is difficult to deny. I would however respond to it in three steps.
First, under the current VAT scheme that prevails in many countries, these activities are already taxed and very often at the standard rate. Since the VAT would disappear in a world with a PCT, everything else equal, the poorest households would pay less for them (since they would not pay any tax at all) than what they pay now.
Second, ideally from a liberal perspective, the arbitrary judgment of the state should not be left to determine what is a consumption activity that should not be (or less) taxed. Assuming that we have to tax one way or another various forms of economic activities, it is best to avoid making any value judgment about what is “worth” or “good.” This is the same reasoning as for the taxation of conspicuous consumption: we don’t want the tax scheme to rely on subjective judgments that may reflect prejudices or controversial philosophical views.
Third, irrespective of how resources are used for consumption, we have good reason to think that past some level, these resources would be better used either for investment purposes or for the provision of public goods that benefit all. This applies to buying an expensive car, going to a very expensive university, or going to the theater for the tenth time in the month.
To conclude this long essay, let me contemplate how the PCT fares with respect to Hayek’s argument against progressive taxes. As the reader will remember, Hayek’s argument is basically that it is unfair that two identical productive activities with the same economic value don’t generate the same income to their producers. This doesn’t apply to the PCT because in this case, our two lawyers will ultimately earn the same amount for their identical services. However, we would extend Hayek’s argument by noting that the value of money is ultimately entirely dependent on what it can buy. If both lawyers receive the sum of $Y but one is significantly richer than the other (and thus consumes significantly more), the purchasing power of $Y for the richest lawyer is less than for the poorest lawyer.
I think a convincing answer is to point out that consumption is only one of the possible uses of money. Money can buy goods and services, but can also buy shares or —figuratively— interests. A PCT has no effect on these uses. What is true is that it changes the tradeoffs between consuming and saving by increasing the opportunity cost of the former (and thus decreasing the opportunity cost of the latter). That makes the unfairness of the tax less salient because people can still choose to save more to “preserve” their income and use it later (and benefit from the returns on their savings). It changes the incentives but over the long run, not necessarily what people can buy with their money.
Bottom line: if you want a progressive tax system but are concerned about the inefficiency and unfairness of a progressive income tax, a progressive consumption tax is an attractive alternative. I’ve put implementation issues aside, as well as some other arguments in favor of it, so what I said is far from exhaustive. Still, it should provide a solid basis in defense of such a scheme, especially from a liberal perspective.
[1] Perplexity estimates that around 120 countries out of 195 have a form or another of PIT. The most significant exceptions are some countries of Central and Eastern Europe that are using flat rates (Hungary, Bulgaria, Russia) and Gulf states that don’t have any income tax.
[2] For a good overview, see this paper by Mankiw, Weinzierl, and Yagan.
[3] F. A. Hayek, The Constitution of Liberty: The Definitive Edition, (Routledge, 1960 [2020]): Chapter 4.
[4] Ibid., p. 275.
[5] With Hayek’s original formulation, LeBron James should receive the same income whether he lives in our society where many people highly value excelling at basketball or in an alternative world where almost no one is interested in basketball. What matters is not the work that is produced itself, but the value it creates for society. Hayek also gestures toward the inefficiency of a PIT, besides the standard argument about the effects on incentives to work and invest. He notes that a PIT will lead to a misallocation of resources as it distorts the price signal (since the net remuneration varies with the time rate at which earning accrues) and possibly to a reduction of the division of labor. I put these considerations aside.
[6] Kenneth Rogoff suggests not setting any threshold and providing instead a large lump-sum transfer (e.g., a universal income) to the poorest households.
[7] See, for instance, Frank 2009.
It’s not exactly clear whether the principle “equal pay for equal work” really is “the only universally recognized principle of economic justice.”
If justice is to mean anything I would also include “luck, which is pervasive in economic life (and in life in general), should be shared by all to the extent that everyone enjoys at least a basic level of dignity, security and care”.
This is great. I've been intrigued by the idea of a PCT and agree with it in principle. I do find myself thinking more about how this would change things on the supply side. Surely this would have substantial impacts on luxury goods. And while I agree that in aggregate perhaps the theater and the arts would not see large impacts, the margins may matter here. In the US I'm thinking of community theater organizations, for example. I wonder if a PCT in practice would end up with similar problems of complexity as a PIT, with carveouts and other incentives that end up causing a different package of issues.