Skip to content

Hill – Mankiw 9th Edn Chapter 12 – The Design of the Tax System

A commentary on Mankiw 9th Edn Chapter 12 – The Design of the Tax System (Mankiw 9th edition)

Mankiw, N. G. (2021) Principles of microeconomics (9th ed.)
Principles of economics (9th ed.)
Mason, OH: South-Western Cengage Learning.

Rod Hill

University of New Brunswick, Saint John campus

Saint John, New Brunswick, Canada


Chapter 12 – The Design of the Tax System

Here are some things to consider when reading this chapter.

  1. The choice of language when discussing taxes and government

The second part of the Commentary on Chapter 6 noted the choice of certain words. Like other textbook authors, Mankiw writes of the burden of taxes that are imposed by governments – words that convey a negative tone that isn’t used when referring to the costs of the goods and services people get from the private sector. Chapter 12 also uses this terminology, while adding, for good measure, the phrases the “tax bite” (p. 225) and “the burden of government” (p. 236). This choice of terminology harkens back to pre-democratic times. More neutral words could be used that would be more compatible with taxes levied by democratic governments accountable to the population.

  1. Judging the fairness of the tax system

(a) The benefits principle

As Mankiw explains (p. 235), the fairness of a tax system can be judged using two different principles. The benefits principle links government expenditures to the benefits people receive from them, claiming people should pay in accordance with benefits received. However, there are two practical problems with this.

In their text on Public sector economics, Stiglitz and Rosengard explain the first problem: “For the most part, economists have not been attracted to the benefit approach to taxation, largely because it is impossible to identify the magnitude of the benefits received by different individuals” (2015, p. 529). Mankiw sketches out a plausible story of how “the wealthy benefit more from public services”, but how much more they benefit remains unclear.

The second problem deals with government efforts to redistribute income. The traditional view is that the benefit principle is applicable when considering the benefits people get from government expenditures on goods and services, but inapplicable to spending whose goal is to redistribute incomes, e.g. transfer payments. As a result, “[f]or benefit taxation to be equitable, it must be assumed that a ‘proper’ state of distribution exists to begin with. This is a serious shortcoming since in practice, there is no separation between the taxes used to finance public services and the taxes used to redistribute income” (Musgrave and Musgrave, 1989, p. 219).

Mankiw addresses this directly with an argument building on his earlier discussion (p. 214) of antipoverty programs as a public good. Again, just how much people of different income levels benefit, on average, from poverty reduction, and thus should contribute to antipoverty programs, is unclear. But Mankiw’s basic point is that everyone can potentially benefit. The traditional approach implicitly assumes narrowly self-interested individuals who don’t care about the distribution of income in the society where they live.

Despite these issues, the benefit principle remains relevant. In democratic countries, tax and expenditure policies will end up approximately applying the benefit rule in this sense: “People, or some majority thereof, would not be willing to sustain a fiscal program if, on balance, they did not benefit therefrom” (Musgrave and Musgrave, 1989, p. 220).

(b) The ability-to-pay principle

This principle judges the fairness of taxes based on how a households’ net taxes compare with its ability to pay, regardless of the pattern of expenditures and benefits. This requires measuring ability to pay, such as some measure of household income, adjusted for its composition (because larger households require more income to maintain the same standard of living for each person in it). This income is then compared with estimates of taxes paid and transfers received.

Horizontal equity is the principle that individuals who are the same be treated in the same way by the tax system. This requires considering household composition and relevant individual characteristics, e.g. age, disability, medical needs, expenses incurred in earning income. Individual tax systems vary from country to country in attempting to implement this. Vertical equity is the principle that individuals with a higher ability to pay should pay more, although how much more is a matter of judgement.

(i) Comprehensive income and income for tax purposes

Mankiw doesn’t define what income is or how economists think of income for tax purposes. Income is defined as the sum of the two things that income can be used for: consumption and net savings (or equivalently, changes in net worth, which can be positive or negative). Traditionally, economists have favoured the concept of comprehensive income as the best way to compare different people’s ability to pay (e.g Stiglitz and Rosengard, 2015, p. 677). In principle, this would include things that people produce in the household for their own consumption as well as the services provided by owner occupied housing, both of which contribute to the household’s consumption. In practice, income tax systems largely focus on monetary flows, so these are not considered when determining income for tax purposes.

Comprehensive income would include accrued capital gains/losses, i.e. the change in the value of the assets the people own, whether or not they sell them and realize the capital gain/loss. (See Mankiw p. 241, question 4.) Again, focusing on monetary flows, only realized capital gains/losses are considered in the US tax system, without adjustments for inflation.

Income can be divided into market income and transfers from governments. Market income consists of wages and salaries, interest income, business income (of business proprietors or partners), dividends, realized capital gains, and income from private pensions. Transfers from governments can be monetary or in the form of in-kind services whose value is measured by the cost of providing them. In the latter case, this is conceptually equivalent to the government giving cash to households who, it is assumed, would then purchase those services at that price. (As described in the next section, official measures of US household income include some, but not all, of these services. As well, income measured for tax purposes varies from country to country.)

(ii) How progressive is the US tax and transfer system?

Mankiw’s Table 5, The Burden of Federal Taxes (p. 236), is offered as a way for readers to judge for themselves whether the ‘tax burden’ is fairly distributed or not. This requires a judgement about ‘vertical equity’: how people with different abilities to pay should be treated by the tax system. Mankiw’s discussion (pp. 235-6) states that income is the basis upon which taxpayers should be compared when thinking about tax equity.

Why then does Mankiw construct his data in Table 5 around the narrow concept of market income, rather than some more comprehensive definition? He offers no explanation, yet this excludes income from all social insurance benefits: unemployment insurance, Worker’s Compensation, Social Security, and in-kind benefits from Medicare. In contrast, the Congressional Budget Office (CBO) publication, from which he takes his data, defines income to include those social insurance benefits.

In its definition of income, the CBO excludes “means tested” benefits, that is benefits which depend on income in the same way that tax payments depend on income. These benefits, whether cash or in-kind, “directly affect the distribution of household income” in the same way that taxes do. In effect, they are treated as negative taxes intended to change the after-tax-and-transfer distribution of income (CBO 2018, p. 1).

Means-tested transfers are cash payments and in-kind benefits from federal, state, and local govern­ments that are designed to provide assistance to individ­uals and families with low income and few assets. They include benefits from government assistance programs such as Medicaid and the Children’s Health Insurance Program (CHIP), the Supplemental Nutrition Assistance Program (SNAP, formerly known as the Food Stamp program), and Supplemental Security Income (SSI). (CBO 2018, p. 1)

Social insurance benefits, which are financed by payroll taxes and other means, are not means-tested and are established for purposes other than directly affecting the post-tax distribution of income. Of course, they do indirectly affect the distribution of both pretax and post-tax income, as do many other government policies. In their absence, the incomes of the unemployed, of injured workers, of the elderly would certainly be significantly different, as would income inequality.

In Table 5, Mankiw treats all transfers as part of the tax system and therefore as negative taxes, while excluding them from his measure of income. Compared with the method used by the Congressional Budget Office, Mankiw’s choices give a different impression of how vertical equity is implemented in the US tax and transfer system.

What would Table 5 look like using the CBO’s methods, defining income more broadly and considering only means tested transfers as negative taxes? The result is shown in the fifth column in Table 12.1 (federal taxes less means-tested transfers as a percent of income). The corresponding values in Mankiw’s table are given in the last column for comparison.

How transfers should be treated is open for debate. If you agree with Mankiw that all transfers should be included as negative taxes, isn’t the CBO’s broader definition of income still a better indicator of the ability to pay than market income? The design of the tax system reflects this view; most cash transfers included in social insurance income are included in taxable income. The second-to-last column in Table 12.1 shows federal taxes less total transfers as a percent of income. It’s clear that Mankiw’s selection of market income, which he does not attempt to justify, portrays the tax and transfer system’s redistribution towards the bottom three quintiles in the most generous light. Would this tend to influence students’ judgements about the fairness of the tax and transfer system?

 Table 12.1: Alternative calculations of the incidence of federal taxes and transfers, 2014

Quintile Average Household Market Income Average Household Income (CBO definition) Federal Taxes, % of Income Federal Taxes less Means-Tested Transfers, % of Income Federal Taxes less all Transfers, % of Income Federal Taxes less all Transfers, % of Household Market Income (Mankiw’s measure)
Lowest $14,800 $19,200 2.1 -62.0 -84.4 -109.5
Second $30,600 $42,100 9.0 -5.7 -33.0 -45.4
Middle $56,400 $68,700 14.0 9.3 -8.4 -10.3
Fourth $92,200 $104,500 17.8 16.1 4.4 5.0
Highest $270,900 $281,400 26.7 26.3 22.5 23.4
Top 1% $1,764,200 $1,773,600 33.6 33.5 33.0 33.2

Source: Mankiw Table 5, p. 236; CBO (2016) and supplemental file and authors calculations. Household income is adjusted for household size to create the income groupings.


 (iii) A progressive tax and transfer system and redistribution to reduce income inequality

This point may belong in the Commentary for Chapter 20 on income inequality, but it’s worth noting here. In their commentary on the US tax and transfer system (as measured by the Congressional Budget Office calculations described earlier), William Gale and Zachary Obstfeld (2019) remark:

Relative to other countries, the U.S. has a fairly progressive tax system but less than average redistribution. This is because the U.S. generates far less revenue, 26% of GDP, than the other OECD countries who collect an average of 33% of GDP. Taken together, U.S. taxes and transfers do less to redistribute income than 26 out of 30 OECD countries.

A judgement about the fairness of a tax and transfer system must surely include the fairness of the post-tax-and-transfer income distribution that it produces. The information in Table 12.1 does not tell us anything directly about that. The Commentary for Chapter 20 (Income Inequality and Poverty) will show evidence to support Gale and Obstfeld’s claim.

  1. Taxing high incomes

Mankiw summarizes the responses of a group of prominent economists in 2019 to a survey about whether they agree or disagree with this statement: “Raising the top federal marginal tax on earned personal income to 70% (and holding the rest of the current tax code, including the top bracket definition, fixed) would raise substantially more revenue (federal and state, combined) without lowering economic activity.” Here, ‘earned income’ broadly refers to wages, salaries, bonuses, and income from self-employment, but excludes income from government transfers and (crucially) income from investments – interest, dividends, capital gains. (See the detailed responses here.)

This survey question was likely prompted by proposals around that time to raise the marginal tax rate on high-income recipients. For example, also in 2019, two prominent economists, Emmanuel Saez and Gabriel Zucman, published a book proposing a 75 percent top marginal rate combining federal and state income taxes and the corporation income tax. However, they specified that “[e]very income source should be subject to the progressive individual income tax”. For example, “[t]here is no compelling reason to tax [inflation-adjusted, or real] capital gains less than other income sources. The practice merely encourages the wealthy to reclassify their labor income and business profits into capital gains” (p. 138). The survey question, which proposes to vastly increase the gap between the top tax rate on labour income and capital income, would greatly incentivize this kind of tax avoidance.

The survey question also ignores Saez and Zucman’s warning that “hiking top tax rates without any other change to the tax code or to enforcement would be a bad idea. The supply of tax dodges in circulation is too large. Before we can effectively tax the wealthy more, avoidance must be curtailed” (p. 135).

Nor did Saez and Zucman claim there would be no effect on economic activity (as this survey question specifies), merely that the effect on taxable income would be “generally quite low” (p. 133). As will be seen in Chapter 18, which considers people’s decisions about time spent working, a higher tax rate could even induce people to work longer hours. In any case, it’s really a question of whether the result is an acceptable trade-off between greater equity and a reduction in total incomes.

It’s hard not to conclude that the survey question posed to the experts was been worded to encourage disagreement with raising this tax rate, thus giving the impression that expert consensus is against high marginal tax rates on top incomes. Why then does Mankiw choose to summarize results of this loaded question and to present it without discussion?

  1. What’s missing?

(i) Why is there a lower tax rate on capital gains compared with ordinary income?

Mankiw doesn’t mention the preferential treatment given to realized capital gains compared with ordinary income. This contrasts with his explanation for a lower tax rate on dividend income (p. 229).

A variety of tax rates apply to realized capital gains depending on the taxpayer’s income, ranging from zero, to 15 percent or 20 percent. (Up to $500,000 of capital gains on the sale of a principal residence can be entirely exempt from tax.) In contrast, the top marginal personal income tax rate is 37 percent (as of 2023). Why the lower rate? In their public finance text, Stiglitz and Rosengard (2015, p. 705) explain that preferential treatment is “largely a result of political pressures by wealthy individuals who receive much of their income in the form of capital gains.” Incidentally, this preferential treatment of capital gains income in the personal income tax is not unique to the United States. In 14 other OECD countries, capital gains are not even taxed at all (Harding 2013, Tables 9 & 10).

The US tax system does not adjust capital gains for inflation. If inflation is fairly high and asset prices are just keeping up with inflation, people may be taxed on capital ‘gains’ which are actually losses in real terms. However, as Stiglitz and Rosengard write, the lower tax rate is “a very imperfect substitute for indexing: especially given the low inflation rates today, assets held for only a short period have almost no inflationary component, and thus the much lower rate at which they are taxed cannot be justified by inflation alone” (2015, p. 661). In their judgement, the benefits to taxpayers of postponing taxes until the realization of capital gains has more than offset the disadvantage of the lack of indexation for inflation.

(ii) A discussion of a wealth tax

There is no wealth tax in the United States, but the idea has been debated in US politics since the 2016 Presidential campaign, when it was advocated by Senators Elizabeth Warren and Bernie Sanders. The idea has also been put forward by prominent economists such as Emmanuel Saez and Gabriel Zucman (For example, see Saez and Zucman 2019, pp. 172-75). Mankiw himself has written critically about the idea (Mankiw 2021) and has participated in a debate on the issue.

On the other hand, perhaps its absence is not surprising. There is no discussion anywhere in the text of the concept of wealth nor of its highly unequal distribution in the United States. (See the Commentary on Chapter 20 for information on wealth inequality in the United States.)

Indeed, the terminology Mankiw (and many other writers) use invites confusion between income and wealth. In the discussion of income and the income tax, Mankiw repeatedly uses phrases such as “the richest 1 percent” or “the richest Americans”. However, the words rich and poor refer to individuals’ wealth, not their incomes. For example, it’s possible for some people to have high incomes but low wealth if they spend much of their income and save little and/or borrow a lot. Similarly, Mankiw writes (p. 236) that the “debate over tax policy often concerns whether the wealthy pay their fair share”, when he likely means those with the highest incomes.


Causa, Orsetta and Mikkel Hermansen (2017) Income Redistribution through Taxes and Transfers across OECD countries, OECD Economics Department Working Papers No. 1453, available here.

Congressional Budget Office (2018) The Distribution of Household Income, 2014. Washington DC. Available here with supplementary data files.

Internal Revenue Service (2022) Topic No. 409 Capital Gains and Capital Losses, available here.

Gale, William and Zachary Obstfeld (2019) “Are Taxes (And Also Spending) Progressive?”, available here.

Harding, Michelle (2013) ‘Taxation of Dividend, Interest, and Capital Gain Income’, OECD Taxation Working Papers No. 19, available here.

Mankiw, N. Gregory (2021) ‘How to Increase Taxes on the Rich (If You Must)’, pp. 137-40 in Blanchard, Olivier and Dani Rodrik, editors (2021) Combating Inequality: Rethinking Governments Role, MIT Press.

Musgrave, Richard A. and Peggy B. Musgrave (1989) Public Finance in Theory and Practice, fifth edition, McGraw-Hill.

Saez, Emmanuel and Gabriel Zucman (2019) The Triumph of Injustice: How the Rich Dodge Taxes and How to Make Them Pay, WW Norton.

Stiglitz, Joseph and Jay Rosengard (2015) Economics of the Public Sector, WW Norton.

Respond to this commentary

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Please note that your email address will not be published.