Punitive Marginal Tax Rates and A Partial Appeal to The Economics Literature

Alexandria Ocasio-Cortez hit headlines last week for advocating marginal income tax rates “as high as 60% or 70%” on those earning $10 million plus per year. Under her plan, revenues from such a policy would be put towards funding a “Green New Deal.”

Matt Yglesias, Paul Krugman and Noah Smith were quick out of the blocks to defend the idea of massive marginal tax hikes on high earners as simply sensible, mainstream economics. They appealed to the work of economists Peter Diamond, Emmanuel Saez, Thomas Piketty and others, who have set out the case for very high marginal tax rates on top incomes in academic journals over the last two decades.

These economists have indeed recommended the optimal marginal tax rate for the top 1% of income earners in the U.S. should be a combined (federal, state and local taxes) rate of 73 percent or higher – designed with the aim of maximizing revenue from top taxpayers.

But their recommendation is not analogous to jacking up marginal federal income tax rates on very high earners in our current code. Furthermore, their result depends on highly contentious philosophical positions and economic assumptions.

A question of philosophy, not economics

Where does their type of result come from? The most important driver is your view on the role of government and redistribution.

Diamond, Saez and others think government can assess the utility that different income groups obtain from keeping more of their own money. They think the government can then aggregate the usefulness of income for different groups to develop an overall “social welfare function,” aiming to allocate our incomes to maximize the welfare of society.

Assuming the very rich do not find additional income very useful, they argue we should attach zero weight to the welfare of the rich when setting tax policy. The only thing that matters is setting a rate to maximize revenues from the wealthy to redistribute to those further down the income scale.

Now, one could challenge this on practical grounds (would a Green New Deal really transfer resources to the poor? Are the rich only useful for their tax dollars?) But more on that later.

The key point is that the redistributive tastes of the economists overwhelmingly drive their result. Given the actual current tax system is nothing like their ideal, such tastes do not seem to reflect the preferences of the public.

Don’t take my word for it. Emmanuel Saez himself, in an older paper on this issue co-authored with Jonathan Gruber, set out different redistributive tastes governments might hold. These included:

1. A Rawlsian view – where the government cares only about the poorest members of society (a policy of maximizing total revenue to redistribute)
2. A Progressive Liberal view – where the government assumes the social weight we put on individuals declines as income rises, right down to zero for those at the top (a policy of maximizing revenue from the very rich)
3. A Conservative utilitarian view – treating the rich and middle-classes with equal social weight, but those with very low incomes as in need of extra assistance (a policy of more limited redistribution)
4. No redistribution at all (a policy designed to raise revenue to maximize efficiency with no concern for equity).

Gruber and Saez calculated the optimal marginal tax rates for each of these agendas presuming government could design a new tax code to raise the average level of revenue collected in the 1980s, given their own calculations about the responsiveness to taxes of different income groups.

The result for the optimal marginal tax rate on the rich (those earning $100k and above) was indeed 73 percent for the Rawlsian and Progressive Liberal outlook. But for a Conservative utilitarian approach, it was just 30 percent. And for a government that did not want to redistribute at all, the optimal rate would be just 3 percent.

What does the 73 percent optimal tax rate result really mean?

Thinking a top 73 percent marginal tax rate optimal then is overwhelmingly driven by your philosophical priors. As Greg Mankiw has previously intimated, it’s not clear that ordinary people share the view of the rich held by progressive liberals.

But, importantly, it is also a bait and switch for Krugman and Smith to use this result as implicit support for 73 percent marginal income tax rates being added to today’s tax code as proposed by Ocasio-Cortez.

Jonathan Gruber and Emmanuel Saez’s paper used data from the 1980s tax reform to estimate the responsiveness of different broad income groups to changes in tax rates. These calculated elasticity figures (which showed the rich more responsive than others to changes in tax rates) were then plugged into the social welfare functions described above to estimate what optimal tax rates should be according to a government’s redistributive objective.

But the results represent the optimal marginal tax rate if we had just a single tax on all income to replace existing taxes. This is very different from adding a top new rate of 73 percent rate for the federal income tax, as Ocasio-Cortez appeared to endorse. The 73 percent result assumes that in the new tax system “the social planner is free to reshape the tax system and remove all the deductions and exemptions embodied in the current law.” This would make it more difficult for people to tax plan or avoid high rates by changing the timing of charitable donations and realized capital gains, for example.

Helpfully, Gruber and Saez set out what the optimal total tax rate would be if all existing deductions and exemptions were assumed sacrosanct because they were politically difficult to abolish. In that case, the revenue maximizing, optimal marginal total tax rate (even under a progressive worldview) would be just 49 percent. This is only slightly above the 45 percent combined top marginal rate they observed the US tax system actually delivered at that time.

In fact, Saez and Gruber’s calculations, finding that the rich (and particularly high-income itemizers) are much more responsive to tax changes than the middle-class or the poor, imply that marginal tax rates on the highest income groups should be lower than those faced further down the income scale. The main policy implication of the Saez-Gruber work is that tax rates on all groups earning gross income below $100,000 should be jacked up to fund more redistribution to the poorest, if you’re a good progressive. Good luck to Miss Ocasio-Cortez making that argument!

What about the more recent paper by Peter Diamond and Emmanuel Saez cited by Krugman? Here, the 73 percent optimal marginal tax figure for top earners (those earning over $300,000) comes as part of a recommended package where marginal tax rates rise with income and peak for highest earners.

This result is again driven by a progressive social welfare function, but the optimal rising marginal rates result comes from the economists assuming a much weaker responsiveness of taxable income to changes in the tax rate than Saez’s earlier work. AEI economists have previously shown that the assumption used by Diamond and Saez of an elasticity of just 0.25 is far too low relative to other literature. But Diamond and Saez wave this concern away by saying that, ideally, governments can reform the tax code to minimize tax avoidance.

Making this heroic assumption, again, makes a big difference to the results. Diamond and Saez acknowledge that if they took the current tax system as given, with all its deductions and exemptions and assuming that state, local and payroll taxes were fixed, then the revenue-maximizing total marginal tax rate would be 54 percent (were top taxpayers as responsive as Saez previously believed).

This would mean something like a 48 percent top federal marginal income tax rate – certainly higher than the 37 percent top income tax rate we see for 2019, but way, way lower than the idea of tacking on a 73 percent rate for earners of $10 million plus proposed by Ocasio-Cortez.

It’s true that some other work – particularly that of Piketty, Saez and Stantcheva – have similarly recommended top marginal tax rates of between 71 percent and 83 percent. But their results use elasticities of tax responsiveness for the whole population, not just higher earners. Their "optimal" results depend on the U.S. government essentially eliminating the ability to tax plan through fundamental reform, including simultaneous huge hikes to capital gains and taxes on corporations too.  And they postulate that top pay for the very rich essentially just arises through socially-wasteful rent-seeking, meaning it doesn’t matter if the activity is discouraged.

Strangely, none of Krugman, Yglesias or Smith highlight that these economists’ calculated optimal rates assume that fundamental tax reform would eliminate almost all deductions and exemptions. This is one of the reasons why using Diamond-Saez’s work to back up Ocasio-Cortez’s idea while also comparing her proposed rate to tax rates in 1950s is so misleading (deductions were numerous back then, making the gap between statutory and effective rates huge.)

The U.K.: a case study

As an aside on the elasticity point, the U.K. has in recent years undertaken an experiment on first hiking its top rate of income tax from 40 percent to 50 percent, and then lowering it from 50 percent to 45 percent. The government’s rationale for the latter move, in the face of strong pressure, was that the elasticity of taxable income to the net-of-tax rate was somewhere between 0.4 and 0.7. This is substantially higher than Diamond and Saez’s 0.25. As such, the U.K. government believed that cutting the rate would barely lose revenue, but would be good for the rich themselves, and for broader economic health.

What happened? As I wrote in January 2014:

Cutting the 50p rate to 45p, as implemented by George Osborne, was only estimated to reduce the exchequer revenues by around £100 million after behavioral effects, including steps to avoid the tax, had been considered. Early indications after the tax was implemented suggested that the behavioral effects might be more significant still. In pure revenue terms, HMRC figures show that in 2011/12 and 2012/13 the amount collected from top income taxpayers was £41.3 billion and £41.6 billion respectively under the 50p rate before jumping to £49.4 billion in 2013/14, when the top rate was cut to 45p.

Of course, much of this may have been due to forestalling of income and other activities based on knowledge of the planned rate changes – but at the very least the ease with which those top rate taxpayers were able to rearrange their tax affairs should put significant doubt in the minds of those who believe that a permanent rate would lead to significant extra revenues.

What are the rich good for?

Perhaps the biggest problem with the analysis of Krugman and others though is that it views the responses of the rich to tax rates in a very static sense. Results are largely driven by how useful we consider current, existing income to different groups. Little thought is put into the long-term incentives to earn income in the first place. Yet tax rates could, on the margin, affect people’s decisions to invest in human capital or undertake the development of new ideas.

After all, income later in life is one “reward” or payoff for hard work or taking risks through entrepreneurial activity. It stands to reason that hiking top tax rates reduces the financial payoff to such activity and so may deter it. We know that superstar inventors are very responsive to tax rates in terms of their location decisions, as are star scientists. Diamond and Saez themselves acknowledge too that the long-term elasticity of income to net-of-tax rates could well be higher than they envisage because of deterring human capital accumulation.

Yet in Paul Krugman’s column, he implies that the usefulness of the rich to the poor is purely the tax revenue the former provide to be redistributed through government. As John Cochrane notes, this completely ignores the question of how people get rich in a market economy: by providing goods and services people want and need, and hence generating consumer surplus. If on the margin high tax rates deter a potential entrepreneur from deciding to set up the next Amazon, the loss to social welfare would be huge.

Krugman and Piketty seek to diminish this potential effect by looking at broad economy-wide growth rates historically under different tax rate regimes. Growth was good in the 1950s, they say, so high tax rates are evidently not that damaging. Sure, taxes are not the be-all and end-all. But, as noted by Magness, there was a huge difference between statutory tax rates and effective tax rates in the 1950s. Again, one cannot on the one hand claim that the 1950s shows high tax rates were fine for growth while also appealing to Diamond-Saez’s work which recommends eliminating the deductions which existed in the 50s.

Besides, there is another body of work – not least papers by Karel Mertens – that finds top marginal income tax rates *do* matter for GDP growth. And as Charles Jones has noted, if we accept new ideas drive economic growth and acknowledge after-tax income is a financial reward for innovation, then the optimal tax rate would be much, much lower than Saez suggests (Jones estimates 28 percent), precisely because we all benefit from the better products and higher GDP that result.   

Conclusion

As I hope this piece has demonstrated then, the results of Diamond-Saez’s work:

a)   Are dependent on a progressive worldview that is seemingly rejected through the revealed preference of voters

b)   Are predicated on a wholesale tax reform including the elimination of deductions, exemptions and opportunities for avoidance (unlike when the U.S. previously had high marginal rates)

c)   Are dependent on the assumption of less responsiveness of high-income individuals to tax rates than found in most studies

d)   Ignore the potential impact that high tax rates might have on future human capital accumulation or entrepreneurial activity

Krugman, Yglesias and Smith could use the Diamond-Saez work to say “there’s a progressive case for major tax reform, including high tax rates across the distribution, eliminating all deduction and very high rates on top earners.” Alternatively, they could say “there’s a progressive case for modestly higher top tax rates within the current code.” But they cannot claim simultaneously Ocasio-Cortez’s ideas merely echo the 1950s *and* reflect the work of Diamond-Saez.

Perhaps more importantly, they cannot claim those of us with different philosophical views, and hence different preferences for redistribution, are somehow ignorant of economics.