VoxEU Column Poverty and Income Inequality

Capital taxation in the 21st century

Piketty's justification for his proposed wealth tax relies on the notion that the rate of return on capital exceeds economic growth. This column challenges this basis, arguing that it fails to account for risk. The authors also examine the relative merits of a consumption tax, which may be more valid.

In Capital in the Twenty-First Century (Piketty 2014) Thomas Piketty presents a compelling story of the ascendency of capital and the powerlessness of the market forces to arrest the growing threat to democracy from growing wealth – and its increasing concentration in the hands of the few. Piketty prescribes a global, coordinated wealth tax as the antidote to this dystopian trend, arguing that only such a direct assault on wealth concentration will succeed where the other policies of governments – that already play large roles – in their respective economies have failed.

We take issue with Piketty’s facts, logic, and policy conclusions.  The support for the view that there will be an inevitable rise in capital’s share is lacking.  In addition, even if the world were as Piketty describes, a wealth tax finds little or no support either in Piketty’s own work or elsewhere in the literature.

A note on data and data sources

A number of authors have raised question concerning the data analysis that underpins Piketty's dystopian view.  We illustrate these concerns with an in-depth investigation of recent US data.  As discussed in our paper (Auerbach and Hassett 2015) – and as is evident from the comparison in Figure 1 – in translating data from other sources on US wealth concentration, Piketty manipulates the timing of observations in a manner that overstates the strength and smoothness of the recent upward trend, and obscures a downtrend present in his source data at the end of the sample period.1

Figure 1. Share of wealth, top 1%

r vs. g

In Capital, the relationship between the return to capital r and the economic growth rate g is presented as a fundamental relationship determining our economic path. The basic syllogism is

  • r exceeds g,
  • saving generated by this high rate of return causes capital and wealth to grow faster than the economy, and
  • capital income grows as a share of income because the rate of return does not fall sufficiently fast with capital deepening. 

But how should one measure r? Absent fundamental risk or capital taxation, all rates of return should be equal. But with market risk and risk-aversion, the expected return will include a risk premium, so that markets will outperform safe investments on average but will do worse in bad states of nature where resources are particularly highly valued. Thus excluding returns in bad states is certainly not the way to measure r, and even including all returns misses the point that one shouldn’t weigh good and bad outcomes equally.

Further, the rate of return to investors relevant to Piketty’s comparisons to the rate of economic growth is the after-tax return. Capital income taxed away by the government does not accumulate in the hands of the few.

Piketty’s time series on after-tax rates of return faces two shortcomings.

  • First, the tax rates used to calculate the after-tax returns on capital in Piketty and Zucman (2014) are average tax rates. But given the book’s focus on the share of wealth held by the top 1% and even the top 0.1%, the relevant tax policy parameter seems to be the top marginal tax rate rather than the average tax rate.
  • Second, Piketty calculates the return on capital based on national accounts data and does not adjust for risk.

These decisions are crucial to the analysis.

In our paper, we construct an alternative time-series on the after-tax rates of return in the US that takes into account these two criticisms. To derive the after-tax rate of return from the pre-tax rate of return, we simulated using NBER’s TAXSIM the increase in tax liability that would result from an increase of 1% in the interest income accruing to the top 1% of income earners in each year, and then calculated the effective tax rate that would be paid on the average dollar of that additional income.  To proxy for the risk-free pre-tax rate of return on capital, we use historical time-series data on ten-year treasury bond yields.

Figure 2 graphs both this alternative series and the series on pre- and post-tax returns on capital from Piketty and Zucman (2014). As one can see, the pre-tax rate of return using the alternative specification is substantially lower than the national accounts based series from Piketty and Zucman (2014). Likewise, the post-tax rate of return using the top marginal tax rate is substantially lower than their post-tax rate. Indeed, the alternative post-tax rate of return remains consistently lower than GNP growth. From this perspective, the apocalyptic r > g 'exploding wealth inequality' scenario does not look especially likely.

Figure 2. Four Different Computations of "r"

Global wealth taxation and its alternatives

Piketty (pp. 528-530) argues passionately for a substantial annual wealth tax, with a progressive rate structure with marginal tax rates of two, five, or even 10% on the highest wealth levels. Putting forward such a proposal is truly a bold step, taking place as it does against a backdrop of intense international tax competition. It is a bold step, also, in terms of the logical support one can find for it elsewhere in the book or in the substantial literature on the design of optimal tax systems. Indeed, Piketty turns a key insight from this literature on its head.

An important development in the literature on taxation and saving is a fuller understanding of the relationships between different tax bases – in particular consumption taxes, labour income taxes, and capital income taxes. While simple models may equate taxes on labour and consumption, a major difference is the ability of consumption taxes to hit existing sources of wealth, attributable to rents, inheritances, disguised labour income, etc. This broader consumption tax base can provide a tax system that is more efficient (Auerbach and Kotlikoff 1987), and – with a progressive rate structure – a reasonably progressive distribution of the lifetime tax burden (Altig et al. 2001). As is now also well understood, a consumption tax differs from a capital income tax in its treatment of capital income only by its exemption of the safe rate of return on investment. Thus, consumption taxes hit wealth without interfering with the incentive to save associated with the intertemporal terms of trade. Wealth taxes, on the other hand, effectively tax the safe rate of return on investment because they do not depend on actual rates of return, thereby incurring the intertemporal distortion but forgoing tax on other components of the rate of return. Thus, consumption taxes reduce the value of wealth – just as wealth taxes do – so we see no basis to argue that wealth taxes have a greater capacity to limit the power that wealth provides.

A distinct question is how certain expenditures – for example political campaign contributions – should be taxed. But this question actually highlights a further potential advantage of a consumption tax relative to a wealth tax. A wealth tax reduces the purchasing power of wealth that is spent currently by the same amount, regardless of how the funds are used. Under a consumption tax, an additional level of tax could be added by defining political contributions to be consumption, so that both the act of contributing and the expenditures of contributed funds would be taxed. Thus, a consumption tax is potentially a more powerful tool.  In summary, tax policy can play an important role in addressing inequality, but we find little support for Piketty’s proposal, either in his own work or elsewhere in the literature.

Piketty’s book rests on three pillars:

  • the idea that there is a steady and inexorable upward trend in the share of wealth going to those at the top,
  • the idea that r>g induces a path wherein the wealthy gradually absorb all of economic output, and
  • the idea that a wealth tax is a logical response to these conditions. 

Upon careful scrutiny, each of these pillars falls.


Altig D, A Auerbach, L Kotlikoff, K Smetters and J Walliser (2001), “Simulating Fundamental Tax Reform in the United States,” American Economic Review 91 (3): 574-595.

Auerbach, A and K Hassett (2015), “Capital Taxation in the 21st Century,” NBER Working Paper 20871.

Auerbach, A and L Kotlikoff (1987), Dynamic Fiscal Policy, Cambridge University Press.

Bricker J, J Krimmel, A Henriques and J Sabelhaus (2014), "Measuring Income and Wealth at the Top Using Administrative and Survey Data", Presentation at the European Central Bank, November 18.

Kopczuk, W (2014), “What do We Know about the Evolution of Top Wealth Shares in the United States?” NBER Working Paper No. 20735.

Piketty, T and G Zucman (2014), “Capital is Back: Wealth-Income Ratios in Rich Countries, 1700-2010”, Quarterly Journal of Economics 129 (3): 1255-1310.

Saez, E and G Zucman (2014), “Wealth Inequality in the United States Since 1913: Evidence from Capitalized Income Data,” NBER Working Paper No. 20625.


1 This is not to suggest that there is no evidence in favor of an increasing recent share of wealth at the top. For example, Saez and Zucman (2014) find an increasing share from estimates based on capitalizing flows of capital income. Such a trend is much less evident when one uses alternative approaches based on estate tax returns or survey evidence.  The sources of these differences and a consideration of which approach might be most accurate are considered by Saez and Zucman (who favor the capitalized income method) as well as Kopczuk (2014) and Bricker et al. (2014), who argue in favor of the survey-based method of measurement.

4,409 Reads