VoxEU Column Taxation

A welfarist role for non-welfarist rules

Under a welfarist approach, tax policy is judged on its implications for the well-being of those in the society to which it applies. An implicit vulnerability of this approach is that judgements are based on necessarily incomplete cost and benefit calculations. This column investigates people’s preferences for welfarist and non-welfarist approaches by exploring responses to envy. A narrow majority of respondents reject a redistribution of resources that raises overall welfare by assuaging envy. These respondents seem to be using non-welfarist principles to encode concerns about indirect policy consequences.

The technocracy is under attack, and economists are paradigmatic technocrats. Our starting point in any policy discussion is to identify and estimate the costs and benefits of an option, basing our recommendations (if we make them) on what the data tell us through the lens of our well-specified models. This focus on facts and quantifiable effects is one of our great strengths, for it excludes fanciful thinking and untestable logic, forcing political arguments to confront economic reality.

But some of our own leading lights have warned of a vulnerability in our technocratic approach – our cost and benefit calculations are inescapably incomplete. Friedrich Hayek (1973) wrote that the "irremediable ignorance of most of the particular facts which determine the processes of society" prevents us from anticipating "all the costs of achieving particular results." In this warning, Hayek was echoing a long line of thinkers tracing back to Edmund Burke, David Hume, and Adam Smith. A similar limitation affects our personal moral judgments, as philosophers since at least Mill (1871) have pointed out. Even if I wish to take the best action in a given situation, knowing what the best action is may be often (even, most of the time) impossible.

What can we do in the face of our ignorance as policymakers or individuals?

The answer given by the thinkers just cited, and many more, is to learn from the past. Societies adopt principles that complement cost-benefit calculations in decision-making. These principles come from a (perhaps unknowable) process of societal learning from experience, passed down across generations and taught in simple ways that are memorable and clear. John Harsanyi (1992) makes this point in the context of someone trying to act so as to maximise overall welfare – "rule utilitarianism is free to choose a moral code that judges the moral value of individual actions partly in terms of nonconsequentialist criteria if use of such criteria increases social utility". Perhaps the simplest example is the stricture do not lie. Lying often, in fact, seems like precisely the right thing to do from a narrow cost-benefit standpoint. But we hesitate to lie, and we teach our children to hesitate to lie, because we know a society – or an individual – that condones routine lying ends up worse off.

Tax economists may have a claim to be the most technocratic of all. The reigning Mirrlees (1971) model of tax theory assumes what is called a ‘welfarist’ objective function in which the appeal of a tax policy is judged by its implications for the well-being of those in the society to which it applies. In other words, direct cost-benefit calculations are all that matter for determining whether a tax policy is a good idea or not. The case for this approach is compelling to many – after all, why would it make sense to include anything other than direct cost-benefit calculations? As Louis Kaplow and Steven Shavell (2001) have famously shown in this context, "any non-welfarist method of policy assessment violates the Pareto principle". Liam Murphy and Thomas Nagel (2002) have forcefully supported this logic.

In a new paper, I point out that this current practice in the economics of taxation is particularly vulnerable to Hayek’s critique, as no analysis – no matter how comprehensive – can ever hope to capture the full set of costs and benefits associated with tax policy (Weinzierl 2017). In particular, we simply do not know how to model the effects of tax policy on the long-run norms and structure of the economy. We struggle quite a bit, in fact, to understand even the short-term effects of tax policy in partial equilibrium. Our optimal tax models strike many as complex, and optimal tax theorists do their best to capture the most important factors in the analysis, but we are far from omniscient.

In the face of these limits to our knowledge, I suggest that historically prominent non-welfarist principles for taxation, such as the Equal Sacrifice principle advocated by Mill (1871) or the Classical Benefit-Based Taxation principle favoured by Smith (1776), are just the sort of rules passed down by society that capture some of the unforeseeable consequences of taxation. And I note that greater public support for these principles than for conventional objectives assumed in optimal tax research, as documented in some of my previous work (Weinzierl 2014, 2016a, 2016b), is consistent with these principles, carrying with them some accumulated wisdom of society.

This all may seem rather abstract, so I turn now towards something more familiar – envy. The proper policy response to the existence of envy has been an age-old problem for utilitarian tax theorists. If envy of the rich makes others suffer, high incomes act like sources of negative externalities. According to a strict cost-benefit calculation, optimal taxation is therefore more progressive in the presence of envy (Boskin and Sheshinski 1978, Oswald 1995). But most people – even most utilitarians – are uncomfortable with the idea of taxing some to assuage the envy of others. What explains this judgement, contrary to the welfarist logic?

I surveyed several hundred Americans on Amazon’s MTurk platform, presenting them with two hypothetical situations in which this tension over envy is made real, as shown in Figure 1.

Figure 1 Hypothetical situations

Note that in Situation 2, both Person F and Person G have more money, though the gap between them has increased. Person G’s well-being is also greater in Situation 2, but Person F’s well-being is lower. Importantly, Person F’s well-being falls more than Person G’s well-being rises when we move from Situation 1 to Situation 2, so Situation 1 would be preferred under a welfarist objective. I then asked respondents to take the perspective of objective observers, as if they were policymakers, and answer the question: “Which situation do you think is better?” In response, 54% chose Situation 2, despite its lower levels of overall well-being and for the worse off person.

In other words, I found that a majority of respondents reject a redistribution of resources that raises overall welfare by assuaging envy, contrary to the conventional welfarist analysis. Moreover, respondents were more likely to reject such a redistribution if I explicitly mentioned the role of envy in the situations and if they support libertarianism – a philosophy that doubts the sufficiency of welfarist cost-benefit calculations more generally. Based on these patterns, it seems respondents were using non-welfarist principles to encode concerns about indirect welfare consequences of policy. In particular, neither Equal Sacrifice nor Classical Benefit-based Taxation would countenance redistribution to assuage envy, so respondents’ support for these principles is consistent with their preferences over the hypothetical situations in my survey.

The lesson I take from these findings is not that technocratic analysis is the wrong path. Rather, truly sophisticated technocratic analysis must take into account its own limitations. Given our limited knowledge, as individuals and as policymakers, it is wise that we seek any supplement to our direct cost-benefit calculations. rules passed down by our societies are certainly not to be trusted blindly – history is full of sometimes horribly mistaken beliefs and practices – but they may carry wisdom that will help us do our jobs, as proud technocrats, a bit better.


Boskin and Sheshinski (1978),"Optimal income redistribution when individual welfare depends upon relative income", Quarterly Journal of Economics: 589-601.

Burke, E (1790), Reflections on the Revolution in France, Penguin.

Harsanyi, J C (1992), "Game and decision theoretic models in ethics", chapter 19 in R J Aumann and S Hart (eds), Handbook of Game Theory, Elsevier.

Hayek, F A (1960), The Constitution of Liberty, Chicago.

Hayek, F A (1973), Law, Legislation, and Liberty, Chicago.

Hume, D (1738) A Treatise of Human Nature, Oxford (2007).

Kaplow, L and S Shavell (2001), "Any non-welfarist method of policy assessment violates the Pareto principle", Journal of Political Economy 109(2).

Mill, J S (1871), Principles of Political Economy, Oxford University Press (1994).

Mirrlees, J A (1971), "An exploration in the theory of optimal income taxation", Review of Economic Studies 38: 175-208.

Murphy, L and T Nagel (2002), The Myth of Ownership, Oxford.

Oswald, A (1983), "Altruism, jealousy, and the theory of optimal non-linear taxation", Journal of Public Economics 20: 77-87.

Smith, A (1759), Theory of Moral Sentiments.

Weinzierl, M (2014), "The promise of positive optimal taxation: Normative diversity and a role for equal sacrifice", Journal of Public Economics 118: 128-142.

Weinzierl, M (2016a), "Revisiting the classical view of benefit based taxation", Economic Journal (forthcoming).

Weinzierl, M (2016b), "Popular acceptance of inequality due to brute luck and support for classical benefit-based taxation", NBER Working paper 22462.

Weinzierl, M (2017), “A welfarist role for non-welfarist rules”, NBER Working paper 23587.

2,003 Reads