Getting to grips with tax

Maria Racionero

Associate Professor Maria Racionero has been grappling with the theory of taxation for over a decade. As she tells Stephen Green, it is a field of enquiry that has exercised some of the big names in economic theory, and which continues to provide fruitful new avenues for her own research.

Tax may be proverbially one of the only certainties in life, but what is equally certain is that tax is never going to be perfect. Think of the familiar complaints: “this tax is unfair, it’s immoral; it puts too much burden on the poor;” or “the rich are getting away with it;” or “tax is too complicated – it’s too hard to understand”; or “there are too many taxes”. “Tax doesn’t have to be taxing”, proclaimed a recent advertising campaign by the UK Tax Office. Maybe not – for the man in the street. Yet the conundrum of trying to balance these many, often conflicting, challenges continues to tax some of the leading minds in economics. The likes of James Mirrlees, Joseph Stiglitz and George Akerlof, Nobel Laureates all, have played significant roles in the theoretical developments of the last few decades.

Associate Professor Maria Racionero of the Research School of Economics has continued in this distinguished tradition by making optimal taxation theory the central focus of her research throughout her career. This is an area of economics that seeks to define the best way to achieve the objectives of taxation – raising revenue for redistribution and government administration – whilst minimising distortions (the inevitable changes to individuals’ economic behaviour).

“The desirable characteristics of a tax system are usually listed as efficiency, equity, administrative simplicity and transparency, or political responsibility,” she explains. “The ones that have mostly concerned the optimal tax literature are efficiency and equity. How can you achieve a certain degree of equity at the least possible cost – that is, the least possible inefficiency cost.”

The giant in optimal tax theory is James Mirrlees, the Scottish economist who (with William Vickrey) won the Nobel Prize for Economics in 1996. As Maria explains, his 1971 paper, An Exploration in the Theory of Optimum Income Taxation, was a watershed in the history of the field, introducing the importance of asymmetric information in understanding taxation policy:

“Mirrlees’ contribution really marks out a before and after in optimal taxation. Before him, the models were often based on ad hoc restrictions on the tax instruments that could be used. There was no particular reason why there should be a linear income tax as opposed to a lump sum tax, which is the efficient tax by definition – where individuals cannot affect their liability by changing their behaviour. The inefficiency cost of taxes is due to the fact that an individual can alter their tax liability by changing their behaviour. In the case of income tax, they can work less. With a commodity tax, they can consume less. Mirrlees established that the absence of perfect information prevents us from using ideal taxes with no efficiency costs.”

Imperfect or asymmetric information describes the situation where one economic agent knows something the other does not – information is asymmetric because it is not equally shared. It has come to be such an important factor in understanding much of public economics because it concerns the inability of a policy-maker to know some of the factors governing an individual’s economic behaviour. It can only make judgments based on the incomplete information it has:

“In Mirrlees’ model, individuals differ in innate ability, which is private information. The government cannot observe the innate ability of individuals… What it does observe is the income. That is the government observes the combination of innate ability and the amount of effort an individual puts in but cannot distinguish between the two.”

Inevitably, income is an imperfect guide to the innate ability, or indeed other characteristics of the individual that might affect his or her economic behaviour. This fact obviously limits the extent to which the efficiency of any tax system can be maximised. At the same time, understanding how individuals of differing ability behave under certain tax conditions can help avoid overly distortionary policies. One of the main results of the literature, demonstrated in a 1982 paper by Joseph Stiglitz, is quite counter-intuitive. Using a model with just two types of individual: high and low ability; he showed that it is actually optimal to impose distortions by means of a positive marginal tax rate, on low ability types. In other words, tax the lower earners more for each additional dollar. As Maria points out, whilst on the face of it this looks like trying to raise revenue from the lower income sectors of the population, it is actually about continuing to raise revenue from the higher earners:

“The important result of this literature is that it allows you to explain apparently counter-intuitive results like [this]. It is not to raise revenue from them [the lower earners], but to raise revenue from the higher earners by inducing the individuals at the top of the income distribution not to mimic the low ability types.”

Maria’s own interaction with optimal taxation began early in her career with a paper drawing on her PhD thesis which expanded on the Mirrlees and Stiglitz models by adding an additional characteristic to the traditional one of innate ability – this was the individual’s preference, and in particular, preference for leisure. One of the great advantages of an economic model is that it can simplify reality to a sufficient extent to powerfully elucidate otherwise latent underlying phenomena. At the same time, it might exclude other important factors that could be similarly revealing. Thus any field of economic research builds by making adjustments to the original theory.

Inevitably, then, introducing a second dimension into an economic model brings with it a great increase in complexity, and not just in a technical sense, as Racionero explains:

“The Mirrlees and Stiglitz models are based on the fact that with a single characteristic, although you do not observe that characteristic, you do know how individuals behave when there is a particular condition which these fulfil. This makes the analysis much simpler. When you have two characteristics, this condition [known as the ‘Spence-Mirrlees’ or ‘single-crossing’ condition] no longer holds – you have many more cases to deal with. But there is an additional conceptual and philosophical issue when you look at preferences for leisure. Innate ability is considered as a variable that you would like to compensate people [of lower ability] for. With preferences for leisure there is more of a discussion: we don’t know if we should compensate people for their preferences, or hold them responsible for them. It’s a grey area.”

Many optimal taxation models have a broadly utilitarian social objective – ie, maximising the sum of individual utilities. As long as everyone has the same or similar preferences this works, but if individuals have different preferences you can run into problems:

“If you add the utilities of individuals with different preferences, whilst you will tend to redistribute from high to low ability types, at the same time you could be redistributing from people who like working to people who like leisure.”

Plenty has been written on this issue in the last couple of decades, but Maria and her co-authors did not take a particular view in their paper, and worked within the utilitarian framework, but gave different weighting to different preferences. One of the primary results from this research ran contrary to previous literature and was again somewhat counter-intuitive:

“For certain weights we showed that it could be optimal to impose subsidies at the top of the distribution, where all the previous results in the literature said that it had to be zero at the top – it could not be negative.’

This paper, co-authored by Robin Boadway, Maurice Marchand and Pierre Pestieau, and published in the Journal of Public Economic Theory in 2002, has proved to be Maria’s most cited and has been a jumping off point for further research, not least by Maria herself. In a subsequent paper she examined preferences in greater depth – looking at the underlying reasons for preferences, again in the context of imperfect information, for example:

“People may have a high preference for leisure but for some this may be because of a disability, for others it may be just laziness, but they are indistinguishable because they display the same preference. So we tried to see whether the observation of some commodity consumption patterns – for instance health related goods - would allow us to distinguish those with the unobservable disability and so provide a rationale for commodity taxation.”

In all optimal taxation models, the key element is that asymmetry of information limits the extent of redistribution that can be achieved. One of the means used to tackle this problem and try to extend the area of distribution is a mechanism known as ‘tagging’, where an observable characteristic can be used as a proxy for the unobservable, underlying variable that could define a group’s tax status. Much of the time, the observable variable has no normative value in itself, it just happens that there is a useful correlation. However, sometimes the observable variable has relevance in itself. For example, in some recent research, Maria was using time needs as the distinguishing characteristic of different groups:

“If there is any correlation between time need and the distribution of ability, you may wish to use time needs for the information it conveys about the distribution of ability, but at the same time you may wish to compensate individuals for those time needs [for example, people who have longer to commute, or who have children or parents to care for]. However, you may not be able to do that completely because you are already using the information in another way.”

The consideration of time needs also brings into play questions of responsibility again: for example, although parents might need more time to care for their children, they also have a choice about how many children they have and it could be argued, have a degree of responsibility for how much of their time childcare demands. Having touched briefly on the question in this research, Maria was prompted to explore the question of time needs in more depth, and this has led to her current research project in partnership with fellow ANU economist and labour economics specialist Professor Alison Booth and Professor Pierre Pestieau of the University of Liege.

The project, which was awarded an Australian Research Council Discovery Grant earlier this year, aims to make use of elements of the labour economics literature and recent evidence about time use to develop a more sophisticated model of time allocation within the optimal taxation framework.

“The optimal tax literature has a very simple underlying model of the use of time: people either work, or have leisure – there is nothing else,” says Maria. “Working on tagging with time-needs made it apparent to me that there was a need to develop a more sophisticated treatment of time. Within labour economics there are models (for example the Becker model) where people have many more uses for their time, but this has not been embedded in the optimal tax literature. So that is how we came up with this project.”

The second more applied element of the project will focus on the specific area of parental time use, making use of the databases of the Australian Time Use and the Household, Income and Labour Dynamics in Australia (HILDA) surveys.

“What we want to do is comment on how people use time, in particular looking closely at parents’ use of time and disentangling the different aspects of time use,” says Maria. “We are trying to look further into how they employ time and distinguishing between those activities that are more akin to labour such as home production, cooking or minding the children; those that are more like leisure; and those that are more about human capital investment – such as reading or teaching. It’s not just a question of how people use time, but how this can inform childcare policy. A particular question has been highlighted recently with the parental leave scheme debate: should mothers or carers be given a flat rate, or should the rate depend on their earnings? These kinds of questions should be easier to answer by understanding how different parents, with different characteristics, particularly education, employ their time.”

Optimal taxation theory has been criticised because deriving clear-cut answers to policy questions from it can be difficult, although as Maria has pointed out, it does provide some useful overall insights. By applying theory in a more closely circumscribed applied area, as with the questions of childcare policy, researchers can make useful contributions to policy development. But, as Maria emphasises:

“It is always important to have the theoretical background. A theoretical model will tell us the things we have to care about – the main issues and the forces and parameters that we need to estimate. I believe in empirical work based on sound theoretical research. Sometimes you see empirical research which just says – look, this happens. But why? How do you explain what is happening? You always need both.”


[1] Mirrlees, James A, 1971. An Exploration in the Theory of Optimum Income Taxation, Review of Economic Studies, Wiley Blackwell, vol. 38(114), pages 175-208, April

[2] Stiglitz, Joseph E., 1982. Utilitarianism and horizontal equity : The case for random taxation, Journal of Public Economics, Elsevier, vol. 18(1), pages 1-33, June

[3] Robin Boadway & Maurice Marchand & Pierre Pestieau & María del Mar Racionero, 2002. Optimal Redistribution with Heterogeneous Preferences for Leisure, Journal of Public Economic Theory, Association for Public Economic Theory, vol. 4(4), pages 475-498, October

Updated:   14 October 2014 / Responsible Officer:  CBE Communications and Outreach / Page Contact:  College Web Team