In this article, Tony Atkinson (Nuffield College Oxford and LSE), a past president of the RES, reviews the work of the IFS and James Mirrlees published in 2012 under the title Tax by Design.
A major report on the reform of UK taxation, orchestrated by the Institute for Fiscal Studies (IFS), written by leading public finance economists, led by a Nobel Prize-winning Cambridge professor. For those with long memories, this sounds familiar. In 1978, James Meade and the Meade Committee produced The Structure and Reform of Direct Taxation. A third of a century later, the IFS has done it again, with a team led by James Mirrlees, which has produced the report, Tax by Design, and a set of commissioned studies published as Dimensions of Tax Design. Like the Meade Report, the Mirrlees Review is an impressive study and will be a standard reference for years to come.
Indeed, the Meade and Mirrlees reports have much in common. Both take a broad view of the issues and are firmly grounded in economic theory. Both review teams have blended the contributions of senior scholars with those of up-and-coming younger researchers. Both demonstrate the value in this politically-sensitive field of an independently financed inquiry. There are also important differences. One is scale. The report of the Meade Committee was contained in one volume; the Mirrlees Review has produced two substantial volumes with a total of 1,880 pages — nearly a quarter of the length of UK primary tax legislation! A second important difference arises from the fact that public economics has moved on. The Mirrlees review rests heavily on empirical findings, reflecting the revolution in public finance achieved through intensive use of micro-data and the development of micro-econometric techniques.
What does the Mirrlees review recommend?
The Review set out ‘to identify reforms that would make the tax system more efficient, while raising roughly the same amount of revenue … and while redistributing resources … to roughly the same degree. Our motivation [is] to unlock significant potential welfare gains’ (p.2). Their ‘vision of a good tax system’ to achieve this objective includes a progressive income tax, exempting the ‘normal return to savings’, with a coherent rate structure, a single integrated transfer system for those with low incomes or high needs, a largely uniform value-added tax (VAT), with additional taxes on alcohol, tobacco and road congestion, a lifetime wealth transfer tax, and a single rate of corporation tax, exempting the ‘normal return on investment’.
The Review’s ‘vision’ is presented as being of general validity. In contrast, the process of transition from the current state of taxation depends on the circumstances of the country in question. In the case of the UK, the Review is highly critical of the tax system, which it describes as ‘a jumble of tax rates, a lack of a coherent vision of the tax base, and arbitrary discrimination across different types of economic activities’ (pp. 478-9). The recommended ‘reform package’ for the UK includes (only an abbreviated summary):
- Merging personal income tax with social security contributions;
- Replacing income-tested transfers by a single integrated benefit;
- Exempting interest on bank and building society accounts, making a "rate-of-return allowance" for equities, business assets and rental property; and otherwise taxing capital income at the same rate as earned income; introducing an allowance for corporate equity into the corporation tax;
- Removing nearly all zero and reduced rates of VAT and introducing a tax on financial services equivalent to VAT;
- Introducing a tax on the current value of domestic property, and a land value tax for business, to replace existing council tax (on housing) and Stamp Duty, and business property tax.
Each of the Review proposals warrants careful attention, and they are already having an influence on Government policy. Here I make no attempt to go through them exhaustively; instead I have picked out just three aspects that are interesting in their own right and that raise questions about the present state of public economics.
Optimal taxation and VAT
Given the importance of the Chair’s role in the development of optimal tax theory, it is scarcely surprising that the Review is heavily influenced by this literature. The Review recognises that ‘optimal tax theory has its limitations’ but says that ‘it is nevertheless a powerful tool and, throughout this volume, the conclusions of optimal tax theory will inform the way in which we discuss policy’ (p. 39). One example is the proposal for a more broad-based VAT. Optimal tax theory is not the only strand in the argument for uniform rates of VAT. The Review attaches weight to the administrative advantages and the probability that a uniform tax would be less vulnerable to lobbying pressure — both of which seem to me strong arguments. But it also devotes considerable space to the optimal taxation analysis — which seems more open to debate.
The argument regarding a uniform VAT illustrates well one of the key principles of the Review, ‘the need to think of the tax system as just that — a system’ (p.45). If indirect taxes are considered in isolation, they appear to be a classic battleground between efficiency (tax more heavily goods that are inelastic in demand) and equity (exempt necessities). But once we introduce the possibility of levying direct taxes, the role of indirect taxes changes. If the only source of income differences is earning capacity, and there are no restrictions on the government’s ability to levy non-linear direct taxes on earnings, then — under certain conditions — optimality can be achieved with uniform indirect taxes. (The conditions include the absence of externalities and for this reason the Review recommends additional taxes on alcohol and tobacco.) Pragmatically, the Review devises a tax reform package that broadens the base for VAT, removing zero rating (notably for food) ‘to raise net revenue for the Exchequer and to redistribute more resources from better-off households to less well-off households’ (p. 217).
The optimal tax approach provides a rigorous framework within which to assess such proposals, operating with an explicit objective function and constraints defined by a fully-specified economic model. However, the validity of the argument depends crucially on the appropriateness of the underlying model. In the present case, this is open to question, since the model assumes perfect competition, whereas food retailing in the UK is highly concentrated: the top four supermarkets in the UK have a market share of over 75 per cent. These firms are unlikely to act as perfect competitors, and market structure may affect the incidence of the tax, a subject that receives surprisingly little attention in the Review. The Review assumes that the incidence of the VAT reform is fully on retail prices. Such an assumption is valid where there is perfect competition and constant costs of production, but ceases to be so when these conditions do not hold. A rise in the tax on food may be shifted backwards onto producers. Or oligopolistic supermarkets may raise their prices more than the tax. The non-economist readers of the Review might well wonder what will happen to their grocery bills, and whether the researchers have spoken to supermarket chains to see how they might react to the extension of VAT to food.
Does the existence of imperfect competition affect the optimal design of indirect taxation? At the time of the monopolistic competition revolution in the 1930s, Austin and Joan Robinson pointed out that the tendency for imperfectly competitive firms to charge more than marginal cost creates a distortion that can be corrected by a subsidy — i.e. taxing the good less. The situation is just like that of an externality, and in the same way as other externalities it modifies the conditions for an optimal tax. In the Mirrlees Review, the issue is mentioned only in a footnote (p. 156n), to be dismissed. In my view, this is too hasty. The important conclusion is that the design of VAT has to consider the degree of market competition, and, until this has been taken into account, we should remain agnostic about the strength of the optimal tax argument for extending VAT to food.
The controversial top tax rate
A second application of optimal tax arguments in the Review is to the progression of income tax and the top rate of income tax. This is a controversial topic. In 2010, late in the Labour Government, the top rate was raised from 40 to 50 per cent. The Coalition Government announced in March 2012 its intention to return to a lower top rate, arguing that the 50 per cent rate had raised little additional revenue. The official HM Revenue and Customs study of the revenue impact described its findings as consistent with those contained in the Mirrlees Review.
The Mirrlees Review does indeed say that ‘it is not clear whether the 50 per cent rate will raise any revenue at all’ (p. 109). This statement draws on the background chapter by Mike Brewer, Emmanuel Saez and Andrew Shephard, where they estimate that the elasticity of taxable income for the highest 1 per cent is 0.46. Combined with information about the shape of the upper tail of the distribution, this implies a revenue-maximising overall tax rate of 56.6 per cent. But the overall ‘effective tax rate’ includes National Insurance contributions and indirect taxes, so that the authors conclude that the income tax rate on its own should be no higher than 40 per cent.
There are, however, several reasons for caution regarding this conclusion. First, the Mirrlees Review stressed that the estimate is tentative: ‘there is no escaping the uncertainty around the estimate of a 40 per cent revenue-maximizing income tax rate’ (p. 109). The reported standard error for the estimated elasticity implies a 95 per cent confidence interval for the revenue-maximising tax rate from 46 to 74 per cent, which is a wide range. Secondly, we should consider the implications of the arithmetic that took the Mirrlees Review from a top tax rate of 56.6 per cent down to 40 per cent. What is being calculated here is the total tax wedge between £1 paid by the employer and £1 of goods consumed. Employer and employee social security contributions and value added tax at 20 per cent together bring the 56.6 per cent down to 39 per cent. But this, of course, means that the optimal income tax rate depends on what other taxes those affected are paying. If the marginal earnings come from self-employment, or people are paid via a company, the income tax rate should be not 39, but 46 per cent. Suppose that they spend half their extra income on zero-rated goods, like books or food? Then the optimal tax rate becomes 51 per cent.
Moreover, as in the case for taxing food, the Review's recommendations are affected by considerations absent from the underlying model. The estimate of the taxable elasticity is calculated assuming away any inter-dependencies between the incomes of different people. It is based on the changes in the incomes of those affected (say, the top 1 per cent) relative to the incomes of those not affected by the tax change. The hypothesis is that this group — the bottom 99 per cent — earn the same amount as in the absence of the tax change. However, this rules out some of the most powerful arguments — in both directions. On the side of the current Chancellor is the view that the top 1 per cent increase their incomes through increased effort, and that increased effort creates jobs for others. So the revenue effect should include the taxes collected on these new employees (and saving on transfers paid). This effect is potentially large, and could justify low tax rates. It could be right — from a revenue point of view — that Warren Buffett pays less tax than his secretary if his actions create many more jobs in, say, the rail freight industry. In the opposite direction, it has been argued that the inter-dependence is negative: that the increase in income of the top 1 per cent comes at the expense of other taxpayers. Suppose that we think back to the 1960s theories of the separation of ownership and control developed by Oliver Williamson, Robin Marris and others. In these models, managers are assumed to be concerned with their remuneration but also with other dimensions, such as the scale or rate of growth of their firms. They allocate their efforts to maximise a utility function that has several arguments. Where tax rates are high, there is a low return to effort spent on negotiating higher remuneration, and CEOs concentrate on expansion. But cuts in taxation mean that they switch efforts back to securing a larger share of the profits. This increase in remuneration comes at the expense of the shareholders (and lower growth). So against the increase in managerial pay has to be set the smaller amounts received by others, which means that the optimal top tax rate is higher.
These considerations take us beyond the analysis in the Review and — as with the proposal to tax food — may lead to rather different conclusions from those in their report.
Integration of income tax/social security contributions and behavioural public economics
In his Foundations of Economic Analysis, Paul Samuelson wrote that ‘to a man like Edgeworth, steeped as he was in the Utilitarian tradition, individual utility — nay social utility — was as real as his morning jam’. Edgeworth, who died in 1926, would have been quite comfortable if he had lived to read the Mirrlees Review. Household behaviour is largely governed by utility maximisation and social welfare is typically assessed in terms of the sum of transformed individual utilities. Understanding of both aspects — the explanation of individual behaviour, and the formulation of social objectives — has however moved on since Edgeworth. This is recognised at a number of points in the Review, but I would give greater emphasis to the alternatives to utilitarianism. There are potentially major implications for both positive analysis and normative analysis of taxation.
One of the key messages of behavioural public economics is that context matters in ways that are not recognized in standard analysis. For instance, I referred earlier to the effective marginal tax rate as the combined impact of all taxes on earnings paid by employers and employees and the indirect taxes paid when the earnings are spent. All of these are clearly relevant, but simply adding them assumes that context does not matter. Workers are assumed to react in the same way to the tax wedge regardless of whether it is paid by their employer, by themselves or by some other member of their household doing the weekly shopping. This is the maintained assumption in the argument for one of the Review recommendations — the integration of personal income and social security contributions — but it is not evident that this assumption holds. Recent research has found, for example, that tax ‘salience’ may lead people to respond differently to different forms of the same tax schedule. Even though the Review sees the link between contributions and entitlements as ‘vanishingly weak’ (p. 127), workers may not see it that way, and may respond differently to social security contributions than to income tax.
If context does matter, should this be exploited when planning taxes? If distraction means that the goose does not notice the feathers being plucked, should we use this device to reduce the hissing? Or should the government seek to make taxes more apparent? This brings us to the normative basis for the Review, where there are several distinctions to be drawn and these could usefully have been made more explicit. The first is between outcomes and process. Outcomes appear in the social welfare function, but process often features prominently in debates about taxation. One of the major arguments made in the Review for the integration of personal income tax and social security contributions is that of transparency. This is a judgment about process — one that I find quite appealing. In contrast, the argument in favour of integration on grounds of administrative simplicity is an argument in terms of outcomes: the cost savings would raise social welfare. In that case, we have to consider the different ways in which outcomes can be assessed. We may decide to focus on individual well-being, but this does not necessarily mean experienced utility. For many years, most recently in The Idea of Justice, Amartya Sen has argued for considering alternative evaluative bases, notably individual capabilities, defined broadly as the freedom that people have to function in key dimensions. Well-being assessed in terms of capabilities may lead to different conclusions.
Moreover, outcomes may be evaluated according to other criteria than well-being. A good example is gender equality. Taxes and transfers can contribute, either manifestly or latently, to reducing gender inequality. This is not considered in the Review, but influences our judgments about a number of the proposals discussed. The within-household distribution of income may be affected by the balance between direct and indirect taxation. Extending VAT to food may leave worse off those in the household who do the grocery shopping. When considering the income-testing of child benefit, we have to remember that an express intention of the legislation was to aid women by making the benefit payable to the mother in the first instance.
The two volumes produced by the Mirrlees Review provide valuable policy analysis and demonstrate the vitality of modern public economics. They are worthy successors to the Meade Report and represent a new landmark in the field. At the same time, my reading of the three topics considered here (just three of many covered by the Review) is that public economics has tended to become separated from other branches of economics (such as industrial organisation andlabour economics) — and from other disciplines (such as moral philosophy and psychology). It risks being too narrow in its approach and, by focusing the analysis too sharply, missing important parts of the story.
Note: This is a revised and shortened version of an article that appeared in the Journal of Economic Literature, 50(3), Sept. 2012.