Draft: 16.10.10
3
The Taxation of Labour Earnings
Taxing income from earnings is the biggest source of revenue for most OECD governments. Together with the benefit and tax credit system, it also does most of the heavy lifting in redistributing resources from richer to poorer households.
In this chapter, we set out some broad principles and evidence on the taxation of labour earnings, in particular on the inevitable trade-offs between the desire to redistribute resources to those with low incomes and/or high needs and the desire to promote economic efficiency, especially work incentives. Inevitable as they are, we will see that the severity of these trade-offs can be minimised by designing the tax system well. The next two chaptersfocus in more detail on options for reform.
In deciding how to tax earned income, we must decide both the tax base and schedule of tax rates—in other words, which income to tax and at what rates. For most employees in the UK, the tax base is relatively straightforward. With the exception of exemptions for most pension contributions, childcare vouchers and small items of work-related expenses, income from earnings is subject to the statutory income tax schedule. However, for some other earners (usually those on higher incomes or self-employed), more exemptions and deductions can be used to reduce taxable earnings. Avoidance schemes may also pass earned income through lower-tax jurisdictions.
In principle, the tax base should include all forms of remuneration, including benefits in kind, and deduct all costs of generating earnings, including work and training expenses. Of course, there are difficulties in the valuation of some in-kind benefits and it can be hard to distinguish some expenses from consumption, e.g. company cars and childcare. The choice of tax base may very well affect the composition of remuneration—the split between salaries and fringe benefits, for example. There is good evidence that the base-broadening reforms in the US in the 1980s[1] made it easier to raise revenue by increasing the tax rate on higher incomes.
Any reform to the tax rate schedule must recognise how individual taxpayers can change behaviour to minimise their losses from changes in the tax rate on their earned income. Indeed, one argument for looking at the tax system as a whole, as we do throughout this book, is to bring tax base and tax rate design issues together. A good tax system will tax all sources of remuneration. It may well provide an allowance for saving to avoid distorting when people choose to spend, but a good system will tax all sources of income, including income from investments, and will usually want to do so without favouring one sort of income over another. So, in this and the next two chapters, we are concerned with the taxation of personal income – the direct tax system. We will deal with the precise way this interacts with the taxation of savings and with business taxation in later chapters.
Earnings taxation includes all income taxes paid, credits and benefits received, and employer taxes on earnings. This broad definition allows us to focus on the overall ‘wedge’ between the disposable income that an individual worker has to spend or save and the cost of that worker to an employer. It also sets up our discussion of the interaction between and the integration of different income taxes and benefits, to which we return in chapter six.
Well-being depends on much more than measured income and this has to be accounted for in designing a good tax system. One person may have higher earnings than another because he or she is more productive or because he or she works harder.For most people working harder - by working longer hours for example –will reduce well-beingif it is not accompanied by more income. Different people will also have different needs and, for any given income and work effort, will achieve different levels of well-being depending on these needs. It is the this more general measure of well-being,which incorporates income, work effort and needs, that is a key input into our approach to earnings tax design. Furthermore, society will place different weights on incomebeing transferred to different people. The aim is not simply tochoose taxes so as to maximise the sum of utilities in society. Instead a ‘preference for equality’ is included in our measure of social welfare that acknowledges that society generally places more weight on gains to the poor and needy.
The maximisation of this measure of ‘social welfare’ stands a long way from the maximisation of some crude measure of aggregate income. It allows for work effort, needs and inequality. However, by acknowledging that people may respond to tax changes by adjusting the level of their work effort, it alsoexplicitly acknowledges the importance of incentives to generate income in the economy.A balance has to be struck between the level of work effort generating earnings in the economy – the size of the pie - and the degree of redistribution to the poor and needy.
If the tax authority could observe an individual’s productive capacity and needs directly, the tax design problem would be considerably eased. For example, taxes could be levied on those with a high capacity to earn—irrespective of their actual earnings—thereby preventing them from reducing their tax payments by working less. In the real world, immutable measures of productive ability and needs are not available.[2]Some measure of actual earnings, total income or expenditure is typically used to approximate ability or earnings capacity, while characteristics such as family size and age are used to approximate needs. Hence the equity-efficiency trade-off: if the tax rate on earnings is set too high, individuals may choose to earn less. The more the government can make the tax system contingent on observable factors closely related to abilities and needs, the smaller the welfare losses from taxation. We will certainly explore systems that allow taxes to vary with such characteristics.
If society cares about inequality and poverty, there will always be some willingness to sacrifice a part of national income in order to achieve distributional objectives. The fundamental design issue is to minimise such losses while raising sufficient revenue to finance desired public services and satisfy concerns over inequality and poverty.
Our goal in this chapter is to lay out principles for balancing economic efficiency and redistributive objectives in a coherent tax and benefit system. To do this we will need to address two key questions. The first is positive: how do individuals’ earnings decisions respond to taxes and benefits? The second is normative: given how people behave, what tax and benefit system would best meet policy goals?
In choosing a set of tax rates, a government may not be concerned only with redistribution and economic efficiency. It may also make social judgements—for example, wishing to favour the working poor over the non-working poor, or married parents over cohabiting ones. Even so, some tax arrangements dominate others by achieving the particular social objectives in a more efficient manner.
Over the longer term, the taxation of earnings can also affect choices over careers, education and training and whether to be self-employed or an employee. All affect the pattern of earnings and employment over the working life of an individual. Consequently, in our design of the rate schedule, we have to balance a broad set of incentive effects with the goal of raising tax revenue and redistributing to those with greater needs. By looking at the tax system as a whole our conclusions will present recommendations for a tax system that provides a coherent treatment of savings, pensions, human capital investments and earnings.
In the remainder of this chapter, we begin by discussing the evidence regarding employment, hours worked, and the responsiveness of earnings decisions to financial incentives, which shape the environment in which the tax system must operate. We then develop the central principles for the design of the rateschedule.
3.1.KEY FACTS ABOUT EMPLOYMENT, EARNINGS, AND LABOUR SUPPLY
Our discussion is motivated by the substantial changes in labour market behaviour that have occurred in the UK and similar economies over the past three decades. These point to the pivotal stages in an individual’s lifetime where tax rates impinge most powerfully on work decisions and on earnings determination more generally.
Labour supply responses—people’s decisions whether to take paid work, whether to work longer or harder, or whether to seek a job with a different wage—are a key input in the design of income taxation. Understanding these will inform us as to the scale of the equity-efficiency trade-off. They determine the size of the distortions that the taxation of labour income causes and the amount of revenue it raises. In the short run labour demand considerations – the number of jobs available, especially in recessions -may often dominate individual supply decisions. In the medium run the demand for workers will adjust and we cannot hope to say how best to design earnings taxes and benefits without knowing how people will respond in their labour supply decisions.
3.1.1.Trends in Employment and Hours of Work
Employment rates have changed significantly over the last 30 years in the UK and in many other developed economies. For example, Figure 3.1 shows a fall at younger and older ages in the UK, but a rise for people between their mid-20s and early 50s. The initial fall in employment at older ages has been partially reversed but remains below the levels of employment in the 1970s. We will see that these aggregate employment measures disguise rather different trends for men and women.
Employment changes are only part of the story. Changes in total labour supply in the economy can be separated into changes in the rate of employment and changes in effort per worker. These two components are referred to as the extensive and intensive margins respectively withthe level of effort typically measured by hours per worker. The distinction between the extensive and intensive margins of labour supply – between whether to work and how much to work - will be seen to play a major role in tax and benefit design.
Figure 3.1.Employment rates by age in the UK over time
Source: Blundell, Bozio, and Laroque,2010.
Three features characterise most of the variation in employment and hours of work across developed economiessuch as the UK over the past 30 years:
- an increase in the proportion of women in or seeking paid work;
- a decline in employment among men, especially at older ages;
- a decline in employment among men and women in their late teens and early 20s.
The rise in female labour supply reflects a fall in fertility rates, a trend towards marriage, and an increase in the proportion of mothers in paid work. The decline in male employment reflects less-educated educated and lower-skilled workers withdrawing from the labour market in the years before they reach state pension age, as well as more early retirement among higher earners (often with generous pension arrangements). Finally, the expansion of post-16 school and university/college education has increased the average age at which working life starts and increased the skills of the working population.
Looking at labour supply differences across countries can also be very informative. Figure 3.2a shows that the main differences in male employment in France, The UK and the USare concentrated at younger and older ages. Figure 3.2b suggests a little more variation in female employment but a strikingly similar alignment among women in their peak earnings years (their 30s and 40s) across these three countries.
Figure 3.2a.Employment by age in the UK, the US, and France in 2007: men
Source: Blundell, Bozio, and Laroque,2010.
Figure 3.2b.Employment by age in the UK, the US, and France in 2007: women
Source: Blundell, Bozio, and Laroque,2010.
Figure 3.2c.Average Total Hours of Market Work by age in the UK, the US, and France in 2007: women
Source: Blundell, Bozio, and Laroque (2010).
For many women with children the key decision is not just whether to work, but how many hours to work.[3] Figure 3.2c shows that hours of work dip at ages where there tend to be young children in the family and this dip is more pronounced in the UK than in the US or France. These differences between labour supply at the extensive and intensive margin for women with children will bear heavily on our suggested directions for earnings taxation reform in the next chapter.
A dominant characteristic of these figures is the strong variation in labour supply for men and for women in their late 50s and 60s. In most developed countries, labour market activity has fallen at older ages—with some reversal recently. In the UK, relatively poor and wealthy individuals who are relatively poor or wealthy are more likely to leave employment early than those in the middle of the wealth distribution. Figure 3.3 shows this most clearly. Broadly speaking, the poor are more likely to move onto disability (or means-tested) benefits,while the rich are more likely to retire early and live on private pension income. Those in the middle are more likely to remain in paid work.
Figure 3.3.Early retirement and inactivity by age in the UK by wealth quintile: men
Note: ELSA (English
Longitudinal Study of Ageing, 2002) sample of men.Wealth quintiles are defined within each five-year age group.
Source: Banks and Casanova,2003.
As was suggested by Figure 3.1, in the years before the recent financial crisis, the trend toward early retirement had partly reversed. Tax and pension incentives have played a role in this and maybe expected to continue to do so in the longer run.[4]
3.1.2 How Earnings Decisions Respond to Tax Changes
Changes in the tax rate on earnings will affect people’s behaviour in two ways, through the income effect and the substitution effect. Take the example of a reduction in a tax rate. For any workers earning enough to pay this rate, the cut will increase their income, allowing them to work fewer hours (or more generally, to supply less effort) by making it easier to maintain a given standard of living. This is the income effect. At the same time, workers who now face a lowermarginal tax rate will take home more of every extra pound they earn, encouraging them to work more. This is the substitution effect. The two effects offset each other and theory alone cannot tell us which will dominate.
An exception occurs at the extensive margin. Consider someone currently out of employment. A reduction in a tax rate on earnings cannot make work less attractive. In either case we need credible evidence to tell us how powerful these effects will be.
When it comes to looking for improvements in the tax system we will be mainly concerned with revenue-neutral reforms. Any tax reduction will have to be financed by a reduction in generosity elsewhere. In this case the income effect will tend to balance out across peoplewhile the substitution effect will remain. In terms of efficiency gains it is therefore the size of the substitution effect that will be important. Both effects will come in to play in terms of the impact on total hours supplied in the economy and on earnings.
Summarisingand simplifying, the available evidence[5] suggests that:
- Substitution effects are generally larger than income effects: taxes reduce labour supply.
- Especially for low earners, responses are larger at the extensive margin—employment—than at the intensive margin—hours of work.
- Both income effects and substitution effects, both at the intensive and extensive margins, are largest for women with school-age children and for those aged over 50.
As we have noted above, whether to work and for how many hours are not the only ways people respond to tax changes. They may change the amount of effort they put into each hour they work. They may decide on a different occupation or to take more training, or perhaps to become self-employed. Finally, they may respond by finding ways of avoiding or evading earnings taxation.[6]To capture the welfare cost of taxation across all these different margins of response, we can look directly at how a change in the tax rate affects the level of taxable incomedeclared – the taxable income elasticity.The taxable income elasticity subsumes the income and substitution effects described above, and captures any costly measure taken to avoid tax. In general, it also provides a simple and direct measurement of the welfare cost of earnings tax reform.[7]