A new generation of tax credits in April 2003 introduces a 'light touch' system for assessing eligibility. This represents an important innovation in the UK's welfare system aiming to reconcile sensitivity to need with unobtrusiveness - not an easy task. This project, carried out by Peter Whiteford (Australia), Michael Mendelson (Canada) and Jane Millar (UK) considered how this objective can be pursued, by setting the latest UK developments alongside parallel experience in Australia and Canada, which have longer-standing experience of using the tax system to deliver financial support to low- to middle-income families.

• Australians assess entitlement to their Family Tax Benefit on the basis of advance estimates of family income for the year the benefit is received. Once actual income is known, an adjustment can result in an additional payment or a debt to be repaid; the latter has caused wide controversy.
• Canadians receive their Child Tax Benefit according to income in the previous year. This avoids the need for adjustment, creating a simple but very non-responsive system; this seems to be acceptable in the context of a relatively low level of benefits and a back-up system of social assistance.
• Comparing design and delivery in Australia and Canada with that in the UK highlights the advantages and disadvantages of the different trade-offs between simplicity and responsiveness, in relation to: administrative burden; transparency, intrusiveness and compliance costs for recipients; equity; and work incentives.
• The UK's new tax credit regime has consciously drawn positive elements from both these countries, aiming to avoid controversial overpayments as in Australia and lack of responsiveness as in Canada.
• The researchers conclude that:

- Overall the reporting requirements do seem to have a 'light touch', but how this works in practice will depend on how many recipients both experience and report changes in income and circumstances. So far, little is known about either of these.

- The UK and Australian systems are both more complex and more intrusive for recipients than Canada's system.

- The impact of the new UK tax credits on work incentives is difficult to predict.

- The annual reconciliation is very new to the UK benefits system. A key lesson from the Australian experience is that it will be critical for the Inland Revenue to ensure that all potential recipients have the right information and advice at the right time.

The tax system, the government argues, can deliver a "light touch and non-stigmatising" income test (HM Treasury (2002), The Child and Working Tax Credits, The Modernisation of Britain's Tax and Benefit System, Paper No. 10, London: The Treasury, page 4). But any income test requires difficult decisions about what is counted as income, whose income in a family unit is counted, over what time period income is measured, how long awards should last and how responsive the system should be to changes in income and circumstances during the period of the award. Achieving a light touch means designing a system which, on the one hand, provides families with the help they need at the time when they need it, while, on the other hand, minimising intrusive, complex, costly and administratively burdensome procedures.

In designing the UK system, the Government explicitly drew on experiences of comparable tax-based systems in Canada and Australia, endeavouring to "steer a course between the two" (ibid., page 22). This research examined the Australian and Canadian systems in detail, looking at their design and delivery and comparing how these countries have tackled the competing objectives of simplicity and responsiveness.

In the new system, rather than basing assessments on income in the previous year, recipients must estimate their taxable income for the year ahead. At the end of that year, entitlements are adjusted to reflect actual income. People who have already received their credits through fortnightly payments must pay some of them back if they earned more than estimated, or receive more if they earned less. This 'reconciliation' sparked high-profile political controversy when at the end of the first year well over a third of recipient families received a repayment bill. As a result, the government waived the first A$1000 of any debt owed, reducing the number with repayment debts to about 10 per cent of claimants.

The unexpectedly high level of overpayments indicates the risk of a system based on advance estimates. Most often overpayments were due to individuals having jobs with fluctuating earnings or second earners increasing their pay/hours of work. In addition, it seems that families were more likely to underestimate their income than they were to overestimate; indeed most families did not get their estimates right.

In the second year of the system, there appeared to be some reduction of overpayments, but these still affected an estimated 33 per cent of families (compared with 39 per cent in the first year). One remedial measure has been to allow families to vary the amount they receive over the year and so reduce the likelihood that they will be overpaid.

The amount to be paid is based on net family income in the calendar year before the time a tax return is filed (in April), which determines benefits for the year (from July to June) ahead. There are no mid-year adjustments due to changes in income, although mid-year corrections are required for changes in family composition. Families experiencing sharp drops of income during the year may claim provincial social assistance to top up their incomes, providing the drop takes them below thresholds of eligibility for such schemes. Other than this emergency assistance for people who find themselves in dire straits, there is no adjustment once the year's tax-benefit levels have been set: income falls are not compensated for in the current year, and nor do income rises trigger in-year reductions or repayments.

This failure to make adjustments means monthly payments can in some cases be based on income received up to two years previously. However, this lack of responsiveness has not been an issue in Canada. This may be because the level of the benefits is lower than in Australia or the UK. Alternatively, it may because the simplicity of the system is much more valued by recipients than responsiveness to mid-year income changes.

The amount of tax credit to be paid is initially calculated with respect to the previous tax year's annual gross income. The award runs for up to the next 12 months, with an annual renewal at the end of the tax year. Recipients are required to report certain specified changes in circumstances - in which adults head the family, in discontinuing or significantly reducing the cost of childcare - during the period of an award. Families can choose to report other changes (e.g. children leaving home, change in usual hours of work) but they are not required to do so until the end of the tax year. However, changes increasing entitlement must normally be reported during the year, because any associated adjustment to the award will only be backdated for up to three months before the date the change is notified.

The system reconciling awards with actual income is made less painful by the decision that the first £2,500 of any rise in income, compared with the previous year, will be ignored. With that proviso, where changes are reported in-year, awards will be cumulatively adjusted (in a similar way as PAYE adjustments are made for income tax). At the annual re-assessment, overpayments will normally be recovered through a reduction of next year's award.

If recovery from future awards is not possible, or is inappropriate, the debt may be recovered either directly or, from April 2005, by an adjustment to the PAYE code. Underpayments will be paid as a lump sum.

Overall the reporting requirements do seem to have a 'light touch', with very few requirements to report changes in circumstances and no compulsory requirement to notify any changes in income. But how this works in practice will depend on how many recipients experience changes in income or circumstances during the year and on how many will report them. So far, little is known about either of these. About 7 per cent of families may have a change in the number of adults, but we do not know how many families will experience other relevant changes in circumstances. Nor is there good information on income changes, although the Treasury estimates that about 750,000 households will have a rise in income that would affect their awards. In respect of reporting changes, much will depend on families knowing and understanding the rules about which changes need to be reported and when. This in turn will depend to a great extent on the effectiveness of the Inland Revenue's advertising and awareness campaign. Many decades of evidence on take-up of benefits shows how difficult it can be to get the right information to the right people at the right time.

This suggests that it is likely that many changes will go unreported and so the end-of-year reconciliation is likely to be the main point when changes will be taken into account. This was true in Australia and, as happened there, the end-of-year reconciliation could therefore involve a substantial number of families. This will certainly be the case in 2003/2004 (the first year of the new scheme), because in this particular case initial awards will be based on incomes in the last-but-one financial year.

However, the UK proposals for dealing with end of year reconciliation have been designed to minimise the risk of the kind of problems experienced in Australia. The £2,500 disregard will minimise both the number and level of overpayments. The proposed method of recovery of overpayments - by adjustments to subsequent awards - will spread the repayment of any debts that are incurred and avoid the single large bill that has been so unpopular in Australia. On the downside, however, this will mean that families who have deductions for previous overpayments will receive payments that bear less relation to their current incomes during the repayment period.

One of the purposes of the UK design is to move away from a system in which each change in income and circumstances must be reported as it happens in order to adjust weekly payments. Such a system may be equitable, in the sense that awards relate to income in the same way for all claimants. It may be more or less transparent, depending on the complexity of the reporting requirements. But it is highly intrusive for recipients and it also has substantial administrative costs. It may also have a more direct negative impact on work incentives, since awards will be reduced directly as extra income is earned, although this can partly be offset by high earnings disregards.

The new UK design also rejects the Canadian model - which is similar to the Family Credit/Working Families Tax Credit design - in which income is measured over one set period and awards paid over another, regardless of changes in income. Such a system is very simple to administer, is non-intrusive for recipients and does not involve them in significant compliance costs. It is equitable in the sense that it treats families with the same income and circumstances in the same way, although it is last tax year's income rather than current income that applies. The negative impact on work incentives is reduced because awards do not respond immediately to increases in income.

The UK and Australian systems have much in common, except that Australia uses prospective and the UK retrospective income for the initial assessment. Each then seeks to respond to some changes as they happen and others in an annual reconciliation. These are therefore complex systems, both in administration and in compliance costs, and involve more intrusiveness for recipients than Canada's system. This is especially true in the UK, where there are several tax credits and in-work benefits. There are also issues of equity and work incentives for the UK. Equity is compromised by the existence of the disregard, as this means that not everyone with the same income and circumstances in the current year ends up with the same award. The impact on work incentives is difficult to predict. It could be positive in that the disregard means that some increase in earnings is protected from a reduction in the tax credit award. On the other hand, the system is not very transparent and if people cannot understand the system they may be reluctant to risk their awards by increasing their earnings.

The annual reconciliation is very new to the UK benefits system and indeed to most UK taxpayers, who currently pay their income tax through PAYE and do not complete annual tax returns (unlike Canada and Australia). The Australian experience warns us of some of the problems that can arise.

A key lesson for the Inland Revenue is that it will be critical to ensure that all potential recipients have the right information and advice at the right time.