The history of a single New Labour policy innovation—tax credits—contains almost the full spectrum of the party's success and failure in domestic policy over the past ten years. On the one hand, in the policy's eight-year span it has directed over £75bn into the pockets of lower-paid workers (and some non-workers), helping make work pay for those towards the bottom of the pile and preventing inequality from rising far more sharply. It has become one of the anchors of Labour's new "Anglo-social model," which aims to combine relatively free markets, including labour markets, with improved social protection. On the other hand, the distortions created by the presentational politics of tax credits, the "top-down" otherworldliness of the politicians and officials who designed them and yet another public sector computer failure have combined to cast a long shadow over the policy.
The tax credit story has other elements. It is a morality tale about what happens when good intentions are not matched by a close understanding of the lives of those whom the policy will affect. It is also a story about the influence of US public policy tools on New Labour, especially those used by President Clinton. And, finally, it gives some insight into the working methods of Gordon Brown, who has nurtured the policy from the mid-1990s, when it first emerged as a leading New Labour ambition.
Instead of expanding the income supplements inherited from the Tories, Labour opted for an ambitious new tax credit system with the promise of partly merging the tax and benefit systems. Why? Part of the answer is the influence of America's earned income tax credit (EITC), which was introduced in 1975 but greatly expanded by Bill Clinton in 1993. After 20 years in which welfare had been fiscally starved by Washington, intellectually undermined by dependency theorists and culturally stigmatised by the media, it appeared that Clinton had hit on a new way of distributing money to the poor that could repel these various opponents.
In the US, working families file a tax return at the end of the year, and when their income is below a given threshold they receive a "tax credit," typically as a single lump sum. The fact that the payment is made through the tax system ensures that it is associated with work, which strengthens the incentive to remain in employment. But it also helps stave off the accusation that this is an increase in "welfare." It is a liberal policy for a conservative, anti-welfare society.
This chimed with New Labour's own pessimism about the British electorate's appetite for redistribution—especially after the 1992 election defeat—and its need to fashion a more hard-headed and effective social democracy. That meant not only rethinking tax-and-spend macroeconomics, but also the passive and poorly targeted welfare system. One of the things that most impressed Brown about the EITC was its stress on rewarding work as opposed to "something for nothing" welfare. Given the level of workless households in the mid-1990s, Britain needed such a policy even more than the US.
Labour's 1997 manifesto was rather vague about this—pledging to "examine the interaction of the tax and benefit systems… to fulfil our objectives of promoting work incentives, reducing poverty and welfare dependency"—but soon after taking office Brown set about building the British equivalent of the EITC. His first step, within days of entering the treasury, was to ask Martin Taylor, then chief executive of Barclays, to look into how this could be done. Taylor reported back in March 1998, and his recommendations were predictably close to what the Brownites had envisaged. Warnings concerning the administrative challenge were muted, and Taylor now concedes, "If I had my time over again, I'd shout louder about the dangers involved, and warn them against overloading the tax system."
However, the Taylor report did note a number of features of the EITC that would not translate into a British context. Crucially, the EITC tends to be paid annually, while the British benefit system is based around weekly payments. Moreover, overpayment of tax is common in the US, and most taxpayers file an annual return in order to receive a rebate. This means most Americans are familiar with navigating the tax system. By contrast, only 30 per cent of British taxpayers file a return, and this is a different segment of the population from those who rely on benefits.
Nevertheless, Brown was attracted by the fact that a British version of the EITC could be sold as a "tax" rather than a welfare income supplement, and that it would be delivered through the pay packet. That meant running this cluster of benefits through the tax system, and thus the inland revenue. This form of delivery had the additional benefit that the policy would sit within the treasury's fiefdom, and so Brown's people could retain oversight of its design.
Short history of income supplements
The idea of creating a single, integrated system for taxes and benefits first emerged in the US in the 1960s, and had its intellectual foundations in the negative income tax scheme proposed in Milton Friedman's 1962 Capitalism and Freedom. This scheme would require everyone to file an annual tax return; anyone falling under a certain threshold would receive a tax rebate as a percentage of the difference between their earnings and that threshold. So if the negative income tax rate were 50 per cent and the threshold $10,000, someone earning $6,000 would then receive an additional $2,000 (50 per cent of $4,000) from the state. At least in Friedman's mind, all other welfare schemes could then be abolished.
A negative income tax system in Britain was considered by the Heath and Callaghan governments, and the idea remained fashionable in many countries throughout the 1970s. But it was held back by lack of computerisation. Only after Nigel Lawson had allowed PAYE to be computerised would a negative income tax have been possible in Britain, but by then the idea had run out of steam. By the early 1990s, the notion of a fully integrated tax and benefit system was rarely discussed. In his report, Martin Taylor explained that "with the significant exception of my study of a tax credit for working families… I have not pursued the line of full-blown integration."
In addition to the technological challenge, there is a deeper reason for resisting integration: tax and benefits remain intrinsically separate functions of government; you cannot view one merely as an inverse of the other. Each of the two departments concerned is unaccustomed to the culture, problems and—most importantly—users of the other. To put it simply, tax is paid by people who have enough money; benefits are claimed by those who don't. More tangibly, benefits are paid to households, while income tax (in Britain) is paid by individuals. Finally, there is the crucial distinction between the annual calculation of tax and the weekly budgeting of low-income families.
But if tax-benefit integration had fallen out of favour, the need to make work pay at the bottom end of the labour market had become more urgent as unemployment took off in the 1970s and 1980s. In 1969, Richard Nixon declared that, "What America needs now is not more welfare, but more workfare." In the US, "workfare" became the term to describe all attempts to challenge welfare dependency and push people into work or training with various combinations of sticks and carrots. Labour's New Deal policy is the heir to this tradition.
In Britain, Keith Joseph introduced the family income supplement (FIS) in 1971 to top up the earnings of low earners and thus help to beat the "unemployment trap," which occurs when someone out of work faces no or little financial incentive to find work because of the withdrawal of benefits. Like any other benefit, the FIS was paid via the department for social services (DSS). Although controversial at the time, the FIS applied to a tiny proportion of workers, and take-up was less than 50 per cent. Some critics complained that the FIS replaced the unemployment trap with a poverty trap, in which incentives to work for more hours or money were undermined by a sharp fall in benefits as income from work rose. Others, especially on the left, argued that in the absence of a minimum wage, this policy simply subsidised bad employers.
Still, FIS remained little more than a hint of a different direction for British welfare policy until Norman Fowler's reforms of 1988. From here on, income supplements became a central part of British welfare policy. Fowler significantly expanded the FIS, retitling it "family credit," but it was still treated as a benefit payment for working families administered by the DSS. This system was still in place in May 1997, by which time it was paying out £2.3bn a year.
Today, David Willetts, one of the leading Tory thinkers in this area, accuses Brown of adopting a "year zero mentality in which he refused to consider the Conservative policies that were already in place." Brown undoubtedly could have taken Fowler's family credit system and pumped more money into it. But there is little doubt that the welfare system that existed when Labour took power was inadequate to the challenge of getting more people into work.
Where unemployment may have been the more obvious problem throughout the 1980s and early 1990s, it was nonemployment (where disability or dependent children force individuals out of the labour market) that had become a more pressing issue by the time Brown entered the treasury. It was especially important to get single mothers into the labour market. By May 1998, take-up of family credit—which had begun at 57 per cent in 1988—was still only at 70 per cent, and the payments and income threshold remained too low to make work significantly more attractive than benefits, especially given the loss of housing benefit that acted as a major disincentive to taking work.
A cultural and administrative gulf still existed between the benefit system and the labour market, and the Tories had run out of ideas to do anything about it. As Paul Gregg, a Bristol University economist hired by Brown in 1995 to work on welfare policy, puts it, "the welfare state was still much too passive under the Tories." Brown arrived with a basket of reforms aimed at making low-wage work more attractive: the New Deal, the minimum wage, a 10 per cent tax band and a system of tax credits.
Tax credits in two phases
The first phase of tax credits, which consisted of the working families tax credit (WFTC) and the disabled persons tax credit, arrived in October 1999, 18 months after the Taylor report. Taylor had recommended that the family credit initially be used as the model for the WFTC, and the conditions were virtually identical. Eligibility was still limited to parents in work, recipients still had to work for a minimum 16 hours a week and payment was still calculated on a six-monthly basis. The WFTC was heavily focused on making it easier for low-income parents to take work, especially through supporting childcare costs. Someone working part-time on the minimum wage with two school-age children would see their income roughly doubled by the WFTC, but the credit would be slowly withdrawn as their income increased, falling to zero at a weekly income of around £260.
Apart from the larger sums of money involved, this was mainly a rebranding exercise facilitated by transferring the unit responsible for family credit from the DSS to the inland revenue. Beneath the presentation, tax credits remain a means-tested income supplement, paid in addition to universal benefits like child benefit. But from Gordon Brown's point of view, the big difference for the recipient was that WFTC appeared in their pay packet and was thus associated with work. The desired perception of the policy—in the eyes of both the recipient and the voters—determined its mode of delivery. (Brown initially attempted to exploit the rebranding even further, by accounting for tax credits as tax revenue lost rather than public spending on welfare. The Office for National Statistics refused to accept this.)
By several of its own criteria, the WFTC was a success. Take-up stood at 65 per cent of an eligible 1.5m people—a slightly smaller percentage than the family credit's high-water mark, but in absolute numbers far more households were receiving the benefit. Employment among lone parents also started to rise. Because the WFTC was so closely modelled on family credit, recipients were familiar with how to make claims. The claimant simply declared their circumstances and earnings twice a year, and a calculation of entitlement was made for the next six months, regardless of any changes in income that occurred along the way. Brown poured £6bn a year into the WFTC, more than double what had been spent on family credit. The WFTC was afflicted by significant overpayments—10 to 14 per cent a year—but this was mainly a result of claimant error and fraud, not the system's complexity.
However, it is at this point in the story that bigger problems emerge. In his 2000 budget, Brown announced a second phase of tax credits, to take effect from April 2003. The plan was to split the WFTC into two new types of tax credit. First, there would be the working tax credit (WTC), payable to workers on low incomes. This was an income supplement that appeared in the pay packet, as the WFTC had done. The WTC was also a mechanism for assisting with the costs of childcare that may result from being in work. Second, there was the child tax credit (CTC), payable to most families with children, whether the parents were in work or on benefits. This was paid directly to the recipient. Partly as a result of the fact that eligibility had now been expanded to workers without children and to parents not working, the number of claimants would rise sharply—it currently stands at 6m households.
In addition to making more money available to families, the 2003 reform sought to further integrate the tax and benefit systems. But this created serious complexities for claimants. Rather than reporting their earnings every six months and then receiving the appropriate benefit (as they had under both family credit and the WFTC), the new system calculates entitlement for the whole year on the basis of the previous year's income, and responds to subsequent variations in income and earnings a year later. This drags claimants into a labyrinth similar to the one occupied by self-assessment taxpayers, and makes overpayments a normal feature of the system. Moreover, claimants now have to volunteer information about changes in circumstances (earnings, childcare, who they are living with and so on); under WFTC they merely had to provide information when asked. It has taken people a long time to register this change.
The practical difficulties encountered by claimants are crystallised in the claim form itself, which requires individuals to compile statistics from various sources. Aside from earnings over the previous year, the form asks how much individuals have contributed to PAYE, pensions, social security and gift aid. And calculating childcare costs means individuals having to try to account for fluctuations in circumstances, such as school holidays. The guidance notes themselves run to 50 pages.
The mismatch between the psychology of individuals whose finances work on a weekly basis and a computer-based bureaucracy that can work only in terms of April-to-April tax years remains a problem to this day. Mike Brewer of the Institute for Fiscal Studies, who worked on welfare reform for the treasury from 1999-2000, says: "The case for the 1999 policy was good; the case for the 2003 reform seems much weaker. I wonder if the chancellor regrets it."
Exactly what happened between the announcement of the new tax credit schemes and their disastrous launch in April 2003 is the stuff of Whitehall legend. The challenge of building an IT system that could cope with these new demands fell to the vast American company EDS, and it notoriously failed to rise to it. Back in 1994, the revenue had signed a strategic partnership agreement with EDS covering all IT projects for the next ten years. The relationship had always worked smoothly, including over the three-and-a-half year lifespan of the WFTC. EDS began work on the new tax credit system in summer 2000, and in late 2002 the Office of Government Commerce reported that it was "an exemplar of good programme management."
According to Nicholas Montagu, head of the inland revenue, the new system needed to be "about six or seven times more complex than the system required to bring in tax self-assessment." EDS therefore needed a lot of time to build and test the new system. The usual practice in such situations is to build a "clone" system and run trials on it. But fearing that the launch date would be missed, the treasury started to grow impatient. There are reports that Brown intervened personally to speed the process up, leading the trial period to be compressed.
The relationship between the political and delivery wings of Whitehall was now at its most dysfunctional. The special advisers and civil servants responsible for policy design were largely unfamiliar with the delivery channels and with the lives of benefit recipients. There are also questions about the role of Dawn Primarolo, who as paymaster general was the main minister responsible for the new policy. In June 2003, Montagu told the chair of the treasury select committee that he had not spoken to Primarolo at all between autumn 2002 and March 2003. She was evidently unable either to spot or to stave off the trouble brewing inside the revenue. There are accusations that she even exacerbated it, by taking decisions on policy design too close to the launch.
A big design flaw
Problems with the system began in the months preceding April 2003, with claimants unsure whether their claims had been processed or not. Many never received notification of the size of their awards before receiving their first payment, which had the serious implication for low earners of making it difficult to plan ahead. But problems escalated once the payment system started to operate. Only a few hours after its launch, Montagu received a report that the system was running slowly. Claimants themselves realised something was amiss soon afterwards, when a large number of the first payments, scheduled for the 7th and 8th of April, were made a day or two late. Backlogs developed, leading to payments being made later and later, and much of the processing work being done by hand. Delays varied from a few days to several months: over 400,000 applicants received their first payment later than they had been told to expect, while by 2nd July 220,000 people had still not been paid. The revenue was forced to make 200,000 emergency payments, and call centres were swamped, with over 2m calls made on some days. The inland revenue had received a rude awakening to its new sphere of responsibility, as women arrived in tax offices with their children asking how they were expected to feed them that week. Primarolo apologised to the House of Commons on 28th April, and by the end of the year, EDS's contract had been transferred to Cap Gemini.
One of the biggest flaws in the design resulted from badly underestimating the degree of income fluctuation that low earners can experience in a year. This has led to an in-built overpayment bias, and while IT problems were responsible for about £100m of overpayments in the first year, the bigger difficulty for claimants lay in the fact that it wasn't clear when an overpayment was caused by government error, when it was caused by their own error, and when it was a normal feature of the system. The inland revenue produces an estimate for the claimant's income for the tax year ahead, and if their income rises above that estimate during the tax year, they will be overpaid for the remainder of that year, and owe this money back to the revenue at the end of it. The 2005 pre-budget report sought to deal with this underlying problem by raising the income "disregard"—the amount incomes can rise by mid-year without leading to a reclaim being made on the tax credit—from £2,500 to £25,000. This has cost the treasury £500m a year.
In addition, overpayments can occur as a result of claimant error or fraud. If the claimant fails to inform the revenue of a rise in earnings or change in circumstances at the end of the tax year, then payments will continue on the basis of the previous year's declaration, leading to overpayment. One common error has been to forget to include the income of one's partner in the claim.
Even if the system does start to function perfectly, large numbers of overpayments will still occur and will need to be straightened out retrospectively. This will cause considerable stress for families who don't know what they might or might not owe to the government in a few months' time. The government, meanwhile, faces the problem of collecting money from individuals who may be reluctant to return what they thought was theirs.
What went wrong (and right)?
In the three years for which there are data on the new tax credits, the evidence is as follows. The revenue pays out £17bn a year in credits to 6m families (making it the second largest item of social spending after the state pension), and has been overpaying close to £2bn and underpaying by about £500m a year. Of these overpayments, the treasury accepts that there is about £800m a year that it may never get back. This is in addition to a further £1bn mistakenly or fraudulently claimed each year—giving a total of £5.4bn "misdirected" over just three years.
The system is in urgent need of reform, but it can also boast notable achievements. Above all, there is unprecedented take-up, now at 82 per cent overall, and 97 per cent for families earning less than £10,000 a year, compared with 72 per cent among family credit's (far fewer) eligible applicants. (The government has marketed the policy very heavily, spending £5.37m in 2006-07 alone.) The policy has improved work incentives too, as shown by the rise in lone-parent employment from 42 per cent in 1997 to 57 per cent today. (The net effect of tax and benefit changes since 1997 has been to raise incomes for the poorest 20 per cent of families by 12 per cent, while cutting them by 5 per cent for the top 10 per cent—and tax credits are one of the main reasons.)
With four in ten families now paying no net tax, Brown's goal of making work pay for those at the bottom of the income scale has worked. The question is whether it was necessary to use a system with an almost 15 per cent inaccuracy rate to achieve this.
David Harker, chief executive of Citizens Advice, says, "This is an untenable system. An annualised system doesn't provide the stability of income required by low-income families." Most experts agree that bringing the policy into line with the financial habits of its users can now only be achieved through scrapping the current system and starting again. But this is inconceivable as things stand. The best option is to consider how the government can create greater dialogue with claimants than the current system allows. The system needs to feel less punitive and to offer explanations to claimants before demanding overpayments.
Tax credits have become a punchbag for both left and right. The left might have been expected to back a policy that has increased the incomes of the poor so sharply, but the delivery failure and the subsequent pressure on hundreds of thousands of hard-pressed families has alienated potential supporters. Moreover, according to many Labour MPs, even those claimants who have had a positive experience tend not to associate their extra cash with Labour policy.
How could a flagship progressive policy run into so many problems? Buildings designed according to the way users are assumed to behave cause the sort of misery witnessed on many modernist housing estates. The architects of tax credits were guilty of an analogous error. The policy was designed as if its delivery mechanisms would fall into place and its users would conform to an economic model. And perhaps more than any other big New Labour initiative, the policy was built around presentation—in this case the need to call a benefit a tax and to pay it through the pay packet. (In fact, tax credits ceased to be paid via the pay packet after March 2006.)
Brown's people had no experience of governing when they entered the treasury, and treated resistance to their ideas from civil servants with suspicion. The idea of tax credits was a simple one, but the fact that it couldn't be delivered in a simple fashion was somebody else's problem and then somebody else's fault. EDS publicly accepted a large share of the blame, along with a £71m penalty, but only in the hope of cashing in on the next round of IT procurement. Responsibility for the larger error—that of designing a system around annualised income calculations with a delayed clawback when circumstances change—lies somewhere deep inside the treasury.
If there is a lesson, it is that political ends and means cannot simply be jammed together through sheer force of political will, especially when the state is trying to do complex, customised things for 6m households. Moreover, one has to wonder if a policy device that can be traced back to Milton Friedman could ever be that neatly adapted to the goals of the centre left, despite the Clinton precedent. Friedman imagined a minimalist, perhaps even brutalist, way of simplifying economic incentives across society. The more the Brownites sought to humanise that vision, the more complexity they added, until the system failed many of the people it was intended to help and gave a good policy a bad name.
Given the problems that have dogged the system, and the fact that they were so clearly predictable and predicted, some feel that Gordon Brown has got off lightly. The Tories never succeeded in launching a coherent assault on the policy or dressing down Brown in the Commons—although both Primarolo and Tony Blair have apologised to the House for the system's failings. But by weathering the storm, Brown has succeeded in altering the terms of debate about British welfare policy. As Paul Gregg puts it, "There has been a cultural change, in that work incentives are now recognised as a core aspect of a welfare system. Both parties now accept that. I view that as a vindication of the direction of travel."
Taking the long view, Brown may feel that the reforms have been worth the pain. After all, the Tories have said they will keep the system and try to manage it better. His decade as chancellor has seen 600,000 children lifted out of poverty, compared to the doubling of child poverty that had occurred over the previous 20 years. For all their flaws, tax credits have played an important role in this.
Tax credit success
Britain's tax credit scheme has succeeding in redistributing billions to low earners. It has also been designed to minimise the unemployment and poverty traps, by ensuring that the speed at which benefits are withdrawn as income rises does not create disincentives to take employment, or to work harder or longer.
The unemployment trap has been softened by tax credits, although the fact that housing benefit is no longer removed once a recipient enters the labour market has helped too. As for poverty traps, they can never be entirely eliminated, but tax credits have shifted them further up the income scale. While the working tax credit disappears for a single person without children once their salary hits £13,000 (creating a possible disincentive to seek a higher income), child tax credit tapers away only gradually between salaries of £14,495 and £55,000.
How does it work in practice? A family with two children, and neither parent in work, would be entitled to an annual child tax credit (CTC) of £4,325. When one or both parents takes a job, the household becomes entitled to working tax credit (WTC): if one parent takes a job at the minimum wage—currently £5.35 an hour—for 35 hours a week, their annual income rises by a total of £6,567, thanks to tax credits.
Tax credit failure
The complexity of the tax credit system seems to have created more than its fair share of mistakes. Consider Michelle George, a lone parent with two children from Luton. Shortly after her divorce in 2003, George took a job in telemarketing on £7.20 an hour, and applied for tax credits. With the WTC, the CTC and support for childcare for her youngest child, she was initially receiving £125 a week, to supplement her after-tax income of £110. When her younger child started school shortly afterwards, she phoned her tax office to inform them that she wished to cancel her claim for childcare. The cancellation was not properly processed, but she was not told, and so in February 2004 she received a computerised awards notice, informing her that she owed the revenue £1,500, and her payments were automatically terminated.
This decision was eventually overturned and George was promised a rebate (which has not arrived). But even if no errors had been made, George believes her case illustrates a problem with the system. "Nobody knows what they're entitled to, and you don't know whether you've had an overpayment or not. It's just too complicated." Combine that with the insensitivity of automated notices, and you have a system that does not feel at all user-friendly.