What are the options for reforming pensions taxation?

05 May 2022

There are various reforms that could improve the targeting of pension saving incentives or alleviate inequities in the current system.

Private pensions are treated favourably by the tax system. There are good reasons to encourage saving for retirement, but current policies are expensive and not well targeted. There are some features where clear improvements could be made, and a common issue raised in public debates is whether the tax treatment of private pensions is unfair because those on higher earnings receive more tax relief than those on lower earnings. Here we discuss some guiding principles and the pros and cons of specific reform options.

How and why are pensions treated favourably? 

Private pensions are treated favourably by the tax system. People can take 25% of their pension savings free of income tax; pension contributions made by employers escape National Insurance contributions (NICs) completely; and private pensions are relatively lightly taxed on death. 

Depending on what pensions are being compared to, two other features of their tax treatment can also be viewed as advantageous. 

  • First, income tax is levied when money is received from a pension, rather than when money is contributed to a pension. People who face a lower tax rate in retirement than when they are working therefore pay less tax than if the tax were levied up front on earnings contributed to a pension. So, for example, a higher-rate income tax payer who saved £1 would lower their income tax bill at that point by 40p. If that person were a basic-rate taxpayer in retirement, they would pay just 20p income tax when they took the £1 as pension income. That is, they would pay 20p less in tax on that £1 than if the income were taxed up front. But there is a good case for applying tax when money is taken out of a pension (and available to spend) rather than when it is earned: this allows the government to levy more tax on those whose pensions make higher investment returns, without discouraging saving. If taxing money when it is spent is taken as the benchmark, the fact that some people would have faced a higher tax rate if tax were levied earlier would not be viewed as a tax break. 
  • Second, there is no income tax or capital gains tax levied on investment returns within pension funds. This is a tax advantage relative to the default tax treatment of savings, which involves taxing investment returns as they are received. However, that default system is arguably not an appropriate benchmark because it would penalise saving, and in practice the returns accruing on most other assets are not taxed (or not taxed in full) either. 

An important justification for a preferential tax treatment of pensions is to incentivise an appropriate level of saving for retirement. There are concerns that individuals would not save enough if left to their own devices – for example, because they are too short-sighted or impatient or do not have enough information to understand their retirement resource needs – and that they might fall back on the means-tested benefit system (at the taxpayer’s expense) in retirement if they have not made enough private provision for themselves. There is therefore an argument for the government to encourage individuals to save for the future, and offering tax subsidies for saving in a pension is one way of doing this. Since money saved in a pension cannot normally be accessed until a minimum age (55 until 2028), this encourages saving specifically for later life. However, the particular ways in which pension saving is currently tax favoured are not necessarily well targeted or effective. 

Should the system be reformed?

A reasonable starting point for the taxation of pensions and other savings would be for the tax system to interfere as little as possible with people’s saving decisions. One challenge for the taxation of pensions is how to achieve this. This includes how best to avoid a situation where pension saving is taxed twice – both when money is paid into the pension and when income is received from the pension – and also a situation where pension saving is never taxed – i.e. where tax is collected neither when money is paid into the pension nor when the income is received from it. This issue is key for evaluating many commonly proposed reforms, such as to income tax relief on pension contributions.

However, as noted above, there are also reasons for the government to encourage pension saving, and the tax system is one tool for doing this. Any such incentives should be well targeted, the meaning of which depends on whether the government’s aim is to incentivise all pension saving equally or, for example, to give incentives only, or especially, for certain groups – most obviously those who would otherwise be most at risk of under-saving for retirement – or up to certain limits. Whether any incentives are effective will also depend on how well people understand the incentives and on whether they respond by saving more and using those savings appropriately in retirement. Many arguments for pensions tax reform reflect concerns that the current tax incentives are not well designed.

Below we briefly discuss the main options for reforms to pension taxation. We consider whether they would improve the targeting and effectiveness of incentives for pension saving given different possible objectives, and whether they would alleviate any undesirable incentives or inequities in the current system. We also consider the administrative practicality of reforms, which can differ between pensions of different types.

When fully evaluating a reform, there is also a range of other factors to consider, including: the implications for government revenue and the management of the public finances; how the taxation of pensions fits with the taxation of other forms of saving; the winners and losers from a change; and the likely effect on retirement resources, which can be automatically increased (or decreased) when tax treatment is made more (or less) generous even if people don’t change their saving behaviour. 

Designing pension tax reform can be difficult because money flows into and out of people’s pensions over several decades – often almost their entire adult lives. This means that reform proposals often raise the question of how existing pension wealth would be treated in the transition from the old to the new system. This in turn raises a number of important issues, from administrative practicality and the credibility of long-term policy commitments to the equitable distribution of the tax burden across generations and the fairness of ‘retrospectively’ levying more tax on current pensioners’ past saving. 

Option: Restrict up-front income tax relief to the basic rate

One commonly proposed reform is to restrict the income tax relief on pension contributions to the basic rate of income tax in order that higher earners – specifically, those paying higher rates of income tax – do not receive higher relief. 

However, such arguments are often misguided because they focus on the tax treatment of pension contributions in isolation from the tax treatment of the pension income they finance. Pension contributions are excluded from taxable income because pension income is taxed when it is received: in effect, the tax due on earnings paid into a pension is deferred until the money (plus any returns earned in the interim) is withdrawn from the fund. 

Restricting tax relief to the basic rate for all would mean that people paying higher-rate income tax both in working life and in retirement would pay 20% tax on their pension contributions when they are made and still pay 40% tax on their pension income when it is drawn – effectively paying income tax on their pension saving twice. This would be unfair and a significant disincentive to save in a pension for these people. 

One might want to limit income tax relief to the basic rate for those higher-rate taxpayers who will be basic-rate taxpayers in retirement, so that they do not gain from being able to defer paying income tax. Receiving higher-rate relief on contributions and then paying only basic-rate tax on pension income – and the majority of higher-rate taxpayers in any given year are likely to be basic-rate taxpayers in retirement – is often considered almost self-evidently unfair, though in fact the issues are complex and it is debatable whether such outcomes really are undesirable. But in practice it would be impossible to identify which higher-rate taxpayers will be basic-rate taxpayers in retirement and to reduce the rate of tax relief just for them.

If the argument for reform is that people should not face a different tax rate on pension income in retirement from the rate of relief they received in working life then this would also apply, for example, to those who are basic-rate taxpayers in working life but pay no income tax in retirement. One way to prevent people receiving a different rate of relief when making pension contributions from the tax rate they pay in retirement would be to move to a uniform rate of relief on pension contributions and a uniform rate of tax on pension income. But it is debatable whether that would be an improvement on giving relief and levying tax at an individual’s marginal rate.

There would also be a major practical difficulty with restricting the rate of tax relief. Charging tax on contributions to higher-rate taxpayers’ pensions but not basic-rate taxpayers’ pensions would require HMRC to know the size of the pension contribution for each individual. But in defined benefit pension schemes, employers do not make a separate, measurable contribution in respect of each employee: instead, the employer makes whatever aggregate contribution to the scheme is needed (over and above aggregate employee contributions and investment returns) for the scheme as a whole to be able to expect to meet its future obligations to all scheme members. Valuing the annual ‘contribution’ for each individual would require valuing the future pension benefits they accrue during the year, which is a complicated exercise that is currently done only in a crude way for assessment against the annual allowance. Perhaps most obviously, the value of an addition to an individual’s future annual pension income will depend on their life expectancy, which in turn will depend on many factors. 

Option: Flat rate of relief

Another reform option that is often discussed is introducing a flat rate of up-front income tax relief on pension contributions – for example, at 25% or 30% – for all individuals irrespective of their current income tax rate. The proposal to restrict relief to the basic rate (discussed above) is really just an example of this in which the flat rate of relief is set equal to the basic rate of income tax. As with those proposals, the argument to restrict relief is often misguided in that it looks at the tax treatment of pension contributions in isolation from the tax treatment of the pension income they finance. Giving a flat rate of relief would also face the same practical difficulty in requiring a valuation of annual benefits accrued by each individual in defined benefit pension schemes. 

Proposals for a flat rate of relief usually envisage a rate somewhere in between the basic and higher rates of income tax (indeed, it is sometimes chosen to be revenue-neutral, such that the reduction in relief for higher- and additional-rate taxpayers is just enough to pay for the increase in relief for basic-rate and non-taxpayers). It would therefore increase the incentive to save in a pension for basic-rate and non-taxpayers and reduce the incentive for higher- and additional-rate taxpayers. People who were higher- or additional-rate taxpayers while working and also either higher- or additional-rate taxpayers in retirement would be taxed twice, paying some income tax up front as well as income tax in retirement. While the government may wish to redistribute towards lower earners, and to encourage retirement saving particularly among these individuals, it is unclear why one would ever want to double-tax the pension saving of higher earners. Redistribution from higher to lower earners could be achieved far more easily and effectively by changing the rates and thresholds of income tax and NICs if that were the objective.

Option: Reform the tax-free lump sum

The 25% that can be taken from a pension tax-free, normally in the form of a lump sum, is arguably the most transparent tax incentive to save for retirement. But it has three properties that seem hard to rationalise:

  • It encourages people to take a big part of their pension as a lump sum rather than a regular income throughout their retirement.
  • It is more generous to those on higher incomes in retirement: specifically, it is worth more to higher-rate taxpayers than to basic-rate taxpayers, and worth nothing to those who have too little income in retirement to pay income tax. This last group is particularly big: in 2021–22, 43% of those aged 66 and over did not pay income tax. 
  • It is limited only by the overall lifetime cap on tax-free pension savings – subsidising further pension saving even among those who already have, say, £750,000 in a pension pot.

Each of these features could be changed.

The government could allow 25% (or any other percentage) of anything received from a private pension to be tax-free – without requiring it to be taken as a lump sum. It could also make the 25% free only of basic-rate tax, not all income tax, so that it is equally generous to all taxpayers (though it would still not benefit non-taxpayers).

Instead – or as well – the tax-free element could be limited to a certain cash amount. This would continue to encourage people to build up a pension the government deemed adequate, but stop providing further subsidies for those who already have more than this, without preventing them from saving more in a pension if they wish to without this subsidy. To prevent charges of retrospective taxation, the government could consider exempting pension savings already in place that would exceed the cap. Limiting the tax-free lump sum would target tax relief more closely at encouraging those at risk of under-saving to save more, but would be harder to administer as, for people with more than one private pension, it would require monitoring the total amount they had taken tax-free across all of their pensions, rather than simply being able to take 25% tax-free from each. 

Another way of reforming the tax-free lump sum would be to replace it with a bonus top-up to accumulated pension wealth that could be paid by the government when pension income is accessed. At the moment, for every £1 taken from a pension, 25p is tax-free if taken as a lump sum, saving a basic-rate taxpayer 5p (20% of 25p) and a higher-rate taxpayer 10p (40% of 25p) in income tax. Instead, the government could top up pension savings by, say, 5%, and then charge income tax on the full amount taken from the pension. For a basic-rate taxpayer, a 5p top-up for every £1 of pension savings would be broadly equivalent in value to the tax-free lump sum. This would combine the features of the reforms described above. It would stop favouring lump sums over regular incomes. It would benefit those who were not income taxpayers in retirement, and not be worth more to higher-rate taxpayers than to basic-rate taxpayers. And the bonus could be capped, so that only the first (say) £500,000 of pension wealth was topped up, though again this would be more complex to administer than an uncapped top-up.

Option: Introduce NICs on employer pension contributions

Currently, employer pension contributions escape employer and employee NICs entirely: NICs are levied neither on the contributions going into the pension nor on the income received from it. 

This clearly encourages employers to provide a pension for their employees. But there is no corresponding support for the pension saving of the self-employed. And while we may wish to incentivise employer pension contributions, it is not clear we should want to do so by more than we incentivise employee contributions – giving rise in the process to opaque ‘salary sacrifice’ arrangements, whereby some employee contributions are in effect dressed up as employer contributions in order to avoid employee and employer NICs. 

The size of the incentive for pension provision also varies in undesirable ways because it is related to NICs rates. The incentive is weaker where employees either earn too little to attract NICs or earn above the upper earnings limit (at which point the employee NICs rate falls) than where their earnings are in between those levels. And the incentive is weaker for small employers whose employer NICs liability is already covered by the NICs employment allowance, and for employees aged under 21 for whom employer NICs are typically not payable. There are other such cases too. It also means that the incentive has become stronger over time as headline NICs rates have risen – an outcome that was almost certainly unintended. When, in September 2021, the government announced the introduction of the health and social care levy (in effect an increase in NICs rates), it seems unlikely (and it was not stated) that it was aiming to increase the relative subsidy for employers to make pension contributions, but that was an effect.

There are better ways to incentivise pension saving, such as through a simple top-up to pensions as described in the ‘Reform the tax-free lump sum’ section above.

Administratively, introducing employer NICs on employers’ pension contributions would be relatively easy because employer NICs are largely flat-rate – as long as the government were willing to charge NICs on employer contributions even in respect of some earnings that are not otherwise subject to employer NICs. 

Introducing employee NICs on employers’ pension contributions would be relatively straightforward for defined contribution schemes but harder for defined benefit schemes, because applying the appropriate NICs rate for each individual in these arrangements would require valuing the employer’s ‘contribution’ in respect of each individual – the same challenge of valuing the increase in their future pension rights as discussed above in the context of restricting up-front income tax relief to the basic rate. 

Option: Introduce NICs on pension income

No NICs are levied on employer pension contributions or on the resulting pension income (and no health and social care levy will be either, when it is introduced). In the previous section, we discussed why this is a poorly targeted way to provide relief and how it could be removed by adding NICs on employer contributions. The obvious alternative way of taxing employer contributions is to impose NICs on the pension income that they generate. Levying NICs on pension income rather than on pension contributions – that is, adopting the approach used for income tax – would have advantages. It would avoid the need to value contributions to defined benefit pensions. It would mean that, if people achieved particularly high returns on their pension savings, the exchequer would take a share. It would deliver tax revenue when people are older, making it easier to manage the public finances, since the revenue stream would more closely match demands on public spending. And it would be a big step towards integrating income tax and NICs.

In so far as introducing NICs on pension income would tax existing pension wealth (as well as future accruals), it would be an economically efficient way to raise revenue (because prior saving could not be changed in response). But it would face the charge of retrospective taxation, undermining the legitimate expectations of those who had saved in a pension on the understanding that it would be free of NICs in retirement.

For pension income derived from employer contributions, it would merely be undoing the NICs subsidy given to date, albeit arguably retrospectively for existing pension wealth.

But for pension income derived from employee contributions (which had already been subject to employer and employee NICs) or from contributions by the self-employed (which had been subject to self-employed NICs), it would be outright double taxation, in effect levying NICs on the same pension saving twice.

There are several possible ways of trying to address, albeit partially, these concerns.

  • One would be to levy NICs at a reduced rate, as a simple, pragmatic compromise.
  • Another would be to levy NICs at a higher rate on the kinds of pensions that are typically financed more by employer contributions, such as defined benefit pensions, than on other pension income.
  • The IFS-led Mirrlees Review proposed introducing NICs at a low rate on pension income and gradually increasing it, while immediately removing NICs from employee (and self-employed) pension contributions so that future contributions were not taxed twice. In the long run, this would bring the NICs treatment of pension contributions into line with the income tax treatment, with up-front relief on all contributions and tax levied on pension income instead (and with the added advantage of moving further towards the integration of income tax and NICs) while – for better or worse – not imposing a big, unexpected tax rise on current pensioners and those approaching retirement. However, such a transition could take decades, opening up the political risk that future governments might not follow through with the plan. And the transitional arrangements would mean that, while the reform generated significant revenue in the long run, it would actually cost money up front.

To the extent that NICs are imposed on pension income – whether that be in full or at a reduced rate or gradually increased – they should be reduced/removed from employee pension contributions in order to prevent double taxation of future pension contributions.

Option: Reform taxation of inherited pension savings

Pension wealth that is bequeathed at death is highly tax-favoured relative to other assets. It is not subject to inheritance tax. And if the pension saver dies before age 75, then the person inheriting the pension is not liable for income tax on money received from the pension.

It is hard to find a good reason for either of these tax advantages. They unfairly favour those who inherit pension wealth rather than other forms of wealth, and inefficiently encourage people to keep their wealth in pensions. They create the absurd situation where people have an incentive to use anything except their pension to pay for their retirement, so that their wealth can be bequeathed in a tax-advantaged way. This has not so far been a major concern, because most of those dying had already converted their pension wealth into a regular income in retirement; but the introduction of ‘pension freedoms’ in 2015, essentially removing the requirement to do that, makes this a potentially viable course of action for future pensioners and potentially expensive for the government in future.

Bequeathed pension pots should be brought within inheritance tax, and income tax should be levied on money received from an inherited pension regardless of the age of the pension saver at death. And it should be done swiftly, before even more individuals place significant sums into defined contribution pensions in the expectation that they will be able to bequeath them on favourable terms. 

Option: Change limits on pension saving

Since 2010, the government has reduced the annual and lifetime limits on tax-privileged pension saving. This has substantially reduced the overall cost of pensions tax reliefs, and in particular reduced the extent to which those on very high incomes and those wanting to save relatively large amounts in a pension can benefit from tax reliefs.

It is reasonable that the government may only want to subsidise pension saving up to a point. Tightening the limit on the amount that can be saved in a registered pension over a lifetime is one way of doing this. However, there are drawbacks to the lifetime allowance, which would be exacerbated if it were lowered. The lifetime allowance applies to the value of the fund accumulated, including investment returns as well as pension contributions. Individuals can therefore inadvertently exceed the lifetime allowance even if they stop actively contributing while still some way below it. This is opaque and the penal tax rates that result from exceeding the lifetime allowance are arguably unfair. Where people do understand the risk of breaching the allowance, it can distort their behaviour – for example, leading them to invest their pension savings in lower-risk, lower-return assets or even to start withdrawing money from their pension earlier, in order to avoid accidentally exceeding the limit. 

Reducing the annual allowance makes even less sense than reducing the lifetime allowance. For a given level of lifetime contributions, it is not clear why the government would want to penalise making occasional large contributions rather than frequent smaller ones. Indeed, from an individual’s perspective, it may be preferable to contribute considerably more when earnings are high than to contribute a steady amount each year of working life. The annual allowance is therefore harsher on individuals who have particularly uncertain or volatile incomes – such as the self-employed – than on others. 

If the government wants to reduce the amount of subsidy to pension saving available to each person, there are probably better approaches than further reducing the amounts that can be saved in a registered pension scheme. For example, the government could set a lower cash cap on the tax-free lump sum, and then increase the overall amount that can be saved in a pension given that the extra pension saving would not attract this tax break.

In recent years, the government has reduced the annual allowance especially sharply for very high earners, in an attempt to raise revenue from this well-off group. (The annual allowance is gradually reduced for those whose income including pension contributions is above £240,000, falling from £40,000 to £4,000 for those whose income is above £312,000.) However, this is a very complicated way of achieving this objective, with many unintended consequences. For example, it is very inflexible for people in defined benefit schemes, who typically have less control over the amount deemed to have been contributed to their pension each year and can find themselves facing large unexpected tax bills and/or strong disincentives to work more. More fundamentally, it is not clear why very high earners should not be allowed to contribute as much to a pension (and get as much tax subsidy) as merely high earners. It would be better to remove the ‘tapering’ of the annual allowance (so that it is the same for everyone) and, if desired, raise more revenue from very high earners in another way.

Option: Better align limits for defined benefit and defined contribution pensions

The lifetime and annual allowances are more generous for those in defined benefit schemes than for those in defined contribution schemes. 

For example, the 2021–22 lifetime allowance of £1,073,100 would enable a 65-year-old, at 2021 annuity rates, to take a tax-free lump sum of £268,275 and receive an RPI-linked annual pension of about £22,200 (or an annual pension fixed in cash terms of about £39,500) from their defined contribution pension. In contrast, for someone in a defined benefit pension arrangement, a £268,275 lump sum and an annual RPI-linked pension of £40,200 – nearly twice the maximum defined contribution pension – is deemed to be equivalent to a pension pot of £1.07 million (since defined benefit pension schemes are deemed to have a pot size 20 times the annual pension).

This difference is hard to justify. It is a consequence of the crude and inaccurate way that wealth held in, and contributions to, defined benefit pensions are estimated from an individual’s rights to future pension income. The limits for defined benefit and defined contribution pensions could be better aligned by increasing the factors applied to convert expected defined benefit pension income into pension wealth (for assessment against the lifetime allowance) and to convert increases in expected defined benefit pension income into deemed annual pension contributions (for assessment against the annual allowance). However, given that these factors would have to be increased substantially (nearly doubled) to achieve full alignment, the government would also need to consider increasing the annual and lifetime allowances, otherwise many more people in defined benefit schemes would hit the annual and lifetime limits. In other words, the allowances could effectively be made less generous for defined benefit pensions and more generous for defined contribution pensions.

This article is part of a project funded by the Trust, more information is available here.