Wealth inequality and the Hills Report: a critical assessment

This think piece by ASI fellow Tim Worstall critically examines the National Equality Panel’s ‘Hills Report’, with particular emphasis on its treatment of wealth inequality and the gender pay gap. He argues that not only have the report’s authors directly ignored Office of National Statistics guidelines on how to measure the gender pay gap, but that they have also hugely overstated income and wealth inequality in the UK by failing to take account of the effects of the welfare state.


The National Equality Panel recently released its inaugural report. Called the Hills Report after the panel’s chairman, it manages to specifically and directly ignore strictures offered by the Office of National Statistics (ONS) on how to measure the gender pay gap, overstates income inequality and wildly and grossly overstates wealth inequality. They have also ignored all of the things that we already do to try and narrow the wealth gap.

The major claim picked up by the newspapers – probably because it was emphasized in the National Equality Panel’s own publicity about the report – is that “household wealth of the top 10% of the population stands at £853,000 and more – over 100 times higher than the wealth of the poorest 10%, which is £8,800 or below”(1). A more accurate estimate of that wealth gap would be somewhere between five and ten times higher. The error comes from the report’s authors ignoring the major source of both income and wealth for the poor in our society: the welfare state.

Minor matters

It has long been a favourite trick of those who talk about the gender pay gap to conflate various extant pay gaps and attribute them all to gender. This is political prestidigitation with statistics to make the problem seem larger and more important than is actually the case. Greater problems are more likely to have political attention paid to them. And so with the gender pay statistics, a running theme has been to add together two quite different things.

The first is the real gender pay gap. It is true that women in general are paid less than men in general. While the methods used to calculate this are still somewhat suspect – taking no account of different levels of human capital, choices about jobs done and so on – the Annual Survey of Hours and Earnings (ASHE) from the ONS suggests that a gap does exist. On the basis of hourly wages without overtime it is certainly the case that women get paid less than men on both median and mean average measures.

The second is the part-time pay gap. Those who work part-time tend to receive less compensation per hour of labour than those who work full-time. We can see this from the ASHE statistics. As with the gender pay gap, the reason this is the case is a complex matter. But it is indeed the case that men who work part-time get paid less than men who work full-time.

There is a third fact we must recognize: more women than men work part-time. This is almost certainly a result of the way in which women in our society do most of the childcare (and indeed, caring for other family members as well). Whether this is how one would wish the world to be is less important than the recognition that for now, this is the way it is.

Putting these three things together we can see that if we compared the wages of all men with those of all women then we would not be describing the gender pay gap. We would be describing both the gender pay gap and the part -time pay gap. If were to call this the gender pay gap we would be misleading those to whom we were presenting our statistics, whether deliberately or through ignorance.

After some years of various people doing exactly this, the UK Statistics Authority, in the form of Sir Michael Scholar (2), wrote to the Equalities Minister, Harriet Harman, to point out that this technique “risks giving a misleading quantification of the gender pay gap”. If we look at the statistics one can see his point. The average Ms. Harman was using suggests a 23% gender pay gap while the ONS puts it at 12.8%. The ONS statistics use only full-time workers (and thus measure the pure gender pay gap) while the former figure conflates both gender and part- and full- time. It is thus inaccurate for the government and pressure groups to use the 23% figure, but this does not stop them doing so.

Sir Michael’s actions were a serious injuction and as close as we get to a mandarin slapping wrists – which is why it is a surprise to see the Hills Report doing the same thing just six months after this public admonition. In the report’s discussions of hourly gender pay, figures for all men and women working are used, both full- and part-time. This gives us a 21% pay gap (we have a new set of ASHE statistics to calculate it from). And that, as Sir Michael pointed out, is a “misleading quantification of the gender pay gap”. The report does go on to look at only full-time workers, but it does so in the context of weekly wages, not hourly ones. The authors note that women tend not to work as many hours a week as men but then sail on happily to ignore their own point and point to the gender pay gap again without adjusting for differences in such working hours.

To put it mildly, this does not engender confidence in how they have presented other information in the report. Why would a report to Harriet Harman depend upon a statistical method which she herself has been expressly told not to use?

A digression

It is extremely important that when we look at a problem we make sure that the information we have about it is accurate. It is not enough just to look at what the current situation is; we must also look at the effects of what we are already doing about it, or attempting to do about it. After all, there is the possibility that our current actions are making the problem worse. For example, with the gender pay gap there is at least some evidence that longer maternity leaves increase the size of the gap.

In our discussion of income and wealth gaps, however, the story of John Edwards is instructive. No, not the more tawdry tales of recent months, but his Vice Presidential campaign in 2004 and his more recent attempt to gain the Democratic nomination for President. In both, his stump speeches would point to the US poverty rate (at that time 12 and 13%) and insist that a wealthy nation could do better. Obviously, it is possible that it could, and whether it ought to is certainly a valid political question. Edwards would then go on to propose the following three policies.

  • Firstly, a substantial rise in the Earned Income Tax Credit (EITC). This is very similar to our own Working Tax Credit system, which is not that surprising as our system is based on the US one.
  • Secondly, a further one million Section 8 housing vouchers. This is similar in intent to our Housing Benefit. As with our own experience with council estates, the Americans realized that dispersing the poor through the wider society avoided creating pockets of deprivation like the vertical slums put up by the US Department of Housing and Development and now referred to as “the projects”.
  • Thirdly, a large expansion of Medicaid, the US programme that provides health insurance to the poor.

However, these policies would not change the number defined as being in poverty by one single person. If spending on these programs was doubled, or even tripled, there would still be that 12-13% of the US population ‘in poverty’. This is because there is something very strange about the way that poverty is defined in the US.

The poverty line in the US is an absolute number: three times the food budget for a family in the early 1960s up-rated for inflation. It is not, like the poverty line in all other OECD states, a relative measure – we use less than 60% of equivalized median household income as our measure, for example. More importantly, the US poverty number includes in the income that qualifies you as poor only market income and direct cash grants. It does not include any poverty alleviation that comes through the tax system and does not include any aid that comes in kind.

The EITC is paid through the tax system, while health care and housing are both benefits in kind. Therefore they are not included. It is possible to raise the provision of these things, which do indeed reduce poverty, without changing the number under the poverty line by one iota. Indeed, this is largely what the US has been doing since the 1970s. The EITC, for example, was introduced 1975. Before this the major poverty alleviation program was what we now call “welfare” – simple cash payments to those who didn’t have enough money. A bipartisan change came in around the time of the EITC determining that we should provide benefits to the poor not in cash, but either through the tax system or as benefits in kind.

All of these programmes have been expanded in the decades since then, and the US poverty rate has barely changed. This is not a surprise when the system doesn’t actually take account, when measuring poverty alleviation, of the things that are done to alleviate poverty. One estimate is that the $40 billion a year spent through the EITC lifts 5 million families above the poverty line. Yet in the official statistics they’re still under it.

It is possible to look at the US poverty figures and conclude that something went drastically wrong in the mid 1970s. The number under the poverty line had been falling sharply since World War II and then essentially flat-lined. Indeed many do look at those numbers and then leap to the conclusion that it was Reagan, or the neo-liberals, or some other hated figure or group, that caused poverty levels to stop falling. A more accurate analysis is that before this date we included all the effects of what was spent upon poverty alleviation before calculating how much poverty there was. Now we include only a minor portion of it. It is therefore of little surprise that spending on poverty alleviation has continued to rise without there being much change in the poverty rate.

This is not just some recondite point about statistics. If you were a politican, like John Edwards, you could advocate massively increased spending on poverty alleviation while knowing that this increase would have absolutely no effect upon the metric you were using to measure the problem. Which politician would not like to have an issue which calls for ever greater effort but which can never be solved?

Ending the digression

Of course, we in the UK don’t do anything quite so silly in our measurements of poverty and what we spend upon its alleviation. We use total income as our measurement: income after all taxes and benefits have been included. Don’t we?

Well, actually, no, we don’t. In the UK have universal healthcare, the NHS. Everyone is entitled to the treatment they need – as long as it’s not too expensive – as of right. This matters little when we try to look at static levels of inequality or poverty here since everyone has access to the same services. However, if we want to compare either income or wealth inequality over time, then it needs to be taken into the calculations.

Furthermore, we need to look very carefully at the Hills Report’s method of looking at income and wealth inequality. For in fact, in their look at wealth inequality, they manage to entirely ignore not just the effect of the NHS but the entire apparatus of what we call the welfare state. When we include these effects, far from the wealth gap being 100:1 a more accurate estimate would be 5:1 or 10:1.


In the following sections I am not going to try and explain the intricacies of each and every aspect of the welfare state. Who is eligible for what under which precise circumstances is not part of the remit of this paper. I intend to take just a few examples to show the point. I am also not aiming to state an authoritative set of figures. The aim is simply to explain the basic idea, and thus the flaw in the Hills Report’s estimates.

There is also one conceptual point that needs to be understood: a stream of payments that you are entitled to is indeed wealth. You may not be able to sell it, you may not be able to transfer it, but we can and do calculate what the capital value of that stream of payment is.

Consider, for example, a private pension plan. At some point, by law, you must convert this into an annuity. An annuity is a stream of payments and in this case one that will last for the rest of your life. As a rough example, these days, a payment of £100,000 will secure, for a 65 year old man, a stream of payments of £6,000 per year until his death. It is thus entirely logical to turn around and run the calculation the other way: a stream of payments of £6,000 a year until death for a 65-year old man has a capital value of £100,000. It is, in short, wealth of £100,000.

Thus, if the State old age pension were £6,000 a year, then each and every person eligible for that pension would have wealth of £100,000. And it is this point which the Hills Report has failed to include in its calculations of the wealth gap. The welfare state provides almost all of us with some such wealth. It provides some of us with very much more than others. The wealth gap is thus closed by the fact that we have this welfare state. By ignoring it, we make the John Edwards mistake, failing to account for what we are already doing to reduce the perceived problem.


As Polly Toynbee says in one of her recent columns (3):

The excellent plan in the green paper suggests a long-term solution, fair and sensible. On retirement anyone with the money would pay a lump sum – around £20,000 – to cover all future care, at home or nursing home. They would never need to pay another penny. If they own a home but have no money, the sum can be taken from their estate after death. Those with neither savings nor property would be paid for by the state.

At present the system is that the poor get such services free (as they would in the future), while those with either high incomes or high levels of wealth have to pay for such services. There are many stories of people running down their savings to pay for care homes, with the local council picking up the bill once the money runs out. Or of people selling houses to meet such care bills. What the green paper proposal gives us is an estimate of the capital value of that option that the poor have and the rich do not. It is only an option of course, since not everyone moves into a care home: even now many of us die in our own beds. But the value of that option is clearly £20,000. Those who have that sum must pay it, those who do not have it do not: but both get the same services. Not having to pay thus equates to wealth of £20,000.

Old age pension

Puzzlingly, the Hills Report considers only private pension plans to be wealth that should be counted. They look at “personal marketable wealth” and refer to Wealth and Assets Survey (4), in which we find the following two points being made:

The wealth from pensions in receipt is calculated as the present value of the future income stream that the individual will receive.

This is the same method that I am using to provide an estimate of the Net Present Value (NPV) of other state derived income streams.

We are also told that:

The figures in this chapter relate to private pension wealth only. Since wealth from state pensions is more evenly distributed than wealth from private pensions (as a result of the contribution and benefit formulae), the distribution of total pension wealth (i.e. state plus private) will be more even than that described in the figures below.

How interesting that everyone decides not to use the state pension in the calculation of assets or wealth then, isn’t it? As a rough idea of what the state pension is worth at a net present value, assume that life expectancy at 65 is 15 years. This isn’t too far off the truth. Also assume that the pension itself is £5,000 a year. It isn’t, it’s less than this, but the minimum pension guarantee (which is going to be received by those with no other wealth) is higher, so to a certain extent it all comes out in the wash. The other thing we need to know to get an NPV is the discount rate. Currently the long gilt yields are just over 4%, which would give us an NPV of £60,000. However, the state pension (in contrast to the usual annuities bought with private pension plans) is inflation upgraded each year. So the more appropriate interest rate would be that on inflation indexed gilts, around 1% at present. This gives us an NPV for the state pension of £75,000.


As a round number it costs £2,000 a year to provide the NHS to each adult in Britain. There are indeed countries where health care must be purchased out of disposable income: thus not having to pay this insurance bill for the NHS is a source of wealth. Assuming only a 25 year receipt of this free at the point of use health care (these are all simply examples, remember, not attempts to provide strictly accurate numbers) this has an NPV of £46,000, again at that 1% interest rate. As the poor in income are net receivers of benefits from the tax and benefit system it seems logical to state that they thus have an asset of £46,000 or more.

There are many more such benefits available as part of the welfare state, but rather than try to provide an exhaustive list (education free at the point of use, etc, etc, all things which in other countries or at other times have had to be paid for), here is just one more example…

According to the Hills Report a household in the bottom 10% of the income distribution has an average market income of £4,700 a year. After benefits (but not including the education, pension or health care benefits above) the income rises to £14,300. Call that £10,000 a year as a round number and at our 1% discount rate for 25 years that income stream has an NPV of £228,000.

Marketable and non-marketable wealth

It is true that the Hills Report tries to, or at least says it tries to, look at marketable wealth. But private pension plans are not marketable: you’re not allowed to sell them. So they are at least slightly in breach of this idea. But much more important than this is that marketable wealth isn’t the important determinant.

For example, owning a house which you can sell is to have marketable wealth. Living in social housing or being in receipt of housing benefit is not marketable. But both are wealth: both lead to you living in better housing than you would without that asset. And if you sell your house and bank the money you will not get social housing or housing benefit. Only those who do not have marketable wealth receive either. And the wealth that comes from subsidized social housing or, indeed, housing benefit itself can be calculated.

There is one other point to make about this “wealth” that comes from the welfare state. All of these things, the state pension, the NHS, help with housing, benefits such as jobseekers’ allowance, tax credits and the rest of the system: all of them are supported precisely because they make the poor wealthier. That is their purpose. To suggest that while we spend a couple of hundred billion pounds a year on the entire system, we cannot count, and should not include the increase in wealth which it provides seems remarkable, if not unsupportable. It is to make the John Edwards mistake again.


The Hills Report states that the wealth gap between the 10th and 90th percentile is of the order of 1:100. It says that the average 10th percentile household has assets of £8,800 and the 90th £853,000. But it is only possible to reach this conclusion by ignoring all of the things that we already do to redistribute wealth.

Just as we do with income inequality, we should measure wealth inequality after the influence of the tax and benefit systems. The benefit system provides a number of income streams to the poor and we can calculate their net present value by treating them as an annuity.

Combining the value of just the NHS and the state old age pension every individual has wealth of over £100,000. This must of course be added to the wealth of both poor and rich but it brings the 90:10 wealth ratio down to 10:1.

Looking purely at the income support available to an average 10th percentile household the value of their annual receipts from the welfare state is some quarter of a million pounds when capitalized. This lowers that 90:10 wealth gap to somewhere under 5:1.

It much be stressed that all of the calculations and examples here are only indicative. No attempt has been made to develop authoritative statistics on wealth inequality. But my examples nevertheless clearly illustrate the argument presented here.

Overall then, and in conclusion, if we are to attempt to measure the wealth gap in the UK, we must measure it after the influence of the tax and benefit system on the wealth gap, just as we do with income inequality.

It might be righteous and just that we should do more to close this wealth gap. It might be that what we already do could or should be done better. But before we attempt to answer those questions we have to analyze what it is that we are already doing and the effects this has on the wealth gap.

The Hills Report neglects entirely to do this and is therefore an unhelpful contribution to the debate.