Buried in two new official reports is evidence of a defect that is warping British democracy. Fortunately, it would cost little to correct; so the Treasury can relax. As it happens, reform would also remove a clear, if unintended, bias to the Conservatives. It would make British politics fairer.
The case for reform flows from a basic flaw in the way the new constituency boundaries have been drawn up for the next election. We have known for some years that electoral registers across Britain miss out millions of eligible voters. What the new data show is where the missing voters live. They are concentrated far more in Labour seats than Conservative seats. This affects the way constituency boundaries are drawn. This in turn significantly affects the numbers of seats each party wins.
We should not be surprised that the bias has passed unnoticed, for it requires separate spreadsheets from different official sources to be combined. Separately they look innocuous; combine them and mix in figures for party support in each new constituency, and the source and size of the bias become clear.
Removing the bias would reduce the Conservative majority by 22. In 2019, it won 80 more seats than all other parties combined. The boundary changes have increased that to 96. But if the Boundary Commissions had been charged with equalising the populations of each constituency rather than electorates, the Tories would be defending a majority of 74 at the next election, down six, not up 16, on their majority four years ago. A 22-seat variation in a close election could well be decisive. Indeed, the bias in the present system helped the Tories over the line in three of the last four general elections.
To see why, we need to compare two separately published figures for each of newly-revised constituencies that will send MPs to Westminster at the next election. The electorate figures have been supplied by the Boundary Commissions, while the population figures, and detailed demographic data for England and Wales, were quietly released by the House of Commons Library in mid-August.
According to the 2021 census, the total population of mainland Britain was 65 million. This includes 13.5 million people under the age of 18, leaving 51.5 million residents of voting age. However, just 46.3 million are registered to vote – a shortfall of 5.2 million. This is not an exact figure for eligible adults missing from the register. Some residents counted by the census are not entitled to vote. On the other hand, the census itself misses out some people. Estimates have to be made for the numbers who don’t complete the census form; and, of course, neither the census nor the register tell us how many people live in Britain illegally.
However, the 5.3 million shortfall is more likely to be an underestimate than an overestimate of the eligible adults missing from the register. Four years ago, the Electoral Commission conducted a detailed study of the quality of the register and concluded that the parliamentary register was “85 per cent complete”. The missing 15 per cent represented 7.7 million adults.
The Commission found that these missing adults were far more likely to be found among:
- Ethnic minority than white residents
- Private renters than home owners
- Younger than older adults
The Electoral Commission told the national story. Now we are able to see how the shortfall varies from seat to seat and affects Britain’s political geography. Our new constituencies are supposedly “equal”, yet some MPs are serving the needs and tackling the problems of far larger populations than others. The new figures show not only that the variations in population are far greater than those in electorates – but that the variations match the findings of the Electoral Commission.
If we take the 575 new seats in England and Wales (for which the House of Commons has provided detailed demographic data) and omit the three constituencies where the specific rule for equalising electorates has not been applied (Ynys Mon and the two new constituencies in the Isle of Wight), electorates range from just under 70,000 (South Shields) to just over 77,000 (North Shropshire). All new seats have electorates within 5 per cent of the median of 73,500. This is what the legislation requires
In contrast, populations range from 85,000 (New Forest West) to 152,000 (Birmingham Ladywood).. A part of this vast discrepancy is caused by the age profile. New Forst West has around 13,000 children who are too young to vote; in Birmingham Ladywood the number is 39,000. That leaves New Forest West with 72,000 adults (very close to its electorate of 71,000), while Ladywood has 113,000 adults, a far cry from its electorate of less than 77.000.
This gulf would not surprise anyone examining the Electoral Commission’s report: Ladywood not only has far more children than New Forest West, it has far more ethnic minority residents and far fewer owner-occupiers and people over 65.
These two constituencies are examples of a wider pattern. Here are the figures for (a) the 50 seats where the figures for the electorate most closely match the adult population and (b) the 50 seats where the number of adults exceeds the electorate most.
Ethnicity:
- Seats where electorate and adult population are most similar: 91 per cent white, nine per cent non-white.
- Seats where the adult population exceeds the electorate most: 51 per cent white, 49 per cent non-white.
Housing:
- 41 per cent own their homes owned outright, 14 per cent rent privately.
- 19 per cent own their homes owned, 34 per cent rent privately.
Age:
- 46 per cent are over fifty years old, 18 per cent aged 18-34.
- 27 per cent are over fifty, 31 per cent aged 18-34.
The impact of these differences on the parties is stark. The seats with the fewest unregistered adults are classic Tory territory: notably older and whiter than the average, and with more people who own their homes outright.
In contrast Labour is strong in the seats with the most unregistered adults: younger and ethnically more diverse and with many more people who rent their homes.
What if we equalised constituency populations rather than constituency electorates? This would mean each MP representing broadly the same number of residents, including children (who, though too young to vote, consume services that tend to figure prominently in MPs’ caseloads, such as housing, health and education.)
The average population of constituencies in England and Wales is just over 103,000. But in Conservative-held seats the average is 97,000, while in Labour seats it is 114,000. With equal populations, Labour would gain 11 extra seats, while the Tories win 11 fewer.
Source for 2019 result under new boundaries: Greg Cook*
The changes would make little national difference in a landslide election. But had a population-based system been in operation it might have prevented David Cameron winning outright in 2015, and forced Theresa May out in 2017. Back in 2010, Gordon Brown might even have been able to remain in office with Liberal Democrat support. Had any of things happened, the implications for Britain would have been profound – not least for Brexit. A close election next year could also have very different political consequences under a population-based system of setting boundaries.
It’s not only the political balance of the House of Commons that would be altered. So would the geographical balance. Londoners are younger, more diverse, and far more likely to rent privately than people in any other region. As a result, the population of the average London constituency, more than 117,000, is way above that of any other region, and 14,000 above the national average. It will elect 75 MPs at the next election. In population terms, it should elect 85. This is where we would find the lion’s share of the extra Labour MPs that would be elected under this system.
I can envisage two non-party-political objections. (Partisan views shouldn’t count when designing democracy, even if MPs with a clear vested interest in the matter are bound to work out what reform would mean to them personally.)
Objection 1: If a population-based system is such a good idea, wouldn’t other countries be doing it already? As a matter of fact, they are. The number of seats each US State has in the House of Representatives is determined by the results of each decadal census. The parliaments of France, Germany, India, Canada and New Zealand also have census-based systems. Australia is an outlier: it counts “enrolled voters”, which is like Britain, but Australia misses only 5 per cent of eligible adults, compared with Britain’s 15 per cent. For Britain to switch to a population-based division of constituencies would be to adopt best international practice.
Objection 2: Whatever other countries do, we should honour our own traditions, not trash them. I’m glad you raised this point. In 1885, in one of the series of democratic reforms of the Victorian era, Parliament decided to back the principle of equal populations. Indeed, it was the Conservatives who pushed hardest for this. This remained the case for sixty years, when the switch was made to equalising electorates. A Government committee recommended the change on the basis that, with the advent of universal adult franchise, “the two criteria produce virtually identical results” for adults.
This in now plainly not the case. For once, traditionalists and reformers can march hand-in-hand, demanding a return to one particular Victorian value. Eighty years ago, in one vital respect, British democracy lost its way. Parliament should seize the opportunity as soon as possible to find it again.
* Greg Cook is an independent election analyst and former head of political strategy for the Labour Party
Peter Kellner. The Politics Counter 