The power of place

Pay varies widely across England, with average wages in 2024 ranging from £610 per week in Liskeard to £1,130 per week in London. Why does pay vary so much? Are high-earning places like London populated with workers who would earn more no matter where they worked? Or would the jobs in London pay lots more to all kinds of workers?

The answer to this question is important for a huge range of government policies, from regional policy and industrial strategy to housebuilding and transport investment. For example, if workers will earn nearly the same wherever they work then it might make sense to build houses wherever it is cheapest to do so. This would be a world of large ‘people effects’, where the inequality in earnings between places is driven by the distribution of different types of workers. In contrast, if the jobs are much better in some places than others, it would make sense to build the houses within reach of them. This would be a world of large ‘place effects’.

Previous research on the UK has found that earnings are unequal across labour markets mostly because high-earning workers cluster in certain labour markets – i.e. mostly because of ‘people effects’. We use a new dataset and new techniques to revisit this question. We employ the Longitudinal Education Outcomes (LEO) dataset, which comprises nearly every worker born after 1985 who went to school and work in England and merges their educational records, their tax records and their employment history. We can follow these workers as their earnings and careers progress and as they move across employers and between the 155 English commuting zones (so called ‘Travel to Work Areas’, or TTWAs, which we refer to interchangeably as places, regions and labour markets).

We find that a typical full-time early-career worker (aged 22 to 36) moving from a low-paying labour market (such as Dudley at the 25^th centile) to a high-paying one (such as Harrogate at the 75^th centile) would get a pay rise of around £1,300 per year, or 5 per cent of earnings. More broadly, about one-third of the variance in regional average pay is due to ‘place effects’ – that is, employers paying higher wages in some places than in others to otherwise identical workers. This is much higher than the 1-12 per cent that previous studies have found. The difference is due to our focus on younger workers, new measurement techniques, and the fact that we are using a new, much bigger dataset than previous research.

These pay boosts are proportionally similar for both university graduates and non-graduates. What’s more, they appear to be portable to some extent. For early-career workers, a worker who spends a year working in London and then moves to another major city can take a 10 per cent pay boost with them.

A further 42 per cent of regional pay inequality is driven by the sorting of the highest-earning-potential workers (with the highest ‘people effects’) into the highest-paying labour markets (with the highest ‘place effects’). This is an important phenomenon in explaining regional pay inequality, but it has hardly any impact on average earnings (it leads to a boost of around 0.1 per cent).

So we have found that place effects are important. But what in turn do these place effects represent? What drives them? In common with previous studies, we find that bigger labour markets pay higher wages to any given worker (part of the so-called ‘agglomeration effects’). But the effect is small: a doubling of the size of the labour market boosts the pay a typical worker will get by around 3.9 per cent. Labour market size only explains around 24 per cent of the variation in pay premia across TTWAs, so there is much else at play. For example, Cambridge is the same size as Leicester but pays its workers 23 per cent more.

Another reason for differences in pay across the country is that industries are unevenly distributed across labour markets – for example, high-paying finance jobs are concentrated in London, Slough and Heathrow and Luton, while low-paying hospitality is particularly important in Redruth and Truro, Bideford and St Austell and Newquay. If industries pay differently and are unevenly distributed, it could be true that industrial composition explains the variation in pay premia. But it doesn’t: it only explains around 10 per cent, with worker sorting explaining 8 per cent. Instead, we observe that jobs in any given industry tend to pay more in high-paying labour markets. The same is true for firms of different sizes:  bigger establishments pay more, but (again) it is differences in pay premia for firms of any given size that drive the inequality we see across labour markets.

So our new research tells us that the same workers earn very different wages in different labour markets, and this has little to do with how big the labour markets are or the industries they comprise. It’s hard to unpack these place effects much more with the LEO dataset, but we can use other datasets and research from other countries to make progress.

One possibility is that workers in different places are in different occupations. Our work looked at how the industry mix of places explains regional pay differences, but industries, and the firms within them, comprise a wide range of jobs (or occupations). For example, everyone working in Tesco is in the retail industry, but the jobs that workers do at their headquarters in Welwyn Garden City will be very different to those in their superstore in Wigan. The LEO dataset does not measure workers’ occupations, but analysis of the Annual Survey of Hours and Earnings (ASHE) suggests that occupational composition explains only slightly more of the pay premium than industry does – around 14 per cent. However, the occupational classification in ASHE is not very granular – lumping a store manager in the same category as the CEO.

Other evidence strongly suggests that the type of firms and the jobs are key to explaining regional pay premia. First, we know that productivity varies hugely across firms – with firms in the top decile of productivity within a sector being typically more than eight times more productive than those in the bottom decile. Second, recent academic evidence from the US and France shows that the best firms cluster strongly together, and that it is those firms, not the amenities of the place they are in, that explain most of the pay premia in those locations. For example, there are few intrinsic differences between Cambridge and Leicester, but for historical reasons, employers like Arm or Darktrace have chosen to cluster in the former more than in the latter. Third, evidence from the US also documents how large firms have separated their functions across space, for example concentrating all their management or R&D functions in one place. Consistent with this, we show that the density of large corporate headquarters is correlated with labour market pay premia.

Our main new finding – that typical early-career workers earn much higher pay in labour markets with the best firms – implies that there are large benefits to moving workers into these top labour markets.early-career Pay and productivity are highly correlated, so raising total pay is not only important in and of itself, but will raise national GDP. And this suggests several points for policy makers. First, a pro-growth approach to housebuilding would concentrate the new properties, as in recent government plans, in the best-paying labour markets. And these houses should be for workers of all kinds – not just the best-paid ones. In particular, it is important to build social and affordable housing in high-wage areas; to do otherwise would increase inequality with very little benefit in terms of total GDP. Lastly, while part of the higher wages that result will be eaten up in the form of higher housing costs, these are nonetheless worth paying, especially when the benefits for workers can be long-lasting. Moreover, a reformed property tax could capture these higher rents to pay for public services or tax cuts elsewhere.

But relocating workers can only go so far. To further reduce regional pay inequality, well-paying jobs would also need to move to the workers. This doesn’t mean moving whole industries, but rather encouraging productive firms and especially the high-value functions within them – such as their headquarters – to spread to new locations in the country. This has been tried before with mixed outcomes. But our results suggest that the benefits are bigger than previously thought, so researchers and policymakers should revisit this question anew.