To provide some context for our Wealth Gap Risk Register, we wanted to answer one key question: how big is the wealth gap in the UK, and what has happened to it in the last few years? To find out, we asked Ben Tippet, an economist at King’s College London, to look at the data. The Flourish ‘story’ below sets out what he found. There’s a full methodology underneath.
Methodology
This report is based on two sources of data: waves 3-7 of the Wealth and Assets Survey (WAS) provided publicly by the Office for National Statistics (ONS), and the 2011 to 2019 Sunday Times Rich List (STRL). Wealth is defined as the sum of financial, physical, property, pension and business assets minus their debts. Individual-level data is used.
All wealth values are adjusted to control for inflation. The CPI-H from the ONS is used. Wealth for all years is expressed in 2019 £ values.
The ONS wealth survey data significantly underestimates wealth at the very top of the distribution, as wealthier households are less likely to respond to the survey (differential unit non-response), may under-report their wealth (item non-response) or may simply be missing from the sample (ONS, 2022). For example, the wealth of the richest person in 2018-2020 is £78m, which is less than that of the least wealthy person on the STRL (£100m). For this reason, it is standard practice in the academic literature to combine STRL and survey data and to fill in the missing data by assuming that the top tail of the wealth distribution follows a Pareto type 1 distribution (Vermeulen, 2018). This is done for all waves.
This report does this by taking the 99th percentile (individuals with wealth above £2.1m in 2018-2020) as the cutoff at which the data is adjusted. Using the methodology outlined by Vermeulen (2018), it finds that £2.25 trillion in wealth is missing from the top 1% of the wealth distribution in the latest year. The top 1% share of wealth increases from 15% to 25% – to a level more broadly consistent with that of the World Inequality Database, which has a top 1% wealth share of 21% in the UK for that year.
As the WAS conducts its survey waves in periods of two years, we take the middle year as the reference point (e.g. 2011 refers to the 2010-2012 WAS wave). We ignore the first wave of the WAS (2006-2008) due to data issues in collecting physical wealth, and the second wave (2008-2010) due to data issues with individual-level wealth holdings. We therefore treat the third wave of the WAS (2010-2012) as the baseline. This also helps to control for the high volatility of wealth values during the financial crisis, particularly at the top.
All means, medians and totals are weighted using cross-sectional survey weights.
The WAS does not include business assets, nor does it combine its data with STRL data, so these findings may not be directly comparable.
This dataset includes all individual respondents to the survey, and has not been processed to exclude children. As a result, the mean and median wealth figures for each decile are lower than the ONS publications, which use household level data.
We keep in individuals with negative wealth, but we do not assume a growth rate for deciles where the mean/median has a negative value.