United Kingdom

Persistent low income

Graphs on this page:

Supporting information:

Key points

  • A sixth of the population – around 10 million people – are in the poorest fifth of households at least two years in three.
  • Most households who spend some time in the poorest fifth move in and out of the poorest fifth, with only a minority (around 20%) persistently in the poorest fifth.  One implication of this is that around a third of the population have, within the latest four-year period, been in the poorest fifth in at least one of the four years.

Why this indicator was originally chosen

The duration of time spent on a very low income can have a considerable effect on the deprivation of a person or family.  Some people move in and out of low income whilst others remain persistently in low income.

As well as those continuously on low income, this indicator begins to measure the extent of those intermittently on low income, experiencing what is usually referred to as ‘income churning’ – repeated movements into and out of low income.

Since it is based not on monitoring people continuously but only once a year, the indicator under-estimates the numbers who are ‘churning’.  It is also important to note that, while many individuals do alter their position in the income distribution from year to year, most changes are small. 1  Income ‘churning’ is generally not about moving from low income in one period to rather high income in the next, but of income changes of perhaps just a few pounds a week. 2

Definitions and data sources

For each three-year period, the first graph shows the proportion of people who were in the poorest fifth of households in at least two years out of three.  The bars are split to show those in the poorest fifth in all three years, those in two consecutive years only, and those in the first and the third year only.

The second graph looks at transitions in and out of the poorest fifth.  For the latest four-year period, it classifies individuals who have experienced at least one year of being in the poorest fifth of households in that period according to the number of times they switched into or out of the poorest fifth.

The data source for both graphs is the British Household Panel Survey (BHPS) and relates to Great Britain (the usual dataset for analysing low income – Households Below Average Income – cannot be used as, unlike BHPS, it does not sample the same households each year).  Income is net disposable income before deducting housing costs 3, equivalised (adjusted) for the size of the household 4.  Note that the labelling is done according to the year that the relevant surveys ended although the actual surveys are from September to May.  So, for example, the label “2003-05” refers to the three surveys September 2002 to May 2003, September 2003 to May 2004, and September 2004 to May 2005.

Previous versions of the two graphs in this indicator used a low income threshold of “60% of median household income” rather than “in the lowest fifth of household incomes”.  After much analysis and discussion, it has been decided that the time trends using such a threshold are not reliable.  This is for a number of reasons, namely:

  • A relatively high number of the households in BHPS have a household income which is close to the 60% of median threshold.  Given that the sample size of BHPS is relatively small, with consequent uncertainties about the precise level of median household income, it effectively becomes a matter of chance how many of these households in a particular year fall either just below or just above the 60% threshold.
  • The time trends in the first graph when using a 60% of median threshold are rather different depending on what methods (e.g. OECD or MacClements) are used to adjust the household incomes for household size and composition, with the differences being much greater than might have been expected.  This could either be because of the bullet point above or because there is something a bit odd with how one or other of these methods has been implemented in the BHPS dataset.
  • The results of analysing the BHPS dataset using the 60% of median threshold are rather different from those published by the Department of Work and Pensions (DWP) in its Low income dynamics publications (which is based on the BHPS dataset).  The basic reason for this is known: DWP has not been using the standard household income data from the BHPS dataset, apparently because it was not aware that it existed.  It is, however, not clear why the DWP’s methods should produce materially different results.  In summary, it would appear that there are problems with how either DWP or the BHPS team is doing its calculations. 5

Use of a low income threshold of “in the lowest fifth of household incomes” removes the first of these problems.  It also somewhat ameliorates the impact of the other two problems by effectively assuming that the number of households in low income is constant over time.  Note, however, that the resulting trends over time are therefore purely an indication of how often people are moving in or out of the poorest fifth, whereas the trends using a threshold of “60% of median household income” would be an indication of the combined effects of how often people are moving in or out of low income and how the total number who are in low income is changing over time. 6  Which of these is the more insightful is a matter of opinion, but the important point to understand is that they are certainly different.

Overall adequacy of the indicator: limited.  As well as the issues discussed above, BHPS is a much smaller survey than that used by Households Below Average Income, suffers from a loss of members over time, and the remaining members by definition change in nature over time (as they are growing older).  The data is also not directly comparable with the low income data in the other indicators on this website.

External links

See the 2006 Joseph Rowntree Foundation report entitled The persistence of poverty across generations.

Relevant 2007 Public Service Agreements

Overall aim:  Halve the number of children in poverty by 2010-11, on the way to eradicating child poverty by 2020.

Lead department

HM Treasury.

Official national targets

Reduce by a half the number of children living in relative low-income by 2010/11.

Other indicators of progress

Number of children in absolute low-income households.

Number of children in relative low-income households and in material deprivation.

Previous 2004 targets

Halve the number of children in relative low-income households between 1998/99 and 2010/11, on the way to eradicating child poverty by 2020, including:

  • reducing the proportion of children in workless households by 5% between spring 2005 and spring 2008; and
  • increasing the proportion of parents with care on Income Support and income-based Jobseeker’s Allowance who receive maintenance for their children by 65% by March 2008.

By 2008, be paying Pension Credit to at least 3.2 million pensioner households.  While maintaining a focus on the most disadvantaged by ensuring that at least 2.2 million of these households are in receipt of the Guarantee Credit.

The numbers

Graph 1

Years All 3 years 2 consecutive years 2 non-consecutive years
1992 - 19949% 7% 2%
1993 - 1995 10% 7% 2%
1994 - 1996 10% 6% 2%
1995 - 1997 10% 7% 2%
1996 - 1998 11% 6% 2%
1997 - 1999 10% 6% 2%
1998 - 2000 10% 6% 2%
1999 - 2001 10% 6% 3%
2000 - 2002 10% 6% 2%
2001 - 2003 9% 7% 2%
2002 - 20049% 7% 2%
2003 - 20059% 7% 3%
2004-20068% 7% 3%
2005-20078% 7% 2%

Graph 2

Number of transitions in and out of the poorest fifth over a four-year period Percent
Persistently in the poorest fifth (no changes) 18%
One transition 44%
Two transitions 31%
Three transitions 7%
Total 100%
1. Hills J, Income and wealth: the latest evidence, Joseph Rowntree Foundation, 1998. 
2. Churning has a number of causes.  The most common reasons for changes in household income are rises or losses of earnings by household members, though this is not always the head of household, but also spouses and even children.  Changes in receipt of state benefits also account for a part of the picture.  Longer term moves into or out of low income are often caused by changes in the composition of a household, for example children leaving home or being born.  For women and children, marital separation and divorce often cause considerable falls in income, whilst for husbands real income tends to change much less. 
3. Incomes after deducting housing costs are not available in the dataset. 
4. Using the same OECD method of equivalisation used in all the low income graphs on this website. 
5. We made DWP aware of this problem in early 2008.  They decided that they would need to meet the BHPS team to discuss and resolve it but our understanding is that it has not yet been resolved.  It will remain unresolved until and unless DWP and the BHPS team agree a common method of analysis. 
6. For example, if the total number of people in households below 60% of median household income were falling over time, then this would tend to result in a downward time trend for this indicator if it used the “60% of median household income” threshold but would have no obvious effect using the “in the lowest fifth of household incomes” threshold.