Much of the recent policy debate surrounding poverty in Britain focuses on income as a measure of living standards. This report looks at household expenditure as a way of measuring poverty. This alternative to using income data has been largely overlooked in the mainstream poverty debate in the UK.
Economic theory suggests that household expenditure is an important measure of financial well-being. Using 30 years of data from household surveys, this report:
- shows the trends in poverty in Britain since the 1970s when household expenditure, rather than household income, is used as a measure of financial well-being;
- investigates how using spending, rather than income, as a measure of well-being alters our view of who is poor;
- examines the spending levels of the lowest income households;
- analyses whether low-income pensioners' spending on basic and non-basic items increased as a result of the large increases in entitlements to means-tested benefits since 1999.
This study by researchers from the Institute of Fiscal Studies examines trends in poverty over the past 30 years, using expenditure to measure poverty rather than income. It also examines spending levels among the lowest-income households, to assess whether income or spending would give a better guide to their living standards. Finally, it examines whether pensioners’ spending has increased as a result of higher means-tested benefits. Key findings were that:
Recent poverty trends are very different when spending is used as a measure of living standards rather than income. Spending poverty rates (the proportion of people with spending below 60% of median expenditure) have risen since 1997, even though income poverty rates (income below 60% of median income) have fallen:
- spending poverty rose from 20% to 22% between 1996/97 and 2002/03 – up by 12% – compared with a fall in income poverty, from 25% to 22% (down by 12%);
- child poverty rates measured using expenditure rose by 11% between 1996/97 and 2002/03, but fell by 15% when measured using income;
- pensioner poverty has remained roughly unchanged measured on spending, but has fallen rapidly measured on income.
- Further back, trends in income and spending poverty also showed differences. The rise in relative poverty over the 1980s was much smaller when measured using spending than the rise in income poverty.
- Income and spending poverty are distinct states; only about half of those who would be classified as poor by one would be so classified by the other. Self-employed people and job-seekers are more likely to be classified as poor when measured by income rather than spending. For pensioners, the situation is reversed.
- The lowest-income households have disproportionately high expenditure. Among households in the bottom 1% of income distribution, average spending is greater than at any point until one-third of the way up income distribution, suggesting that spending might be a better guide to their welfare than income.
- Though they remain low spenders on average, pensioners have increased their spending – particularly on non-essentials – in response to higher benefits.
Much of the recent policy debate surrounding poverty in Britain has focused on income as a measure of living standards. This report considers an alternative for measuring poverty that has been largely overlooked in the mainstream poverty debate in the UK: namely, household expenditure.
By examining trends in spending poverty, the intention was to learn more about longer-term poverty than by looking at income alone. This is because not all people with low incomes are long-term poor. By borrowing (or dipping into savings) when income is low, and saving (or repaying debt) when incomes are higher, people tend to achieve a more steady standard of living. Spending is therefore arguably a more reliable guide to longer-term resources than income measured in household surveys.
The study used data from the Family Expenditure Survey, the Expenditure and Food Survey and the Family Resources Survey to examine trends in poverty in Britain between 1974 and 2002/03.
Poverty and living standards
Real living standards have increased both over time and across generations, irrespective of whether the focus is on income or expenditure. However, this growth has not been uniform for all people. In the 1980s, households at the bottom of both the income and spending distributions tended to experience lower rates of income and spending growth than those at the top. This pushed up substantially the relative poverty rates for both expenditure and income. But the rise in relative poverty measured by spending was less rapid than the rise in income poverty (see Figure 1).
Since Labour has been in power, the story has been very different: low-income households have experienced more rapid income growth than middle-income households, which has reduced income poverty. However, low-spending households have continued to see their expenditure rise less quickly than that of middle-spending households. This means that spending poverty has continued to rise. Since 1996/97, income poverty rates have fallen from 25% to 22% (in 2002/03), a fall of 12%. However, spending poverty rates have risen from 20% to 22%, a rise of 12%.
A similar story emerges for child poverty, one of the Government’s key targets. Between 1996/97 and 2002/03, the child poverty rate measured using income fell from 34% to 29%, a drop of 15%. Based on expenditure, though, it increased from 24% to 27%, a rise of 11%. For another target group, pensioners, poverty rates are much higher and more stable when measured using spending rather than income. Using spending to measure living standards, pensioner poverty has not fallen since 1996/97.
Figure 1: Income and expenditure poverty rates, 1961 to 2002/03
Who is poor?
Using spending as a measure also alters the perception of who is poor. In 2002/03, the income poverty rate for self-employed people (23%) was much higher than the spending poverty rate (13%). The picture was similar for those seeking work. This suggests that low incomes were a transitory state for some of these people, and that they were using savings or borrowing to maintain their standard of living. For retired and pensioner households, spending poverty rates were much higher than income poverty rates, highlighting the fact that many older people spend well below their income level.
Income poverty and spending poverty are quite distinct states: only around half of people classified as poor on one measure would simultaneously be classified as poor on the other. Very low-income households (those in around the bottom 2% of the income distribution) on average have levels of spending much higher than their income alone would suggest (see Figure 2). This phenomenon could be due to two factors: either very low-income households have only temporarily low incomes, and are running down savings or accumulating debt to fund their expenditure; or very low incomes are measured with error. In contrast, very low-spending households tend to have very low incomes, commensurate with their spending levels. This suggests that spending might be a preferable measure of well-being for the very poorest in society. A more reliable impression of those with the lowest standards of living might be obtained by examining those recorded at the bottom of the spending distribution rather than the bottom of the income distribution.
Increased benefit entitlements for pensioners – the effect on expenditure
Since 1999, a series of reforms to the means-tested benefits system have benefited poor pensioner households: the introduction of the Minimum Income Guarantee and Pension Credit; the equalisation of pensioner premiums in Income Support at the highest rate for all pensioners; real increases in the state pension; and winter fuel payments. Between April 1996 and April 2002, the real entitlements to state benefits for non-disabled pensioners aged 60-74 with no private income rose by around 25%. For those aged 80 or more, the rise was around 15%.
Using statistical techniques to isolate the effects of these policy changes on pensioner spending, the study estimated that the introduction of the Minimum Income Guarantee in 1999 raised the spending of eligible pensioners by around 10% relative to non-eligible pensioners. This additional spending went almost entirely on ‘non-basic’ items, i.e. goods and services other than food, housing and fuel.
The equalisation of the age premiums in Income Support at the highest rate in 2001 also led to an increase in spending among the groups who benefited. The spending of those aged under 80 rose by around 8 to 10% more than the spending of those aged 80 or more. Again, almost all of this extra spending was on non-essentials.
Figure 2: Expenditure among the lowest-income half of households
The study assessed trends in poverty when using expenditure as a measure of well-being, and found some substantial differences to the familiar trends in income poverty, particularly since 1996/97. The results of the study highlight the need to place the Government’s reduction in income poverty, particularly for children and pensioners, in a wider context. The findings suggest that while government policy has resulted in some equalisation of the distribution of income, it is yet to reduce spending inequalities, which better reflect longer-term differences in society. The authors conclude that there is a strong case for the Government or other bodies to monitor trends in spending poverty carefully and regularly, as a complement to other measures of well-being.
About the project
The study was carried out by Mike Brewer, Alissa Goodman and Andrew Leicester at the Institute for Fiscal Studies. It used data from the Family Expenditure Survey and Expenditure and Food Survey (1961 to 2002/03) and the Family Resources Survey (1993/94 to 2002/03) to examine trends in relative expenditure and income poverty. (A household is defined as being ‘in poverty’ if its total income or spending of all household members is less than 60% of the median value in a given year.) The study focused on incomes after housing costs, and expenditure excluding housing. The analysis of pensioner spending used a difference-in-difference approach to assess whether the spending of a ‘treatment’ group (those pensioners most affected by benefit reforms) rose faster than that of a ‘control’ group (those pensioners relatively unaffected). This made use of straightforward multivariate regression techniques.