Analysis of the experiences of over 7,900 people interviewed annually between 1991 and 1994 reveals that there is a lot of fluctuation in income from one year to the next, but little long-range movement from being poor to becoming rich or vice versa. The study, by Sarah Jarvis and Stephen P Jenkins of the ESRC Research Centre on Micro-Social Change at the University of Essex, found that:
- Only about forty per cent of the population are in the same tenth of income distribution from one year to the next, but over seventy per cent remain in the same income band, or in the one higher or one lower.
- Fluctuations in income are experienced by people at all income levels. There is some evidence that mobility is greater in the very poorest and the very richest income groups and, perhaps surprisingly, for elderly people compared to non-elderly people.
- Variations in income are such that as many as one-third of the population experience a low income (below half average income) at least once over a four-year period. About one-tenth have a low income in at least three of the four years surveyed. About one-twentieth have low incomes at all four periods. This group mainly comprises single pensioners and families with children headed by either a couple or a lone parent not in work.
- Almost a third of those who leave the low income bracket have low income again within a year; ninety per cent of those in the poorest tenth of the population remain in the bottom three-tenths a year later.
- It is not just losing or getting a job which are important for income changes but also demographic events such as marriage, divorce, death of a partner and birth of a child.
- When married or cohabiting partners separate, the incomes of separating wives fall by almost a fifth on average over the following year, whereas separating husbands' incomes remain about the same on average. The children of these partnerships experience income decreases of about one sixth.
- After a marital split, separating wives are more likely to be looking after children, not doing paid work and receiving welfare benefits than are separating husbands. Only a quarter of separating wives with dependent children receive maintenance income from their former partner.
The Department of Social Security's Households Below Average Income statistics tell us each year how many people are at a given income level but do not tell us about people's movements between income levels from one year to the next. Using a new data source, four annual 'waves' of interviews for the British Household Panel Survey, this project provides new evidence about the dynamics of personal income.
How much income mobility is there in Britain?
Just under 40 per cent of the population were in the same tenth of income distribution when interviewed in 1991 and 1992 but over 70 per cent remained in the same income band, one higher or one lower. Very similar patterns are revealed when one looks at the income changes between 1992 and 1993 and between 1993 and 1994.
One-half of those in the poorest tenth of the income distribution at one point in time are no longer in the poorest tenth in the next period. However, about one half of those moving out of the poorest tenth move only to the second poorest tenth and almost 90 per cent remain in the poorest three-tenths. By contrast about two-thirds of those in the richest tenth remain in the richest tenth.
The longer the period over which each person's income is averaged, the greater the possibility that temporary income changes or measurement errors are smoothed out and inequality in averaged income reflects 'permanent' differences in income between people. Measuring income inequality using the 'Gini coefficient' (a commonly used inequality index), shows inequality for wave 1 (1991) incomes as 0.31. However, using the four-wave average of each person's income, the Gini coefficient falls by almost ten per cent to 0.28. This reduction in inequality, reflecting income mobility, is of about the same magnitude as the equalising impact of direct taxation on household's taxable income, but a lot smaller than the increase in inequality in Britain between 1979 and 1990/91 (about a third according to the Gini coefficient).
The study looked at a number of ways of measuring instability in income. The amount of movement between different levels of income differed according to which measure was used, but was greatest for elderly people relative to other people.
Movements in and out of low income
Although only a minority of the population have a low income in any given year, many more people experience low income at least once over a four-year period. If the low income cut-off is defined as £127 per week (equal to half the average income in wave 1 (1991) ) the proportion of people in a low income in wave 1 was 18 per cent and the proportion of people with a low income in wave 4 was 14 per cent. However, only 4.3 per cent of the sample had low income on this basis at all four interviews (and 68.7 per cent did not have low income at any interview over the four-year period).
Almost one-tenth of the sample (9.8 per cent) had a low income at three or more interviews, almost one-fifth (17.9 per cent) had a low income at two or more interviews, and almost one-third (31.3 per cent) had low income at one or more interviews. If, alternatively, the low income threshold is defined to be the income level which separates the poorest fifth from the richest four-fifths of the population (the poorest quintile), then the corresponding estimates are 14 per cent, 23 per cent and 36 per cent respectively. For both low income threshold definitions, the proportion of the population which is touched by low income over a four-year period is much larger than the proportion with a low income at any one point in time.
The exit rate from low income after one year with low income is 54 per cent if the low income threshold is defined as half the average 1991 income. The exit rate after two interviews reporting a low income is 51 per cent. These rates imply that for a group starting a low income spell, just under one-half (46 per cent) still have low income after one year, and about one-fifth (22 per cent) still have low income after two years (i.e. after the third interview reporting low income). That is, almost four-fifths of the low income group no longer have low income after two years.
The low income re-entry rate one year after leaving low income is 29 per cent but more than halves, to 11 per cent, two years after leaving. The rates imply that, for a group of people starting a spell of low income, 29 per cent will have another low income spell after one year, and more than one-third (36 per cent) will have fallen below the threshold again after two years. Thus just over one half of the group will have incomes above the cut-off for at least two years (three interviews). When the low income cut-off is the poorest quintile, magnitudes are different but patterns are similar.
The exit rates, if looked at on their own, might suggest that the majority of people falling into low income will spend only a couple of years in this situation. However, the path out of low income is not a one-way up-escalator: the re-entry estimates remind us that there is a not insignificant chance of finding oneself on the down escalator to low income again. The study derived better predictions of the number of times people have a low income over a given time period when it incorporated the possibility of repeated low income spells.
Characteristics of those on low income
The small group of people who have low income at all four interviews between 1991 and 1995 mostly comprises single pensioners and unemployed couple families with children and lone-parent families. The characteristics of those entering low income and those escaping low income are broadly similar to those of the group staying on low income. (The income ranges round about the low income threshold are disproportionately occupied by elderly people and non-working families and, since most income mobility is relatively short-range, these groups are most likely to make a transition.) However, some differences stand out: for example, amongst the low income 'escapers' there are higher proportions of childless couples and single adults than there are amongst the population as a whole.
The number of earners increased for 18 per cent of low-income 'escapers' compared with about 12 per cent of the sample as a whole. For low-income entrants, the proportion with a decrease in the number of earners is more than twice the average sample incidence, 30 per cent compared to 12 to 13 per cent.
Increases in the numbers of earners may arise through either an existing household member getting a job, or the arrival of a new partner who also works, or both. Since the incidence of household composition change is about average for low income escapers, this suggests that getting a job plays a particularly important role in taking people out of low income. This story fits best for the escapers who are in non-working families with children.
Marital splits and income changes
In the year after separating from their husband or cohabiting partner, women suffer substantial falls in real income, whereas separating men experience much smaller income changes. Comparing incomes at the pre-separation interview with incomes at the post-separation interview, incomes fell by 18 per cent on average for separating women, whereas incomes increased by 2 per cent on average for separating men. There were dependent children in about one-half of the former partnerships, and more than 90 per cent of these children lived with their mother after the separation. Their income decrease was about 14 per cent on average.
|Characteristics at interview prior to split||median percentage change in net income between pre and post-split interview|
|Separating husbands||Separating wives|
|Age (2 years):|
|35 or older||10||-21|
Table 1 shows that amongst separating women, the average income loss in the year after the marital split was much the same regardless of age, whether they had dependent children or whether they were married or cohabiting before the marital split. Amongst separating men, those with income increases after the marital split were more likely to have legally married, be at least 35 years old, and to have had dependent children.
Underlying the contrasting income changes between separating men and women are differential changes in the composition of household income packages. After the separation two-thirds of men were in paid work but only half of women (and the proportion of separating women who stop working after the marital split, 15 per cent, is 50 per cent higher than the proportion of husbands who stop). Reflecting this, almost one-half of separated women received income support or housing benefit in the year following the separation, compared to less than one-quarter of men. Moreover, only one-quarter of women with dependent children received any maintenance income from their former partner after the separation.
Another reason why men are better off than women after a marital split is that the same income has to support more people in women's households than men's households, since the children of separating couples typically stay with their mothers. Although the estimated impact on real income of these demographic changes depends on the specific formula used to adjust incomes for differences in household composition, the study found that separating women always did worse than did separating men.
About the study
The research is based on the first four interview waves of the British Household Panel Survey (BHPS). A sample of some 5,500 households was interviewed in the autumn of 1991, and the people in those households have been followed and they and their adult co-residents interviewed every year subsequently and information derived about the incomes and circumstances of all household members. The subsample used in this analysis includes the 7,910 people from complete respondent households who were present in all four waves. Income is defined as equivalised household net income on a per person basis, and expressed in constant January 1995 prices. The variable is constructed in the same way as the 'income before housing costs' measure in the Department of Social Security's Households Below Average Income statistics.