What pushes unpaid carers into poverty?

No one deserves to live in poverty. Yet millions of people who provide care for their loved ones find themselves without sufficient income to participate in the very society that they sustain. As a society, we are biting the hand that feeds us.

Using data from the Understanding Society survey,¹ this report investigates the dynamics of poverty among unpaid carers between 2011 and 2012, and 2019 and 2020. We find that:

• Unpaid carers transition into poverty at a higher rate than adults overall, and exit poverty at a lower rate once there. For both unpaid child-carers and unpaid social-carers, poverty entry rates are particularly high among new carers, and higher still among those who became a carer unexpectedly, for these carers, the risk of entering poverty is twice as high as it is for the adult population as a whole, at 12% compared to 6%.
• Unpaid carers transition out of poverty at a lower rate than adults overall, and exit rates once there are slightly lower still for new carers.
• Unpaid carers are more likely to re-enter poverty after they exit, and more likely to remain in poverty after they enter, than adults overall. New unpaid child-carers, particularly those who became carers unexpectedly, have even higher persistence rates, while new unpaid social-carers have higher recurrence rates.
• Examining the factors that push unpaid carers into poverty, labour-market events like moving out of paid work or experiencing a reduction in household earnings account for the majority of transitions into poverty among unpaid child-carers, coinciding with around two-thirds of poverty entries. Falls in benefit income are also significant, although less so for new child-carers, for whom (unsurprisingly) an increase in the number of children, and so a higher number of mouths to feed, is the single most significant event.
• Meanwhile, among unpaid social-carers, falls in income from benefits like Universal Credit account for the majority of poverty entries. However, labour-market events are also significant (particularly among unpaid social-carers who were previously in paid work, for whom the carer exiting paid work is the single most significant event), as are reductions in other household income.
• Labour-market events are less common triggers for pulling unpaid social carers out of poverty.

To address the link between poverty and caring, we need to better understand what’s driving the hardship that carers face. Given events in the labour market and falls in benefit income are key drivers, we need a new deal for carers. This should help carers juggle work and care, through flexible working support and paid care leave.

Reforms to the benefits system must provide a solid safety net and recognition of the difficult, skilled and life-saving work that carers contribute to society day in, day out, that prevents them doing paid work.

Unpaid carers face an elevated risk of being in poverty: in 2019 and 2020, around one-quarter of unpaid child-carers (24%) and unpaid social-carers (28%) were living in relative poverty after housing costs, compared to 17% of the overall adult population.² As a result, 1 in in 4 adults living in poverty were unpaid carers, 1.9 million child-carers and 0.9 million social-carers. To this we can add those living in the same household as a carer, including millions of disabled and elderly people, and virtually all children. Unpaid carers have been particularly struggling during the cost-of-living crisis (Carers UK, 2022).

We know that much of the hardship faced by unpaid carers reflects a tension between paid work and unpaid care, along with inadequacies in the benefits system and statutory leave policies (Thompson et al., 2023; Casey, 2023). In particular, many unpaid carers are forced to exit paid work or reduce their working hours, often with little in the way of income replacement.

This is only part of the picture, however. Some unpaid carers may fall into poverty for other reasons even if they stay in paid work, and those who exit paid work may find that multiple life events coincide to push them into poverty. Others may not be in paid work to begin with.

This report uses data from the Understanding Society survey to quantify the risk of falling into poverty and the prospect of escaping it, and to identify the trigger events that are associated with these transitions. Our analysis builds on the Department for Work and Pensions’ methodology for analysing income dynamics (DWP, 2023; Jenkins and Rigg, 2001; Taylor and Schmuecker, 2023).

Following our previous work on the caring penalty (Thompson et al., 2023), we distinguish between unpaid social-carers (those providing at least 20 hours of unpaid care per week to someone who is sick, disabled or elderly) and unpaid child-carers (the adult in each household with children who is primarily responsible for the children). Poverty is defined as relative poverty after housing costs, with disability benefit income excluded for analysis on unpaid social-carers. You can read more details in the 'Methodology' section.

One difficulty in studying the relationship between poverty and unpaid care is establishing causation. While becoming an unpaid carer will push many people into poverty by reducing their income, an effect that we quantified in Thompson et al. (2023) for those previously in paid work, income and factors contributing to income levels may themselves influence whether or not somebody becomes an unpaid carer.³ Indeed, if individuals face a degree of choice, they are likely to take into account the potential effect of caring on household income.

One way to help disentangle the relationship between poverty and unpaid care is to focus on new unpaid carers, defined as those who have been providing unpaid care for up to 3 years. Most of the life events that we would expect to affect the impact of unpaid care on poverty status, such as exiting paid work and having children, may be more likely to occur in the first few years of being an unpaid carer, even if scarring effects persist for longer.

So, while differences in outcomes between unpaid carers and the overall population can help uncover relationships between unpaid care and poverty, differences in outcomes between new unpaid carers and unpaid carers as a whole can help reveal whether it is caring itself that is driving the relationship rather than other factors.

Still, new unpaid carers are not a random group, meaning that comparisons will not show us the pure impact of unpaid care on poverty. To get closer to this analytical ideal, we can attempt to identify individuals who became unpaid carers unexpectedly. In the case of unpaid child care, these are individuals caring for a child who is not their natural or adoptive child, for example due to the death or sickness of a parent, by the intervention of social services and other arrangements leading to kinship care, or through family formation and breakdown.

Unexpected social-carers are those for whom the person receiving care experienced a sudden deterioration in health. To the extent that these carers had less choice and less time to make alternative arrangements, caring can be considered an external ‘shock’ that should be less affected by other factors. While the selection of these carers will still not be fully random, comparing their outcomes to those of all unpaid carers can at least help us understand the direction in which any selection bias is operating.

For each of unpaid child-carers and unpaid social-carers, we therefore have 3 groups: all carers, new carers and unexpected carers, each a subgroup of the group preceding it. For some analysis, we also present results for new unpaid carers who were in paid work in the previous year (some of whom may be unexpected carers). The methodology explains how we define these groups in more detail, and Appendix A contains a selection of demographic breakdowns of each group, focusing on age, sex and work status, as well as showing the size of each group.

Typical unpaid child-carers tend to be women, 35–44 years old, in couple households, and in paid work. The predominance of women, who comprise 97% of unpaid child-carers, is particularly striking, though this partly reflects the way that these carers are identified in the data with only 1 adult in each household counted as child-carer. Unpaid social-carers are also predominantly women, but less so than unpaid child-carers. They also tend to be older than average, with 65+ being the most common age bracket, and the majority are not in paid work.

Unexpected carers in both groups more closely resemble the overall adult population in a number of dimensions, validating our treatment of this group as a more random distribution of individuals that is less subject to selection bias. In particular, unexpected carers contain a higher proportion of men, though women still make up the majority.

One exception is that unexpected social-carers tend to be older than unpaid social-carers overall, though due to the way this group is defined, based on changes in the health status of the person receiving care, this likely reflects a correlation between age and health rather than to selection mechanisms per se.

Poverty entries

Figure 1 shows the number of unpaid carers who transitioned into poverty each year from 2011 and 2012 to 2019 and 2020, as a proportion of those who were not in poverty in the previous year. This is known as the poverty entry rate. The entry rate for all adults stayed relatively stable at 6–7% over the period. The rates for unpaid carers were higher, between 8% and 10% for unpaid child-carers, and between 9% and 12% for unpaid social-carers.

To understand what is driving these patterns, we can examine the events that most often coincide with and can affect transitions in and out of poverty. These ‘trigger events’ only represent proximate causes, they help us understand which components of household income change, but not the underlying reasons for these changes. Trigger events are also proximate in a temporal sense: they only include events that coincide with a poverty transition, when in reality important events may have occurred leading up to the transition, spanning from the years immediately prior all the way back to before the survey respondent was born.

So while trigger events are informative, particularly for comparing poverty dynamics between groups, they need to be complemented by a deeper understanding of the causes of poverty and the context in which transitions occur. This includes an appreciation for how different components of income interact.

For example, falls in household income due to demographic and earnings-related changes should, in theory, elicit offsetting changes in entitlements to means-tested benefits for low-income households. However, we know that the benefits system often fails to provide an adequate safety net: around half of people in households receiving Universal Credit or legacy benefits are in poverty, due in part to the series of welfare reforms introduced in the 2010s (JRF, 2024).

In addition, some households may be ineligible for these benefits despite being in poverty, for example, if their savings exceed the capital limit of Universal Credit, while others may not claim their entitlement, for example because the system is confusing or laced with stigma (Schmuecker and Elliot, 2023). Similarly, leave policies may not sufficiently replace earnings, particularly in the case of unpaid social care, for which no statutory paid leave is available at all (Thompson et al., 2023).

For households already receiving benefits, there are multiple reasons why a household may fall into poverty as a result of benefit income falling.⁵ Most notably, the welfare reforms mentioned above, including the benefit cap (introduced in 2013, with a further reduction in 2016), the social sector size criteria (also known as the ‘Bedroom Tax’, 2013), the two-child limit (2017), the removal of the family element (2017) and various changes to disability assessments, led to significant reductions in benefit income for millions of households.⁶

A means-tested benefit can also fall when household earnings increase, and although this in itself should not leave the household worse off, it can trigger knock-on effects on other benefits. Other aspects of a means test, such as savings and income from other sources, can also have a bearing. In addition, working-age benefit recipients may be subject to sanctions if they are deemed to have violated their claimant commitment.

And when it comes to Universal Credit, the lag in the monthly assessment can mean that recipients with volatile earnings face a simultaneous drop in earnings and benefits.

Trigger events and poverty entries

For each type of unpaid carer, Figure 8 shows a range of trigger events and the proportion of transitions into poverty that coincided with each event from 2011 and 2012 to 2019 and 2020. All events reduce the measure of household income used to determine a household’s poverty status, labour-market and benefit events through income, demographic events through the needs adjustment (equivalisation) and other events through housing costs as well as income.

In addition to showing results for new carers and those who became carers unexpectedly, this analysis includes a breakdown for new carers who were in paid work in the previous year.

For adults as a whole, labour-market events, someone in the household moving out of paid work, or staying in work but experiencing a reduction in earnings, account for the majority of transitions into poverty, with falls in another household member’s earnings representing the most common event. Labour-market events coincide with an even larger proportion of poverty entries for unpaid child-carers.

This difference is mostly due to household members other than the carer experiencing a fall in earnings, but if we focus on new unpaid child-carers who were previously in paid work, the most significant event is carers themselves exiting work, coinciding with around one-third (35%) of poverty entries.

Falls in benefit income are also associated with a slightly higher proportion of poverty entries for unpaid child-carers than for adults overall, 41% compared to 38%, with means-tested benefits the most common source. However, the opposite is true if we focus on new child-carers, for whom benefit-related events account for only one-quarter (25%) of poverty entries.

For these carers, demographic events are more significant, particularly an increase in the number of children, which is unsurprisingly the single most significant event for new child-carers, though increases in the number of adults are also significant for unexpected child-carers.

Falls in benefit income account for a larger proportion of entries into poverty for unpaid social-carers than adults in general. This difference is spread across all categories of benefits, including means-tested and disability benefits,⁷ which include Carer’s Allowance.⁸

The flipside is that labour-market events coincide with a smaller proportion of poverty entries for unpaid social-carers, reflecting the fact that unpaid social-carers are less likely to be in paid work even before entering poverty.⁹

Importantly, then, the relative frequency of benefit-related triggers as compared to labour-market triggers does not mean that the labour market is functioning well for this group; on the contrary, the high levels of poverty among unpaid social-carers largely reflects their exclusion from the labour market.

In any case, if we focus on unpaid social-carers who were previously in paid work, labour-market events become more significant, particularly the carer’s own work status, which comprises the single most significant event.

Interestingly, exiting paid work is a less common trigger for new child-carers who were previously in work than the previously working informal social-carers, even though going on maternity leave would be classified as a work exit in this analysis. This will in part reflect the availability of Statutory Maternity Leave, for which there is no equivalent in the case of unpaid social care.

Falls in other household income, which includes income from private pensions and investments, are more significant for unpaid social-carers than they are for adults overall, coinciding with a quarter (25%) of poverty entries. This event is less significant for unpaid child-carers than for adults overall, likely reflecting differences in age between the two groups. Increases in housing costs account for between 10% and 20% of poverty entries for all groups.

The analysis in this section focuses on the proportion of poverty entries and exits that are associated with particular events. For example, we found that 60% of poverty entries among unpaid social-carers involve a fall in benefit income. However, this does not tell us whether benefit-related events represent such a high share of poverty entries among this group because they happen particularly often, or because they are particularly likely to move people into poverty when they do happen.

To understand which of these effects is at play, we can examine the incidence of each event among those at risk of a transition into poverty (event frequency) and the probability of transitioning into poverty when the event occurs (transition risk), alongside the share of transitions associated with the event.¹⁰ This is shown in full in Appendix B, with summary figures presented here (Figure 10). Note that like transition rates, transition risks will tend to be higher for poverty exits than poverty entries even if a similar number of people are affected, simply because a minority of people are in poverty at any given time.

We have seen that labour-market events make up a larger proportion of poverty entries than any other event category (68%) for unpaid child-carers. Appendix B shows that this is the result both of adverse labour-market events being more common than other events, affecting 33% of those not in poverty in the previous year, and also presenting the highest risk of pushing child-carers into poverty when they do happen, with 19% of those experiencing an adverse labour-market event moving into poverty.

Meanwhile, among unpaid social-carers, the high proportion of poverty entries associated with benefit-related events (60%) mainly reflects their high frequency, affecting 40% of those not in poverty in the previous year. Their transition risk is actually lower than that of labour-market events, at 16% compared to 20%.

Thus, for both groups of carers, falls in income resulting from labour-market events are particularly detrimental when they occur. The reason that they do not represent a higher proportion of poverty entries among unpaid social-carers is because benefit-related events are more likely to occur in the first place, reflecting the lower proportion of this group who are in paid work.

The analysis in this report has shown that unpaid carers face elevated risks of falling into poverty, and weakened prospects of escaping it once there. Carers who do exit poverty are disproportionately likely to re-enter poverty, and those who enter poverty are disproportionately likely to remain in poverty. For unpaid child-carers, labour-market events are the most common triggers for moving in and out of poverty, whereas unpaid social-carers are more reliant on the benefit system.

All of this paints a bleak picture, but we should not accept it as a fact of life. It reflects a failure of the labour market to provide flexible, stable, well-paid jobs; a failure of leave policies to support people while they take time off and to give them a route back into paid work; and a failure of the benefit system to ensure that nobody falls below a decent standard of living.

These failures impact people across the population, but unpaid carers are among those who suffer the most, along with the people they care for, including millions of children, disabled people, and others who are unable to care for themselves.

Data

We use Waves 2–12 of the UK Household Longitudinal Study, also known as Understanding Society. Since a poverty transition can only be identified if we know an individual’s poverty status in the prior year, the first transition is between Waves 2 and 3. The final transition is between Waves 10 and 11, to ensure that we can examine poverty persistence and recurrence in the following year while using a consistent time period throughout the analysis. A transition is recorded as occurring in the second wave (year) of the transition.

Definitions

In each household with children, an unpaid child-carer is defined as the adult who is primarily responsible for looking after the children. An unpaid social-carer is defined as an adult who provides at least 20 hour per week of care to someone who is sick, disabled or elderly, whether in the same household or another household. In a transition between 2 years, an individual need only meet 1 of these definitions in the second year to be counted as an unpaid carer.¹¹

Unexpected social-carers are identified as those who are within 3 years of becoming a carer according to the definition above, were previously providing no unpaid care, and who care for someone that experienced a deterioration in health over the previous 2 years such that they became limited in performing moderate activities or more limited than before. We designate an individual as an unexpected child-carer if they are within 3 years of becoming a carer, and are not a biological or adoptive parent of the children for whom they care. These individuals could have become carers following the death or sickness of a parent, by the intervention of social services and other arrangements leading to kinship care, or through family formation and breakdown.

Poverty is defined as having a net equivalised household income after housing costs that is less than 60% of the national median in a given year.¹² Household income and its components are uprated by an adjusted version of the Consumer Price Index (CPI), which is also used in DWP’s Households Below Average Income (HBAI) statistics. Analysis on unpaid social-carers removes particular disability benefits from household income, namely Disability Living Allowance, Personal Independence Payment and Attendance Allowance, since these are intended to cover the additional costs associated with disability. A poverty transition is only counted if household income clears the poverty line by a margin of at least 10%.

Trigger events

* Changes to or from 0 and changes that involve moving from a positive to a negative number or vice versa are not subject to the 20% threshold, but are still subject to the £10 threshold.

^ Means-tested benefits include Universal Credit and legacy benefits, plus Council Tax Reduction. Old-age benefits include State Pension, Pension Credit, and widows pensions and allowances. Disability benefits include Disability Living Allowance, Personal Independence Payment, Attendance Allowance, Carer’s Allowance, War Disablement Pension, Incapacity Benefit, Severe Disablement Allowance, Industrial Injury Disablement Allowance, and 'other disability related benefit[s] or payment[s]' as recorded by the relevant Understanding Society variable. Child-related benefit income includes Child Benefit, Maternity Allowance and Foster/Guardian Allowance. Other benefits include any reported benefit not included in these categories, as well any remaining difference between total benefit income and all reported benefits.

Source: Understanding Society, Waves 3–11

Notes: Pooled longitudinal analysis. ‘Unpaid sandwich carers’ are those performing both unpaid child care and unpaid social care, and are included in the proportions for both groups.

Source: Understanding Society, Waves 3–11

Notes: Pooled longitudinal analysis. Denominator of all figures is all unpaid carers.

1. The UK Household Longitudinal Study
2. Based on Understanding Society. Poverty rates derived from DWP’s Households Below Average Income (HBAI) dataset will differ slightly. See Fisher et al. (2019).
3. See Thompson et al. (2023) for a discussion. This is partially addressed in our analytic approach, since we are examining individuals who move into poverty rather than those who were in poverty to begin with. However, there are a number of reasons why these determinants may still play a role. First, there is likely to be a strong correlation between poverty entry and previous experience of poverty. Second, because the poverty line is binary, poverty entries will include households that already had low incomes but were not technically in poverty, so correlations between unpaid care and household income can have a bearing on the results. Third, the discrete nature of the data, consisting of annual snapshots rather than a continuous record, means that we cannot be sure that a poverty entry did not precede an individual becoming an unpaid carer. This is particularly relevant since we count an unpaid carer as moving into poverty regardless of whether they were an unpaid carer in the previous year, in order to capture the effects of unpaid care in the first year of caring. Lastly, whether or not somebody becomes an unpaid carer may be influenced by income potential (and its determinants), not just income (and its determinants) at the time.
4. These differences are small, though this is due to the exit rate having a larger denominator than the entry rate, as explained in the next paragraph.
5. These changes are not due to demographic changes, which are classified as separate trigger events. See Appendix B for more details.
6. In isolation, benefit freezes would most likely not meet the criteria for a trigger event, particularly the condition that income falls by at least 20% year-on-year. However, they could combine with other events, and likely increase the poverty entry rates associated with labour-market and demographic events by weakening the safety net.
7. This is despite the fact that a change in particular disability benefits, namely Disability Living Allowance, Personal Independence Payment and Attendance Allowance, would not in itself move the household of an unpaid social-carer into or out of poverty, since these benefits are subtracted from the household income of these carers when it comes to determining poverty status.
8. Carer’s Allowance makes up a relatively small proportion of entry trigger events. However, this largely reflects the low rate of the benefit, at just £76.65 per week in 2023/24, which means that few households are likely to be kept above the poverty line by Carer’s Allowance alone, and therefore to enter poverty when the benefit falls. This is particularly true in our analysis because a household needs to fall below 10% of the poverty line to count as entering poverty.
9. See Section 2 of the House of Commons Library report on informal carers for an analysis of informal care and employment.
10. Being at risk of entering poverty simply means that an individual was not in poverty in the previous year, and conversely for exiting poverty. Transition risk refers to the conditional probability of poverty exit as well as poverty entry, even though the term ‘risk’ refers to undesirable outcomes in ordinary usage.
11. This approach varies from DWP’s methodology, in which characteristics are defined based on the first year of a transition. The characteristics analysed by DWP are essentially demographic, whereas caring is an activity that can begin and end. Our approach ensures that the life events that occur when somebody first becomes a carer are taken into account.
12. The measure of after housing costs income in Understanding Society differs in some respects from that in DWP’s Households Below Average Income (HBAI) dataset, which forms the basis of the official poverty statistics. See Fisher et al. (2019).

The author would like to thank Isabel Taylor and Peter Matejic for quality assurance and analytical guidance, and Patrick Calver and Abby Jitendra for helpful feedback.

Carers UK (2022) Under Pressure: Caring and the cost of living crisis

Casey (2023) Unpaid care and poverty: unpaid carers' priorities for change through participatory co-design

Department for Work and Pensions (2023) Income Dynamics: income movements and the persistence of low income, 2010 to 2021

Fisher, P. Fumagalli, L. Buck, N. and Avram, S. (2019) Understanding Society and its income data.

Jenkins, S.P. and Rigg, J.A. (2001) The Dynamics of Poverty in Britain

Joseph Rowntree Foundation (2024) UK Poverty 2024: The essential guide to understanding poverty in the UK

Schmuecker, K. and Elliot, J. (2023) On a low income, but not claiming means-tested benefits

Taylor, I. and Schmuecker, K. (2023) What protects people from very deep poverty, and what makes it more likely?

Thompson, S. Jitendra, A. and Woodruff, L. (2023) The caring penalty

If you are using this document in your own writing, our preferred citation is:

Thompson, S. (2024) What pushes unpaid carers into poverty? York: Joseph Rowntree Foundation.