Care, Kinship and the Life Course

The last lesson established that fewer births and longer lives reorganised the shape of an adult life.

This lesson is about the part of that reorganisation that no institution was designed for: a period in midlife, lasting years, in which a person is responsible for a parent — and frequently for a child at the same time.

It is done overwhelmingly by families, overwhelmingly by women, and it is counted by nobody (see 8.5.1).

One week, and the sentence at the end of it.

Rekha is fifty-two. She works four days a week as a practice manager. Her son is nineteen and living at home while he studies locally. Her mother is eighty-four and lives forty minutes away.

Monday. Two calls to her mother, one to establish that the medication was taken and one because the first was not answered. Thursday is her non-working day, which is not a day off: shopping, the pharmacy, sorting the week's tablets into a box, changing the bed, a conversation with a neighbour who has noticed something.

Wednesday at 2 a.m. her mother falls. Not badly. Rekha drives over, stays until five, and is at work by nine.

Friday she takes annual leave to attend a needs assessment. She has used eleven days of annual leave this year on appointments.

Saturday she has the conversation with her brother, who lives two hundred miles away, phones on Sundays, and believes the situation is under control.

Sunday her son mentions that he needs to talk about something and she says she will come back to it.

Add it up. Between the calls, the driving, the appointments, the administration, the sorting and the nights, she is providing something in the region of twenty hours of care a week, on top of four days of paid work.

Now the accounting.

None of those hours appear in any measure of economic output (see 8.5.1's production boundary).

She reduced her hours three years ago, which reduced her pension contributions permanently, and her earnings by roughly a fifth for what will be the highest-earning decade of her life.

She is not recorded anywhere as a worker in the care sector, although the hours she supplies exceed those of a part-time care assistant.

And here is the sentence at the end of the assessment.

The assessment recorded that family support was available, and the care package was set accordingly — fewer funded hours than would have been provided had she lived two hundred miles away, like her brother.

Read that carefully, because it is the structural feature of the whole system. In a needs-assessed care system, the formal provision is calibrated to what the family is already doing. The more care a daughter supplies, the less the state supplies. The system does not merely rely on unpaid care; it is designed around it, and it responds to increases in it by withdrawing.

Why is there no institution for this?

Because the phase did not exist when the institutions were built.

The pension systems, the employment protections, the leave entitlements and the welfare states of the mid-twentieth century were designed around a life course with three phases: education, then continuous full-time employment, then a short retirement ending in death.

That model assumed something that has ceased to be true. It assumed a household with an adult not in paid employment, available to absorb whatever contingencies arose — and it assumed that most people would not have surviving parents in their eighties, because most people did not (see 9.2.1).

Both assumptions failed at once. Women entered paid employment, removing the contingency reserve. Life expectancy rose, creating a long period of frailty at the end of life. And the two changes were not coordinated, because nobody was coordinating them.

Which is 8.5.3's crisis of care, located precisely: the hours were being supplied by someone, they are no longer available in the same quantity, and no institution took over the difference.

The life course as a sociological object

Four principles, from the study that established the field.

The founding work followed children who had been young during the Great Depression into old age, and asked how a historical event experienced at different ages produced different lives. The framework it generated has four parts, and they are all useful.

Historical time and place. A life is shaped by the specific period and society it is lived in. Two people with identical characteristics born ten years apart can have very different lives, because the labour market they entered differed (see 9.1.4 on scarring).

Timing. The consequence of an event depends on when in a life it happens. Unemployment at twenty, at forty and at fifty-eight are three different events. A child's experience of the Depression depended on how old they were when it hit — old enough to work, or young enough to be dependent through the worst years.

Linked lives. Lives are lived interdependently, so a transition in one person's life produces transitions in others'. A grandmother's retirement enables a mother's employment; a father's job loss changes a child's schooling; a daughter's promotion is declined because of a parent's diagnosis. Rekha's week is entirely made of linked lives.

And agency within constraint. People construct their lives through choices, within the opportunities and limits their time and place supply — which is 1.3.1's structure and agency, made into a research programme.

And the sociological point underneath all four: the life course is institutionally organised, not biologically given. Education, employment and retirement are separated by legal and administrative thresholds — school leaving ages, pension ages, eligibility rules — and the tripartite pattern that seems natural is a construction of the twentieth century.

That construction is now loosening. Education recurs, employment is interrupted, retirement is partial and phased, and the sequence is less standard than it was — which produces both freedom and the loss of the defaults that made planning possible.

And a methodological warning that this field cannot escape.

Three things vary together in any observation of people over time: their age, the period in which they are observed, and the cohort they were born into.

And any one of the three is exactly determined by the other two. Age equals period minus birth year. Which means the three effects cannot be statistically separated without imposing an assumption — the identification is not merely difficult, it is impossible from the data alone.

Every claim distinguishing "young people today are different" (cohort) from "people are like this when young" (age) from "everyone changed recently" (period) therefore rests on an assumption, and the assumption does the work.

This matters far beyond the family, and it is the standing reason to be sceptical of generational claims (see 7.6.4). When someone tells you a generation is distinctive, ask how they separated the cohort effect from the age effect — and note that the honest answer involves an assumption they should state.

Who provides care

The distribution, and the three facts that determine the policy.

One — informal carers supply most long-term care in every country that has measured it. Estimates vary with definition, and the great majority of people needing sustained help receive it primarily from family. Formal services are the minority contribution almost everywhere, including in countries with generous systems.

Two — the distribution is gendered and it is more gendered at higher intensities. Women constitute around two-thirds of carers overall, and a higher share of those providing intensive personal care — washing, dressing, toileting, night-time supervision. Men's contribution is larger in the tasks that are more deferrable and less intimate, which is 8.5.1's pattern again.

Three — the costs to carers are concentrated above a threshold.

Light caring — a few hours a week — has small measured effects on employment, earnings and health.

Intensive caring, conventionally identified at around twenty hours a week or more, has substantial ones: reduced employment, reduced hours, reduced earnings, permanently reduced pension accumulation, and worse measured health.

The health finding requires the usual care (see 7.6.2). Carers are in worse health than non-carers, and part of that is selection — people who become carers differ beforehand, and caring is concentrated in ages when health declines anyway. Designs following the same people through the onset of caring find real effects, smaller than the cross-sectional gap, and concentrated at high intensity.

And the sandwich generation is real and is a minority at any moment — commonly estimated at somewhere between a tenth and a quarter of adults in midlife, depending sharply on how it is defined.

But the snapshot understates it, for 9.2.1's reason. Far more people will occupy that position at some point than occupy it now — which is the developmental cycle point, and it means the experience is close to normal across a life course while looking exceptional in a cross-section.

Grandparents, and the finding that shows how institutional this all is.

Grandparental childcare is enormous in volume and almost invisible in policy. Across European countries, something like half of grandparents provide some care for grandchildren, and a substantial minority provide it regularly.

And the distribution across countries is the interesting part.

Intensive, daily grandparental childcare is highest in countries where formal childcare provision is scarce and expensive — concentrated in southern Europe.

Occasional, supplementary grandparental care is highest where formal provision is extensive and affordable — concentrated in the Nordic countries.

So the two are substitutes, and the pattern is institutional rather than cultural. Where the state provides, grandparents help. Where it does not, grandparents are the childcare system, with the consequences for their own employment, earnings and pensions that follow.

Which means a claim that some societies have stronger family values than others, evidenced by grandparental care, is at least partly a claim about the availability of nurseries — and 7.6.2's question about what else could produce the pattern is answered by looking at the provision map.

Kinship beyond the household, and what shape it now takes.

Two structural changes have altered what a kin network looks like.

More generations are alive at once, and each has fewer members. Falling fertility means fewer siblings, fewer cousins, fewer aunts and uncles; rising longevity means three, four and occasionally five generations coexisting. The network has become taller and narrower — described as a beanpole shape.

The consequence is that vertical relationships have become more important and horizontal ones thinner. A person today is more likely to have a living grandparent and less likely to have four siblings than at any previous point — which changes who is available when help is needed, and concentrates obligation on fewer people.

And financial transfers run predominantly downward. Contrary to the assumption that ageing populations extract resources from the young, the flow of money within families in rich countries is overwhelmingly from older generations to younger ones — deposits, help with rent, support during education, gifts at life events (see 8.3.3).

Which produces an important asymmetry. Money flows down; care flows up. Older generations transfer resources to their children and receive time from them, and both flows are unequally distributed — the families with most to transfer downward are also the families best able to purchase the care that would otherwise flow upward.

Elder care systems: three models, none of which has solved it.

Family obligation, sometimes legally enforced. A number of countries maintain filial responsibility provisions requiring adult children to support parents, with varying enforcement. The system's cost is low and it falls on daughters.

Means-tested residual provision. The state provides for those with insufficient assets, after an assessment — which produces the paradox in the story, and a strong incentive to divest assets.

And social insurance. Several countries have established dedicated long-term care insurance, funded by contributions, providing benefits on an assessed need basis without means-testing. These systems provide considerably more and have not eliminated the family's role, and their costs have risen faster than projected in every case.

Two features recur across all three.

Cost disease (see 8.5.3). Care is labour-intensive and resists productivity growth, so its relative cost rises permanently. No funding settlement is stable, and the recurring political crisis is arithmetic rather than a series of failures.

And migrant labour. Where formal care is provided at wages the domestic workforce will not accept, it is staffed by migrant workers — with the displacement of care down an international chain that 8.5.3 described.

And a note on cash-for-care schemes, which several countries have adopted: paying a benefit that the recipient may use to purchase care, including from a family member. They increase choice and recognise family care as work. They also, at the rates typically set, pay family carers well below the wage they gave up, and can entrench a carer's withdrawal from the labour market — which is a real trade-off rather than a design flaw.

And the dependency ratio, which is quoted constantly and is a weaker measure than it appears.

The old-age dependency ratio — people above pension age divided by people of working age — is the standard summary of population ageing, and it is an accounting artefact of two chosen thresholds.

Three problems.

It counts by age, not by dependency. A healthy, employed sixty-eight-year-old is counted as dependent; an unemployed thirty-year-old is counted as supporting them.

It ignores changes in health. If people are healthier at a given age than a generation ago, a ratio built on a fixed age threshold overstates the change. Alternative measures built on remaining life expectancy — counting people as "old" when they have a given number of years left rather than at a fixed birthday — show much less dramatic ageing.

And what matters for care is not the number of old people but the number needing help. Whether the extra years of life are healthy years is the question, and the evidence on this is genuinely mixed: the hypothesis that morbidity would compress into a short period before death has partial support on some measures and not others, with disability-free life expectancy rising in some countries and periods and stagnating in others.

None of this means ageing is not a real challenge. It means the standard number overstates it and the useful number — healthy life expectancy — is the one to ask for.

Because the largest care system in every country is the one with no budget, no workforce plan and no measurement — and it is being relied upon by systems that assume it while making it harder to supply.

Four things to carry.

Care assessment that counts family availability transfers cost onto families, and it does so progressively in the wrong direction — the more a daughter does, the less is provided.

The costs to carers are concentrated at high intensity, which means the policy-relevant threshold is identifiable and interventions can be targeted at it.

Grandparental care and formal childcare are substitutes, which means a claim about family values is frequently a claim about provision.

And the standard ageing statistic is weaker than the one nobody quotes. Ask for healthy life expectancy.

Three questions.

When care is described as a family responsibility, ask which family member — the answer is a daughter or a daughter-in-law about two-thirds of the time, and more at higher intensity.

When a country is described as having strong family traditions, check its childcare provision.

And when a generational claim is made, ask how age, period and cohort were separated — the answer is an assumption, and it should be stated.

One closing observation. Rekha's twenty hours are not in the national accounts, not in the care workforce statistics, not in her pension, and explicitly deducted from her mother's entitlement.

Every one of those four is a measurement decision (see 7.2.2), and together they constitute the single clearest case in this course of an activity being made invisible by the way it is counted — and then relied upon precisely because it is invisible.

A midlife week supplying around twenty hours of care alongside four days of paid work appears in no measure of output, has permanently reduced a pension, and ends with an assessment recording that family support was available — so the funded package was set lower. In needs-assessed systems, formal provision is calibrated to what the family already does.

The phase exists because the institutions were built for a life course with three phases and two assumptions that both failed: a household with an adult available to absorb contingencies, and few surviving parents in advanced old age.

The life course framework has four principles — historical time and place, timing within a life, linked lives, and agency within constraint — and its underlying claim is that the life course is institutionally organised rather than biologically given, with the tripartite pattern a twentieth-century construction now loosening.

And age, period and cohort cannot be statistically separated, because any one is determined by the other two. Every generational claim rests on an assumption that should be stated.

On who provides care: informal carers supply most long-term care everywhere measured; the distribution is gendered and more so at higher intensities; and the costs — employment, earnings, permanent pension loss, health — are concentrated above roughly twenty hours a week, with the health effects real, smaller than the cross-sectional gap, and partly selected. The sandwich generation is a minority at any moment and close to normal across a life course.

Grandparental childcare and formal provision are substitutes: daily grandparental care is highest where nurseries are scarce, occasional care where they are plentiful — so a claim about family values is often a claim about provision.

Kin networks have become taller and narrower, making vertical ties more important and horizontal ones thinner. Money flows down and care flows up, and the families best able to transfer downward are best able to purchase the care that would otherwise flow up.

Three elder care models — family obligation, means-tested residual, and social insurance — and none has solved it, for the cost disease reason; all rely on migrant labour where wages are low; and cash-for-care schemes recognise family care while typically paying below the forgone wage.

And the old-age dependency ratio counts age rather than dependency, ignores health change, and overstates ageing relative to measures built on remaining life expectancy. Healthy life expectancy is the number to ask for.

Life course — the institutionally organised sequence of life phases; not biologically given.

Linked lives — the interdependence by which a transition in one life produces transitions in others.

Timing — the principle that an event's consequence depends on when in a life it occurs.

De-standardisation — the loosening of the education–employment–retirement sequence.

Age–period–cohort identification problem — the impossibility of separating the three without an assumption.

Informal care — unpaid care by family and friends; the majority of long-term care everywhere measured.

Care intensity threshold — roughly twenty hours a week, above which employment, earnings and health effects concentrate.

Sandwich generation — simultaneous responsibility for a child and a parent; a minority at any moment and common across a life course.

Beanpole family — more generations alive with fewer members in each.

Filial responsibility — legal obligation on adult children to support parents.

Cash-for-care — a benefit purchasing care, including from family members, typically below the forgone wage.

Old-age dependency ratio — an accounting artefact of two age thresholds. Healthy life expectancy — the measure that answers the question the ratio is used for.

One — do the accounting. For any carer you know, estimate the weekly hours and multiply by the local hourly rate for a care worker. Then ask where that figure appears.

Two — check the assessment. Find out whether your country's care needs assessment takes account of available family support. If it does, work out what that means for a daughter who lives nearby.

Three — separate the three effects. Take a claim about a generation and try to distinguish age, period and cohort. Note the assumption you had to make.

Four — map the substitution. Compare grandparental childcare rates and formal childcare provision across three countries. The relationship will be visible.

Five — ask for the right number. Find your country's old-age dependency ratio projection, and then its healthy life expectancy trend. The second answers the question the first is used to raise.