Centre for Market and Public Organisation

Centre for Market and Public Organisation

Understanding the effect of public policy on fertility

Mike Brewer (Institute for Fiscal Studies)

Anita Ratcliffe (CMPO, University of Bristol)

Sarah Smith (CMPO and IFS)

The project

• Falling total fertility rates sparking interest in pro-natalist policies (France, Italy, Japan, Germany)

• Phase 1: Understanding trends in fertility in the UK

– What have been the main changes in fertility behaviour?

– (How) Do these trends vary by education?

• Phase 2: The impact of welfare reform on fertility

– New Labour substantially increased financial help for families (working families tax credit, child tax credit)

– Does fertility respond to financial incentives?

Main Findings

• Phase 1

• Distinctive cohort trends in fertility

– 1935-45 fall in higher-order births

– 1945-55 rise in childlessness & one-child families

– 1955-62 fertility patterns relatively stable

– 1962 - ? further postponement of fertility

• Biggest falls in fertility are among educated women (19+)

• Phase 2

• Preliminary results suggest a small, significant effect of WFTC on fertility

Phase 1: Cohort trends in fertility

• Pooled cross-section data

– Family Expenditure Survey 1968 – 2003

– Family Resources Survey 1990 – 2004

• Own-child method

– Use age of mother and children to infer age of birth; birth order

– Construct (age-specific) parity progression ratios by cohort

1955 cohort: What proportion have a first birth at age 25?

Combine “current” information: those aged 25 who haveone child aged 0 in 1980 survey…

… and “retrospective” information:1981 survey – those aged 26 whose oldest child is aged 11982 survey – those aged 27 whose oldest child is aged 2and so on…

Use survival analysis to estimate proportion of each cohort who havea birth of order n by age x

Current and retrospective information is assumed to be equally valid;regression analysis shows no systematic effect ofdistance of survey year on estimated probability of birth

Survey year Age of cohort1970 151971 161972 171973 181974 191975 201976 211977 221978 231979 241980 251981 261982 271983 281984 291985 301986 311987 321988 331989 341990 351991 361992 37

Constructing fertility histories

Potential measurement error/ selection problems

• Infant mortality

• Household re-formation

– Rely on the fact that most children remain with mother

• Children leaving home

– Only looking at births 15 – 37

• Advantages

– Long time-series (cohorts born 1935 – 1975)

– Large sample sizes, allow us to look at individual date of birth cohorts

– Information on birth order

– Demographic information – region, education, employment

Annual total fertility rate = number of children a woman would haveif she had the age-specific birth rates in that year

• Average family size falls – but by less than TFR • Phase 1 (1935 – 1950) – decline in higher order births• Phase 2 (1945 – 1955) – decline in first and second births• Phase 3 (1955 – 1962) – period of stability• Phase 4 ??

1.5

1.7

52

2.2

52

.52

.75

Com

ple

ted

fam

ily s

ize

1935 1940 1945 1950 1955 1960 1965Date of birth

all women women with children

Completed family size, by cohort

0.1

.2.3

.4.5

Pro

por

tion

of

coh

ort

1935 1940 1945 1950 1955 1960 1965Date of birth

childless onechild

twochild threeplus

• Delay in family formation beginning with 1945 cohort

1.8

.6.4

.20

Pro

por

tion

of

coh

ort

1935 1940 1945 1950 1955 1960 1965 1970 1975Date of birth

first birth second birth

Proportion giving birth before age 30

• Northern Ireland experienced biggest fall in completed family size• Of the three other regions:• Wales has highest completed family size• Scotland experienced a greater fall than England and Wales

3.5

32

.52

1.5

1C

ompl

ete

d f

amily

siz

e

1945 1950 1955 1960 1965Date of birth

England Wales

N Ireland Scotland

Completed family size, by cohort and region

Decline in fertility cannot be explained just in terms of greater participation in higher education (19+).

High education group have experienced a greater (absolute and relative) decline in fertility

Completed family size, by cohort and education

11

.25

1.5

1.7

52

2.2

5P

rop

ortio

n o

f co

ho

rt

1945 1950 1955 1960 1965Date of birth

low education high education

Greater rise in childlessness among high education group, particularly cohorts born 1950 – 55

.1.2

.3.4

.5

1945 1950 1955 1960 19651945 1950 1955 1960 1965

low education high education

childless onechildtwochild threeplus

Pro

por

tion

of

coh

ort

Date of birth

Graphs by ed

Completed family size, by cohort and education

Greater delay in child-bearing among higher educated women

0.2

.4.6

.8

1945 1950 1955 1960 1965 1970 19751945 1950 1955 1960 1965 1970 1975

low education high education

first birth second birth

Pro

por

tion

of

coh

ort

Date of birth

Graphs by ed

Proportion giving birth before age 30

A tale of four “cohorts”

• Women born 1935 – 45: Family then part-time work

Average age at first birth fell, deliberate choice to limit family size to two (pill)

Little increase in full-time employment, but more women with kids working part-time

• Women born 1945 – 55: Career then family

Average age at first birth rose, increase in childlessness & one-child families

Increase in full-time work (particularly educated women), but not among women with children

• Women born 1955 – 65: Career and family (for some)

Average age at first birth continued to rise, but rates of childbearing fairly constant

Rise in proportion of women with children who work full-time, particularly educated women

• Women born 1965 onwards

Huge increase in proportion of women going on to higher education

Average age at first birth rising faster, implications for proportion remaining childless, one-child families?

Phase 2: The effect of WFTC on fertility

• Working Families Tax Credit introduced in 1999

– Higher benefits for (low-income) families with children

– In theory, should raise fertility

• Evaluation strategy: Difference-in-differences

– Compare fertility “before” and “after” the reform for couples affected by WFTC reform: the “treatment” group = low education

– Contrast with change over the same time period for couples not affected by the reform: the “control” group = high education

• Data

– FES/FRS – on an annual rather than a cohort basis

– BHPS – look at employment/fertility transitions

Phase 2: The effect of WFTC on fertility

• Sample issues

• Women aged 20 – 37 who were in a couple at the time of interview– Lone parents

• Reform has positive employment effects, which may reduce fertility

• Fertility complicated by partnership formation issues

• Before = April 1985 – March 1999; After = April 2000 – March 2002

• Low education, treatment group defined as both partners left school at the compulsory school leaving age

Phase 2: The effect of WFTC on fertility

One child Two children Three children

1998/9 £13.87 £25.14 £36.41

2002/03 £28.97 £40.77 £52.41

• Big increase in median weekly payments to couples with children (2006 prices), bigger for first children

• Magnitude of increase correlated with education

Phase 2: The effect of WFTC on fertility

Phase 2: The effect of WFTC on fertility

• Regression analysis

– Controlling for age of mother, education, presence and age of kids, region

– Allowing for common trend, differential trend & common change in trend

• Results

– The probability of a birth increases by 0.75 percentage points among the treatment group following the introduction of WFTC (statistically significant at the 10% level)

– The effect is greater for first births. The probability of a first birth increases by 1.77 points (statistically significant at the 5% level). There is no significant effect for women who already have children

– There is no difference in effect by the age of the mother

– NB we cannot distinguish tempo from quantity effects

Further work

• Project finishes Dec 2006

• Robustness checks on regression analysis

– Define treatment and control in terms of income

– Experiment with spurious reforms

• Use panel data (BHPS) to examine employment/fertility transitions

• Finish writing up

• Disseminate!