Page 1
The data
Education and the contemporary ‘baby boom’: Evidence from the Northern Ireland Longitudinal
Study
Patrick McGregor Patricia McKee
“Predicting Short Run Changes in Fertility in Northern Ireland”
a project funded by
The authors are grateful to the NILS team at NISRA for their assistance
Page 2
2
Education and the contemporary ‘baby boom’: Evidence from the Northern Ireland Longitudinal Study
• Overview• Education and fertility• The NILS• The data• The statistical model• Results• Conclusion
Page 3
Overview 3
A baby-boom fuelled by rising fertility rates and immigration has pushed the population of the United Kingdom to more than 61 million for the first time.
The Times, 28 August 2009
Page 4
The data 4
Total Births
Age Specific Fertility Rates
20000
22000
24000
26000
28000
30000
32000
1977 1982 1987 1992 1997 2002 2007
year
birt
hs r
egis
tere
d
0
25
50
75
100
125
150
175
200
1977 1982 1987 1992 1997 2002 2007
Ag
e-s
pe
cific
fert
ility
ra
te
15-19 20-24 25-29 30-34 35-39 40-44
Fertility in Northern Ireland
Page 5
Education and fertility 5
Education and fertility
Static economic analysis
Max ST
Children assumed to be normal ‘goods’ so: but possibly or possibly not!
The home production framework allows the cost of children to be expressed as a function of the parents’ wages and their respective shares in the costs of producing child quality
sqnUU ,,sc snqI
0 qn 0n
Page 6
Education and fertility 6
Dynamic economic analysis The ‘user cost’ of a child now is a function of a
sequence of prices such as the female wage rate
The optimal profile of a woman’s stock of human capital will be jointly determined with the timing of the births of her children.
Any empirical analysis should permit the demographic profile to vary with educational attainment
Page 7
The NILS 7
The NILS
• The NILS potential mothers: those women with health card registrations, aged 16-44 years and whose DOB is one of the 104 in the systematic sample
• Registrations downloaded biannually and constitute potential panel members
• Details of any birth to a NILS mother are forwarded by the GRO to the NILS
• 2001 Census: An attempt is made to link the Census details of all NILS mothers
Page 8
The NILS 8
Population on Census Day, 2001
Total Population
Census 1,685,267 (4.84% Imputed)
BSO 1,768,473 (4.94% list inflation)
Women in NILS aged 16-44 years
Census 101,034 ( Census x 104/365.25)
BSO 107,874
list inflation 6,840 (6.3%)
Expected imputed census records 4,890 (101,034 x 0.0484)
Fertility Panel with census records 93,601
BSO – FP 14,273
Mismatch = 14,273 – 6,840 – 4,890 = 2,543
Page 9
The data 9
Year N % pattern01 5083 3.5 100000001-02 5167 3.6 110000001-03 5771 4.0 111000001-04 4523 3.1 111100001-05 4804 3.3 111110001-06 4451 3.1 111111001-07 76878 53.0 111111102-07 4416 3.1 011111103-07 4353 3.0 001111104-07 4649 3.2 000111105-07 5067 3.5 000011106-07 5796 4.0 0000011 07 6504 4.5 0000001
Presence of women in the Fertility Panel:the effect of Age
born 1957 1977 1978 1991age 2001 44 24 23 10age 2007 50 30 29 16
The fertility panel essentially is the 1957 – 77 cohort and ignores the 1978 – 91 cohort due to the absence of educational information
Page 10
The data 10
Sources of Information available for the NILS
Fertility Panel
101
Census 91
4,653
7,019
Census 01
GRO
BSO
61,263
24,041
32,960
530
26,710
Page 11
The data 11
Sources of Parity in the Fertility Panel
Parity from Freq. PercentCensus - offspring 116,327 74.0GRO births 10,141 6.591 census 6,432 4.1Count births 97-01 178 0.1Census + n of births 158 0.1Missing 24,041 15.3
Total Women 157,277 100.0
Page 12
The data 12
Educational Qualifications
• Level 0
• Level 1
• Level 2
• Level 3
• Level 4
• Level 5
No qualifications
GCSE grade D-G; 1-4 CSEs grade 1;
1-4 ‘O’ level passes; NVQ level 1
5+ CSEs grade 1; 5+ GCSEs grade A-C;
5+ ‘O’ level passes; NVQ level 2
2+ ‘A’ levels; 4+ AS levels;
NVQ level 3 or GNVQ Advanced
First degree; NVQ level 4; HNC ; HND
Higher degree; NVQ level 5
Page 13
The data 13
2325
20
10
22
0
5
10
15
20
25
30
no quals level 1 level 2 level 3 level 4/5
Education level
Pro
port
ion
of s
ampl
e
The Distribution of Educational Attainment
Page 14
The data 14Source: 2001 Census
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
19
81
19
82
19
83
19
84
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
year
pro
po
rtio
n
none lev 1 lev 2 lev 3 uni deg
Distribution of Educational Attainment by Year that Woman was aged 24
Page 15
The data 15
0
100
200
300
400
500
600
1997 1999 2001 2003 2005 2007
Year
Gro
ss
We
ek
ly p
ay
(£
)
63.0
64.0
65.0
66.0
67.0
68.0
Act
ivit
y R
ate
Q1 median Q3 Activity Rate
Women in the Labour Market
Source: DETINI
Page 16
The data 16
First Birth
22
25
28
31
34
1997 1999 2001 2003 2005 2007
Year
mea
n a
ge
of
mo
ther
no quals level 1
Third Birth
22
25
28
31
34
1997 1999 2001 2003 2005 2007
Yearm
ean
ag
e o
f m
oth
erlevel 4/5
Second birth
22
25
28
31
34
1997 1999 2001 2003 2005 2007
Year
mea
n a
ge
of
mo
ther
level 2 level 3
Mean Age of Mother at Time of Birth
Page 17
The Statistical Model 17
The Statistical Model
• Raftery, AE, Lewis, SM and Aghajanian, A (1995).• Demand or Ideation? Evidence from the Iranian
Marital Fertility Decline, Demography, vol. 32.• Data: 1977 Iran Fertility Survey• “ each woman-year of exposure is treated as a
separate case” • Five clocks:
– Age Period Cohort Parity Duration
Page 18
The Statistical Model 18
The Logit Model
iiiii
iiiiii
XEDUCATIONECONOMIC
TIMENSINTERACTIODURATIONPARITYAGEB
/
76
54321*
0*0
0*1
ii
ii
BifB
BifB
EDUCATION includes interactions with PARITY, DURATION and TIME so demographic profile can vary with educational attainment
TIME captures the change in quantum for women aged > 23
Page 19
Results 19
Demographic
age 237.053age2 -309.685age3 182.198age4 -41.369
par1 2.078par2 2.595par3 2.161pargt3 0.899
dur1 -3.974dur2 -2.483dur3 -1.661dur4 -1.783dur5 -1.540
Demographic interactions
agepar1 -1.420agepar2 -2.663agepar3 -2.441agepargt3 -1.341
agedur1 3.512agedur2 3.162agedur3 2.442agedur4 2.283agedur5 1.927
Education level01constant 0.651quality -2.003timetrend -0.173quality*tt 0.343
Demographic par0 -0.384par1 -0.386par2 -0.178par3 -0.096
dur1 0.046dur2 -0.234dur3 -0.195dur4 -0.091dur5 -0.140durgt5 -0.03
Logit Regression Results Education level 45constant 1.438quality -5.470timetrend -0.145quality*tt 0.702
interactionspar0 -0.010par1 0.258par2 0.238par3 0.094
dur1 0.138dur2 0.484dur3 0.468dur4 0.557dur5 0.394durgt5 0.394
Timeyr2002 -0.457yr2004 -0.427yr2005 -1.473yr2006 -0.111timetrend 0.667
Economicfwage-1 -1.423factivityrate-1 -0.971Interest-1 0.244
Othercatholic 0.169constant -2.758
Page 20
Results 20
Demographic
age 237.053age2 -309.685age3 182.198age4 -41.369
par1 2.078par2 2.595par3 2.161pargt3 0.899
dur1 -3.974dur2 -2.483dur3 -1.661dur4 -1.783dur5 -1.540
Demographic interactions
agepar1 -1.420agepar2 -2.663agepar3 -2.441agepargt3 -1.341
agedur1 3.512agedur2 3.162agedur3 2.442agedur4 2.283agedur5 1.927
Education level01constant 0.651quality -2.003timetrend -0.173quality*tt 0.343
Demographic par0 -0.384par1 -0.386par2 -0.178par3 -0.096
dur1 0.046dur2 -0.234dur3 -0.195dur4 -0.091dur5 -0.140durgt5 -0.03
Logit Regression Results Education level 45constant 1.438quality -5.470timetrend -0.145quality*tt 0.702
interactionspar0 -0.010par1 0.258par2 0.238par3 0.094
dur1 0.138dur2 0.484dur3 0.468dur4 0.557dur5 0.394durgt5 0.394
Timeyr2002 -0.457yr2004 -0.427yr2005 -1.473yr2006 -0.111timetrend 0.667
Economicfwage-1 -1.423factivityrate-1 -0.971Interest-1 0.244
Othercatholic 0.169constant -2.758
The demographic profile of the base educational category is well determined with parity=0 and durgt5 base
Page 21
Results 21
Work and childbearing tend to be separate rather than combined activities; wage rate not significant although divided by quartiles; interest rate?
Demographic
age 237.053age2 -309.685age3 182.198age4 -41.369
par1 2.078par2 2.595par3 2.161pargt3 0.899
dur1 -3.974dur2 -2.483dur3 -1.661dur4 -1.783dur5 -1.540
Demographic interactions
agepar1 -1.420agepar2 -2.663agepar3 -2.441agepargt3 -1.341
agedur1 3.512agedur2 3.162agedur3 2.442agedur4 2.283agedur5 1.927
Education level01constant 0.651quality -2.003timetrend -0.173quality*tt 0.343
Demographic par0 -0.384par1 -0.386par2 -0.178par3 -0.096
dur1 0.046dur2 -0.234dur3 -0.195dur4 -0.091dur5 -0.140durgt5 -0.03
Logit Regression Results Education level 45constant 1.438quality -5.470timetrend -0.145quality*tt 0.702
interactionspar0 -0.010par1 0.258par2 0.238par3 0.094
dur1 0.138dur2 0.484dur3 0.468dur4 0.557dur5 0.394durgt5 0.394
Timeyr2002 -0.457yr2004 -0.427yr2005 -1.473yr2006 -0.111timetrend 0.667
Economicfwage-1 -1.423factivityrate-1 -0.971Interest-1 0.244
Othercatholic 0.169constant -2.758
Page 22
Results 22
Timeyr2002 -0.457yr2004 -0.427yr2005 -1.473yr2006 -0.111timetrend 0.667
Economicfwage-1 -1.423factivityrate-1 -0.971Interest-1 0.244
Othercatholic 0.169constant -2.758
There is a fluctuating increase in the quantum of those 24 and over
Demographic
age 237.053age2 -309.685age3 182.198age4 -41.369
par1 2.078par2 2.595par3 2.161pargt3 0.899
dur1 -3.974dur2 -2.483dur3 -1.661dur4 -1.783dur5 -1.540
Demographic interactions
agepar1 -1.420agepar2 -2.663agepar3 -2.441agepargt3 -1.341
agedur1 3.512agedur2 3.162agedur3 2.442agedur4 2.283agedur5 1.927
Education level01constant 0.651quality -2.003timetrend -0.173quality*tt 0.343
Demographic par0 -0.384par1 -0.386par2 -0.178par3 -0.096
dur1 0.046dur2 -0.234dur3 -0.195dur4 -0.091dur5 -0.140durgt5 -0.03
Logit Regression Results Education level 45constant 1.438quality -5.470timetrend -0.145quality*tt 0.702
interactionspar0 -0.010par1 0.258par2 0.238par3 0.094
dur1 0.138dur2 0.484dur3 0.468dur4 0.557dur5 0.394durgt5 0.394
Page 23
Results 23
Timeyr2002 -0.457yr2004 -0.427yr2005 -1.473yr2006 -0.111timetrend 0.667
Economicfwage-1 -1.423factivityrate-1 -0.971Interest-1 0.244
Othercatholic 0.169constant -2.758
For those from Northern Ireland
Demographic
age 237.053age2 -309.685age3 182.198age4 -41.369
par1 2.078par2 2.595par3 2.161pargt3 0.899
dur1 -3.974dur2 -2.483dur3 -1.661dur4 -1.783dur5 -1.540
Demographic interactions
agepar1 -1.420agepar2 -2.663agepar3 -2.441agepargt3 -1.341
agedur1 3.512agedur2 3.162agedur3 2.442agedur4 2.283agedur5 1.927
Education level01constant 0.651quality -2.003timetrend -0.173quality*tt 0.343
Demographic par0 -0.384par1 -0.386par2 -0.178par3 -0.096
dur1 0.046dur2 -0.234dur3 -0.195dur4 -0.091dur5 -0.140durgt5 -0.03
Logit Regression Results Education level 45constant 1.438quality -5.470timetrend -0.145quality*tt 0.702
interactionspar0 -0.010par1 0.258par2 0.238par3 0.094
dur1 0.138dur2 0.484dur3 0.468dur4 0.557dur5 0.394durgt5 0.394
Page 24
Results 24
Timeyr2002 -0.457yr2004 -0.427yr2005 -1.473yr2006 -0.111timetrend 0.667
Economicfwage-1 -1.423factivityrate-1 -0.971Interest-1 0.244
Othercatholic 0.169constant -2.758
The differential quantum effect of education for those aged >23 in the 1957 – 77 cohort
Demographic
age 237.053age2 -309.685age3 182.198age4 -41.369
par1 2.078par2 2.595par3 2.161pargt3 0.899
dur1 -3.974dur2 -2.483dur3 -1.661dur4 -1.783dur5 -1.540
Demographic interactions
agepar1 -1.420agepar2 -2.663agepar3 -2.441agepargt3 -1.341
agedur1 3.512agedur2 3.162agedur3 2.442agedur4 2.283agedur5 1.927
Education level01constant 0.651quality -2.003timetrend -0.173quality*tt 0.343
Demographic par0 -0.384par1 -0.386par2 -0.178par3 -0.096
dur1 0.046dur2 -0.234dur3 -0.195dur4 -0.091dur5 -0.140durgt5 -0.03
Logit Regression Results Education level 45constant 1.438quality -5.470timetrend -0.145quality*tt 0.702
interactionspar0 -0.010par1 0.258par2 0.238par3 0.094
dur1 0.138dur2 0.484dur3 0.468dur4 0.557dur5 0.394durgt5 0.394
Page 25
Results 25
Education level01constant 0.651quality -2.003timetrend -0.173quality*tt 0.343
Demographic par0 -0.384par1 -0.386par2 -0.178par3 -0.096
dur1 0.046dur2 -0.234dur3 -0.195dur4 -0.091dur5 -0.140durgt5 -0.03
Timeyr2002 -0.457yr2004 -0.427yr2005 -1.473yr2006 -0.111timetrend 0.667
Economicfwage-1 -1.423factivityrate-1 -0.971Interest-1 0.244
Othercatholic 0.169constant -2.758
The differential tempo effect of education for those aged >23 in the 1957 – 77 cohort
Demographic
age 237.053age2 -309.685age3 182.198age4 -41.369
par1 2.078par2 2.595par3 2.161pargt3 0.899
dur1 -3.974dur2 -2.483dur3 -1.661dur4 -1.783dur5 -1.540
Demographic interactions
agepar1 -1.420agepar2 -2.663agepar3 -2.441agepargt3 -1.341
agedur1 3.512agedur2 3.162agedur3 2.442agedur4 2.283agedur5 1.927
Logit Regression Results Education level 45constant 1.438quality -5.470timetrend -0.145quality*tt 0.702
interactionspar0 -0.010par1 0.258par2 0.238par3 0.094
dur1 0.138dur2 0.484dur3 0.468dur4 0.557dur5 0.394durgt5 0.394
Page 26
Results 26
Marginal Effects
0,1ˆ1,1ˆ, educXBPeducXBPME iiiiieduc
Where is the vector of values of all the explanatory variables except for educ
iX
For ith woman:
Educational variables are ordered:
noqual base degree
% sample 48 30 22
The marginal effect is taken as the mean of individual marginal effects over the particular category and the base
Page 27
The data 27
Parity 0 1 2 3 mean par at 24No quals 0.043 0.084 0.039 0.033 0.98others 0.066 0.132 0.049 0.037 0.51Degree 0.069 0.193 0.073 0.050 0.10
Table: Probability of Birth 2001-7
Marginal Effect of Education
No qualifications
-0.05
0
0.05
0.1
0.15
24 26 28 30 32 34 36 38 40 42 44
Age
Mar
gin
al e
ffec
t
Degrees
-0.05
0
0.05
0.1
0.15
24 26 28 30 32 34 36 38 40 42 44
Age
Mar
gin
al e
ffec
t
par=0 tt only
par=1 tt only
par=2 tt only
par=0 + dem
par=1 + dem
par=2 + dem
Page 28
Results 28
Conclusions
• The total marginal effect of educational attainment , including both quantum and tempo effects, is almost always positive
• It peaks at 0.16 for women aged 33 with parity = 1 for the degree case; for parity = 0 the marginal effect is about one third of this
• In the degree case the quantum effects are negative up to age 28 possibly reflecting the decline in degree ‘quality’
• For the poorly qualified the pattern is similar, except that the peak for parity = 1 is earlier, at 30 and smaller, at 0.11
• For most ages, the marginal effect is greater for parity 2 than 0
Page 29
Results 29
parity = 0
0
10
20
30
40
50
60
70
80
90
100
24 26 28 30 32 34 36 38 40 42 44
age
%
noqual
middle
degree
parity = 1
0
5
10
15
20
25
30
35
40
24 26 28 30 32 34 36 38 40 42 44
age
%
noqual
middle
degree
parity = 2
0
5
10
15
20
25
30
35
40
24 26 28 30 32 34 36 38 40 42 44
age
%
noqual
middle
degree
parity = 3
0
5
10
15
20
25
30
35
40
24 26 28 30 32 34 36 38 40 42 44
age
%
noqual
middle
degree
Parity Proportions by Age and Education
Page 30
The data 30
Births and Birth Events
N of births Freq. Percent
0 371,113 84.521 66,944 15.252 995 0.23
3+ 24 0.01
Total 439,076 100
In the analysis birth events rather than births per se are examined