A case study on divergence and convergence of fertility behaviour in Switzerland The value of detailed fertility The value of detailed fertility data data Dr Marion Burkimsher Dr Marion Burkimsher Universities of Geneva and Universities of Geneva and Lausanne Lausanne
56
Embed
A case study on divergence and convergence of fertility behaviour in Switzerland The value of detailed fertility data Dr Marion Burkimsher Universities.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
A case study on divergence and convergence of fertility
behaviour in Switzerland
The value of detailed fertility dataThe value of detailed fertility data
Dr Marion BurkimsherDr Marion Burkimsher
Universities of Geneva and LausanneUniversities of Geneva and Lausanne
GlacierGlacier
SnowSnow
MeltwaterMeltwater
SubliminSublimin-ation-ation PopulationPopulation
BirthsBirths
DeathsDeaths
ImmigrImmigr-ation-ation
EmigrEmigr-ation-ation
GlaciologyGlaciology DemographyDemography
Questions to exploreQuestions to explore
• How are Switzerland's fertility data resources superior to those in How are Switzerland's fertility data resources superior to those in
many other countries?many other countries?
• But what are the weaknesses, and how can they be overcome?But what are the weaknesses, and how can they be overcome?
• Is age at childbearing still rising in Switzerland?Is age at childbearing still rising in Switzerland?
• Why has there been a recent rise in the total fertility rate?Why has there been a recent rise in the total fertility rate?
• Is there more variability in the age at which women have their Is there more variability in the age at which women have their
children now than in the past?children now than in the past?
• Is there more variability in how many children they have than in Is there more variability in how many children they have than in
the past?the past?
Structure of talkStructure of talk
• Introduction to Swiss demographic featuresIntroduction to Swiss demographic features
• Data overview and estimation of biological parityData overview and estimation of biological parity
• Trends in age at birth of each parityTrends in age at birth of each parity
• Period trends in fertility rates by parityPeriod trends in fertility rates by parity
• Cohort fertility ratesCohort fertility rates
• Predictions of fertility ratesPredictions of fertility rates
• Overview of answers to questions posed and conclusionsOverview of answers to questions posed and conclusions
Structure of talkStructure of talk
• Introduction to Swiss demographic featuresIntroduction to Swiss demographic features
• Data overview and estimation of biological parityData overview and estimation of biological parity
• Trends in age at birth of each parityTrends in age at birth of each parity
• Period trends in fertility rates by parityPeriod trends in fertility rates by parity
• Cohort fertility ratesCohort fertility rates
• Predictions of fertility ratesPredictions of fertility rates
• Overview of answers to questions posed and conclusionsOverview of answers to questions posed and conclusions
• Introduction to Swiss demographic featuresIntroduction to Swiss demographic features
• Data overview and estimation of biological parityData overview and estimation of biological parity
• Trends in age at birth of each parityTrends in age at birth of each parity
• Period trends in fertility rates by parityPeriod trends in fertility rates by parity
• Cohort fertility ratesCohort fertility rates
• Predictions of fertility ratesPredictions of fertility rates
• Overview of answers to questions posed and conclusionsOverview of answers to questions posed and conclusions
Data sourcesData sources
BEVNAT - registration of births by age of motherBEVNAT - registration of births by age of mother
ESPOP - population of Swiss women by age at mid-yearESPOP - population of Swiss women by age at mid-year
Census 2000 - answers to the question “Are you the father or Census 2000 - answers to the question “Are you the father or mother of one or several children? If so, how many and what mother of one or several children? If so, how many and what years were they born in?”years were they born in?”
Surveys including fertility question(s): Surveys including fertility question(s):
Fertility and Family Survey (FFS)Fertility and Family Survey (FFS)
World Values Survey/European Values Study (WVS/EVS)World Values Survey/European Values Study (WVS/EVS)
European Social Survey (ESS)European Social Survey (ESS)
Generations and Gender Survey (2013)Generations and Gender Survey (2013)
Birth Registration dataBirth Registration data
1969-1997 Computerised database, but parity only registered 1969-1997 Computerised database, but parity only registered ““au sein du lit actuelau sein du lit actuel”, ie. birth order within the current marriage. ”, ie. birth order within the current marriage. Births outside marriage were all registered as parity 0.Births outside marriage were all registered as parity 0.
1998-2004 Biological parity started to be registered in addition to 1998-2004 Biological parity started to be registered in addition to parity within current marriage, but quite a lot of unknowns.parity within current marriage, but quite a lot of unknowns.
2005-2009.. Accurate recording of both biological parity and parity 2005-2009.. Accurate recording of both biological parity and parity within current marriage. Important as large growth in number of within current marriage. Important as large growth in number of births outside marriage and births in second/subsequent births outside marriage and births in second/subsequent marriages.marriages.
Re-processing of pre-2005 data to obtain estimates of true Re-processing of pre-2005 data to obtain estimates of true biological paritybiological parity
Process of modelling biological parity from 1998-2008 data Process of modelling biological parity from 1998-2008 data to apply to pre-1998 datato apply to pre-1998 data
Assume that proportion of births outside marriage attributable to Assume that proportion of births outside marriage attributable to each biological parity is age-dependent, ie 100% of births to girls each biological parity is age-dependent, ie 100% of births to girls <=15 are 1st births and this % declines with increasing age of <=15 are 1st births and this % declines with increasing age of mothermother
Similarly, where parity in marriage is not equal to biological parity Similarly, where parity in marriage is not equal to biological parity this will be age-dependent. Women have had more possibility for this will be age-dependent. Women have had more possibility for multiple marriages / births outside marriage as they get older! multiple marriages / births outside marriage as they get older!
With 1998-2004 data used only data where biological parity was With 1998-2004 data used only data where biological parity was knownknown
Assumptions for data 1998-2004Assumptions for data 1998-2004
If biological parity was recorded it was considered correctIf biological parity was recorded it was considered correct
The distribution of parities which were recorded as unknown The distribution of parities which were recorded as unknown
follows the same distribution pattern as in the pre-1998 data follows the same distribution pattern as in the pre-1998 data
model (an extrapolation from the 1998-2008 data)model (an extrapolation from the 1998-2008 data)
ModelModel
Calculated the mean % for each age and each parity from the Calculated the mean % for each age and each parity from the
1998-2008 data, then for pre-1998 data re-assigned each parity 1998-2008 data, then for pre-1998 data re-assigned each parity
using these percentages.using these percentages.
Even though there may have been a trend over time, it would be Even though there may have been a trend over time, it would be
dangerous to extrapolate that trend backwards in time - would it dangerous to extrapolate that trend backwards in time - would it
continue linearly, for how long, etc…??continue linearly, for how long, etc…??
This may mean that forThis may mean that for
births outside marriage at older ages - too many assigned to births outside marriage at older ages - too many assigned to
parities greater than 1 (but not so many of them in past)parities greater than 1 (but not so many of them in past)
births within marriage - too many re-assigned to a higher parity births within marriage - too many re-assigned to a higher parity
than there actually werethan there actually were
Attribution of births outside marriage to biological parities
In 1970 there was 11 years between modal age of In 1970 there was 11 years between modal age of 1st and 4th births1st and 4th births
In 2007 the difference was only 2 years!In 2007 the difference was only 2 years!
Change in age at first birth - deciles
18
20
22
24
26
28
30
32
34
36
38
1969 1989 2009
Year
Age at first birth
1st decile
2nd decile
3rd decile
4th decile
Median
6th decile
7th decile
8th decile
9th decile
Median has increased steadily from 23 to 29 over 40 year period (6 years)First decile has gone up from 19 to 22 (3 years)9th decile has gone up from 30 to 36 (6 years)
Change in mean age at birth of each parity 1969-2007
Parity 1 increase started 1994Parity 2 increase started 1995Parity 3 increase started 1998Parity 4 increase started 2001Parity 5+ increase started 2001
Reversal in order! Parity 1 had least variability,Reversal in order! Parity 1 had least variability,now most; high parities were most variable, now leastnow most; high parities were most variable, now least
Structure of talkStructure of talk
• Introduction to Swiss demographic featuresIntroduction to Swiss demographic features
• Data overview and estimation of biological parityData overview and estimation of biological parity
• Trends in age at birth of each parityTrends in age at birth of each parity
• Period trends in fertility rates by parityPeriod trends in fertility rates by parity
• Cohort fertility ratesCohort fertility rates
• Predictions of fertility ratesPredictions of fertility rates
• Overview of answers to questions posed and conclusionsOverview of answers to questions posed and conclusions
In the 2000 Swiss census, each person was asked how many children they had had.
For women born in 1960 - who were, therefore, aged 40, and so approaching the end of their childbearing years - the mean
number of children they had had was 1.73
The Population Reference Bureau’s definition of TFR is “The average number of children that would be born alive to a woman during her lifetime if she were to pass through her childbearing years conforming to the age-specific fertility rates of a given year”
But the mean TFR for the period 1980-1999 was 1.53!
Why the big difference between 1.53 and 1.73?
Possible reasons for mismatch between Possible reasons for mismatch between
period fertility rates and cohort fertility ratesperiod fertility rates and cohort fertility rates
Data errors:Data errors:
Birth registrationsBirth registrations
Population totals by agePopulation totals by age
““Sampling” errors in the censusSampling” errors in the census
Change in population between years of birth and censusChange in population between years of birth and census
Differential mortalityDifferential mortality
Immigration and emigrationImmigration and emigration
Postponement of childbearingPostponement of childbearing
Tempo-adjusted TFR is Tempo-adjusted TFR is the value expected if there had been no the value expected if there had been no change in age at childbearing: change in age at childbearing:
Raw FRRaw FR
(1-(1-rrpp) )
where where rrpp denotes the rate of change in the period mean age at denotes the rate of change in the period mean age at childbearing in year childbearing in year tt..
Needs to be applied to each parity separately because each parity Needs to be applied to each parity separately because each parity may be affected by postponement at different times and to may be affected by postponement at different times and to differing degrees!differing degrees!
Used 5 year moving average of delays for each parity, except for Used 5 year moving average of delays for each parity, except for 2008 (3 yr moving average) and 2009 (extrapolation)2008 (3 yr moving average) and 2009 (extrapolation)
Females born in 1980 Females born in 1975 Females born in 1970Females born in 1965 Females born in 1960 Females born in 1955Females born in 1950 Females born in 1945 Females born in 1940
Cohort sizes have changed considerably (>30% for cohorts 1965 and 1970 during the course of their reproductive life)!
There has been large net immigrationComparison suggests fertility rate of residentsand newcomers is identical
Model for estimating cohort fertility Model for estimating cohort fertility before full reproductive life is completebefore full reproductive life is complete
Can use it as soon as cohort has passed age of peak fertility for Can use it as soon as cohort has passed age of peak fertility for parity 1 (ideally also parity 2)parity 1 (ideally also parity 2)
Use cohort data to complete the curve with current year’s period Use cohort data to complete the curve with current year’s period data. This will ‘probably’ give an data. This will ‘probably’ give an under-estimateunder-estimate of total fertility of of total fertility of youngest cohorts, as ongoing postponement will cause ongoing youngest cohorts, as ongoing postponement will cause ongoing depression of fertility rates (for a while).depression of fertility rates (for a while).
Can be improved by making Bongaarts-Feeney correction to this Can be improved by making Bongaarts-Feeney correction to this added period data. However, this might give an added period data. However, this might give an over-estimateover-estimate of of the total cohort fertility, as there is ‘likely’ to be a slowing down on the total cohort fertility, as there is ‘likely’ to be a slowing down on postponement in the future.postponement in the future.
Changes in family sizes by cohort
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
19201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975Cohort (year of birth)
Proportion with each parity
Childless C+ 1 child C+ 2 children C+ 3 children C+ 4 children C+ 5+ children C+
Dashed lines - extrapolated with period dataDashed lines - extrapolated with period data
Continuous lines - extrapolated with period data with Continuous lines - extrapolated with period data with B-F correction assuming ongoing postponementB-F correction assuming ongoing postponement
MismatchesMismatches
Could be errors in the biological parity model - need to checkCould be errors in the biological parity model - need to check
Could be due to population movementsCould be due to population movements
Could be due to different measures of the resident population inCould be due to different measures of the resident population in
the census and that used to calculate fertility ratesthe census and that used to calculate fertility rates
Or…?Or…?
Decline in proportion of childless women in women >40 after 2000
The proportion childless in 2000 was derived from the the census 2000 ignoring the women who did not declare the number of childlren they had had.The proportion childless in 2009 was the proportion childless in the census 2000 with the addition of births deduced from the parity 1 fertility rates for the years 2001-2009.
Gini coefficient of children per woman and children per mother, census 2000 data and BEVNAT+ extrapolation
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
1930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975Cohort (year of birth)
Gini coefficient
Children/woman census Children/woman BEVNATChildren/mother census Children/mother BEVNAT
Structure of talkStructure of talk
• Introduction to Swiss demographic featuresIntroduction to Swiss demographic features
• Data overview and estimation of biological parityData overview and estimation of biological parity
• Trends in age at birth of each parityTrends in age at birth of each parity
• Period trends in fertility rates by parityPeriod trends in fertility rates by parity
• Cohort fertility ratesCohort fertility rates
• Predictions of fertility ratesPredictions of fertility rates
• Overview of answers to questions posed and conclusionsOverview of answers to questions posed and conclusions
Doubtful, because delay will stop “soon” and then TFR will Doubtful, because delay will stop “soon” and then TFR will
automatically rise, starting with parity 1 fertility rateautomatically rise, starting with parity 1 fertility rate
Likely to rise to 1.7, perhaps even higher??Likely to rise to 1.7, perhaps even higher??
This will happen automatically when postponement halts - This will happen automatically when postponement halts -
no influence from government policies!no influence from government policies!
Year OFS estimated TFR
2010 1.52
2020 1.53
2030 1.52
2040 1.52
2050 1.52
Structure of talkStructure of talk
• Introduction to Swiss demographic featuresIntroduction to Swiss demographic features
• Data overview and estimation of biological parityData overview and estimation of biological parity
• Trends in age at birth of each parityTrends in age at birth of each parity
• Period trends in fertility rates by parityPeriod trends in fertility rates by parity
• Cohort fertility ratesCohort fertility rates
• Predictions of fertility ratesPredictions of fertility rates
• Overview of answers to questions posed and conclusionsOverview of answers to questions posed and conclusions
Questions exploredQuestions explored
• How are Switzerland's fertility data resources superior to those in How are Switzerland's fertility data resources superior to those in
many other countries? many other countries? Biological parity, fertility question in censusBiological parity, fertility question in census
• But what are the weaknesses, and how can they be overcome? But what are the weaknesses, and how can they be overcome?
Modelling of biological parity pre-1998Modelling of biological parity pre-1998
• Is age at childbearing still rising in Switzerland? Is age at childbearing still rising in Switzerland? Yes, sustainedYes, sustained
• Why has there been a recent rise in the total fertility rate? Why has there been a recent rise in the total fertility rate?
Rise in parity 1 fertility rate followed by rise in parity 2 fertility rateRise in parity 1 fertility rate followed by rise in parity 2 fertility rate
• Is there more variability in the age at which women have their Is there more variability in the age at which women have their
children now than in the past? children now than in the past? Yes, especially first birthsYes, especially first births
• Is there more variability in how many children they have than in Is there more variability in how many children they have than in
the past? the past? NoNo
Other snippetsOther snippets
• Change in cohort size can be used to track net immigration, if Change in cohort size can be used to track net immigration, if population registers are good (as in Switzerland)population registers are good (as in Switzerland)• Population decline has not started to happen, despite a TFR well Population decline has not started to happen, despite a TFR well below replacement level for nearly 4 decades, because death below replacement level for nearly 4 decades, because death rates have been similarly deflated because of the rise in life rates have been similarly deflated because of the rise in life expectancy – this may also slow / cease at some time in the futureexpectancy – this may also slow / cease at some time in the future• The final distribution of parities of the cohort of women born in The final distribution of parities of the cohort of women born in the early 1970s is predicted to be: childless 18 - 20%;the early 1970s is predicted to be: childless 18 - 20%;1 child 21- 23%; 2 children around 40% or a little higher; 1 child 21- 23%; 2 children around 40% or a little higher; 3 children 14%; 4 children 2 - 3%; higher parities 1 - 2%3 children 14%; 4 children 2 - 3%; higher parities 1 - 2%• Compared to neighbouring countries, the childlessness rate in Compared to neighbouring countries, the childlessness rate in Switzerland is high (though not quite as high as in W Germany); Switzerland is high (though not quite as high as in W Germany); however 1-child families are rather rare in Switzerland, though however 1-child families are rather rare in Switzerland, though showing a tendency to increase. The proportions with three and showing a tendency to increase. The proportions with three and more children are similar to Italy, Austria and West Germany, more children are similar to Italy, Austria and West Germany, while France has significantly more larger familieswhile France has significantly more larger families
Final conclusions (1)Final conclusions (1)
It would appear that women born post-war in Switzerland are It would appear that women born post-war in Switzerland are
maintaining a very steady distribution of family sizesmaintaining a very steady distribution of family sizes
They are achieving this by giving birth over an increasingly wide They are achieving this by giving birth over an increasingly wide
age range - especially with more births at higher agesage range - especially with more births at higher ages
The parity 1 fertility rate, which can be considered a measure of The parity 1 fertility rate, which can be considered a measure of
propensity for childlessness, is the most variable over time, propensity for childlessness, is the most variable over time,
probably reacting to economic influences: however, postponed probably reacting to economic influences: however, postponed
births are generally recouped at higher ages - increasingly even to births are generally recouped at higher ages - increasingly even to
women over 40. Once the parity 1 rate starts to rise, the parity 2 women over 40. Once the parity 1 rate starts to rise, the parity 2
rate follows a couple of years later..rate follows a couple of years later..
Final conclusions (2)Final conclusions (2)
There was There was convergenceconvergence in family sizes in family sizes up to the cohort born in up to the cohort born in 1946 and since then there has been stability1946 and since then there has been stability
This stability in parity distributions has been achieved by a marked This stability in parity distributions has been achieved by a marked divergencedivergence in timing in timing of births of births
There has been a marked There has been a marked convergenceconvergence in the mean age at in the mean age at births of all paritiesbirths of all parities: women who are going to have larger : women who are going to have larger families start childbearing at a younger age; those who will have families start childbearing at a younger age; those who will have smaller families start latersmaller families start later