sub-replacement fertility: why pure postponement models are inadequate elizabeth sowers & ron...

1
Sub-replacement Fertility: Why Pure Postponement Models are Inadequate Elizabeth Sowers & Ron Lesthaeghe, UC Irvine •Period Total Fertility Rates (PTFRs) for Western Europe from 1960-2003 have have fallen to below replacement level. •Period measures of fertility can be substantially deflated by a shift in the timing of childbirth to older ages 1 (“postponement effect”). •Previous research has tried to estimate the magnitude of a TFR without the birth tempo effect. •Both the Bongaarts-Feeney model (1998) and the Kohler-Philipov adjustment (2001) models assume full recuperation of delayed births, and do not account for situations where postponed fertility is not being made up, or for the disparity in “catching up” among countries. •Research Question: Do pure postponement models have any predictive value for period TFRs? Background and Research Question Analysis of Cohort Fertility Schedules by Age •The graphs below show there is great postponement of births, but that the extent to which fertility declines at young ages is made-up later in life varies greatly. •The cohort analysis is based on the comparison of cumulated age-specific fertility schedules for each cohort to that of the benchmark cohort born 1940-44. •“Catching up” countries have reduced the deficit at age 40 (def (40)) compared to the deficit at age 30 (def(30)), whereas countries that are not catching up have values of def(40) that are equal or larger than d(30) (bottom graphs). D eficits C C FR Finland -600 -500 -400 -300 -200 -100 0 100 <20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+ A ge B rackets and B aseline (1940-44) Deficets, Thousands 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 D eficits C C FR N etherlands -1000 -800 -600 -400 -200 0 200 <20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+ A ge B rackets and B aseline (1940-44) Deficets, Thousands of Bi 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 D eficets C C FR Spain -1200 -1000 -800 -600 -400 -200 0 200 <20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+ A ge B rackets and B aseline (1940-44) Deficets, Thousands of B 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 D eficits C C FR P ortugal -900 -800 -700 -600 -500 -400 -300 -200 -100 0 100 200 <20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+ A ge B rackets and B aseline (1940-44) Deficets, Thousands 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 1980-1984 “Catching Up” Countries like Belgium, Denmark, Finland, France, the Netherlands, Sweden Not “Catching Up” Countries like Austria, Greece, Italy, Portugal, Spain Regression Analysis •The regression analysis uses 3 factors to predict the PTFRs 30 years after the birth of a cohort: CTFR for the baseline cohort, the trough parameter def (30), and the gap reduction parameter def(40)- def(30). • Model 3 explains 79.3% of the variance by regressing the CTFRs of the baseline cohort, the level of the trough parameter (postponement effect), and the recuperation factor on the PTFRs. •The graphs above plot the PTFR against the value predicted by the model, showing that each model is a substantial improvement over the previous one. M odel 1 M odel 2 M odel 3 G ap R eduction -- -- -.917*** Trough P aram eter-- .851*** 1.414*** B aseline C TFR 0.651*** .752*** 1.107*** Constant 318.361 371.068 -286.525 R S quare 0.505 0.673 0.793 R egression M odels Model 1 Model 2 Model 3 Conclusion •Fertility models based on pure postponement are inadequate, and this analysis has shown that the factor of differential catching up is essential to explain the differences in PTFRs. •Countries with lowest-low fertility have not exhibited any catching up after age 30. Countries with PTFRs that have not fallen below 1.50 either 1 Bongaarts & Feeney: 1998; Frejka & Sardon: 2006). Fertility data source: Council of Europe (2004) Recent Demographic Developments in Europe – 2004 Edition (CD). ( Council of Europe: Strasbourg, France). For more information on postponement and catching up, see Lesthaeghe & Willems (1999), Lesthaeghe (2001), Frejka & Calot (2001), and Sobotka (2003). Data and References

Upload: gilbert-wright

Post on 05-Jan-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Sub-replacement Fertility: Why Pure Postponement Models are Inadequate Elizabeth Sowers & Ron Lesthaeghe, UC Irvine Period Total Fertility Rates (PTFRs)

Sub-replacement Fertility: Why Pure Postponement Models are InadequateElizabeth Sowers & Ron Lesthaeghe, UC Irvine

•Period Total Fertility Rates (PTFRs) for Western Europe from 1960-2003 have have fallen to below replacement level.

•Period measures of fertility can be substantially deflated by a shift in the timing of childbirth to older ages1 (“postponement effect”).

•Previous research has tried to estimate the magnitude of a TFR without the birth tempo effect.

•Both the Bongaarts-Feeney model (1998) and the Kohler-Philipov adjustment (2001) models assume full recuperation of delayed births, and do not account for situations where postponed fertility is not being made up, or for the disparity in “catching up” among countries.

•Research Question: Do pure postponement models have any predictive value for period TFRs?

Background and Research Question Analysis of Cohort Fertility Schedules by Age

•The graphs below show there is great postponement of births, but that the extent to which fertility declines at young ages is made-up later in life varies greatly.

•The cohort analysis is based on the comparison of cumulated age-specific fertility schedules for each cohort to that of the benchmark cohort born 1940-44.

•“Catching up” countries have reduced the deficit at age 40 (def (40)) compared to the deficit at age 30 (def(30)), whereas countries that are not catching up have values of def(40) that are equal or larger than d(30) (bottom graphs).

•Clearly, postponement is occurring, but not all countries are recuperating from the deficit of births at early ages.

Deficits CCFR Finland

-600

-500

-400

-300

-200

-100

0

100

<20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+

Age Brackets and Baseline (1940-44)

Def

icet

s, T

ho

usa

nd

s o

f B

irth

s

1945-1949

1950-1954

1955-1959

1960-1964

1965-1969

1970-1974

1975-1979

1980-1984

Deficits CCFR Netherlands

-1000

-800

-600

-400

-200

0

200

<20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+

Age Brackets and Baseline (1940-44)

Def

icet

s, T

ho

usa

nd

s o

f B

irth

s

1945-1949

1950-1954

1955-1959

1960-1964

1965-1969

1970-1974

1975-1979

1980-1984

Deficets CCFR Spain

-1200

-1000

-800

-600

-400

-200

0

200

<20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+

Age Brackets and Baseline (1940-44)

Def

icet

s, T

ho

usa

nd

s o

f B

irth

s

1945-1949

1950-1954

1955-1959

1960-1964

1965-1969

1970-1974

1975-1979

1980-1984

Deficits CCFR Portugal

-900-800-700-600-500-400-300-200-100

0100200

<20 <20-24 <20-30 <20-34 <20-39 <20-44 <20-45+

Age Brackets and Baseline (1940-44)

Def

icet

s, T

ho

usa

nd

s o

f B

irth

s

1945-1949

1950-1954

1955-1959

1960-1964

1965-1969

1970-1974

1975-1979

1980-1984

“Catching Up”

Countries like Belgium, Denmark,

Finland, France, the Netherlands,

Sweden

Not “Catching Up”

Countries like Austria, Greece, Italy, Portugal,

Spain

Regression Analysis

•The regression analysis uses 3 factors to predict the PTFRs 30 years after the birth of a cohort: CTFR for the baseline cohort, the trough parameter def (30), and the gap reduction parameter def(40)-def(30).

• Model 3 explains 79.3% of the variance by regressing the CTFRs of the baseline cohort, the level of the trough parameter (postponement effect), and the recuperation factor on the PTFRs.

•The graphs above plot the PTFR against the value predicted by the model, showing that each model is a substantial improvement over the previous one.

Model 1 Model 2 Model 3

Gap Reduction -- -- -.917***

Trough Parameter -- .851*** 1.414***

Baseline CTFR 0.651*** .752*** 1.107***

Constant 318.361 371.068 -286.525

R Square 0.505 0.673 0.793

Regression Models

Model 1 Model 2 Model 3

Conclusion•Fertility models based on pure postponement are inadequate, and this analysis has shown that the factor of differential catching up is essential to explain the differences in PTFRs.

•Countries with lowest-low fertility have not exhibited any catching up after age 30. Countries with PTFRs that have not fallen below 1.50 either have high baseline CTFRs or clearly exhibit catching up.

1 Bongaarts & Feeney: 1998; Frejka & Sardon: 2006).

Fertility data source: Council of Europe (2004) Recent Demographic Developments in Europe – 2004 Edition (CD). (Council of Europe: Strasbourg, France).

For more information on postponement and catching up, see Lesthaeghe & Willems (1999), Lesthaeghe (2001), Frejka & Calot (2001), and Sobotka (2003).

Data and References