The Relationship between Consumption, Labor Supply and Fertility Theory and Evidence from Japan

1
The Relationship between Consumption, Labor
Supply and Fertility - Theory and Evidence from
Japan

• Reiko Aoki
• Center for Intergenerational Studies, IER,
  Hitotsubashi University
• Yoko Konishi
• Research Institute of Economy, Trade and Industry

2
Motivation

• Positive relationship between total fertility
  rate (TFR) and female labor participation rate
  (FLPR) among OECD countries
• But there are OECD countries with negative
  relationship time series
• Similar phenomenon in Japan
• Negative time series relationship
• Positive cross sectional relationship

3
Selected OECD Countries

• Negative time series relationship

The Netherlands 1970
Total Fertility Rate
Norway 1970
U.S.A. 1970
JAPAN 1970
Female Labor Supply Ages 15 to 64 ()
NOTE) 5 time points 1970, 1980, 1985, 1990, and
2000.
4
Japanese cross section
NOTE) Pink points are TYPE1(low declining rate in
TFR and high level of TFR and female labor
supply). Blue points are TYPE7(high declining
rate in TFR and low level in TFR and female labor
supply). Sources) Ministry of Internal Affairs
and Communications Employment Status Survey,"
National Institute of Population and Social
Security Research Indicators of Fertility by
Prefecture in 1970-1985, and Health, Labor and
Welfare Ministry Population Survey Report.
5
Consumer Optimization

• Both consumption and childrearing takes time
  (labor) and goods
• Optimization with redefined opportunity set

Number of children Consumption
experience Utility function
6
Opportunity Set
7
Effects of Higher Quality
8
We further index consumption by quality,
Q. Utility function
where z measures quality of consumption.
First-order condition for utility maximization
are
9
Effect of Higher Quality
10
Changes over Time
11
Time Series and Cross Section

• Time series
• Both quality and wages changing
• More children but less time devoted to
  childrearing
• Negative relationship between consumption and
  number of children
• Cross section
• Regional differences in quality of consumption
• Negative relationship between consumption and
  number of children

12
Higher Quality and Wages- General Equilibrium
Model

• High and low ( standard) quality goods
• Heterogeneous consumers
• Different preference over children and quality
  adjusted consumption
• Skilled or unskilled worker
• Skilled and unskilled labor
• High and low quality goods
• High and low quality income

13
Consumer

• Each consumer is skilled or unskilled

14
Production and Labor Market

• CRS production and competitive market

• Labor Markets ( skilled and unskilled )
• Supply from consumers
• Demand
• Unskilled from standard product production
• Skilled from high quality product production

15
Equilibrium
Figure 4
16
Effect of Higher Quality to Wages
Figure 5
17
Empirical Analysis

• We examine the empirical evidence to support the
  theoretical implications of the previous sections
  by Japanese prefectures (regions) data.
• We estimate the equations that assume that
  regional TFR is affected by regional variables
  that reflect quality or variety of consumption
  goods.
• We employ the fixed effects model to take into
  account the unobserved heterogeneity among
  regions.

18
Correlation Coefficient (2005) 47 prefectures
The correlation coefficient is close to 0.5
between TFR and FLRP. This is similar to the
phenomenon observed in OECD countries in recent
years.
19
Figure 7 Correlation Coefficient between TFR and
FLPR
Figure 2plots correlation coefficients between
regional TFR and FLRP by the 8 years from 1970 to
2005. The coefficient is negative for 1975 but is
positive thereafter.
20
Data Description
We use data from 47 prefectures for years 1970,
75, 80, 85, 90, 2000, and 2005 (Okinawa
prefecture is not included in 1970).
Table 1. Description of Variables
21
Regression Model 1 Cross sectional regression
Table 3. Estimation Coefficient of FLPR

• FLPR coefficient is no longer significant at the
  5 level when consumption variables added
• However, the coefficient is significantly
  positive when cross sections are pooled for 1975
  -2005 with 0.066.( indicates significance at
  5 level)

22
Regression Models Panel regression in Table 4
(next slide)
23
(No Transcript)
24
Fixed Effects Model Findings Estimation (Table 4)

• FLPR has a statistically significant negative
  effect on TFR in the fixed effect model 2 (column
  3), once we take both time-variant regional
  heterogeneity of consumption and time-invariant
  heterogeneity.
• Much of the distinction between the urban and
  rural areas in fertility patterns (Council for
  Gender Equality, Special Committee on the
  Declining Birthrate and Gender-Equal
  Participation, 2006b), can be explained by the
  differences in consumption patterns.

25
Concluding Remarks (1)

• Consider two factors
• Time necessary to consume and raise children
• Quality of consumption goods
• Wages and quality changes over time negative
  relationship between TFR and FLRP
• Quality differ by region positive relationship
  between TFR and FLRP
• General equilibrium model shows quality
  improvement leading to higher wages

26
Concluding Remarks (2)

• Japanese cross section from 8 different points in
  time (every five years from 1970 -- 2005)
• When variables that capture consumption quality
  (household leisure and entertainment
  expenditures, automobile ownership, and number of
  department stores) are added to cross regression
  of TFR on FLPR
• 1) consumption variables have significant
  coefficients
• 2) BUT, the FLPR coefficient is no longer
  significant at the 5 level.
• In fixed effects model, FLPR has a statistically
  significant negative effect on TFR, and also
  consumption variables are statistically
  significant.
• The results are consistent with our new model as
  well as traditional economic models of the
  relation between TFR and FLPR.

27
Appendix
Table gives the prefecture names which has max or
min value of Mean in 1975 and 2005.
28
Correlation Coefficients (1975)
Correlation Coefficients (2005)