Page 134 - IJPS-11-5

International Journal of
Population Studies Regional disparities and fertility rates

2006). Population size has a negative effect in all models, 3.3. Time series model
with statistical significance at p<0.01 in all but the SDM To investigate the impact of regional disparities resulting
Panel, where it is significant at p<0.05. This implies that in from economic growth on fertility rates, this study
regions with larger populations, individuals are more likely collected data spanning 20 years, from 2000 to 2019, for
to pursue personal career goals, cultural activities, and time series analysis. As mentioned earlier, the time series
other values rather than focusing on family (Kulu et al., model employed in this study includes the lagged total
2009), as these regions tend to have more cultural facilities fertility rate as a controlled variable. Therefore, regression
and job opportunities. The proportion of the population of analysis was conducted using data starting from 2001.
reproductive age, which represents the potential fertility of
a region, is found to have a positive impact on the regional Table 6 presents the results of the time series analysis.
total fertility rate (p<0.01), and this result is consistent The DW statistic, which indicates autocorrelation in error
across all models. The unemployment rate, reflecting labor terms, shows values close to 1, suggesting the presence
market instability, has a negative impact on the regional of positive autocorrelation. Regarding model goodness-
total fertility rate in the pooled model that disregards the of-fit, the models that control for the Theil coefficient
spatial and temporal characteristics of the data. However, demonstrate higher explanatory power than those that
it shows a positive impact in the spatial panel models. control for the Gini coefficient. Among them, Model
This aligns with Schultz’s (1973) rational choice theory, 5, which includes lagged variables and controls for the
which suggests that labor market participation can be Theil coefficient, exhibits the highest explanatory power
considered an opportunity cost of childbirth. In fact, some at 69.76%. The Durbin h statistic, which evaluates model
studies conducted in advanced countries have shown that fitness when controlling for a lagged dependent variable
an increase in the unemployment rate can have a positive in the time series model, varies depending on whether the
impact on the total fertility rate (Schmitt, 2008). GINI and THEIL coefficient is used. In Model 4, which
controls for regional inequality using the Gini coefficient,
Analyzing the results of the SDM panel model provides the Durbin h statistic is statistically significant at the p<0.01
insights into how independent variables in neighboring level, indicating the presence of serial autocorrelation. In
regions influence the total fertility rate of a specific region. contrast, Model 5, which controls for regional disparities
The spatial interaction coefficient for GRDP (W_GRDP) using the Theil coefficient, shows no statistically
shows a significant negative impact at p<0.01, indicating significant Durbin h value, suggesting the absence of serial
that as the economic level of neighboring regions autocorrelation. Therefore, the final interpretation of the
increases, the total fertility rate in the specific region regression results is based on Model 5, which includes
decreases. This suggests that when neighboring regions lagged variables and controls for the Theil coefficient.
have higher economic levels, various factors – such as
the outmigration of reproductive-age populations – Model 1, which only includes the economic growth
come into play, contributing to lower fertility rates in rate as a predictor, shows a statistically significant positive
adjacent areas. Although the spatial lag coefficient for effect on the total fertility rate at the p<0.05 level. However,
population density (W_Density) has a positive value, it in Models 2 through 5, which additionally account for
is not statistically significant. The spatial lag coefficient regional disparities, the impact of economic growth on
for population size (W_Population) is statistically fertility is no longer statistically significant. In Models 4 and
significant at p<0.01 and shows a positive impact. This 5, which control for the lagged total fertility rate, the lagged
finding implies that as the population size of neighboring term exerts a statistically significant positive influence
regions increases, the likelihood of marriage and on the present year’s total fertility rate at the p<0.10 and
childbirth among reproductive-age groups in the specific p<0.05 levels, respectively. Considering that the lagged
region increases. Conversely, a high proportion of the total fertility rate reflects unobserved socioeconomic and
reproductive-age population in neighboring regions may cultural factors influencing fertility, the estimated effects
lead to the outmigration of youth from the focal region, of economic growth and regional disparities on fertility
which can contribute to a decline its fertility rate. This – when this variable is controlled for – carry significant
inference is supported by the significant negative spatial explanatory power.
lag coefficient for the proportion of the reproductive-age Models 2 and 4, which use the Gini coefficient control
population (W_Youth; p<0.01). The spatial lag coefficient to capture regional disparities, reveal a statistically
for the unemployment rate (W_Unemployment), significant negative impact on the total fertility rate at the
representing the labor market conditions in neighboring p<0.01 and p<0.05 levels, respectively. Similarly, Models
regions, shows a negative effect but does not reach 3 and 5, which control for regional disparities using the
statistical significance. Theil coefficient, also exhibit a negative effect, with all

Volume 11 Issue 5 (2025) 128 https://doi.org/10.36922/ijps.8157