Priorities for the Japanese Economy in 2016 (January 2016)
Is the Decentralization of the Population a Viable Measure to Raise Fertility?
Vice Chairman & Vice President
Japan's economic policy for 2016 will be guided by the "new three arrows" or three key numerical targets set out in Abenomics 2.0. Along with achieving 600 trillion yen in gross domestic product (GDP), the government aims to raise the fertility rate to 1.8. Recently, in the face of a falling population, we often hear an argument that calls for decentralization as a means to put the brakes on the decline, pointing to the lower fertility in urban areas and higher fertility in rural areas. In what follows, I would like to take a new look at whether the decentralization of the population can be a viable fertility-boosting measure.
Will population concentration in metropolitan areas lead to lower fertility?
The argument for raising fertility by means of decentralization has its main basis in the negative correlation observed between population density and fertility. Indeed, a cross-sectional ordinary least squares (OLS) regression analysis, which is designed to explain the total fertility rate (TFR) in each prefecture as a function of population density, found a negative correlation between the TFR and population density--i.e., the higher the population density, the lower the fertility--when performed on prefectural data collected every five years for the period 1960-2010 (see Table 1A-(1)). However, the negative correlation may be a spurious one and actually a result of various factors unique to each prefecture such as values and cultural traditions. Thus, I estimated a fixed-effects (FE) model that controls for prefecture-specific factors, which also found a negative correlation between population density and fertility (see Table 1A-(2)).
However, when the same estimations were performed on the data for the last 20-year period from 1990, when Japan's low fertility rate grew into a serious problem, the sign of the coefficient of population density obtained using the FE model controlled for prefecture-specific factors turned positive (see Table 1B-(2)), although cross-sectional regression results show a significant correlation between the two variables. In other words, as far as recent years are concerned, the higher (lower) the population density of a prefecture, the higher (lower) the TFR.
|Estimation period||(1) OLS||(2) FE|
|Note: Estimates obtained using the OLS and FE regression models based on prefectural panel data collected every five years. TFR is the dependent variable, while the logarithm of population density and year dummies are independent variables. Figures in the parentheses are the standard errors of estimates. The three asterisks (***) and two asterisks (**) indicate that the estimate is statistically significant at the 1% and 5% levels respectively.|
I do not intend to interpret those simple estimation results as suggesting that greater population concentration in major cities would lead to higher fertility. This may be due to the influence of time-series changes in various factors such as industrial structure, the ratio of the self-employed to the total workforce, and the availability of family support (e.g., three-generation households, grandparents or relatives living nearby). Also, the possibility of selection bias arising from the geographical movement of people cannot be excluded, as in the case of any analysis based on aggregate data. However, the above results point to the presence of factors other than population concentration behind the regional differences in fertility. Or, at least, they suggest that it is dangerous to make important policy decisions based on a simple correlation identified in a cross-sectional regression analysis.
Quantitative relationship between population distribution and Japan's TFR
Many theoretical and empirical studies have been conducted on the determinants for fertility. Based on representative survey research papers, I have attempted to identify whether and how urbanization and other geographical factors impact fertility. However, some papers conclude that population density and housing prices have nothing to do with time-series changes in fertility or cross-country differences (Feyrer et al., 2008), whereas others found that changes in housing costs due to urbanization have had an impact on fertility (Guinnane, 2011). As such, it is hard to say that there is any robust empirical stylized fact.
However, as far as Japan is concerned, a close examination of data by prefecture and year finds that regardless of prefectures, a long-term downward trend in the time series of TFR is the dominant factor for low fertility in recent years and the impact of concentration in urban areas has been extremely small, if any, in quantitative terms. For instance, even if Japan's population composition by prefecture in 2010 were the same as that in 1970 when TFR was 2.13, TFR in 2010 would be 1.40--hardly different from the actual rate of 1.39--based on a simple calculation. The drop in Japan's TFR (by 0.74 percentage points between 1970 and 2010) is mostly attributable to the nationwide downward trend in fertility (i.e., "within effect"), with variation in the population composition by prefecture (i.e., "reallocation effect") accounting for a mere 0.01 percentage point decrease (or contributing to approximately 2% of the decline in TFR) (see Table 2).
|Decline in TFR||Reallocation effect||Within effect|
|Note: The reallocation effect is the difference between pro forma TFR in 2010, calculated assuming that each prefecture's share in the total population remains unchanged from the first year of each period, and actual TFR. The within effect is the degree of contribution by declining fertility in each prefecture.|
Conversely, we can say that even if we were able to bring the distribution of population in Japan back to its state in or around 1970 by dispersing the population from major cities to rural regions, the resulting impact would be a mere 0.01 percentage point increase in TFR. If the government aims to develop effective policy solutions for raising Japan's overall fertility significantly to around 1.8, it needs to explore and identify time-series factors that are common to all prefectures and contribute to more than 90% of the decline in fertility, and then adopt measures targeted at such factors.
A policy mix for achieving two goals: Higher productivity and higher fertility
The ongoing shift in the industrial structure--i.e., an increase in the share of service industries that are urban-oriented in nature--is closely related with the geographical distribution of the population. Population concentration specifically in Tokyo, as compared to the Tokyo metropolitan area, became conspicuous in the mid-1990s, and this coincided with the accelerated growth of service industries. What we see here is a reciprocal causal relationship: a shift to a services-oriented economy brings greater concentration in urban areas, while urbanization generates a greater variety of services.
Furthermore, in service industries where production and consumption occur simultaneously, productivity is closely linked with population density. Indeed, the tendency of firms located in urban areas to show higher productivity is much more prominent in the service sector than in the manufacturing sector (Morikawa 2014). Meanwhile, in the case of knowledge- and information-intensive service industries, employment density in the location of businesses is a major determinant for productivity (Morikawa 2015). Therefore, agglomeration of economic activities in densely populated major cities is desirable in terms of improving productivity in service industries, which holds the key to increasing Japan's growth potential at a time when the total population is shrinking.
As discussed above, we cannot be conclusive about the presence of any negative correlation between population concentration and fertility. However, such a relationship, if present, poses a trade-off between a growth policy aiming at improving productivity in the service industries and a policy seeking to raise fertility. In this case, assigning different policy tools to different policy goals is the way to respond to the situation, according to the guiding principles of the optimal policy mix. Specifically, the government should promote population concentration as a means to improve economic efficiency, and apply public policy measures that will likely have a direct impact on fertility (i.e., increasing childcare facilities, enhancing public education, improving commuting infrastructure, etc.) with a particular focus on densely populated areas, instead of promoting decentralization that would only have an indirect impact. The same argument can apply in pursuing the two goals of promoting women's work-life balance (Morikawa 2016).
The government's recently announced set of policy measures, dubbed the "Urgent Policies to Realize a Society in Which All Citizens are Dynamically Engaged," call for, inter alia, securing employment for youths, enhancing childcare services, providing an environment to facilitate family support for childcare by three generations living under one roof or nearby, and making more scholarships available. All in all, those measures are pointed to the right direction.
- Feyrer, James, Bruce Sacerdote, and Ariel Dora Stern (2008), "Will the Stork Return to Europe and Japan? Understanding Fertility within Developed Nations," Journal of Economic Perspectives, Vol. 22, No. 3, pp. 3-22.
- Guinnane, Timothy W. (2011), "The Historical Fertility Transition: A Guide for Economists," Journal of Economic Literature, Vol. 49, No. 3, pp. 589-614.
- Morikawa, Masayuki (2014), Productivity in Service Industries: Empirical analyses using microdata, Nippon Hyoronsha Co., Ltd.
- Morikawa, Masayuki (2015), "Location and Productivity of Knowledge- and Information-intensive Services," RIETI Discussion Paper 15-J-050.
- Morikawa, Masayuki (2016), "Productivity in the Service Sector and Labor Market," Japan Labor Review, No. 666, pp. 16-26.
January 13, 2016
Article(s) by this author
Outside directors, firm behaviour, and performance: Evidence from a quasi-natural experiment in Japan
March 30, 2020［VoxEU Column］
February 28, 2020［RIETI Report］
February 13, 2020［VoxEU Column］
December 26, 2019［RIETI Report］
December 26, 2019［Priorities for the Japanese Economy in 2020 (January 2020)］