Abstract
Japan: Selected Issues
Impact of Demographics on Growth and Inflation in Japan1
Is Japan’s aging and, more recently, declining population hampering growth and reflation? Exploiting demographic variation in prefectural data we find that: (i) aging has a significant negative impact on labor productivity, mainly through total factor productivity (TFP), and tends to lower overall inflation, (ii) a shrinking population is associated with a weak but significant and negative impact on TFP, but does not have a significant impact on overall inflation. In sum, aging seems to have a stronger adverse impact on productivity and inflation than population decline. While, the results also indicate that migration could help raise labor productivity, addressing low growth and inflation in Japan requires a coordinated package of measures that strengthens the monetary transmission and lifts potential growth.
A. Introduction
Mounting demographic headwinds have raised policymakers’ concerns about their macroeconomic impact. Over the past five years Japan’s population declined by close to 1 million, the equivalent of the population of San Francisco, while the share of the population aged 65 or above rose from 12 to 26 percent since 1990. The potential economic impact of an aging and shrinking population has increasingly caught the attention of policymakers in Japan. The government recently announced a new growth strategy which includes efforts to raise the fertility rate, boost female and older worker labor force participation and increase the number of high skilled foreign workers. Meanwhile, former Governor Shirakawa of the Bank of Japan voiced concerns that demographic headwinds have contributed to Japan’s deflationary mindset.
Indeed, when comparing Japan with other G7 countries, the country appears to be an outlier both in terms of demographics and macroeconomic performance (Figure 1). Specifically, the total labor force participation rate has fallen, more than offsetting the rise in the working age population as Japan’s working age population grew by close to 6 percent between 1995 and 2014, while the labor force shrank by 1 percent. This is because the share of the working age population (defined as 15 years and above) that is older than 65 has increased substantially.
Demographic and Macroeconomic Comparison of G7 Countries
Citation: IMF Staff Country Reports 2016, 268; 10.5089/9781475522525.002.A004
Demographic and Macroeconomic Comparison of G7 Countries
Citation: IMF Staff Country Reports 2016, 268; 10.5089/9781475522525.002.A004
Demographic and Macroeconomic Comparison of G7 Countries
Citation: IMF Staff Country Reports 2016, 268; 10.5089/9781475522525.002.A004
Few studies have looked at the impact of aging and population growth on labor productivity in Japan.2 Moreover, while the impact of demographics on inflation in Japan has been explored primarily through theoretical papers, there are few empirical studies that tries to establish a relationship. There are several channels through which population growth and aging could potentially impact growth and inflation:
Growth. First, a shrinking and aging population reduces the amount of labor available for production. While a declining population would decrease the size of the working-age population, aging would negatively impact the overall labor force participation rate. Second, demographics could impact labor productivity. A larger population could have a positive impact on productivity through positive network effects and greater economies of scale. On the other hand, higher population growth could put strains on the availability of capital and natural resources, hence reducing labor productivity. Moreover, aging could negatively impact productivity through the depreciation of human capital.
Inflation. There is little consensus about the relationship between demographics and inflation with some pointing to a positive relationship (as supply falls while aggregate demand remains supported through permanent income effects) and others finding a negative relationship. Many theoretical studies are skeptical of any relationship at all as monetary policy could offset the impact in either direction, just like other disturbances to the inflation path. However, in a situation where policy is constrained by the zero-lower bound and the natural rate of interest may also have fallen significantly, recent studies have started to revisit this monetarist dogma.3 Aging could also lead to a shift from market to regulated prices and hence affect overall measured inflation.
B. Data and Empirical Methodology
Demographics and labor productivity. To estimate the effects of aging and population size on labor productivity we first decompose prefectural labor productivity into its total factor productivity (TFP) and capital-labor ratio components.4 Overall labor productivity growth and its two-sub components are then regressed on the change in prefectural dependency ratios and population size.5 Moreover, to estimate the impact of inter-prefectural migration on labor productivity, a five year average of net migration inflows (as a share of prefectural population) is included in the regression.6
Demographics and inflation. Similarly, to assess the impact of aging and population growth on inflation, we estimate a panel regression using overall prefectural consumer price inflation (CPI) as well as its sub-components. Since monetary policy is targeting national CPI inflation and could in theory mitigate any demographic impact on inflation, prefectural inflation was measured relative to national inflation.7 The regressors (i.e., the old-age dependency ratio and population growth) were also measured relative to prefectural averages.
C. Results
Aging has a negative impact on labor productivity by lowering TFP growth, while population has a neutral impact through higher TFP growth but lower capital-labor ratios. Table 1 shows the results from the labor productivity regressions:
Aging. The change in the old age dependency ratio has a significant and negative impact on labor productivity, stemming from its negative effect on TFP growth. The point estimate suggests that a 10 percentage point increase in the old age dependency ratio lowers TFP by 11.6 percent.
Population size. More populous prefectures are associated with higher TFP growth but lower growth in the capital-labor ratio, resulting in an insignificant impact on overall labor productivity. A one percent increase in the relative prefectural population results in a 0.08 percent increase in TFP growth.
Migration. Net migration is positively associated with TFP growth but negatively related to capital-labor growth. A 1 percentage point increase in net migration (as a share of population) increases overall labor productivity by 0.7 percent, reflecting an increase in TFP by 2.5 percent and a decrease in the capital-labor ratio by 1.6 percent.
Panel Regression on Labor Productivity, Aging and Population Size
Panel Regression on Labor Productivity, Aging and Population Size
| Explanatory variables | (1) Δlog(Y/L) | (2) Δlog(A) | (3) Δlog(K/L) |
|---|---|---|---|
| Δold age dependency | −0.851** | −1.157* | −0.214 |
| (0.338) | (0.664) | (0.493) | |
| 5 year average net migration in percent of population | 0.724* | 2.491*** | −1.654*** |
| (0.361) | (0.691) | (0.567) | |
| Lagged log population | −0.00632 | 0.0883** | −0.0924** |
| (0.0199) | (0.0407) | (0.0366) | |
| Control Variable | |||
| Δservice in percent of GDP | −1.367*** | −2.274*** | 0.945*** |
| (0.133) | (0.227) | (0.115) | |
| Lagged on dependent variables | Yes | Yes | Yes |
| Time Dummy | Yes | Yes | Yes |
| Constant | 0.164 | −1.042* | 1.251** |
| (0.298) | (0.595) | (0.532) | |
| Observations | 1,380 | 1,380 | 1,380 |
| R-squared | 0.643 | 0.585 | 0.451 |
| Number of id | 46 | 46 | 46 |
Panel Regression on Labor Productivity, Aging and Population Size
| Explanatory variables | (1) Δlog(Y/L) | (2) Δlog(A) | (3) Δlog(K/L) |
|---|---|---|---|
| Δold age dependency | −0.851** | −1.157* | −0.214 |
| (0.338) | (0.664) | (0.493) | |
| 5 year average net migration in percent of population | 0.724* | 2.491*** | −1.654*** |
| (0.361) | (0.691) | (0.567) | |
| Lagged log population | −0.00632 | 0.0883** | −0.0924** |
| (0.0199) | (0.0407) | (0.0366) | |
| Control Variable | |||
| Δservice in percent of GDP | −1.367*** | −2.274*** | 0.945*** |
| (0.133) | (0.227) | (0.115) | |
| Lagged on dependent variables | Yes | Yes | Yes |
| Time Dummy | Yes | Yes | Yes |
| Constant | 0.164 | −1.042* | 1.251** |
| (0.298) | (0.595) | (0.532) | |
| Observations | 1,380 | 1,380 | 1,380 |
| R-squared | 0.643 | 0.585 | 0.451 |
| Number of id | 46 | 46 | 46 |
Faster aging prefectures tend to have lower overall inflation. Table 2 shows the results from the CPI panel regressions. Aging has a negative and statically significant impact on overall inflation, while population growth has a positive but insignificant impact. The regressions also indicate some interesting results on the impact of aging and population growth on the sub components of the CPI basket:
Aging and consumption preferences. The empirical evidence suggests that faster aging prefectures have higher inflation of medical services but lower inflation in transportation and communication costs. This is consistent with average expenditure shares of older and younger households in Japan, which shows that older households spend relatively more on medical care, but relatively less on transportation services.
Population growth and supply constraints. The results also indicate that prefectures with relatively higher population growth tend to have higher inflation in housing. This appears reasonable given that housing supply may be sticky, at least in the short run. However, the negative impact on energy and transportation inflation seems harder to understand.
Panel Regression on Inflation, Aging and Population Growth
Panel Regression on Inflation, Aging and Population Growth
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | |
|---|---|---|---|---|---|---|---|---|---|---|
| Relative CPI inflation | ||||||||||
| Independent variables | Overall | Medical Care | housing | fuel, energy, and | reading and recreation | trans&commu | cloths&foot wear | education | furniture | food |
| Δ relative old age dependency | −0.149* | 0.151* | −0.0109 | −0.0988 | −0.0294 | −0.574*** | −0.272 | 0.0394 | −0.233 | 0.00826 |
| (0.0824) | (0.0885) | (0.226) | (0.242) | (0.0961) | (0.146) | (0.230) | (0.171) | (0.194) | (0.102) | |
| Δ relative population | 0.101 | 0.0319 | 0.369* | −0.405*** | 0.195*** | −0.184* | 0.446*** | 0.0306 | 0.0515 | 0.0661 |
| (0.0726) | (0.0600) | (0.191) | (0.129) | (0.0608) | (0.106) | (0.124) | (0.130) | (0.158) | (0.0882) | |
| Control variable | ||||||||||
| Δlog in relative GDP | −0.000858 | 0.00596 | −0.0228 | 0.00573 | 0.0102 | 0.00211 | −0.00296 | 0.0181 | −0.0219 | 0.00138 |
| (0.00569) | (0.00757) | (0.0153) | (0.0131) | (0.00896) | (0.00837) | (0.0161) | (0.0184) | (0.0149) | (0.0110) | |
| Lagged on dependent variables | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Time Dummy | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Constant | 0.000657 | −0.000577 | 0.000254 | −0.00565** | 0.000200 | 5.99e-05 | 0.00267 | −2.94e-05 | −0.00144 | 0.00200 |
| (0.000698) | (0.000876) | (0.00133) | (0.00272) | (0.000924) | (0.000903) | (0.00196) | (0.00107) | (0.00270) | (0.00126) | |
| Observations | 1,472 | 1,472 | 1,472 | 1,472 | 1,469 | 1,472 | 1,472 | 1,472 | 1,472 | 1,472 |
| R-squared | 0.076 | 0.065 | 0.136 | 0.040 | 0.030 | 0.091 | 0.025 | 0.148 | 0.053 | 0.053 |
| Number of id 1/ | 46 | 46 | 46 | 46 | 46 | 46 | 46 | 46 | 46 | 46 |
Panel Regression on Inflation, Aging and Population Growth
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | |
|---|---|---|---|---|---|---|---|---|---|---|
| Relative CPI inflation | ||||||||||
| Independent variables | Overall | Medical Care | housing | fuel, energy, and | reading and recreation | trans&commu | cloths&foot wear | education | furniture | food |
| Δ relative old age dependency | −0.149* | 0.151* | −0.0109 | −0.0988 | −0.0294 | −0.574*** | −0.272 | 0.0394 | −0.233 | 0.00826 |
| (0.0824) | (0.0885) | (0.226) | (0.242) | (0.0961) | (0.146) | (0.230) | (0.171) | (0.194) | (0.102) | |
| Δ relative population | 0.101 | 0.0319 | 0.369* | −0.405*** | 0.195*** | −0.184* | 0.446*** | 0.0306 | 0.0515 | 0.0661 |
| (0.0726) | (0.0600) | (0.191) | (0.129) | (0.0608) | (0.106) | (0.124) | (0.130) | (0.158) | (0.0882) | |
| Control variable | ||||||||||
| Δlog in relative GDP | −0.000858 | 0.00596 | −0.0228 | 0.00573 | 0.0102 | 0.00211 | −0.00296 | 0.0181 | −0.0219 | 0.00138 |
| (0.00569) | (0.00757) | (0.0153) | (0.0131) | (0.00896) | (0.00837) | (0.0161) | (0.0184) | (0.0149) | (0.0110) | |
| Lagged on dependent variables | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Time Dummy | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Constant | 0.000657 | −0.000577 | 0.000254 | −0.00565** | 0.000200 | 5.99e-05 | 0.00267 | −2.94e-05 | −0.00144 | 0.00200 |
| (0.000698) | (0.000876) | (0.00133) | (0.00272) | (0.000924) | (0.000903) | (0.00196) | (0.00107) | (0.00270) | (0.00126) | |
| Observations | 1,472 | 1,472 | 1,472 | 1,472 | 1,469 | 1,472 | 1,472 | 1,472 | 1,472 | 1,472 |
| R-squared | 0.076 | 0.065 | 0.136 | 0.040 | 0.030 | 0.091 | 0.025 | 0.148 | 0.053 | 0.053 |
| Number of id 1/ | 46 | 46 | 46 | 46 | 46 | 46 | 46 | 46 | 46 | 46 |
D. Conclusions and Policy Implications
In sum, the results suggest that aging seems to have a stronger adverse impact on productivity and inflation than population decline. While, the results also indicate that internal migration could help raise labor productivity, addressing low growth and inflation in Japan requires a coordinated package of measures that strengthens the monetary transmission and lifts potential growth.
Impact of Aging and Population Growth on Japan’s Labor Force
Citation: IMF Staff Country Reports 2016, 268; 10.5089/9781475522525.002.A004
Impact of Aging and Population Growth on Japan’s Labor Force
Citation: IMF Staff Country Reports 2016, 268; 10.5089/9781475522525.002.A004
Impact of Aging and Population Growth on Japan’s Labor Force
Citation: IMF Staff Country Reports 2016, 268; 10.5089/9781475522525.002.A004
References
Anderson, D, D. Botman, B. Hunt, 2014, “Is Japan’s Population Aging Deflationary,” IMF Working Paper WP14/139.
Bloom, D and J. Finlay, 2008, “Demographic Change and Economic Growth in Asia,” PGDA Working Paper No.41.
Feyrer, J. 2007, “Demographics and Productivity,” The Review of Economics and Statistics 89(1):100–109.
Kyoji, F, 2010, “Service Sector Productivity in Japan: The key to future economic growth,” RIETI Policy Discussion Paper Series 10-P-007.
Kyoji, F and M. Tatsuji, 2015, “Aging, Interregional Income Inequality, and Industrial Structure: An empricial analysis based on the R-JIP Database and the R-LTES Database,” RIETI Discussion Paper Series 15-E-022.
Kyoji, F and M. Tatsuji, 2015, “Regional Factor Inputs and Convergence in Japan: A macro-level analysis, 1955-2008,” RIETI Discussion Paper Series 15-E-123.
Prepared by Dennis Botman, Yihan Liu, and Niklas Westelius (all APD).
One exception is Kyoji and Tatsuji (2015) who find that aged prefectures tend to have lower labor productivity but argue that aging does not systematically reduce TFP levels.
See for example Juselius and Takats, 2015. “Can demography affect inflation and monetary policy?” BIS Working Papers 485, Bank for International Settlements.
Data on prefectural real value added and labor and capital stocks are from the Research Institute of Economy, Trade and Industry (REITI). The time span of the date is from 1970 to 2009.
The panel regressions also include prefectural fixed effects, time dummies, and the size of the services sector. The latter was included to account for the fact that labor productivity tends to be lower in this sector. Population size is measured relative to the average prefectural population.
To mitigate the potential endogeneity problem if migration is driven by prefectural wage differentials, lagged migration and population growth is used.
The regression also includes prefectural fixed effects and prefectural real growth relative to national GDP growth.
