Population and Immigration in Japan: Projections with Time Series and Demographic Causal Analysis Until 2044

Michelli Okada; Nícolas Volgarine Scaraboto

DOI: 10.22167/2675-6528-202603295

Article derived from a Course Conclusion Work (TCC), with content based on the student’s original work and adapted to the editorial format of the E&S Magazine with the support of the ResumeAI tool, an artificial intelligence solution developed by Instituto Pecege for textual synthesis and organization.

Summary

Population aging and low fertility rates present complex demographic challenges in Japan, accompanied by the expansion of foreign immigration as a response to labor supply constraints. The study aimed to analyze Japanese demographic trends and project their evolution until 2044, with an emphasis on the relationships between population decline, fertility, and immigration. The ARIMA and Prophet time series models were applied to historical data from the World Bank, the National Institute of Population and Social Security Research of Japan, and the Immigration Services Agency of Japan, covering the period from 2003 to 2023 for the population series and from 2014 to 2024 for the immigration series. The analysis included descriptive statistics, Pearson correlation, augmented Dickey-Fuller stationarity test, and Granger causality tests, in addition to predictive evaluation using MAE and MAPE metrics. The results showed the continuation of population reduction and the growth of the immigrant contingent in projections up to 2044. Prophet presented lower MAE and MAPE values on the test sets, although the short duration of the immigration series limited generalization capacity. Temporal causality tests did not identify statistically significant evidence of precedence between population and immigration. It was concluded that immigration can contribute to mitigating labor supply constraints, but it does not, in isolation, constitute a sufficient mechanism to reverse Japan’s population decline.

Keywords: Population decline; Demography; Fertility; Migration flows; Time series.

1. Introduction

Japan, a country with a rich cultural heritage, is currently facing a complex demographic challenge characterized by an aging population and a low birth rate (Nakatani, 2023). This dynamic imposes significant pressures on the country’s social and economic structure.

In 2024, Japan’s total population was 123.84 million inhabitants, representing a decrease of 0.51 million compared to the previous year. The fertility rate reached a record low of 1.30 children per woman (IPSS, 2024). Furthermore, approximately 29.4% of the population consists of individuals over 65 years old, classifying the country as hyper-aged (Statistical Handbook of Japan, 2024). This scenario results in a reduction of the active population, directly impacting economic growth (United Nations, 2024).

To mitigate the population decline and address the labor shortage, the Japanese government has been implementing policies to encourage foreign immigration. Programs such as the Specified Skilled Worker (SSW) have expanded the possibilities for stay and visa renewal for skilled workers, allowing, in some categories, the transition to long-term residency (Ministry of Foreign Affairs of Japan, 2026). These initiatives are part of a governmental strategy to confront the worker shortage in strategic economic sectors. The rate of immigrants has grown approximately 40% in the last decade, reaching 4.38 million in 2024 (OECD, 2024). Although immigrants represent about 2.5% of the resident population, their numbers have been increasing significantly, highlighting the relevance of immigration for the country’s demographic challenges (Ministry of Justice of Japan, 2024).

However, it is not yet clear whether the expansion of immigration constitutes a structural transformation capable of reversing the population decline trajectory. Classic studies suggest that, although immigration can temporarily mitigate the reduction in the economically active population, its effects on aggregate population growth are limited when fertility levels remain persistently below replacement (Coleman, 2006; Lutz, Sanderson & Scherbov, 2008). Additionally, empirical evidence indicates that, over time, immigrants tend to converge to the reproductive patterns of the destination society, diminishing the potential impact on birth rates (Massey et al., 1993; Ogawa et al., 2016).

Fertility dynamics, in turn, cannot be understood exclusively by demographic or cultural factors. The decision to have children involves economic constraints, such as the direct costs of upbringing and education, the availability of care services, and the opportunity cost associated with labor market participation. From the economic perspective of the family, reproductive decisions are influenced by resource and time constraints (Becker, 1960; Becker; Lewis, 1973). In the Japanese context, the costs associated with children’s education, housing, and the difficulty of reconciling work and family are obstacles to family formation (OECD, 2024a). Recent evidence for Japan reinforces this perspective, showing that the expansion of public childcare capacity is associated with an increase in the birth rate (Fukai and Toriyabe, 2025). Dynamic economic models also demonstrate that childcare policies can simultaneously affect fertility decisions and maternal labor supply (Sakamoto, 2025).

In parallel, the contribution of immigration to the Japanese labor market is growing, but its effectiveness depends on integration conditions. Population aging and the reduction in the labor supply have led to changes in foreign labor recruitment policies (OECD, 2024b). However, linguistic barriers and differences in labor market practices can limit immigrants’ integration and occupational mobility, with Japanese proficiency being an important requirement in a large part of the labor market (OECD, 2024b).

In this sense, the relationship between the growth of immigration and the Japanese population decline constitutes an open empirical question, especially when analyzed from the perspective of long-term quantitative and predictive models. The absence of studies that articulate recent historical series with future projections limits the understanding of whether immigration constitutes a structural mechanism for demographic reversal or merely a conjunctural response to labor shortages.

In light of this context, the present study seeks to contribute to the academic debate and to the formulation of public policies aimed at Japan’s demographic sustainability. To this end, the objective is to analyze, through time series models, the dynamics between immigration growth and population decline in Japan, projecting demographic scenarios up to the year 2044. The ARIMA (AutoRegressive Integrated Moving Average) and Prophet models will be employed, widely used in the literature for forecasting economic and demographic time series (Hyndman & Athanasopoulos, 2021; Taylor & Letham, 2018), integrating historical analysis, predictive modeling, and examination of correlations between demographic variables and causality.

2. Material and Methods

The present study adopted a quantitative and predictive approach to analyze the dynamics between immigration growth and population decline in Japan, projecting demographic scenarios until 2044. The methodology was structured in three stages: exploratory analysis and preparation of historical series; predictive modeling with ARIMA and Prophet; and statistical analysis of the relationships between demographic variables.

Historical data were obtained from official sources. Information on the number of resident immigrants in Japan (2014-2024) was extracted from the Immigration Services Agency of Japan (2024). Data on population growth and total population (2003-2023) were collected from the World Bank (2023).

The migratory database (2014-2024) consisted of 201 observations and 36 variables, including continents, countries, total migration value, sex, and age groups (0 to 80 years). The population database (1960-2023) contained 267 observations and 68 variables, such as country, indicator codes, and years.

In the exploratory analysis stage, the migration spreadsheets, originally in Japanese, were translated and concatenated. Population data were filtered for 2003-2023. In both sets, the existence of missing values and inconsistencies was verified to ensure the quality of the database (Hyndman and Athanasopoulos, 2018).

For graphical visualization, the Matplotlib and Seaborn libraries of the Python language were used. Line graphs were generated for population and immigration evolution, and heat maps for the age distribution of immigrants. Descriptive statistics (mean, median, dispersion, and position) were calculated to synthesize the trends of the series.

The predictive modeling employed the time series models ARIMA (Box and Jenkins, 1976; Hyndman and Athanasopoulos, 2021) and Prophet (Taylor and Letham, 2018). ARIMA analyzed autocorrelations and historical patterns, while Prophet captured non-linear trends. The joint choice aimed for complementarity and greater reliability of projections.

The databases were structured with a temporal variable (ds) and a target variable (y). For population analysis, the data were divided into training (2003–2019) and testing (2020–2023). For immigration, the division was into training (2014-2022) and testing (2023–2024).

Immigratory data, obtained in heterogeneous formats, required manual translation, standardization, and consolidation. This restricted the immigration series to 2014-2024, limiting the test period to two years (2023–2024). The models were calibrated and validated using metrics such as MAPE and MAE (Hyndman and Athanasopoulos, 2018).

Based on the adjusted models, future projections were made for the next 20 years. Subsequently, a comparative graphical visualization of the proportion between the Japanese population and immigrants was performed. To complement, the relationships between fertility rate, elderly population, and total immigration were analyzed, using data from the World Bank (2014-2023).

The assessment of linear relationships between variables was conducted using Pearson’s correlation coefficient (r). To investigate temporal precedence, the Granger causality test (Granger, 1969) was applied, considering an annual lag. The directions population to immigration and immigration to population were evaluated, and the tests were repeated using the first differences of the series.

The use of only one annual lag in the Granger test was determined by the reduced number of observations (2014-2023, ten observations). The test results were interpreted in an exploratory manner, without inferring definitive causality. All methodological steps were implemented using the Python programming language, with libraries such as Pandas, NumPy, Statsmodels, Seaborn, Matplotlib, and Scikit-learn. The development environment was Google Colaboratory (Carneiro et al., 2018).

3. Results and Discussion

The analysis of Japanese demographic trends and the projection of their evolution until 2044 revealed a complex scenario, characterized by a persistent population decline and a continuous growth in the immigrant contingent. The results obtained through the ARIMA and Prophet time series models, along with correlation and causality analysis, allowed for an in-depth understanding of the underlying dynamics of these movements. Although immigration appears to be a mitigating factor for labor shortages, the evidence suggests that, on its own, it is not sufficient to reverse the trajectory of aging and population reduction in Japan.

Descriptive statistics of the data

The exploratory analysis of immigration data, covering the period from 2014 to 2024, highlighted Asia as the main origin of immigrants in Japan, with volumes close to 4 million people in recent years. This predominance reflects geographical proximity, historical ties, and the specific demands of the Japanese labor market (Shiraishi, 2019). South America appears in second place, with approximately 300,000 immigrants, while Africa and Oceania registered marginal volumes, less than 50,000 people.

A sharp drop in migratory flow was observed between 2020 and 2021, attributed to the effects of the COVID-19 pandemic, which imposed travel restrictions, border closures, and resulted in layoffs or the return of immigrants to their home countries. However, a significant recovery was verified in 2022 and 2023, possibly driven by the easing of post-pandemic migration policies. These patterns indicate the sensitivity of migratory flows to global events and policy responses.

The age distribution of immigrants revealed that the foreign population in Japan is predominantly young and economically active. The age groups of 21 to 35 years concentrated the largest volumes of immigrants, with continuous growth throughout the decade analyzed. This profile suggests a strong association of immigration with temporary work programs, opportunities in sectors with a shortage of local workers, and the permanence of university students and interns after graduation (Liu-Farrer, 2020).

The presence of immigrant children (0 to 15 years old) remained stable, with slight growth, indicating the occurrence of family immigration, albeit in smaller proportions. This stability contrasts with the rapid growth of the more active age groups, reinforcing the perception that immigration is predominantly driven by the demand for labor in specific sectors of the Japanese economy.

Regarding Japan’s overall demographic evolution, historical data from 2003 to 2023 evidenced a pattern of continuous population decline. The maximum recorded population was 128.1 million in 2010, followed by a steady decrease that reached 124.5 million in 2023. This reduction of approximately 3.6 million inhabitants in 15 years underscores the severity of the country’s demographic crisis, driven by one of the lowest birth rates in the world, with an average of 1.3 children per woman (IPSS, 2024).

Population aging is also a critical factor, with about 29.4% of the population over 65 years old, classifying Japan as a hyper-aged country (Statistical Handbook of Japan, 2024). Events such as the 2011 earthquake, tsunami, and nuclear accident (Zheng et al., 2022), and the COVID-19 pandemic in 2020, may have accelerated existing population decline trends, intensifying pressure on the workforce and the social security system.

Descriptive statistics of the Japanese population, from 2003 to 2023, indicated a median of 127.14 million inhabitants. However, the most recent years (2021-2023) registered values below the first quartile, positioning the current population among the lowest 25% of historical records. This total variation of 3.55 million inhabitants in 20 years highlights the magnitude of the demographic decline and the urgency of effective public policies.

In contrast, total immigration showed accelerated growth in the same period. The annual average of immigrants was 2.83 million, with the median slightly higher at 2.96 million, indicating that higher volumes were concentrated in recent years. This growth was particularly intense from 2020 onwards, with the number of immigrants reaching a record 4.39 million in 2024, which demonstrates a significant shift in the pace of foreign entry into the country.

The age group analysis of immigrants in 2023 revealed that the 0 to 20 years range represented 12.7% of the total, with an annual average of 84,349 immigrants. The 21 to 40 years range was the most dominant, corresponding to 52.3% of the total, with an average of 263,494 immigrants and a growth of 78.4% since 2014. The 41 to 60 years group, in turn, registered an average of 189,459 immigrants, with a less concentrated distribution, reinforcing the young profile of the immigrant population in contrast to the aging native population.

Predictive modeling

Population data

The predictive modeling of the Japanese population with the Prophet model, using data from 2003 to 2019 for training and 2020 to 2023 for testing, projected an estimate of 115.28 million inhabitants in 2044. The 95% confidence interval for this projection ranged between 109.85 and 120.30 million. The model’s accuracy was evaluated with a Mean Absolute Error (MAE) of 0.185 million and a Mean Absolute Percentage Error (MAPE) of 0.15%, indicating a good fit to historical data and reinforcing the trend of population decline in the coming decades.

The ARIMA model, in turn, was applied after identifying the need for two differenciations (d=2) to stabilize the population series, due to the absence of stationarity. Trained with data from 2003 to 2019 and tested from 2020 to 2023, the ARIMA(1,2,1) model was selected based on the Akaike Information Criterion (AIC). This model projected a central value of 111.82 million inhabitants in 2044, with a 95% confidence interval between 98.95 and 124.69 million. The accuracy results were MAE of 0.234 million and MAPE of 0.18%, slightly lower than those of Prophet.

Immigration data

For the immigration series, the Prophet model, trained with data from 2014 to 2022 and tested from 2023 to 2024, projected continuous growth in the number of immigrants in Japan, reaching approximately 15.18 million in 2044. The estimated average annual growth rate was 7.68%. The accuracy of Prophet for immigration was an MAE of 0.115 million and MAPE of 2.48% in the test period. This projection reflects the expectation that the immigrant population will continue to grow, driven by more flexible migration policies and the aging of the Japanese population.

However, the Prophet’s projection for immigration should be interpreted with caution, as the Japanese political scenario underwent significant changes from the second half of 2025, with the adoption of a more restrictive orientation regarding immigration (The Japan Times, 2025; The Diplomat, 2025). In February 2026, the Japan Immigration Services Agency established stricter criteria for granting permanent residency, including additional requirements for minimum visa duration and tax regularity (The Japan Times, 2026). Such structural breaks in the trajectory of migratory easing introduce an additional degree of uncertainty into long-term projections.

The ARIMA(1,2,1) model was also applied to the immigration series, with training from 2014 to 2022 and validation from 2023 to 2024. The central projection for 2044 was 9.71 million immigrants, with a 95% confidence interval ranging from -0.26 to 19.67 million. The ARIMA accuracy for immigration was MAE of 0.166 million and MAPE of 3.61%. The differencing order (d=2) was adopted to make the series stationary, but the reduced number of available observations (only 10 years) may impact the stability of projections and widen the confidence intervals, demanding caution in interpreting long-term results.

General comparison between models

The comparison between the Prophet and ARIMA models revealed that Prophet showed better short-term accuracy for both series, with lower MAE and MAPE than ARIMA. For population data, Prophet achieved an MAE of 0.185 million and MAPE of 0.15%, while ARIMA(1,2,1) recorded an MAE of 0.234 million and MAPE of 0.18%. For immigration data, Prophet presented an MAE of 0.115 million and MAPE of 2.48%, outperforming ARIMA(1,2,1) with an MAE of 0.166 million and MAPE of 3.61%.

Despite Prophet’s superiority in short-term accuracy, its reliability in long-term immigration projections is questionable due to confidence intervals with negative lower bounds, which is statistically impossible for a count of people. Prophet’s confidence interval for immigration in 2044 showed a range exceeding 50 million, while ARIMA, although also with a negative lower bound (-0.26 million), exhibited a more controlled range. For population data, both models maintained plausible confidence intervals, with Prophet showing a narrower range (109.85-120.30 million) compared to ARIMA (98.95-124.69 million). The joint use of both models, as adopted in this study, is recommended as it allows for analytical complementarity, mitigating the limitations of each.

The projections for 2044, for both population and immigration, assume the continuity of observed historical patterns. However, it is essential to recognize that changes in migratory policies, economic conditions, reproductive behavior, technological advancements, international crises, or alterations in the integration conditions of foreigners can produce future trajectories significantly different from those projected. The short duration of the immigration series (2014-2024) and the limited two-year test period (2023-2024) for evaluating immigration models reduce the generalizability and robustness of conclusions about the superiority of one model over the other in the long term.

Correlation between Demographic Variables

The analysis of the correlation matrix between fertility rate, elderly population, and total immigration, in the period from 2014 to 2023, complemented the projections, revealing important relationships. A strong negative correlation (-0.83) was found between the fertility rate and the elderly population, indicating that the increase in the elderly contingent tends to be accompanied by the maintenance or intensification of low birth rates. This pattern corroborates the idea of a “second demographic transition”, where fertility remains below replacement level, increasing pressure on the age balance of the population (Lutz et al., 2008).

The moderately negative correlation (-0.60) between fertility and immigration suggests that the growth of the immigrant contingent occurred in parallel with the maintenance of low fertility levels. However, this association does not allow us to infer that the increase in immigration causes the reduction in fertility. Such a relationship may, in fact, reflect simultaneous demographic and economic trends or the influence of factors not directly incorporated into the model, and should be interpreted as a statistical association and not as evidence of a causal effect.

The persistence of low fertility in Japan can be understood in light of long-term economic and institutional factors. The decision to have children involves not only individual preferences but also direct costs of raising and educating them, the availability of care services, and the opportunity costs related to labor market participation. The OECD (2024a) highlights that the costs associated with children’s education, housing, and the difficulty of balancing work and family are significant obstacles to family formation in the Japanese context (Becker, 1960; Becker & Lewis, 1973).

Recent empirical evidence for Japan reinforces this perspective, indicating that the expansion of public childcare capacity is associated with increases in the birth rate, especially for first and second births (Fukai & Toriyabe, 2025). Furthermore, dynamic economic models demonstrate that childcare policies can simultaneously affect fertility decisions and maternal labor supply (Sakamoto, 2025). This suggests that the availability of care infrastructure can reduce the constraints faced by families in balancing work and parenthood, and that pronatalist policies should go beyond financial benefits, focusing on reducing time costs and reconciliation difficulties.

On the other hand, the strong positive correlation (0.73) between the elderly population and immigration aligns with projections of a significant increase in immigration, mainly in response to labor market demands in critical sectors such as healthcare and construction (Harper, 2016). This pattern reinforces the interpretation that immigration, in the Japanese context, acts as a palliative measure to compensate for the labor shortage resulting from population aging, but not as a direct instrument to encourage birth rates.

The contribution of immigration to the Japanese labor market, however, must be analyzed in conjunction with integration conditions. The OECD (2024b) highlights that, although Japan has expanded its mechanisms for recruiting foreign workers, Japanese language proficiency remains a relevant barrier. Surveys with employers indicate that insufficient linguistic skills are one of the main difficulties in hiring foreigners, limiting the match between professional qualifications and the occupation held.

Thus, the economic impact of immigration depends not only on the numerical growth of the foreign contingent but also on linguistic integration, occupational mobility, and the recognition of qualifications. These dimensions are particularly relevant for Japan, where traditional labor market characteristics may hinder the integration of foreign workers trained abroad (OECD, 2024b). By articulating quantitative projections with observed statistical relationships, it is found that the trends of population aging and declining fertility are unlikely to be reversed solely by increasing immigration, reinforcing the need for comprehensive public policies aimed at stimulating birth rates (Jones, 2024).

Granger causality test

To deepen the analysis of the relationship between population and immigration, the Granger causality test was applied with an annual lag, evaluating both directions. In the population to immigration direction, the test presented an F-statistic of 2.3931 and a p-value of 0.1728. As the p-value was higher than 0.05, the null hypothesis of absence of Granger causality was not rejected. In the inverse direction, immigration to population, an F-statistic of 0.3036 and a p-value of 0.6015 were obtained, also without rejecting the null hypothesis.

The results indicate that, considering an annual lag, no statistically significant evidence was found that past values of one variable contribute to the prediction of the other. Even when applying the test to the first differences of the series, the results remained non-significant, with p-values of 0.2444 for the population to immigration direction and 0.3059 for the inverse direction. This means that transforming the series into annual variations did not alter the main conclusion.

The interpretation of these results should be particularly cautious due to the small sample size. The common period between the series comprised only 2014 to 2023, totaling ten annual observations. This limited amount of data reduces the degrees of freedom and the statistical power of the tests, preventing the identification of statistically robust temporal relationships. Thus, the absence of significance does not allow us to assert the non-existence of any causal relationship between population and immigration, but only that insufficient statistical evidence was found to identify it in the analyzed sample. The results therefore contribute to delimiting the causal interpretation of the study, indicating that, although the observed negative correlation between population and immigration points to opposite trajectories, the Granger tests did not provide evidence of statistically significant temporal precedence.

In summary, the findings of this research demonstrate that Japan will continue to face a population decline and a sharp aging until 2044, despite the projected growth of immigration. Immigration, although crucial for mitigating labor shortages in specific sectors, is not configured as an isolated mechanism capable of reversing the demographic trajectory. Low fertility, in turn, is intrinsically linked to economic and institutional factors, such as the costs of raising children and the availability of childcare services. Thus, Japan’s demographic sustainability will depend on an integrated strategy that combines policies supporting parenthood with the effective attraction and integration of immigrants, recognizing the analytical limitations imposed by the length of historical series and the univariate nature of the models.

4. Conclusion

This study analyzed Japanese demographic trends and projected their evolution until 2044, focusing on the relationships between population decline, fertility, and immigration. Japan’s continued population reduction was observed, with projections indicating a population of approximately 115.28 million inhabitants in 2044 by the Prophet model and 111.82 million by ARIMA. In parallel, a continuous growth in the immigrant contingent was observed, projected to reach about 15.18 million by Prophet and 9.71 million by ARIMA in the same period. Correlation analysis revealed a strong negative association between fertility rate and the elderly population, and a positive correlation between the elderly population and immigration, suggesting that immigration acts as a response to aging. However, Granger causality tests did not identify statistically significant evidence of temporal precedence between population and immigration, indicating that, although immigration may mitigate labor supply constraints, it does not, in isolation, constitute a sufficient mechanism to reverse Japan’s population decline. Low fertility, in turn, was found to be intrinsically linked to economic and institutional factors, such as the costs of raising children and the availability of childcare services.

Despite the contributions to the understanding of Japanese demographic dynamics, the study acknowledges important limitations. The short duration of the immigration series, which covered only ten years for joint analysis, reduced the statistical power of causality tests and the generalizability of long-term projections. Additionally, the predominantly univariate nature of ARIMA and Prophet models did not allow for the direct incorporation of economic and institutional variables that influence fertility and immigration. The projections for 2044, especially for immigration, should be interpreted with caution, considering the uncertainty introduced by recent changes in migration policies and the confidence intervals with statistically implausible lower bounds. It is suggested that future studies explore multivariate models, such as ARIMAX or VAR, and machine learning models, incorporating longer time series and data disaggregated by age and sex, in order to provide a more comprehensive analysis of demographic mechanisms and their interactions with socioeconomic factors and public policies.

Bibliographic References

Instituto Nacional de Pesquisa Populacional e Segurança Social do Japão [IPSS]. 2024. Projeções Populacionais para o Japão: 2024. Disponível em: <https://www.ipss.go.jp>. Acesso em: 10 mar. 2025.

Ministério das Relações Exteriores do Japão. 2026. Specified Skilled Worker (SSW). Tokyo: MOFA. Disponível em: <https://www.mofa.go.jp/mofaj/ca/fna/ssw/us/index.html>. Acesso em: 21 fev. 2026.

Nakatani, H. 2023. Ageing and shrinking population: The looming demographic challenges of super-aged and super-low fertility society starting from Asia. Glob Health Med. Disponível em: <http://doi.org/10.35772/ghm.2023.01057>. Acesso em: 9 mar. 2025.

Organisation for Economic Co-operation and Development [OCDE].

Statistics Bureau of Japan. Statistical Handbook of Japan. 2024. Disponível em: <https://www.stat.go.jp/english/index.html>. Acesso em: 10 mar. 2025.

United Nations, 2024 [Full reference not found in original document]

Article originating from the Final Course Project of the Specialization in Data Science and Analytics of the MBA USP/Esalq

To learn more about the course, click here and access the MBX Academy platform