## wpiea2025055-print-pdf

## Source details

**Canonical URL:** [wpiea2025055-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025055-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025055-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025055-print-pdf.pdf.json)

---

### I. Introduction — scope, motivation, and identification
- Scope:
  - Analysis covers 42 countries in Europe over the period 1960–2022.
  - Dependent variable: annual percent change in real GDP per capita.
  - Key explanatory variable: total fertility rate (average number of births per woman).
  - Instrument for fertility: comparative abortion index (Forman-Rabinovici and Sommer (2018) dataset; 0–1 scale).
- Motivation and identification challenge:
  - Europe average total fertility rate is 1.46, below the approximate replacement rate of 2.18.
  - Fertility transitions are projected to accelerate, shrinking population and increasing elderly share.
  - Fertility is endogenous to income (reverse causality, persistence in income, and measurement error).
  - IV strategy: use comparative abortion index as plausibly exogenous variation to identify causal impact of fertility on real GDP per capita growth.

### II. Key descriptive statistics and data features
- Fertility and population trends:
  - Europe average total fertility rate: 1.46.
  - Total fertility rate projected to decline to an average of 1.37 by 2100 (Bhattacharjee et al., 2024).
  - Country variation in 1960–2022 sample: from 1.1 to 6.5; sample mean fertility = 2.
  - A total fertility rate below 1.5 is considered “very low” and below 1.3 is classified as “ultra-low”.
- Sample-level statistics (selected):
  - Sample average for economic growth: 2.5 percent (minimum -31.2 percent; maximum 48 percent).
  - Total fertility rate: Observations 2,644; Mean 2; Std. dev. 0.7; Minimum 1.1; Maximum 6.5.
  - Comparative abortion index: Observations 2,148; Mean 0.9; Std. dev. 0.2; Minimum 0; Maximum 1.0.
  - Real GDP per capita: Observations 2,105; Mean 23,398.6; Std. dev. 19,943.3; Minimum 698.2; Maximum 112,417.9.
  - Female labor force participation: Observations 1,568; Mean 50.0; Std. dev. 11.3; Minimum 13.5; Maximum 85.5.
  - Infant mortality: Observations 2,406; Mean 14.1; Std. dev. 16.1; Minimum 1.5; Maximum 171.4.
  - Educational attainments: Observations 1,358; Mean 10.7; Std. dev. 1.9; Minimum 4.5; Maximum 14.3.
  - Government stability index: Observations 1,413; Mean 7.6; Std. dev. 1.7; Minimum 2.0; Maximum 11.5.
- Abortion index handling:
  - Comparative abortion index values from 1992 and 2013 are used to represent prior and subsequent periods to maximize sample; assumes temporal stability in the index.

### III. Empirical strategy
- OLS specification with country and time fixed effects:
  - 훥y_it = β1 + β2 tfr_it−1 + β3 X_it + η_i + μ_t + ε_it
  - Controls (X_it): lagged real GDP per capita, share of industry in GDP, trade openness, population, educational attainments, infant mortality rate, female labor force participation rate, government stability.
  - Robust standard errors clustered at country level.
- Endogeneity concern and IV approach:
  - Reverse causality and measurement error likely bias OLS.
  - Instrument: comparative abortion index; first-stage shows index is statistically significant predictor of total fertility rate and Kleibergen Paap F-statistics > 10.
  - Two-stage least squares (2SLS/IV) implemented to identify causal effect of fertility on real GDP per capita growth.

### IV. Main empirical findings (OLS and 2SLS-IV)
- Selected regression coefficients and standard errors (reported exactly as in source):
  - OLS:
    - Total fertility rate: -1.928*** [0.471]
    - Real GDP per capita: -4.831*** [1.042]
  - OLS (alternative specification):
    - Total fertility rate: -2.290*** [0.648]
    - Real GDP per capita: -7.460*** [0.981]
  - 2SLS-IV (total fertility rate instrumented with comparative abortion index):
    - Total fertility rate: 4.023* [1.538]
    - Real GDP per capita: -1.720*** [0.360]
    - Share of industry: 0.068*** [0.018]
    - Trade openness: 0.012** [0.003]
    - Population: 0.004 [0.151]
    - Educational attainments: 0.243* [0.076]
    - Infant mortality: -0.201* [0.074]
    - Female labor force participation: 0.058*** [0.015]
    - Government stability: 0.267** [0.084]
    - Number of observations: 1,107; Number of countries: 42; Country FE: Yes; Time FE: Yes; R^2: 0.42
- Interpretation:
  - OLS shows a negative and statistically significant association between fertility and real GDP per capita growth.
  - 2SLS-IV reverses the sign: instrumented total fertility rate is positive and statistically significant (4.023*), with absolute magnitude nearly twice as large relative to OLS estimates.
  - Reversal suggests severe endogeneity bias in OLS (reverse causality and/or measurement error attenuating OLS estimates).
  - IV evidence implies higher fertility makes a strong positive contribution to real GDP per capita growth in Europe overall.
- Subsample and robustness:
  - Disaggregation by income group (advanced vs developing economies in Europe) produces positive coefficients on fertility across groups, but statistical significance is achieved only in advanced economies.
  - First-stage diagnostics: Kleibergen Paap F-statistics significantly above 10 (instrument strength).
  - Results remain broadly unchanged when using alternative sample period 1992–2013.

### V. Cross-country IV estimates (selected summary from Appendix Table A2)
- 2SLS-IV coefficients and standard errors (All / AE / EM reported where available):
  - Total fertility rate: 4.023* [1.538], 2.301* [0.846], 16.487 [10.538]
  - Real GDP per capita: -1.720*** [0.360], -1.269*** [0.282], -1.129*** [0.247]
  - Share of industry: 0.068*** [0.018], 0.093*** [0.016], 0.208*** [0.022]
  - Trade openness: 0.012** [0.003], 0.012*** [0.003], 0.012 [0.018]
  - Educational attainments: 0.243* [0.076], 0.052 [0.081], 2.266* [0.855]
  - Infant mortality: -0.201* [0.074], -0.334*** [0.095], -0.221* [0.084]
  - Female labor force participation: 0.058*** [0.015], 0.004 [0.016], 0.075*** [0.014]
  - Government stability: 0.267** [0.084], 0.148 [0.082], 0.084** [0.014]
  - Number of observations: 1,107 (All), 808 (AE), 299 (EM)
  - Number of countries: 42 (All), 27 (AE), 15 (EM)
  - R2: 0.42 (All), 0.57 (AE), 0.26 (EM)
  - Comparative abortion index first-stage: R2 0.3969, Adjusted R2 0.3737, F-test statistic 53.1437
- Note: Dependent variable is real GDP per capita growth. Robust standard errors reported in brackets. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

### VI. Lithuania macrosimulation — setup, projections, and outcomes
- Purpose and methodology:
  - Macrosimulation model follows Ashraf, Weil, and Wilde (2013) and Karra, Canning, and Wilde (2017); parameterized with country-specific data and UN data.
  - Model incorporates dynamic evolution of population age structure and female labor force participation; assumes fertility behavior exogenous.
- Scenario definitions:
  - Medium-variant scenario (baseline fertility path).
  - Low-variant scenario: assumes fertility will remain 0.5 children below fertility in the medium variant over the projection period.
- Historical and projected Lithuanian demographics (reported exactly):
  - The total fertility rate in Lithuania declined by 37.4 percent from 2.03 to 1.27 births per woman between 1990 and 2023.
  - Projected total fertility rate: set to decline to 0.88 births per woman by 2050.
  - Population: 3.7 million in 1990; 2.8 million in 2023; projected to be less than 2.3 million by 2050.
  - Old-age dependency rate projected to rise from 16.5 percent to 57.9 percent over the same period (1990 to 2050).
- Macrosimulation outcomes (key projection reported exactly):
  - A reduction in fertility lowers the level of population, the share of working-age adults, and the level of income per capita in Lithuania by as much as 17.6 percent under the low-fertility scenario relative to the baseline over the course of the next 75 years.
  - The ratio of real GDP per capita under the low-variant scenario to that under the medium-variant scenario:
    - Peaks at 1.1242 in 2055.
    - Declines to 0.9481 by 2100.
  - Interpretation: A shrinking labor force due to low fertility and population aging will turn the contribution of labor to economic growth negative, making productivity improvements the key factor in growth dynamics.

### VII. Long-term care, labor supply, and demographic pressures
- Europe-wide old-age dependency context:
  - Old-age dependency ratio (population aged 65 and over as a share of population aged 20–64) already accounts for 37.3 percent in Europe.
  - Projected to be 55.2 percent by 2050 and 59.1 percent by 2070.
- Caregiving implications:
  - Much of the growing demand for long-term care is met by unpaid family caregivers, putting pressure, especially on women, to exit the labor force.
  - Long-term caregiving effects on labor supply and income growth are important to consider in fast-aging countries.

### VIII. Policy implications and recommendations
- Main policy takeaway:
  - The downward fertility transition in Europe, accompanied by fast-aging population, is a drag on income per capita growth.
- Policy levers suggested:
  - Family-friendly interventions to raise fertility:
    - Reduce the cost of education.
    - Provide childcare facilities.
  - Labor market reforms:
    - Increase flexibility through teleworking and alternative working hour arrangements to better gender-balance childcare responsibilities and increase labor force participation of women with children.
  - Immigration and integration:
    - Use immigration and appropriate migrant integration policies to increase fertility, expand the workforce, and boost potential growth in aging societies.
- Contextual note:
  - International experience provides little hope for demographic reversal in the short run; social norms and reproductive preferences change slowly, so no quick fixes exist for ultra-low fertility.

### IX. Summary conclusions
- Fertility trends and growth:
  - OLS estimates indicate a negative relationship between fertility and real GDP per capita growth.
  - IV estimates (instrumenting fertility with comparative abortion index) show a statistically significant positive impact of fertility on real GDP per capita growth, with a larger magnitude than OLS—implying downward fertility transition is a drag on income per capita growth in Europe.
- Heterogeneity:
  - Positive IV effects are present across country groups; statistical significance mainly in advanced economies.
- Policy direction:
  - Family support, childcare, education affordability, labor market flexibility, and immigration/integration policies are logical levers to mitigate negative economic effects of ultra-low fertility.

*Source: wpiea2025055-print-pdf - 2022. This means that the downward fertility transition across Europe, accompanied by fast-aging*

### 2022. This means that the downward fertility transition across Europe, accompanied by fast-aging

### wpiea2025055-print-pdf - 2022. This means that the downward fertility transition across Europe, accompanied by fast-aging

### I. Introduction — scope, motivation, and identification
- Scope:
  - Analysis covers 42 countries in Europe over the period 1960–2022.
  - Dependent variable: annual percent change in real GDP per capita.
  - Key explanatory variable: total fertility rate (average number of births per woman).
  - Instrument for fertility: comparative abortion index (Forman-Rabinovici and Sommer (2018) dataset; 0–1 scale).
- Motivation and main identification challenge:
  - Europe average total fertility rate is 1.46, below the approximate replacement rate of 2.18.
  - Fertility transitions are projected to accelerate, shrinking population and increasing elderly share.
  - Fertility is endogenous to income (reverse causality, persistence in income, and measurement error).
  - Instrumental variable (IV) strategy: use comparative abortion index as plausibly exogenous variation to identify causal impact of fertility on real GDP per capita growth.

### II. Key descriptive statistics and data features
- Fertility and population trends:
  - Europe average total fertility rate: 1.46.
  - Total fertility rate projected to decline to an average of 1.37 by 2100 (Bhattacharjee et al., 2024).
  - Country variation in 1960–2022 sample: from 1.1 to 6.5; sample mean fertility = 2.
  - A total fertility rate below 1.5 is considered “very low” and below 1.3 is classified as “ultra-low”.
- Sample-level statistics (selected):
  - Sample average for economic growth: 2.5 percent (minimum -31.2 percent; maximum 48 percent).
  - Total fertility rate: Observations 2,644; Mean 2; Std. dev. 0.7; Minimum 1.1; Maximum 6.5.
  - Comparative abortion index: Observations 2,148; Mean 0.9; Std. dev. 0.2; Minimum 0; Maximum 1.0.
  - Real GDP per capita: Observations 2,105; Mean 23,398.6; Std. dev. 19,943.3; Minimum 698.2; Maximum 112,417.9.
  - Female labor force participation: Observations 1,568; Mean 50.0; Std. dev. 11.3; Minimum 13.5; Maximum 85.5.
  - Infant mortality: Observations 2,406; Mean 14.1; Std. dev. 16.1; Minimum 1.5; Maximum 171.4.
  - Educational attainments: Observations 1,358; Mean 10.7; Std. dev. 1.9; Minimum 4.5; Maximum 14.3.
  - Government stability index: Observations 1,413; Mean 7.6; Std. dev. 1.7; Minimum 2.0; Maximum 11.5.
- Abortion index handling:
  - Comparative abortion index values from 1992 and 2013 are used to represent prior and subsequent periods to maximize sample; assumes temporal stability in the index.

### III. Empirical strategy
- OLS specification with country and time fixed effects:
  - 훥y_it = β1 + β2 tfr_it−1 + β3 X_it + η_i + μ_t + ε_it
  - Controls (X_it): lagged real GDP per capita, share of industry in GDP, trade openness, population, educational attainments, infant mortality rate, female labor force participation rate, government stability.
  - Robust standard errors clustered at country level.
- Endogeneity concern and IV approach:
  - Reverse causality and measurement error likely bias OLS.
  - Instrument: comparative abortion index; first-stage shows index is statistically significant predictor of total fertility rate and Kleibergen Paap F-statistics > 10.
  - Two-stage least squares (2SLS/IV) implemented to identify causal effect of fertility on real GDP per capita growth.

### IV. Main empirical findings (OLS and 2SLS-IV)
- Table 2 (selected coefficients and standard errors shown exactly as in source):
  - OLS:
    - Total fertility rate: -1.928*** [0.471]
    - Real GDP per capita: -4.831*** [1.042]
  - OLS (alternative specification):
    - Total fertility rate: -2.290*** [0.648]
    - Real GDP per capita: -7.460*** [0.981]
  - 2SLS-IV (total fertility rate instrumented with comparative abortion index):
    - Total fertility rate: 4.023* [1.538]
    - Real GDP per capita: -1.720*** [0.360]
    - Share of industry: 0.068*** [0.018]
    - Trade openness: 0.012** [0.003]
    - Population: 0.004 [0.151]
    - Educational attainments: 0.243* [0.076]
    - Infant mortality: -0.201* [0.074]
    - Female labor force participation: 0.058*** [0.015]
    - Government stability: 0.267** [0.084]
    - Number of observations: 1,107; Number of countries: 42; Country FE: Yes; Time FE: Yes; R^2: 0.42
- Interpretation of results:
  - OLS shows a negative and statistically significant association between fertility and real GDP per capita growth.
  - 2SLS-IV reverses the sign: instrumented total fertility rate is positive and statistically significant (4.023*), with absolute magnitude nearly twice as large relative to OLS estimates.
  - This reversal suggests severe endogeneity bias in OLS (reverse causality and/or measurement error attenuating OLS estimates).
  - IV evidence implies higher fertility makes a strong positive contribution to real GDP per capita growth in Europe overall.
- Subsample results:
  - Disaggregation by income group (advanced vs developing economies in Europe) produces positive coefficients on fertility across groups, but statistical significance is achieved only in advanced economies.
  - Magnitude of coefficient is larger in developing countries but not statistically significant at conventional levels.
- Robustness:
  - First-stage diagnostics: Kleibergen Paap F-statistics significantly above 10 (instrument strength).
  - Results remain broadly unchanged when using alternative sample period 1992–2013 (note: source states these results remain broadly unchanged).

### V. Macrosimulation for Lithuania — outline and key projection
- Purpose:
  - Build a macrosimulation model (following Ashraf, Weil, and Wilde (2013) and Karra, Canning, and Wilde (2017)) to examine fertility’s impact on real GDP per capita growth in Lithuania.
  - Model accounts for changes in population age structure and female labor force participation.
- Key projection reported:
  - A reduction in fertility lowers the level of population, the share of working-age adults, and the level of income per capita in Lithuania by as much as 17.6 percent under the low-fertility scenario relative to the baseline over the course of the next 75 years.
- Recent Lithuanian fertility change (reported exactly):
  - The total fertility rate in Lithuania declined by 37.4 percent from 2.03 to 1.27 births per woman between 1990 and

### VI. Policy implications and recommendations
- Main policy takeaway:
  - The downward fertility transition in Europe, accompanied by fast-aging population, is a drag on income per capita growth.
- Policy measures suggested (drawn from evidence and literature cited):
  - Family-friendly interventions to raise fertility:
    - Reduce the cost of education.
    - Provide childcare facilities (evidence cited: Bauernschuster, Hener, and Rainer, 2016; Sobotka, Matysiak, and Brzozowska, 2019; UN, 2021; Zhang et al., 2023).
  - Labor market reforms:
    - Increase flexibility through teleworking and alternative working hour arrangements to better gender-balance childcare responsibilities and increase labor force participation of women with children.
  - Immigration policy:
    - Use immigration and appropriate migrant integration policies to increase fertility, expand the workforce, and boost potential growth in aging societies (evidence cited: Bove and Elia, 2017; Mitaritonna, Orefice, and Peri, 2017; Dustmann and Preston, 2019; Tabellini, 2020).
- Contextual note:
  - International experience provides little hope for demographic reversal in the short run; social norms and reproductive preferences change slowly, so no quick fixes exist for ultra-low fertility.

### VII. Summary conclusions
- Fertility trends and growth:
  - OLS estimates indicate a negative relationship between fertility and real GDP per capita growth.
  - IV estimates (instrumenting fertility with comparative abortion index) show a statistically significant positive impact of fertility on real GDP per capita growth, with a larger magnitude than OLS—implying downward fertility transition is a drag on income per capita growth in Europe.
- Heterogeneity:
  - Positive IV effects are present across country groups; statistical significance mainly in advanced economies.
- Policy direction:
  - Family support, childcare, education affordability, labor market flexibility, and immigration/integration policies are logical levers to mitigate negative economic effects of ultra-low fertility.

*Source: wpiea2025055-print-pdf - 2022. This means that the downward fertility transition across Europe, accompanied by fast-aging*

### 2023. It is now one of the lowest fertility rates in Europe and set to decline to 0.88 births per

### wpiea2025055-print-pdf - 2023. It is now one of the lowest fertility rates in Europe and set to decline to 0.88 births per

### Key findings (cross-country and causal inference)
- Using the comparative abortion index as an instrument for the total fertility rate, fertility has a significant positive effect on real GDP per capita growth after controlling for economic, demographic, social and political factors in a sample of 42 European countries over the period 1960–2022.
- The downward fertility transition in Europe, coupled with fast-aging population, is a drag on income per capita growth.
- Empirical 2SLS-IV results (Appendix Table A2) highlight:
  - Total fertility rate coefficient: 4.023* (All), 2.301* (AE), 16.487 (EM) with reported standard errors [1.538], [0.846], [10.538].
  - Real GDP per capita coefficient: -1.720*** (All), -1.269*** (AE), -1.129*** (EM) with standard errors [0.360], [0.282], [0.247].
  - Share of industry: 0.068*** (All), 0.093*** (AE), 0.208*** (EM) with standard errors [0.018], [0.016], [0.022].
  - Trade openness: 0.012** (All), 0.012*** (AE), 0.012 (EM) with standard errors [0.003], [0.003], [0.018].
  - Educational attainments: 0.243* (All), 0.052 (AE), 2.266* (EM) with standard errors [0.076], [0.081], [0.855].
  - Infant mortality: -0.201* (All), -0.334*** (AE), -0.221* (EM) with standard errors [0.074], [0.095], [0.084].
  - Female labor force participation: 0.058*** (All), 0.004 (AE), 0.075*** (EM) with standard errors [0.015], [0.016], [0.014].
  - Government stability: 0.267** (All), 0.148 (AE), 0.084** (EM) with standard errors [0.084], [0.082], [0.014].
  - Number of observations: 1,107 (All), 808 (AE), 299 (EM).
  - Number of countries: 42 (All), 27 (AE), 15 (EM).
  - R2: 0.42 (All), 0.57 (AE), 0.26 (EM).
  - Comparative abortion index first-stage statistics: R2 0.3969, Adjusted R2 0.3737, F-test statistic 53.1437.
- Note: The dependent variable in these regressions is economic growth as measured by real GDP per capita growth. Robust standard errors reported in brackets. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

### Lithuania macrosimulation: setup and assumptions
- Model parameterization:
  - Parameterized using country-specific data and estimations.
  - Assumes fertility behavior to be exogenous.
  - Incorporates dynamic evolution of population age structure and female labor force participation.
  - Methodology draws on UN data and the Ashraf, Weil, and Wilde (2013) methodology.
- Scenarios modeled:
  - Medium-variant scenario (baseline fertility path).
  - Low-variant scenario: assumes fertility will remain 0.5 children below fertility in the medium variant over the projection period.

### Lithuania demographic projections and economic implications
- Historical and projected fertility and population:
  - Total fertility rate in Lithuania declined by 37.4 percent from 2.03 to 1.27 births per woman between 1990 and 2023.
  - Projected total fertility rate: set to decline to 0.88 births per woman by 2050.
  - Population projections: 3.7 million in 1990; 2.8 million in 2023; projected to be less than 2.3 million by 2050.
  - Old-age dependency rate projected to rise from 16.5 percent to 57.9 percent over the same period (1990 to 2050).
- Macrosimulation outcomes:
  - Reduction in fertility lowers the level of population and the share of working-age adults in the population over the next 75 years.
  - The ratio of real GDP per capita under the low-variant scenario to that under the medium-variant scenario:
    - Peaks at 1.1242 in 2055.
    - Declines to 0.9481 by 2100.
  - Income per capita is projected to fall by 17.6 percent with lower fertility relative to the baseline (low-variant vs medium-variant) over the modeled horizon.
  - Interpretation: A shrinking labor force due to low fertility and population aging will turn the contribution of labor to economic growth negative, making productivity improvements the key factor in growth dynamics.

### Long-term care, labor supply, and demographic pressures
- Europe-wide old-age dependency context:
  - Old-age dependency ratio (population aged 65 and over as a share of population aged 20–64) already accounts for 37.3 percent in Europe.
  - Projected to be 55.2 percent by 2050 and 59.1 percent by 2070.
- Caregiving implications:
  - Much of the growing demand for long-term care is met by unpaid family caregivers, putting pressure, especially on women, to exit the labor force.
  - Long-term caregiving effects on labor supply and income growth are important to consider in fast-aging countries.

### Policy implications and recommendations
- Boosting fertility can help lessen the decline in labor force and raise growth potential in fast-aging countries, but demographic change is slow and there are no quick fixes to ultra-low fertility.
- Policy levers suggested by literature (not modeled here but referenced as evidence):
  - “Family-friendly” policy interventions aimed at reducing the cost of education and improving the provision of childcare facilities can help raise fertility rates.
  - Labor market reforms to increase flexibility through teleworking and alternative working hour arrangements could help better gender-balance childcare responsibilities and boost labor force participation of women with children.
  - Immigration plays an important role in increasing fertility, expanding the workforce, and boosting potential growth in an aging society.
- The paper concludes that demography is not destiny, but changing social norms and reproductive preferences slowly limits short-run demographic reversals.

*Source: Author’s estimations.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025055-print-pdf.pdf_