## IMF Working Paper No. WP/2025/247 — “Demographics and Consumption in Asia Toward 2050”

## Source details

**Canonical URL:** [IMF Working Paper No. WP/2025/247 — “Demographics and Consumption in Asia Toward 2050”](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025247-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025247-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025247-source-pdf.pdf.json)

---

### Demographic transformation in Asia
- Asia is the most populous region and stands at the forefront of a demographic transformation that will reshape economic and social structures across the region.
- Economies such as Japan and Korea are already experiencing population decline and rapid aging.
- Even economies with growing populations, such as India and the Philippines, will confront sharp shifts in age structure in the coming decades.

### China example: falling births and cascading demand effects
- China: the age 0 population, or the number of babies born in a year, halved between 2017 and 2023, despite the abolition of the one-child policy in 2016.
- Observed industry responses in China:
  - The number of OBGYN hospitals rose through 2018 and declined thereafter.
  - After several years of lag, the number of kindergartens started to decrease.
- Implication: reductions in births imply predictable, lagged declines in demand for baby-related services (hospitals → kindergartens → universities as cohorts age).

### Research question, contribution, and data coverage
- Central question: How do demographics change future consumption in both the total and the composition across consumption categories?
- Contribution:
  - Studies both total consumption and its composition across economies, enabling consistent cross-economy comparisons.
  - Combines UN (2024) population projections and household consumption surveys across seven economies: China, India, Japan, Korea, Philippines, Singapore, and Tuvalu.
  - Develops a conservative estimation method to harmonize household consumption surveys and align their units with population-projection data.
  - Fills a gap in the literature which has mostly studied total consumption or individual categories separately or focused on single-country CGE analyses.
- Countries analyzed: China, India, Japan, Korea, Philippines, Singapore, and Tuvalu.
- Population projection source: UN (2024), cohort-component method projecting population by age and sex from 2024 to 2100; baseline uses the median-variant projection.

### Key descriptive population facts (2024 to 2050)
- Some economies, like India and the Philippines, are expected to expand by more than 15 percent.
- Japan’s population will decline by roughly 15 percent.
- Korea’s population will decline by about 13 percent.
- China’s population will decline by about 11 percent.
- The Philippines’ population is projected to grow by about 16 percent overall, but with a markedly smaller 0–4 cohort relative to adjacent cohorts.
- Population shapes:
  - China (2024): sharp drop at ages 0–4; by 2050 the “cliff” moves to around 30–34.
  - Japan: youngest cohort well under half the size of middle-aged cohorts.
  - Korea: youngest cohort roughly one quarter of the most populous cohort.
  - Singapore: population below age 10 is close to one third of that at ages 25–29.
  - Tuvalu: profile closer to a classic pyramid but highly sensitive to outward migration.

### Consumption data and age-profile regularities
- Consumption categories mapped to COICOP high-level groups: Food and Beverage; Alcohol and Tobacco; Clothing; Housing; Utilities; Furnishing; Health; Transport; Information; Recreation; Education; Restaurants and Hotels; Insurance; Other.
- Household consumption survey sources and years:
  - China 2020: China Family Panel Studies, Institute of Social Science Survey of Peking University, China
  - India 2023-24: Household Consumption Expenditure Survey, Ministry of Statistics and Programme Implementation, India
  - Japan 2019: National Survey of Family Income, Consumption and Wealth, Ministry of Internal Affairs and Communications, Japan
  - Korea 2024: Household Income and Expenditure Survey, Statistics Korea, Korea
  - Philippines 2023: Family Income and Expenditure Survey, Philippine Statistical Authority, Philippines
  - Singapore 2023: Household Expenditure Survey, Department of Statistics, Singapore
  - Tuvalu 2022: Household Income and Expenditure Survey Report, Central Statistics Division, Tuvalu
- Age-profile regularities (illustrated for Japan, 2019):
  - Education consumption concentrated in households with household heads in their 40s to 60s.
  - Health’s share increases with age.
  - Transport’s share declines with age.

### Baseline methodology (partial equilibrium decomposition)
- Aggregate consumption in category k at time τ equals the dot product of the population vector pτ and the age-specific per-capita consumption vector c{k,τ}.
- Change in aggregate consumption from time 0 to t decomposes into:
  - A demographic term: change due to population composition changes holding per-capita age-specific consumption constant.
  - A residual term: changes in per-capita consumption patterns across ages (general equilibrium effects).
- Baseline assumption: per-capita consumption at each age cohort remains constant over time (c{k,t} = c{k,0}), isolating demographic effects (partial equilibrium).
- Interpretation:
  - Under the baseline, aggregate consumption growth mostly follows population growth; compositional effects are limited but can be large in cases where middle-aged cohorts (which consume most) decline drastically (example: Singapore).
  - Aging tends to increase growth of old-dependent consumption categories (e.g., Health and Furnishing) relative to young-dependent ones (e.g., Education and Transport).
  - Differences between consumption growth and population growth approximate per-capita consumption growth; under constant per-capita consumption this difference captures compositional effects of demographics.

### Estimation approaches for different survey structures
- When all household members’ ages are available (China, India, Korea):
  - Individual consumption is estimated by dividing household consumption equally across household members (ci = m_i / n_i).
  - Age-specific average consumption c{k,0}(a) computed as the average across individuals of age a.
  - Alternative assumption for Education: attribute all education expenditure to those below 18 when present (Annex II); this yields larger declines in Education and Total consumption but similar qualitative conclusions.
- When only household head’s age is available (Japan, Philippines, Singapore, Tuvalu):
  - Aggregate at household-unit level; use supplemental household data to construct h0 and estimate household counts over time.
  - Assume constant probability to become household head: P(a) ≔ h0(a) / p0(a); ht(a) ≔ P(a) p_t(a).
  - Rescale household consumption by household size: m{k,t}(a) ≔ s_t / s_0 * m{k,0}(a).
  - Aggregate consumption computed in household units: hτ′ m{k,τ} = pτ′ c{k,τ}.

### Baseline results — total consumption
- Baseline isolates demographic effects by assuming constant consumption per capita at each age cohort, c_{k,t} = c_{k,0}, so growth of consumption would equal population growth absent compositional effects.
- Deviations between consumption growth and population growth reflect compositional effects of demographics (age-structure shifts).
- Population-decreasing economies:
  - Japan, Korea, and China: total consumption declines faster than population. The blue bar for Total lies below the benchmark in red in the figures, implying real consumption declines more rapidly than population growth and per-capita consumption growth is negative over the horizon.
  - Aging and declines in the middle-aged population put additional downward pressure on consumption beyond population decline.
- Population-increasing economies:
  - India and the Philippines: total consumption grows faster than population because the middle-aged population increases and they consume more than other cohorts.
  - Singapore: despite positive population growth, total consumption declines because the decline in the middle-aged population is so fast that negative contributions from the 30s and early 40s offset positive contributions from older cohorts.
- Age-wise decomposition:
  - Contributions defined as in equation (13): ccnttcii... (notation preserved from source). Figure 4 (Singapore) shows negative contributions from middle-aged cohorts leading to overall consumption decline despite population growth.

### Baseline results — composition of consumption
- General pattern:
  - Health consumption tends to grow faster than Education in many economies.
  - Education spending is concentrated in a short parenting-age window; as cohorts pass parenting ages, Education demand declines sharply (example: Japan, Figure 5).
  - Health consumption increases with age; aging therefore raises Health demand (example: Korea, Figure 5).
- Transport:
  - Tends to decline faster than Health but slower than Education; includes commuting and work-related mobility that falls with age.
- Economy-specific patterns:
  - Korea: Food & Beverage consumption does not decline much with age, so its consumption declines less than population.
  - India and Singapore: Housing consumption tends to increase with age, so Housing grows faster than other categories.
- Summary implication:
  - When aging is slow and population rises broadly, most categories move with population.
  - When aging is rapid and consumption for some categories concentrates in narrow age cohorts, demographics can drive both total consumption and composition.

### Key baseline implications and mechanisms
- Compositional effects can be large when middle-aged cohorts change rapidly because they tend to be the largest consumers.
- Aging shifts demand toward old-dependent categories (Health, Furnishing) and away from young-dependent categories (Education, Transport).
- Harmonized household-survey-based implied total consumption, when compared to private consumption expenditure in IMF (2025), ranges from 60 to 100 percent in most cases (Annex III sanity check).

### Extensions — China growth-adjusted consumption path (section 5.1)
- Baseline assumption relaxed: per capita real consumption at each age cohort is not constant; incorporate economic growth by making China’s per capita consumption a weighted average of China’s and Korea’s consumption at 0:
  - c_{k,t}^{CHN} = w c_{k,0}^{KOR} + (1−w) c_{k,0}^{CHN} (equation (14)).
- Weight w disciplined by projected GDP per capita at PPP in 2050, y_{t}^{CHN} (PwC (2017)), and GDP per capita in 2024 for China y_{0}^{CHN} and Korea y_{0}^{KOR} (WEO, IMF (2025)), via:
  - y_{t}^{CHN} = w y_{0}^{KOR} + (1−w) y_{0}^{CHN} (equation (15)).
- Given values:
  - y_{t}^{CHN} ≈ 46550
  - y_{0}^{KOR} ≈ 62648
  - y_{0}^{CHN} ≈ 26879
  - Resulting weight: w ≈ .55
- Main results:
  - Economic growth can offset demographic headwinds: compared to baseline where consumption growth was uniformly negative, all categories exhibit positive growth under the growth-adjusted path.
  - Categories Furnishing, Recreation, and Restaurants & Hotels grow more than 200 percent from 2024 to 2050 (around 3 percent if annualized).
  - Korea’s consumption per capita in 2024 is roughly 5 times larger than China’s for Recreation; with w = .55, a 200–300 percent increase is plausible.
  - Some categories (e.g., Furnishing) see reinforcing effects of economic growth and aging; others (Food & Beverage, Education) see smaller gains because Korea and China per capita consumption are similar in 2024.

### Extensions — Korea fertility scenarios (section 5.2)
- Context:
  - Declining total fertility rate is common; Korea is one of the lowest in Asia as of 2025.
  - UN (2024) projects gradual recovery but subject to uncertainty.
- Experiment:
  - Replace median-variant fertility in Korea’s population projection with UN (2024) high and low fertility paths.
  - Low-fertility path: total fertility rate around half of median path.
  - High-fertility path: total fertility rate reaches 1.5 by 2050.
- Results:
  - High-fertility scenario: decline in consumption is mitigated significantly across categories; higher fertility cushions decline and can turn Health consumption growth positive even though population declines.
  - Low-fertility scenario: consumption declines, notably Education declines by more than 40 percent.
- Mechanism:
  - Younger cohorts are directly affected by fertility changes; categories concentrated in young cohorts (Education) are most impacted.

### Extensions — Tuvalu migration scenario (section 5.3)
- Context:
  - Migration has limited impact for large-population economies but can be significant for small populations.
  - Tuvalu example: Falepili Union (2023) allows up to 280 Tuvaluan citizens to relocate to Australia; first visa lottery conducted in 2025. UN (2024) has not incorporated its impact.
  - 280 visas are around 3 percent of the 10,000 population.
- Scenario:
  - Baseline projection: net emigrants decline to less than 50 and population recovers to the 2025 level by 2050.
  - Extension assumption: at least half of the 280 people with emigration visas relocate abroad through 2050.
  - Resulting population decline: roughly 25 percent by 2050.
  - Note: population projections simulated using experimental UN portal; age distribution of emigrants in that simulation not publicly available, but resulting population distribution suggests middle-aged and young-aged populations emigrate most.
- Results:
  - Consumption declines at a similar speed to population across broad categories.
  - Categories dependent on middle-aged households—Education, Restaurants & Hotels, and Transport—exhibit the largest declines.
- Implication:
  - Alternative migration assumptions can change demographics and consumption drastically for small-population economies, highlighting the large economic impact of emigration.

### Conclusion and policy-relevant takeaways
- Main quantitative findings:
  - Compositional demographic effects can break the direct link between population and total consumption when middle-aged cohorts that support most consumption decline drastically.
  - Aging increases demand for old-dependent categories (Health, Furnishing) and reduces demand for young-dependent categories (Education, Transport).
- Uncertainty and mitigating factors:
  - Economic growth can offset demographic headwinds and substantially alter consumption profiles (example: China growth-adjusted scenario).
  - Deep demographic parameters—total fertility rate and net migration—are uncertain and can materially alter consumption projections (Korea fertility scenarios, Tuvalu migration scenario).
- Policy implications:
  - Structural policies that support economic growth can offset negative demographic impacts on consumption demand in many categories.
  - Policies influencing fertility and migration can have profound implications for consumption dynamics and should be considered in long-term planning for sectoral demand and fiscal pressures.
  - Small-population economies should consider migration policy carefully given its outsized demographic and consumption impact.

*Source: wpiea2025247-source-pdf — IMF Working Paper No. WP/2025/247 — “Demographics and Consumption in Asia Toward 2050”.*

### 1. Introduction

### 1. Introduction

### Demographic transformation in Asia
- Asia is the most populous region and stands at the forefront of a demographic transformation that will reshape economic and social structures across the region.
- Economies such as Japan and Korea are already experiencing population decline and rapid aging.
- Even economies with growing populations, such as India and the Philippines, will confront sharp shifts in age structure in the coming decades.

### China example: falling births and cascading demand effects
- China: the age 0 population, or the number of babies born in a year, halved between 2017 and 2023, despite the abolition of the one-child policy in 2016.
- Observed industry responses in China:
  - The number of OBGYN hospitals rose through 2018 and declined thereafter.
  - After several years of lag, the number of kindergartens started to decrease.
- Implication: reductions in births imply predictable, lagged declines in demand for baby-related services (hospitals → kindergartens → universities as cohorts age).

### Research question and contribution
- Central question: How do demographics change future consumption in both the total and the composition across consumption categories?
- Contribution:
  - Studies both total consumption and its composition across economies, enabling consistent cross-economy comparisons.
  - Combines UN (2024) population projections and household consumption surveys across seven economies: China, India, Japan, Korea, Philippines, Singapore, and Tuvalu.
  - Develops a conservative estimation method to harmonize household consumption surveys and align their units with population-projection data.
  - Fills a gap in the literature which has mostly studied total consumption or individual categories separately or focused on single-country CGE analyses.

### Policy relevance
- Understanding demographic implications for consumption informs long-lived public investment and policy choices (e.g., building schools, roads, training doctors).
- Population projection is one of the most reliable forward-looking data inputs for government budgeting, structural reforms, and industrial policy.
- Different demographic structures imply different dynamism across sectors like education, recreation, and health services.

### Data coverage and key descriptive facts
- Countries analyzed: China, India, Japan, Korea, Philippines, Singapore, and Tuvalu—coverage limited by availability of age information in household consumption surveys.
- Population projection source: UN (2024), cohort-component method projecting population by age and sex from 2024 to 2100; baseline uses the median-variant projection.
- Cross-country population dynamics (2024 to 2050):
  - Some economies, like India and the Philippines, are expected to expand by more than 15 percent.
  - Japan’s population will decline by roughly 15 percent.
  - Korea’s population will decline by about 13 percent.
  - China’s population will decline by about 11 percent.
  - The Philippines’ population is projected to grow by about 16 percent overall, but with a markedly smaller 0–4 cohort relative to adjacent cohorts.
- Population shapes:
  - China (2024): sharp drop at ages 0–4; by 2050 the “cliff” moves to around 30–34.
  - Japan: youngest cohort well under half the size of middle-aged cohorts.
  - Korea: youngest cohort roughly one quarter of the most populous cohort.
  - Singapore: population below age 10 is close to one third of that at ages 25–29.
  - Tuvalu: profile closer to a classic pyramid but highly sensitive to outward migration.

### Consumption data and patterns
- Household consumption surveys record household-level expenditures by detailed category plus demographic information; categories are mapped to COICOP high-level groups:
  - Food and Beverage; Alcohol and Tobacco; Clothing; Housing; Utilities; Furnishing; Health; Transport; Information; Recreation; Education; Restaurants and Hotels; Insurance; Other.
- Table of household consumption survey sources and years:
  - China 2020: China Family Panel Studies, Institute of Social Science Survey of Peking University, China
  - India 2023-24: Household Consumption Expenditure Survey, Ministry of Statistics and Programme Implementation, India
  - Japan 2019: National Survey of Family Income, Consumption and Wealth, Ministry of Internal Affairs and Communications, Japan
  - Korea 2024: Household Income and Expenditure Survey, Statistics Korea, Korea
  - Philippines 2023: Family Income and Expenditure Survey, Philippine Statistical Authority, Philippines
  - Singapore 2023: Household Expenditure Survey, Department of Statistics, Singapore
  - Tuvalu 2022: Household Income and Expenditure Survey Report, Central Statistics Division, Tuvalu
- Age-profile regularities (illustrated for Japan, 2019):
  - Education consumption concentrated in households with household heads in their 40s to 60s.
  - Health’s share increases with age.
  - Transport’s share declines with age.

### Baseline methodology (partial equilibrium decomposition)
- Decomposition: aggregate consumption in category k at time τ equals the dot product of the population vector pτ and the age-specific per-capita consumption vector c{k,τ}.
- Change in aggregate consumption from time 0 to t decomposes into:
  - A demographic term: change due to population composition changes holding per-capita age-specific consumption constant.
  - A residual term: changes in per-capita consumption patterns across ages (general equilibrium effects).
- Baseline assumption: per-capita consumption at each age cohort remains constant over time (c{k,t} = c{k,0}), isolating demographic effects (partial equilibrium).
- Interpretation:
  - Under the baseline, aggregate consumption growth mostly follows population growth; compositional effects are limited but can be large in cases where middle-aged cohorts (which consume most) decline drastically (example: Singapore).
  - Aging tends to increase growth of old-dependent consumption categories (e.g., Health and Furnishing) relative to young-dependent ones (e.g., Education and Transport).
  - Differences between consumption growth and population growth approximate per-capita consumption growth; under constant per-capita consumption this difference captures compositional effects of demographics.

### Estimation approaches for different survey structures
- When all household members’ ages are available (China, India, Korea):
  - Individual consumption is estimated by dividing household consumption equally across household members (ci = m_i / n_i).
  - Age-specific average consumption c{k,0}(a) computed as the average across individuals of age a.
  - Alternative assumption for Education: attribute all education expenditure to those below 18 when present (Annex II); this yields larger declines in Education and Total consumption but similar qualitative conclusions.
- When only household head’s age is available (Japan, Philippines, Singapore, Tuvalu):
  - Aggregate at household-unit level; use supplemental household data (MIAC (2020), UN (2020), MTI (2020), CSDGT (2021)) to construct h0 and estimate household counts over time.
  - Assume constant probability to become household head: P(a) ≔ h0(a) / p0(a); ht(a) ≔ P(a) p_t(a).
  - Rescale household consumption by household size: m{k,t}(a) ≔ s_t / s_0 * m{k,0}(a).
  - Aggregate consumption computed in household units: hτ′ m{k,τ} = pτ′ c{k,τ}.

### Key baseline findings (summary)
- Total consumption growth mostly follows population growth; compositional effects are smaller in aggregate but important for specific categories.
- Middle-aged population is central to total consumption; large declines in middle-aged cohorts can reverse signs of population vs. consumption growth (e.g., Singapore).
- Aging commonly raises growth in old-dependent categories (Health, Furnishing) and lowers growth in young-dependent categories (Education, Transport).
- Harmonized household-survey-based implied total consumption, when compared to private consumption expenditure in IMF (2025), ranges from 60 to 100 percent in most cases (Annex III sanity check).

### Extensions and robustness exercises
- Allow per-capita consumption patterns to change with economic development: economic growth can offset demographic headwinds, especially for categories sensitive to income levels (Recreation; Restaurants & Hotels; Furnishing).
- Scenario analyses altering UN population-projection deep parameters (total fertility rate and net migration) show these factors can significantly impact consumption projections; Korea and Tuvalu used as examples in scenario analyses.
- Measurement challenges and limitations are discussed (age grouping alignment, within-household allocation, China interviewing only those over 9 years old, sensitivity to migration for small economies).

### Paper organization
- Section 2 describes the data.
- Section 3 explains the methodology.
- Section 4 shows the baseline results.
- Section 5 illustrates the extensions.
- Section 6 concludes.

*Source: wpiea2025247-source-pdf — 1. Introduction.*

### 4. Baseline results

### 4. Baseline results

### Total consumption: baseline findings
- Baseline isolates demographic effects by assuming constant consumption per capita at each age cohort, 푐푐
푘푘푡푡 = 푐푐
푘푘0, so growth of consumption would equal population growth absent compositional effects.
- Deviations between consumption growth and population growth reflect compositional effects of demographics (age-structure shifts).
- Population-decreasing economies (examples):
  - Japan, Korea, and China: total consumption declines faster than population. The blue bar for Total lies below the benchmark in red in the figures, implying real consumption declines more rapidly than population growth and per-capita consumption growth is negative over the horizon.
  - Aging and declines in the middle-aged population put additional downward pressure on consumption beyond population decline.
- Population-increasing economies (examples):
  - India and the Philippines: total consumption grows faster than population because the middle-aged population increases and they consume more than other cohorts.
  - Singapore: despite positive population growth, total consumption declines because the decline in the middle-aged population is so fast that negative contributions from the 30s and early 40s offset positive contributions from older cohorts.
- Age-wise decomposition (notation):
  - Define contributions as in equation (13): 푐푐푐푐푛푛푡푡푐푐푖푖푐푐푐푐푡푡푖푖  푐푐푛푛
푘푘
(
푎푎
)
≔
퐶퐶
푘푘푡푡
(
푎푎
)
−퐶퐶
푘푘0
(
푎푎
)
∑
퐶퐶
푘푘0
(
푎푎
)
푎푎
, and 퐶퐶
푘푘휏휏
(
푎푎
)
≔푝푝
휏휏
(
푎푎
)
푐푐
푘푘휏휏
(
푎푎
)
, 휏휏=0,푡푡.
- Figure 4 (Singapore) shows negative contributions from middle-aged cohorts leading to overall consumption decline despite population growth.

### Composition of consumption: baseline patterns
- General pattern: Health consumption tends to grow faster than Education in many economies.
  - Education spending is concentrated in a short parenting-age window; as cohorts pass parenting ages, Education demand declines sharply (example: Japan, Figure 5).
  - Health consumption increases with age; aging therefore raises Health demand (example: Korea, Figure 5).
- Transport:
  - Tends to decline faster than Health but slower than Education; includes commuting and work-related mobility that falls with age.
- Other economy-specific patterns:
  - Korea: Food & Beverage consumption does not decline much with age, so its consumption declines less than population.
  - India and Singapore: Housing consumption tends to increase with age, so Housing grows faster than other categories.
- Summary implication:
  - When aging is slow and population rises broadly, most categories move with population.
  - When aging is rapid and consumption for some categories concentrates in narrow age cohorts, demographics can drive both total consumption and composition.

### Key baseline implications and mechanisms
- Compositional effects can be large when middle-aged cohorts change rapidly because they tend to be the largest consumers.
- Aging shifts demand toward old-dependent categories (Health, Furnishing) and away from young-dependent categories (Education, Transport).

---

### 5. Extensions

### 5.1 China: alternative consumption path with economic growth
- Baseline assumption relaxed: per capita real consumption at each age cohort is not constant; incorporate economic growth by making China’s per capita consumption a weighted average of China’s and Korea’s consumption at 0:
  - 푐푐
푘푘푡푡
퐶퐶퐶퐶 = 푤푤 푐푐
푘푘0
퐾퐾퐾퐾 + (1−푤푤) 푐푐
푘푘0
퐶퐶퐶퐶  (equation (14)).
- Weight 푤푤 disciplined by projected GDP per capita at PPP in 2050, 푦푦
푡푡
퐶퐶퐶퐶 (PwC (2017)), and GDP per capita in 2024 for China 푦푦
0
퐶퐶퐶퐶 and Korea 푦푦
0
퐾퐾퐾퐾 (WEO, IMF (2025)), via:
  - 푦푦
푡푡
퐶퐶퐶퐶 = 푤푤 푦푦
0
퐾퐾퐾퐾 + (1−푤푤) 푦푦
0
퐶퐶퐶퐶  (equation (15)).
- Given values:
  - 푦푦
푡푡
퐶퐶퐶퐶 ≈ 46550
  - 푦푦
0
퐾퐾퐾퐾 ≈ 62648
  - 푦푦
0
퐶퐶퐶퐶 ≈ 26879
  - Resulting weight: 푤푤 ≈ .55
- Main results:
  - Economic growth can offset demographic headwinds: compared to baseline where consumption growth was uniformly negative, all categories exhibit positive growth under the growth-adjusted path.
  - Categories Furnishing, Recreation, and Restaurants & Hotels grow more than 200 percent from 2024 to 2050 (around 3 percent if annualized).
  - Arithmetic illustration: Korea’s consumption per capita in 2024 is roughly 5 times larger than China’s for Recreation; with weight 푤푤 = .55, a 200–300 percent increase is plausible.
  - Some categories (e.g., Furnishing) see reinforcing effects of economic growth and aging; others (Food & Beverage, Education) see smaller gains because Korea and China per capita consumption are similar in 2024.
- Interpretation:
  - Economic growth can produce large shifts in consumption profiles and can quantitatively offset demographic effects, especially for categories where older cohorts in advanced economies consume more than in emerging markets.

### 5.2 Korea: alternative fertility path
- Context:
  - Declining total fertility rate is common; Korea is one of the lowest in Asia as of 2025.
  - UN (2024) projects gradual recovery but subject to uncertainty.
- Experiment:
  - Replace median-variant fertility in Korea’s population projection with UN (2024) high and low fertility paths.
  - Low-fertility path: total fertility rate around half of median path.
  - High-fertility path: total fertility rate reaches 1.5 by 2050.
- Results:
  - High-fertility scenario: decline in consumption is mitigated significantly across categories; higher fertility cushions decline and can turn Health consumption growth positive even though population declines.
  - Low-fertility scenario: consumption declines, notably Education declines by more than 40 percent.
- Mechanism:
  - Younger cohorts are directly affected by fertility changes; categories concentrated in young cohorts (Education) are most impacted.

### 5.3 Tuvalu: alternative migration path
- Context:
  - Migration has limited impact for large-population economies but can be significant for small populations.
  - Tuvalu example: Falepili Union (2023) allows up to 280 Tuvaluan citizens to relocate to Australia; first visa lottery conducted in 2025. UN (2024) has not incorporated its impact.
  - 280 visas are around 3 percent of the 10,000 population.
- Scenario:
  - Baseline projection: net emigrants decline to less than 50 and population recovers to the 2025 level by 2050.
  - Extension assumption: at least half of the 280 people with emigration visas relocate abroad through 2050.
  - Resulting population decline: roughly 25 percent by 2050.
  - Note: population projections simulated using experimental UN portal; age distribution of emigrants in that simulation not publicly available, but resulting population distribution suggests middle-aged and young-aged populations emigrate most.
- Results:
  - Consumption declines at a similar speed to population across broad categories.
  - Categories dependent on middle-aged households—Education, Restaurants & Hotels, and Transport—exhibit the largest declines.
- Implication:
  - Alternative migration assumptions can change demographics and consumption drastically for small-population economies, highlighting the large economic impact of emigration.

---

### 6. Conclusion and policy-relevant takeaways
- Main quantitative findings:
  - Compositional demographic effects can break the direct link between population and total consumption when middle-aged cohorts that support most consumption decline drastically.
  - Aging increases demand for old-dependent categories (Health, Furnishing) and reduces demand for young-dependent categories (Education, Transport).
- Uncertainty and mitigating factors:
  - Economic growth can offset demographic headwinds and substantially alter consumption profiles (example: China growth-adjusted scenario).
  - Deep demographic parameters—total fertility rate and net migration—are uncertain and can materially alter consumption projections (Korea fertility scenarios, Tuvalu migration scenario).
- Policy implications (inferred from results):
  - Structural policies that support economic growth can offset negative demographic impacts on consumption demand in many categories.
  - Policies influencing fertility and migration can have profound implications for consumption dynamics and should be considered in long-term planning for sectoral demand and fiscal pressures.
  - Small-population economies should consider migration policy carefully given its outsized demographic and consumption impact.

*IMF Working Paper No. WP/2025/247 — “Demographics and Consumption in Asia Toward 2050”*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025247-source-pdf.pdf_