## 1. Government Investment Elasticities

## Source details

**Canonical URL:** [1. Government Investment Elasticities](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10110.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10110.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10110.pdf.json)

---

### Background and context
- Japan experienced a sharp slowdown in the 1990s: average growth declined to 1.2 percent from an earlier average of 3.9 percent.
- Between 1989 and 1991 the stock market declined by 41 percent and credit flow declined by 71 percent.
- Fifteen spending stimulus packages were implemented between 1990 and 2008.
- Aggregate announced stimulus components (about 28 percent of GDP in total) included:
  - public works and social infrastructure related projects, including land acquisition (14.2 percent of 2000 GDP);
  - credit guarantees and augmentation of credit lines to banks for loans to small and medium-sized enterprises and for the housing sector (8.5 percent of GDP);
  - employment assistance and cash transfers (2.1 percent of GDP);
  - tax measures (3.3 percent of GDP).
- Local governments:
  - comprise 60 percent of spending and an even larger share of public investment;
  - implemented about 80 percent of public works projects and financed about 65 percent of them during the nineties.
- Structural balances deteriorated from a surplus of about 1/5 percent of GDP in 1990 to a deficit of 6 percent of GDP in 2000. Main contributors:
  - declining taxes (3 percentage points),
  - increases in social security costs (3½ percentage points),
  - the remaining 1 percent of GDP due to government spending on land and capital transfers.

### Literature evidence on multipliers (select findings)
- VAR-based short-run spending multiplier estimates for Japan have ranged from 0.4 to 0.7.
- Kuttner and Posen (2002) find a cumulative four-year spending multiplier of 2.0.
- Hida and others (2008) find the public investment multiplier peaks at 1.1 at a two-year horizon.
- Several studies document heterogeneity over time and across types of spending; some find declining multipliers over time.

### Channels and constraints affecting multiplier size
- Factors supporting limited crowding out:
  - High gross government debt but low net debt by industrial-country standards.
  - Debt predominantly domestically held by long-term institutional investors, containing default and inflation risks.
  - Low short-term interest rates, though long-term rates increased, indicating some crowding out.
- Factors reducing fiscal effectiveness:
  - Credit crunch and deleveraging by firms limiting private investment response.
  - An appreciating yen limiting investment growth.
  - Composition effects from an aging population shifting spending toward social security with lower multipliers.
- Efficiency and allocation concerns:
  - Public investment often allocated toward social infrastructure in rural areas (e.g., rural roads, agriculture) with lower marginal productivity than urban projects.
  - Political economy factors and local special interest groups influenced allocation, sometimes favoring rural areas over metropolitan centers.

### Empirical approach used in this study
- Data: annual time-series at the prefecture level, 1990–2000, with detailed components of general government public investment and local government expenditures.
- Estimation model (log-levels, within-prefecture variation):
  - Yc,t = a_c + d_t + p * F_c,t + X_c,t + u_c,t
  - Yc,t: real per capita value added of prefecture c in year t (logs)
  - F_c,t: government expenditure of prefecture c in year t (logs)
  - X_c,t: vector of control variables at prefecture-year level
  - a_c: prefecture fixed effects; d_t: year fixed effects
- Identification features:
  - Year fixed effects control for national shocks including aggregate monetary policy.
  - Cyclically adjusted fiscal variables follow Blanchard and Perotti (2002).
  - Endogeneity addressed by treating government expenditure as endogenous and using SYS-GMM (Blundell and Bond, 1998), instrumenting levels with lagged first-differences.
  - Converting elasticity to absolute multiplier: multiply elasticity estimate by the inverse of the share of government expenditures in GDP; sample shares are 0.14 for local government expenditures and 0.12 for general government investment.

### Key empirical result: government investment elasticity
- Baseline SYS-GMM estimate (general government public investment elasticity):
  - elasticity = 0.05 (statistically significant at the 1 percent level).

### Government investment elasticities (Table 1) — baseline and components
- Baseline elasticity estimates range from 0.051*** to 0.030** across specifications; preferred estimate reported is 0.033*** (standard error (0.013)).
- Lagged GDP coefficient examples: 0.732*** (0.039) to 0.581*** (0.040); final specification shows 0.588*** (0.040).
- Tax revenue coefficients significant negative in some specifications: -0.138*** (0.023) and -0.130*** (0.025).
- Private investment coefficients positive and significant in extended specifications: 0.097*** (0.026), 0.100*** (0.026), 0.068*** (0.023), 0.064*** (0.021).
- Exports positive and significant in some specifications: 0.038* (0.020) and 0.043** (0.021).
- Government investment externality (other prefectures’ investment) coefficient: 0.006 (0.040), not statistically significant.
- Estimation method: SYS-GMM using instruments dating 2 years and further back. AR(2) and Hansen J p-values reported; observations up to 470 (410 in some columns).

### Public investment multiplier estimates (section B)
- Impact multiplier (regional): 0.28.
  - Calculation: Multiplier (0.28) = Elasticity (0.033) / Public Investment to GDP ratio (0.12).
- Medium-term cumulative multiplier for a fully persistent shock: 0.67.
  - Calculation: Medium-term multiplier (0.67) = Multiplier (0.28) / (1 - coefficient on lagged dependent variable (0.588)).

### Multipliers by type of investment and administrative level (Table 2)
- By type — elasticity (standard error) and implied multiplier:
  - Livelihood: Elasticity 0.018* (0.011); Multiplier 0.35. (Multiply by 1/(0.053).)
  - Industry: Elasticity 0.003 (0.005); Multiplier 0.15. (Multiply by 1/(0.026).)
  - Agriculture: Elasticity 0.029** (0.012); Multiplier 2.04. (Multiply by 1/(0.015).)
  - Land Conservation: Elasticity 0.011 (0.010); Multiplier 1.15. (Multiply by 1/(0.012).)
  - Other: Elasticity 0.005 (0.004); Multiplier 0.41. (Multiply by 1/(0.012).)
- By administrative level — elasticity (standard error) and implied multiplier:
  - Central Government: Elasticity 0.008 (0.008); Multiplier 0.26. (Multiply by 1/(0.040).)
  - Prefecture Government: Elasticity 0.003 (0.012); Multiplier 0.14. (Multiply by 1/(0.036).)
  - City Government: Elasticity 0.033*** (0.012); Multiplier 0.78. (Multiply by 1/(0.045).)
- Observations: 410. Prefecture and year fixed effects included. Method: SYS-GMM.

### Local government expenditure multipliers (Table 3)
- Aggregate local government expenditures:
  - Elasticity estimate: 0.036* (0.02). Implied multiplier: 0.26.
  - Implied cumulative multiplier reported about 0.53.
- Component-level elasticities and implied multipliers:
  - Transfers to Firms: Elasticity 0.045** (0.02); Implied multiplier 3.46. (Multiply by 1/(0.013).)
  - Ordinary Construction: Elasticity 0.036*** (0.01); Implied multiplier 0.76. (Multiply by 1/(0.047).)
  - Social Assistance: Elasticity -0.001 (0.01); Implied multiplier -0.25. (Multiply by 1/(0.004).)
  - Government Personnel: Elasticity -0.011 (0.07); Implied multiplier -0.28. (Multiply by 1/(0.039).)
- Observations: 410. AR(2) and Hansen J p-values indicate instrument validity (examples: AR(2) p-value 0.219, Hansen J p-value 0.983 for aggregate). Method: SYS-GMM.

### Robustness analysis (section D)
- Panel VAR (one lag): public investment multiplier ≈ 0.24; local government expenditure multiplier ≈ 0.49 (Appendix Table 11).
- Level-differences specification (direct multiplier):
  - Public investment multiplier: 0.67; 95 percent confidence interval band 0.3 to 1.04 (Appendix Table 12).
  - Local government spending multiplier: 0.58; 95 percent confidence interval band .12 and .99.
- Quadratic term in log specification: quadratic coefficient positive and significant for both public investment and local government expenditure; sample-average multipliers estimated 0.26 (public investment) and 0.21 (local government expenditure).
- Interaction with output gap: interaction term negative but not statistically significant.

### Crowding-out effects and private sector responses (Table 5, section A)
- Effects of total government investment:
  - On private investment: coefficient 0.219 (0.144) — positive but statistically insignificant.
  - On private consumption: coefficient -0.099* (0.06) — a 1 percent increase in government investment decreased private consumption by 0.1 percent.
- Effects of local government expenditures:
  - On private investment: coefficient -0.518*** (0.195) — a 10 percent increase in local government expenditures decreased private investment by more than 5.2 percent on average.
  - On private consumption: coefficient 0.063 (0.192) — positive but statistically insignificant.
- Controls include tax revenue, public debt, exports, employment, output, and private investment where appropriate. Method: SYS-GMM.

### Regional allocation, heterogeneity, and equity vs. efficiency (section B)
- Prefecture-specific multipliers estimated by interacting government investment with prefecture fixed effects; kernel density of 47 prefecture-specific multipliers shows substantial dispersion.
- Nonparametric local polynomial estimates:
  - Relationship between public investment per capita (1990–2000) and prefecture-specific multiplier: upward-sloping — more investment went to regions with higher multipliers.
  - Relationship between prefecture GDP per capita (1990–2000) and investment per capita: downward-sloping — evidence of equity motive (more investment to poorer regions).
- Interpretation: allocation reflects equity motives but does not appear to explain low aggregate multipliers through misallocation to low-multiplier regions; tradeoffs with multiplier effects seem limited.

### Marginal productivity of public capital and evolution of multipliers (section C)
- Production function estimates (Table 6):
  - Constant returns to scale (column (1)): Private capital elasticity 0.283** (0.13); Public capital elasticity 0.323*** (0.10). Observations 368.
  - Flexible functional form (column (2)): Private capital elasticity 0.580*** (0.18); Public capital elasticity 0.461*** (0.13). Observations 368.
- Marginal product of public capital (elasticity × output-to-public-capital ratio) shows a declining trend for 1975–2000 with strong declines in 1975–85 and 1990–2000.
- Evolution of government expenditure multiplier over time (Table 7):
  - 1975–1989: Government expenditures coefficient 0.102*** (0.03); Implied multiplier 0.8. Observations 690.
  - 1990–2000: Government expenditures coefficient 0.087*** (0.02); Implied multiplier 0.6. Observations 470.
  - Interpretation: government expenditure multiplier about 20 to 25 percent larger in 1975–1989 relative to 1990–2000; decline consistent with lower marginal product of public capital.

### Other factors discussed
- Duration of shocks matters: permanent fiscal changes have larger effects than transitory ones.
- Financing method matters: distortionary tax financing can reduce output while deficit financing may increase output; context-specific implications discussed for Japan.
- Zero lower bound: multiplier largest when monetary policy constrained; in Japan nominal interest rate around 8 percent at start of 1990s and reached zero bound only towards end of decade.

### Conclusions and policy-relevant implications (section VI)
- Baseline finding: investment multipliers in Japan are low and have declined over time but are higher than multipliers for public consumption.
- Local (city) government investment multipliers are higher than central and prefecture government multipliers; decentralized investment appears more effective.
- Multipliers vary substantially by type of investment: agriculture and certain local “livelihood” investments show higher multipliers than industry-focused (infrastructure) investment in these estimates.
- Factors contributing to low multipliers include:
  - Evidence of crowding-out effects, particularly for local government spending on private investment.
  - Declining marginal productivity of public capital due to overinvestment and large existing public capital stock.
  - Allocation patterns motivated by equity did not evidently worsen aggregate multipliers materially.
- Policy considerations:
  - Prioritize forms and administrative levels of public investment that yield higher multipliers (e.g., city-level “livelihood” projects).
  - Take into account declining marginal returns to public capital when designing stimulus via public investment.

### Appendix I: Fiscal stimulus packages (1990–2008) — aggregate figures and components
- Total size of stimulus package: 205.5
- Total size/GDP (%) (average): 2.2, 2.7, 1.2, 3.1, 1.4, 2.9, 3.3, 4.7, 3.6, 2.2, 1.2, 0.9, 3.1, 2.2, 5.1, 2.3
- Selected components (values preserved as listed):
  - Tax cut: 0.2
  - Cash transfer to households: 0.7, 2.0, 2.7
  - Government investments (Ig) to build up social infrastructure: 6.3, 7.2, 2.0, 3.7, 5.4, 9.1, 7.7, 8.1, 6.8, 5.2, 4.2, 3.4, 1.9, 2.2, 73.1
  - Public works involving central government: 4.5, 5.6, 1.5, 3.4, 6.7, 1.6, 3.6, 2.9, 2.6, 2.3, 2.6, 1.0, 38.2
  - Public works by local governments: 1.8, 1.6, 0.5, 0.3, 1.0, 1.5, 6.7
  - Science and technology: 0.3, 11.1, 1.2, 11.2, 0.3, 0.2, 0.1, 6.4
  - Education and social welfare: 11.1, 0.6, 0.5, 0.7, 0.5, 0.7, 5.1
  - Alternative energy and environment: 1.6, 10.6, 0.6, 3.8
  - Natural disaster relief: 5.1, 1.4, 11.3, 1.6, 0.5, 1.7, 0.4, 13.0
  - Other government measures: 4.5, 5.8, 4.0, 5.7, 1.5, 5.2, 4.4, 9.1, 1.3, 5.8, 5.8, 0.0, 11.8, 9.6, 22.7, 107.1
    - Acquisition of land for public use: 1.6, 1.6, 0.3, 2.3, 3.2, 1.1, 10.1
    - Employment support: 0.0, 0.0, 0.0, 0.1, 1.0, 1.0, 1.3, 1.3, 0.9, 0.3, 5.9
    - Expansion of policy lending for housing sector: 0.8, 1.8, 2.9, 1.2, 0.5, 1.2, 10.4
    - Expansion of policy lending and government guarantees for non-financial sector (small and medium size businesses): 1.2, 1.9, 0.8, 1.4, 1.4, 1.4, 2.0, 5.9, 7.4, 4.5, 4.5, 10.9, 9.1, 21.8, 74.2
    - Others: 0.9, 0.5, 0.8, 0.1, 1.2, 0.9, 0.5, 0.6, 6.5
- Aggregate of key stimulus components:
  - (Ig + Tax cuts + Cash transfer): 98.4
  - (Ig + Tax cuts + Cash transfer) / GDP (%) (average): 1.3, 1.5, 0.4, 2.0, 0.4, 1.1, 1.8, 0.4, 0.4, 2.4, 2.9, 1.4, 1.0, 0.0, 0.9, 0.7, 0.4, 0.8, 1.1
- Nominal GDP (series as reported): 483.8, 480.7, 480.7, 487.0, 496.5, 496.5, 496.5, 508.4, 513.3, 503.3, 503.3, 499.5, 504.1, 493.6, 493.6, 489.9, 526.9, 526.9
- Central government bond issuance in supplementary budgets (series as reported): 2.3, 2.2, 3.6, 2.2, 2.8, 4.7, 6.1, 2.3, 7.6, 2.0, 1.7, 0.0, 5.0, 0.4, n.a., 5.2, 2.9

### Appendix II: Data description and investment classification
- Data scope and sources:
  - Annual prefecture-level data for 47 Japanese prefectures during 1990–2000 from the Japan Statistical Yearbook.
  - Government investment data from the annual report on administrative investment (Ministry of Home Affairs); expenditure-based covering maintenance and repair of facilities, improvement projects (including cost of land and compensation), office expenses, planning and surveys.
  - Local government finance data from the annual statistical report on local government finance (Ministry of Home Affairs).
  - Prefecture-level GDP, export, employment, and investment data from the Japan Statistical Yearbook.
- Investment categories and shares (exact labels and shares):
  - Livelihood Investment (44% of total investment): city, town and village roads; streets; city planning; housing; environment sanitation; welfare (including works of hospitals); educational facilities; water supplies.
  - Industry Investment (22% of total investment): national highways and prefectural roads; harbors; airports; industrial water.
  - Agricultural Investment (12% of total investment): agriculture, forestry, fishery.
  - Land Conservation Investment (10% of total investment): forest and river conservation; seashore conservation.
  - Other Investment (12% of total investment): unemployment measures; disaster restoration; government office repairs; railways; subways electricity; gas; residential land formation; other.

### Appendix III: Alternative estimation strategies and multiplier estimates
- Panel VAR (one-lag, six variables) — Appendix Table 11:
  - Total Investment Expenditures:
    - Elasticity Estimate: 0.030
    - G/Y dY/dG: 0.121
    - Multiplier Estimate: 0.25
    - 95% Confidence Interval: [-0.07; 0.57]
  - Local Government Expenditures:
    - Elasticity Estimate: 0.067
    - G/Y dY/dG: 0.138
    - Multiplier Estimate: 0.49
    - 95% Confidence Interval: [0.18; 0.80]
- Dynamic panel first-difference level estimates — Appendix Table 12:
  - Estimated dY/dG Multiplier — Total Investment Expenditures: 0.66; 95% Confidence Interval: [0.29; 1.04]
  - Estimated dY/dG Multiplier — Local Government Expenditures: 0.58; 95% Confidence Interval: [0.12; 0.99]
- Qualitative summary: panel VAR estimates similar to dynamic panel regressions; on average government expenditure multiplier during the 1990s was not significantly larger than 1.

*Source: _wp10110 - 1. Government Investment Elasticities (IMF working paper excerpt).*

### 1. Government Investment Elasticities ..................................................................................

### 1. Government Investment Elasticities

### Background and context
- Japan experienced a sharp slowdown in the 1990s: average growth declined to 1.2 percent from an earlier average of 3.9 percent.
- Between 1989 and 1991 the stock market declined by 41 percent and credit flow declined by 71 percent.
- Fifteen spending stimulus packages were implemented between 1990 and 2008.
- Aggregate announced stimulus components (about 28 percent of GDP in total) included:
  - public works and social infrastructure related projects, including land acquisition (14.2 percent of 2000 GDP);
  - credit guarantees and augmentation of credit lines to banks for loans to small and medium-sized enterprises and for the housing sector (8.5 percent of GDP);
  - employment assistance and cash transfers (2.1 percent of GDP);
  - tax measures (3.3 percent of GDP).
- Local governments:
  - comprise 60 percent of spending and an even larger share of public investment;
  - implemented about 80 percent of public works projects and financed about 65 percent of them during the nineties.
- Structural balances deteriorated from a surplus of about 1/5 percent of GDP in 1990 to a deficit of 6 percent of GDP in 2000. Main contributors to the increase in the fiscal deficit were:
  - declining taxes (3 percentage points),
  - increases in social security costs (3½ percentage points),
  - the remaining 1 percent of GDP due to government spending on land and capital transfers.

### Literature evidence on multipliers (select findings)
- VAR-based short-run spending multiplier estimates for Japan have ranged from 0.4 to 0.7.
- Kuttner and Posen (2002), using a structural VAR identification, find a cumulative four-year spending multiplier of 2.0.
- Hida and others (2008) find the public investment multiplier peaks at 1.1 at a two-year horizon.
- Several studies document heterogeneity over time and across types of spending; some find declining multipliers over time.

### Channels and constraints affecting multiplier size (as discussed)
- Factors supporting limited crowding out:
  - High gross government debt but low net debt by industrial-country standards.
  - Debt predominantly domestically held by long-term institutional investors, containing default and inflation risks.
  - Low short-term interest rates, though long-term rates increased, indicating some crowding out.
- Factors reducing fiscal effectiveness:
  - Credit crunch and deleveraging by firms limiting private investment response.
  - An appreciating yen limiting investment growth.
  - Composition effects from an aging population shifting spending toward social security with lower multipliers.
- Efficiency and allocation concerns:
  - Public investment often allocated toward social infrastructure in rural areas (e.g., rural roads, agriculture) with lower marginal productivity than urban projects.
  - Political economy factors and local special interest groups influenced allocation, sometimes favoring rural areas over metropolitan centers.

### Empirical approach used in this study
- Data: annual time-series at the prefecture level, 1990–2000, with detailed components of general government public investment and local government expenditures.
- Estimation model (log-levels, within-prefecture variation):
  - Yc,t = a_c + d_t + p * F_c,t + X_c,t + u_c,t
  - Yc,t: real per capita value added of prefecture c in year t (logs)
  - F_c,t: government expenditure of prefecture c in year t (logs)
  - X_c,t: vector of control variables at prefecture-year level
  - a_c: prefecture fixed effects; d_t: year fixed effects
- Identification features:
  - Year fixed effects control for national shocks including aggregate monetary policy, mitigating bias from common shocks.
  - Cyclically adjusted fiscal variables follow Blanchard and Perotti (2002).
  - Endogeneity addressed by treating government expenditure as endogenous and using SYS-GMM (Blundell and Bond, 1998), instrumenting levels with lagged first-differences.
  - Note on converting elasticity to absolute multiplier: multiply elasticity estimate by the inverse of the share of government expenditures in GDP; sample shares are 0.14 for local government expenditures and 0.12 for general government investment.

### Key empirical result: government investment elasticity
- Baseline SYS-GMM estimate (general government public investment elasticity):
  - elasticity = 0.05 (statistically significant at the 1 percent level).

*Source: _wp10110 - 1. Government Investment Elasticities (IMF working paper excerpt).*

### 0.2 percent. The inter-quartile range of the local government deficit, as a share of GDP, is 0.1 percent to

### _wp10110 - 0.2 percent. The inter-quartile range of the local government deficit, as a share of GDP, is 0.1 percent to

### Key finding
- Local government deficit (as a share of GDP): 0.2 percent.

### Distributional detail
- Inter-quartile range of the local government deficit (as a share of GDP): 0.1 percent to

*Source: _wp10110 - 0.2 percent. The inter-quartile range of the local government deficit, as a share of GDP, is 0.1 percent to*

### 0.3 percent.

### _wp10110 - 0.3 percent.

### Government investment elasticities (Table 1)
- Baseline elasticity estimates for government investment range from 0.051*** to 0.030** across specifications; preferred estimate reported is 0.033*** (standard error (0.013)).
- Lagged GDP coefficient ranges reported (examples): 0.732*** (0.039) to 0.581*** (0.040); final specification shows 0.588*** (0.040).
- Tax revenue coefficients become significant negative in some specifications: -0.138*** and -0.130*** (standard errors (0.023) and (0.025)).
- Private investment enters positively and significantly in extended specifications: 0.097*** (0.026), 0.100*** (0.026), 0.068*** (0.023), 0.064*** (0.021).
- Exports positive and significant in some specifications: 0.038* (0.020) and 0.043** (0.021).
- Government investment externality (sum of other prefectures’ investment) coefficient reported as 0.006 (0.040) and is not statistically significant.
- Estimation method: SYS-GMM (Blundell and Bond, 1998) using instruments dating 2 years and further back. AR(2) p-values and Hansen J p-values reported across columns (e.g., AR(2) p-value 0.558, Hansen J p-value 0.782 in column (1)). Prefecture and year fixed effects included. Observations up to 470 (410 in some columns).

### Public investment multiplier estimates (section B)
- Impact multiplier (regional) calculated: 0.28.
  - Calculation noted: Multiplier (0.28) = Elasticity (0.033) / Public Investment to GDP ratio (0.12).
- Medium-term cumulative multiplier for a fully persistent shock: 0.67.
  - Calculation noted: Medium-term multiplier (0.67) = Multiplier (0.28) / (1 - coefficient on lagged dependent variable (0.588)).

### Multipliers by type of investment and administrative level (Table 2)
- By type — elasticity (standard error) and implied multiplier:
  - Livelihood: Elasticity 0.018* (0.011); Multiplier 0.35. (Note: multiply by 1/(0.053) to obtain dY/dG on expenditure scale.)
  - Industry: Elasticity 0.003 (0.005); Multiplier 0.15. (Multiply by 1/(0.026).)
  - Agriculture: Elasticity 0.029** (0.012); Multiplier 2.04. (Multiply by 1/(0.015).)
  - Land Conservation: Elasticity 0.011 (0.010); Multiplier 1.15. (Multiply by 1/(0.012).)
  - Other: Elasticity 0.005 (0.004); Multiplier 0.41. (Multiply by 1/(0.012).)
- By administrative level — elasticity (standard error) and implied multiplier:
  - Central Government: Elasticity 0.008 (0.008); Multiplier 0.26. (Multiply by 1/(0.040).)
  - Prefecture Government: Elasticity 0.003 (0.012); Multiplier 0.14. (Multiply by 1/(0.036).)
  - City Government: Elasticity 0.033*** (0.012); Multiplier 0.78. (Multiply by 1/(0.045).)
- Observations: 410. Prefecture and year fixed effects included. Method: SYS-GMM.

### Local government expenditure multipliers (Table 3)
- Aggregate local government expenditures:
  - Elasticity estimate: 0.036* (0.02). Implied multiplier: 0.26.
  - Implied cumulative multiplier reported about 0.53.
- Component-level elasticities and implied multipliers:
  - Transfers to Firms: Elasticity 0.045** (0.02); Implied multiplier 3.46. (Multiply by 1/(0.013).)
  - Ordinary Construction: Elasticity 0.036*** (0.01); Implied multiplier 0.76. (Multiply by 1/(0.047).)
  - Social Assistance: Elasticity -0.001 (0.01); Implied multiplier -0.25. (Multiply by 1/(0.004).)
  - Government Personnel: Elasticity -0.011 (0.07); Implied multiplier -0.28. (Multiply by 1/(0.039).)
- Observations: 410. AR(2) and Hansen J p-values indicate instrument validity (examples: AR(2) p-value 0.219, Hansen J p-value 0.983 for aggregate). Method: SYS-GMM.

### Robustness analysis (section D)
- Panel VAR (one lag): public investment multiplier ≈ 0.24; local government expenditure multiplier ≈ 0.49 (Appendix Table 11).
- Level-differences specification (direct multiplier):
  - Public investment multiplier: 0.67 with 95 percent confidence interval band 0.3 to 1.04 (Appendix Table 12).
  - Local government spending multiplier: 0.58 with 95 percent confidence interval band .12 and .99.
- Quadratic term in log specification: quadratic coefficient positive and significant for both public investment and local government expenditure; sample-average multipliers estimated 0.26 (public investment) and 0.21 (local government expenditure).
- Interaction with output gap: interaction term negative (consistent with smaller multipliers when output gap positive) but not statistically significant.

### Crowding-out effects and private sector responses (Table 5, section A)
- Effects of total government investment:
  - On private investment: coefficient 0.219 (0.144) — positive but statistically insignificant.
  - On private consumption: coefficient -0.099* (0.06) — a 1 percent increase in government investment decreased private consumption by 0.1 percent.
- Effects of local government expenditures:
  - On private investment: coefficient -0.518*** (0.195) — a 10 percent increase in local government expenditures decreased private investment by more than 5.2 percent on average.
  - On private consumption: coefficient 0.063 (0.192) — positive but statistically insignificant.
- Controls: prefecture and year fixed effects; additional control variables include tax revenue, public debt, exports, employment, output, and private investment where appropriate. Method: SYS-GMM.

### Regional allocation, heterogeneity, and equity vs. efficiency (section B)
- Prefecture-specific multipliers estimated by interacting government investment with prefecture fixed effects; kernel density of 47 prefecture-specific multipliers shows substantial dispersion (Figure 4).
- Nonparametric local polynomial estimates:
  - Relationship between public investment per capita (1990–2000) and prefecture-specific multiplier: upward-sloping (Figure 5) — more investment went to regions with higher multipliers.
  - Relationship between prefecture GDP per capita (1990–2000) and investment per capita: downward-sloping (Figure 6) — evidence of equity motive (more investment to poorer regions).
- Interpretation: allocation reflects equity motives but does not appear to explain low aggregate multipliers through misallocation to low-multiplier regions; tradeoffs with multiplier effects seem limited.

### Marginal productivity of public capital and evolution of multipliers (section C)
- Production function estimates (Table 6):
  - Imposing constant returns to scale (column (1)): Private capital elasticity 0.283** (0.13); Public capital elasticity 0.323*** (0.10). Observations 368.
  - Flexible functional form (column (2)): Private capital elasticity 0.580*** (0.18); Public capital elasticity 0.461*** (0.13). Observations 368.
- Marginal product of public capital constructed as elasticity × output-to-public-capital ratio shows a declining trend for 1975–2000 with strong declines in 1975–85 and 1990–2000 (Figure 7).
- Evolution of government expenditure multiplier over time (Table 7):
  - 1975–1989: Government expenditures coefficient 0.102*** (0.03); Implied multiplier 0.8. Observations 690.
  - 1990–2000: Government expenditures coefficient 0.087*** (0.02); Implied multiplier 0.6. Observations 470.
  - Interpretation: government expenditure multiplier about 20 to 25 percent larger in 1975–1989 relative to 1990–2000; decline consistent with lower marginal product of public capital.

### Other factors discussed
- Duration of shocks matters: permanent fiscal changes have larger effects than transitory ones.
- Financing method matters: distortionary tax financing can reduce output while deficit financing may increase output; context-specific implications discussed for Japan (large deficits and tax cuts in the period).
- Zero lower bound: multiplier largest when monetary policy constrained; in Japan nominal interest rate around 8 percent at start of 1990s and reached zero bound only towards end of decade.

### Conclusions and policy-relevant implications (section VI)
- Baseline finding: investment multipliers in Japan are low and have declined over time but are higher than multipliers for public consumption.
- Local (city) government investment multipliers are higher than central and prefecture government multipliers; decentralized investment appears more effective.
- Multipliers vary substantially by type of investment: agriculture and certain local “livelihood” investments show higher multipliers than industry-focused (infrastructure) investment in these estimates.
- Factors contributing to low multipliers include:
  - Evidence of crowding-out effects, particularly for local government spending on private investment.
  - Declining marginal productivity of public capital due to overinvestment and large existing public capital stock, contributing to lower multipliers over time.
  - Allocation patterns motivated by equity did not evidently worsen aggregate multipliers materially.
- Policy considerations: prioritize forms and administrative levels of public investment that yield higher multipliers (e.g., city-level “livelihood” projects) and take into account declining marginal returns to public capital when designing stimulus via public investment.

*Source: Fund staff estimates.*

### Appendix I: Fiscal Stimulus Packages in Japan since 1990s

### Appendix I: Fiscal Stimulus Packages in Japan since 1990s

### Fiscal stimulus packages (1990–2008) — aggregate figures and components
- Appendix Table 8 reports detailed stimulus-package line items and magnitudes for packages between 1992 and 2008, with a total summary column showing:
  - Total size of stimulus package: 205.5
  - Total size/GDP (%) (average): 2.2, 2.7, 1.2, 3.1, 1.4, 2.9, 3.3, 4.7, 3.6, 2.2, 1.2, 0.9, 3.1, 2.2, 5.1, 2.3
- Components (selected aggregates preserved exactly as in the source):
  - Tax cut: 0.2. Note entries include "Ad hoc personal income tax cut", "Permanent tax cut in personal and corporate income tax", and "Other tax cuts included in the stimulus package".
  - Cash transfer to households: 0.7, 2.0, 2.7 (appearing in package breakdowns).
  - Government investments (Ig) to build up social infrastructure: 6.3, 7.2, 2.0, 3.7, 5.4, 9.1, 7.7, 8.1, 6.8, 5.2, 4.2, 3.4, 1.9, 2.2, 73.1 (sequence as listed in table).
  - Public works involving central government: 4.5, 5.6, 1.5, 3.4, 6.7, 1.6, 3.6, 2.9, 2.6, 2.3, 2.6, 1.0, 38.2 (as listed).
  - Public works by local governments: 1.8, 1.6, 0.5, 0.3, 1.0, 1.5, 6.7 (as listed).
  - Science and technology: 0.3, 11.1, 1.2, 11.2, 0.3, 0.2, 0.1, 6.4 (as listed).
  - Education and social welfare: 11.1, 0.6, 0.5, 0.7, 0.5, 0.7, 5.1 (as listed).
  - Alternative energy and environment: 1.6, 10.6, 0.6, 3.8 (as listed).
  - Natural disaster relief: 5.1, 1.4, 11.3, 1.6, 0.5, 1.7, 0.4, 13.0 (as listed).
  - Other government measures: 4.5, 5.8, 4.0, 5.7, 1.5, 5.2, 4.4, 9.1, 1.3, 5.8, 5.8, 0.0, 11.8, 9.6, 22.7, 107.1 (as listed).
    - Acquisition of land for public use: 1.6, 1.6, 0.3, 2.3, 3.2, 1.1, 10.1 (as listed).
    - Employment support: 0.0, 0.0, 0.0, 0.1, 1.0, 1.0, 1.3, 1.3, 0.9, 0.3, 5.9 (as listed).
    - Expansion of policy lending for housing sector: 0.8, 1.8, 2.9, 1.2, 0.5, 1.2, 10.4 (as listed).
    - Expansion of policy lending and government guarantees for non-financial sector (small and medium size businesses): 1.2, 1.9, 0.8, 1.4, 1.4, 1.4, 2.0, 5.9, 7.4, 4.5, 4.5, 10.9, 9.1, 21.8, 74.2 (as listed).
    - Others: 0.9, 0.5, 0.8, 0.1, 1.2, 0.9, 0.5, 0.6, 6.5 (as listed).
- Aggregate of key stimulus components:
  - (Ig + Tax cuts + Cash transfer): 98.4 (aggregate column entry).
  - (Ig + Tax cuts + Cash transfer) / GDP (%) (average): 1.3, 1.5, 0.4, 2.0, 0.4, 1.1, 1.8, 0.4, 0.4, 2.4, 2.9, 1.4, 1.0, 0.0, 0.9, 0.7, 0.4, 0.8, 1.1 (sequence as listed).
- Nominal GDP (series as reported): 483.8, 480.7, 480.7, 487.0, 496.5, 496.5, 496.5, 508.4, 513.3, 503.3, 503.3, 499.5, 504.1, 493.6, 493.6, 489.9, 526.9, 526.9
- Central government bond issuance in supplementary budgets (series as reported): 2.3, 2.2, 3.6, 2.2, 2.8, 4.7, 6.1, 2.3, 7.6, 2.0, 1.7, 0.0, 5.0, 0.4, n.a., 5.2, 2.9

Notes included in the table (verbatim):
- Source: Nakagawa (2009).
- 1/ There was an economic package in June 1999 to boost employment by 700 thousand jobs by deregulations and so on, involving almost no additional budgetary outlays. Therefore, this package is not listed in the table.
- 2/ Nominal GDP for 2008 is an estimation by the Japanese government.
- 3/ Tax cuts announced in the 1998 November stimulus package was implemented from FY1999 and lifted in FY2007.

### Cross-sectional correlations (within-prefecture)
- Appendix Table 9 — Correlation matrix (values preserved exactly):
  - GDP to: Gov. Investment 0.4098; Gov. Tax Revenue 0.6444; Government Debt -0.0835; Exports 0.2873; Private Investment 0.3282; Employment 0.0942
  - Gov. Investment to: Gov. Tax Revenue 0.4314; Government Debt -0.0390; Exports 0.1349; Private Investment 0.3450; Employment -0.1191
  - Gov. Tax Revenue to: Government Debt -0.0888; Exports 0.2470; Private Investment 0.3860; Employment 0.0016
  - Government Debt to: Exports 0.0028; Private Investment -0.0096; Employment 0.0127
  - Exports to: Private Investment 0.1245; Employment -0.0317
  - Private Investment to: Employment -0.0726
- Source: Authors’ calculations.

### Appendix II: Data description and investment classification

### Data scope and sources
- Annual prefecture-level data for 47 Japanese prefectures during the 1990–2000 period were obtained from the Japan Statistical Yearbook.
- Government investment data are obtained from the annual report on administrative investment from the Ministry of Home Affairs; the investment data are expenditure-based and cover:
  - maintenance and repair of facilities,
  - improvement projects (including cost of land and compensation),
  - office expenses,
  - planning and surveys.
- Local government finance data (government expenditures and government tax revenues) are from the annual statistical report on local government finance of the Ministry of Home Affairs and are based on reports submitted by local public bodies.
- Prefecture-level GDP, export, employment, and investment data are also from the Japan Statistical Yearbook.
- Appendix Table 10 provides the detailed description and classification of investment measures used in the empirical analysis.

### Investment categories and shares (exact labels and shares)
- Livelihood Investment (44% of total investment)
  - city, town and village roads
  - streets
  - city planning
  - housing
  - environment sanitation
  - welfare (including works of hospitals)
  - educational facilities
  - water supplies
  - Source: Japan Statistical Yearbook
- Industry Investment (22% of total investment)
  - national highways and prefectural roads
  - harbors
  - airports
  - industrial water
  - Source: Japan Statistical Yearbook
- Agricultural Investment (12% of total investment)
  - agriculture, forestry, fishery
  - Source: Japan Statistical Yearbook
- Land Conservation Investment (10% of total investment)
  - forest and river conservation
  - seashore conservation
  - Source: Japan Statistical Yearbook
- Other Investment (12% of total investment)
  - unemployment measures
  - disaster restoration
  - government office repairs
  - railways
  - subways electricity
  - gas
  - residential land formation
  - other
  - Source: Japan Statistical Yearbook

### Appendix III: Alternative estimation strategies and multiplier estimates

### Panel VAR estimation (log-log specification)
- Model specification:
  - One-lag panel VAR with six variables: GDP, government expenditures, tax revenues, private investment, employment, and exports.
  - Estimation by Generalized Methods of Moments using programs written by Love and Ziccino (2006).
- Key qualitative result:
  - "the panel VAR estimates are similar to the estimates obtained from the dynamic panel regressions" and "the panel VAR estimates confirm that on average the government expenditure multiplier during the 1990s was not significantly larger than 1."
- Appendix Table 11 — Panel VAR estimates of the Government Spending Multiplier (values preserved exactly):
  - Total Investment Expenditures
    - Elasticity Estimate: 0.030
    - G/Y dY/dG: 0.121
    - Multiplier Estimate: 0.25
    - 95% Confidence Interval: [-0.07; 0.57]
  - Local Government Expenditures
    - Elasticity Estimate: 0.067
    - G/Y dY/dG: 0.138
    - Multiplier Estimate: 0.49
    - 95% Confidence Interval: [0.18; 0.80]
- Source: Authors’ calculations.

### Dynamic panel first-difference level estimation (levels specification)
- Model specification:
  - Equation in first differences: ΔYc,t = φ ΔYc,t-1 + dt + p * ΔGc,t + DT ΔXc,t + Δec,t
  - All variables specified in levels (not logs); dY/dG multiplier equals estimated coefficient p.
  - First-differences used because levels of the series indicated unit-root dynamics.
- Appendix Table 12 — Dynamic Panel First-Difference Level Estimates of the Government Spending Multiplier (values preserved exactly):
  - Estimated dY/dG Multiplier — Total Investment Expenditures: 0.66; 95% Confidence Interval: [0.29; 1.04]
  - Estimated dY/dG Multiplier — Local Government Expenditures: 0.58; 95% Confidence Interval: [0.12; 0.99]
- Source: Authors’ calculations.

*Source: _wp10110 - Appendix I: Fiscal Stimulus Packages in Japan since 1990s (IMF PDF)._

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10110.pdf_