## 1jpnea2023002

## Source details

**Canonical URL:** [1jpnea2023002](https://www.imf.org/-/media/files/publications/cr/2023/english/1jpnea2023002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2023/english/1jpnea2023002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2023/english/1jpnea2023002.pdf.json)

---

### Corporate Savings and Cash Holdings in Japan — Drivers, Evidence, and Policy
- Stylized context:
  - Corporate savings in Japan have risen since 2000 in line with profits; a large share of additional savings was kept as cash holdings (cash and short-term investments) rather than used for investment.
  - Firms deleveraged: corporate leverage ratio declined from the peak of 45 percent of total assets to about 30 percent; after leverage stabilized around 2004, firms began to accumulate cash and other liquid assets.
  - Established drivers in literature: precautionary motives under funding uncertainty, FDI that is hard to collateralize, and weaker corporate governance in Japan.
- Additional drivers tested (firm-level, public and private samples):
  - Rising share of intangible capital (intangible capital is hard to collateralize).
  - Declining corporate tax rates (statutory corporate tax rate declined from 40 percent in 2004 to around 30 percent in 2016).
- Data and methods:
  - Public firms: Compustat-Global; private firms: Orbis (cleaned Orbis database from Diez et al. 2021).
  - Sample covers around 70 percent of firms surveyed in the Financial Statements Statistics of Corporations by Industry (FSSCI); Orbis coverage increases over time starting from 2003.
  - Cash holdings proxied by Cash and Short-term Investments (CHE) in Compustat-Global and Cash and Cash Equivalent in Orbis (share of total assets).
  - Intangible capital estimated via perpetual inventory: intangible investment = R&D investment + 30 percent of selling and administrative expenses; book intangible assets added in robustness checks.
  - Effective tax rate (ETR) calculated from tax payments and pre-tax income; ETR used alongside statutory rates.
  - Fixed-effect panel regressions estimated separately for public and private firms; standard controls included; intangible share and tax rates lagged by one year; variables winsorized at 1 percent and 99 percent.
- Key empirical magnitudes:
  - Intangible share association with cash/total assets:
    - Public firms: 1 percentage point increase in intangible share → about 0.24 percentage point increase in cash / total assets.
    - Private firms: 1 percentage point increase in intangible share → about 0.27 percentage point increase in cash / total assets.
    - Regression coefficients (selected):
      - Public firms (Lagged) intangible share = 0.232***.
      - Private firms (Lagged) intangible share = 0.256***.
  - Effective tax rate association with cash/total assets:
    - Negative relationship: lower effective tax rates associated with higher cash holdings.
    - Regression coefficients (selected):
      - Public firms (Lagged) effective tax rate = -0.025***.
      - Private firms (Lagged) effective tax rate = -0.002***.
    - Effect of tax cuts stronger among public firms; private-firm effect noisier and subject to attenuation bias (excluding tax rate outliers (<20% or >80%) increases coefficient magnitude).
    - Auxiliary regression: tax cuts do not boost investment.
  - Decomposition of rise in cash holdings since 2004:
    - Public firms: average cash holdings increased by about 7.5 percentage points since 2004, of which:
      - 1.7 percentage points explained by rising intangible share.
      - 0.3 percentage points explained by lower tax rates.
    - Private firms: average cash holdings increased by about 5.4 percentage points since 2004, of which:
      - 1.5 percentage points explained by rising intangible share.
  - Selected regression statistics:
    - Public firms (Table 1, column 1): (Lagged) intangible share = 0.232*** (standard error 0.018); (Lagged) effective tax rate = -0.025*** (standard error 0.003); Cash flow volatility = 0.821*** (standard error 0.267); log(Total assets) = 0.017*** (standard error 0.005); Capex / Total assets = -0.337*** (standard error 0.031); Debt / Total assets = -0.212*** (standard error 0.021); Adj. R-squared = 0.835; Firm FE = Yes; Year FE = No; N = 42982 (Cash / Total assets).
    - Private firms (Table 2, column 1): (Lagged) intangible share = 0.256*** (standard error 0.008); (Lagged) effective tax rate = -0.002*** (standard error 0.001); Cash flow volatility = 0.119* (standard error 0.064); log(Total assets) = 0.020*** (standard error 0.005); Capex / Total assets = -0.273*** (standard error 0.009); Debt / Total assets = -0.061*** (standard error 0.007); Adj. R-squared = 0.828; Firm FE = Yes; Year FE = No; N = 49795 (Cash / Total assets).
- Robustness and interpretation:
  - Adding book value of intangible assets lowers but does not eliminate the intangible coefficient significance.
  - Statutory tax rates are statistically significant and imply larger effects than ETRs, suggesting attenuation bias in ETRs.
- Policy implications and recommendations:
  - Reduce financing frictions that raise precautionary cash holdings, especially for intangible-intensive sectors:
    - Develop venture capital markets to support start-ups and intangible-intensive firms.
    - Government support for high-growth SMEs.
    - Expand well-designed R&D tax incentives for larger firms.
  - Financial market reforms favoring intangible-intensive projects:
    - Promote cash flow-based lending (CBL) relying on operating profits rather than physical collateral; CBL could relieve collateral constraints for intangible-heavy firms.
    - Recognize adoption challenges for CBL in Japan due to rarity of corporate reorganization; complementary reforms could include improving labor flexibility and developing venture capital/private equity markets.
  - Use of intellectual property (IP)-backed loans:
    - IP-protected assets (patents, trademarks, copyrights, design rights) can serve as collateral.
    - The Japan Patent Office (JPO) provides IP Business Valuation reports to banks by request.
    - Prudential regulation that treats IP-backed loans as riskier and does not count them fully toward regulatory capital may deter development of IP-backed lending.
    - Possible regulatory responses:
      - Revise Basel III regulations to account for financing needs of intangible-intensive firms while preserving financial stability (requires international cooperation and is complex/lengthy).
      - Allow banks to transfer part of loan risk to insurance companies (example: British Business Bank investigation).
  - Re-evaluate broad-based corporate tax cuts:
    - Corporate tax cuts improved profits but mostly increased corporate cash holdings rather than investment or wages.
    - Given significant fiscal cost and high public debt, broad-based corporate tax cuts should be reconsidered for efficiency.

### A New Growth Engine for Japan: Women in STEM — Model, Results, and Policies
- Motivation and key findings:
  - Female labor force participation rose by 10 percentage points in the last 10 years, approaching the highest level among G7 countries.
  - Share of women enrolled in STEM fields in university in Japan is around 7 percent (lowest among G7).
  - Bridging the gender gap in STEM fields can boost TFP growth by 20 percent and consumption-equivalent welfare by 4 percent in Japan.
  - Innovation appears to be slowing in Japan, with patent grants declining since 2012; tapping underutilized female STEM talent is a promising channel to raise potential growth amid demographic headwinds.
- Model and calibration:
  - Endogenous growth model with quality ladders and heterogeneous STEM (research) talent drawn from a Pareto distribution; STEM workers = researchers.
  - Two discrimination channels:
    - Schooling bias: uniform discount factor on women’s research talent (lowers productivity).
    - Labor market discrimination: uniform discount factor on female researchers’ wages (no effect on productivity).
  - Baseline calibration assumes all misallocation comes from labor market discrimination (conservative); robustness includes half-distortion from schooling bias.
  - Parameters (as reported):
    - α: 1-labor share — PWT10, 0.4
    - ρ: Discount rate — 0.02
    - γ: Risk Aversion (CRRA) — Hall (2009), 2.0
    - θ: Pareto shape parameter — Basic Survey on Wage Structure, 3.5
    - η = 0.024: TFP growth rate — PWT10
    - λ = 2.4: % of STEM workers among male: 10% — Basic Survey on Wage Structure
    - δ = 0.4: % of STEM workers among female: 2% — Basic Survey on Wage Structure
  - Equilibrium: agents choose worker vs researcher given wages and talent; researcher cutoff ε* determined by wage comparisons; aggregate idea arrival rate z(t) = η ⋅ HR(t).
- Quantitative results:
  - Removing discrimination against female researchers increases total number of researchers by about 20 percent.
  - TFP growth:
    - Baseline: TFP growth rises by 20 percent under removal of barriers to female STEM participation.
    - If half of distortion comes from schooling bias, TFP growth rises by 26 percent.
  - Welfare (consumption-equivalent, CE):
    - Average consumption-equivalent welfare rises by about 4 percent under removal of barriers.
    - Welfare gains from a 10-percentage point increase in female LFPR are about 3.5 percent given current labor market structure.
  - Distributional outcomes:
    - Old female researchers: CE Welfare 12.5 [4.2~12.5]
    - New female researchers: -5.3 [-5.3,4.2]
    - Remaining male researchers: 4.2
    - Male researchers can be worse off due to crowding-out; female researchers gain substantially.
    - Workers (non-researchers) see welfare improvement around 4.2 percent.
    - If schooling barriers are also removed (half of distortion), average welfare rises by 4.5 percent.
- Policy implications and recommendations:
  - Address financial and non-financial barriers to female STEM participation:
    - Reduce explicit pay gaps: increase wage transparency and reduce pay gaps through policy and public-sector leadership.
    - Tackle implicit pay gaps and workplace constraints:
      - Family care burdens: 30 percent of women quit their jobs when they give birth.
      - Maternity harassment: one out of every five pregnant full-time working women have experienced maternity harassment (Japan Institute for Labour Policy and Training).
      - Promote paternity leave uptake (government target to double number of men taking paternity leave by 2025); consider mandatory leave policies; reduce working hours; adopt flexible work arrangements (teleworking); reform hiring and promotion toward merit-based systems rather than seniority.
    - Schooling, culture, and stereotypes:
      - Counter gender stereotyping from early education onward; training for teachers and parents.
      - Support female scientists through targeted research grants and special quotas at public universities.
      - Promote role models and mentoring.
  - Quotas: presented as a corrective measure where gender bias exists in STEM.

### Box 2 — Japan: Consumption-Equivalent Welfare (Definition and Implications)
- Definition and measurement:
  - Welfare change ω(ε) defined as extra consumption making a consumer indifferent between new and old equilibrium following Lucas (1987).
  - Welfare equality condition (as presented):
    - ∫0∞ e−ρt ⋅ [ (1 + ω(ε)) ⋅ cto(t)(ε) ](1−γ)/(1−γ) dt = ∫0∞ e−ρt ⋅ [ cno(t)(ε) ](1−γ)/(1−γ) dt
  - Average welfare change ω derived from aggregate integrals across talent distribution f(ε).
- Key implications:
  - Welfare changes individualized: ω depends on research talent ε and gender.
  - Average welfare change aggregates individual utilities via the provided integral expressions.
  - Box frames welfare in consumption-equivalent terms and embeds heterogeneity via f(ε).

### High Working-Age Poverty and the Case for an Earned Income Tax Credit (EITC)
- Background and problem statement:
  - Japan’s unemployment rate remains relatively low compared to other OECD countries, while Japan’s poverty rate of the working-age population has risen since the 1980s and is one of the higher ones among OECD countries.
  - Working-age population: people between 18 and 65 years old.
  - Poverty rate defined as share with income below half of the median household income (from Comprehensive Survey of Living Conditions by MHLW, conducted once every three years).
- Link with non-regular employment:
  - Non-regular workers (fixed-term contract, part-time, dispatched) account for about 40 percent of total employment (IPSS, 2019).
  - Non-regular workers: unstable employment, low wages, poor development opportunities.
  - Average annual income of non-regular workers falls below 2 million JPY and is less than half of regular workers’ income.
  - Domestically, "working poor" understood as annual income less than 2 million JPY.
- Public assistance program features and shortcomings:
  - Public assistance (Public Assistance Act, 1950) comprises eight types; livelihood assistance is major program.
  - Assistance amount = minimum living expenses − household income; means-tested; assets generally must be sold unless exceptions apply; able persons required to use ability to work where adequate jobs exist.
  - Key statistics (as of July 2016):
    - About 1.6 million households received public assistance.
    - Elderly households take largest share (over 50 percent).
    - Households with a disabled or sick individual account for about 26 percent of recipients.
    - Single-mother households account for about 6 percent of recipients.
    - Average duration: about 10 percent of recipients have been on the program for less than a year; about 30 percent have been on the program for more than ten years.
  - Shortcomings: more than half of recipients are elderly; large share disabled or sick suggests inadequate income support for working poor; calculation method creates work disincentives (worker whose annual income is less than minimum living expenses may be induced not to work).
- Case for considering an EITC:
  - Four principal problems motivating EITC:
    1. Low unemployment rate but high poverty rate of the working-age population.
    2. Widespread non-regular employment.
    3. Limited income support provided by public assistance for working poor.
    4. Inherent work disincentives created by the public assistance program.
  - Additional motivations:
    - EITC programs widespread in advanced economies (US, UK long-standing; Germany, Denmark, Sweden examples).
    - As of 2009, 16 of 30 OECD countries operated EITC programs (Immervoll and Pearson, 2009).
    - EITC can address work disincentives in means-tested programs and support working poor.
    - Could help address aging by boosting employment and labor supply:
      - 2017 elderly (65+) accounted for about 27.6 percent of total population.
      - Share of elderly projected to reach about 38.4% in 2065.
    - EITC aligns with Prime Minister Fumio Kishida’s “New Form of Capitalism” objective.
- Theoretical rationale:
  - Extensions of Mirrlees (1971) show EITC can be part of optimal schedule when labor supply responds on intensive and extensive margins; two-dimensional informational asymmetries exist; individual failures present; or non-welfarist social welfare functions considered.
  - Trade-off: EITC increases employment by inducing non-workers to enter labor force; effect on wages split between workers and employers depends on labor demand elasticity; higher worker capture of EITC raises poverty alleviation but may lower employment impact.
- International experience and program scale:
  - Designs vary: tax concessions, refundable credits, eligibility tied to children, minimum hours, etc.; about half of programs restrict eligibility to working families with children.
  - Examples of scale:
    - US 2019: 25 million working households received about $63 billion; average benefit ~ $2,500.
    - UK 2016-17: 2.9 million working families received about £19.3 billion; average benefit ~ £6,700.
- Key policy issues and design choices:
  - Primary objective: address in-work poverty (short run) vs promote employment (long run); trade-offs exist.
  - Address existing anomalies before introducing EITC:
    - Labor taxes: high labor income taxes (payroll taxes) at low earnings counter EITC objectives; consider targeted tax reductions.
    - Out-of-work benefits: where eligibility is loose or levels generous, strengthen link between benefit levels and earnings; reassess public assistance eligibility (currently independent of employment status).
  - Design elements:
    - Definition of income: for poverty alleviation, use total household income; for promoting employment, individual earnings may be preferable.
    - Linking benefits to family structure: for poverty reduction, link to household income and family composition.
    - Working hours / earnings thresholds: thresholds strengthen incentives and reduce fiscal cost but increase administrative burden and may exclude vulnerable workers.
    - Frequency and timing of payments: frequent (monthly) payments can alleviate liquidity constraints; US experience showed preference for annual payments but informational constraints affect choices.
    - Treatment of self-employed: verification of income difficult; including self-employed important for poverty alleviation but requires measures to ensure tax compliance and prevent fraud.
- Implementation considerations:
  - Administrative capacity prerequisites:
    - High levels of tax filing among target population and strong personal income tax assessment system reduce administrative cost of EITC.
    - In Japan, most employees do not file annual tax returns because labor income taxes are withheld and adjusted by employers; investment income often withheld at source.
    - Information sharing among ministries, local governments, and public corporations is insufficient: National Tax Agency collects employer income info but coverage partial; local governments gather resident income data but not centrally aggregated.
    - Delivery of payments challenging because taxpayers do not register bank account information with National Tax Agency.
  - Administrative solutions proposed:
    - Make annual tax returns mandatory while providing pre-filled returns.
    - Utilize the My Number taxpayer identification system (launched 2016) to improve information collection, sharing, and management.
    - Link bank accounts to My Number for benefit delivery.
  - Outreach and take-up:
    - Ensure eligible individuals are aware of benefits and understand incentives via informational outreach and peer networks.
    - Simplify tax and benefit system to make interactions clear.
  - Monitoring and evaluation:
    - Allocate resources to monitor employment and wage outcomes across household and skill types pre- and post-introduction.

### Banks’ Exposure to Transition Risks and Policy Responses
- Banks’ exposure and sectoral lending:
  - Japanese G-SIBs’ exposure to energy and utilities is in the 6-8 percent range.
  - Most smaller banks have smaller exposures to these sectors.
  - Transition risk credit costs concentrated in high emission sectors: energy, power, steel, and automotive (Bank of Japan and Financial Services Agency (2022) scenario-based analysis).
  - That analysis suggests “the annual average of estimated transition risk credit costs [...] is considerably lower than the average annual net profits of each bank”.
- Firm-level vulnerability to carbon pricing (ICR dynamics):
  - With carbon taxes at $75 or $150 on scope 1 and 2 emissions, the share of firms with ICRs below 1 would increase by 7 percent and 14 percent, respectively.
  - Most exposed sectors: energy, utility, and materials (high direct emissions).
  - Downstream sectors significantly impacted via purchased power and electricity emissions.
  - There is a significant weak tail of firms following the pandemic that is not able to cover interest payments out of profits.
- Policy implications and recommendations:
  - Corporate-level actions:
    - Incorporate anticipated public policies requiring much lower emissions into business strategies.
    - Improve climate-related corporate disclosures.
  - Financial sector actions:
    - Banks should consider transition risks in lending to allocate capital in line with climate targets.
    - Efforts by Japanese policymakers and financial institutions to quantify and manage climate financial risks are appropriate.
  - Public sector actions:
    - Invest in decarbonization and support green technologies.
    - Provide predictable emission pricing pathways.
    - Support green and transition finance.
    - Provide targeted policies to help hard-to-abate and high-emission sectors where transition is particularly challenging.

*Prepared by Seho Kim (University of Maryland), Pablo Lopez-Murphy, and Rui Xu (both APD). Source: IMF staff compilation (chapter: Drivers of Corporate Cash Holdings in Japan) and IMF staff estimates and text from "A NEW GROWTH ENGINE FOR JAPAN: WOMEN IN STEM FIELDS"; additional chapter content as provided in 1jpnea2023002.*

### 1. Corporate Savings and Cash Holdings in Japan ______________________________________ 6

### Drivers of Corporate Cash Holdings in Japan

### Introduction
- Corporate savings in Japan have risen since 2000 in line with profits; a large share of additional savings was kept as cash holdings (cash and short-term investments) rather than used for investment.
- Households’ net savings declined before the pandemic while corporations moved from net borrowers to net savers after the 1990s asset bubble burst.
- Firms deleveraged: corporate leverage ratio declined from the peak of 45 percent of total assets to about 30 percent; after leverage stabilized around 2004, firms began to accumulate cash and other liquid assets.
- Established drivers in the literature include precautionary motives under funding uncertainty, FDI that is hard to collateralize, and weaker corporate governance in Japan.
- This study tests two additional drivers using firm-level data for publicly listed and private Japanese firms:
  - Rising share of intangible capital (motivated by literature showing intangible capital is hard to collateralize).
  - Declining corporate tax rates (statutory corporate tax rate declined from 40 percent in 2004 to around 30 percent in 2016).

### Data and Methodology
- Data sources:
  - Public firms: Compustat-Global.
  - Private firms: Orbis (cleaned Orbis database from Diez et al. 2021).
- Sample covers around 70 percent of firms surveyed in the Financial Statements Statistics of Corporations by Industry (FSSCI); Orbis coverage increases over time starting from 2003.
- Cash holdings defined as cash deposits plus other liquid assets (share of total assets), captured as Cash and Short-term Investments (CHE) in Compustat-Global and Cash and Cash Equivalent in Orbis.
- Intangible capital estimated by accumulating intangible investment using the perpetual inventory method; intangible investment = R&D investment + 30 percent of selling and administrative expenses. Book intangible assets are added in robustness checks in some specifications.
- Effective tax rate (ETR) calculated using tax payments and pre-tax income; ETR used in addition to statutory tax rates.
- Empirical approach:
  - Fixed-effect panel regressions estimated separately for public and private firm samples.
  - Standard controls: cash flow volatility, market-to-book ratio, total assets (log), cash flow / total assets, capital expenditures / total assets, dividend dummy, net working capital / total assets, leverage (debt / total assets).
  - Intangible share and tax rates lagged by one year; firm fixed effects included. Year fixed effects not included in tax rate analysis; recession dummy included instead.
  - Variables winsorized at 1 percent and 99 percent.

### Empirical Results — Stylized Findings and Magnitudes
- Association between intangible capital and cash holdings:
  - Positive relationship for both public and private firms (see Figure 2 stylized facts).
  - A 1 percentage point increase in intangible share is associated with about:
    - 0.24 percentage point increase in cash / total assets among public firms.
    - 0.27 percentage point increase in cash / total assets among private firms.
  - Regression coefficients (selected specifications):
    - Public firms: (Lagged) intangible share = 0.232*** (Table 1, column 1).
    - Private firms: (Lagged) intangible share = 0.256*** (Table 2, column 1).
- Association between effective tax rates and cash holdings:
  - Negative relationship: lower effective tax rates associated with higher cash holdings, implying firms saved tax reductions rather than investing them.
  - Regression coefficients (selected specifications):
    - Public firms: (Lagged) effective tax rate = -0.025*** (Table 1, column 1).
    - Private firms: (Lagged) effective tax rate = -0.002*** (Table 2, column 1).
  - Effect of tax cuts stronger among public firms; effect among private firms is noisier and subject to attenuation bias (excluding tax rate outliers (<20% or >80%) increases coefficient magnitude).
  - Auxiliary regression: tax cuts do not boost investment.
- Contribution to the rise in cash holdings since 2004 (decomposition):
  - Public firms: average cash holdings increased by about 7.5 percentage points since 2004, of which:
    - 1.7 percentage points explained by rising intangible share.
    - 0.3 percentage points explained by lower tax rates.
  - Private firms: average cash holdings increased by about 5.4 percentage points since 2004, of which:
    - 1.5 percentage points explained by rising intangible share.
- Robustness:
  - Adding book value of intangible assets lowers but does not eliminate the intangible coefficient significance.
  - Statutory tax rates are statistically significant and imply larger effects than ETRs, suggesting attenuation bias in ETRs.

### Policy Implications and Recommendations
- Reduce financing frictions that raise precautionary cash holdings, especially for intangible-intensive sectors:
  - Standard innovation policies:
    - Develop venture capital markets to support start-ups and intangible-intensive firms.
    - Government support for high-growth SMEs.
    - Expansion of well-designed R&D tax incentives for larger firms.
  - Financial market reforms favoring intangible-intensive projects:
    - Promote cash flow-based lending (CBL), which relies on firms’ operating profits rather than physical collateral; CBL could relieve collateral constraints for intangible-heavy firms.
    - Recognize challenges in adopting CBL in Japan due to rarity of corporate reorganization; complementary reforms could include improving labor flexibility and developing venture capital/private equity markets.
  - Use of intellectual property (IP)-backed loans:
    - IP-protected assets (patents, trademarks, copyrights, design rights) can serve as collateral.
    - The Japan Patent Office (JPO) provides IP Business Valuation reports to banks by request.
    - Prudential regulation that treats IP-backed loans as riskier and does not count them fully toward regulatory capital may deter development of IP-backed lending.
    - Possible regulatory responses:
      - Revise Basel III regulations to account for financing needs of intangible-intensive firms while preserving financial stability (requires international cooperation and is complex/lengthy).
      - Allow banks to transfer part of loan risk to insurance companies (example: British Business Bank investigation).
- Re-evaluate broad-based corporate tax cuts:
  - Corporate tax cuts improved profits but mostly increased corporate cash holdings rather than investment or wages.
  - Given significant fiscal cost and high public debt, broad-based corporate tax cuts should be reconsidered for efficiency.

### Key Regression and Sample Statistics (selected)
- Public firms (Compustat Global) regression highlights (Table 1, column 1):
  - (Lagged) intangible share = 0.232*** (standard error 0.018).
  - (Lagged) effective tax rate = -0.025*** (standard error 0.003).
  - Cash flow volatility = 0.821*** (standard error 0.267).
  - log(Total assets) = 0.017*** (standard error 0.005).
  - Capex / Total assets = -0.337*** (standard error 0.031).
  - Debt / Total assets = -0.212*** (standard error 0.021).
  - Adj. R-squared = 0.835; Firm FE = Yes; Year FE = No; N = 42982 (Cash / Total assets).
- Private firms (Orbis) regression highlights (Table 2, column 1):
  - (Lagged) intangible share = 0.256*** (standard error 0.008).
  - (Lagged) effective tax rate = -0.002*** (standard error 0.001).
  - Cash flow volatility = 0.119* (standard error 0.064).
  - log(Total assets) = 0.020*** (standard error 0.005).
  - Capex / Total assets = -0.273*** (standard error 0.009).
  - Debt / Total assets = -0.061*** (standard error 0.007).
  - Adj. R-squared = 0.828; Firm FE = Yes; Year FE = No; N = 49795 (Cash / Total assets).

*Prepared by Seho Kim (University of Maryland), Pablo Lopez-Murphy, and Rui Xu (both APD).*
*Source: IMF staff compilation (chapter: Drivers of Corporate Cash Holdings in Japan).*

### References

### A NEW GROWTH ENGINE FOR JAPAN: WOMEN IN STEM FIELDS

### Key findings and motivation
- Female labor force participation rose by 10 percentage points in the last 10 years, approaching the highest level among G7 countries.
- Share of women enrolled in STEM fields in university in Japan is around 7 percent (lowest among G7).
- Bridging the gender gap in STEM fields can boost TFP growth by 20 percent and consumption-equivalent welfare by 4 percent in Japan.
- Innovation appears to be slowing in Japan, with patent grants declining since 2012; tapping underutilized female STEM talent is a promising channel to raise potential growth amid demographic headwinds.

### Model and calibration
- Model: An endogenous growth model with quality ladders and heterogeneous STEM (research) talent drawn from a Pareto distribution. STEM workers = researchers; research increases machine quality and drives growth.
- Discrimination modeled in two ways:
  - Schooling bias: uniform discount factor on women’s research talent (lowers productivity).
  - Labor market discrimination: uniform discount factor on female researchers’ wages (no effect on productivity).
- Baseline calibration assumes all misallocation comes from labor market discrimination (conservative estimate). Robustness considers scenario where half of discrimination comes from schooling bias.
- Parameters set externally / independently / estimated:
  - α: 1-labor share — PWT10, 0.4
  - ρ: Discount rate — 0.02
  - γ: Risk Aversion (CRRA) — Hall (2009), 2.0
  - θ: Pareto shape parameter — Basic Survey on Wage Structure, 3.5
  - η = 0.024: TFP growth rate — PWT10
  - λ = 2.4: % of STEM workers among male: 10% — Basic Survey on Wage Structure
  - δ = 0.4: % of STEM workers among female: 2% — Basic Survey on Wage Structure
- Equilibrium features: agents choose worker vs researcher given wages and talent; researcher cutoff ε* determined by wage comparisons; aggregate idea arrival rate z(t) = η ⋅ HR(t).

### Quantitative results
- Removing discrimination against female researchers increases total number of researchers by about 20 percent (additional female researchers outnumber replaced male researchers).
- TFP growth:
  - Baseline: TFP growth rises by 20 percent under removal of barriers to female STEM participation.
  - If half of distortion comes from schooling (i.e., schooling barriers also removed), TFP growth rises by 26 percent.
- Welfare (consumption-equivalent, CE):
  - Average consumption-equivalent welfare rises by about 4 percent under removal of barriers.
  - For comparison: welfare gains from a 10-percentage point increase in female LFPR are about 3.5 percent given current labor market structure.
- Distributional welfare impacts (Table as reported):
  - Old female researchers: CE Welfare 12.5 [4.2~12.5]
  - New female researchers: -5.3 [-5.3,4.2]
  - Remaining male researchers: 4.2
  - Male researchers who dropped out: (values given in table; net effects depend on individual research talent)
  - (Table heading: Japan: Welfare Impact of Removing Barriers to Female STEM Workers — Source: IMF staff estimates.)
- Detailed outcomes:
  - Male researchers can be worse off: crowding-out effect from new female researchers can dominate growth benefits for male researchers (reported welfare loss of 5.3 percent for male researchers in text).
  - Female researchers gain substantially (reported welfare gain up to 12.5 percent).
  - Workers (non-researchers) see welfare improvement in line with faster TFP growth by about 4.2 percent.
  - If schooling barriers are also removed (half of distortion), average welfare rises by 4.5 percent.

### Policy implications and recommendations
- Address both financial and non-financial barriers to female STEM participation:
  - Explicit pay gaps: Japan’s gender pay gap is the third highest among OECD countries; increase wage transparency and reduce pay gaps through policy and public-sector leadership.
  - Implicit pay gaps and workplace constraints:
    - Family care burdens reduce female full-time employment: 30 percent of women quit their jobs when they give birth.
    - Maternity harassment: one out of every five pregnant full-time working women have experienced maternity harassment (Japan Institute for Labour Policy and Training).
    - Policy responses: promote paternity leave uptake (government target to double number of men taking paternity leave by 2025); consider mandatory leave policies; reduce working hours; adopt flexible work arrangements (teleworking); reform hiring and promotion toward merit-based systems rather than seniority to facilitate reentry of women into full-time employment.
  - Schooling, culture, and stereotypes:
    - Counter gender stereotyping from early education onward; training and workshops for teachers and parents.
    - Support female scientists through targeted research grants and special quotas at public universities to correct for gender bias (Tokyo Tech’s special quotas mentioned as an example).
    - Promote role models and mentoring: invite successful female scientists to schools; senior women mentor younger colleagues on balancing work and family.
- Quotas: In the presence of gender bias in STEM, quotas are presented as a corrective measure to improve diversity and representation.

### Mechanism and interpretation
- Discrimination raises the talent cutoff for female researchers; fewer women enter STEM, creating talent misallocation that reduces long-term growth by lowering aggregate research talent.
- Removing discrimination increases female researcher wages and incentivizes STEM entry, increasing aggregate research input even if unit wages fall and some marginal male researchers exit research.
- Schooling bias matters more for growth than wage discrimination because it reduces female researchers’ productivity directly.

*Italic: Source — IMF staff estimates and text from "A NEW GROWTH ENGINE FOR JAPAN: WOMEN IN STEM FIELDS" (References and chapter content as provided).*

### Box 2. Japan: Consumption-Equivalent Welfare

### Box 2. Japan: Consumption-Equivalent Welfare

### Definition and measurement
- As in Lucas (1987), an individual’s welfare change is defined as the amount of extra consumption 휔휔(휖휖) that a rational consumer would require in order to be indifferent between the new equilibrium (after removing the barriers to women in STEM) and the old equilibrium.
- 휔휔(휖휖) would differ across individuals depending on their research talent 휖휖 and gender.
- Welfare equality condition (as presented):
  - ∫0∞ 푒푒−휌휌푡푡 ⋅ [ (1 + 휔휔(휖휖)) ⋅ 푐푐푡푡표표표표표표(휖휖) ](1−훾훾)/(1−훾훾) 푑푑푡푡 = ∫0∞ 푒푒−휌휌푡푡 ⋅ [ 푐푐푡푡푛푛푛푛푛푛(휖휖) ](1−훾훾)/(1−훾훾) 푑푑푡푡
- The average welfare change 휔휔 is calculated using the total welfare of all individuals:
  - ∫0∞ 푒푒−휌휌푡푡 ⋅ ∫1∞ [ (1 + 휔휔) ⋅ 푐푐푡푡표표표표표표(휖휖) ](1−훾훾)/(1−훾훾) 푓푓(휖휖) 푑푑휖휖 ⋅ 푑푑푡푡 = ∫0∞ 푒푒−휌휌푡푡 ⋅ ∫1∞ [ 푐푐푡푡푛푛푛푛푛푛(휖휖) ](1−훾훾)/(1−훾훾) 푓푓(휖휖) 푑푑휖휖 ⋅ 푑푑푡푡

### Key implications from the box
- Welfare changes are individualized: 휔휔 depends on research talent 휖휖 and gender.
- Average welfare change 휔휔 is derived from aggregating individual utilities across the population using the provided integral expressions.
- The box frames welfare in consumption-equivalent terms following Lucas (1987) and embeds population heterogeneity via the talent distribution f(휖휖).

### References (as listed in the source)
- Bendels, M. H. K., Ruth Müller, Doerthe Brueggmann and David A. Groneberg. 2018. “Gender disparities in high-quality research revealed by Nature Index journals.” Plos One
- Griliches, Zvi. 1992. “The search for R&D spillovers.” Scandinavian Journal of Economics, 94: S29–S47
- Homma, M. K., R. Motohashi, and H. Ohtsubo. 2013. “Japan’s lagging gender equality.” Science 340(6131), 428-430
- Hsieh, C. T., E. Hurst, C. I. Jones, P. J. Klenow. 2019. “The allocation of talent and US economic growth.” Econometrica. 87(5):1439-74.
- Jones, C. I. 1995. “Time series tests of endogenous growth models.” Quarterly Journal of Economics 110, no. 2:495–525.
- Jones, C. I. 2002. “Sources of U.S. economic growth in a world of ideas.” American Economic Review 92(1):220–39.
- Lucas, R. E. 1987. “Models of Business Cycles.” Wiley-Blackwell.

*Source: 1jpnea2023002 - Box 2. Japan: Consumption-Equivalent Welfare (PDF chapter).*

### References

### References

### Content overview
- This section provides the bibliographic references cited in the chapter.
- It indicates scholarly and policy sources drawn on in the analysis, including working papers, journal articles, IMF reports, and discussion papers.
- The references support empirical methods (e.g., Nelson–Siegel state-space yield-curve models and panel estimations) and contextual background on monetary policy, bank profitability, and spillovers.

*Source: 1jpnea2023002 - References*

### 1.      The high poverty rate of the working-age population is a long-standing social and

### 1jpnea2023002 - 1.      The high poverty rate of the working-age population is a long-standing social and

### Background and problem statement
- Japan’s unemployment rate continues to remain relatively low compared to other OECD countries, while Japan’s poverty rate of the working-age population is one of the higher ones among OECD countries after a gradual rise since the 1980s.
- The working-age population refers to those people between 18 and 65 years old.
- The poverty rate is defined as the share of the people whose income falls below half of the median household income. Japan’s poverty rate is derived from the Comprehensive Survey of Living Conditions conducted once every three years by the Ministry of Health, Labour, and Welfare (MHLW) of Japan.

### Link between non-regular employment and working-age poverty
- The number of non-regular workers (fixed-term contract workers, part-time workers, and dispatched workers) has increased in recent decades, accounting for about 40 percent of total employment (IPSS, 2019).
- Non-regular workers face unstable employment, low wages, and poor development opportunities.
- The average annual income of non-regular workers falls below 2 million JPY and is less than half of that of regular workers.
- Domestically, the "working poor" in Japan is understood to be those whose annual income is less than 2 million JPY.

### Public assistance program: coverage and incentives
- The public assistance program is founded on the Public Assistance Act (1950) and comprises eight types of assistance: livelihood, education, housing, medical, long-term care, maternity, occupational, and funeral assistance; livelihood assistance is the major program.
- Assistance amount is calculated by subtracting household income from minimum living expenses (which vary by region, family structure, etc.); livelihood assistance is provided as cash transfers.
- Public assistance is means-tested; assets must generally be sold unless certain utilization conditions apply; able persons are required to use their ability to work where adequate jobs exist.
- Key statistics (as of July 2016):
  - About 1.6 million households received public assistance.
  - Elderly households take the largest share (over 50 percent).
  - Households with a disabled or sick individual account for about 26 percent of recipients.
  - Single-mother households account for about 6 percent of recipients.
  - Average duration on the program: about 10 percent of recipients have been on the program for less than a year; about 30 percent have been on the program for more than ten years.
- Program shortcomings identified:
  - More than half of recipients are elderly and a large share are disabled or sick, suggesting inadequate income support for the working poor.
  - The calculation method (subtracting household income from minimum living expenses) creates inherent work disincentives; a worker whose annual income is less than the minimum living expenses may be induced to not work at all.

### Case for considering an Earned Income Tax Credit (EITC) in Japan
- Four principal problems suggest consideration of an EITC:
  1. Low unemployment rate but high poverty rate of the working-age population.
  2. Widespread non-regular employment.
  3. Limited income support provided by the public assistance program for the working poor.
  4. Inherent work disincentives created by the public assistance program for the working poor.
- Additional motivations:
  - EITC programs have grown in popularity in advanced economies (AEs). Examples:
    - US and UK have had EITC programs for about four or more decades.
    - Germany included EITC as part of broader labor market reforms in the early 2000s.
    - Denmark and Sweden have implemented EITC programs since 2004 and 2007, respectively.
    - As of 2009, 16 of the 30 OECD countries have operated EITC programs (Immervoll and Pearson, 2009).
  - EITC can address concerns about work disincentives inherent in means-tested anti-poverty programs and support the working poor.
  - Introducing an EITC can help address aging population challenges by boosting employment and labor supply:
    - In 2017, the elderly (aged 65 years or over) accounted for about 27.6 percent of the total population.
    - The share of the elderly has been projected to reach as high as about 38.4% in 2065.
  - EITC aligns with Prime Minister Fumio Kishida’s “New Form of Capitalism” objective of a virtuous cycle for growth through labor market policies and a more equitable distribution of benefits.

### Theoretical rationale for EITC
- Extensions of the Mirrlees (1971) optimal income taxation model suggest an EITC can be part of the optimal income tax schedule when:
  - Labor supply responds on both the intensive (hours) and extensive (participation) margins (Diamond, 1980; Saez, 2002).
  - Two-dimensional informational asymmetries exist between government and individuals (Choné and Laroque, 2010).
  - Individual failures are present (Gerritsen, 2016; Farhi and Gabaix, 2020; Lockwood, 2020).
  - Non-welfarist social welfare functions are considered (Kanbur et al., 1994; An and Coady, 2022).
- Trade-off highlighted by standard labor supply theory:
  - Introducing an EITC tends to increase aggregate employment by inducing non-workers to enter the labor force.
  - This can increase post-EITC wages facing workers but decrease post-EITC wages facing employers; the division depends on labor demand elasticity.
  - Greater share of EITC benefit captured by workers raises poverty alleviation but can lower employment impact, and vice versa.

### International experience and scale of EITC programs
- EITC programs vary widely in design: tax concessions, social contribution reductions, refundable tax credits, eligibility criteria tied to children, minimum hours, work income, or entry into employment.
- About half of EITC programs restrict eligibility to working families with children.
- Income assessment can be individual- or family-based.
- Examples of program scale and impact:
  - US in 2019: 25 million working households received about $63 billion nationwide, with an average benefit of approximately $2,500.
  - UK in 2016-17: 2.9 million working families received about £19.3 billion, with an average benefit of around £6,700.

### Key policy issues and design considerations
- Primary objective articulation:
  - Decide whether primary objective is addressing in-work poverty or promoting employment; trade-offs exist.
  - Given Japan’s low unemployment but high working-age poverty, addressing in-work poverty could be the short-run primary objective; boosting labor force participation may be prioritized in the long run due to ageing.
- Address existing policy anomalies before introducing EITC:
  - Labor taxes: High labor income taxes (e.g., payroll taxes) at low earnings levels run counter to EITC objectives; consider tax reductions targeted at low earners and examine the personal income tax base and rate schedule.
  - Out-of-work benefits: Where eligibility for out-of-work benefits is not strict and levels are generous, reforms should strengthen the link between benefit levels and earnings; reassess public assistance eligibility criteria and benefit levels (eligibility in Japan is currently independent of employment status).
- Design choices tied to objectives:
  - Definition of income:
    - For poverty alleviation: link benefit levels to total household income (including capital income and earnings from employment or self-employment).
    - For promoting employment: individual earnings may be a more appropriate income base.
  - Linking benefits to family structure:
    - To address poverty, link benefits to family composition (e.g., number of children) and household income rather than individual income alone.
    - Although Japan’s income tax is levied on an individual basis, delivering family-based credits is possible; administrative difficulty must not undermine take-up rates.
  - Use of working hours or earnings thresholds:
    - Thresholds provide stronger incentives to work more hours and reduce fiscal cost, but increase administrative burden and may exclude vulnerable workers; could target thresholds at less vulnerable groups and condition additional benefits on job search/training.
  - Frequency and timing of payments:
    - Frequent (e.g., monthly) or contemporaneous payments can alleviate liquidity constraints and tighten perceived link between income and labor supply decisions; US experience showed preference for annual payments but information constraints may influence choices.
  - Treatment of the self-employed:
    - Verifying self-employed income is difficult and invites fraudulent claims; including the self-employed is important for poverty alleviation but must be accompanied by measures to ensure tax compliance and prevent fraud.

*Source: 1jpnea2023002 - 1.      The high poverty rate of the working-age population is a long-standing social and economic issue in Japan (IMF).*

### 13.      Implementing an EITC. The effectiveness of an EITC in achieving underlying policy

### 13.      Implementing an EITC. The effectiveness of an EITC in achieving underlying policy objectives requires that country authorities have the capacity to implement the program.

### Administrative capacity and preconditions
- Effectiveness of an EITC requires:
  - high levels of tax filing among the target population, and
  - a strong system of assessment for the personal income tax.
- If most taxpayers are required to lodge income tax returns or an effective withholding tax system is already in place, an EITC program should be relatively inexpensive to administer. Otherwise, the cost of introducing an EITC can be substantial.
- Specific implementation context in Japan:
  - Most employees in Japan do not file annual tax returns. This is because labor income taxes are withheld at source and adjusted annually by their employers, and investment income is withheld at source in many cases.
  - Information sharing is insufficient among the ministries, local governments, and public corporations:
    - Japan’s National Tax Agency collects employees’ income information from employers, but its coverage is partial.
    - Local governments gather data of residents’ incomes, but such data are not aggregated by a central agency.
  - Delivery of payments is challenging because taxpayers do not register information of their bank accounts with Japan’s National Tax Agency yet, despite transfers between bank accounts being widely-accepted means of payment.

### Administrative solutions proposed
- The three administrative challenges can be overcome by:
  - making annual tax returns mandatory while providing individuals with pre-filled tax returns;
  - utilizing the My Number taxpayer identification system launched in 2016 to improve information collection, sharing and management;
  - linking the bank account to the My Number of each taxpayer for the purpose of delivering benefits.

### Overcoming information constraints and knowledge barriers
- The EITC’s effectiveness also requires that eligible individuals:
  - are aware of these benefits, and
  - understand the inherent incentives for increasing their labor supply.
- Country authorities should address information constraints or knowledge barriers through:
  - informational outreach efforts, and
  - leveraging other sources such as peer networks (Maag, 2005; Chetty et al., 2013).
- It is important to simplify the broader tax and benefit system within which an EITC is embedded and ensure that potential beneficiaries can easily understand how various components of the tax and benefit system interact (Azmat, 2014).

### Monitoring and evaluation
- Adequate resources need to be allocated to monitoring and evaluating how the EITC is impacting employment and wage levels to ensure the policy is achieving its objectives and to avoid unintended adverse consequences.
- Monitoring and evaluation requires:
  - careful analysis of employment and wage outcomes for different household and skill types both before and after the policy is introduced.

*Source: 1jpnea2023002 - 13. Implementing an EITC. The effectiveness of an EITC in achieving underlying policy objectives*

### 12.      Banks’ exposure to transitions risks is determined by their lending portfolio and other

### 12.      Banks’ exposure to transitions risks is determined by their lending portfolio and other asset holdings.

### Banks’ exposure and sectoral lending
- Japanese G-SIBs’ exposure to energy and utilities is in the 6-8 percent range.
- Most smaller banks have smaller exposures to these sectors.
- There are substantial bank loan exposures to other, less affected sectors; a full analysis would require detailed loan exposure data.
- Bank of Japan and Financial Services Agency (2022) scenario-based analysis suggests that “the annual average of estimated transition risk credit costs [...] is considerably lower than the average annual net profits of each bank”.
- That analysis also shows that transition risk credit costs are concentrated in high emission sectors including energy, power, steel, and automotive.

### Firm-level vulnerability to carbon pricing (ICR dynamics)
- Figure: Share of Firms with Debt at Risk Covering Scope 1 and 2 Emissions (In percent, ICR<1 by carbon tax levels, excluding financial sector).
- With carbon taxes at $75 or $150 on scope 1 and 2 emissions, the share of firms with ICRs below 1 would increase by 7 percent and 14 percent, respectively.
- Sectoral exposure to carbon pricing:
  - Most exposed sectors: energy, utility, and materials (high direct emissions).
  - Downstream sectors are significantly impacted through emissions related to purchased power and electricity.
- There is a significant weak tail of firms following the pandemic that is not able to cover interest payments out of profits.

### Policy implications and recommendations
- Corporate-level actions:
  - Incorporate anticipated public policies requiring much lower emissions into business strategies.
  - Improve climate-related corporate disclosures.
- Financial sector actions:
  - Banks need to consider transition risks in their lending decisions to help allocate capital in line with climate targets.
  - Efforts by Japanese policymakers and financial institutions to quantify and manage climate financial risks are appropriate.
- Public sector actions:
  - Invest in decarbonization and support for green technologies.
  - Provide predictable emission pricing pathways.
  - Support green and transition finance.
  - Provide targeted policies to help hard-to-abate and high-emission sectors where the transition may be particularly challenging.

*Source: IMF staff summary of chapter content.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2023/english/1jpnea2023002.pdf_