## sipea2024027

## Source details

**Canonical URL:** [sipea2024027](https://www.imf.org/-/media/files/publications/selected-issues-papers/2024/english/sipea2024027.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/selected-issues-papers/2024/english/sipea2024027.pdf.md)
- [Structured JSON version](/-/media/files/publications/selected-issues-papers/2024/english/sipea2024027.pdf.json)

---

### Overview and policy context
- Japan’s startup ecosystem has grown gradually, supported by the government’s "Startup Development Five-Year Plan".
- Plan pillars:
  - 1) building human resources and networks for creating startups;
  - 2) strengthening funding for startups and diversifying exit strategies;
  - 3) promoting open innovation.
- Japan Investment Corporation: launched a 200-billion-yen venture growth fund in 2023 to support later stage startups and target early-stage startups beyond deep tech and life sciences.
- Tokyo: emerged as a top 20 city for the global VC ecosystem and ranks as the third city in terms of fund value (PitchBook).

### Stylized facts: scale, industry composition, and geography
- Unicorns and valuations (2023):
  - Japan: 7 unicorns; valuation of unicorns about US$9.8 billion.
  - U.S.: 653 unicorns; valuation of unicorns about US$2 trillion.
- Geographic concentration:
  - Tokyo metro area accounts for about 80% of total startup funding in Japan.
- Industry composition:
  - Most startups operate in information technology (IT), followed by health care, and business products and services (B2B).
  - IT/SaaS share appears somewhat higher in Japan than the global average.
  - CleanTech startups are less prevalent in Japan than in the global sample.
- Investor origin and sector focus (2010–2023):
  - U.S. investors account for about 50 percent of investment into Japan’s startups.
  - U.K. investors account for about 10 percent.
  - Japanese investors’ share declined to about 5 percent in recent years.
  - Most capital invested in Japan flows to IT and health care sectors; more recently, B2B and B2C startups have seen rising capital inflows.
- Corporate venture capital (CVC):
  - Between 2015 and 2022, Japanese CVCs invested in at least half of all VC deals in Japan, peaking at 62 percent in 2020.
- Stage of funding:
  - Most deals are in seed and early-stage VC, though the share of later-stage VC has increased in recent years.

### Aggregate cross-country dataset and empirical approach
- Coverage and data:
  - 30 large advanced and emerging market economies.
  - Time coverage: 2000 to 2022.
  - Firm performance data: PitchBook.
  - Macro and structural variables include GDP growth, inflation, firm dynamics (entry and exit rates), entrepreneurship (share of self-employed who are employers), average job tenure.
- Econometric specification:
  - Panel regressions estimated with the Arellano-Bover/Blundell-Bond linear dynamic panel data estimator with robust standard errors.
  - Dependent variables: capital invested (mean) in firms or valuation (mean) of firms (millions of U.S. dollars).
  - Independent variables: country-specific macro and structural variables, including lagged dependent variable.

### Aggregate cross-country findings (structural drivers)
- Capital invested (mean):
  - Higher share of entrepreneurship (lagged) is associated with higher capital invested (mean).
  - Better firm dynamism (higher entry rate, lagged) is associated with higher capital invested (mean).
- Valuations (mean):
  - Higher share of entrepreneurship (lagged) is associated with higher valuations (mean).
  - Better firm dynamism (higher entry and exit rates, lagged) is associated with higher valuations (mean).
- Selected numeric coefficients and reported figures (as presented):
  - Valuation (lagged): -0.013 0.0883* 0.0236 -0.00293 0.101***.
  - Real GDP growth (lagged): 5.479 4.218 3.148 -0.335 3.363.
  - Inflation (lagged): 6.924 -2.614 4.331 1.020 0.402.
  - Enterpreneurship (lagged): 32.44* 12.73.
  - Exit rate (lagged): 17.12 18.35*.
  - Entry rate (lagged): 5.619*.
  - Observations reported (valuation regressions): 17050 121433 34418.
  - Capital invested (lagged) coefficients: -0.0399 0.0803** 0.00502 0.0144 0.103***.
  - Real GDP growth (lagged) in capital regressions: 2.971* 1.213 1.84 -0.806 -0.84.
  - Enterpreneurship (lagged) in capital regressions: 17.37*** 15.40***.
  - Entry rate (lagged) in capital regressions: 9.061*.
  - Constant terms and standard errors are reported in the original tables.

### Firm-level analysis: sample, identification, and empirical strategy
- PitchBook firm-level sample and definitions:
  - Nearly 4 million companies globally in PitchBook; sample limited to startups as of end-2023.
  - Startups defined as companies less than 10 years old and backed by any early-stage funding at least once.
  - Final country sample (12 countries): Australia, Canada, France, Germany, India, Italy, Japan, Korea, Singapore, Spain, Sweden, United Kingdom.
  - U.S. excluded due to data limitations.
- Outcome variables (Oi): log of total patent documents, log of number of employees, and the exit probability (via IPO or M&A) as reported by PitchBook.
- Treatment variable (Well-funded): dummy equal to 1 if total capital raised per employee by startup i is above the country-industry median; 0 otherwise.
- Controls (Xi): firm age and number of employees (number of employees dropped when it is the outcome).
- Identification and estimation:
  - Country-sector fixed effects included.
  - Coarsened exact matching to match well-funded and non-well-funded startups by country, industry, age, and number of employees.
  - Well-funded treated as endogenous; instrumented with Zi: (i) total number of active investors, (ii) years since the first funding, and (iii) years since the last funding.
  - Auxiliary regression: Well-funded = β Zi + error; test for cov(error, outcome error) = 0 uses χ2 distribution.

### Firm-level results: impact of funding on startup performance (global and Japan)
- Main global sample findings:
  - Well-funded startups have higher total patent documents, are larger (more employees), and have higher exit probabilities (via IPO or M&A).
  - Quantitative magnitudes (as reported):
    - Well-funded startups are predicted to have 1.5 times more employees compared to non-well-funded startups.
    - Well-funded startups have 43 percentage points higher probability of exit compared to non-well-funded startups.
- Global regressions — selected coefficients (Table 3):
  - Treatment Variable = Well-Funded:
    - log(patent docs): 0.744* (standard error 0.408)
    - log(# employees): 1.540*** (standard error 0.171)
    - exit: 43.848*** (standard error 1.774)
    - IPO: 14.500*** (standard error 3.108)
    - MA: 39.903*** (standard error 1.819)
  - Controls and other stats:
    - Observations: 1,854 across reported columns.
    - chi2 - p-value (rho=0): 0.3280000
- Global alternative treatment-specification (# Active Investors and timing):
  - # Active Investors effects:
    - log(patent docs): 0.033*** (0.007)
    - log(# employees): 0.064*** (0.006)
    - exit: 0.035*** (0.007)
    - IPO: 0.041*** (0.008)
    - MA: 0.026*** (0.006)
  - Years since first funding and years since last funding show positive and negative effects respectively on outcomes (coefficients and standard errors reported in source).
  - Observations: 1,854.
- Japan-specific findings (Table 4):
  - Treatment Variable = Well-Funded (Japan sample):
    - log(patent docs): 3.443*** (0.283)
    - log(# employees): 0.848*** (0.173)
    - exit: 51.863*** (3.026)
    - IPO: 23.050*** (2.172)
    - MA: 46.359*** (5.190)
  - Observations: 302 across reported columns.
  - chi2 - p-value (rho=0): 00000
- Japan-specific alternative specification (# Active Investors, timing):
  - # Active Investors (Japan):
    - log(patent docs): 0.001 (0.006)
    - log(# employees): 0.054*** (0.009)
    - exit: 0.018** (0.009)
  - Years since first funding and years since last funding show mixed effects; years since last funding negative and significant for several outcomes.
  - Observations: 302.

- Endogeneity and auxiliary tests:
  - The p-value of the χ2 statistic for the null hypothesis of uncorrelated errors in main and auxiliary regressions is strongly rejected in the majority of specifications (χ2 p-values reported in Table 3).
  - The number of active investors is statistically significant in predicting whether a startup is well-funded (reported in lower panel of Table 3).

### Role of cultural factors (risk-taking) in mediating funding effects
- Cultural proxies (Hofstede, 2013):
  - (i) uncertainty avoidance — society’s tolerance for uncertainty and ambiguity;
  - (ii) power distance — how much a society delegates power to authority and accepts unequal power distribution.
- Key interaction findings:
  - The positive impact of availability of funding on startup exit is higher in countries that reward risk-taking behavior.
  - Predicted impact of funding on IPO exits is higher in countries with less uncertainty avoidance.
  - Predicted impact of funding on M&A exits is higher in countries with less power distance.
- Selected interaction coefficients (Table 5):
  - Well-Funded: 46.022*** (1.484) and 48.184*** (1.412)
  - Well-Funded * Country Characteristic: -1.617 (1.671) and -4.376*** (1.138)
  - Dependent variable variants report Well-Funded coefficients and Well-Funded * Country Characteristic interactions with significance levels and standard errors as presented in source.
  - Observations: 1,854; various chi2 - p-value (rho=0) reported (for example: 0.003600 and 0.00285 in cited specifications).

### Caveats and data limitations
- Cross-sectional design cannot track startups’ performance following funding rounds.
- Data vintage: end-2023; includes surviving startups only; firms that exited before end-2023 are not observed (survivor bias remains).
- Analysis examines availability of funding but not cost, conditionality, or investor type due to data limitations.
- Identification is within country-industry; this sharpens identification but overlooks potential variation across countries and/or industries.
- Does not determine whether private equity funding complements or substitutes other financing types (for example, debt financing).
- Exploiting country-industry variation partly mitigates bias from omitting other funding sources.
- Matching diagnostics:
  - Two-sided t-test p-values before and after matching:
    - Age: p-value 0.05 before matching and 1.00 after matching.
    - Total number of employees: p-value 0.003 before matching and 0.42 after matching.
  - Matching drops 13 percent of the firms in the final sample.

### Policy-relevant implications and recommendations
- Strengthen entrepreneurship and firm dynamism:
  - Higher entry and exit rates support higher capital investment and firm valuations.
- Broaden access to diverse funding sources, especially later-stage financing:
  - Policies facilitating access to diverse funding sources can help create larger startups and raise valuations (consistent with the 200-billion-yen venture growth fund initiative).
- Encourage risk-taking culture and reduce institutional obstacles:
  - Reducing uncertainty avoidance and power distance tendencies could amplify the effectiveness of funding in generating successful exits.
- Promote geographic diversification of VC activity within Japan:
  - Reducing Tokyo-centric concentration (about 80 percent concentrated in Tokyo) may lower concentration risks.
- Labor market reform to support entrepreneurship:
  - A more flexible labor market, gradual shift away from lifelong employment, reduced labor-market dualism, encouragement of merit-based promotions, and facilitation of job mobility can encourage entrepreneurship and improve allocation of talent.
- Improve firm dynamism and reduce zombie firms:
  - Greater firm entry and exit, reduced personal liabilities, and gradual reduction of zombie firms could improve allocation of capital and labor to more productive ventures, boosting productivity and growth.

*Prepared by Salih Fendoglu (MCM) and TengTeng Xu (APD); April 15, 2024 — STARTUPS AND VENTURE CAPITAL IN JAPAN: HOW TO GROW (chapter: 1. Capital Investment and Structural Characteristics).*

### 1. Capital Investment and Structural Characteristics ___________________________________ 6

### 1. Capital Investment and Structural Characteristics

### Overview and policy context
- Japan’s startup ecosystem has grown gradually, supported by the government’s "Startup Development Five-Year Plan".
- The plan focuses on three main pillars:
  - 1) building human resources and networks for creating startups;
  - 2) strengthening funding for startups and diversifying exit strategies;
  - 3) promoting open innovation.
- The Japan Investment Corporation launched a 200-billion-yen venture growth fund in 2023 to support later stage startups and target early-stage startups beyond deep tech and life sciences.
- Tokyo has emerged as a top 20 city for the global VC ecosystem and ranks as the third city in terms of fund value (PitchBook).

### Stylized facts: scale, industry composition, and geography
- Number of unicorns and valuations (2023):
  - Japan: 7 unicorns; valuation of unicorns about US$9.8 billion.
  - U.S.: 653 unicorns; valuation of unicorns about US$2 trillion.
- Concentration:
  - Tokyo metro area accounts for about 80% of total startup funding in Japan.
- Industry composition:
  - Most startups operate in information technology (IT), followed by health care, and business products and services (B2B).
  - IT/SaaS share appears somewhat higher in Japan than the global average.
  - CleanTech startups are less prevalent in Japan than in the global sample.
- Investor origin and sector focus (2010–2023):
  - U.S. investors account for about 50 percent of investment into Japan’s startups.
  - U.K. investors account for about 10 percent.
  - Japanese investors’ share declined to about 5 percent in recent years.
  - Most capital invested in Japan flows to IT and health care sectors; more recently, B2B and B2C startups have seen rising capital inflows.
- Corporate venture capital (CVC):
  - Between 2015 and 2022, Japanese CVCs invested in at least half of all VC deals in Japan, peaking at 62 percent in 2020.
- Stage of funding:
  - Most deals are in seed and early-stage VC, though the share of later-stage VC has increased in recent years.

### Aggregate cross-country dataset and empirical approach
- Cross-country aggregate database:
  - 30 large advanced and emerging market economies.
  - Time coverage: 2000 to 2022.
  - Firm performance data: PitchBook.
  - Macro and structural variables: GDP growth, inflation, firm dynamics (entry and exit rates), entrepreneurship (share of self-employed who are employers), average job tenure.
- Econometric specification:
  - Panel regressions estimated with the Arellano-Bover/Blundell-Bond linear dynamic panel data estimator with robust standard errors.
  - Dependent variables: capital invested (mean) in firms or valuation (mean) of firms (millions of U.S. dollars).
  - Independent variables: country-specific macro and structural variables, including lagged dependent variable.

### Aggregate cross-country findings (structural drivers)
- Capital invested (mean):
  - Higher share of entrepreneurship (lagged) is associated with higher capital invested (mean).
  - Better firm dynamism (higher entry rate, lagged) is associated with higher capital invested (mean).
- Valuations (mean):
  - Higher share of entrepreneurship (lagged) is associated with higher valuations (mean).
  - Better firm dynamism (higher entry and exit rates, lagged) is associated with higher valuations (mean).
- Selected numeric values from regression tables (preserved as reported):
  - Valuation (lagged): -0.013 0.0883* 0.0236 -0.00293 0.101*** (standard errors shown in parentheses in source).
  - Real GDP growth (lagged): 5.479 4.218 3.148 -0.335 3.363.
  - Inflation (lagged): 6.924 -2.614 4.331 1.020 0.402.
  - Enterpreneurship (lagged): 32.44* 12.73.
  - Exit rate (lagged): 17.12 18.35*.
  - Entry rate (lagged): 5.619*.
  - Observations reported (valuation regressions): 17050 121433 34418 (as presented in table).
  - Capital invested (lagged) coefficients reported: -0.0399 0.0803** 0.00502 0.0144 0.103***.
  - Real GDP growth (lagged) in capital regressions: 2.971* 1.213 1.84 -0.806 -0.84.
  - Enterpreneurship (lagged) in capital regressions: 17.37*** 15.40***.
  - Entry rate (lagged) in capital regressions: 9.061*.
  - Constant terms and standard errors are reported in the original tables.

### Firm-level analysis: sample and identification
- Firm-level PitchBook sample:
  - Nearly 4 million companies globally in PitchBook; sample limited to startups as of end-2023.
  - Startups defined as companies less than 10 years old and backed by any early-stage funding (venture capital, accelerator/incubator, or angel) at least once.
  - Final country sample (12 countries with largest coverage): Australia, Canada, France, Germany, India, Italy, Japan, Korea, Singapore, Spain, Sweden, United Kingdom.
  - U.S. excluded due to data limitations.
- Empirical strategy:
  - Outcome variables (Oi): log of total patent documents, log of number of employees, and the exit probability (via IPO or M&A) as reported by PitchBook.
  - Treatment variable (Well-funded): dummy equal to 1 if total capital raised per employee by startup i is above the country-industry median; 0 otherwise.
  - Controls (Xi): firm age and number of employees (number of employees dropped when it is the outcome).
  - Country-sector fixed effects included.
  - Coarsened exact matching: match well-funded startups to non-well-funded startups in same country and industry with same age and number of employees.
  - Endogeneity addressed by treating Well-funded as endogenous and instrumenting with Zi: (i) total number of active investors, (ii) years since the first funding, and (iii) years since the last funding.
  - Auxiliary regression: Well-funded = β Zi + error; test for cov(error, outcome error) = 0 uses χ2 distribution.

### Firm-level results: impact of funding on startup performance
- Main findings (global sample):
  - Well-funded startups have higher total patent documents, are larger (more employees), and have higher exit probabilities (via IPO or M&A).
  - Quantitative magnitudes preserved as reported:
    - Well-funded startups are predicted to have 1.5 times more employees compared to non-well-funded startups.
    - Well-funded startups have 43 percentage points higher probability of exit compared to non-well-funded startups.
  - Endogeneity tests:
    - The p-value of the χ2 statistic for the null hypothesis of uncorrelated errors in main and auxiliary regressions is strongly rejected in the majority of specifications (χ2 p-values reported in Table 3 in source).
  - The number of active investors is statistically significant in predicting whether a startup is well-funded (reported in lower panel of Table 3).
- Japan-specific sample:
  - Results hold qualitatively for Japan.
  - For Japan, the impact of availability of funding on the number of patents is larger and more precisely estimated.
  - Estimated impacts on other outcome variables for Japan are not materially different from the global sample (see Table 4 in source).

### Role of cultural factors (risk-taking) in mediating funding effects
- Augmented model includes country-level cultural proxies from Hofstede (2013):
  - (i) uncertainty avoidance — society’s tolerance for uncertainty and ambiguity;
  - (ii) power distance — how much a society delegates power to authority and accepts unequal power distribution.
- Key result:
  - The positive impact of availability of funding on startup exit is higher in countries that reward risk-taking behavior.
  - Specifically:
    - Predicted impact of funding on IPO exits is higher in countries with less uncertainty avoidance.
    - Predicted impact of funding on M&A exits is higher in countries with less power distance.

### Policy-relevant implications (from findings)
- Strengthening entrepreneurship and firm dynamism (higher entry and exit rates) supports higher capital investment and firm valuations.
- Policies facilitating access to diverse funding sources—especially to broaden later-stage financing—can help create larger startups and raise valuations (consistent with Japan Investment Corporation’s 200-billion-yen venture growth fund initiative).
- Encouraging risk-taking culture and reducing institutional obstacles to risky entrepreneurship could amplify the effectiveness of funding in generating successful exits.
- Promoting geographic diversification of VC activity within Japan may reduce concentration risks from the current Tokyo-centric funding distribution (about 80 percent concentrated in Tokyo).

*Prepared by Salih Fendoglu (MCM) and TengTeng Xu (APD); April 15, 2024 — STARTUPS AND VENTURE CAPITAL IN JAPAN: HOW TO GROW (chapter: 1. Capital Investment and Structural Characteristics).*

### 18.      The results should be read with some caveats, in large part due to data limitations.

### sipea2024027 - 18.      The results should be read with some caveats, in large part due to data limitations.

### Caveats and data limitations
- The analysis is cross-sectional and cannot track startups’ performance following funding rounds.
- The study uses a particular vintage of the data, i.e., end-2023, that includes surviving startups only; data for firms that had exited before end-2023 is not available, hence survivor bias remains.
- The study examines availability of funding but does not explore cost, conditionality of funding, or type of investors due to data limitations.
- Estimated impacts are identified within country-industry to sharpen identification, at the expense of overlooking potential variation across countries and/or industries.
- The analysis does not determine whether private equity funding is a complement or substitute to other financing types (for example, debt financing).
- Exploiting country-industry variation in the identification strategy partly mitigates potential bias due to omitting other potential sources of funding.
- Matching procedure notes:
  - Two-sided t-test of means for treated and control samples imply for age a p-value of 0.05 before matching and 1.00 after the matching.
  - For total number of employees, a p-value of 0.003 before matching and 0.42 after matching.
  - Matching drops 13 percent of the firms in the final sample.

### Empirical findings (high-level, from reported regressions and tables)
- Global regressions (Table 3: Does Availability of Funding Affect Startup Performance?) — selected coefficients and statistics:
  - Treatment Variable = Well-Funded:
    - log(patent docs): 0.744* (standard error 0.408)
    - log(# employees): 1.540*** (standard error 0.171)
    - exit: 43.848*** (standard error 1.774)
    - IPO: 14.500*** (standard error 3.108)
    - MA: 39.903*** (standard error 1.819)
  - Controls:
    - log (# employees) coefficient on relevant equations: 0.159*** (0.030); additional entries 11.635*** (0.471), 4.833*** (0.409), 7.229*** (0.689)
    - log(age): 0.677*** (0.041); additional entries 0.285*** (0.105), -6.517*** (1.174), -0.237 (0.770), -7.174*** (1.116)
  - Observations: 1,854 across reported columns.
  - chi2 - p-value (rho=0): 0.3280000

- Global regressions — alternative treatment-variable specification using funding timing and investor activity:
  - # Active Investors:
    - log(patent docs): 0.033*** (0.007)
    - log(# employees): 0.064*** (0.006)
    - exit: 0.035*** (0.007)
    - IPO: 0.041*** (0.008)
    - MA: 0.026*** (0.006)
  - Years since first funding:
    - log(patent docs): 0.045*** (0.010)
    - log(# employees): 0.018 (0.012)
    - exit: 0.035*** (0.008)
    - IPO: 0.029** (0.011)
    - MA: 0.040*** (0.008)
  - Years since last funding:
    - log(patent docs): -0.124*** (0.026)
    - log(# employees): -0.108*** (0.026)
    - exit: -0.356*** (0.024)
    - IPO: -0.138*** (0.024)
    - MA: -0.305*** (0.023)
  - Observations: 1,854 across reported columns.

- Japan-specific regressions (Table 4: Japan: Does Availability of Funding Affect Startup Performance?) — selected coefficients and statistics:
  - Treatment Variable = Well-Funded (Japan sample):
    - log(patent docs): 3.443*** (0.283)
    - log(# employees): 0.848*** (0.173)
    - exit: 51.863*** (3.026)
    - IPO: 23.050*** (2.172)
    - MA: 46.359*** (5.190)
  - Controls:
    - log (# employees): 0.0571 (0.073); additional entries 1.802*** (1.083), 4.750*** (0.955), 7.187*** (1.269)
    - log(age): 0.730*** (0.214); additional entries 0.575*** (0.141), -3.726 (3.613), 1.739 (2.174), -7.524** (3.339)
  - Observations: 302 across reported columns.
  - chi2 - p-value (rho=0): 00000

- Japan-specific alternative specification (# Active Investors, years since funding):
  - # Active Investors (Japan):
    - log(patent docs): 0.001 (0.006)
    - log(# employees): 0.054*** (0.009)
    - exit: 0.018** (0.009)
    - IPO: 0.011 (0.010)
    - MA: 0.011 (0.011)
  - Years since first funding:
    - log(patent docs): -0.016 (0.027)
    - log(# employees): 0.040 (0.040)
    - exit: 0.057* (0.030)
    - IPO: 0.085** (0.037)
    - MA: 0.045 (0.038)
  - Years since last funding:
    - log(patent docs): -0.080* (0.041)
    - log(# employees): -0.146*** (0.054)
    - exit: -0.432*** (0.047)
    - IPO: -0.237*** (0.050)
    - MA: -0.307*** (0.064)
  - Observations: 302 across reported columns.

- Risk culture and exits (Table 5: Global: Does Risk Culture Matter for Startup Exit?) — interaction results:
  - Country Characteristic: Uncertainty Avoidance / Power Distance — selected coefficients:
    - Well-Funded: 46.022*** (1.484) and 48.184*** (1.412)
    - Well-Funded * Country Characteristic: -1.617 (1.671) and -4.376*** (1.138)
    - log (# employees): 11.999*** (0.459) and 12.090*** (0.428)
    - log(age): -3.198** (1.452) and -3.230** (1.463)
    - Observations: 1,854; chi2 - p-value (rho=0): 00
  - Dependent Variable: Successful Exit / MA Exit / IPO Exit — sample coefficients include:
    - Well-Funded: 16.574*** (4.084) and 11.204*** (2.660)
    - Well-Funded * Country Characteristic: -5.241** (2.178) and 2.344 (1.957)
    - log (# employees): 4.293*** (0.446) and 4.221*** (0.459)
    - log(age): -3.943*** (1.266) and -4.065*** (1.388)
    - Observations: 1,854; chi2 - p-value (rho=0): 0.003600 and 0.00285
    - Additional reported specification:
      - Well-Funded: 40.367*** (3.285) and 48.153*** (2.244)
      - Well-Funded * Country Characteristic: 2.514 (3.479) and -7.293*** (1.496)
      - log (# employees): 8.005*** (0.755) and 8.158*** (0.706)
      - log(age): -0.920 (2.123) and -0.680 (2.072)
      - Observations: 1,854; chi2 - p-value (rho=0): 00

### Policy implications (paragraphs 19–21)
- Equity funding importance:
  - The results highlight the importance of equity funding in supporting startups to grow and eventually exit in Japan.
  - Better access to equity funding is crucial for startups to grow, innovate, and exit successfully.
  - Angel or venture capital investment provides private equity financing when startups lack access to capital markets, bank loans, or other debt instruments at early stages, and offers value-added services (for example, operational and market insights).
- Labor market reform:
  - A more flexible labor market is crucial for entrepreneurship and innovation.
  - A gradual shift away from the lifelong employment system could encourage talented individuals to set up startups and allow second chances after failure.
  - Reducing labor-market dualism, encouraging merit-based promotions, and facilitating more job mobility can encourage entrepreneurship, associated with higher capital investment and firm valuations at the country level.
- Firm dynamism:
  - Greater firm dynamism can support startups and innovation.
  - Dynamic firm entry and exit and reduced personal liabilities can encourage entrepreneurship, innovation, and more efficient allocation of resources.
  - A gradual reduction of zombie firms could help improve allocation of capital and labor to more productive ventures, boosting productivity and growth.

*Source: sipea2024027 - 18.      The results should be read with some caveats, in large part due to data limitations.*

---


_Source: https://www.imf.org/-/media/files/publications/selected-issues-papers/2024/english/sipea2024027.pdf_