## _wp13196

## Source details

**Canonical URL:** [_wp13196](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13196.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13196.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13196.pdf.json)

---

### I. Introduction and objectives
- 10-year average unemployment rate in Japan: rose from 2.5 percent in the 1980s, to 3 percent in the 1990s, and to 4.7 percent in the 2000s.
- Unemployment rate prior to the 1997 recession: stayed well below 4 percent; rose by 1.2 percent to around 5 percent since then.
- Change in unemployment rate between 2006 and 2009: U.K.: +2.1%; U.S.: +4.7%; Japan: +1.2%.
- Beveridge curve: shifted outward since the late 1990s, suggesting increased labor market mismatch (skill and other mismatches).
- Objectives:
  - Calculate mismatch across age groups, occupation, and employment-type (full time versus part time) in Japan for April 2000–April 2013 using the methodology of Sahin et al. (2013).
  - Measure how much the types of mismatch contributed to the rise in the Japanese unemployment rate during the global financial crisis.

### Key empirical findings (overview)
- Age mismatch has steadily declined.
- Occupational mismatch and employment-type mismatch display countercyclical patterns.
- Occupational mismatch increased sharply by about 1.5 to 2-fold during the 2008 global financial crisis and has stayed relatively high since then.
- Occupational mismatch accounted for a significant portion (20-40 percent) of the recent rise in the unemployment rate.
- The magnitude of occupational mismatch in Japan is comparable to that in the U.K. (Patterson et al. 2013) and the U.S. (Sahin et al. 2013).

### Data and basic statistics (April 2000–April 2013 unless noted)
- Aggregate and monthly averages (levels):
  - Hires: 151,998
  - Unemployed: 2,324,788
  - Vacancies: 1,643,031
- Hires / Unemployed / Vacancies — Shares (Percent) by Age:
  - 15-24 years old: 17.3 / 12.6 / 22.1
  - 25-34 years old: 29.2 / 27.7 / 27.0
  - 35-44 years old: 21.7 / 19.0 / 21.1
  - 45-54 years old: 16.9 / 17.4 / 14.6
  - 55-64 years old: 13.4 / 20.6 / 10.9
  - 65 years old and over: 1.6 / 2.7 / 4.3
- Hires / Unemployed / Vacancies — Shares (Percent) by Employment Type:
  - Full-Time: 68.3 / 62.3 / 54.2
  - Part-Time: 31.7 / 37.7 / 45.8
- Hires / Unemployed / Vacancies — Shares (Percent) by Occupation:
  - Production processing: 34.1 / 31.8 / 26.7
  - Clerks: 19.2 / 9.0 / 10.6
  - Professionals/Engineers: 16.3 / 13.4 / 23.2
  - Sales: 10.5 / 12.7 / 15.5
  - Services: 10.2 / 7.2 / 15.0
  - Transportation, information, communications: 5.9 / 4.1 / 5.4
  - Defense: 2.5 / 0.8 / 2.9
  - Agricultural, forestry, fishery: 1.0 / 0.5 / 0.5
  - Managers: 0.2 / 0.4 / 0.2
- Observations:
  - On average, about 2.3 million unemployed workers and 1.6 million vacancies per month, resulting in about 0.15 million monthly hires.
  - Younger age groups (15-24, 25-34, 35-44) have higher shares of hires relative to their share of unemployment.
  - Full-time hires: 68.3 percent of hirings are for full-time workers; while 62.3 percent of unemployed workers look for full-time positions, 68.3 percent of hires are full-time.

### Empirical strategy and matching framework
- Two-step strategy:
  - Measure mismatch across occupations, age groups, and full-time/part-time jobs following Sahin et al. (2013).
  - Compute counterfactual unemployment in the absence of mismatch to measure contribution to changes in unemployment.
- Island-specific matching function:
  - h_it(v_it,u_it) = Φ_t φ_it v_it^{α} u_it^{1−α}, where φ_it is island-specific matching efficiency, Φ_t is aggregate matching efficiency, v_it vacancies, u_it unemployed.
  - Vacancy share, α, is set to 0.4 (estimated around 0.4; 0.34 based on GMM; results checked for α in 0.3–0.5).
- Efficient allocation condition (under homogenous productivities and separations):
  - φ_it (v_it / u*_it)^{α} = φ_jt (v_jt / u*_jt)^{α}.

### Mismatch indices (definitions and interpretation)
- Theory-based indices:
  - M_{φt} = (h*_t − h_t)/h*_t = 1 − ∏_{i=1}^I [ (φ̄_t / φ_it) (v_it / v_t)^{α} (u_it / u_t)^{1−α} ] where φ̄_t = [ ∑_{i∈I} φ_it^{1/α} (v_it / v_t) ]^{α}.
  - M_t (no heterogeneity in matching efficiency): M_t = 1 − ∏_{i=1}^I (v_it / v_t)^{α} (u_it / u_t)^{1−α}.
- Conventional statistical mismatch index:
  - M_{st} = 1/2 ∑_{i∈I} | v_it / v_t − u_it / u_t | (Jackman and Roper, 1987).
- Interpretation: indices measure fractions of hires lost because of mismatch.

### Mismatch unemployment and counterfactual law of motion
- Counterfactual unemployment law of motion:
  - u^*_{t+1} = (1 − s_t − f^*_t) u^*_t + s_t, where s_t is observed job separation rate and f^*_t is counterfactual job-finding rate.
  - f^*_t = f_t / [1 − M_{φt} (u_t / u^*_t)^{α}], where f_t is observed job-finding rate.
- Since M_{φt} ∈ (0,1) and u_t / u^*_t > 1, f^*_t > f_t, implying counterfactual unemployment is lower than observed. The gap u_t − u^*_t is mismatch unemployment.

### Key empirical results (detailed)
- General findings:
  - 68.3 percent of the hirings are for full-time workers.
  - Age mismatch declined steadily since 2001.
  - Employment-type mismatch is countercyclical and increased during/after the 2008 global financial crisis.
  - Occupational mismatch rose substantially during the 2008 recession and remained higher than pre-recession levels.
- Age mismatch (quantified):
  - In 2000, around 9 percent of hires were lost due to age mismatch; by 2007 this decreased to around 2.5 percent.
  - Contribution to change in unemployment rate between 2006 and July 2009 (Change in Unemployment Rate = 1.38 percent):
    - Age mismatch:
      - M_{φ}: −0.02 percent; Percentage explained: 1.7
      - M: 0.01 percent; Percentage explained: 0.9
    - Text note: contribution ranged between −0.02 to 0.01 percent out of 1.38 percent.
- Employment-type mismatch:
  - Peaked at 7.4 percent in April 2002.
  - Dropped sharply from 6 percent to 1 percent in 2004; stayed around 0.5 percent until increasing during the 2008 crisis and remaining around 3 percent thereafter.
  - Contribution to change in unemployment rate between 2006 and July 2009:
    - M_{φ}: 0.19 percent; Percentage explained: 13.9
    - M: 0.06 percent; Percentage explained: 4.0
- Occupational mismatch:
  - Based on 9 occupational categories: average 6.7 percent in 2006 rising to a peak of 12.9 percent in February 2009; average since the 2008 recession remained high at 10 percent.
  - Based on 65 occupation categories: rose from the 2006 average of 19 percent to 26 percent in February 2009 (increase of 7 percent).
  - Contribution to change in unemployment rate between 2006 and July 2009:
    - Occupation (9 categories):
      - M_{φ}: 0.37 percent; Percentage explained: 26.7
      - M: 0.32 percent; Percentage explained: 23.2
    - Occupation (65 categories):
      - M_{φ}: 0.58 percent; Percentage explained: 41.8
      - M: 0.32 percent; Percentage explained: 23.2
  - Interpretation: occupational mismatch accounted for approximately 20-40 percent of the rise in Japanese unemployment during 2008-2009.
- Robustness to vacancy share α:
  - When α ranges from 0.3 to 0.5, age, employment-type, and occupational mismatch explain 1.0-2.7 percent, 12.1-15.9 percent, and 24.3-32.2 percent of the rise in the unemployment rate, respectively.
- Cross-country comparison:
  - U.S. contribution (reported in Sahin et al. (2013)): 11.1–29.3 percent depending on disaggregation.
  - U.K. contribution (reported in Patterson et al. (2013)): 17.1–26.9 percent depending on disaggregation.
  - Note: cross-country comparisons are sensitive to disaggregation levels.

### Descriptive matching and labor market measures (selected statistics)
- Aggregate matching and rates:
  - Aggregate matching: h_t = Θ_t v_t^{α} u_t^{1−α}.
  - Job-finding rate: f_t = h_t / u_t = Θ_t (v_t / u_t)^{α}.
  - Vacancy yield: y_t = h_t / v_t = Θ_t (u_t / v_t)^{1−α}.
  - Estimated vacancy share for Japanese data around 0.4 (0.34 based on GMM; other estimates higher).
- Patterns by group:
  - Age groups:
    - 15-24: highest job-finding rates; highest vacancy-unemployment ratios.
    - 65+: large increase in vacancy-unemployment ratio between 2001 and 2008 but lower job-finding probabilities; job-finding rates for older groups improved modestly.
    - Job-finding rates for 15-54 are more cyclical than for 55-64 and 65+.
  - Full-time vs Part-time:
    - Historically, probability of finding a job higher for part-time workers than full-time workers.
    - Job-finding rate for part-time declined after 2004 and remained stable through the 2008 recession; probabilities for part-time and full-time became almost the same after the global financial crisis.
    - Vacancy yields: easier to fill full-time positions than part-time positions from firms’ perspective.
  - Occupations:
    - High job-finding rates in production processing occupations and agricultural, forestry, and fishery workers.
    - Higher vacancy yields in agricultural, forestry and fishery occupations and clerical positions.
- Data sources and notes:
  - Unemployed, vacancies, and hires by occupation, age, and employment type from monthly Employment Referrals for General Workers (Ippan Shokugyo Shokai Jokyo), MHLW (excludes newly graduated college workers).
  - The survey’s unemployed count represents approximately 80 percent of the number reported by the Cabinet Office for April 2000–April 2013.
  - Employment-type data series were seasonally adjusted in source; other series seasonally adjusted using X-12-ARIMA by the author.
  - Part-time workers: hours worked less than regular workers at same establishment; definitions per MHLW.
  - Job separation rate from MHLW’s Monthly Labour Survey (about 33,000 establishments).

### Part vs Full-Time Mismatch — definitions and measures
- Observed sector-specific job separation rates identity:
  - (1−s_i) ≡ (1−∆)(1−δ_i), where ∆ is the aggregate job destruction rate and δ_i is the sector-specific job destruction rate.
- Three mismatch measures compared:
  - Simple Mismatch Index (M_s) based on equation (6) — measure used by statistical agencies in Japan.
  - Baseline Mismatch (M) in equation (5) — mismatch in the absence of heterogeneity in matching efficiencies.
  - Preferred Mismatch Index (M_φ) in equation (4).
- Key observations:
  - M_φ, M, and M_s generally capture similar trends in mismatch in Japan.
  - M_φ captures the rise in mismatch after the 2011 earthquake in Japan while M_s fails to do so.
  - M_φ has a clear interpretation as a fraction of hires lost owing to mismatch.

### Robustness, limitations, and organization
- Robustness:
  - Analysis accounts for heterogeneity in matching efficiency across markets.
  - Results robust to vacancy share α in range 0.3–0.5.
- Limitations:
  - Assumes homogeneity in productivities and job separation rates across submarkets due to data limitations.
  - Could not analyze additional dimensions (male vs. female, regular vs. nonregular, prefectures) because of hiring data limitations.
- Paper organization:
  - Section II: literature review.
  - Section III: data description.
  - Section IV: empirical strategy to measure mismatch and its contribution to unemployment.
  - Section V: empirical results.
  - Section VI: conclusions and discussion of potential policy measures.

### Policy conclusions and recommendations
- Main conclusion: labor market mismatch is a significant issue in Japan; occupational mismatch in particular contributed substantially (20-40 percent) to the rise in unemployment during 2008-2009.
- Policy implications and directions:
  - Focus policy efforts on reducing occupational mismatch to limit rises in unemployment.
  - Strengthen and enhance training programs to help unemployed workers acquire skills necessary for in-demand occupations; build on existing programs and public employment security offices.
  - Improve job matching processes and platforms; consider opening job-matching services to private sector providers.
  - Consider structural reforms to increase labor market flexibility:
    - Reduce firing costs for regular workers while promoting mid-career recruitment to raise job mobility.
    - Recognize trade-offs: lower firing costs may increase flows into unemployment in the short term but can increase long-run job mobility and potentially overall job security (flexicurity), especially if coupled with safety nets (e.g., severance payments).
  - Further research recommended to assess applicability of evidence from other advanced countries to Japan and to quantify net effects of reforms on flows between employment and unemployment.

*Source: _wp13196 (IMF Working Paper).*

### References .............................................................................................................

### _wp13196 - References

### I. Introduction
- The 10-year average unemployment rate in Japan rose from 2.5 percent in the 1980s, to 3 percent in the 1990s, and to 4.7 percent in the 2000s.
- Japan’s unemployment rate stayed well below 4 percent prior to the 1997 recession and rose by 1.2 percent to around 5 percent since then.
- Between 2006 and 2009, the change in unemployment rate was: U.K.: +2.1%; U.S.: +4.7%; Japan: +1.2%.
- The Beveridge curve in Japan shifted outward since the late 1990s, suggesting increased labor market mismatch (i.e., unemployed workers not matched to existing job vacancies owing to skill and other mismatches).
- Objectives of the paper:
  - Calculate mismatch across age groups, occupation, and employment-type (full time versus part time) in Japan for April 2000–April 2013 using the methodology of Sahin et al. (2013).
  - Measure how much the types of mismatch contributed to the rise in the Japanese unemployment rate during the global financial crisis.

### Key empirical findings (overview)
- Age mismatch has steadily declined.
- Occupational mismatch and employment-type mismatch display countercyclical patterns.
- Occupational mismatch increased sharply by about 1.5 to 2-fold during the 2008 global financial crisis and has stayed relatively high since then.
- Occupational mismatch accounted for a significant portion (20-40 percent) of the recent rise in the unemployment rate.
- The magnitude of occupational mismatch in Japan is comparable to that in the U.K. (Patterson et al. 2013) and the U.S. (Sahin et al. 2013).

### Robustness and limitations
- Analysis accounts for heterogeneity in matching efficiency across markets (e.g., different occupations).
- Analysis assumes homogeneity in productivities and job separation rates across submarkets due to data limitations.
- U.S. evidence in Sahin et al. (2013) suggests effects of these simplifying assumptions are rather small.
- Additional dimensions (male vs. female, regular vs. nonregular, prefectures) could not be analyzed due to hiring data limitations.

### Policy implications (mentioned)
- Policy measures to address occupational mismatch could build on existing training programs and the matching process.
- Structural reforms to promote labor market flexibility could increase job mobility across occupations over the long term.
- Government policies targeted at reducing occupational mismatch could lower the recent increase in the unemployment rate and potentially raise the economy’s growth potentials.
- The paper does not provide definitive policy recommendations but highlights potential directions.

### II. Literature
- Ito et al. (2009): studied the Japanese Beveridge curve and suggested mismatches increased over time but did not quantify levels and dimensions.
- Higuchi et al. (2012), Sato et al. (2006): used a conventional mismatch index (sum of absolute distance of vacancy and unemployment share across markets); found excess vacancies in technical and specialized occupations in major metropolitan prefectures.
- Limitations of the conventional mismatch index: assumes homogenous matching efficiencies across markets and cannot be converted to implied loss in unemployment rates.
- Sahin et al. (2013): methodology used to measure mismatches as deviations of observed labor market outcomes from socially efficient outcomes; found occupational mismatch accounted for up to 1/3 of the rise in the U.S. unemployment rate during the Great Recession and was more severe than geographic mismatch.
- Patterson et al. (2013): applying Sahin et al. methodology to the U.K., found occupational mismatch accounted for 1/4-1/3 of the recent rise in the unemployment rate.

### III. Data
- Aggregate unemployment rate: Ministry of Internal Affairs and Communications (MIAC), Japan.
- Job separation rate: Monthly Labour Survey by the Ministry of Health, Labour and Welfare (MHLW).
- Group-level unemployment, vacancies, and hires: Employment Referrals for General Workers (survey by MHLW at public employment security offices).
- Occupational classification: nine disaggregated occupation levels based on the 1999 revision; data cover April 2000–April 2013. Robustness check using 65 occupational categories.
- Age brackets for age mismatch: six 10-year groups: 15-24, 25-34, 35-44, 45-54, 55-64, and 65 years old and over. Age data cover January 1996–April 2013.
- Full-time vs. part-time data date back to January 1972 (Appendix A1).
- Availability of disaggregate labor market data limits main analysis to April 2000–April 2013.

### Table 1 — Monthly averages and shares (April 2000–April 2013)
- Total (Monthly Average) (level):
  - Hires: 151,998
  - Unemployed: 2,324,788
  - Vacancies: 1,643,031

- Hires / Unemployed / Vacancies — Shares (Percent) by Age:
  - 15-24 years old: 17.3 / 12.6 / 22.1
  - 25-34 years old: 29.2 / 27.7 / 27.0
  - 35-44 years old: 21.7 / 19.0 / 21.1
  - 45-54 years old: 16.9 / 17.4 / 14.6
  - 55-64 years old: 13.4 / 20.6 / 10.9
  - 65 years old and over: 1.6 / 2.7 / 4.3

- Hires / Unemployed / Vacancies — Shares (Percent) by Employment Type:
  - Full-Time: 68.3 / 62.3 / 54.2
  - Part-Time: 31.7 / 37.7 / 45.8

- Hires / Unemployed / Vacancies — Shares (Percent) by Occupation:
  - Production processing: 34.1 / 31.8 / 26.7
  - Clerks: 19.2 / 9.0 / 10.6
  - Professionals/Engineers: 16.3 / 13.4 / 23.2
  - Sales: 10.5 / 12.7 / 15.5
  - Services: 10.2 / 7.2 / 15.0
  - Transportation, information, communications: 5.9 / 4.1 / 5.4
  - Defense: 2.5 / 0.8 / 2.9
  - Agricultural, forestry, fishery: 1.0 / 0.5 / 0.5
  - Managers: 0.2 / 0.4 / 0.2

- Observations from shares:
  - On average, there are 2.3 million unemployed workers and 1.6 million vacancies per month, resulting in about 0.15 million monthly hires.
  - Younger age groups (15-24, 25-34, 35-44 years old) have a higher share of hires relative to their share of unemployment compared with older age groups (45-54, 55-64, 65+), implying younger job seekers are hired disproportionately.
  - Full-time workers are disproportionately hired more than part-time workers: while 62.3 percent of unemployed workers are looking for full-time positions, 68.3 percent of hires are full-time.

### IV. Organization of the paper (structure)
- Section II: literature review.
- Section III: data description.
- Section IV: empirical strategy to measure mismatch and its contribution to unemployment.
- Section V: empirical results.
- Section VI: conclusions and discussion of potential policy measures.

*Source: IMF working paper content (April 2000–April 2013 data coverage as described in the text).*

### 68.3 percent of the hirings are for full-time workers.  Lastly, some occupation types

### _wp13196 - 68.3 percent of the hirings are for full-time workers.  Lastly, some occupation types

### Empirical strategy
- Two-step strategy:
  - Measure mismatch in the Japanese labor market across occupation types, age groups, and full-time or part-time jobs following Sahin et al. (2013).
  - Measure the contribution of mismatch to changes in the unemployment rate by computing counterfactual unemployment in the absence of mismatch.
- Matching framework:
  - Island-specific matching function: h_it(v_it,u_it) = Φ_t φ_it v_it^{α} u_it^{1−α}, where φ_it is island-specific matching efficiency, Φ_t is aggregate matching efficiency, v_it vacancies, u_it unemployed.
  - Vacancy share, α, is set to 0.4.
  - Under homogenous productivities and separations, efficient allocation satisfies φ_it (v_it / u*_it)^{α} = φ_jt (v_jt / u*_jt)^{α}.
- Mismatch indices:
  - Theory-based indices:
    - M_{φt} = (h*_t − h_t)/h*_t = 1 − ∏_{i=1}^I [ (φ̄_t / φ_it) (v_it / v_t)^{α} (u_it / u_t)^{1−α} ] where φ̄_t = [ ∑_{i∈I} φ_it^{1/α} (v_it / v_t) ]^{α}.
    - M_t (no heterogeneity in matching efficiency): M_t = 1 − ∏_{i=1}^I (v_it / v_t)^{α} (u_it / u_t)^{1−α}.
  - Conventional statistical mismatch index: M_{st} = 1/2 ∑_{i∈I} | v_it / v_t − u_it / u_t | (Jackman and Roper, 1987).
  - The indices measure fractions of hires lost because of mismatch.
- Data requirements: vacancies (v_i), hires (h_i), unemployment (u_i); market-specific matching efficiencies φ_it and vacancy share α estimated via matching functions (Appendix A3).

### Mismatch unemployment (counterfactual method)
- Counterfactual unemployment law of motion:
  - u^*_{t+1} = (1 − s_t − f^*_t) u^*_t + s_t, where s_t is observed job separation rate and f^*_t is counterfactual job-finding rate.
  - f^*_t = f_t / [1 − M_{φt} (u_t / u^*_t)^{α}], where f_t is observed job-finding rate.
- Since M_{φt} ∈ (0,1) and u_t / u^*_t > 1, f^*_t > f_t, implying counterfactual unemployment is lower than observed. The gap u_t − u^*_t is mismatch unemployment.

### Key empirical results
- General findings:
  - 68.3 percent of the hirings are for full-time workers.
  - Age mismatch declined steadily since 2001.
  - Employment-type mismatch is countercyclical and increased during/after the 2008 global financial crisis.
  - Occupational mismatch rose substantially during the 2008 recession and remained higher than pre-recession levels.
- Age mismatch:
  - In 2000, around 9 percent of hires were lost due to age mismatch; by 2007 this decreased to around 2.5 percent.
  - The impact of legislation in 2001 and further amendment in 2007 contributed to declines in age mismatch.
- Employment-type mismatch:
  - Peaked at 7.4 percent in April 2002.
  - Dropped sharply from 6 percent to 1 percent in 2004; stayed around 0.5 percent until increasing during the 2008 crisis and remaining around 3 percent thereafter.
  - The 2004 legislation extending maximum contract length for dispatched workers from one to three years potentially contributed to the decline.
- Occupational mismatch:
  - Based on 9 occupational categories: average 6.7 percent in 2006 rising to a peak of 12.9 percent in February 2009; average since the 2008 recession remained high at 10 percent.
  - Based on 65 occupation categories: rose from the 2006 average of 19 percent to 26 percent in February 2009 (increase of 7 percent).
  - Disaggregate index (65 categories) is always higher than aggregate 9-category index, but both show similar patterns.
- Contributions to the rise in the unemployment rate between 2006 and July 2009 (Change in Unemployment Rate = 1.38 percent):
  - Age mismatch:
    - M_{φ}: −0.02 percent; Percentage explained: 1.7
    - M: 0.01 percent; Percentage explained: 0.9
    - Text note: contribution ranged between −0.02 to 0.01 percent out of 1.38 percent.
  - Full and Part Time mismatch:
    - M_{φ}: 0.19 percent; Percentage explained: 13.9
    - M: 0.06 percent; Percentage explained: 4.0
  - Occupation (9 categories):
    - M_{φ}: 0.37 percent; Percentage explained: 26.7
    - M: 0.32 percent; Percentage explained: 23.2
  - Occupation (65 categories):
    - M_{φ}: 0.58 percent; Percentage explained: 41.8
    - M: 0.32 percent; Percentage explained: 23.2
- Interpretation and robustness:
  - Occupational mismatch accounted for approximately 20-40 percent of the rise in Japanese unemployment during 2008-2009.
  - Results robust to vacancy share α in range 0.3 to 0.5: when α ranges from 0.3 to 0.5, age, employment-type, and occupational mismatch explain 1.0-2.7 percent, 12.1-15.9 percent, and 24.3-32.2 percent of the rise in the unemployment rate, respectively.
  - Comparison with U.K. and U.S.: occupational mismatch in Japan shows a similar countercyclical pattern and comparable magnitude of contribution to unemployment increases (U.S. contribution: 11.1–29.3 percent depending on disaggregation; U.K.: 17.1–26.9 percent depending on disaggregation). Note: cross-country comparisons are sensitive to disaggregation levels.

### Descriptive matching and labor market measures (selected statistics)
- Matching function and parameters:
  - Aggregate matching: h_t = Θ_t v_t^{α} u_t^{1−α}.
  - Job-finding rate: f_t = h_t / u_t = Θ_t (v_t / u_t)^{α}.
  - Vacancy yield: y_t = h_t / v_t = Θ_t (u_t / v_t)^{1−α}.
  - Estimated vacancy share for Japanese data around 0.4 (0.34 based on GMM; other estimates higher). Sahin et al. (2013) and Patterson et al. (2013) use α = 0.5 for U.S. and U.K.
- Job-finding and vacancy-yield patterns (April 2000–April 2013, summarized):
  - Age groups:
    - Youngest (15-24): highest job-finding rates; vacancy-unemployment ratios generally highest.
    - Oldest (65+): large increase in vacancy-unemployment ratio between 2001 and 2008 but lower job-finding probabilities; job-finding rates for older groups improved modestly.
    - Job-finding rates for 15-54 are more cyclical than for 55-64 and 65+.
  - Full-time vs Part-time:
    - Probability of finding a job higher for part-time workers than full-time workers historically.
    - Job-finding rate for part-time declined after 2004 and remained stable through the 2008 recession; probabilities for part-time and full-time became almost the same after the global financial crisis.
    - Vacancy yields: easier to fill full-time positions than part-time positions from firms’ perspective.
  - Occupations:
    - High job-finding rates in production processing occupations and agricultural, forestry, and fishery workers.
    - Higher vacancy yields in agricultural, forestry and fishery occupations and clerical positions.
- Data sources and notes:
  - Unemployed, vacancies, and hires by occupation, age, and employment type from monthly Employment Referrals for General Workers (Ippan Shokugyo Shokai Jokyo), MHLW (excludes newly graduated college workers).
  - The survey’s unemployed count represents approximately 80 percent of the number reported by the Cabinet Office for April 2000–April 2013.
  - Employment-type data series were seasonally adjusted in source; other series seasonally adjusted using X-12-ARIMA by the author.
  - Part-time workers: hours worked less than regular workers at same establishment; definitions per MHLW.
  - Job separation rate from MHLW’s Monthly Labour Survey (about 33,000 establishments).

### Policy conclusions and recommendations
- Main conclusion: labor market mismatch is a significant issue in Japan; occupational mismatch in particular contributed substantially (20-40 percent) to the rise in unemployment during 2008-2009.
- Policy implications:
  - Focus policy efforts on reducing occupational mismatch to limit rises in unemployment.
  - Strengthen and enhance training programs to help unemployed workers acquire skills necessary for in-demand occupations; build on existing programs and public employment security offices.
  - Improve job matching processes and platforms; consider opening job-matching services to private sector providers.
  - Consider structural reforms to increase labor market flexibility:
    - Reduce firing costs for regular workers while promoting mid-career recruitment to raise job mobility.
    - Recognize trade-offs: lower firing costs may increase flows into unemployment in the short term but can increase long-run job mobility and potentially overall job security (flexicurity), especially if coupled with safety nets (e.g., severance payments).
  - Further research recommended to assess applicability of evidence from other advanced countries to Japan and to quantify net effects of reforms on flows between employment and unemployment.

*Source: _wp13196 (IMF Working Paper).*

### Part vs Full-Time Mismatch

### Part vs Full-Time Mismatch

### Definitions and identities
- Observed sector-specific job separation rates:
  - (1−s_i) ≡ (1−∆)(1−δ_i)
  - where ∆ is the aggregate job destruction rate and δ_i is the sector-specific job destruction rate.
- In Figure 11:
  - Simple Mismatch (orange line) shows M_s series.
  - Baseline Mismatch (red line) shows M series.
  - Mismatch (green line) shows M_φ series.

### Three measures of mismatch (M, M_s, M_φ)
- The three measures plotted for comparison:
  - Simple Mismatch Index based on equation (6) (M_s) — i.e., the measure of mismatch used by statistical agencies in Japan.
  - Baseline Mismatch in equation (5) (M) — i.e., mismatch in the absence of heterogeneity in matching efficiencies.
  - Mismatch Index in equation (4) (M_φ).
- Key observations about the three measures:
  - The three measures (M_φ, M, M_s) generally capture similar trends in mismatch in Japan.
  - The preferred occupational mismatch index, M_φ, captures the rise in mismatch after the 2011 earthquake in Japan while the simple mismatch index, M_s, fails to do so.
  - M_φ offers a clear interpretation: a fraction of hires lost owing to mismatch.

*Source: _wp13196 - Part vs Full-Time Mismatch*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13196.pdf_