## 1. Skill Distance Across Occupations

## Source details

**Canonical URL:** [1. Skill Distance Across Occupations](https://www.imf.org/-/media/files/publications/cr/2022/english/1macea2022002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2022/english/1macea2022002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2022/english/1macea2022002.pdf.json)

---

### Introduction and context
- Macao SAR’s economy heavily relies on the gaming sector:
  - In 2019, the gaming sector is estimated to account for around 50 percent of GDP and 22 percent of employment.
  - Since the pandemic, gaming tourism has been depressed, resulting in a 54 percent decline in economic activity during 2020.
  - Government revenues from the gaming sector historically account for about 80 percent of total revenues and plummeted during the COVID-19 crisis.
- Government diversification (GMICZ Master Plan) targets four nongaming sectors:
  - (i) scientific and technological research;
  - (ii) traditional Chinese medicine;
  - (iii) tourism, convention, and exhibition services (MICE);
  - (iv) modern financial services.
- Chapter approach (four steps):
  - assess Macao SAR’s current labor market structure and occupational composition;
  - compare current skills availability with skill requirements of targeted sectors;
  - estimate costs of moving workers from current occupational structure to target one;
  - derive policy implications to address potential skills mismatch.

### Macao SAR labor market structure — key statistics
- Overall occupational composition (2020):
  - Clerks: 26%
  - Service and sales workers: 20%
  - Hawkers, cleaners and custodians: 18%
  - Technicians and associate professionals: 12%
  - Administrators & Managers: 8%
  - Craftsmen and similar workers: 7%
  - Professionals: 5%
  - Machine operators, drivers and assemblers: 4%
- Gaming sector occupational composition:
  - About 84 percent are clerks and service and sales workers, and 6 percent low-skilled workers.
- Outside the gaming sector:
  - Only about 14 percent of the currently employed workers were high-skilled in 2020.
- International comparison:
  - In the United States, about 27 percent of the workers are high skilled.

### Skill requirements of targeted sectors (occupational composition)
- Scientific and technological research (high-tech):
  - Professionals: about 55%
  - Technicians: about 15%
- Financial sector:
  - Clerks: 50%
  - Professionals: 28%
- Traditional Chinese medicine (benchmarked to pharmaceutical and medicine sector):
  - Machine operators, drivers and assemblers: 37%
  - Professionals: 26%
- MICE sector:
  - Service and sales workers: about 57%
- Sectoral overlap and distance:
  - MICE demonstrates the closest occupational composition to Macao SAR’s current composition.
  - Financial services and gaming overlap on clerks, but financial services require a much higher proportion of professionals.
  - Scientific and technological research and traditional Chinese medicine are furthest from Macao SAR’s overall occupational composition due to high professional and manufacturing worker needs.

### Skill distance and implications for occupational mobility
- Skill distance measure:
  - Index range: 0 = identical skill sets, 1 = entirely different skill sets (constructed following Gathmann and Schonberg (2010)).
- Specific distances cited:
  - Skill distance between hawkers, cleaners, and custodians and professionals: about 0.67.
  - Average skill distance between low-skill-requiring occupations and professionals: about 0.73.
  - Average skill distance between middle-skill-requiring occupations and professionals: about 0.44.
- Interpretation:
  - Distances indicate substantial skill mismatches between Macao SAR’s current labor supply and the labor demand of targeted sectors.
  - Important skills for professionals include critical thinking, social perceptiveness, active listening, reading comprehension, and speaking.
  - Important skills for hawkers, cleaners, and custodians include active listening, service orientation, coordination, monitoring, and judgment and decision making.

### Mobility costs and training time required
- Monetary mobility costs (summary):
  - On average, it costs about 30 percent of the average annual wage for middle-skilled workers to move to professionals.
  - On average, it costs about 40 percent of the average annual wage for low-skilled workers to move to professionals.
  - Cross-occupation mobility costs depend on skill distance, worker's initial skill level, and destination occupation’s entry cost.
- Time to achieve occupational mobility through skill upgrading:
  - About two years of additional training for middle-skilled workers to move to professionals.
  - About three years of additional training for low-skilled workers to move to professionals.
- Estimation approach:
  - Mobility costs between occupations are estimated following Cortes and Gallipoli (2016), based on a model of occupational choice delivering a gravity equation linking labor flows to occupation skill distance and transition costs.

### Policy recommendations to overcome barriers to diversification
- Invest in nurturing talent and scale up public spending on education to achieve skill upgrading:
  - Emphasize outcome-oriented occupational and on-the-job training.
  - Strengthen cooperation with employers to ensure training aligns with demand.
  - Achieve greater synergy between higher education and future skill needs; strengthen science-business linkages.
- Complement domestic skill upgrading with talent attraction:
  - Recruit high-skilled non-resident workers to build a deeper pool of qualified workers and facilitate knowledge spillovers.
  - Streamline administrative barriers for hiring high-skilled non-resident workers.
- Leverage integration with the Greater Bay Area (GBA):
  - Access a large pool of skilled labor in the GBA and offshore some economic activities to capitalize on the Mainland’s comparative advantage in manufacturing.

---

### K-means clustering, skill-distance construction, and occupational mobility model

### K-means clustering algorithm and cluster fit
- Implementation details:
  - Initialize cluster centroids 푚푚1, 푚푚2, ..., 푚푚퐺퐺 ∈ 푅푅푁푁 for each cluster.
  - Assign each sector 푘푘 to cluster whose centroid is closest to 푥푥(푘푘): 푐푐(푘푘) = argmin_{g∈{1,2,...,G}} ||푥푥(푘푘) − 푚푚g||^2.
  - Replace centroid 푚푚g by coordinate-wise average: 푚푚̂g = (∑_{k=1}^{M} 1(푐푐(푘푘)=g) 푥푥(푘푘)) / (∑_{k=1}^{M} 1(푐푐(푘푘)=g)).
  - Iterate until convergence.
- Preprocessing and initialization:
  - Normalize each characteristic to have zero mean and unit variance.
  - Use the “k-means ++” algorithm (Arthur and Vassilvitskii (2006)) for initial seeds and perform extensive checks with alternative starting points.
- Cluster-fit measures:
  - Within sum of squares WSS and log(WSS).
  - 휂^2 coefficient (proportional reduction of WSS).
  - Proportional reduction of error (PRE).
- Key finding on optimal k:
  - The analysis suggests k = 8 to be a good value.
  - At k = 8, there is a kink in WSS and log(WSS).
  - 휂^2 coefficient points to a reduction of the WSS by 80% and PRE decreased by about 30% compared with k = 7.

### Clustering results and sector classification
- Data and outcome:
  - 250 sectors (4-digit NAICS) are grouped into 8 clusters using k-means.
  - Representative sectors per group are those closest in occupational composition to the within-group average.
- Notable assignments:
  - MICE is classified into Group 6 (service sectors) — implies Macao SAR’s composition is suitable to develop MICE.
  - Financial services and gaming are both in Group 2 (substantial clerks requirement).
    - Gaming: most needed clerks are gambling cage workers.
    - Financial services: essential clerks include tellers in credit intermediation and bill and account collectors; financial sectors also need many professionals such as lawyers.
  - Scientific and technological research and traditional Chinese medicine fall into Group 8 — require professionals and manufacturing workers.
- Occupational categories used in visualization:
  - Hawkers, cleaners and custodians; Technicians; Agricultural and fishery workers; Service and sales workers; Professionals; Machine operators, drivers and assemblers; Craftsmen and similar workers; Clerks; Administrators & Managers.

### Skill distance: construction and data
- Distance measure (angular separation following Gathmann and Schonberg (2010)):
  - Ang_{WAp}_{ij} = (∑_{d=1}^{D} W_{i d} × W_{j d}) / ( (∑_{d=1}^{D} W_{i d}^2 × ∑_{d=1}^{D} W_{j d}^2)^{1/2} ).
  - Task distance d_{a j}^{task} = 1/2 (1 − Ang_{WAp}_{ij} ).
- Data sources and dimensions:
  - Skill vectors taken from O*Net database.
  - Total number of skill dimensions included: 35.
  - Skill importance scale: 100 to 0 (100 most important; 0 least important).
- Aggregation level:
  - Bilateral skill distance computed at the 2-digit SOC level.
  - A concordance maps Macao SAR occupational classification (9 broad categories) to SOC.

### Occupational mobility cost: model and specification
- Gravity model linking labor flows to skill distance and transition costs:
  - ln(X_{a j} / X_{j a}) = ln T_a + θ ln P_a − ln T_j − θ ln P_j − θ_{a j} ln d_{a j}
    - Where X_{a j} is flow of workers from occupation a to j; T and P are occupation-specific characteristics; d_{a j} is cost of switching occupation.
- Specification of occupation mobility cost:
  - ln d_{a j} = β_1 ln d_{a j}^{task} + β_2 λ_{a j}^{NN} + β_3 λ_{a j}^{RC} + β_4 λ_{a j}^{RM} + β_5 λ_{a j}^{NM} + m_j + ε_{a j}
    - λ dummies equal one if occupations are in different broad task groups and destination occupation is nonroutine cognitive (NN), routine cognitive (RC), routine manual (RM), or nonroutine manual (NM).
    - m_j is destination dummy.
- Data for flows:
  - Flows measured using matched monthly data from the Current Population Survey (CPS) at 2-digit SOC level.

### Converting mobility costs to monetary and education units
- Monetary measure:
  - M_lM_{a j}^{C_ldt} = W_j × (β_1 ln d_{a j}^{task} − 1)
    - W_j is destination occupation’s annual wage (from DSEC).
    - (β_1 ln d_{a j}^{task} − 1) is the payoff workers need to give up when moving from occupation a to j.
- Educational (training years) measure:
  - M_lM_{a j}^{C_ldt} = P_j − P_a / (β_1 ln d_{a j}^{task})
    - P_j is (extra) years of schooling observed in destination occupation j (education requirement from DSEC minus 12 years of compulsory education).

### Empirical estimates and examples (mobility cost magnitudes)
- Computations at 2-digit SOC; selected examples from Annex Table 4:
  - Lowest overall mobility costs (selected entries):
    - Management related occupations → Executives, administrators and managers: Skill Distance 1.037; Costs All Costs 2.768
    - Information and records processing, except financial → Other administrative support occupations, including clerical: Skill Distance 1.021; Costs All Costs 2.777
    - Financial records processing occupations → Other administrative support occupations, including clerical: Skill Distance 1.046; Costs All Costs 2.844
    - Secretaries, stenographers, and typists → Other administrative support occupations, including clerical: Skill Distance 1.061; Costs All Costs 2.884
    - Librarians, social scientists, religious workers → Executives, administrators and managers: Skill Distance 1.081; Costs All Costs 2.887
  - Highest overall mobility costs (selected entries):
    - Machine operators and tenders, not precision → Lawyers and judges: Skill Distance 2.747; All Costs 96.289
    - Fabricators, assemblers and hand working occupations → Lawyers and judges: Skill Distance 2.677; All Costs 93.826
    - Helpers, construction and production occupations → Lawyers and judges: Skill Distance 2.669; All Costs 93.577
    - Other precision production occupations → Lawyers and judges: Skill Distance 2.622; All Costs 91.911
    - Freight, stock and material handlers → Lawyers and judges: Skill Distance 2.552; All Costs 89.477
- Interpretation example:
  - The estimated cost of 1.037 associated with skill costs for the move from “Management related occupations” to “Executives, administrators and managers” implies:
    - A mover’s payoff would be 3.7 percent higher if there were no costs associated with the skill content of occupations.
    - Overall, the payoff to a worker moving between these two occupations would be more than 2.7 times higher if all mobility costs were removed.
- General finding:
  - Estimated mobility costs are substantial, even across occupations with relatively high labor flows.
  - Moves into occupations with high skill-independent entry costs (e.g., lawyers and judges) are associated with prohibitively high estimated mobility costs and very few observed transitions.

### Data annexes and occupational detail
- Annexes include:
  - Annex Table 1: Representative sectors within each group (three representative sectors per group).
  - Annex Table 2: List of 35 skill set dimensions from O*Net (Active Learning; Operation and Control; Active Listening; Operations Analysis; ...; Writing).
  - Annex Table 3: Occupation pairs with the closest and farthest skill distance.
  - Annex Table 5 and 6: Detailed classification of clerks and financial clerks (SOC codes and labels, including 43-3099.00 Financial Clerks, All Other; 43-9051.00 Office Clerks, General; 43-3031.00 Bookkeeping, Accounting, and Auditing Clerks; 43-3040 Gambling Cage Workers; 43-3070 Tellers; etc.).
- Aggregation and concordance:
  - Macao SAR classification aggregated into 9 broad categories and manually concorded with SOC to allow aggregation from SOC to Macao SAR occupational categories.

*Source: IMF staff calculations and analysis in "How to Overcome the Barriers to Economic Diversification: A Labor Market Approach," March 14, 2022.*

### 1. Skill Distance Across Occupations __________________________________________________ 6

### 1. Skill Distance Across Occupations

### Introduction and context
- Macao SAR’s economy heavily relies on the gaming sector:
  - In 2019, the gaming sector is estimated to account for around 50 percent of GDP and 22 percent of employment.
  - Since the pandemic, gaming tourism has been depressed, resulting in a 54 percent decline in economic activity during 2020.
  - Government revenues from the gaming sector historically account for about 80 percent of total revenues and plummeted during the COVID-19 crisis.
- The government’s diversification plan (GMICZ Master Plan) targets development of four nongaming sectors:
  - (i) scientific and technological research;
  - (ii) traditional Chinese medicine;
  - (iii) tourism, convention, and exhibition services (MICE);
  - (iv) modern financial services.
- Chapter approach (four steps):
  - assess Macao SAR’s current labor market structure and occupational composition;
  - compare current skills availability with skill requirements of targeted sectors;
  - estimate costs of moving workers from current occupational structure to target one;
  - derive policy implications to address potential skills mismatch.

### Macao SAR labor market structure — key statistics
- Overall occupational composition (2020):
  - Clerks: 26%
  - Service and sales workers: 20%
  - Hawkers, cleaners and custodians: 18%
  - Technicians and associate professionals: 12%
  - Administrators & Managers: 8%
  - Craftsmen and similar workers: 7%
  - Professionals: 5%
  - Machine operators, drivers and assemblers: 4%
- Gaming sector occupational composition:
  - About 84 percent are clerks and service and sales workers, and 6 percent low-skilled workers.
- Outside the gaming sector:
  - Only about 14 percent of the currently employed workers were high-skilled in 2020.
- International comparison:
  - In the United States, about 27 percent of the workers are high skilled.

### Skill requirements of targeted sectors (occupational composition)
- Scientific and technological research (high-tech):
  - Professionals: about 55%
  - Technicians: about 15%
- Financial sector:
  - Clerks: 50%
  - Professionals: 28%
- Traditional Chinese medicine (benchmarked to pharmaceutical and medicine sector):
  - Machine operators, drivers and assemblers: 37%
  - Professionals: 26%
- MICE sector:
  - Service and sales workers: about 57%
- Sectoral overlap and distance:
  - MICE demonstrates the closest occupational composition to Macao SAR’s current composition.
  - Financial services and gaming overlap on clerks, but financial services require a much higher proportion of professionals.
  - Scientific and technological research and traditional Chinese medicine are furthest from Macao SAR’s overall occupational composition due to high professional and manufacturing worker needs.

### Skill distance and implications for occupational mobility
- Skill distance measure (index range: 0 = identical skill sets, 1 = entirely different skill sets) constructed following Gathmann and Schonberg (2010).
- Specific distances cited:
  - Skill distance between hawkers, cleaners, and custodians and professionals: about 0.67.
  - Average skill distance between low-skill-requiring occupations and professionals: about 0.73.
  - Average skill distance between middle-skill-requiring occupations and professionals: about 0.44.
- Interpretation:
  - These distances indicate substantial skill mismatches between Macao SAR’s current labor supply and the labor demand of targeted sectors.
  - Important skills for professionals include critical thinking, social perceptiveness, active listening, reading comprehension, and speaking.
  - Important skills for hawkers, cleaners, and custodians include active listening, service orientation, coordination, monitoring, and judgment and decision making.

### Mobility costs and training time required
- Monetary mobility costs (summary):
  - On average, it costs about 30 percent of the average annual wage for middle-skilled workers to move to professionals.
  - On average, it costs about 40 percent of the average annual wage for low-skilled workers to move to professionals.
  - Cross-occupation mobility costs depend on skill distance, worker's initial skill level, and destination occupation’s entry cost.
- Time to achieve occupational mobility through skill upgrading:
  - About two years of additional training for middle-skilled workers to move to professionals.
  - About three years of additional training for low-skilled workers to move to professionals.
- Additional notes on estimation:
  - Mobility costs between occupations are estimated following the strategy implemented by Cortes and Gallipoli (2016), based on a model of occupational choice delivering a gravity equation linking labor flows to occupation skill distance and transition costs.

### Policy recommendations to overcome barriers to diversification
- Invest in nurturing talent and scale up public spending on education to achieve skill upgrading.
  - Emphasize outcome-oriented occupational and on-the-job training.
  - Strengthen cooperation with employers to ensure training aligns with demand.
  - Achieve greater synergy between higher education and future skill needs; strengthen science-business linkages.
- Complement domestic skill upgrading with talent attraction:
  - Recruit high-skilled non-resident workers to build a deeper pool of qualified workers and facilitate knowledge spillovers.
  - Streamline administrative barriers for hiring high-skilled non-resident workers.
- Leverage integration with the Greater Bay Area (GBA):
  - Access a large pool of skilled labor in the GBA and offshore some economic activities to capitalize on the Mainland’s comparative advantage in manufacturing.

*Source: IMF staff calculations and analysis in "How to Overcome the Barriers to Economic Diversification: A Labor Market Approach," March 14, 2022.*

### 1. Initialize cluster centroids 푚푚

### 1. Initialize cluster centroids 푚푚

### K-means clustering algorithm (implementation details)
- Steps implemented:
  - Initialize cluster centroids 푚푚1, 푚푚2, ..., 푚푚퐺퐺 ∈ 푅푅푁푁 for each cluster.
  - Assign each sector 푘푘 to cluster whose centroid is closest to 푥푥(푘푘). Cluster assignment is 푐푐(푘푘) ∈ {1,2, ..., 퐺퐺}, 푐푐(푘푘) = argmin_{g∈{1,2,...,G}} ||푥푥(푘푘) − 푚푚g||^2.
  - Replace cluster centroid 푚푚g by the coordinate-wise average of all points in the gth cluster: 푚푚̂g = (∑_{k=1}^{M} 1(푐푐(푘푘)=g) 푥푥(푘푘)) / (∑_{k=1}^{M} 1(푐푐(푘푘)=g)).
  - Iterate steps 2 and 3 until convergence.
- Preprocessing and initialization:
  - Normalize each characteristic to have zero mean and unit variance.
  - Use the “k-means ++” algorithm (Arthur and Vassilvitskii (2006)) for initial seeds and perform extensive checks with alternative starting points.

### Choosing the optimal number of clusters (cluster-fit measures)
- Measures used to evaluate cluster fit:
  - Within sum of squares WSS and its logarithm log(WSS).
  - 휂휂^2 coefficient (measures the proportional reduction of the WSS for each cluster solution k compared with the total sum of squares).
  - Proportional reduction of error (PRE) coefficient (P R E (k) illustrates the proportional reduction of the WSS for cluster solution k compared with the previous solution with k−1 clusters).
- Key findings on optimal k:
  - The analysis suggests k = 8 to be a good value.
  - At k = 8, there is a kink in WSS and log(WSS).
  - 휂휂^2 coefficient points to a reduction of the WSS by 80% and PRE decreased by about 30% compared with k = 7.

### Clustering results and sector classification
- Data and clustering outcome:
  - 250 sectors (4-digit NAICS) are grouped into 8 clusters using k-means (MacQueen et al., 1967).
  - Representative sectors for each group are those with closest occupational composition to the within-group average.
- Notable group assignments and implications:
  - MICE is classified into Group 6, which contains a group of service sectors. The classification suggests Macao SAR’s current composition is suitable to develop the MICE sector as it meets the large demand of service and sales workers to develop the MICE sector.
  - Financial services and gaming sectors are both classified into Group 2; both require a substantial proportion of clerks.
    - In the Gaming sector, the most needed clerks are gambling cage workers.
    - In financial services, the most essential clerks are tellers in credit intermediation and related activities, and bill and account collectors in the non-depository credit intermediation sector. Financial sectors also need many professionals such as lawyers.
  - Scientific and technological research and traditional Chinese medicine sectors fall into Group 8; both require significant support from professionals and manufacturing workers.
- Occupational composition visualization:
  - Occupational categories include: Hawkers, cleaners and custodians; Technicians; Agricultural and fishery workers; Service and sales workers; Professionals; Machine operators, drivers and assemblers; Craftsmen and similar workers; Clerks; Administrators & Managers.
  - The clustering results produce groups with salient occupational-share features.

### Skill distance: construction and data
- Distance measure:
  - Angular separation between occupation skill vectors (following Gathmann and Schonberg (2010)):
    - Ang_{WAp}_{ij} = (∑_{d=1}^{D} W_{i d} × W_{j d}) / ( (∑_{d=1}^{D} W_{i d}^2 × ∑_{d=1}^{D} W_{j d}^2)^{1/2} ).
    - Task distance d_{a j}^{task} = 1/2 (1 − Ang_{WAp}_{ij} ).
  - Skill vectors taken from O*Net database.
  - Total number of skill dimensions included: 35.
  - Skill importance scale: 100 to 0 (100 most important; 0 least important).
- Aggregation level:
  - Bilateral skill distance computed at the 2-digit SOC level to align with occupational mobility analysis.
  - A concordance maps Macao SAR occupational classification (9 broad categories) to SOC to aggregate results.

### Occupational mobility cost: model and specification
- Gravity model relating labor flows to skill distance and transition costs:
  - ln(X_{a j} / X_{j a}) = ln T_a + θ ln P_a − ln T_j − θ ln P_j − θ_{a j} ln d_{a j}
    - Where X_{a j} is flow of workers from occupation a to j; T and P are occupation-specific characteristics; d_{a j} is cost of switching occupation.
- Specification of occupation mobility cost:
  - ln d_{a j} = β_1 ln d_{a j}^{task} + β_2 λ_{a j}^{NN} + β_3 λ_{a j}^{RC} + β_4 λ_{a j}^{RM} + β_5 λ_{a j}^{NM} + m_j + ε_{a j}
    - λ dummies equal one if occupations are in different broad task groups and destination occupation is nonroutine cognitive (NN), routine cognitive (RC), routine manual (RM), or nonroutine manual (NM).
    - m_j is destination dummy.
- Data for flows:
  - Flows measured using matched monthly data from the Current Population Survey (CPS) at 2-digit SOC level.

### Converting mobility costs to monetary and education units
- Monetary measure of skill distance-related mobility cost:
  - M_lM_{a j}^{C_ldt} = W_j × (β_1 ln d_{a j}^{task} − 1)
    - W_j is destination occupation’s annual wage (from DSEC).
    - (β_1 ln d_{a j}^{task} − 1) is the payoff workers need to give up when moving from occupation a to j.
- Educational measure (additional training in years):
  - M_lM_{a j}^{C_ldt} = P_j − P_a / (β_1 ln d_{a j}^{task})
    - P_j is (extra) years of schooling observed in destination occupation j (education requirement from DSEC minus 12 years of compulsory education).

### Empirical estimates and examples (mobility cost magnitudes)
- Skill-distance and mobility cost computations at 2-digit SOC.
- Examples from Annex Table 4 (selected entries shown exactly as reported):
  - Lowest overall mobility costs (examples):
    - Management related occupations → Executives, administrators and managers: Skill Distance 1.037; Costs All Costs 2.768
    - Information and records processing, except financial → Other administrative support occupations, including clerical: Skill Distance 1.021; Costs All Costs 2.777
    - Financial records processing occupations → Other administrative support occupations, including clerical: Skill Distance 1.046; Costs All Costs 2.844
    - Secretaries, stenographers, and typists → Other administrative support occupations, including clerical: Skill Distance 1.061; Costs All Costs 2.884
    - Librarians, social scientists, religious workers → Executives, administrators and managers: Skill Distance 1.081; Costs All Costs 2.887
  - Highest overall mobility costs (examples):
    - Machine operators and tenders, not precision → Lawyers and judges: Skill Distance 2.747; All Costs 96.289
    - Fabricators, assemblers and hand working occupations → Lawyers and judges: Skill Distance 2.677; All Costs 93.826
    - Helpers, construction and production occupations → Lawyers and judges: Skill Distance 2.669; All Costs 93.577
    - Other precision production occupations → Lawyers and judges: Skill Distance 2.622; All Costs 91.911
    - Freight, stock and material handlers → Lawyers and judges: Skill Distance 2.552; All Costs 89.477
- Interpretation example:
  - The estimated cost of 1.037 associated with skill costs for the move from “Management related occupations” to “Executives, administrators and managers” implies:
    - A mover’s payoff would be 3.7 percent higher if there were no costs associated with the skill content of occupations.
    - Overall, the payoff to a worker moving between these two occupations would be more than 2.7 times higher if all mobility costs were removed.
- General findings:
  - Estimated mobility costs are substantial, even across occupations with relatively high labor flows.
  - Moves into occupations with high skill-independent entry costs (e.g., lawyers and judges) are associated with prohibitively high estimated mobility costs and very few observed transitions.

### Data annexes and occupational detail
- Annexes include:
  - Annex Table 1: Representative sectors within each group (three representative sectors per group).
  - Annex Table 2: List of 35 skill set dimensions from O*Net (Active Learning; Operation and Control; Active Listening; Operations Analysis; ...; Writing).
  - Annex Table 3: Occupation pairs with the closest and farthest skill distance (summary of closest and farthest pairs).
  - Annex Table 5 and 6: Detailed classification of clerks and financial clerks (SOC codes and labels, including 43-3099.00 Financial Clerks, All Other; 43-9051.00 Office Clerks, General; 43-3031.00 Bookkeeping, Accounting, and Auditing Clerks; 43-3040 Gambling Cage Workers; 43-3070 Tellers; etc.).
- Aggregation and concordance:
  - Macao SAR classification aggregated into 9 broad categories and manually concorded with SOC to allow aggregation from SOC to Macao SAR occupational categories.

*Source: IMF staff calculations; material excerpted from the IMF chapter on PEOPLE’S REPUBLIC OF CHINA—MACAO SPECIAL ADMINISTRATIVE REGION.*

### Introduction

### Introduction

### Overview
- Economic diversification is a key pillar of the Macao SAR government’s economic policy.  
- The COVID-19 crisis increased urgency to diversify away from the dominant gaming sector.  
- The modern financial industry is a government priority, aiming to leverage Macao SAR’s ties with Portuguese speaking countries (PSC) to occupy the Mainland-PSC niche market.

### Recent developments in Macao SAR's financial sector
- Progress in modernizing financial markets:
  - In 2018, Macao SAR’s bond market was officially established after three years of preparation.
  - As of the end of November 2021, the outstanding bond amount adds up to 249.6 billion MOP, equivalent to around US$31.1 billion, accounting for 15.1 percent of Macao SAR’s total foreign assets in banking sector, or 122.1 percent of 2020 GDP.
- MOX (Macao SAR’s first bond exchange platform) established in 2018, offering issuance, listing, registration, custody, trading, and settlement.
  - As of the end of November 2021, 86 bonds have been listed in MOX, with the total listed bond amounting to US$31 billion 121 percent of 2020 GDP.
  - Out of the 86 listed bonds, 13 bonds were issued and listed in MOX whereas the remaining 73 bonds were only listed, but not issued in MOX.
  - State owned banks and local governments from the Mainland are key issuers.
  - More than 70 percent of the total listed in MOX is in USD.
  - Bonds listed by industry: largest share—financial corporates; followed by real estate and tourism sectors.

### Comparative analysis: structural factors for financial hub development
- Framework: five categories—(i) availability of skilled labor, (ii) quality of infrastructure, (iii) reputability of a jurisdiction, (iv) business environment, and (v) measure of financial development.
- Benchmarks: Hong Kong SAR, Singapore, and the tenth and ninetieth percentiles of OECD countries.
- Data sources include: Worldwide Governance Indicators, World Bank World Development Indicators, International Labor Organization, UN ICT, Numbeo, Economic Freedom Heritage Foundation, Economist Intelligence Unit. (Annexes referenced in the chapter.)
- Key comparative findings:
  - Human capital:
    - Macao SAR has a relatively high share of tertiary educated people compared with OECD countries.
    - Finding skilled labor, particularly with specialized skills, is a challenge for Macao SAR.
  - Physical and ICT infrastructure:
    - Macao SAR has considerable room to improve physical and ICT infrastructure compared to OECD countries.
    - Cyber security preparedness is high.
  - Reputation:
    - Macao SAR performs well on safety and human development.
    - Governance, rule of law, quality of healthcare, pollution, and housing affordability have room to improve relative to OECD countries.
  - Business environment:
    - Business and labor market regulations and financial sector regulation are areas that can be further enhanced.
  - Financial sector development:
    - Despite a healthy banking system, Macao SAR lags OECD countries in local bond market capitalization, market access, and provision of non-life insurance services.

### MOX milestones (selected)
- Nov 2014: Inspection and preparation work started.
- Oct 2018: MOX was established.
- Dec 2018: LUSO International Banking Ltd publicly issued first HKD bond and Bank of China, Macao Branch listed first MOP denominated bonds.
- May 2019: first offshore RMB bond listed.
- Dec 2019: first USD denominated bond issued.
- Jan 2020: MOX signed a MoU with Luxembourg Stock Exchange in Macao.
- Dec 2020: MOX and China Central Depository & Clearing Co., Ltd. officially launched the “CCDC-MOX Bond Information Platform” in Shenzhen.

### Case studies and lessons from established financial centers
- Purpose: Distill policy lessons from Malta, Hong Kong SAR, and Singapore by examining initial conditions, policy measures, and outcomes.

- Malta
  - Initial conditions (late 1980s):
    - High-income country with solid tourism inflows and high living standards.
    - In 1988, domestic bank deposits accounted for 30 percent of GDP and private credit accounted for 23 percent of GDP.
    - Relative weaknesses vs OECD median: share of tertiary educated labor force, financial regulation, property rights, and control of corruption.
  - Implemented policies in 1990s:
    - Reputable legal framework and robust supervision: banking legislation overhaul, stock market expansion, deregulation of interest rates, abolition of offshore/onshore distinction, Malta International Business Authority (MIBA) given supervisory responsibilities.
    - Human capital: education system reforms including 1989 National Minimum Curriculum, decentralization, and labor market opening to EU citizens after 2004 accession.
    - Business environment: structural reforms removing trade barriers, deregulation, privatization, Competition Act (1995), and EU passporting rights after EU accession.
  - Results:
    - Malta narrowed gaps with OECD median in some structural factors.
    - Financial depth growth since late 1980s: domestic bank deposits relative to GDP increased more than 4 times; private credit to GDP grew 3-fold.

- Hong Kong SAR
  - Initial conditions (1970s–1980s):
    - Commercial, manufacturing, and shipping center with four stock exchanges, strong goods exports, free capital flow, and stable exchange rate.
    - Early 1990s: domestic bank deposits at 160.3 percent of GDP and private credit at 149.4 percent of GDP.
    - Outperformed OECD median on business regulation; lagged on tertiary-educated labor force share, political stability, and voice and accountability.
  - Implemented policies:
    - Financial regulation: opening domestic banking market to qualified foreign banks (from 1978), reforms aligning regulation and supervision with international standards, and a strong financial sector safety net.
    - Special bilateral financial linkage with Mainland: Mainland financial opening (1978) increased FDI and loan flows through Hong Kong SAR; various “connect schemes” established to facilitate cross-border transactions and mutual market access.
    - Infrastructure: large investments in public housing, land reclamation, and financial market infrastructure since 1950s; hosts largest FX and OTC interest derivatives market in Asia.
    - Human capital: increased public spending on education since the 1970s, universal primary education, compulsory free education up to age 15 from 1978, and access to foreign specialists via absence of employment restrictions; recent focus on fintech and green finance talent cultivation.

### Policy implications (synthesized from analysis and case studies)
- Address structural bottlenecks to advance financial sector development:
  - Improve availability of skilled and specialized labor through education and talent policies.
  - Upgrade physical and ICT infrastructure, while sustaining cyber security preparedness.
  - Strengthen governance and rule of law, healthcare quality, pollution control, and housing affordability.
  - Reform business and labor regulations to improve the business environment and enhance financial sector regulation.
  - Expand marketable debt, deepen local bond markets, and improve market access and non-life insurance provision.

*Source: 1macea2022002 - Introduction.*

### 13.      Results: Building on its initial strength, the implemented policies improved Hong Kong

### 13. Results: Building on its initial strength, the implemented policies improved Hong Kong SAR’s standing

### Hong Kong SAR — Key findings and outcomes
- Structural improvements relative to OECD median in:
  - financial sector regulation,
  - share of tertiary educated in the labor force,
  - property rights,
  - control of corruption.
- Financial deepening (1990s to 2019):
  - Domestic bank deposits: 367.8 percent of GDP in 2019 (almost doubled since the early 1990s).
  - Private credit: 235.7 percent of GDP in 2019 (almost doubled since the early 1990s).
- Institutional strengths emphasized in Hong Kong SAR’s Financial System Stability Assessment:
  - importance of the rule of law,
  - transparent and efficient institutions,
  - high-quality professional services,
  - strong financial regulatory and supervisory frameworks.
- Notes on indicators:
  - Some indicators are perception based and constructed using survey results.
  - Confidence bands cited in the source: control of corruption (8.9; 9.5) and voice and accountability (3.4; 5.7).

*Sources: ILO; Fraser institute; World Governance Indicators (WGI), D. Kaufman (Natural Resources Governance Institute and Brookings Institutions, and A. Kraay (World Bank), 2017; and EIU.*

### Comparative financial depth context
- Hong Kong SAR Financial Depth (In percent of nominal GDP) — medians for OECD countries used for comparison.
  - Domestic bank deposits: Hong Kong SAR 1991 -> Hong Kong SAR 2019 = 367.8 percent of GDP.
  - Private credit: Hong Kong SAR 1991 -> Hong Kong SAR 2019 = 235.7 percent of GDP.
- Sources for financial depth: IMF FinStats Database 2021, and IMF staff calculations.

---

### Singapore case study

### Initial conditions (1970s–1980s)
- Economy: labor-intensive commercial center relying on foreign capital; financial system based on currency board prior to June 1973.
- Share of GDP: around 20 percent from financial and business services during 1970s and 1980s.
- Financial depth in 1980:
  - Domestic bank deposits: 50 percent of GDP.
  - Private credit: around 70 percent of GDP.
- Early 1990s structural standing: ranked better than OECD median in business regulation, property rights, and control of corruption; less favorable in share of tertiary educated labor force and voice and accountability.

### Implemented policies (selected)
- Targeting niche markets:
  - 1968: establishment of an Asian dollar market (ADM) to fill global trading-hour gap.
  - 1985 Economic Committee identified risk management, fund management, capital markets, reinsurance, and financial and commodity derivatives as growth areas.
  - Continued exploration of wealth management, financial and commodity derivatives, and fintech.
- Robust legal and supervisory framework:
  - Late 1990s: reforms to liberalize the financial sector and strengthen regulatory and supervisory frameworks.
  - Monetary Authority of Singapore (MAS) focused on resilience, innovation, and credibility in oversight.
- Effective public institutions and business environment:
  - Long-range policy planning and cooperation between policymakers, industry actors, and academic experts.
  - "Policy relations" network enabled regulatory predictability and iterative regulatory design to minimize costs while ensuring systemic stability.
- Human capital initiatives:
  - 1979: Skills Development Fund (SDF) financed by employer levy to upgrade employee skills.
  - Financial sector schemes: Financial Sector Talent Development scheme, Financial Training Scheme, Institute of Banking and Finance Standards Training Scheme, Financial Scholarship Program.
  - Favorable treatment of skilled foreign workers.
  - Recent collaboration (IMDA, MAS, Skills Future Singapore, six local universities, five financial associations) to launch the Tech Skills Accelerator Fintech Collective.
- Infrastructure:
  - Capital expenditure rose steadily relative to GDP during the 1970s and 1980s, focusing on infrastructure and public housing development.

### Results
- Structural indicators: implemented policies helped Singapore surpass the OECD median across many areas.
- Financial deepening (1980 to 2019):
  - Domestic bank deposits relative to GDP more than doubled in 2019 compared with 1980.
  - Private credit to GDP ratio increased by about 55 percentage points during the same period.
- Pace of financial deepening in Singapore was much faster than the OECD median.

*Sources: ILO; Fraser institute; World Governance Indicators (WGI), D. Kaufman (Natural Resources Governance Institute and Brookings Institutions, and A. Kraay (World Bank), 2017; and EIU.*

---

### Policies to facilitate development of Macao SAR’s financial hub

### Policy recommendations and areas for government action
- Skill building:
  - Invest in education and training to improve domestic labor force skills and strengthen science-business linkages.
  - Attract foreign talents, especially with financial expertise and skills, to complement domestic labor.
- Physical and ICT infrastructure:
  - Close infrastructure gap vis-a-vis OECD countries, including digital infrastructure, to access skilled labor remotely.
  - Boost supply of public housing and reform regulatory frameworks to increase housing supply by the private sector to improve housing affordability.
- Effective public institutions and business environment:
  - Increase regulatory predictability.
  - Raise the effectiveness of the judiciary.
  - Strengthen protection of property rights.
  - Streamline labor and business regulations to increase attractiveness to foreign investors and facilitate resource reallocation toward expanding sectors.
- Financial supervision and integrity:
  - Balance promotion of financial sector development with the need to preserve financial stability and integrity.
  - Strengthen regulatory and supervisory framework to manage potential risks from sector expansion.
  - Implement measures to identify and mitigate ML/TF risks, particularly risks associated with offshore activities.

---

### Annex highlights — literature and indicators
- Literature findings:
  - Skilled labor positively correlates with importance of major financial centers.
  - Quality of the legal system and property rights are influential for attractiveness of financial centers.
  - High living standards and trade liberalization help attract foreign financial inflows; living standards correlate with infrastructure quality.
  - Business environment is a catalyst for attracting multinational corporations and FDI.
- Indicator domains and sources summarized in Annex III (selected):
  - Human capital: ILO, World Bank Human Capital Index, UN ICT Development index, EIU.
  - Infrastructure: Economist Intelligence Unit-Risk Briefing Database; Worldwide Broadband Speed League.
  - Reputation / Rule of law: EIU; Fraser Institute; Worldwide Governance Indicators.
  - Financial development: IMF FinStats Database 2021; Global Connectedness Index; life insurance premium volume to GDP.

*Italicized source attribution: Content derived from the IMF country chapter on PEOPLE’S REPUBLIC OF CHINA—MACAO SPECIAL ADMINISTRATIVE REGION as provided in the supplied PDF content.*

### 4. Reputation (Living Standard)

### 4. Reputation (Living Standard)

### Human development
- Indicator: Human Development Index (UNDP).
- Dimensions covered: health, education, and standard of living.
- Source link provided in text: https://data.mendeley.com/datasets/53zr7cfyrs/1

### Safety (Crime)
- Indicator: Numbeo Crime Index.
- Survey questions covered:
  - How serious you feel the level of crime is
  - Change of crime level in the past three years
  - Feeling of safety walking during daylight
  - Feeling of safety walking during night
  - How worried you are of being mugged or robbed
  - Worries of having a car stolen (or things from the car)
  - Worries of being physically attacked by strangers
  - Worries of being insulted or pestered by anybody, while in the street or any other public place
  - Worries of being subject to a physical attack because of your skin color, ethnic origin, gender or religion
  - Problem of people using and dealing drugs
  - How much is the problem property crimes?
  - How much is the problem violent crimes?
- Source link: https://www.numbeo.com/crime/

### Political stability
- Sources referenced: EIU-Political stability Risk; Worldwide Governance Indicators; D. Kaufman (Natural Resources Governance Institute and Brookings Institutions), and A. Kraay (World Bank), 2017- (Political Stability).
- Issues covered by the political stability category:
  - Risk of significant social unrest during the next two years
  - Clarity, establishment, and acceptance of constitutional mechanisms for orderly transfer of power
  - Likelihood that an opposition party or group will come to power and cause significant deterioration in business operating conditions
  - Concentration of excessive power in the executive and lack of accountability
  - Risk that international disputes/tensions will negatively affect the economy and/or polity
- EIU link pattern shown: http://www.eiu.com/index.asp?layout=RKArticleVW3&article_id=485802032&country_id=1510000351&refm=rkCtry&page_title=Latest%20alerts

### Pollution (Environmental quality)
- Indicator: Numbeo Pollution.
- Survey topics:
  - Satisfaction with air quality
  - Quality and accessibility of drinking water
  - Water pollution (general)
  - Satisfaction with garbage disposal
  - Whether people find the city clean and tidy
  - Noise pollution and light during the night in the city
  - Green and parks in the city
  - Feeling of being comfortable to spend time in the city because of the pollution
- Source link: https://www.numbeo.com/pollution/

### Property price to income ratio (Housing affordability)
- Indicator: Numbeo property price to income ratio.
- Definition: Basic measure for apartment purchase affordability (lower is better); generally calculated as the ratio of median apartment prices to median familial disposable income, expressed as years of income (variations exist).
- Source link: https://www.numbeo.com/property-investment/indicators_explained.jsp

### Cost of living
- Indicator: Numbeo Cost of Living Index.
- Coverage: Living expenses related to rent and housing, groceries, restaurants, local purchasing powers, and others.
- Database construction: Based on user inputs and manually collected data from authoritative sources (websites of supermarkets, taxi company websites, governmental institutions, newspaper articles, other surveys, etc.).
- Source links:
  - https://www.numbeo.com/cost-of-living/
  - https://www.numbeo.com/common/motivation_and_methodology.jsp

### Health care (Perceived health service quality)
- Indicator: Numbeo Health Care Index.
- Survey elements:
  - Skill and competency of medical staff
  - Speed in completing examination and reports
  - Equipment for modern diagnosis and treatment
  - Accuracy and completeness in filling out reports
  - Friendliness and courtesy of the staff
  - Responsiveness (waitings) in medical institutions
  - Convenience of location
- Source link: https://www.numbeo.com/health-care/

### 5. Business Environment (selected indicators)
- Predictable and non-discriminatory taxes
  - Source: Economist Intelligence Unit-Risk Briefing Database.
  - Focus: Is the tax regime clear and predictable? Risk of discriminatory taxes; corporate tax rate considerations; risk from retroactive taxation.
  - EIU link pattern: http://www.eiu.com/index.asp?layout=RKArticleVW3&article_id=485802032&country_id=1510000351&refm=rkCtry&page_title=Latest%20alerts
- Government effectiveness
  - Source: Worldwide Governance Indicators, D. Kaufman (Natural Resources Governance Institute and Brookings Institutions), and A. Kraay (World Bank), 2017.
  - Definition: Perceptions of the quality of public services, quality of the civil service and its independence from political pressures, quality of policy formulation and implementation, and credibility of the government's commitment to such policies.
  - Source link: https://info.worldbank.org/governance/wgi/Home/Documents
- Financial sector regulation and efficiency
  - Source: Economic Freedom Heritage Foundation.
  - Financial freedom components:
    - Extent of government regulation of financial services
    - Degree of state intervention in banks and other financial firms through direct and indirect ownership
    - Government influence on the allocation of credit
    - Extent of financial and capital market development
    - Openness to foreign competition
  - Heritage link pattern: https://www.heritage.org/index/financial -freedom
- Labor market regulation and efficiency
  - Source: Economist Intelligence Unit-Risk Briefing Database.
  - Areas covered: trade unions, labor strikes, labor skills, meritocratic remuneration, and freedom of association.
  - EIU link pattern: http://www.eiu.com/index.asp?layout=RKArticleVW3&article_id=485802032&country_id=1510000351&refm=rkCtry&page_title=Latest%20alerts
- Business environment (composite)
  - Source: Economist Intelligence Unit-Risk Briefing Database.
  - Composition: Average of indices assessing (i) extent to which authorities favor domestic interest over foreign companies in legal matters, and (ii) impact of government policy on promoting competition and curbing unfair business practices.
  - EIU link pattern: http://www.eiu.com/index.asp?layout=RKArticleVW3&article_id=485802032&country_id=1510000351&refm=rkCtry&page_title=Latest%20alerts
- Business regulation
  - Source: Fraser Institute, Economic-freedom-of-the-world-2021.
  - Construction:
    - Simple average of administrative costs and bureaucracy costs.
    - Administrative requirements based on Global Competitiveness Report question: “Complying with administrative requirements (permits, regulations, reporting) issued by the government in your country is (1 = burdensome, 7 = not burdensome)”.
    - Bureaucracy costs based on “Regulatory Burden Risk Ratings” from IHS Markit measuring risk that normal business operations become more costly due to regulatory environment.
  - Fraser Institute link: fraserinstitute.org/economic-freedom
- Tax Burden
  - Source: Economic Freedom Heritage Foundation.
  - Definition: Composite measure reflecting marginal tax rates on both personal and corporate income and overall level of taxation (direct and indirect) as a percentage of GDP.
  - Component sub-factors:
    - The top marginal tax rate on individual income
    - The top marginal tax rate on corporate income
    - The total tax burden as a percentage of GDP
  - Heritage link: https://www.heritage.org/index/fiscal-freedom

### 6. Financial (Market access and financial depth)
- Market access
  - Source: Economist Intelligence Unit-Risk Briefing Database.
  - Question: Are there restrictions on foreign companies gaining access to local capital markets?
  - EIU link pattern: http://www.eiu.com/index.asp?layout=RKArticleVW3&article_id=485802032&country_id=1510000351&refm=rkCtry&page_title=Latest%20alerts
- Global connectedness index
  - Source: DHL-Global Connectedness Index.
  - Coverage: Measures depth and breadth of international flows of trade, capital, information, and people over the period from 2001 to 2020.
  - Definitions:
    - Depth: compares each cross-border flow to relevant domestic activities.
    - Breadth: evaluates the extent to which flows are distributed broadly around the globe rather than concentrated between specific origins and destinations.
  - Source link: https://www.dhl.com/global-en/spotlight/globalization/global-connectedness-index.html
- Bank Z-score
  - Source: World Bank Global Financial Development Database.
  - Definition: Captures the probability of default of a country's commercial banking system by comparing the buffer of capitalization and returns with the volatility of those returns.
  - Source link: https://www.worldbank.org/en/publication/gfdr/data/global-financial-development-database
- Domestic credit to private sector (% of GDP)
  - Source: World Bank Global Financial Development Database.
  - Definition: Financial resources provided to the private sector (loans, nonequity securities, trade credits, other accounts receivable) that establish a claim for repayment; for some countries includes credit to public enterprises.
  - Source link: https://www.worldbank.org/en/publication/gfdr/data/global-financial-development-database
- Nonlife insurance premium volume to GDP
  - Source: World Bank Global Financial Development Database.
  - Definition: Ratio of nonlife insurance premium volume to GDP; premium volume equals insurer's direct premiums earned (Property/Casualty) or received (Life/Health) during the previous calendar year.
  - Source link: https://www.worldbank.org/en/publication/gfdr/data/global-financial-development-database
- Life insurance premium volume to GDP
  - Source: World Bank Global Financial Development Database.
  - Definition: Ratio of life insurance premium volume to GDP; premium volume equals insurer's direct premiums earned (Property/Casualty) or received (Life/Health) during the previous calendar year.
  - Source link: https://www.worldbank.org/en/publication/gfdr/data/global-financial-development-database
- Deposit money banks' assets to GDP
  - Source: World Bank Global Financial Development Database.
  - Definition: Total assets held by deposit money banks as a share of GDP. Assets include claims on domestic real nonfinancial sector (central, state and local governments, nonfinancial public enterprises and private sector). Deposit money banks comprise commercial banks and other financial institutions that accept transferable deposits, such as demand deposits.
  - Source link: https://www.worldbank.org/en/publication/gfdr/data/global-financial-development-database
- Marketable debt
  - Source: Economist Intelligence Unit-Risk Briefing Database.
  - Question: Is there a liquid, deep local-currency denominated fixed-rate medium-term (five-years or more) bond market in marketable debt (debt that is traded freely)?
  - EIU link pattern: http://www.eiu.com/index.asp?layout=RKArticleVW3&article_id=485802032&country_id=1510000351&refm=rkCtry&page_title=Latest%20alerts

*Source: 1macea2022002 - 4. Reputation (Living Standard)*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2022/english/1macea2022002.pdf_