## 1. Demographics of AI Exposure

## Source details

**Canonical URL:** [1. Demographics of AI Exposure](https://www.imf.org/-/media/files/publications/cr/2025/english/1uryea2025002-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2025/english/1uryea2025002-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2025/english/1uryea2025002-source-pdf.pdf.json)

---

### Executive summary and headline quantitative findings
- Well-targeted reforms in credit markets, trade facilitation, and labor market flexibility could increase output growth by up to 1.5 percentage points over the medium term.
- Advancements in artificial intelligence (AI) could raise GDP by as much as 10 percent over the next 30 years.
- Since 1980, GDP per capita (in PPP) in Uruguay has hovered around 35 percent of the GDP per capita of the US.
- South Korea example: GDP per capita is now nearly 80 percent of that of the United States (contrasted with Uruguay’s ~35 percent).
- Uruguay’s long-run average real GDP growth over the last 60 years: 2.2 percent; comparator LA7 peers: 3.9 percent.
- Average growth over the last 10 years: 1.2 percent per year.
- Potential growth estimates:
  - Baseline potential growth around 2.2 percent.
  - Statistical filtering methods range: 1.9-2.5 percent; median around 2.2 percent.
- Growth-accounting assumptions behind the 2.2 percent potential:
  - Real investment growth around 3-4 percent per year.
  - Labor force growth at 0.2 percent.
  - Contribution from TFP growth around 1 percent.

### Role of structural reforms and exports
- Framing:
  - Structural reforms and export-led growth are central to escaping the middle-income stall via endogenous growth mechanisms (human capital, technological progress, knowledge spillovers).
- Reforms highlighted for Uruguay:
  - Governance, trade openness, financial development, and R&D to progress toward high-income status.
  - Export-led growth to enhance productivity through competition, learning-by-exporting, and resource reallocation.

### Current strengths, gaps, and AI preparedness
- Strengths (selected):
  - Regional leader in governance, macroeconomic stability, and trade openness.
  - High performance on human capital, poverty reduction, and institutional indicators; control of corruption better than many advanced economies in some measures.
  - High ranking on IMF’s AI Preparedness Index (AIPI); top 5 in regulation and ethics.
  - Ranks third in Latin America on CEPAL’s Latin American Artificial Intelligence Index.
  - Higher scores on Government, Technology, Data & Infrastructure in the 2024 Government AI Readiness Index by Oxford Insights compared to the LAC average.
- Gaps and constraints:
  - Convergence with high-income countries stalled; disappointing investment and TFP growth.
  - High youth unemployment and persistent regional disparities (notably northern departments).
  - Limited global integration and participation in global value chains due to regulatory barriers and low competitiveness.
  - Dominance of SOEs in key sectors creating inefficiencies and potential conflicts of interest.
  - Infrastructure bottlenecks in logistics and transport.
  - Financial sector stable but heavily dollarized, restricting credit access for micro, small, and medium-sized enterprises.
  - Firms perceive labor market practices as rigid.
  - Education quality, completion rates, and early learning outcomes need improvement.

### AI implications and broad considerations
- AI effects:
  - Potential to boost long-term productivity and growth while inducing reallocation between labor and capital (task replacement and complementarity).
  - Generative AI (large language models) broadens applications and accessibility; specialized AI models can focus on pattern identification or repetition.
- Literature context: positions contribution alongside Acemoglu (2025), Goldman Sachs (2023), Korinek (2024).

### Methodology notes relevant to projections
- Data and methods:
  - Growth accounting uses Penn World Tables 10.0; decomposes growth into labor, capital, and TFP.
  - Potential output estimation combines multiple statistical filters (Christiano-Fitzgerald band-pass, Butterworth high-pass, Boosted Hodrick-Prescott), a multivariate Holston-Laubach-Williams Kalman filter, and a medium-scale DSGE model estimated for Uruguay (Rabanal and Sbrancia, 2024).
  - Median of methods yields the 2.2 percent potential growth figure.

### Policy implications and priorities (chapter emphasis)
- Priority reforms to harness AI and close competitiveness gaps:
  - Financial deepening to improve credit access for micro, small, and medium-sized enterprises.
  - Trade facilitation and measures to enhance global competitiveness and participation in global value chains.
  - Labor market reforms to increase flexibility while managing social impacts.
  - Reduce regulatory barriers and bureaucracy that constrain firm entry, innovation, and export orientation.
  - Invest in and design policies that foster AI adoption and broader innovation capabilities, leveraging Uruguay’s strengths in AI preparedness and regulation.

---

### Structural gaps, economic complexity, and labor-market rigidity
- Moderate gaps relative to peers in:
  - Labor market regulation, credit markets, and economic complexity.
- Observations:
  - Labor market regulations increasingly perceived as rigid due to centralized wage-setting and hiring practices.
  - Economic Complexity Index (ECI) comparisons (1998–2010 vs 2011–2023) show:
    - Trade gap increased for Uruguay between these periods (divergence in export sophistication and diversity).
    - Technology gap narrowed somewhat but innovation and research capabilities remain challenging.
  - Structural reform indicators normalized 0–1 over 2000–2022; structural gaps computed as absolute difference between frontier and Uruguay.

### AI exposure and labor market: indices, findings, and occupational examples
- Measurement:
  - AI Occupational Exposure (AIOE) index from Felten et al. (2021) used.
  - AI Complementarity index reflects shielding/complementary potential to AI.
  - Occupation classifications: high-exposure/high-complementarity, high-exposure/low-complementarity, low-exposure.
- Uruguay vs peers:
  - Uruguay’s labor force has similar exposure to AI compared with the average Emerging Market economy.
  - Emerging Market economies rank below advanced economies on AI exposure metrics.
  - Uruguay’s AI exposure below Chile and Mexico; similar to Brazil and Colombia.
- Occupational examples:
  - High-exposure/high-complementarity: Lawyers, surgeons, judges, teachers’ aides, childcare workers.
  - High-exposure/low-complementarity: Sales assistants.
  - Low-exposure: Cooks, waiters, hairdressers.

### Demographics, regional distribution, informality, and inequality implications
- Age and education:
  - Older workers less likely to be in high-exposure low-complementary occupations; transitions occur especially between ages 20-40.
  - Additional years of education associated with greater employment share in high-exposure occupations, especially high-complementarity ones.
- Gender:
  - Females tend to work in occupations more exposed to AI than males.
  - A larger share of female workers are also in occupations highly complementary to AI — implying both greater risks and opportunities for women.
- Informality:
  - Informality rate: around 20-25 percent.
  - Most informal workers are in low-exposure occupations.
  - Formal workforce (~75 percent of workers) mixes high- and low-exposure occupations.
  - Risk: AI adoption may displace formal-sector workers and increase informality if not managed, especially where firms face a labor tax wedge.
- Regional concentration:
  - Montevideo houses forty percent of the population with a mixture of exposure/complementarity occupations.
  - Canelones: majority of workers not directly exposed to AI (agriculture and wine production).
- Income distribution:
  - Professional and managerial occupations account for above 15 percent of employed population; these occupations exhibit high exposure and high complementarity.
  - Employment in high-exposure/low-complementarity occupations is increasing across income quantiles (mildly positive trend).
  - High-exposure/high-complementarity employment concentrated in upper-income quantiles, especially the top 20 percent of earners — suggesting AI gains may disproportionately accrue to higher-income earners.

### Sectoral patterns (ISIC Rev. 4)
- Sector variation:
  - Wholesale, retail trade, and services: mix of occupations—some highly exposed and complemented by AI, others with low complementarity.
  - Education: many roles both highly exposed and highly complementary to AI.
  - Health: high complementarity potential (diagnostics, patient care, efficiency).
  - Agriculture, construction, manufacturing: less exposed generally due to manual labor dependence, though innovation initiatives (e.g., INIA) exist.

---

### Potential output gains from closing structural gaps
- Empirical approach:
  - Applies Budina et al. (2023) local projection method (Jordà 2005) for Uruguay to trace cumulative effects over a projection horizon; converts five-year cumulative percent change to annual contribution using compound annual growth rate.
- Estimated gains from closing gaps:
  - Closing gap in governance, business regulation, or external orientation to most advanced country: up to 1 percentage points annual growth.
  - Achieving global frontier score in credit market development: up to 1.2 percentage points annual growth.
  - Labor market reforms enhancing flexibility: could dynamize the economy (weaker statistical significance).
- Model calibration and data sources:
  - Household income distribution and AI exposure: ECH (INE); occupations matched at 4-digit ISCO 08 to Felten et al. (2021).
  - Investor behavior and household balance sheets: 2014 Uruguayan Household Finance Survey (EFHU-2).
  - Historical capital and labor shares: Penn World Table (Feenstra et al., 2015).

### Policy recommendations tied to structural gaps and AI exposure
- Priority reform areas:
  - Governance, business regulation, external orientation — closing gaps could yield up to 1 percentage points annual growth.
  - Credit market development — closing the gap to frontier could yield up to 1.2 percentage points annual growth.
  - Labor market flexibility — reforms in hiring and wage-setting to support dynamism (estimates weaker statistically).
- Complementary policies for AI:
  - Reskilling and upskilling programs, especially for older workers and those in high-exposure low-complementarity occupations.
  - Targeted regional interventions (e.g., for Canelones and Montevideo).
  - Inclusive digital adoption to integrate informal workers and enhance financial inclusion while managing displacement risks.

---

### AI adoption modeling: projected output and productivity effects
- Model approach:
  - Follows Rockall et al. (2024) building on Moll et al. (2022); calibrated with ECH labor market data; results shown as ranges by altering investor share, capital risk, and degree of AI complementarity.
- Estimated output gain over 30 years due to AI adoption: between 8.5 percent and 9 percent.
- Productivity baseline:
  - Acemoglu (2025) forecast: 1.5 percent increase due to AI adoption (annual TFP rise of 0.05 percent, aggregated to 1.5 percent over 30 years).
- Uncertainty note:
  - Model assumes unchanged worker distribution across AI-exposed professions over forecast; abstracts from global trade spillovers.

### Distributional impacts: wages versus wealth
- Baseline inequality: Gini index around 40 (World Bank).
- Expected effects:
  - Wage inequality: expected to lessen in most scenarios; concentration of AI-exposed workers at top reduces labor inequality.
  - Wealth inequality: expected to rise; AI-associated gains accrue predominantly to top earners.
  - Top groups most affected: surge in wealth inequality concentrated in top 5 percentile and stronger for top 0.1 percentile.
  - Overall income: baseline calibration indicates overall positive income impact for all earners relative to a no-AI world despite some labor displacement.

### Sectoral exposure and contribution to growth
- Finding:
  - High value-added sectors driving Uruguay’s growth are highly exposed to AI.
  - Positive correlation between sectors' contribution to GDP growth and AI exposure.
- Risk:
  - Growth driven by AI-exposed sectors may be more volatile due to sensitivity to rapid technological change.

---

### Firm-level environment, innovation, and trade: WBES and other indicators
- Days and constraints (WBES 2024 Uruguay survey):
  - Days to obtain an operating license: 167 days.
  - Days to obtain an import license: 37.4 days.
  - Days to clear direct exports through customs: 2.9 days.
  - Days to clear imports from customs: 3.4 days.
  - Percent of firms identifying business licensing/permits as a major constraint: 25.1 percent.
  - Percent of firms identifying customs and trade regulations as a major constraint: 28.4 percent.
  - Percent of firms identifying labor regulations as a major constraint: 22.5 percent.
  - Percent of firms identifying an inadequately educated workforce as a major constraint: 41.7 percent.
- Innovation and AI preparedness metrics:
  - Percent of firms that introduced a new product/service over last 3 years: 72.6 percent.
  - Percent of firms whose new product/service is also new to the main market: 59.4 percent.
  - Percent of firms that introduced a process innovation over last 3 years: 53.7 percent.
  - Percent of firms that spend on R&D in the last fiscal years: 29.8 percent.
  - Percent of firms using technology licensed from foreign companies: 15.7 percent.
  - Percent of firms having their own web site: 80.6 percent.
  - Percent of firms identifying tax rates as a major constraint: 52.7 percent.
  - Percent of firms perceiving corruption as a major constraint: 24.2 percent.
- Financial and investment financing:
  - Percent of firms with a bank loan/line of credit: 64.1 percent.
  - Percent of firms with a checking or savings account: 98.9 percent.
  - Percent of firms identifying access to finance as a major constraint: 18.6 percent.
  - Percent of firms whose recent loan application was rejected: 2.6 percent.
  - Percent fully credit constrained: 2.6 percent.
  - Percent partially credit constrained: 10.0 percent.
  - Investment financing composition:
    - Investment financed internally: 69 percent.
    - Investment financed by banks: 18.5 percent.
    - Investment financed by supplier/customer credit: 9.0 percent.
    - Investment financed by equity or stock sales: 2.2 percent.
  - Working capital financing composition:
    - Working capital financed internally: 60.3 percent.
    - Working capital financed by banks: 18.0 percent.
    - Working capital financed by supplier/customer credit: 20.2 percent.
    - Percent of firms using banks to finance working capital: 46.1 percent.
- Interpretation:
  - Administrative bottlenecks (operating and import licenses) and skills shortages persist.
  - Customs clearance times are relatively short; licensing delays are a key regulatory bottleneck.
  - Internal financing dominates investment; supplier/customer credit plays a notable role.

### Trade reform simulation (Melitz model) and productivity
- Model calibration:
  - Simplified Melitz (2003) calibrated with WBES observed exporters and estimated TFP; elasticity of substitution for CES demand assumed equal to 4.
  - Trade reform simulated: reduction in trade costs by 30 percent.
- Simulated outcome:
  - Number of exporting firms would increase by about 60 percent in Uruguay.
  - Mechanism: lower trade costs reduce TFP cutoff for exporting; selection and scale-up by high-productivity exporters raise aggregate TFP despite potential entry of lower-TFP exporters.

---

### Opening and operating procedures (Box 4) — administrative burdens
- Opening procedures (Uruguay):
  - Opening procedures: 1,080 hours.
  - LA7 average (opening): 724 hours.
  - LAC average (opening): 1,968 hours.
  - Identification and registration (component): 840 hours.
  - Registration with national entities (component): 72 hours.
  - Sector-specific procedures (component): 5 hours.
- Operating procedures (annual hours):
  - Operating procedures required annually: 414 hours.
  - LAC average (operating): 1,850 hours.
  - LA7 average (operating): 1,155 hours.
  - Employment administration: 201 hours.
  - Taxation: 136 hours.
- Diagnostics and policy implications:
  - Identification and registration are principal sources of opening burden.
  - Operating burdens are relatively low compared with regional averages; employment administration and tax compliance are material sources of annual cost.
  - Recommendations: streamline identification processes, reduce frictions in employment administration, and simplify tax compliance.

---

### The “Exporter Premium” (Annex III) — firm-level heterogeneity
- Regional empirical findings (WBES regressions):
  - Advanced economies/OECD: exporters exhibit lower TFP compared to non-exporters (negative and significant coefficients).
  - Emerging & developing: exporters show a positive TFP effect.
  - LAC and Uruguay: no statistically significant TFP differences between exporters and non-exporters.
  - Across regions, exporters are older, larger, and more likely foreign owned; proportion of skilled workers among exporters is consistently lower.
  - Real annual sales and labor productivity growth significantly higher for exporters in emerging and advanced economies; impact in LAC weaker.
  - Exporter firms are more likely to have a website.
- Uruguay-specific findings:
  - Exporters: non-statistically significant effect on TFP; no significant gains in real annual sales growth or labor productivity growth.
  - Exporters tend to be older, employ more workers, and more likely foreign owned.
  - Uruguayan exporters have highest website adoption rate among firms relative to LAC and OECD.
- Selected regression parameter highlights (exporter indicator coefficients from Annex III. Table 1):
  - TFP (output method): -0.234 0.0418 0.0134 -0.0259 -0.0771***
  - TFP (VA method): -0.288 -0.0385 0.0849*** -0.210*** -0.234***
  - Age of establishment (years): 9.670*** 5.404*** 3.846*** 6.116*** 6.389***
  - Ownership type (foreign vs. domestic): 0.141*** 0.175*** 0.176*** 0.145*** 0.144***
  - Number of workers: 78.91*** 127.9*** 132.9*** 74.30*** 56.55***
  - Proportion of skilled workers, out of: -3.310 -2.136*** -2.320*** -3.259*** -3.210***
  - Real annual sales growth (%): -1.728 0.371 1.223*** 0.628*** 0.549**
  - Real annual labor productivity growth: -1.111 0.576 1.217*** 0.803*** 0.565***
  - Percent of firms having their own web: 26.42*** 24.35*** 25.25*** 12.61*** 8.437***

- Analytical implication:
  - Exporter premium is heterogeneous by region; in Uruguay exporters do not clearly exhibit a TFP or growth premium — policy focus should help exporters convert market access into measurable productivity and sales gains (e.g., technology adoption, skills upgrading, digitalization).

---

### Tax expenditures, fiscal space, and GIMF simulations (selected references and results)
- Tax expenditures overview:
  - Number of active tax expenditure measures in Uruguay: 186.0.
  - Total foregone revenue:
    - As percent of tax revenues: Uruguay 32.3; LAC 28.7; Emerging & developing 17.6; Advanced economies 21.9.
    - As percent of GDP: Uruguay 6.0; LAC 4.1; Emerging & developing 3.1; Advanced economies 5.5.
  - HHI index of measures: Uruguay 421.9; LAC 1342.2; Emerging & developing 1993.3; Advanced economies 1437.2.
- Composition and concentration:
  - Type of tax shares (Uruguay):
    - Goods and services: 48.2 percent.
    - Income: 30.8 percent.
    - Property: 21.0 percent.
  - Type of tax expenditure (Uruguay):
    - Exemption: 59.5 percent.
    - Reduced rate: 13.3 percent.
    - Tax credits, rebates and refunds: 11.6 percent.
  - Type of beneficiary (Uruguay):
    - Businesses: 39.7 percent.
    - Households: 31.3 percent.
  - Ten measures account for nearly half of all tax expenditures.
  - Relative cost: property tax expenditures in the form of exemptions about 150 percent of property tax collected.
- Cross-country patterns:
  - Uruguay stands out for number of measures and rising revenue foregone over time.
  - Cross-country regressions show associations of revenue foregone with GDP per capita, legal/property rights, sound macroeconomics, and free trade.
- GIMF simulation of a reform reducing inefficient tax expenditures:
  - Assumed revenue increase: 0.4 percent of GDP redirected to productive public expenditure (휉휉 calibrated to 80 percent for public investment efficiency).
  - Simulated macro outcomes (deviations relative to baseline over 2025–2037):
    - Real GDP: cumulative increase reaching 0.5 percent compared to baseline by 2037.
    - Real exports: about 0.8 percent above baseline by 2037.
    - Imports of investment goods: almost 3 percent above baseline by 2037.
    - Nominal effective exchange rate (NEER): about -1.2 percent lower compared to baseline in 2037 (depreciation).
    - Public debt-to-GDP declines by about 0.4 percentage points by 2037.
  - Fiscal outcomes:
    - Reform increases total government revenues-to-GDP by about 0.35 percentage points over the period in an illustrative scenario.
    - Government spending as a share of GDP increases while early small initial deficit is eliminated and debt stabilizes lower.
- Policy implications from tax-expenditure analysis:
  - Regular ex-ante cost-benefit analyses and a benchmark tax system recommended.
  - Redirect savings to education, infrastructure, and innovation to boost private investment and long-term growth.
  - Target investment-related incentives (construction, healthcare, housing, energy) and property tax exemptions given their relative cost.
  - Careful calibration of behavioral elasticities and sequencing to manage short-term trade-offs and long-term gains.

*Source: International Monetary Fund — chapter "1. Demographics of AI Exposure" from BOOSTING GROWTH IN URUGUAY THROUGH STRUCTURAL REFORMS AND ARTIFICIAL INTELLIGENCE.*

### 1. Demographics of AI Exposure ______________________________________________________ 11

### 1. Demographics of AI Exposure

### Executive summary and key quantitative findings
- Well-targeted reforms in credit markets, trade facilitation, and labor market flexibility could increase output growth by up to 1.5 percentage points over the medium term.
- Advancements in artificial intelligence (AI) could raise GDP by as much as 10 percent over the next 30 years.
- Since 1980, GDP per capita (in PPP) in Uruguay has hovered around 35 percent of the GDP per capita of the US.
- South Korea example: GDP per capita is now nearly 80 percent of that of the United States (contrasted with Uruguay’s ~35 percent).
- Uruguay’s long-run average real GDP growth over the last 60 years has been 2.2 percent, compared with 3.9 percent for other LA7 peers.
- Average growth over the last 10 years was 1.2 percent per year (impacted by Brazil recession in 2015, Covid-19, and the most severe drought in the last 100 years).
- Potential growth estimates:
  - Baseline potential growth around 2.2 percent given assumptions below (see Growth accounting).
  - Statistical filtering methods over higher-frequency data yield a range of 1.9-2.5 percent for potential output growth; the median of different estimates is around 2.2 percent.
- Growth-accounting assumptions underpinning the 2.2 percent potential:
  - Real investment growth around 3-4 percent per year (the average of the last 25 years).
  - Labor force growth at 0.2 percent (based on historical trend and 2023 census migration flows).
  - Contribution from TFP growth around 1 percent (close to the average over the last 20 years, excluding 2020-2021).

### Role of structural reforms and exports
- Structural reforms and export-led growth are central to escaping the middle-income stall; reforms influence human capital, technological progress, and knowledge spillovers (endogenous growth mechanisms).
- Export-led growth enhances productivity via increased competition, learning-by-exporting, and resource reallocation.
- For Uruguay, reforms in governance, trade openness, financial development, and R&D are highlighted as necessary to progress toward high-income status.

### Current strengths, gaps, and AI preparedness
- Strengths:
  - Uruguay is a regional leader in governance, macroeconomic stability, and trade openness.
  - High performance across human capital, poverty reduction, and institutional indicators; in some measures (e.g., control of corruption) performing better than many advanced economies.
  - Uruguay ranks highly in AI preparedness among Emerging Market economies on the IMF’s AI Preparedness Index (AIPI), with particularly strong performance in regulation and ethics (placing Uruguay in the top 5 for that category).
  - Other indices: Uruguay ranks third in Latin America on the Latin American Artificial Intelligence Index by CEPAL and scores higher on Government, Technology, Data & Infrastructure in the 2024 Government AI Readiness Index by Oxford Insights compared to the LAC average.
- Gaps and constraints:
  - Convergence with high-income countries has stalled; investment and total factor productivity growth have been disappointing.
  - High youth unemployment and persistent regional disparities (notably in northern departments).
  - Limited global integration and participation in global value chains due to regulatory barriers and low competitiveness.
  - Dominance of state-owned enterprises (SOEs) in key sectors contributes to inefficiencies and potential conflicts of interest.
  - Infrastructure bottlenecks in logistics and transport hinder competitiveness.
  - Financial sector is stable but heavily dollarized, restricting credit access for micro, small, and medium-sized enterprises.
  - Firms perceive labor market practices as rigid.
  - Education challenges: quality, completion rates, and early learning outcomes need improvement.

### AI implications and considerations
- AI has potential to boost long-term productivity and growth but will also lead to reallocation between labor and capital; some tasks may be replaced while others are complemented.
- Recent developments, especially generative AI (large language models), broaden AI’s potential applications and accessibility; specialized AI models can focus on pattern identification or repetition.
- The paper positions its contribution within the growing literature on AI’s economic impact (citations noted: Acemoglu (2025), Goldman Sachs (2023), Korinek (2024) as context).

### Methodology notes relevant to projections
- Growth accounting relies on Penn World Tables 10.0 data, decomposing growth into labor, capital, and TFP contributions.
- Potential output estimation combines multiple statistical filters (Christiano-Fitzgerald band-pass, Butterworth high-pass, Boosted Hodrick-Prescott) and a multivariate Holston-Laubach-Williams Kalman filter approach, plus a medium-scale DSGE model estimated for Uruguay (Rabanal and Sbrancia, 2024).
- The median of these multiple methods yields the presented potential growth figure (2.2 percent).

### Policy implications and priorities (as emphasized in the chapter)
- Prioritize structural reforms to close gaps in competitiveness and harness AI benefits, focusing on:
  - Financial deepening to improve credit access for micro, small, and medium-sized enterprises.
  - Trade facilitation and measures to enhance global competitiveness and participation in global value chains.
  - Labor market reforms to increase flexibility while managing social impacts.
  - Reforms to reduce regulatory barriers and bureaucracy that constrain firm entry, innovation, and export orientation.
  - Investments and policies fostering AI adoption and broader innovation capabilities, leveraging Uruguay’s relative strengths in AI preparedness and regulation.

*Source: International Monetary Fund — chapter "1. Demographics of AI Exposure" from BOOSTING GROWTH IN URUGUAY THROUGH STRUCTURAL REFORMS AND ARTIFICIAL INTELLIGENCE.*

### 13. However, moderate gaps compared to peers exist in other areas, such as credit and

### 1uryea2025002-source-pdf - 13. However, moderate gaps compared to peers exist in other areas, such as credit and

### Structural gaps and economic complexity
- Labor market regulation, credit markets, and economic complexity exhibit moderate gaps relative to peers.
- Labor market regulations have been increasingly perceived as rigid over time due the centralized nature of wage-setting and hiring practices (Figure 2).
- Using the Economic Complexity Index (ECI):
  - Periods compared: 1998–2010 and 2011–2023.
  - Trade gap increased in Uruguay between 1998–2010 and 2011–2023, indicating growing divergence in export sophistication and diversity.
  - Technology gap has narrowed somewhat, but Uruguay still faces substantial challenges in innovation and research capabilities.
  - Compared to peers in Latin America and Emerging East and Southeast Asia, Uruguay performs similarly in research and technology gaps but has fallen further behind on trade-related complexity.
- Structural reform indicators are normalized between 0 and 1 over 2000-2022; structural gaps are calculated as the absolute difference between the frontier (maximum value in comparator group) and Uruguay.

### AI exposure and labor market: measures and aggregate findings
- AI exposure definition and measurement:
  - AI Occupational Exposure (AIOE) index from Felten et al. (2021) is used to measure overlap between AI applications and human abilities required by occupations.
  - AI Complementarity index reflects an occupation’s shielding and complementary potential to AI (see Annex).
  - Occupations can be high-exposure/high-complementarity, high-exposure/low-complementarity, or low-exposure.
- Uruguay’s AI exposure relative to peers:
  - Uruguay’s labor force has similar exposure to AI compared with the average Emerging Market economy.
  - Emerging Market economies rank below advanced economies on AI exposure metrics.
  - Uruguay’s AI exposure is below that of Chile and Mexico and more similar to Brazil and Colombia (Bakker et al., 2024).
- Occupational examples illustrating exposure/complementarity:
  - Lawyers: high exposure and high complementarity (textual analysis complements document review; interpersonal tasks remain).
  - Surgeons and judges: high-exposure and high-complementarity due to responsibility and interpersonal interactions.
  - Cooks, waiters, hairdressers: low AI exposure due to hands-on tasks.
  - Sales assistants: highly exposed with low complementarity.
  - Teachers’ aides and childcare workers: high exposure with high complementarity.

### Demographics, regional distribution, informality, and inequality implications
- Demographics and education:
  - As Uruguayan workers age, they are less likely to be in high-exposure low-complementary occupations; transition in career progression occurs particularly between ages 20-40.
  - Additional years of education are associated with greater share of employment in high-exposure occupations, especially those with high complementarity to AI.
- Gender:
  - Females tend to work in occupations that are more exposed to AI than males.
  - A larger share of female workers are also in occupations that are highly complementary to AI compared to males — implying both greater risks and greater opportunities for women from AI adoption.
- Informality:
  - Uruguay’s informality rate: around 20-25 percent.
  - Within the informal workforce, the majority of workers are in low-exposure occupations.
  - The formal workforce (~75 percent of workers) displays a mixture of high exposure and low exposure workers.
  - Risk: AI adoption may cause displacement in the formal sector, potentially increasing informality if not managed, especially where firms pay a cost for formality (labor tax wedge).
- Regional concentration:
  - Montevideo houses forty percent of the population and a mix of occupations across exposure/complementarity categories.
  - Canelones has a majority share of workers not directly exposed to AI (agriculture and wine production).
- Income distribution:
  - Professional and managerial occupations account for above 15 percent of the employed population and exhibit high exposure and high complementarity.
  - Employment in occupations at risk of displacement (high-exposure, low complementarity) is increasing across income quantiles (mildly positive trend).
  - High-exposure, high-complementarity employment is concentrated in upper-income quantiles, especially the top 20 percent of earners — suggesting AI gains may disproportionately accrue to higher-income earners.

### Sectoral patterns
- Sector variation in AI exposure and complementarity (ISIC Rev. 4 classification):
  - Wholesale, retail trade, and services: mix of occupations—some highly exposed and complemented by AI, others with low or no complementarity.
  - Education: many roles both highly exposed and highly complementary to AI — strong potential for AI to enhance learning and administrative tasks.
  - Health: high complementarity potential with improvements in diagnostics, patient care, and efficiency.
  - Agriculture, construction, manufacturing: less exposed to AI due to reliance on manual labor, though innovation exists (e.g., initiatives led by INIA).

### Potential output gains from closing structural gaps
- Methodology:
  - Empirical model estimates from Budina et al. (2023) applied to Uruguay use the local projection method (Jordà 2005), regressing future values of output on current and past values of structural reform indicators and controls.
  - Effects are traced over a projection horizon; cumulative percent change in real output (or employment) over five years is converted to annual contribution using the compound annual growth rate formula.
- Estimated gains:
  - Reforms closing the gap between Uruguay and the most advanced country in governance, business regulation, or external orientation could increase Uruguay’s annual growth by up to 1 percentage points.
  - Achieving the same score as the global frontier in credit market development could increase growth by up to 1.2 percentage points.
  - Labor market reforms that enhance flexibility in hiring and wage-setting practices could dynamize the economy (weaker statistical significance reported).
- Model calibration and data sources:
  - Household income distribution and AI exposure from ECH (household survey) (INE); occupations matched at 4-digit ISCO 08 level to Felten et al. (2021) AIOE index.
  - Investor behavior and household balance sheets from the 2014 Uruguayan Household Finance Survey (EFHU-2).
  - Historical capital and labor share of output from Penn World Table (Feenstra et al., 2015).

### Policy implications and areas for reform
- Priority reform areas indicated by analysis:
  - Governance, business regulation, and external orientation — closing gaps here could yield up to 1 percentage points in annual growth.
  - Credit market development — closing the gap to the frontier could yield up to 1.2 percentage points in annual growth.
  - Labor market flexibility — reforms in hiring and wage-setting could support dynamism and employment, though estimates show weaker statistical significance.
- Complementary policies given AI exposure patterns:
  - Support reskilling and upskilling, especially for older workers who may find reskilling more difficult and for workers in high-exposure low-complementarity occupations.
  - Target regional interventions (e.g., for Canelones and Montevideo) to bridge regional disparities in exposure and benefits from AI.
  - Promote inclusive digital adoption to integrate informal workers and enhance financial inclusion while managing displacement risks in the formal sector.

*URUGUAY — INTERNATIONAL MONETARY FUND.*

### 25. The adoption of AI has also the potential to boost Uruguay’s output. Following the

### 25–35: The adoption of AI, distributional impacts, firm-level environment, and trade reforms in Uruguay

### AI adoption: projected output and productivity effects
- Model approach: Follows Rockall et al. (2024) building on Moll et al. (2022); calibrated using labor market data from the ECH; results shown as a range by altering calibration values for the share of investors, capital risk, and degree of complementarity of AI.
- Estimated output gain due to AI adoption over 30 years: between 8.5 percent and 9 percent.
- Productivity rise baseline: Acemoglu (2025) forecasts a 1.5 percent increase due to AI adoption (noted as an annual TFP rise of 0.05 percent which, if assumed constant, is taken as 1.5 percent over 30 years).
- Results fall within ranges from Cerutti and others (2025) for Emerging Market Economies Latin America over 10 years in a multi-country DSGE model.
- Notes on uncertainty: model assumes the distribution of workers in AI-exposed professions does not change over the forecast period; global benefits via trade linkages and cross-border spillovers are abstracted from the analysis.

### Distributional impacts: wages versus wealth
- Baseline inequality context: Gini index of around 40 (World Bank).
- Expected distributional effects of AI adoption:
  - Wage inequality: expected to lessen in most scenarios; higher concentration of AI-exposed workers at the top of the income distribution reduces labor inequality.
  - Wealth inequality: expected to rise, with AI-associated gains accruing predominantly to top earners as capital income rises.
  - Top groups most affected: surge in wealth inequality concentrated in the top 5 percentile, and even stronger for the top 0.1 percentile, driven by returns to capital.
- Overall income impact: despite labor displacement for some high-wage workers, baseline calibration indicates the overall impact on income for all earners is positive compared to a no-AI world.

### Sectoral exposure to AI and contribution to growth
- Finding: High value-added sectors driving Uruguay’s growth are highly exposed to AI; positive correlation between sectors that significantly contribute to GDP growth and those with higher AI exposure (industry data from INE for 2017, aggregated to 2-digit classification across 108 industries).
- Risk: Growth driven by AI-exposed sectors may become more volatile because these sectors are increasingly sensitive to rapid technological change, potentially leading to sudden shifts in operational capacities and output instability.
- Visual indicators described: Value Added of Industry per Employee plotted against AI exposure, showing higher value added for high AI exposure sectors.

### Firm-level evidence on business environment and innovation (WBES findings)
- Data: World Bank Enterprise Surveys (WBES), including the 2024 Uruguay survey (comparisons across Uruguay, LAC, Emerging & Developing, OECD, Advanced Economies for years 2016–2025 where indicated).
- Key regulatory and labor environment findings:
  - Days to obtain an operating license in Uruguay: 167 days.
  - Days to obtain an import license in Uruguay: 37.4 days.
  - Days to clear direct exports through customs: 2.9 days.
  - Days to clear imports from customs: 3.4 days.
  - Percent of firms identifying business licensing/permits as a major constraint: 25.1 percent.
  - Percent of firms identifying customs and trade regulations as a major constraint: 28.4 percent.
  - Percent of firms identifying labor regulations as a major constraint: 22.5 percent.
  - Percent of firms identifying an inadequately educated workforce as a major constraint: 41.7 percent.
- Implications: administrative bottlenecks (operating and import licenses) and skills shortages are persistent constraints; customs clearance times are relatively short compared to peers but licensing delays are a key regulatory bottleneck.

### Innovation indicators and AI preparedness
- Innovation and technology adoption metrics for Uruguay:
  - Percent of firms that introduced a new product/service over last 3 years: 72.6 percent.
  - Percent of firms whose new product/service is also new to the main market: 59.4 percent.
  - Percent of firms that introduced a process innovation over last 3 years: 53.7 percent.
  - Percent of firms that spend on R&D in the last fiscal years: 29.8 percent.
  - Percent of firms using technology licensed from foreign companies: 15.7 percent.
  - Percent of firms having their own web site: 80.6 percent.
- Interpretation: relatively strong firm-level innovation performance and technology adoption signal comparatively high AI preparedness.
- Broader constraints: 52.7 percent of firms identify tax rates as a major constraint; perceived corruption is relatively low at 24.2 percent.

### Financial sector, access to finance, and investment financing
- Credit and accounts:
  - Percent of firms with a bank loan/line of credit: 64.1 percent.
  - Percent of firms with a checking or savings account: 98.9 percent.
  - Percent of firms identifying access to finance as a major constraint: 18.6 percent.
  - Percent of firms whose recent loan application was rejected: 2.6 percent.
  - Percent of firms that are fully credit constrained: 2.6 percent.
  - Percent of firms that are partially credit constrained: 10.0 percent.
- Investment financing composition:
  - Percent investment financed internally: 69 percent.
  - Percent investment financed by banks: 18.5 percent.
  - Percent investment financed by supplier/customer credit: 9.0 percent.
  - Percent investment financed by equity or stock sales: 2.2 percent.
- Working capital financing:
  - Percent working capital financed internally: 60.3 percent.
  - Percent working capital financed by banks: 18.0 percent.
  - Percent working capital financed by supplier/customer credit: 20.2 percent.
  - Percent of firms using banks to finance working capital: 46.1 percent.
- Interpretation: access to basic financial services is high, internal financing remains the predominant source for investment and working capital, supplier/customer credit plays a larger role than in comparator groups.

### Trade reform simulation and productivity (Box 3)
- Model: Simplified Melitz (2003) model calibrated using WBES (observed exporters and estimated TFP values); elasticity of substitution for CES demand function assumed equal to 4.
- Trade reform simulated: reduction in trade costs by 30 percent.
- Simulated outcome:
  - Number of exporting firms would increase by about 60 percent in Uruguay.
  - Mechanism: lower trade costs reduce the TFP cutoff for exporting, bringing more firms into export markets; selection effects and scale-up by high-productivity exporters typically raise aggregate TFP despite potential entry of lower-TFP exporters.
- Calibration notes: fixed and variable trade costs calibrated to match observed pre-reform export distribution; TFP bins generated using TFP deciles.

*Sources: IMF staff calculations; ECH (INE); PWT.*

### Box 4. Opening and Operating Procedures for Firms in LAC

### Box 4. Opening and Operating Procedures for Firms in LAC

### Summary of empirical comparison (Uruguay vs peers)
- Data source: 2024 Index of Bureaucracy, published by the Adam Smith Center for Economic Freedom (16 countries in Latin America and three in Europe; see Annex C in source for details).
- LA7 definition preserved: LA7 includes Brazil, Chile, Colombia, Mexico, Paraguay, Peru and Uruguay.
- Note: Certain countries with very high values have been omitted from the table in the source to facilitate presentation.

### Opening procedures — key statistics
- Opening procedures for Uruguayan firms: 1,080 hours.
- LA7 average (opening procedures): 724 hours.
- LAC average (opening procedures): 1,968 hours.
- Identification and registration (component of opening procedures) for Uruguay: 840 hours.
- Registration with national entities (component) for Uruguay: 72 hours.
- Sector-specific procedures (component) for Uruguay: 5 hours.
- Comparative observation from source: identification and registration represent a major burden for Uruguay, while Uruguay performs better in registration with national entities and sector-specific procedures compared to peers.

### Operating procedures — key statistics
- Operating procedures required annually for Uruguay: 414 hours.
- LAC average (operating procedures): 1,850 hours.
- LA7 average (operating procedures): 1,155 hours.
- Main components of Uruguay’s operating procedures (annual hours):
  - Employment administration: 201 hours.
  - Taxation: 136 hours.
- Comparative observation from source: Uruguay’s operating-procedures burden is significantly below the LAC and LA7 averages; employment administration and taxation are the largest bureaucratic costs.

### Findings and short diagnostics (from the source)
- Identification and registration procedures are a principal source of administrative burden for firms opening in Uruguay.
- Operating burdens are relatively low in Uruguay compared with regional averages, but employment administration and tax compliance remain material sources of annual bureaucratic costs.
- The source indicates that streamlining administrative processes could enhance Uruguay’s business environment and competitiveness.

### Policy implications and targeted reforms (as stated in the source)
- Streamline identification processes to reduce the opening-procedure burden.
- Reduce frictions in employment administration to lower annual operating costs for firms.
- Simplify tax compliance procedures to reduce annual operating costs.
- Overall recommendation in Box 4 (explicit in source): Streamlining identification processes, employment administration, and tax compliance could further enhance Uruguay’s business environment and competitiveness.

*Source: Box 4. Opening and Operating Procedures for Firms in LAC, 2024 Index of Bureaucracy (Adam Smith Center for Economic Freedom), as summarized in the IMF source document.*

### Annex III. The “Exporter Premium”

### Annex III. The “Exporter Premium”

### Regional empirical findings
- Using the WBES, regression estimates reveal differences in the effects of exporting status across regions (advanced economies/OECD, emerging & developing, LAC, and Uruguay).
- Advanced economies and OECD: exporters exhibit lower total factor productivity (TFP) compared to non-exporters, with negative and statistically significant coefficients for both TFP measures.
- Emerging & developing economies: exporters show a positive TFP effect.
- LAC and Uruguay: no statistically significant TFP differences observed between exporters and non-exporters.
- Across all regions, exporters tend to be older, larger (employ more workers), and more likely to be foreign owned.
- The proportion of skilled workers among exporters is consistently lower across all regions, particularly in advanced economies and the OECD.
- Real annual sales growth and real annual labor productivity growth are significantly higher for exporters in emerging and advanced economies, indicating stronger expansion and efficiency gains there; the impact in LAC is weaker.
- Exporter firms are more likely to have a website across regions, suggesting greater innovation or digital sophistication.

### Uruguay-specific findings and interpretation
- Uruguay displays a slightly distinct pattern, potentially influenced by a much smaller sample size.
- In Uruguay:
  - Exporting firms show a non-statistically significant effect on TFP, implying exporters are not necessarily more productive than non-exporters in the short term.
  - Exporters do not register significant gains in real annual sales growth or labor productivity growth; coefficients are close to zero or negative.
  - Exporters tend to be older, employ more workers, and are more likely to be foreign owned than domestic firms, aligning with broader LAC trends.
  - Uruguayan exporters have the highest website adoption rate among firms, surpassing LAC and even OECD firms, suggesting higher digital integration for exporters in Uruguay.

### Key regression estimates (parameter on exporter indicator from Annex III. Table 1)
- TFP (output method): -0.234 0.0418 0.0134 -0.0259 -0.0771***
- TFP (VA method): -0.288 -0.0385 0.0849*** -0.210*** -0.234***
- Age of the establishment (years): 9.670*** 5.404*** 3.846*** 6.116*** 6.389***
- Ownership type (foreign vs. domestic): 0.141*** 0.175*** 0.176*** 0.145*** 0.144***
- Number of workers: 78.91*** 127.9*** 132.9*** 74.30*** 56.55***
- Proportion of skilled workers, out of: -3.310 -2.136*** -2.320*** -3.259*** -3.210***
- Real annual sales growth (%): -1.728 0.371 1.223*** 0.628*** 0.549**
- Annual employment growth (%): -0.439 -0.376 0.154 -0.254* -0.0429
- Real annual labor productivity growth: -1.111 0.576 1.217*** 0.803*** 0.565***
- Capacity utilization (%)*: 3.152* 3.438*** 1.631*** 1.072*** 0.603*
- Sales per worker (in USD 2009): 91,431*** 14,668 -4,434 30,044*** 48,627***
- Percent of firms having their own web: 26.42*** 24.35*** 25.25*** 12.61*** 8.437***

### Analytical implications
- Heterogeneous exporter premium: the productivity and growth advantages associated with exporting are not universal and vary by development level and region.
- In advanced economies the exporter status may reflect other firm characteristics (e.g., larger scale, different factor mixes) associated with lower measured TFP in these regressions.
- In emerging markets, exporting appears associated with stronger sales and productivity growth, indicating exporting can be a driver of firm dynamism in these contexts.
- For Uruguay, the absence of a pronounced exporter premium in TFP and growth suggests policy attention could focus on enabling exporters to translate market access into measurable productivity and sales gains (e.g., through technology adoption, skills upgrading, and digitalization), building on relatively high website adoption among exporters.

*Source: WBES and IMF staff calculations (Annex III. The “Exporter Premium”).*

### References

### References (Selected Issues Paper on Uruguay — tax expenditures and AI-related literature)

### Overview and context
- Uruguay had around 180 active tax expenditures and an estimated revenue foregone equivalent to about 6 percent of GDP in 2021, the third highest in Latin America.
- The paper investigates composition and trends of tax expenditures in Uruguay, benchmarking results from a cross-country perspective and providing empirical analysis and macroeconomic simulations.
- The most affected taxes in Uruguay’s system are the Value-Added Tax (IVA) and the Corporate Income Tax (IRAE).
- A small subset of measures—just ten—accounts for nearly half of all tax expenditures.

### Key empirical findings and statistics
- Number of active tax expenditure measures in Uruguay: 186.0
- Total foregone revenue:
  - As percent of tax revenues: Uruguay 32.3; LAC 28.7; Emerging & developing 17.6; Advanced economies 21.9
  - As percent of GDP: Uruguay 6.0; LAC 4.1; Emerging & developing 3.1; Advanced economies 5.5
- HHI index of measures (concentration of revenue foregone per measure): Uruguay 421.9; LAC 1342.2; Emerging & developing 1993.3; Advanced economies 1437.2
- Number measures (averages by group): Uruguay 186.0; LAC 103.6; Emerging & developing 91.7; Advanced economies 142.3
- Type of tax (shares in percent) — Uruguay vs. comparators:
  - Goods and services: Uruguay 48.2; LAC 62.8; Emerging & developing 69.2; Advanced economies 34.4
  - Income: Uruguay 30.8; LAC 33.9; Emerging & developing 26.2; Advanced economies 60.7
  - Property: Uruguay 21.0; LAC 3.1; Emerging & developing 1.9; Advanced economies 3.4
  - Multiple: Uruguay 0.0; LAC 0.2; Emerging & developing 0.2; Advanced economies 0.4
  - Not stated: Uruguay 0.0; LAC 0.0; Emerging & developing 2.3; Advanced economies 1.0
  - Other: Uruguay 0.0; LAC 0.0; Emerging & developing 0.1; Advanced economies 10.0
- Type of tax expenditure (shares in percent) — Uruguay:
  - Exemption: 59.5
  - Reduced rate: 13.3
  - Tax credits, rebates and refunds: 11.6
  - Other: 9.6
  - Deduction: 3.6
  - Not stated/unclear: 2.4
  - Deferral: 0.0
  - Multiple: 0.0
  - Zero-rated: 0.0
- Type of beneficiary (shares in percent) — Uruguay:
  - Businesses: 39.7
  - Households: 31.3
  - Other: 16.9
  - Non-profit organizations/NGOs/Philanthropic: 5.9
  - Not stated/unclear: 3.8
  - Multiple: 2.2
  - Public sector: 0.1
  - Churches/religious organizations: 0.0
  - International/Regional/Multilateral organizations: 0.0

### Cross-country perspective and interpretation
- Cross-country panel regressions indicate Uruguay's level of tax expenditures is significantly higher than predicted given its institutional quality, governance indicators, and economic structure.
- Uruguay’s relatively low HHI (421.9) implies tax expenditures are more evenly distributed across measures compared with emerging & developing economies (1993.3) where costs are more concentrated on fewer incentives.
- Advanced economies forgo on average 22 percent of tax revenues; emerging markets about 18 percent; LAC about 29 percent; Uruguay 32 percent.
- When measured as percent of GDP: advanced economies 5.5 percent; emerging markets 3.1 percent; LAC 4.1 percent; Uruguay 6.0 percent.

### Macroeconomic simulation results
- GIMF model simulations suggest that reducing cost-inefficient tax incentives and redirecting resources could:
  - Increase tax revenues by 0.4 percentage points of GDP.
  - Lead to a long-term increase in annual real GDP growth of about 0.5 percent.
- The simulations highlight potential fiscal space creation for productive public expenditure and improvements in long-term growth through reallocation toward education, infrastructure, and innovation.

### Policy analysis and recommendations
- Many incentives have endured for decades; literature suggests regular reevaluation through cost-benefit analysis is beneficial.
- Recommendations include:
  - Conducting regularly scheduled, in-depth cost-benefit analyses based on detailed micro-level data and robust econometric methods to assess the behavioral effects of tax incentives.
  - Considering a comprehensive benchmark tax system to improve transparency and provide a consistent framework for evaluating tax expenditures.
  - Phasing out obsolete or inefficient measures and redirecting resources toward high-priority areas such as education, infrastructure, and innovation to boost private investment, productivity, and long-term growth.
  - Learning from international and regional experience to better understand behavioral responses to incentives.
- Policy design should balance fiscal sustainability with economic competitiveness, ensuring tax incentives support strategic development goals while avoiding undue fiscal pressures.

### Tax expenditure typology (definitions and examples drawn from the paper)
- Exoneraciones (exemptions): Situations explicitly provided for in tax regulations that exempt certain economic events (examples: incomes exempt from both corporate and personal income tax; transactions of goods specifically released from tax obligations).
- Alicuotas reducidas (reduced rates): Lower nominal rates established for specific cases (example: reduced rate in the Value Added Tax).
- Deducciones Especiales (special deductions): Amounts subtracted from the taxable base beyond the reference framework (example: increased deductions in corporate income for R&D, staff training).
- Creditos (credits): Amounts reducible from the determined tax liability or tax quota (example: tax credits for investment promotions in corporate income tax).
- Regimenes simplificados (simplified regimes): Provisions typically for small taxpayers or disadvantaged geographical areas (example: fixed-sum or turnover-based taxes for microenterprises replacing general taxation).

*Source: 1uryea2025002-source-pdf - References*

### 8. The composition of tax expenditures also varies substantially across regions. In

### 1uryea2025002-source-pdf - 8. The composition of tax expenditures also varies substantially across regions. In

### Regional composition of tax expenditures
- Emerging markets and LAC: tax expenditures are predominantly associated with taxes on goods and services; exemptions are the most common type for goods and services taxes.
- Advanced economies: tax expenditures are mainly linked to income taxes; exemptions are the most common type for income taxes.
- Uruguay:
  - Cost of tax deductions on property taxes is significantly higher than other regions.
  - Exemptions are particularly prevalent.
  - Businesses receive the largest share of tax expenditures (contrast with households in advanced economies).

### International patterns, concentration, and cross-country variation
- Revenue foregone (RF) and number of measures:
  - Uruguay: tax expenditures totaling about 6 percent of GDP and more than 30 percent over total revenues in 2021; consistently positioned in the top quartile of countries.
  - Uruguay has increased revenue foregone both as percent of GDP and as percent of revenues over time.
  - Uruguay consistently stands out for the number of tax measures over the years.
- Correlations and dispersion:
  - Scatter plot of number of tax measures versus revenue foregone shows a positive correlation, but with significant dispersion: some countries have high RF with relatively fewer measures (broad-based or high-value incentives).
  - Number of tax expenditure measures and HHI concentration index exhibit an inverse relationship: more measures → lower concentration levels.
- Cross-country econometric associations (Table 3, OLS of RF (log)):
  - Government efficiency: -0.0352 (column 1), -0.0465 (column 2) (standard errors (0.0298), (0.0320)).
  - Legal/property rights: 0.0730** (column 1), 0.0387 (column 2) (standard errors (0.0321), (0.0389)).
  - Sound macroeconomics: 0.122*** (column 1), 0.0960*** (column 2) (standard errors (0.0205), (0.0326)).
  - Free trade: 0.153*** (column 1), 0.0918 (column 2) (standard errors (0.0419), (0.0603)).
  - Regulation: 0.0844** (column 1), -0.0482 (column 2) (standard errors (0.0390), (0.0584)).
  - GDP per capita (log): 0.273***, 0.184***, 0.195***, 0.162***, 0.231***, 0.103 (across specifications) (standard errors (0.0318), (0.0560), (0.0327), (0.0466), (0.0408), (0.0700)).
  - Constant terms reported (e.g., -1.271***, -1.171***, -1.792***, -1.686***, -1.698***, -1.098** with corresponding standard errors).
  - Observations: 1,240 (various specifications), R-squared values: 0.105, 0.109, 0.121, 0.116, 0.108, 0.127.
  - Year fixed effects: YES; Country fixed effects: NO (in reported specifications).
- Interpretation:
  - Countries with higher GDP per capita and stronger legal rights, sound macroeconomics, freer trade, and lighter regulation tend to forgo more revenue through tax expenditures.
  - Government size does not show a statistically significant relationship with RF in these regressions.
  - Caution: GTED data do not apply a standardized benchmark for tax expenditures, limiting comparability.

### Uruguay: scope, composition, and distribution of tax expenditures
- Scale and persistence:
  - Approximately 180 active measures in Uruguay.
  - Estimated cost: 6 percent of GDP in 2024.
  - 2005 benchmark: 4 percent of GDP (or 19 percent of tax revenues) in 2005.
  - Many measures date back to the 1970s and 1980s.
- Distribution of beneficiaries and sectors:
  - Expenditures approximately equally distributed between individuals and firms; very small fraction to non-residents.
  - Sectoral concentration: heavily concentrated in investment-related activities, followed by health, industry, energy, and housing.
- Tax instruments and concentration:
  - Most affected taxes: Value-Added Tax (IVA) and Corporate Income Tax (IRAE) — both represent more than 60 percent of the total revenue foregone.
  - By mechanism:
    - Exemptions: largest share of revenue foregone, estimated at about 60 percent.
    - Tax credits: representing around 20 percent.
  - Concentration across measures:
    - Ten measures account for nearly half of all tax expenditures.
    - Relative to revenue collected, property tax expenditures in the form of exemptions represent about 150 percent of the property tax collected.
  - For IRAE: exemptions and tax credits are about equally split.
  - For IVA: reduced rates and tax credits are also relevant.
- Implication: tax expenditures in Uruguay are highly persistent, concentrated in investment-related incentives, and increasingly significant as a share of public revenue—raising fiscal sustainability and policy-efficiency concerns.

### Regional heterogeneity at the measure level (Table 4 findings)
- Advanced economies (interaction terms): lower RF associated with goods and services (-0.785***), income (-0.579***), deductions (-0.627***), exemptions (-0.528***), reduced rate (-0.250), and other categories; tax credit/refund shows 1.083*** in some specifications; other taxes show 1.663*** in one specification.
- Emerging & developing and LAC:
  - LAC: revenue foregone is lesser for taxes on goods and services and other taxes, but higher for tax exemptions and deductions; “other” beneficiaries show larger benefits in LAC.
  - Exemptions, tax credits/refunds and other mechanisms display distinct patterns across regions per interaction estimates in the tax-measure-level regression (98,905 observations; regression includes year, country, and region fixed effects).

### Effects of a tax reform that reduces tax expenditures: GIMF simulations and scenarios
- Illustrative reform:
  - Assumed revenue increase: 0.4 percent of GDP from reducing inefficient tax expenditures.
  - Assumed use of revenue: increase in productive public expenditure, preserving the fiscal balance.
  - Behavioral caveat: actual macroeconomic impact depends on behavioral responses and demand elasticities for consumption and investment; requires ex-ante calibration by tax authorities.
- GIMF model setup and parameters:
  - Model: Global Integrated Monetary and Fiscal (GIMF), a DSGE model with OLG and rule-of-thumb households, monopolistic competition, sticky prices and wages, trade and financial linkages, fiscal instruments, and public debt dynamics.
  - Public investment efficiency parameter (휉휉) used to convert public investment into public capital; given high quality of public investment management in Uruguay, 휉휉 was calibrated to 80 percent.
- Simulated macroeconomic impacts (IMF staff GIMF simulations; deviations relative to baseline):
  - Time horizon of reported deviations: 2025–2037 (figures display percent deviations and percentage-point deviations for current account).
  - General qualitative simulation outcomes (figures show):
    - Real GDP growth: small initial deviations relative to baseline, with cumulative differences plotted over 2025–2036.
    - Real GDP (percent deviation), private absorption, real household consumption, real private investment, public absorption, real government absorption, real imports and exports, and current account/GDP deviations are presented as percent or percentage-point deviations over 2025–2037.
  - Policy interpretation: when revenues are reallocated to productive public investment with relatively high efficiency (휉휉 = 80 percent), long-term growth can rise via increased private productivity and lower production costs; but short-term costs are possible and depend on behavioral responses.
- Alternative use of additional revenues:
  - Authorities could maintain current public expenditure levels and use revenues to reduce the fiscal deficit—outcomes depend on specifics of reform design and behavioral responses.
- Key policy design note:
  - Reforms should be accurately calibrated to account for behavioral reactions of households and firms and to weigh short-term trade-offs against anticipated long-term benefits.

### Policy implications and priorities
- Potential fiscal space:
  - A modest reform (0.4 percent of GDP revenue gain) illustrates how reducing inefficient tax expenditures can create fiscal space for productive public investment or deficit reduction.
- Reform targeting:
  - Focus on investment-related incentives concentrated in construction, healthcare, housing, and energy as potential targets for rationalization.
  - Consider priority on property tax exemptions given their particularly large relative cost (about 150 percent of property tax collected).
- Design and sequencing:
  - Ex-ante empirical calibration of behavioral elasticities is essential to anticipate short-term losses and long-term gains.
  - Given persistence of longstanding measures, comprehensive reform may be needed; systems with many measures may require broader reforms due to lower concentration per measure.
- Efficiency of reallocations:
  - Where public investment efficiency is high (휉휉 calibrated to 80 percent for Uruguay), reallocating savings to productive investment is more likely to yield stronger growth multipliers and positive long-term output effects.
- Cautions:
  - Cross-country data limitations (no standardized benchmark for tax expenditures in GTED) warrant cautious interpretation of international comparisons and econometric residuals.

*Source: GTED database on tax expenditure, Rendición de Cuentas from Uruguay, Fraser Institute economic freedom indicators, IMF staff calculations as presented in the content unit.*

### 20. The Global Integrated Monetary and Fiscal (GIMF) model from the IMF is used to

### 1uryea2025002-source-pdf - 20. The Global Integrated Monetary and Fiscal (GIMF) model from the IMF is used to 

### Simulation of illustrative reform: macroeconomic dynamics and GDP
- Model: Global Integrated Monetary and Fiscal (GIMF).
- Reform: relatively modest reduction in tax expenditures increases revenues over time which are devoted to productive public expenditure (about 0.4 percent of GDP, over 2 percent higher for government absorption).
- Transmission:
  - Productive public expenditure pushes up productivity, gradually encouraging higher private investment (private investment initially falls due to the reduction in tax expenditures).
  - Increased investment and productivity lead to higher labor demand and production.
  - Higher taxes decrease household consumption initially; consumption recovers over the medium term due to higher labor income from increased productive capacity.
- Real GDP impact:
  - Initially, the difference in GDP growth compared to baseline remains close to zero.
  - Starts increasing from 2029 onward.
  - Reaches a cumulative increase of 0.5 percent compared to baseline by 2037.

### Prices, exchange rates, and monetary policy
- Inflation and policy rate:
  - As taxes increase, inflation experiences a slight initial increase but stabilizes over time.
  - Headline inflation spikes on impact due to the one-off change in the VAT, but inflation is rapidly contained due to the reaction of monetary policy and a disinflationary productivity shock.
  - Monetary policy rate (and real interest rate) only increase slightly in the short term.
- Exchange rates:
  - Nominal effective exchange rate (NEER) depreciates consistently—by about -1. 2 percent lower compared to baseline in 2037—making domestic goods more competitive internationally.
  - Real effective exchange rate (REER) initially appreciates but then stabilizes, suggesting price adjustments partly offset nominal depreciation.

### External sector, trade flows, and current account
- Real exports:
  - Increase steadily, with an increase of about 0.8 percent above baseline by 2037, driven by the currency depreciation and improved competitiveness.
- Real imports:
  - Initially rise but start declining from 2028 onwards, reflecting the depreciation of the currency, weaker domestic consumption and the shift toward greater reliance on domestic production.
  - Imports of consumption goods show a notable decline, indicating household spending cuts affecting demand for foreign products.
  - Imports of investment goods increase significantly, reaching almost 3 percent above baseline by 2037, driven by higher investment.
- Current account:
  - The deviation in the current account balance improves over time, turning positive from 2028 onwards in line with the depreciated exchange rate and reduced household consumption.

### Fiscal outcomes and consolidation
- Revenue and spending:
  - An illustrative scenario assumes that a reform of tax expenditure could increase total government revenues-to-GDP by about 0.35 percentage points over the period.
  - Even with increased government expenditures, simulations show the government would be able to eliminate early on the small initial government deficit, stabilizing at near-zero levels.
  - Government spending as a share of GDP increases.
- Debt:
  - Public debt-to-GDP declines steadily, improving by 0.4 percentage points by 2037 due to higher economic growth.
- Policy implication:
  - Lowering inefficient tax expenditures helps consolidate public finances, reduce debt burdens, and create space for productive public spending without excessive crowding out of private sector activity.

### Policy recommendations, evaluation, and data-driven approaches
- World Bank (2024) recommendations:
  - Efficiencies could be achieved with the revision and modernization of tax incentives.
  - Elimination of ineffective or obsolete tax expenditures would ensure fiscal resources are directed toward policies that generate meaningful economic impact.
  - A systematic review of existing tax incentives is required to assess their contribution to investment, employment, and productivity.
- Evaluation and design:
  - Literature suggests a data-driven approach is essential for improving tax policy design.
  - Conduct cost-benefit analysis grounded on detailed micro-level data and robust econometric methods to understand behavioral effects of tax incentives.
  - Identify which measures drive productive investment and which primarily benefit specific sectors without broader economic gains.
  - Define a clear benchmark tax system (IMF (2019b)) to provide greater transparency and consistency in evaluating tax expenditures and serve as a standardized reference for international comparison.
- Measurement and regular review:
  - Impact of incentives should be measured on relevant outcomes, such as job creation, innovation, or regional development.
  - Effectiveness of tax incentives should be evaluated regularly and phase out those that do not contribute to sustainable and inclusive economic growth.
  - Recent initiatives in other countries that have reformed tax incentives, such as in Chile in 2024, could provide valuable lessons.

*Source: IMF staff simulations and related text in the provided content unit.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2025/english/1uryea2025002-source-pdf.pdf_