## URUGUAY’S POTENTIAL GROWTH: WHAT TO EXPECT AFTER THE PANDEMIC?

## Source details

**Canonical URL:** [URUGUAY’S POTENTIAL GROWTH: WHAT TO EXPECT AFTER THE PANDEMIC?](https://www.imf.org/-/media/files/publications/cr/2022/english/1uryea2022002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2022/english/1uryea2022002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2022/english/1uryea2022002.pdf.json)

---

### A. A long-term view on growth
- Long-run averages and drivers:
  - Uruguay’s average growth since 1963: 2.3 percent per year.
  - Peer countries average growth: 3.6 precent.
  - Population dynamics account for about ½ of Uruguay’s lower growth relative to peers.
  - Volatility driven by regional crises, commodity price cycles, and the pandemic.
- Notable episodes:
  - Latin American debt crisis (early 1980s); Brazil’s crisis (late 1990s); Argentinian crisis of 2001 and Uruguay’s 2002 crisis.
  - Exceptional growth during 2004-14 driven by rebound from 2002 crisis and favorable commodity cycle.
  - Recent stagnation: averaging 0.9 percent in 2015-19 (despite absence of major negative external shock).
- Growth accounting framework (sample and data):
  - Data source: Penn World Tables 10.0.
  - Sample: 1978-2019, 91 countries.
  - Growth decomposed into contributions from labor (employment, hours, human capital), capital (perpetual inventory method), and TFP (residual).
- Key long-run contributions (Table 1, 1978-2019):
  - GDP Growth: 2.5%
  - Employment: 0.5%
  - Human Capital: 0.4%
  - Hours: -0.2%
  - Capital: 1.4%
  - TFP: 0.4%
- Subperiod highlights:
  - 1990-2019 GDP growth: 2.9% (Employment 0.4%, Human Capital 0.3%, Hours -0.3%, Capital 1.5%, TFP 0.9%)
  - 2006-09 GDP growth: 5.4% (Employment 1.5%, Human Capital 0.2%, Hours -0.2%, Capital 2.0%, TFP 1.9%)
  - 2010-14 GDP growth: 4.7% (Employment 0.4%, Human Capital 0.3%, Hours -0.2%, Capital 2.6%, TFP 1.6%)
  - 2015-19 GDP growth: 1.3% (Employment -0.1%, Human Capital 0.4%, Hours -0.2%, Capital 1.0%, TFP 0.2%)
- Aggregate attribution (four decades):
  - Capital accumulation explains nearly 60 percent of Uruguay’s GDP growth.
  - TFP about 15 percent.
  - Employment about 20 percent.
  - Human capital accumulation about 15 percent.
  - Hours worked contributed negatively (about -0.2 percent per year).
- Cross-country comparison (1978–2019, selected):
  - URY GDP growth: 2.5% (Employment 0.5%, Human capital 0.4%, Hours -0.2%, Capital 1.4%, TFP 0.4%)
  - EM non-oil com. exp. GDP growth: 3.6% (Employment 1.1%, Human capital 0.5%, Hours -0.1%, Capital 2.4%, TFP -0.2%)
  - After accounting for population growth, Uruguay’s per capita GDP growth aligned with peers; the lower aggregate GDP growth largely reflects slower population growth (Uruguay population growth: 0.5 percent per year vs 1.1 percent in non-oil commodity exporting EMEs).

### B. Labor and human capital
- Labor contribution drivers:
  - Lower labor contribution relative to peers arises mainly from employment, reflecting slower population growth.
  - Contributions from hours worked and the employment-to-population rate have been lower than peers, signaling frictions in the labor market.
  - High school dropout rates may explain lower human capital accumulation.
  - Uruguay’s labor share decreased in line with the global trend but by less than peers, partly due to long-standing worker protection policies.
- Hours worked:
  - Uruguay average hours worked per week in 2018: 38.4 hours.
  - Average hours worked in EMEs in 2018: 43 hours.
  - Uruguay’s average hours comparable to AE averages despite lower income.
  - Long-term downward trend in hours started in the 1990s and is steeper than in AEs and other EMEs.
- Human capital and the 2002 crisis:
  - The 2002 crisis caused a significant slowdown in years of schooling—especially tertiary and, to a lesser extent, secondary—which was never recovered.
  - This erased Uruguay’s advantage in human capital relative to other EMEs.
  - Hours worked remained 3% below the pre-crisis trend after 2002, adding to long-term reduction.
  - Despite full recovery in GDP levels, labor market issues persisted and were masked by acceleration in capital accumulation during the commodity boom.

### C. The role of the commodity price cycle
- Commodity price dynamics and growth:
  - Commodity export prices rose gradually in early 2000s, paused during 2008-09, resumed up to 2014, then fell rapidly until the pandemic.
  - The commodity boom associated with higher growth in emerging economies including Uruguay; end of boom associated with slowdown.
  - The commodity cycle affected employment and capital accumulation significantly.
- Quantitative assessment (local projections, panel of commodity exporting EMEs, 1960-2019):
  - Commodity export price index s_it from Gruss and Kebhaj (2021) used, with index weights by lagged export shares.
  - Model specification: K = 4 lags of dependent variable; J = 4 lags of commodity terms of trade; H = 5 years horizon; time and country fixed effects; OLS estimation.
  - Results robust to alternative specifications and time periods.

### Effects of commodity export price shocks on output and components
- Impulse-response magnitudes (one percent increase in commodity export prices):
  - GDP: increase by 0.02 percent on impact and 0.1 percent after five years.
  - Investment: raises investment by about 0.3 percent at a five-year horizon.
  - Capital stock: increases by 0.1 percent over a five-year horizon.
  - Employment: increases gradually to about 0.07 percent over a five-year horizon.
  - Average hours worked and TFP: effects not significant and magnitudes small.
- Interpretation:
  - Temporary commodity price shocks can have sizable and long-lasting effects on the level of output primarily through capital accumulation and employment (level effects, not growth rate changes).

### Role of the commodity boom in Uruguay’s 2004–14 growth
- Commodity prices for Uruguay increased on average by 3 percent annually during 2004-14.
- The commodity boom accounts for 1.2 percentage points GDP growth per year above the historical average during 2004-14.
- Disaggregated contributions attributable to the boom:
  - Capital accumulation: 1 percent higher.
  - Employment: 0.7 percent higher.
- Post-boom adjustment:
  - The drop in commodity prices after 2014 accounts for about 0.2 lower growth annually during 2015-19.
  - The post-2014 drop also accounts for negative employment growth and decreasing employment rates observed in that period.

### D. The effects of the pandemic on growth
- Labor market and participation impacts:
  - At the peak of the pandemic, employment in Uruguay fell by 8 percent.
  - 152 thousand workers lost their job.
  - The participation rate dropped by 1.7 percentage points, compared to an average among EM in Latin America of 4.1 percentage points.
  - Those who retained jobs worked significantly fewer hours; in April 2020 people worked on average twelve fewer hours than at the beginning of the year.
- Education and human capital:
  - World Bank estimates: Uruguayan students in 2020 enrolled in primary and secondary education may have lost between 0.79 and 0.63 years of school due to school closures.
  - Because of school closures, the impact on human capital translates into 2.5 percent lower GDP per capita than it would have been otherwise.
- Investment and capital:
  - Investment contracted by 0.5 percent in 2020, implying a negative capital contribution of -1.3% in the year.
  - The impact on the overall capital stock was small.
- Productivity and growth accounting (2020 contributions, Table 4):
  - GDP growth: -5.9%
  - Employment: -1.8%
  - Human capital: 0.0%
  - Hours: -5.7%
  - Capital: -1.3%
  - TFP: 2.9%
- Outlook regarding persistence:
  - If human capital and hours worked remain permanently low (as they did during the 2002 crisis), the level of GDP per worker might never fully recover — implying erosion in GDP per capita but not necessarily long-lasting negative effects on GDP growth.
  - Structural changes from the pandemic (e.g., leveraging advanced Internet access and digitization) might translate into boosts in productivity and investment in the medium-term.

### Post-pandemic growth scenarios and five-year outlook (Table 5: Uruguay: Growth Outlook for the Next Five Years)
- Near-term rebound expectation:
  - Next years likely to see a strong rebound from the pandemic, with potential full recovery in employment and capital and at least a partial recovery in hours worked.
- Commodity cycle scenario:
  - Food prices expected to remain high in the foreseeable future (IMF October 2021 WEO), supporting GDP, employment, and capital accumulation.
  - The commodity cycle could add 1.3 percentage points to the annual GDP growth (average effect in sample of 32 countries over sixty years).
- Growth contribution scenarios (percent):
  - Historical trend (1978-2019): GDP growth 2.5%; Employment 0.5%; Human capital 0.4%; Hours -0.2%; Capital 1.4%; TFP 0.4%.
  - Pre-pandemic trends (2014-19): GDP growth 0.9%; Employment -0.1%; Human capital 0.4%; Hours -0.2%; Capital 1.0%; TFP -0.2%.
  - With high commodity prices (assumes 4.1 percent average increase in export prices for 2020-26): GDP growth 1.3%; Employment 0.5%; Capital 0.8%.
  - With rebound from COVID-19 (assumes part of recovery in 2021; 2022-26 assumes full recovery in employment and capital, partial recovery in hours):
    - GDP growth 0.5%; Employment 0.3%; Human capital 0%; Hours 0.2%; Capital 0%; TFP 0%.
  - Total growth average (2022-26):
    - GDP growth: 2.7%
    - Employment: 0.7%
    - Human capital: 0.4%
    - Hours: 0.0%
    - Capital: 1.8%
    - TFP: -0.2%
- Medium-term requirement:
  - Beyond next five years, structural reforms in key sectors needed to sustain growth (examples: enhancing labor market flexibility, improving efficiency and competitiveness through SOE reform, improving education outcomes, investing in digitalization and green energy).

### E. Structural determinants of inflation and policy implications
- Key structural channels affecting inflation:
  - Labor market rigidity and wage setting:
    - Greater labor market flexibility (competitive wage determination) hypothesized to contribute to a lower long-run inflation rate.
    - Backward-looking wage indexation can lead to higher and stickier inflation and make disinflation initiatives costly.
    - Centralized wage bargaining may increase equilibrium inflation level and stickiness by reducing wage flexibility and inducing backward indexation.
  - Central bank institutions:
    - Central bank independence and transparency associated with lower inflation by committing to price stability and counteracting inflationary bias from discretionary policy.
  - Fiscal discipline:
    - More disciplined fiscal policies (especially linked to the implementation of fiscal rules) help reduce inflation.
- Empirical sample and variables (stylized facts):
  - Sample: 18 Latin American countries, 1995–2018.
  - Fiscal rules: composite indicator equals sum of budget balance rules (BBR), debt rules (DR), expenditure rules (ER), revenue rules (RR) (0–4).
  - Uruguay summary statistics (selected):
    - Inflation: observations 409, mean 0.10, std 0.34, min -0.01, max 6.48.
    - Centralized collective bargaining: obs 348, mean 6.19, std 1.40, min 1.61, max 8.85.
    - Central bank independence: obs 432, mean 0.64, std 0.17, min 0.17, max 0.83.
    - Primary structural balance: obs 291, mean -0.02, std 0.02, min -0.10, max 0.05.
    - Public debt: obs 383, mean 0.41, std 0.22, min 0.04, max 1.81.
    - Fiscal rule: obs 446, mean 0.5, std 0.78, min 0, max 2.
- Empirical findings — structural factors and inflation:
  - Labor market institutions:
    - Wage flexibility and absence of centralized collective bargaining: negative and significant (OLS and GMM); GMM coefficients larger.
    - Numerical interpretation: a 1-unit increase in the wage flexibility score leads to around 20 percent reduction in inflation level (percentage reduction from current inflation level, not percentage point).
    - “Flexibility of labor market regulations” not significant.
  - Central bank institutions:
    - Central bank transparency and independence negative and significant in OLS.
    - In system GMM, central bank transparency remains significant (negative); central bank independence negative but not significant.
    - GMM magnitude: 1-unit increase in central bank transparency score associated with 10 percent reduction in inflation.
  - Fiscal institutions:
    - Higher structural deficits and public debt levels lead to higher inflation; introduction of fiscal rules reduces inflation.
    - Only public debt significant in both OLS and GMM.
    - Fiscal rule variable significant in GMM; primary structural balance not significant.
    - GMM magnitudes: a 10 percentage point increase in the debt-to-GDP ratio associated with 10 percent increase in inflation; implementing one type of fiscal rule reduces inflation by around 18 percent.
- Quantitative estimates for Uruguay (from 2020 inflation level of 9.8 percent, Table 5):
  - Reduce centralized collective bargaining: 99% reduction → Projected inflation rate 4.9
  - Increase flexible wage determination: 83% reduction → Projected inflation rate 5.3
  - Improve central bank transparency: 34% reduction → Projected inflation rate 7.3
  - Implement fiscal rules: 14% reduction → Projected inflation rate 8.6
  - Reduce public debt: 2% reduction → Projected inflation rate 9.6
  - Note: calculations explored variables one at a time and should be interpreted with caution; fiscal rule calculation assumes no fiscal rule for Uruguay.
- Inflation persistence and institutional linkages:
  - Flexible wages and less centralized bargaining reduce inflation stickiness (interaction estimates negative and significant).
  - Central bank independence may reduce stickiness (negative and significant in GMM only); central bank transparency shows negative but insignificant interaction.
  - Fiscal discipline variables do not show significant interaction effects on stickiness; fiscal channels likely do not operate by lowering inflation persistence.
- Policy recommendations (paper conclusion):
  - Enhance the flexibility of wage setting mechanisms (allow greater decentralization).
  - Strengthen central bank transparency and independence.
  - Increase fiscal disciple (including through effective implementation of the fiscal rule).

### F. Export diversification and the decline of basic manufacturing
- Stylized facts on export structure:
  - Uruguay’s goods export diversification declined over past two decades; number of high-RCA exports fell to about 2/3 of its peak level (peak in late 1990s above 150 high-RCA exports).
  - Services exports’ share relatively constant at around 35 percent of total exports (mainly tourism and select other services such as information technology).
  - Agricultural commodities represent 80 percent of goods exports (40 percent food, 40 percent raw materials).
  - Agricultural exports concentrate in meat, soybeans, and forestry products; meat ~20 percent share of total exports.
  - Material manufacturing and miscellaneous manufacturing combined declined from over 35 percent of total goods exports (about 7 percent of GDP) in early 1990s to less than 10 percent of total exports by 2018.
- Causes and implications:
  - Commodity price boom of 2003-14 attracted capital, labor, and entrepreneurial resources away from non-commodity tradables and to the non-tradable sector, reducing competitiveness of non-commodity exports via exchange rate appreciation and real wage increases.
  - Decline in export product complexity index from 44 to 62 suggests decline in basic manufacturing was not followed by industrial upgrades.
  - Commodity-dominated export structures associated with lower long-term growth and greater macro volatility.
- Machine-learning methodology to explore diversification:
  - Algorithms: product-based KNN, country-based KNN, Singular Value Decomposition (SVD).
  - Data: SITC 4-digit export data for 2017 (786 product categories; around 260 countries).
  - Prediction target: RCA scores by SITC 4-digit categories; top-N recommendations produced.
  - Ensemble weights: .6 (product-based KNN), .2 (country-based KNN), .2 (SVD).
- Algorithm performance (Table 1 summary):
  - Number of recommendations: Product-based KNN 144; Country-based KNN 155; SVD 189.
  - Hit rate (percent): Product-based KNN 47.2; Country-based KNN 40.0; SVD 30.2.
  - Top-100 Hit rate (percent): Product-based KNN 7.6; Country-based KNN 32.9; SVD 21.7.
- Sectoral recommendation patterns (selected SITC 1-digit observations, Table 2):
  - Category 0 (Food and live animals): Actual high-RCA exports Num. 33, Share 34.7%; Product-based KNN recommended Num. 34, Share 23.6%.
  - Category 6 (Basic manufacturing (material based)): Actual Num. 18, Share 18.9%; Product-based KNN Num. 30, Share 20.8%.
  - Category 7 (Machinery & transport equipment): Actual Num. 0, Share 0.0%; Product-based KNN Num. 11, Share 7.6%; SVD Num. 31, Share 16.4%.
  - Models recommend expanding diversity in categories 2 (inedible crude materials), 6 (basic manufacturing), 8 (miscellaneous manufacturing), and potentially 7 (machinery & transport equipment).

### G. Top diversification opportunities and top-20 product recommendations
- Top shifting 2-digit categories (product-based KNN, Table 3 & 4 highlights):
  - Increases suggested: 89 Miscellaneous manufactured articles (A 2.1 → B 5.6), 28 Metalliferous ores & metal scrap (A 1.1 → B 4.2), 72 Machinery specialized for particular industries (A 0.0 → B 2.8), 79 Other transport equipment (A 0.0 → B 2.8).
  - Reductions suggested: 04 Cereals & cereal preparations (A 9.5 → B 6.3), 01 Meat & preparations (A 7.4 → B 4.9), 24 Cork & wood (A 4.2 → B 1.4).
- Top-20 SITC 4-digit product recommendations (product-based KNN, Table 5; Predicted RCA; Actual RCA):
  - 5851 Derivatives of Rubber — Predicted RCA 10.9; Actual RCA 0.0
  - 2711 Crude Fertilizer — Predicted RCA 4.8; Actual RCA 0.0
  - 6731 Iron Wire — Predicted RCA 4.5; Actual RCA 0.0
  - 8951 Metal Office Products — Predicted RCA 3.0; Actual RCA 0.0
  - 7212 Harvesting Machines — Predicted RCA 2.3; Actual RCA 0.4
  - 7213 Dairy Machinery — Predicted RCA 2.1; Actual RCA 0.5
  - 7211 Soil Preparation Machinery — Predicted RCA 2.0; Actual RCA 0.1
  - 6577 Textiles Fabrics for Machinery — Predicted RCA 1.9; Actual RCA 0.1
  - 7428 Miscellaneous Pumps — Predicted RCA 1.9; Actual RCA 0.0
  - 7442 Lifting and Loading Machinery — Predicted RCA 1.9; Actual RCA 0.3
  - 7283 Miscellaneous Mineral Working Tools — Predicted RCA 1.8; Actual RCA 0.1
  - 7821 Trucks and Vans — Predicted RCA 1.8; Actual RCA 0.6
  - 7452 Miscellaneous Non-Electrical Machines — Predicted RCA 1.8; Actual RCA 0.1
  - 6282 Transmission Belts — Predicted RCA 1.8; Actual RCA 0.4
  - 7416 Miscellaneous Heating and Cooling Equipment — Predicted RCA 1.8; Actual RCA 0.0
  - 6793 Iron and Steel Forging — Predicted RCA 1.7; Actual RCA 0.0
  - 6785 Iron Tubes — Predicted RCA 1.7; Actual RCA 0.0
  - 7219 Miscellaneous Agricultural Machinery — Predicted RCA 1.7; Actual RCA 0.2
  - 7188 Miscellaneous Engines — Predicted RCA 1.6; Actual RCA 0.0
  - 7429 Miscellaneous Liquid Pump Parts — Predicted RCA 1.6; Actual RCA 0.0
- Product complexity and strategic fit:
  - Top-20 recommended items are concentrated in machinery and equipment, and material manufacturing.
  - The top-20 products have average Product Complexity Index (PCI) higher than 70 percent of SITC 4-digit categories.
  - Uruguay’s existing export portfolio at the 55 percentile in product complexity.
  - Many recommended items are production tools supporting the agricultural exporting sector (e.g., harvesting machines, dairy machinery).
- Implications for diversification strategy:
  - Models indicate potential to rebalance away from agricultural commodities toward higher value-added manufacturing linked to agriculture and raw materials.
  - Diversifying into SITC categories 7 and 8 may require targeted efforts given limited export experience.
  - Next steps: diagnose constraints and market frictions preventing emergence of recommended products; design policy remedies; identify barriers including needs for human capital and physical capital upgrades.

*Source: IMF staff compilation from Penn World Tables 10.0 and referenced material in the content unit.*

### References _____________________________________________________________________________ 16

### URUGUAY’S POTENTIAL GROWTH: WHAT TO EXPECT AFTER THE PANDEMIC?

### A. A Long-term View on Growth
- Uruguay’s average growth since 1963: 2.3 percent per year.
- Peer countries average growth: 3.6 precent.
- Population dynamics account for about ½ of Uruguay’s lower growth relative to peers.
- Volatility drivers: regional crises, commodity price cycles, and the pandemic.
- Notable crises and episodes:
  - Latin American debt crisis (early 1980s)
  - Brazil’s crisis (late 1990s)
  - Argentinian crisis of 2001 and Uruguay’s 2002 crisis
  - Exceptional growth during 2004-14 driven by rebound from 2002 crisis and favorable commodity cycle
  - Recent stagnation: averaging 0.9 percent in 2015-19 (despite absence of major negative external shock)
- Growth accounting framework:
  - Data source: Penn World Tables 10.0
  - Sample: 1978-2019, 91 countries
  - Growth decomposed into contributions from labor, capital, and TFP
  - Labor contribution further decomposed into average hours worked per worker, human capital (proxied by a Mincer equation of the average years of schooling), and number of people employed (driven by demographic dynamics and employment-to-population rate)
  - Capital estimated using the perpetual inventory method; TFP as residual
- Key long-run contributions (Table 1, 1978-2019):
  - GDP Growth: 2.5%
  - Employment: 0.5%
  - Human Capital: 0.4%
  - Hours: -0.2%
  - Capital: 1.4%
  - TFP: 0.4%
- Subperiod contributions (Table 1 highlights):
  - 1990-2019 GDP growth: 2.9% (Employment 0.4%, Human Capital 0.3%, Hours -0.3%, Capital 1.5%, TFP 0.9%)
  - 2006-09 GDP growth: 5.4% (Employment 1.5%, Human Capital 0.2%, Hours -0.2%, Capital 2.0%, TFP 1.9%)
  - 2010-14 GDP growth: 4.7% (Employment 0.4%, Human Capital 0.3%, Hours -0.2%, Capital 2.6%, TFP 1.6%)
  - 2015-19 GDP growth: 1.3% (Employment -0.1%, Human Capital 0.4%, Hours -0.2%, Capital 1.0%, TFP 0.2%)
- Across four decades, capital accumulation explains nearly 60 percent of Uruguay’s GDP growth; TFP about 15 percent; employment about 20 percent; human capital accumulation about 15 percent. Hours worked contributed negatively (about -0.2 percent per year).
- Cross-country comparison (1978–2019, Table 2 highlights):
  - URY GDP growth: 2.5% (Employment 0.5%, Human capital 0.4%, Hours -0.2%, Capital 1.4%, TFP 0.4%)
  - ARG GDP growth: 1.7% (Employment 0.8%, Human capital 0.4%, Hours -0.2%, Capital 1.2%, TFP -0.4%)
  - BRA GDP growth: 2.4% (Employment 1.0%, Human capital 1.0%, Hours -0.2%, Capital 1.4%, TFP -0.7%)
  - CHL GDP growth: 4.3% (Employment 1.1%, Human capital 0.3%, Hours -0.2%, Capital 2.8%, TFP 0.3%)
  - EM non-oil com. exp. GDP growth: 3.6% (Employment 1.1%, Human capital 0.5%, Hours -0.1%, Capital 2.4%, TFP -0.2%)
  - AE GDP growth: 2.9% (Employment 0.7%, Human capital 0.4%, Hours -0.2%, Capital 1.4%, TFP 0.6%)
- After accounting for population growth, Uruguay’s per capita GDP growth aligned with peers; the lower aggregate GDP growth largely reflects slower population growth (Uruguay population growth: 0.5 percent per year vs 1.1 percent in non-oil commodity exporting EMEs).

### B. Labor and Human Capital
- Main findings on labor contribution:
  - The lower labor contribution relative to peers arises mainly from employment, which reflects slower population growth.
  - Contributions from hours worked and the employment-to-population rate have been lower than peers, signaling frictions in the labor market.
  - High school dropout rates may explain lower human capital accumulation.
  - Uruguay’s labor share decreased in line with the global trend but by less than peers, partly due to long-standing worker protection policies.
- Hours worked:
  - Uruguay average hours worked per week in 2018: 38.4 hours.
  - Average hours worked in EMEs in 2018: 43 hours.
  - Uruguay’s average hours comparable to AE averages despite lower income.
  - Long-term downward trend in hours started in the 1990s and is steeper than in AEs and other EMEs.
  - Data note: ILO alternative data display same pattern.
- Human capital and the 2002 crisis:
  - The 2002 crisis caused a significant slowdown in years of schooling—especially tertiary and, to a lesser extent, secondary—which was never recovered.
  - This erased Uruguay’s advantage in human capital relative to other EMEs.
  - Hours worked remained 3% below the pre-crisis trend after 2002, adding to long-term reduction.
  - Despite full recovery in GDP levels, labor market issues persisted and were masked by acceleration in capital accumulation during the commodity boom.

### C. The Role of the Commodity Price Cycle
- Commodity prices and growth:
  - Commodity export prices rose gradually in early 2000s, paused during 2008-09, resumed up to 2014, then fell rapidly until the pandemic.
  - Commodity boom associated with higher growth in emerging economies including Uruguay; end of boom associated with slowdown.
  - The commodity cycle affected employment and capital accumulation significantly.
- Quantitative assessment:
  - The impact of commodity cycle on growth estimated using local projection method (Jorda, 2005) in a panel of commodity exporting EMEs (1960-2019).
  - Commodity export price index s_it from Gruss and Kebhaj (2021) used; index weights commodity prices by lagged export shares for each country.
  - Model specification includes:
    - K = 4 lags of the dependent variable
    - J = 4 lags of the commodity terms of trade
    - H = 5 years (time horizon for impulse response)
    - Time fixed effects and country fixed effects
    - Estimated by OLS
  - Results robust to alternative specifications and time periods, including inclusion/exclusion of global financial crisis years.

### Key Tables and Figures Referenced
- Figures:
  - Figure 1. Hours Worked and Human Capital in Uruguay
  - Figure 2. Recovery from the 2002 Crisis
  - Figure 3. Impulse Responses to a Commodity Export Price Increase
- Tables:
  - Table 1. Growth Accounting
  - Table 2. Growth Accounting. Cross-Country Comparison, 1978-2019
  - Table 3. Growth Accounting. Difference Between 2014-19 and 2010-14
  - Table 4. Growth Accounting During the Pandemic
  - Table 5. Growth Outlook for the Next Five Years

*Source: IMF staff compilation from Penn World Tables 10.0 and referenced material in the content unit.*

### 12.      Swings in export prices have sizable and long-lasting effect on output and its

### 12.      Swings in export prices have sizable and long-lasting effect on output and its components

### Effects of commodity export price shocks on output and components
- A one percent increase in commodity export prices is estimated to increase GDP by 0.02 percent on impact and 0.1 percent after five years.
- Transmission channels and magnitudes:
  - Investment: a one percent temporary increase in commodity export prices raises investment by about 0.3 percent at a five-year horizon.
  - Capital stock: increases by 0.1 percent over a five-year horizon.
  - Employment: increases gradually to about 0.07 percent over a five-year horizon.
  - Average hours worked and TFP: the effect is not significant and magnitudes are small.
- Interpretation: temporary commodity price shocks can have sizable and long-lasting effects on the level of output (although not on growth) primarily through capital accumulation and employment.

### Role of the commodity boom in Uruguay’s 2004–14 growth
- Commodity prices for Uruguay increased on average by 3 percent annually during 2004-14.
- The commodity boom accounts for 1.2 percentage points GDP growth per year above the historical average during 2004-14.
- Disaggregated contributions:
  - Capital accumulation: 1 percent higher (attributable to the commodity boom).
  - Employment: 0.7 percent higher (attributable to the commodity boom).
- Post-boom adjustment:
  - The drop in commodity prices after 2014 accounts for about 0.2 lower growth annually during 2015-19.
  - The post-2014 drop also accounts for the negative employment growth and decreasing employment rates observed in that period.

### The effects of the pandemic on growth
- Labor market and participation:
  - At the peak of the pandemic, employment in Uruguay fell by 8 percent.
  - 152 thousand workers lost their job.
  - The participation rate dropped by 1.7 percentage points, compared to an average among EM in Latin America of 4.1 percentage points.
  - Those who retained jobs worked significantly fewer hours; in April 2020 people worked on average twelve fewer hours than at the beginning of the year.
- Education and human capital:
  - Relying on estimates from a World Bank study, Uruguayan students in 2020 enrolled in primary and secondary education may have lost between 0.79 and 0.63 years of school due to school closures.
  - Because of school closures, the impact on human capital translates into 2.5 percent lower GDP per capita than it would have been otherwise.
- Investment and capital:
  - Investment contracted by 0.5 percent in 2020, implying a negative capital contribution of -1.3% in the year.
  - The impact on the overall capital stock was small.
- Productivity and growth accounting (Table 4: Uruguay: Growth Accounting During the Pandemic):
  - Contributions in 2020:
    - GDP growth: -5.9%
    - Employment: -1.8%
    - Human capital: 0.0%
    - Hours: -5.7%
    - Capital: -1.3%
    - TFP: 2.9%
- Outlook regarding persistence:
  - If human capital and hours worked remain permanently low (as they did during the 2002 crisis), the level of GDP per worker might never fully recover — implying erosion in GDP per capita but not necessarily long-lasting negative effects on GDP growth.
  - Structural changes from the pandemic (e.g., leveraging advanced Internet access and digitization) might translate into boosts in productivity and investment in the medium-term.

### Post-pandemic growth scenarios and five-year outlook
- Near-term rebound:
  - The next years are likely to see a strong rebound from the pandemic, with potential full recovery in employment and capital and at least a partial recovery in hours worked.
- Commodity cycle scenario:
  - Food prices are expected to remain high in the foreseeable future (IMF October 2021 WEO), supporting GDP, employment, and capital accumulation.
  - The commodity cycle could add 1.3 percentage points to the annual GDP growth (average effect of commodity export prices on GDP growth in a sample of 32 countries over sixty years).
- Table 5: Uruguay: Growth Outlook for the Next Five Years (contributions)
  - Historical trend (1978-2019):
    - GDP growth: 2.5%
    - Employment: 0.5%
    - Human capital: 0.4%
    - Hours: -0.2%
    - Capital: 1.4%
    - TFP: 0.4%
  - Pre-pandemic trends (2014-19):
    - GDP growth: 0.9%
    - Employment: -0.1%
    - Human capital: 0.4%
    - Hours: -0.2%
    - Capital: 1.0%
    - TFP: -0.2%
  - With high commodity prices (note: commodity price boom assumes a 4.1 percent average increase in export prices for the period 2020-26):
    - GDP growth: 1.3%
    - Employment: 0.5%
    - Capital: 0.8%
  - With rebound from COVID-19 (assumes part of the recovery occurs in 2021, and for 2022-26 assumes a full recovery in employment and capital, and a partial recovery in hours worked):
    - GDP growth: 0.5%
    - Employment: 0.3%
    - Human capital: 0%
    - Hours: 0.2%
    - Capital: 0%
    - TFP: 0%
  - Total growth average (2022-26):
    - GDP growth: 2.7%
    - Employment: 0.7%
    - Human capital: 0.4%
    - Hours: 0.0%
    - Capital: 1.8%
    - TFP: -0.2%
- Medium-term requirement:
  - Beyond the next five years, structural reforms in key sectors will be needed to sustain growth (examples cited: enhancing labor market flexibility, improving efficiency and competitiveness through SOE reform, improving education outcomes, investing in digitalization and green energy).

### Structural determinants of inflation (summary of hypotheses and implications)
- Key structural channels affecting inflation:
  - Labor market rigidity and wage setting:
    - Greater labor market flexibility (competitive wage determination) is hypothesized to contribute to a lower long-run inflation rate.
    - Backward-looking wage indexation can lead to higher and stickier inflation and make disinflation initiatives costly.
    - Centralized wage bargaining may increase equilibrium inflation level and stickiness by reducing wage flexibility and inducing backward indexation.
  - Central bank institutions:
    - Central bank independence and transparency are associated with lower inflation by committing to price stability and counteracting inflationary bias from discretionary policy.
  - Fiscal discipline:
    - More disciplined fiscal policies (especially linked to the implementation of fiscal rules) help reduce inflation.
- Policy implication noted: implementing reforms along labor market flexibility, fiscal discipline, and central bank institutional quality in Uruguay could have meaningful impact in permanently lowering inflation.

*Source: IMF staff analysis in chapter "Swings in export prices have sizable and long-lasting effect on output and its components" from the Uruguay country report.*

### 9. Responsible fiscal policies can support a lower inflation level. Countries with a large

### 1uryea2022002 - 9. Responsible fiscal policies can support a lower inflation level. Countries with a large

### Key hypothesis and motivation
- A large negative fiscal imbalance may increase the incentive to use inflation as an intangible tax (Cottarelli et al, 1998).
- Deficit spending may bid up labor costs which, combined with wage rigidity, may perpetuate a higher inflation level.
- Hypothesis: lower public debt and government deficit, as well as the presence of fiscal rules, should be associated with lower inflation.
- Responsible fiscal management is especially important for countries dependent on commodity exports (case studies e.g. Holden, 2013).

### Data and stylized facts
- Sample: 18 Latin American countries: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela.
- Time period: 1995–2018 (some structural variables shorter).
- Labor market indicators:
  - “Flexibility of wage determination” (GCR) available from 2007; Fraser Institute’s “centralized collective bargaining” indicator runs from 2000 and harmonizes earlier years.
  - Fraser Institute composite index on “flexibility of labor market regulations”.
- Central bank institutions:
  - Central bank independence (Garriga, 2016) updated to 2012.
  - Central bank transparency (Dincer and Eichengreen, 2014) runs from 1998 to 2014; latest observations extrapolated to 2018.
- Fiscal responsibility:
  - Primary structural balance from WEO used to mitigate business cycle influence.
  - Fiscal rules data from IMF’s Fiscal Rules and Fiscal Councils Database; four types: budget balance rules (BBR), debt rules (DR), expenditure rules (ER), revenue rules (RR). Composite fiscal rule indicator equals sum of these (0–4). Database covers 10 sample countries and updated to 2014; last observations extrapolated to 2018.
- Uruguay: Table 1 summary statistics (variable, observations, mean, std, min, max)
  - Inflation: 409, 0.10, 0.34, -0.01, 6.48
  - Centralized collective bargaining: 348, 6.19, 1.40, 1.61, 8.85
  - Flexibility of labor market regulations: 349, 5.19, 1.11, 2.10, 7.37
  - Wage flexibility: 196, 4.68, 0.89, 2.21, 6.11
  - Central bank transparency: 210, 4.89, 2.10, 1.00, 8.50
  - Central bank independence: 432, 0.64, 0.17, 0.17, 0.83
  - Primary structural balance: 291, -0.02, 0.02, -0.10, 0.05
  - Public debt: 383, 0.41, 0.22, 0.04, 1.81
  - Fiscal rule: 446, 0.5, 0.78, 0, 2

- Cross-sectional correlations (simple, without controls):
  - Negative relationship between labor market flexibility and inflation: more centralized collective bargaining associated with higher inflation; more flexibility of wage determination associated with lower inflation.
  - Central bank transparency appears negatively correlated with inflation; central bank independence shows no clear link (data limitations).
  - Higher structural primary balance appears associated with lower inflation on average.
  - Presence of fiscal rules associated with lower inflation (correlation not significant).
  - Average public debt level shows no clear correlation with inflation in simple plots.

### Empirical method
- Model (equation 1): inflation π_{i,t} = α_i + β π_{i,t-1} + γ X_{i,t} + δ s_{i,t} + ε_{i,t}, where s_{i,t} is structural factor and X_{i,t} are controls.
- Control variables: output gap, primary fiscal balance, monetary base as percent of GDP, import-to-GDP ratio, world inflation.
- Expected signs: output gap and base money positive; import ratio expected negative; world inflation positive.
- Estimation: OLS and system GMM (Blundell and Bond, 1998) to address dynamic panel bias (Nickell, 1981).

### Results — Structural factors and inflation
- Labor market institutions (Table 2):
  - Wage flexibility and absence of centralized collective bargaining: negative and significant in OLS and GMM; GMM coefficients larger.
  - Numerical interpretation: a 1-unit increase in the wage flexibility score (i.e., 1-unit reduction in centralized collective bargaining score) leads to around 20 percent reduction in inflation level. (Note: percentage reduction from current inflation level, not percentage point.)
  - “Flexibility of labor market regulations” not significant.
- Controls:
  - log(world inflation) significant with expected positive sign in most specifications.
  - Output gap mostly positive but not always significant.
  - Import ratio mostly positive and not significant in most specifications (contrary to prior).
  - Higher primary balance mostly associated with higher inflation (not significant)—may reflect business cycle capture.
  - Base money coefficients not significant.
- Central bank institutions (Table 3):
  - Central bank transparency and independence negative and significant in OLS.
  - In system GMM, central bank transparency remains significant (negative), central bank independence coefficient negative but not significant.
  - Magnitude: GMM indicates a 1-unit increase in central bank transparency score associated with 10 percent reduction in inflation.
- Fiscal institutions (Table 4):
  - Higher structural deficits and public debt levels lead to higher inflation; introduction of fiscal rules reduces inflation (consistent with hypothesis).
  - Only public debt is significant in both OLS and GMM.
  - Fiscal rule variable significant in GMM; primary structural balance not significant in either setting.
  - Interpretation: debt level is accumulated fiscal management signal; fiscal rules may both constrain policy and signal seriousness.
  - Magnitudes (GMM): a 10 percentage point increase in the debt-to-GDP ratio is associated with 10 percent increase in inflation, other things equal; implementing one type of fiscal rule reduces inflation by around 18 percent.

### Quantitative estimates for Uruguay (Table 5)
- Estimated percentage reduction in inflation (from 2020 inflation level of 9.8 percent) if Uruguay’s scores reached LAC sample averages:
  - Reduce centralized collective bargaining: 99% reduction → Projected inflation rate (in %) 4.9
  - Increase flexible wage determination: 83% reduction → Projected inflation rate (in %) 5.3
  - Improve central bank transparency: 34% reduction → Projected inflation rate (in %) 7.3
  - Implement fiscal rules: 14% reduction → Projected inflation rate (in %) 8.6
  - Reduce public debt: 2% reduction → Projected inflation rate (in %) 9.6
- Note: these calculations explored variables one at a time and should be interpreted with caution. The fiscal rule calculation assumes no fiscal rule for Uruguay (did not take into account the fiscal rule recently implemented in Uruguay).

### Inflation persistence and institutional linkages
- Structural factors can increase inflation persistence (stickiness):
  - Backward wage indexation to past inflation can make downward adjustment harder and raises output/employment costs of disinflation.
  - Labor market institutions encouraging widespread wage indexation increase stickiness.
  - Lack of central bank transparency and independence can lead to stickier inflation by limiting the central bank’s tools or resolve to change the status quo.

*Italic: International Monetary Fund — Uruguay chapter content unit.*

### 23. To explore the effect of structural factors on inflation stickiness, we modify the

### 1uryea2022002 - 23. To explore the effect of structural factors on inflation stickiness, we modify the

### Methodology: baseline modification to assess inflation stickiness
- Estimation augments the baseline specification with an interaction between lagged inflation and a structural factor:
  - 휋i,t = 훼i + 훽휋i,t−1 + 훾Xi,t + 훿si,t + 휃휋i,t−1·si,t + 휀i,t  (Equation 2)
- Interpretation:
  - 훽 captures baseline inflation stickiness (positive 훽 implies stickiness).
  - 휃 (coefficient on interaction term) positive (negative) means the structural factor increases (decreases) inflation stickiness.

### Findings — Labor market flexibility and inflation stickiness (Table 6)
- Interaction term estimates:
  - Wage flexibility: interaction term is negative and significant in both OLS and GMM specifications (log(inflation) x Wage flexibility = -0.21** (OLS), -0.38** (Dynamic panel/GMM)).
  - Centralized collective bargaining: interaction term is negative and significant (log(inflation) x Centr. collective bargaining = -0.11** (OLS), -0.17** (Dynamic panel/GMM)).
  - Labor market regulations: interaction term negative but insignificant (log(inflation) x Labor market regulations = -0.13 (OLS), -0.13 (Dynamic panel/GMM)).
- Implication:
  - Flexible wages and less centralized bargaining are associated with reduced inflation stickiness, supporting a channel through which flexible wages can lower inflation.
- Selected coefficient context (from Table 6):
  - log(inflation) -1: 1.45*** (Column 1 OLS), 2.21** (Column 2 Dynamic panel), 1.17*** (Column 3 OLS), 1.47** (Column 4 Dynamic panel), 1.31*** (Column 5 OLS), 1.14 (Column 6 Dynamic panel).
  - Wage flexibility -1: -0.86*** (OLS), -1.33** (Dynamic panel).
  - Centralized collective bargaining -1: -0.47*** (OLS), -0.65** (Dynamic panel).
  - Output gap: 0.03* (Column 1), 0.03 (2), -0.00 (3), -0.01 (4), 0.01 (5), 0.00 (6).
  - Import ratio: 0.15 (1), 0.19 (2), 0.21** (3), 0.20 (4), 0.16 (5), 0.12 (6).
  - log(world inflation): 0.44*** (1), 0.46*** (2), 0.46*** (3), 0.53*** (4), 0.39*** (5), 0.48*** (6).
  - Observations: 110 (columns 1–2), 193 (columns 3–6).
  - R-squared reported: 0.51, 0.48, 0.40 (as applicable).

### Findings — Central bank institutions and inflation stickiness (Table 7)
- Identification caveat: limited within-country variation and shorter time coverage for central bank institution variables make identification harder.
- Interaction term estimates:
  - Central bank transparency: interaction term is negative but insignificant (log(inflation) x Central bank transparency = -0.04 (OLS), -0.61 (Dynamic panel)).
  - Central bank independence: interaction term is negative and significant in the GMM estimate (log(inflation) x Central bank independence = -6.54* (Dynamic panel)), but positive and insignificant in OLS (0.32 (OLS)).
- Selected coefficient context (from Table 7):
  - log(inflation) -1: 0.69*** (Column 1), 3.54 (2), 0.34 (3), 4.69* (4).
  - Central bank transparency -1: -0.16 (OLS), -1.64 (Dynamic panel).
  - Central bank independence -1: 0.46 (OLS), -21.35 (Dynamic panel).
  - Output gap: 0.04* (1), 0.11 (2), 0.01 (3), -0.01 (4).
  - Import ratio: -0.33** (1), 6.06 (2), 0.15 (3), 0.76 (4).
  - log(world inflation): 0.24** (1), 1.74 (2), 0.37*** (3), 0.16 (4).
  - Observations: 133 (columns 1–2), 217 (columns 3–4).
  - R-squared reported: 0.42, 0.44.

### Findings — Fiscal responsibility and inflation stickiness (Table 8)
- Interaction term estimates for fiscal variables:
  - Primary structural balance: interaction term not significant (log(inflation) x Structural balance = -4.20 (OLS), 22.75 (Dynamic panel); not significant).
  - Public debt: interaction term not significant (log(inflation) x Public debt = 0.04 (OLS), 0.23 (Dynamic panel); not significant).
  - Fiscal rule: interaction term not significant (log(inflation) x Fiscal rule = 0.03 (OLS), 0.18 (Dynamic panel); not significant).
- Additional identification issue:
  - Adding the interaction term appears to reduce identification for the structural factor itself in the cases of fiscal rules and debt level (coefficients lose precision/significance).
- Implication:
  - Although fiscal discipline correlates with lower inflation (as shown elsewhere in the paper), the mechanism is unlikely to operate through reducing inflation stickiness.
- Selected coefficient context (from Table 8):
  - log(inflation) -1: 0.50*** (1), 0.82 (2), 0.55*** (3), 0.14 (4), 0.54*** (5), 0.18 (6).
  - Primary structural balance: -16.01* (OLS), 40.53 (Dynamic panel).
  - Public debt: 0.41 (OLS), 1.14 (Dynamic panel).
  - Fiscal rule -1: 0.01 (OLS), 0.34 (Dynamic panel).
  - Base money: 0.52 (1), -13.03 (2), 0.00*** (3), 0.00 (4), 0.00*** (5), 0.00 (6).
  - log(world inflation): 0.36*** (1), 0.94 (2), 0.40*** (3), 0.44** (4), 0.36*** (5), 0.56*** (6).
  - Observations: 179 (columns 1–2), 229 (3–4), 236 (5–6).
  - R-squared reported: 0.43, 0.60, 0.61.

### Overall conclusion on inflation stickiness mechanisms
- Summary of empirical evidence:
  - Flexible wage determination and less centralized collective bargaining reduce inflation stickiness (negative and significant interaction terms).
  - Central bank independence may reduce stickiness (negative and significant in GMM only); central bank transparency shows negative but insignificant interaction.
  - Fiscal discipline variables do not show significant interaction effects on stickiness; fiscal channels likely do not operate by lowering inflation persistence.
- Policy-relevant inference:
  - One channel through which structural reforms lower inflation is by reducing inflation persistence when such reforms affect wage-setting and bargaining structures.

### Policy recommendations (from paper conclusion)
- Structural reforms that would support Uruguay’s disinflation plan:
  - Enhance the flexibility of wage setting mechanisms (to allow for greater decentralization).
  - Strengthen central bank transparency and independence.
  - Increase fiscal disciple (including through an effective implementation of the fiscal rule).
- Evidence base:
  - More flexible wage determination, greater central bank transparency and independence, and greater fiscal discipline are each associated with lower inflation to various extents.

### Export diversification: patterns and implications for Uruguay
- Key stylized facts:
  - Uruguay’s goods export diversification has declined over the past two decades; number of high-RCA exports fell to about 2/3 of its peak level (peak in late 1990s above 150 high-RCA exports).
  - Services exports’ share in total exports has been relatively constant over the past three decades at around 35 percent of total exports (mainly tourism and select other services such as information technology).
  - Currently agricultural commodities represent 80 percent of the country’s goods exports (40 percent food, 40 percent raw materials).
  - Agricultural exports concentrate in meat, soybeans, and forestry products; meat ~20 percent share of total exports.
  - Material manufacturing and miscellaneous manufacturing combined declined from over 35 percent of total goods exports (about 7 percent of GDP) in early 1990s to less than 10 percent of total exports by 2018.
- Implications from literature and analysis:
  - Commodity-dominated export structures are associated with lower long-term growth and greater macroeconomic volatility.
  - Diversification should be growth-friendly and aligned with comparative advantage; not all diversification is equally beneficial.
- Machine-learning recommendation (high-level):
  - Algorithms suggest diversification could rebalance away from commodities toward higher value-added manufacturing categories that support agricultural and raw material production.

*URUGUAY INTERNATIONAL MONETARY FUND*

### 12. The decline of basic manufacturing

### 12. The decline of basic manufacturing

### Overview and main concern
- The decline of basic manufacturing is not necessarily a concern by itself, but it is problematic when it is not replaced by other non-commodity exports of higher value added.
- Uruguay experienced a decline in ranking in the export product complexity index from 44 to 62, suggesting the decline in basic manufacturing was not followed by industrial upgrades.

### Causes of export concentration and decline in non-commodity exports
- Commodity price boom of 2003-14 contributed to the decline in non-commodity exports and greater concentration:
  - The boom attracted capital, labor, and entrepreneurial resources away from the non-commodity tradable sector to the commodity sector and to the non-tradable sector due to increased demand and prices.
  - Exchange rate appreciation and real wage increases during the boom reduced competitiveness of the non-commodity exporting sector.
  - The net effect: a more concentrated export structure in primary commodities.
- Summary of drivers (textual synthesis from source):
  - Increased dominance of commodity exports.
  - Real exchange rate appreciation and domestic wage increases reducing competitiveness of traditional non-commodity exports.
  - Few new non-commodity export categories emerged to replace those that lost traction.
- Policy challenge: Reduce disruptive impact of commodity price cycles on the normal industrialization process while maintaining comparative advantage in commodity exports.

### Machine learning methodology to explore export diversification options
- Methodological intuition:
  - Insights about latent comparative advantages can be gained by looking at other countries with similar comparative advantages.
  - A country is likely to have latent comparative advantage in products closely related to existing high-RCA products it already exports.
- Algorithms employed:
  - Product-based K Nearest Neighbors (KNN)
  - Country-based K Nearest Neighbors (KNN)
  - Singular Value Decomposition (SVD)
- Implementation details:
  - Goal: "top-N recommendations" — generate list of N SITC 4-digit product categories with largest predicted RCA for the country.
  - Prediction target: RCA scores by SITC 4-digit categories, using training dataset of export values by country and export volumes.
  - For most implementations, y = 1 (single-year cross-country export data used).
  - Cosine similarity is used as the similarity measure in KNN implementations.
  - SVD optimization minimizes: min_{u_i,v_j} ∑_{r_{ij}∈R}(r_{ij} − u_i · v_j)^2 + λ(‖u_i‖^2 + ‖v_j‖^2), solved with stochastic gradient descent using the python Surprise library.

### Algorithmic strengths and weaknesses
- KNN (product-based and country-based):
  - Intuitive and easy to implement.
  - Drawbacks: limited scalability as m and n increase (O(n) time for similarity calculations); poor performance with sparse data because similarities among vectors become inaccurate when many missing entries exist.
  - Product-based KNN likely superior to country-based KNN in this context because n > m, making relatedness among products easier to identify than similar countries.
- SVD (matrix factorization):
  - Addresses sparsity by representing countries and products as combinations of latent factors (U, S, V′).
  - Latent factors are not human-interpretable in practice but efficiently characterize the data.
  - Recommended products are those with highest predicted r̂_{ij} = û_i · v̂_j.

### Application to Uruguay (data and setup)
- Data: SITC 4-digit export data for 2017.
  - 786 product categories.
  - around 260 countries.
- Summary statistics from Table 1 (algorithm recommendations):
  - Number of recommendations:
    - Product-based KNN: 144
    - Country-based KNN: 155
    - SVD: 189
  - Hit rate (in percent):
    - Product-based KNN: 47.2
    - Country-based KNN: 40.0
    - SVD: 30.2
  - Top-100 Hit rate (in percent):
    - Product-based KNN: 7.6
    - Country-based KNN: 32.9
    - SVD: 21.7
- Weighted ensemble used in analysis:
  - Weights: .6 (product-based KNN), .2 (country-based KNN), .2 (SVD).

### Model recommendations and sectoral implications
- General recommendation:
  - Models recommend a more balanced export portfolio, shifting somewhat away from agricultural commodities and towards non-commodity manufacturing exports.
- Observations by SITC 1-digit category (selected figures from Table 2):
  - Category 0 (Food and live animals):
    - Actual high-RCA exports (2017): Num. 33, Share 34.7%
    - Product-based KNN recommended: Num. 34, Share 23.6%
    - Country-based KNN recommended: Num. 63, Share 40.6%
    - SVD recommended: Num. 29, Share 15.3%
  - Category 2 (Inedible crude materials exc. fuel):
    - Actual: Num. 14, Share 14.7%
    - Product-based KNN: Num. 25, Share 17.4%
    - Country-based KNN: Num. 28, Share 18.1%
    - SVD: Num. 44, Share 23.3%
  - Category 6 (Basic manufacturing (material based)):
    - Actual: Num. 18, Share 18.9%
    - Product-based KNN: Num. 30, Share 20.8%
    - Country-based KNN: Num. 27, Share 17.4%
    - SVD: Num. 39, Share 20.6%
  - Category 7 (Machinery & transport equipment):
    - Actual: Num. 0, Share 0.0%
    - Product-based KNN: Num. 11, Share 7.6%
    - Country-based KNN: Num. 6, Share 3.9%
    - SVD: Num. 31, Share 16.4%
  - Category 8 (Miscellaneous manufacturing):
    - Actual: Num. 4, Share 4.2%
    - Product-based KNN: Num. 15, Share 10.4%
    - Country-based KNN: Num. 4, Share 2.6%
    - SVD: Num. 17, Share 9.0%
- Interpretation:
  - Models detect strong existing comparative advantages in SITC category 0 but limited further diversification potential within that category.
  - To raise the total number of high-RCA exports, Uruguay would need to expand diversity in categories 2 (inedible crude materials), 6 (basic manufacturing), and 8 (other manufacturing), and potentially in category 7 (machinery and transport equipment), where Uruguay currently has no high-RCA exports but model recommendations suggest latent comparative advantages.

### Top diversification opportunities and contractions (product-based KNN results)
- Top 10 SITC 2-digit Categories with Increased Shares in Total High-RCA Exports (Table 3) — (A = Actual share, B = Rec. share, Difference B − A):
  - 89 Miscellaneous manufactured articles: A 2.1, B 5.6, Difference 3.5
  - 28 Metalliferous ores & metal scrap: A 1.1, B 4.2, Difference 3.1
  - 72 Machinery specialized for particular industries: A 0.0, B 2.8, Difference 2.8
  - 79 Other transport equipment: A 0.0, B 2.8, Difference 2.8
  - 65 Textile yarn: A 4.2, B 6.9, Difference 2.7
  - 87 Professional & scientific instruments: A 0.0, B 2.1, Difference 2.1
  - 42 Fixed vegetable oils & fats: A 0.0, B 1.4, Difference 1.4
  - 25 Pulp & wastepaper: A 1.1, B 2.1, Difference 1.0
  - 84 Apparel & clothing accessories: A 1.1, B 2.1, Difference 1.0
  - 26 Textile fibers and their wastes: A 3.2, B 4.2, Difference 1.0
- Top 10 SITC 2-digit Categories with Reduced Shares in Total High-RCA Exports (Table 4) — (A = Actual share, B = Rec. share, Difference B − A):
  - 59 Chemical materials & products: A 5.3, B 4.2, Difference -1.1
  - 22 Oil seeds & oleaginous fruit: A 2.1, B 0.7, Difference -1.4
  - 02 Dairy products & birds’ eggs: A 4.2, B 2.8, Difference -1.4
  - 05 Vegetables & fruit: A 4.2, B 2.8, Difference -1.4
  - 55 Oils & perfume; toilet & cleansing preparations: A 3.2, B 1.4, Difference -1.8
  - 53 Dyeing, tanning & coloring materials: A 4.2, B 2.1, Difference -2.1
  - 01 Meat & preparations: A 7.4, B 4.9, Difference -2.5
  - 12 Tobacco & tobacco manufactures: A 4.2, B 1.4, Difference -2.8
  - 24 Cork & wood: A 4.2, B 1.4, Difference -2.8
  - 04 Cereals & cereal preparations: A 9.5, B 6.3, Difference -3.2

### Implications for policy and diversification strategy
- Models indicate potential to rebalance Uruguay’s export portfolio toward greater participation in non-commodity manufacturing and select inedible crude materials.
- Policy implications (drawn from analysis in source):
  - Address vulnerabilities from commodity-dependent specialization by promoting industrial upgrading and emergence of new non-commodity export categories.
  - Mitigate impacts of commodity price cycles on industrialization while preserving comparative advantage in commodities.
  - Use model recommendations (product-based KNN, country-based KNN, SVD) to identify concrete 4-digit SITC products where Uruguay may have latent comparative advantage and target policies to develop capabilities for those products.

*Source: Chapter 12, "The decline of basic manufacturing", 1uryea2022002 - 12. The decline of basic manufacturing*

### 35. Table 5 provides greater granularity by focusing on the top 20 recommendations of

### 1uryea2022002 - 35. Table 5 provides greater granularity by focusing on the top 20 recommendations of

### Top-20 SITC 4-Digit Product Recommendations (Predicted RCA; Actual RCA)
- According to the results from the product-based KNN (products with Actual RCA < 1):
  - 5851 Derivatives of Rubber — Predicted RCA 10.9; Actual RCA 0.0
  - 2711 Crude Fertilizer — Predicted RCA 4.8; Actual RCA 0.0
  - 6731 Iron Wire — Predicted RCA 4.5; Actual RCA 0.0
  - 8951 Metal Office Products — Predicted RCA 3.0; Actual RCA 0.0
  - 7212 Harvesting Machines — Predicted RCA 2.3; Actual RCA 0.4
  - 7213 Dairy Machinery — Predicted RCA 2.1; Actual RCA 0.5
  - 7211 Soil Preparation Machinery — Predicted RCA 2.0; Actual RCA 0.1
  - 6577 Textiles Fabrics for Machinery — Predicted RCA 1.9; Actual RCA 0.1
  - 7428 Miscellaneous Pumps — Predicted RCA 1.9; Actual RCA 0.0
  - 7442 Lifting and Loading Machinery — Predicted RCA 1.9; Actual RCA 0.3
  - 7283 Miscellaneous Mineral Working Tools — Predicted RCA 1.8; Actual RCA 0.1
  - 7821 Trucks and Vans — Predicted RCA 1.8; Actual RCA 0.6
  - 7452 Miscellaneous Non-Electrical Machines — Predicted RCA 1.8; Actual RCA 0.1
  - 6282 Transmission Belts — Predicted RCA 1.8; Actual RCA 0.4
  - 7416 Miscellaneous Heating and Cooling Equipment — Predicted RCA 1.8; Actual RCA 0.0
  - 6793 Iron and Steel Forging — Predicted RCA 1.7; Actual RCA 0.0
  - 6785 Iron Tubes — Predicted RCA 1.7; Actual RCA 0.0
  - 7219 Miscellaneous Agricultural Machinery — Predicted RCA 1.7; Actual RCA 0.2
  - 7188 Miscellaneous Engines — Predicted RCA 1.6; Actual RCA 0.0
  - 7429 Miscellaneous Liquid Pump Parts — Predicted RCA 1.6; Actual RCA 0.0

### Product complexity and sectoral focus
- Many top-20 recommended items are in machinery and equipment, and material manufacturing.
- The top-20 products recommended have an average Product Complexity Index (PCI) that is higher than 70 percent of the SITC 4-digit product categories (according to the ranking by the Observatory of Economic Complexity).
- Uruguay’s existing export portfolio is at the 55 percentile in the product complexity ranking.
- Many recommended items are production tools that support the agricultural exporting sector (examples cited: harvesting machines, dairy machinery, textile fabrics for machinery).

### Findings on export structure and implications
- Uruguay has become increasingly concentrated in the agricultural sector over the past 20 years.
- Since the late 1990s, the number of high-RCA export products in SITC 4-digit categories has dropped by close to 40 percent.
- The agricultural commodity price boom has been a key contributing factor, attracting resources away from manufacturing and the normal industrialization process.
- Diversification potential exists in higher value-added manufacturing linked to agriculture and other traditional exports.
- Manufacturing categories that provide production means for the agricultural and raw material sectors are identified as promising diversification areas.
- Products recommended are relatively more knowledge-intensive than Uruguay’s current export profile but remain linked to the country’s comparative advantages in agriculture.
- Moving into recommended products could leverage domestic product and customer networks, opportunities for knowledge exchange, and feedback loops for product research & development.
- Prices of agriculture-related manufacturing products are less variable than those of raw commodities, potentially shielding the economy from commodity price volatility to some extent.

### Policy implications and next steps
- Diversifying into SITC categories 7 and 8 may require targeted efforts, given Uruguay’s relatively limited export experience in these categories.
- More in-depth analysis is needed to:
  - Identify the main constraints and market frictions preventing the emergence of the recommended products and sectors.
  - Design related policy remedies to support diversification.
  - Identify key barriers to diversification despite apparent latent comparative advantages (including the need for maintenance and upgrades of human capital and physical capital).

*Source: 1uryea2022002*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2022/english/1uryea2022002.pdf_