## _wp14150

## Source details

**Canonical URL:** [_wp14150](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14150.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14150.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14150.pdf.json)

---

### I. Introduction
- EAC member countries are Burundi, Kenya, Rwanda, Tanzania, and Uganda.
- The EAC’s (unweighted) average growth rate in 2004–13 was 6.2 percent, placing it in the top one-fifth of the distribution of 10-year growth rate episodes experienced by all countries worldwide since 1960.
- Purpose: analyze disaggregated sources of past growth (composition of output and exports by sector, indicators of product quality and sophistication) to glean clues about the sustainability of recent strong growth.
- Note: Analysis relies on past growth and does not incorporate implications of possible exploitation of recent natural resource finds.

### II. Overall growth performance and reversion to the mean
- Complementary indicators corroborating GDP growth:
  - Electricity consumption (long series for Kenya and Tanzania) grew rapidly with acceleration during the past ten years.
  - Health outcomes improved: infant mortality rates and life expectancy improved considerably in most countries.
  - Fiscal revenues and private sector credit (in domestic currency at constant prices) rose rapidly; indices set to 100 in year 2000.
- Ten-year growth rarity and predictive exercise:
  - Worldwide distribution of 10-year growth episodes (1950–2010) has mean 3.8 percent and standard deviation 2.6 percent.
  - EAC unweighted average 2004–13 growth = 6.2 percent — within the top one-fifth percent of the worldwide distribution of decade-long episodes since 1950.
  - Regression estimated on decade-long growth for a panel of 162 countries over 1960–2010 (with decade fixed effects): estimated coefficients and statistics:
    - ߚ = 0.23
    - ߤ = 2.8
    - ߪ = 2.54
  - Using the estimated model and input g_past = 6.2, the conditional simulation for next-decade growth is:
    - g_next = 4.2 + ߝ, where ߝ ~ N(0, 2.54^2)
  - Interpretation: conditional on historical persistence patterns, there is a little less than a one in four chance of another decade of growth above 6 percent for the EAC.

### III. Structural transformation, diversification, and rising sophistication
- Long-run sectoral trends (1970–2010, UN national accounts, seven broad sectors):
  - Agriculture share declined from almost one half (unweighted average) in 1970 to about a third in 2010.
  - Decline in agriculture was steeper in the EAC than in Sub-Saharan Africa (SSA) or low-income countries (LIC) worldwide.
  - Largest gains in sector shares observed in transportation and construction; manufacturing gains limited; mining remains a relatively small sector.
- Sectoral composition and changes since 2000 (12-sector decomposition, 2000–11, IMF staff database):
  - Aggregate (EAC unweighted average) sector shares, 2000 vs. 2010:
    - Agriculture: 36% (2000) → 28% (2010)
    - Mining: 1% → 1%
    - Manufacturing: 9% → 9%
    - Utilities: 2% → 2%
    - Construction: 6% → 8%
    - Wholesale trade: 11% → 12%
    - Hotels and restaurants: 5% → 5%
    - Transport and communications: 6% → 9%
    - Financial services: 2% → 3%
    - Real estate & business services: 7% → 7%
    - Public administration: 6% → 7%
    - Other: 9% → 9%
  - Sectoral growth rates, 2000–10 (average annual):
    - Agriculture: 3.2 percent
    - Transport and communications: 10.8 percent
    - (All sectors experienced positive growth on average during the period.)
  - Sector share changes (2000–10) in percentage points (selected largest changes):
    - Agriculture: -7.1
    - Construction: +2.0
    - Transport and communications: +1.7
    - Wholesale trade: +1.5
    - Public administration: +1.2

### IV. Exports: rising share of GDP and geographic diversification
- Exports as a share of GDP for the EAC rose from 12 percent in 1990 to 19 percent in 2010.
- Increase in export/GDP particularly pronounced in Tanzania and Uganda; to a lesser extent in Kenya and Rwanda.
- Partner diversification indicators:
  - Theil index of partner concentration declined (higher diversification).
  - Share of top ten destination countries in total exports declined.

### V. Product diversification of exports and structure of exports
- Agriculture’s share in exports fell from 4/5 (80 percent) in the 1970s–90s to 2/3 (66.666... percent) in 2010, with mirroring gains in manufacturing.
- EAC aggregate snapshots (percent of total exports of goods):
  - 1970: Agriculture 80%, Manufacturing 6%, Mining 14%
  - 1990: Agriculture 85%, Manufacturing 12%, Mining 3%
  - 2009: Agriculture 66%, Manufacturing 23%, Mining 11%
- SSA aggregate snapshots (percent of total exports of goods):
  - 1970: Agriculture 65%, Manufacturing 15%, Mining 20%
  - 1990: Agriculture 51%, Manufacturing 27%, Mining 22%
  - 2009: Agriculture 44%, Manufacturing 29%, Mining 27%

### VI. Number of distinct export products
- The number of distinct export products increased substantially, especially since the early 1990s and particularly in manufacturing.
- Example: Uganda’s distinct products rose from about one hundred in 1980 to more than five hundred in 2010, with more than three hundred products in manufacturing.
- Robustness checks: counts repeated for products with export revenues exceeding US$1 million and US$10 million; results confirm sizable increases, particularly in Kenya, Tanzania, and Uganda.
- Time span of product counts: 1962–2010 (UN COMTRADE, authors’ calculations).

### VII. Sophistication of export products
- EAC countries have generally moved toward exporting more “sophisticated” products, defined as those predominantly produced by advanced economies.
- Sophistication index construction (4-digit SITC, two-step method):
  - Step 1: For each product, a “productivity” index is the weighted average of per capita GDP of exporting countries, weights are the share of the product in each country’s total exports.
  - Step 2: A country’s sophistication index is the weighted average of its export products’ productivity indices, with weights equal to value shares of products in the country’s total exports.
- Empirical findings:
  - The sophistication index of each EAC country rose visibly between 1990 and 2010.
  - Gains for most EAC countries were noticeably greater than the average for sub-Saharan Africa and other regions.
  - Kenya experienced the smallest gain (it started with the highest sophistication index in 1990).
  - By 2010, Kenya and Uganda were slightly above the sub-Saharan average; Tanzania was close to it.
  - Specific plotted index ranges include values between 6.0 and 9.5 for regions/countries in 1990 and 2010.

### VIII. Quality of individual exported items
- Quality measured as the ratio of the value of exported items by the country to the average value of such item exported by all other countries (unit-value-based quality adjusted by Henn et al. 2013).
- Summary measures indicate a trend toward improved quality within each export category for EAC countries.
- Example: In 2009, Burundi’s coffee quality was 0.8 (on a scale from 0 to 1) and coffee represented 80 percent of Burundi’s goods exports in the database.
- Overall pattern:
  - No clear rising trend in quality across all items—consistent with many products being primary commodities or goods involving limited processing.
  - Quality is measured relative to trading partners, suggesting EAC countries broadly keep pace with competitors.
  - Shares of the top ten products in total exports reveal greater diversification in 2009 than in 1980.

### IX. Sectoral output and export composition changes (selected country-level snapshots)
- Sectoral output database:
  - Value added series in constant prices for 12 economy-wide sectors for Sub-Saharan Africa, period 2000–2011.
  - 12 sectors: agriculture (including forestry and fishing); mining and quarrying; manufacturing; utilities (electricity and water); construction; trade and repairs; hotels and restaurants; transportation and communication; financial intermediation; real estate and business activities; public administration; and other. FISIM and net taxes are excluded.
- Selected country export composition changes (shares in percent):
  - Burundi: 1970 — Agriculture 95%, Manufacture 2%, Mining 3%; 2010 — Agriculture 88%, Manufacture 11%, Mining 1%
  - Kenya: 1970 — Agriculture 80%, Manufacture 17%, Mining 3%; 2010 — Agriculture 63%, Manufacture 34%, Mining 3%
  - Rwanda: 1970 — Agriculture 39%, Manufacture 1%, Mining 60%; 2010 — Agriculture 41%, Manufacture 22%, Mining 37%
  - Tanzania: 1970 — Agriculture 90%, Manufacture 7%, Mining 3%; 2010 — Agriculture 65%, Manufacture 24%, Mining 11%
  - Uganda: 1970 — Agriculture 99%, Manufacture 1%, Mining 0%; 2010 — Agriculture 73%, Manufacture 27%, Mining 0%

### X. Interpretation, implications, and outlook
- Recent decade: accelerated overall economic growth and structural transformation in EAC member countries.
- Observed structural shifts:
  - The share of agriculture has fallen substantially.
  - Gains broadly distributed: largest gains to construction, transportation, and wholesale trade; manufacturing and mining posted modest gains.
  - Rapid growth has not been driven by a narrow range of products (e.g., natural resources).
- Improvements over time:
  - Export sophistication and quality of items exported have improved, more noticeably during the past decade.
- Cautionary notes and outlook:
  - The pattern suggests consumer and investment demand for more sophisticated goods and services are beginning to be met as per capita incomes rise.
  - No clear production-side “winners” yet that embed a durable comparative advantage beyond the region.
  - No major quality improvements vis-à-vis competitor countries where progress is also occurring.
  - These observations counsel caution about projecting continued rapid growth far into the future.
  - Findings are consistent with expected continued healthy growth but with an eventual slowdown to a more moderate pace, consistent with the statistical pattern for “reversion to the mean.”

### XI. Methodology highlights (data and indices)
- Export data source: UN COMTRADE, SITC1 classification at the four-digit level, period 1962–2010.
- Theil index of export diversification:
  - For product-level Theil index, x_k is export value of product k, n is number of products, μ is the products’ average dollar value.
  - For partner-level Theil index, x_k is export value to partner k, n is number of partners for each exporter.
- Sophistication index:
  - Product productivity P_k computed as weighted average of per capita GDP of exporters, weights = share of product k in each country’s total exports.
  - Country sophistication is the value-share-weighted average of product productivity across the country’s export products.
- Export quality measurement (Henn et al. 2013):
  - Quality derived from unit values (value/quantity), adjusted for production costs and selection bias from distance; coefficients estimated following Hallak and Schott (2011).
  - Quality estimates normalized by their 90th percentile in the relevant product-year combination and aggregated using current trade values as weights; normalization repeated at each aggregation level.

*Source: _wp14150 - References .............................................................................................................*

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

### _wp14150 - References .............................................................................................................

### I. Introduction
- EAC member countries are Burundi, Kenya, Rwanda, Tanzania, and Uganda.
- The EAC’s (unweighted) average growth rate in 2004–13 was 6.2 percent, placing it in the top one-fifth of the distribution of 10-year growth rate episodes experienced by all countries worldwide since 1960.
- Purpose: analyze disaggregated sources of past growth (composition of output and exports by sector, indicators of product quality and sophistication) to glean clues about the sustainability of recent strong growth.
- Note: Analysis relies on past growth and does not incorporate implications of possible exploitation of recent natural resource finds.

### II. Overall growth performance and reversion to the mean
- Corroboration of GDP growth using complementary indicators:
  - Electricity consumption (long series for Kenya and Tanzania) grew rapidly with acceleration during the past ten years (Figure 3).
  - Health outcomes improved: infant mortality rates and life expectancy improved considerably in most countries (Figure 4).
  - Fiscal revenues and private sector credit (in domestic currency at constant prices) rose rapidly; indices set to 100 in year 2000 (Figure 5).
- Ten-year growth rarity and predictive exercise:
  - Worldwide distribution of 10-year growth episodes (1950–2010) has mean 3.8 percent and standard deviation 2.6 percent.
  - EAC unweighted average 2004–13 growth = 6.2 percent — within the top one-fifth percent of the worldwide distribution of decade-long episodes since 1950.
  - Regression estimated on decade-long growth for a panel of 162 countries over 1960–2010 (with decade fixed effects): estimated coefficients and statistics:
    - ߚ = 0.23
    - ߤ = 2.8
    - ߪ = 2.54
  - Using the estimated model and input g_past = 6.2, the conditional simulation for next-decade growth is:
    - g_next = 4.2 + ߝ, where ߝ ~ N(0, 2.54^2)
  - Interpretation: conditional on historical persistence patterns, there is a little less than a one in four chance of another decade of growth above 6 percent for the EAC.

### III. Structural transformation, diversification, and rising sophistication
- Long-run sectoral trends (1970–2010) using UN national accounts (seven broad sectors):
  - Agriculture share declined from almost one half (unweighted average) in 1970 to about a third in 2010.
  - Decline in agriculture was steeper in the EAC than in Sub-Saharan Africa (SSA) or low-income countries (LIC) worldwide (Figure 7).
  - Largest gains in sector shares observed in transportation and construction; manufacturing gains limited; mining remains a relatively small sector.
- Sectoral composition and changes since 2000 (12-sector decomposition, 2000–11, IMF staff database):
  - Aggregate (EAC unweighted average) sector shares, 2000 vs. 2010:
    - Agriculture: 36% (2000) → 28% (2010)
    - Mining: 1% → 1%
    - Manufacturing: 9% → 9%
    - Utilities: 2% → 2%
    - Construction: 6% → 8%
    - Wholesale trade: 11% → 12%
    - Hotels and restaurants: 5% → 5%
    - Transport and communications: 6% → 9%
    - Financial services: 2% → 3%
    - Real estate & business services: 7% → 7%
    - Public administration: 6% → 7%
    - Other: 9% → 9%
  - Sectoral growth rates, 2000–10 (average annual):
    - Agriculture: 3.2 percent
    - Transport and communications: 10.8 percent
    - (All sectors experienced positive growth on average during the period.)
  - Sector share changes (2000–10) in percentage points (selected largest changes):
    - Agriculture: -7.1
    - Construction: +2.0
    - Transport and communications: +1.7
    - Wholesale trade: +1.5
    - Public administration: +1.2
- Exports: rising share of GDP and geographic diversification:
  - Exports as a share of GDP for the EAC rose from 12 percent in 1990 to 19 percent in 2010 (Figure 9).
  - Increase in export/GDP particularly pronounced in Tanzania and Uganda; to a lesser extent in Kenya and Rwanda.
  - Partner diversification indicators:
    - Theil index of partner concentration declined (higher diversification).
    - Share of top ten destination countries in total exports declined (Figure 10).
- Product diversification of exports:
  - Agriculture’s share in exports fell from 4/5 (80 percent) in the 1970s–90s to 2/3 (66.666... percent) in 2010, with mirroring gains in manufacturing (Figure 11).

### IV. Key implications summarized from the analyzed evidence
- Recent growth in the EAC (2004–13) was robust and corroborated by multiple production and development indicators (electricity consumption, health outcomes, fiscal revenues, private credit).
- The magnitude of the past-decade growth (6.2 percent) is historically uncommon; based on past decade-to-decade persistence patterns, continuation at that pace into the next decade is uncertain (estimated conditional central value 4.2 percent with innovation variance corresponding to standard deviation 2.54).
- Structural transformation is evident:
  - Substantial decline in agriculture’s GDP share (36% → 28% between 2000 and 2010).
  - Broad-based gains across construction, transport and communications, wholesale trade, and public administration.
  - Manufacturing share showed limited change.
- Trade integration and diversification trends:
  - Exports increased from 12 percent to 19 percent of GDP between 1990 and 2010.
  - Geographical diversification of export destinations increased (declining Theil index; declining share of top ten partners).
  - Export product mix shifted away from agriculture (from 4/5 to 2/3 of exports), with increasing manufacturing presence.

*Source: _wp14150 - References .............................................................................................................*

### Appendix Figure A2 provides individual country information)

### _wp14150 - Appendix Figure A2 provides individual country information)

### Export product diversification: long-run trends and measures
- Exported products have become more diversified in all EAC countries since the early 1960s, and at a more rapid pace than the Sub-Saharan average, with a steeper trend since the early 1990s.
- The Theil index shows that by mid-2000s export products were at least as diversified in each EAC country as in the average country in Sub-Saharan Africa, with Kenya and Tanzania substantially ahead of their peers.
- Alternative measures (e.g., the share of the top ten products in total exports of goods) confirm the diversification results.
- Figures and measures used:
  - Theil index (product-level and partner-level) computed using UN COMTRADE SITC4 data for 1962–2010.
  - Share of Top 10 Products (time series shown alongside Theil index).

### Structure of exports (selected aggregate shares shown in Figure 11)
- EAC aggregate snapshots (percent of total exports of goods):
  - 1970: Agriculture 80%, Manufacturing 6%, Mining 14%
  - 1990: Agriculture 85%, Manufacturing 12%, Mining 3%
  - 2009: Agriculture 66%, Manufacturing 23%, Mining 11%
- SSA aggregate snapshots (percent of total exports of goods):
  - 1970: Agriculture 65%, Manufacturing 15%, Mining 20%
  - 1990: Agriculture 51%, Manufacturing 27%, Mining 22%
  - 2009: Agriculture 44%, Manufacturing 29%, Mining 27%

### Number of distinct export products (Figure 13)
- The sheer number of export products increased substantially, especially since the early 1990s and particularly in manufacturing.
- Example: Uganda’s distinct products rose from about one hundred in 1980 to more than five hundred in 2010, with more than three hundred products in manufacturing.
- Robustness checks: counts repeated for products with export revenues exceeding US$1 million and US$10 million; results confirm sizable increases, particularly in Kenya, Tanzania, and Uganda.
- Time span of product counts: 1962–2010 (UN COMTRADE, authors’ calculations).

### Sophistication of export products (Figure 14)
- EAC countries have generally moved toward exporting more “sophisticated” products, defined as those predominantly produced by advanced economies.
- Sophistication index constructed at 4-digit SITC level using a two-step method:
  - Step 1: For each product, a “productivity” index is the weighted average of per capita GDP of exporting countries, weights are the share of the product in each country’s total exports.
  - Step 2: A country’s sophistication index is the weighted average of its export products’ productivity indices, with weights equal to value shares of products in the country’s total exports.
- Empirical findings:
  - The sophistication index of each EAC country rose visibly between 1990 and 2010.
  - Gains for most EAC countries were noticeably greater than the average for sub-Saharan Africa and other regions.
  - Kenya experienced the smallest gain (it started with the highest sophistication index in 1990).
  - By 2010, Kenya and Uganda were slightly above the sub-Saharan average; Tanzania was close to it.
- Specific plotted index ranges shown in Figure 14 include values between 6.0 and 9.5 for regions/countries in 1990 and 2010.

### Quality of individual exported items (Figure 15)
- Quality is measured as the ratio of the value of exported items by the country to the average value of such item exported by all other countries (unit-value-based quality adjusted by Henn et al. 2013).
- Summary measures indicate a trend toward improved quality within each export category for EAC countries.
- Example: In 2009, Burundi’s coffee quality was 0.8 (on a scale from 0 to 1) and coffee represented 80 percent of Burundi’s goods exports in the database.
- Overall pattern:
  - No clear rising trend in quality across all items—consistent with many products being primary commodities or goods involving limited processing.
  - Quality is measured relative to trading partners, suggesting EAC countries broadly keep pace with competitors.
  - Shares of the top ten products in total exports reveal greater diversification in 2009 than in 1980.

### Sectoral output and export composition changes (selected country-level snapshots)
- Sectoral output database:
  - Value added series in constant prices for 12 economy-wide sectors for Sub-Saharan Africa, period 2000–2011.
  - 12 sectors: agriculture (including forestry and fishing); mining and quarrying; manufacturing; utilities (electricity and water); construction; trade and repairs; hotels and restaurants; transportation and communication; financial intermediation; real estate and business activities; public administration; and other. FISIM and net taxes are excluded.
- Selected country export composition changes (Figure A2; shares in percent):
  - Burundi: 1970 — Agriculture 95%, Manufacture 2%, Mining 3%; 2010 — Agriculture 88%, Manufacture 11%, Mining 1%
  - Kenya: 1970 — Agriculture 80%, Manufacture 17%, Mining 3%; 2010 — Agriculture 63%, Manufacture 34%, Mining 3%
  - Rwanda: 1970 — Agriculture 39%, Manufacture 1%, Mining 60%; 2010 — Agriculture 41%, Manufacture 22%, Mining 37%
  - Tanzania: 1970 — Agriculture 90%, Manufacture 7%, Mining 3%; 2010 — Agriculture 65%, Manufacture 24%, Mining 11%
  - Uganda: 1970 — Agriculture 99%, Manufacture 1%, Mining 0%; 2010 — Agriculture 73%, Manufacture 27%, Mining 0%
- Additional snapshot (1990 shares shown in figure for comparison across countries).

### Interpretation, implications, and outlook (paper’s summary and conclusions)
- Recent decade: accelerated overall economic growth and structural transformation in EAC member countries.
- Observed structural shifts:
  - The share of agriculture has fallen substantially.
  - Gains broadly distributed: largest gains to construction, transportation, and wholesale trade; manufacturing and mining posted modest gains.
  - Rapid growth has not been driven by a narrow range of products (e.g., natural resources).
- Improvements over time:
  - Export sophistication and quality of items exported have improved, more noticeably during the past decade.
- Cautionary notes and outlook:
  - The pattern suggests consumer and investment demand for more sophisticated goods and services are beginning to be met as per capita incomes rise.
  - No clear production-side “winners” yet that embed a durable comparative advantage beyond the region.
  - No major quality improvements vis-à-vis competitor countries where progress is also occurring.
  - These observations counsel caution about projecting continued rapid growth far into the future.
  - The paper’s findings are consistent with expected continued healthy growth but with an eventual slowdown to a more moderate pace, consistent with the statistical pattern for “reversion to the mean” reported elsewhere in the paper.

### Methodology highlights (data and indices)
- Export data source: UN COMTRADE, SITC1 classification at the four-digit level, period 1962–2010.
- Theil index of export diversification:
  - For product-level Theil index, x_k is export value of product k, n is number of products, μ is the products’ average dollar value.
  - For partner-level Theil index, x_k is export value to partner k, n is number of partners for each exporter.
- Sophistication index:
  - Product productivity P_k computed as weighted average of per capita GDP of exporters, weights = share of product k in each country’s total exports (formula provided in text).
  - Country sophistication is the value-share-weighted average of product productivity across the country’s export products (formula provided in text).
- Export quality measurement (Henn et al. 2013):
  - Quality derived from unit values (value/quantity), adjusted for production costs and selection bias from distance; coefficients estimated following Hallak and Schott (2011).
  - Quality estimates normalized by their 90th percentile in the relevant product-year combination and aggregated using current trade values as weights; normalization repeated at each aggregation level.

*Source: UN COMTRADE, Penn World Table, Henn et al. (2013), and authors’ calculations as presented in the Appendix Figures and text of the provided IMF working paper content.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14150.pdf_