## _wp15237

## Source details

**Canonical URL:** [_wp15237](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15237.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15237.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15237.pdf.json)

---

### Introduction — scope and research question
- Investigates whether countries experiencing natural resource booms in the 2000s have overcome slow growth (the "curse of natural resources").
- Argues traditional tests using total GDP per person can be misleading during booms because total GDP is dominated by a temporary surge in the resource sector.
- Proposes a two-part alternative approach:
  - Use non-resource GDP growth (value-added outside hydrocarbon and mineral sectors) as the metric for testing the curse.
  - Use each country’s pre-boom growth as the counterfactual comparison period rather than a broad cross-section of other countries.
- Rationale: examine sustainable growth trends in the non-resource economy and provide a country-specific test that complements cross-section approaches.

### Key empirical summary (Introduction)
- Using growth in per-capita GDP in the non-resource economy before the boom as the reference period:
  - 11 countries had lower growth during the boom years.
  - 7 countries had higher growth during the boom years.
  - In one case growth was statistically significantly lower during the boom.
  - There were no cases where growth was statistically significantly higher during the boom.
- Conclusion summarized: Growth in value-added per-capita or labor productivity outside hydrocarbons and minerals remains sluggish in resource-intensive economies on average; higher public investment has not, on average, translated into higher payoffs for non-resource growth.

### Model framing and mechanisms (definition of the curse)
- Curse defined as a causal effect from resource rents R that reduces performance in the non-resource economy.
- Competing forces in the model:
  - Pro-curse: Dutch disease — resource wealth depresses other traded activity (α falls).
  - Anti-curse: public capital investment (H) including education/infrastructure that raises non-resource productivity.
- Public capital accumulation equation preserved as in source:
  - H_τ = H_{τ−1} [1 + α_{τ−1} + γ((1−a) R_{τ−1}) − β]    (equation as presented)
- Steady-state growth rate of H:
  - α + γ(Ψ(1−a) R)    (11)
- Condition for a natural resource curse (steady-state):
  - (∂α / ∂R) + γ'(·) Ψ (1−a) < 0    (12)
- Policy levers to offset negative channel:
  - invest a high fraction of resource rents (high 1−a)
  - keep corruption low (high Ψ)
  - choose efficient investments (high γ'(·))

### Model structure — supply, prices, investment, demand, equilibrium
- Production structure:
  - Three sectors: traded manufacturing (μ), non-traded (ν), and resource sector producing value-added equal to R without employing domestic resources.
  - Labor-augmenting technical change drives perpetual growth; human capital stock H multiplies employment in non-resource sectors.
  - Normalize labor force to 1; α is share of labor in traded sector.
  - Intensive-form sector functions: x_μ = μ(k_μ), x_ν = ν(k_ν).
- Price and zero-profit conditions:
  - π_ν = β_ν_ω(ω, ρ) ω + β_ν_ρ(ω, ρ) ρ    (2)
  - 1 = β_μ_ω(ω, ρ) ω + β_μ_ρ(ω, ρ) ρ    (3)
- Human capital growth (restated):
  - H_t = H_{t−1} [1 + α_{t−1} + γ(Ψ(1−a) R_{t−1}) − β]    (4)
- Demand (per-effective-worker) definitions (Φ = 1 / ((1 + β)(1 + θ))):
  - c_μ_t = Φ(ω_t + a R_t)    (5)
  - c_ν_t = (1/π_ν_t) β Φ(ω_t + a R_t)    (6)
  - c_μ_{t+1} = θ (1 + ρ) Φ(ω_t + a R_t)    (7)
  - c_ν_{t+1} = (1/π_ν_{t+1}) θ (1 + ρ) β Φ(ω_t + a R_t)    (8)
- Equilibrium condition in non-traded sector (difference equation):
  - c_ν^y_t + c_ν^o_t / [1 + α_{t−1} + γ(Ψ(1−a) R_{t−1}) − β] = (1−α_t) ν(·)    (10)
  - Difference equation stable; higher R associated with a lower steady-state α via demand shifts.
- GDP bookkeeping:
  - GDP/H = α μ(·) + π_ν (1−α) ν(·) + R/H
  - Total GDP = H [ α μ(·) + π_ν (1−α) ν(·) ] + R
  - In steady state non-resource GDP per-capita grows at same rate as H (Ĥ); total GDP per-capita grows at Ĥ (1 − σ_R).

### Operational testing definitions for booms
- Operational definition: a curse exists when there is a negative association between the size of the resource boom and growth of GDP per-capita in the non-resource economy during the boom.
- Empirical regression for boom test:
  - ln(Ψ)_i_t − ln(Ψ)_i_{t−1} = a_1_i + a_2_i δ^1_i + ε_i    (13)
  - Ψ = constant-price GDP per person excluding hydrocarbons and minerals; δ^1_i = 1 during the boom, 0 in counterfactual.
- For post-Soviet transitions, control for common U-shaped GDP path using year dummies or quadratic trend before testing country-specific deviations.

### Data sources, sample selection, and dating booms
- Exports of hydrocarbons and major metals/minerals from COMTRADE (revision 1), using mirror imports; merged with IMF national accounts (WEO).
- Value-added from natural resources: national accounts where possible; otherwise extrapolated using export growth rates or estimated as 95 percent of export value when necessary.
- Initial candidate criteria:
  - natural resource exports > 5 percent of GDP for at least one year;
  - population > 1 million in 1990.
- Retained countries required forward 5-year moving average > backward 5-year moving average by 6 percentage points of GDP, around or after year 2000, for at least two years.
- Initial selection 32 countries; 6 dropped for inadequate or implausible data: DR Congo, R Congo, Guinea, Papua New Guinea, Sudan, Togo.
- Counterfactual and boom periods chosen to avoid confounding events (wars, regime shifts, structural breaks); examples: Chad counterfactual begins 1990, Algeria counterfactual begins 1986, Mongolia counterfactual begins 1993, Mauritania counterfactual begins 1986.

### Sample retained and period dates (selected list as in source)
- Countries and Counterfactual | Boom periods (as provided):
  - Algeria: Counterfactual 1986-1997; Boom 1998-2011
  - Bolivia: Counterfactual 1987-1998; Boom 1999-2011
  - Chad: Counterfactual 1990-2002; Boom 2003-2011
  - Equatorial Guinea: Counterfactual 1980-1991; Boom 1992-2011
  - Mongolia: Counterfactual 1992-2005; Boom 2006-2011
  - Qatar: Counterfactual 1987-1997; Boom 1998-2011
  - Saudi Arabia: Counterfactual 1986-2001; Boom 2002-2011
  - United Arab Emirates: Counterfactual 1995-2001; Boom 2002-2011
  - Zambia: Counterfactual 1997-2002; Boom 2003-2011
  - (Full list of 18 countries provided in source.)

### Main empirical findings (country regressions and aggregates)
- Country-by-country regressions (ln(Y_it) − ln(Y_it−1) = θ1_i + θ2_i d1_i + O_i) reported for 18 countries.
  - Across 18 cases:
    - 11 showed lower non-resource growth during boom than before.
    - 7 showed higher non-resource growth during boom than before.
    - Only 1 case (Bolivia) had a statistically significant decline in non-resource growth during the boom (Bolivia: Growth Before the Boom (a1) = 2.384***; Change in Growth During Boom (a2) = -2.496***).
    - No case with statistically significantly higher growth during the boom relative to pre-boom.
- Aggregated synchronization (indexed to 100 at t=0):
  - Median total GDP rises strongly during booms.
  - Median non-resource GDP is essentially flat over the boom period.
  - No tendency for acceleration in non-resource GDP growth during later years of booms.
- Country-level aggregated result:
  - Only 5 of 18 countries show non-resource growth over 2 percent per year during booms (sum of coefficients in Table 9).
  - Examples:
    - Zambia: non-resource real per-capita growth 1.9 percent per year during boom; 2.1 percent during counterfactual.
    - Saudi Arabia: total GDP per-capita grew since boom start in 2002, but non-oil real GDP per-capita growth not rapid during boom and not noticeably different than counterfactual.

### Comparison with 1970s booms and lagged-impact hypothesis
- Five-country 1970s series (Saudi Arabia, Algeria, Bolivia, Libya, Trinidad and Tobago) showed non-oil GDP surged at start of 1970s oil boom but both total and non-oil GDP per person reverted to pre-boom levels eventually; took 17 years to play out in 1970s boom.
- Nine 1970s-boom countries: setting 1986 non-oil GDP per person = 100, median GDP per-person = 123 by 2000 (implying average annual growth rate 1.5 percent); mean growth = 0.58 percent over same interval.
- Conclusion: current booms show little evidence of outperforming 1970s counterparts; historical evidence undermines expectation of a long gestation payoff.
- Lagged-impact case studies:
  - Equatorial Guinea: boom started 1994, growth accelerated around 2005 (9 years after start).
  - Papua New Guinea: sharp recovery in 2006, long after its boom began.
  - Overall: aggregate evidence shows no general tendency for non-resource GDP growth to accelerate late in boom periods.

### Saving, investment, and public investment findings (period averages)
- Natural resource production (value-added share of GDP), mean values:
  - Mean counterfactual 0.154 | Mean boom 0.301 | Mean difference 0.148
- Saving rates (fraction of GDP), means:
  - Mean counterfactual 0.16 | Mean boom 0.27 | Mean change 0.11
  - Mean excluding Chad: 0.17 | 0.32 | 0.15
- Current account (fraction of GDP), means:
  - Mean counterfactual -0.06 | Mean boom -0.01 | Mean change 0.06
  - Mean excluding Chad: -0.06 | 0.04 | 0.10
- Investment rates (fraction of GDP), means:
  - Mean counterfactual 0.22 | Mean boom 0.27 | Mean change 0.05
  - Mean excluding Chad: 0.23 | 0.27 | 0.05
- Public investment rates (period means):
  - Mean counterfactual 0.09 | Mean boom 0.10 | Mean change 0.01
  - Mean excluding Equatorial Guinea: 0.07 | 0.09 | 0.02
- Interpretation:
  - Saving rates rose substantially during booms; current accounts shifted toward surplus.
  - Domestic investment rates rose modestly or stayed similar; public investment share rose slightly on average.
  - Evidence consistent with substantial saving held offshore (sovereign wealth funds) and with low returns to domestic investment in the non-resource economy rather than lack of saving or investment effort.

### Labor productivity and population dynamics — UAE and Qatar case details
- United Arab Emirates (2002–2011, GDP at constant 2007 Prices, Million Dirhams)
  - 2002: Oil GDP 265728; Non Oil 402889; Total 668618
  - 2011: Oil GDP 315410; Non Oil 667316; Total 982725
  - Ave. Annual Growth: Oil 1.9% ; Non Oil 5.8% ; Total 4.4%
  - Workers growth: Oil 7.8% ; Non Oil 9.5% ; Total 9.5%
  - Labor Productivity growth: Oil -5.9% ; Non Oil -3.7% ; Total -5.1%
- Qatar (2006/7–2012, GDP at constant 2004 Prices, Million Qatari Riyal)
  - 2006/7: Oil GDP 77662; Non Oil 93089; Total 170751
  - 2012: Oil GDP 147726; Non Oil 192919; Total 340646
  - Ave. Annual Growth: Oil 12.4% ; Non Oil 14.2% ; Total 13.4%
  - Economically Active Workers growth: Oil 16.7% ; Non Oil 12.9% ; Total 13.2%
  - Labor Productivity growth: Oil -3.7% ; Non Oil 1.1% ; Total 0.2%
- Implication: rapid in-migration and employment growth can dilute GDP-per-person and value-added-per-worker gains even when non-resource GDP expands rapidly.

### Post-Soviet countries — U-shaped transition and boom impacts
- U-shaped profile for total GDP in transition economies confirmed empirically using year-dummy and quadratic specifications:
  - Quadratic results: year coefficient -0.187; year^2 coefficient 0.0038; N = 193; Adj R-squared = 0.57.
- Resource-intensive post-Soviet countries examined: Azerbaijan, Kazakhstan, Russia, Turkmenistan (and Tajikistan referenced in interactions).
- Estimation allowing country-specific quadratic terms finds the five resource-intensive countries experienced slower non-resource GDP growth relative to resource-poor comparators; growth statistically significantly slower for all except Azerbaijan.
- Conclusion: little evidence that resource booms in post-Soviet economies served to raise non-resource GDP above what it would have been without the booms.

### Aggregate conclusion and caveats
- Dominant finding: across 18 booming economies, non-resource GDP per-capita growth did not, on average, increase during booms relative to pre-boom periods; 11 countries showed lower growth, 7 higher, only 1 statistically significant decline, zero statistically significant increases.
- Evidence suggests that despite higher saving and investment rates (and some higher public investment), payoffs to non-resource growth have been limited — consistent with low domestic returns to investment or Dutch disease effects offsetting investment benefits.
- Caveats discussed in source:
  - Potential selection bias and counterfactual dating considered case-by-case; no systematic bias identified on measurable items.
  - Possibility of lagged payoffs acknowledged but historical 1970s evidence and post-2000 patterns provide limited support for a general long-gestation reversal.
  - Welfare effects (social spending, transfers, sovereign wealth accumulation) are distinct from GDP-per-capita growth; these are relevant but do not negate the empirical focus on non-resource per-capita growth for diversification policy goals.

*Source: _wp15237 — IMF working paper, Introduction and Sections 1.1–2.2 (content excerpt provided).*

### 1.  Introduction .......................................................................................................

### _wp15237 - 1.  Introduction

### Scope and research question
- Investigates whether countries that have experienced natural resource booms in recent years have managed to overcome slow growth, sometimes known as the curse of natural resources.
- Motivated by the large rise in global prices for metals and minerals over the past decade and new discoveries of minerals and hydrocarbons that spawned booming conditions in resource-rich developing countries.
- Argues that traditional tests (comparing total GDP per person across a large cross-section) can be misleading during booms because total GDP is likely dominated by a temporary surge in the resource sector.
- Proposes a two-part alternative approach:
  - Use non-resource GDP growth (value-added outside hydrocarbon and mineral sectors) as the relevant metric for testing the curse.
  - Use each country’s pre-boom growth as the counterfactual comparison period rather than a broad cross-section of other countries.
- Rationale: this lifts the veil of the booming sector to examine more sustainable growth trends and offers a simple, country-specific test that complements cross-section approaches.

### Key empirical summary (from the Introduction)
- Using growth in per-capita GDP in the non-resource economy before the boom as the reference period:
  - 11 countries had lower growth during the boom years.
  - 7 countries had higher growth during the boom years.
  - In one case growth was statistically significantly lower during the boom.
  - There were no cases where growth was statistically significantly higher during the boom.
- Conclusion (summarized): Growth in value-added per-capita or labor productivity outside the hydrocarbon and mineral sectors remains sluggish in resource-intensive economies on average. Higher public investment has not, on average, translated into higher payoffs for non-resource growth.

### Literature context and past evidence
- Traditional literature typically examines total GDP growth per-capita to test the curse hypothesis (Van de Ploeg (2011) review).
- Example cited: The Economist (June 30th 2012) used total GDP surge (Angola 2004–2008: "its GDP surged by an average of 17% a year") as evidence of success—illustrating the misleading nature of total-GDP-based assessments.
- Past studies (Gelb and Associates (1988)) of 1970s booms (Algeria, Ecuador, Indonesia, Iran, Nigeria, Trinidad and Tobago, Venezuela) found GDP growth lower than expected given the size of booms and domestic investment; subsequent slumps brought GDP per person back to pre-boom levels.
- Empirical associations:
  - Harding and Venables (2010): non-resource exports decreased by 35-70 percent in response to resource windfalls (data for 137 countries, 1975-2007).
  - Kareem Ismail (2010): oil windfalls were associated with a decline of 3.4 percent in value-added in manufacturing.
  - Brahmbhatt, Canuto and Vostroknutova (2010): countries with large resource sectors tended to have smaller traded sectors.
  - Using COMTRADE data (1970-2008) for 98 countries and measuring non-resource exports by subtracting hydrocarbon and mineral exports from total exports, a regression of average annual growth in GDP per capita on non-resource export growth yields:
    - R2 = 50 percent
    - estimated coefficient = 0.34
    - t-ratio = 9.99
  - Note: the above regression is not claimed to be causal but suggests an important connection between non-resource export growth and overall growth.

### Strategy and contribution
- Defines the curse as something that happens (or not) in the non-resource economy; conducts tests using empirical proxies for non-resource GDP.
- Uses country-specific pre-boom periods as counterfactuals to test whether non-resource growth during booms is abnormally low or high relative to the country’s own history.
- Acknowledges arbitrariness in choosing pre-boom comparison periods but argues plausible comparison periods exist and that results are sufficiently consistent across cases to draw conclusions.
- Paper structure: model (next section) that defines the issues and testing approaches; implementation details; application to economies with significant booms in the 2000s; results and policy discussion.

### 1.1 Model — framing and mechanisms
- Purpose: provide a specific definition and associated test for the curse. The framework posits competing forces: pro-curse and anti-curse. A curse exists when pro-curse forces dominate.
- Pro-curse mechanism chosen in the model: Dutch disease (resource wealth depresses other traded activity).
  - Supporting evidence cited in the literature (see above Harding and Venables; Ismail; Brahmbhatt et al.).
  - Non-resource export growth is empirically associated with overall economic growth, so suppression of non-resource exports is important for growth outcomes.
- Anti-curse mechanism chosen: public capital investment (including education, infrastructure, human capital).
  - Gelb and Associates (1988): two-thirds of investment in six resource-rich economies was directed towards infrastructure or human capital (p. 137).
  - Gylfason, Herbertsson, and Zoega (1999) and Bravo-Ortega and De Gregorio (2005) emphasize education as a policy to overcome the curse.
  - Berg, Portillo, Yang and Zanna (2012) and Gelb (2010) reviewed policy options; conditional recommendation is greater investment if returns to public capital are high; Gelb emphasizes higher public investment with caution and fiscal restraint.
- Model features:
  - Public capital (denoted ) aggregates infrastructure and human capital into a single stock.
  - Accumulation depends on employment in the traded sector (learning-by-doing) and investment out of resource revenues.
  - Accumulation equation (as presented):
    - 
      
      =
      −1
      [1 +
      −1
      +((1−)
      −1
      )−] (1)
  - Interpretation of components:
    - : share of labor in the traded sector (learning-by-doing effect).
    - (.): efficiency/function of government investment out of resource rents.
    - : fraction of resource rents distributed as dividends (policy choice).
    - : natural resource rents (size of windfall).
    - : depreciation.
  - Government financing and behavior:
    - Government cannot borrow without limit on world capital markets to finance public investments—rationale: information asymmetries and lack of immediate revenue streams from public goods.
    - Public investment is modeled as a function of natural resource rents.
    - Government choice of  determines the share of resource revenues distributed as dividends versus retained for public investment; if government maximized long-run growth it would choose  = 0 (invest all rents), but in practice  may be fixed or determined by political economy.
- Model type: overlapping-generations growth model (generations live two periods: working and receiving a wage; retiring in second period).
- Model organization in the paper: supply side, demand side, equilibrium and dynamic solution, propositions about effects of resource booms on growth, and definitions of a curse.

### Conceptual implications and testing
- The model frames the curse as a balance between Dutch disease-like forces that reduce traded-sector activity and anti-curse forces (public capital accumulation, education) that can raise non-resource productivity.
- Testing for a curse therefore centers on observed behavior of non-resource GDP (and related indicators such as non-resource exports, saving and investment, and labor productivity) during boom periods relative to pre-boom benchmarks.
- Paper outlines implementation and empirical tests (described in later sections) that operationalize these model-based definitions.

*Source: _wp15237 - 1.  Introduction (IMF PDF content supplied).*

### 1.2  Supply side

### 1.2  Supply side

### Production structure and sectors
- Three sectors: traded manufacturing (superscript ’μ’), non-traded (superscript ’ν’), and a natural resource sector producing value-added equal to ’R’ without employing domestic resources.
- Resource output sold on world markets at an exogenous price π_r; units of R chosen so the price term does not appear explicitly.
- No distinction between resource discoveries and rises in commodity prices.

### Production functions
- Traded and non-traded sectors employ labor and capital:
  - X_μ = M(L_μ, K_μ)
  - X_ν = N(L_ν, K_ν)
- Labor-augmenting technical change is the source of perpetual growth.
- Human capital variable H represents the stock of economically useful knowledge.
- Accumulation of knowledge is a by-product of employment in the traded manufacturing sector; H multiplies employment in both non-resource sectors.
- Normalize total labor force to 1; let α represent the share of labor in the traded sector. Then:
  - X_μ = M(α H, K_μ)
  - X_ν = N((1−α) H, K_ν)
- Intensive-form (homogeneous degree one):
  - x_μ = μ(k_μ)
  - x_ν = ν(k_ν)
  - where k_μ = K_μ / (α H), k_ν = K_ν / ((1−α) H)

### Capital market equilibrium
- Capital employed in each sector until value marginal product of capital per effective worker equals the world real interest rate; no adjustment costs.
- Conditions:
  - p_ν ν'_μ( K_ν / ((1−α) H) ) = ρ
  - μ'_μ( K_μ / (α H) ) = ρ

---

### 1.3  Prices
- Price of manufactures is the numeraire set equal to 1. π_ν is the ratio of the price of the non-traded good to manufactures.
- Competition and free entry ⇒ zero profits.
- Zero profit conditions with β_j_i(ω, ρ) denoting unit cost functions for factor i in production of good j:
  - π_ν = β_ν_ω(ω, ρ) ω + β_ν_ρ(ω, ρ) ρ    (2)
  - 1 = β_μ_ω(ω, ρ) ω + β_μ_ρ(ω, ρ) ρ    (3)
- For given world real interest rate ρ (and world price of traded good set to 1), these solve for wage ω and π_ν as functions of ρ.

---

### 1.4  Investment and Growth
- Two major sources of growth:
  1. Labor-augmenting technical change generated by structure of employment (dependence on α), which can render the economy vulnerable to a resource curse.
  2. Government investment of part of resource revenue to augment human capital.
- Given resource revenue R, fraction α (denoted here as a) is distributed to the population; rest 1−a is invested by government in human capital.
- Parameter Ψ allows analysis of government investment efficiency; parameter γ captures the productivity of that investment.
- Human capital growth equation:
  - H_t = H_{t−1} [1 + α_{t−1} + γ(Ψ(1−a) R_{t−1}) − β]    (4)
  - (Parameters and variables preserved as in source.)

---

### 1.5  Demand side
- Consumers solve inter-temporal consumption problem; each generation works when young and receives wage. Government transfers fraction a of resource revenue to the public: a R.
- Consumers save for retirement at world rate of interest.
- Lifetime utility maximization:
  - Max U = [ln(c_μ_t) + β ln(c_ν_t)] + θ [ln(c_μ_{t+1}) + β ln(c_ν_{t+1})]
  - Budget constraint (young):
    - c_μ_t + π_ν_t c_ν_t + 1/(1+ρ) [ c_μ_{t+1} + π_ν_{t+1} c_ν_{t+1} ] = ω_t + a R_t
- Define Φ = 1 / ((1 + β)(1 + θ))
- Four demand functions (expressed per-effective-worker, dividing by H_t):
  - c_μ_t = C_μ_t / H_t = Φ(ω_t + a R_t)    (5)
  - c_ν_t = C_ν_t / H_t = (1/π_ν_t) β Φ(ω_t + a R_t)    (6)
  - c_μ_{t+1} = C_μ_{t+1} / H_t = θ (1 + ρ) Φ(ω_t + a R_t)    (7)
  - c_ν_{t+1} = C_ν_{t+1} / H_t = (1/π_ν_{t+1}) θ (1 + ρ) β Φ(ω_t + a R_t)    (8)
- Total demand for non-traded good in period t:
  - D_ν_t = c_ν^y_t H_t + c_ν^o_t H_{t−1}

---

### 1.6  Equilibrium
- Aggregate demand = supply in non-traded sector:
  - c_ν^y_t H_t + c_ν^o_t H_{t−1} = N[(1−α_t) H_t, K_ν_t]
- Dividing by effective workers in period t:
  - c_ν^y_t + c_ν^o_t (H_{t−1} / H_t) = (1−α_t) ν(·)    (9)
- Substitute H_t / H_{t−1} from (4) to form a difference equation in α:
  - c_ν^y_t + c_ν^o_t / [1 + α_{t−1} + γ(Ψ(1−a) R_{t−1}) − β] = (1−α_t) ν(·)    (10)
- Notes:
  - c_ν^y_t and c_ν^o_t are functions of a and R, so a and R affect α via both numerator and denominator of the left-hand side.
  - The difference equation has slope positive but less than one in α_t, α_{t−1} space; hence the difference equation is stable.
  - A rise in natural resource dividend raises demand for non-traded products and shifts the equation down so that higher R is associated with a lower steady-state α.

---

### 1.7  Housekeeping - GDP
- GDP per-effective-worker measured in prices of manufactures:
  - GDP/H = α μ(·) + π_ν (1−α) ν(·) + R/H
- Total GDP:
  - GDP = H [ α μ(·) + π_ν (1−α) ν(·) ] + R
- Since population normalized to 1, above gives GDP per-capita.
- In steady state:
  - Non-resource GDP per-capita grows at same rate as H (Ĥ).
  - Total GDP per-capita grows at rate Ĥ (1 − σ_R), where σ_R is the share of resource GDP in total GDP.
  - Steady state has constant α and constant growth rate of H.

---

### 1.8  Testing for the Curse - steady state growth
- Two channels through which R affects steady-state growth rate of H:
  1. Via α (α = function of R): higher R can reduce α and thus reduce human capital growth.
  2. Via direct investment of R: higher R can raise H growth through government investment.
- Steady-state growth rate of H:
  - α + γ(Ψ(1−a) R)    (11)
- Condition for a natural resource curse (long-term per-capita growth inversely associated with level of R):
  - (∂α / ∂R) + γ'(·) Ψ (1−a) < 0    (12)
  - Interpretation: negative effect of ∂α/∂R must dominate positive effect from invested rents.
- Policy levers to counteract the negative channel:
  - (a) invest a high fraction of resource rents (high 1−a)
  - (b) keep corruption low (high Ψ)
  - (c) choose efficient investments (high γ'(·))
- Curse defined as causal effect from R to the non-resource economy (not including resource sector itself). A curse exists when R causally reduces performance in the rest of the economy.

---

### 1.9  Testing for the Curse - resource booms
- Two illustrative cases:
  - Figure 2 (curse): boom immediately raises total GDP by the boom amount, but growth rate of non-resource GDP slows during the boom; after boom ends total GDP returns to the non-resource path and may be below the counterfactual.
  - Figure 3 (no curse / overcoming): non-resource GDP grows faster during the boom so that after the boom total GDP is higher than the no-boom counterfactual.
- Operational definition for booms:
  - A curse exists when there is a negative association between the size of the natural resource boom and growth of GDP per-capita in the non-resource economy during the boom period.
  - Equivalent: a curse exists when level of non-resource GDP after the boom is lower than it would have been had the boom never occurred.
- Empirical testing approach:
  - Regress mean growth rates of non-resource real GDP per-capita on a dummy δ^1_i that is 1 during the boom (0 during counterfactual):
    - ln(Ψ)_i_t − ln(Ψ)_i_{t−1} = a_1_i + a_2_i δ^1_i + ε_i    (13)
  - Significance of a_2_i tests whether growth was significantly higher during the boom.
  - Ψ is constant price GDP per person in the rest of the economy (value added in all sectors excluding hydrocarbons and minerals).
- Special treatment for post-Soviet and Eastern European countries:
  - These economies often follow a U-shaped GDP path due to structural transformations; tests control for the common U-shape and then ask whether, over and above this, countries with booms experienced higher growth.

---

### 2.0.1  Data
- Exports of hydrocarbons and major metals/minerals from COMTRADE (revision 1), using mirror imports from trading partners.
- COMTRADE merged with IMF national accounts aggregates from World Economic Outlook database.
- COMTRADE coverage up to 1962-2011 (varies by country); data limitations constrain time periods studied.
- Value-added from natural resources taken from national accounts where possible; when missing, growth rates in export data were used to extrapolate national accounts backward.
- If no national accounts data available, value-added estimated as 95 percent of export value, based on separate analysis showing value-added ≈ 95 percent of export sales in years with both data types.

### 2.0.2  Selection method
- Initial candidate criteria:
  - Natural resource exports as share of GDP exceeded 5 percent for at least one year in the sample.
  - Population greater than 1 million in 1990.
- Construct backward and forward 5-year moving average time series on natural resource production as share of GDP for each country.
- Retain country if forward average exceeded backward average by 6 percentage points of GDP, around or after year 2000, for at least two years.
- Corroboration checks: value-added in natural resources / GDP from national accounts; value-added in natural resources / non-natural-resource GDP; value-added in natural resources per head.
- Threshold robustness checked between 5 percent and 7 percent of GDP.
- Initial selection yielded 32 countries; six dropped for inadequate or implausible data: DR Congo, R Congo, Guinea, Papua New Guinea, Sudan, Togo.

### 2.0.3  Determining boom and counterfactual dates
- For each country, time divided into boom, counterfactual, and possibly a further excluded period (wars, disqualifying events).
- Boom defined as sustained rise in natural resource export revenues.
  - Initial year: first year revenues start rising sustainably.
  - Final year: year revenues fall back to pre-boom level.
  - Some booms continue through data end (2011).
- Counterfactual periods adjusted to avoid major disqualifying events (wars, regime shifts, sharp collapses).
  - Examples:
    - Chad: counterfactual begins with Idriss Deby regime in 1990 to avoid earlier civil war and unrest.
    - Angola: no suitable counterfactual due to pre-1990 "hard control" period vs post-1990 liberalization.
    - Algeria: counterfactual begins in 1989 to avoid post-1986 oil price collapse effects.
    - Laos: counterfactual begins in 1990 after cessation of Soviet aid in 1989.
    - Mauritania: counterfactual begins in 1986 to avoid Sahel droughts and protracted war in 1970s.

*Italicized attribution: Source: _wp15237 - 1.2  Supply side*

### annex part of Western Sahara. Following Ndulu et. al. (2008), 1986 was

### _wp15237 - annex part of Western Sahara. Following Ndulu et. al. (2008), 1986 was

### Data selection and counterfactual periods
- Counterfactual start years chosen to avoid confounding events:
  - Mauritania: 1986 (first year considered free of anti-growth syndromes).
  - Mongolia: 1993 (to avoid socialist planning and disruptive transition during 1991-1992).
  - Mozambique: 1993 (after 1990 constitution established a market-based economy and civil war ended in 1992).
  - Oman: 1976 (after the defeat of the Dhofar Rebellion).
- Sudan excluded entirely due to state-breakdown assessment in Ndulu et al. (2008).
- Graphical convention described:
  - Boom period = shaded area to the right.
  - Counterfactual period = white area immediately to the left of the boom.
  - Earlier period = shaded area to the left of the counterfactual.

### Regression results and aggregate findings
- Table 9 reports estimates of equation 15 for 18 countries.
- Key empirical findings:
  - 11 of the 18 countries show lower growth during the boom period than before.
  - 1 country shows a statistically significant decline (Bolivia).
  - 7 countries show higher growth during the boom, but none of these are statistically significant.
  - Conclusion: little compelling evidence to reject the null of no change during the boom period.
- On average across booms:
  - Natural resource production rose from 15 to 30 percent of the economy comparing boom to counterfactual periods (Table 4).
  - Average boom duration to date: 11 years.
- Aggregated synchronization by years since start of boom (Figure 5):
  - Median total GDP indexed to 100 at t=0 rises strongly during boom.
  - Median non-resource GDP indexed to 100 at t=0 is essentially flat over boom period.
  - No tendency for acceleration in non-resource GDP growth during later years of booms.
- Country-level aggregated calculation:
  - Only 5 of the 18 countries show non-resource growth over 2 percent per year during booms (sum of coefficients in Table 9).
- Illustrative country cases:
  - Zambia: non-resource real per-capita growth 1.9 percent per year during boom; 2.1 percent during counterfactual.
  - Saudi Arabia: total GDP per-capita growing since boom started in 2002, but non-oil real GDP per-capita growth not rapid during boom and not noticeably different than counterfactual.

### Comparison with the 1970s booms
- Five-country series (Saudi Arabia, Algeria, Bolivia, Libya, Trinidad and Tobago) indexed to 100 in 1973 (Figure 6):
  - Non-oil GDP surged at start of 1970s oil boom along with total GDP.
  - Both total and non-oil GDP per person reverted to pre-boom levels eventually; this took 17 years to play out in the 1970s boom.
- Nine 1970s-boom countries (Algeria, Bahrain, Kuwait, Libya, Oman, Qatar, Saudi Arabia, United Arab Emirates, Venezuela):
  - Setting 1986 non-oil GDP per person = 100, median GDP per-person = 123 by 2000.
  - Median implies an average annual growth rate of 1.5 percent.
  - Mean GDP per-person growth = 0.58 percent (over the same interval).
- Conclusion: currently booming economies show little evidence of outperforming 1970s counterparts; long gestation hypothesis undermined by 1970s evidence.

### Possible bias from country selection and excluded cases
- Reasons for exclusion and associated observations:
  - Bahrain, Kuwait, Venezuela: high resource production in 2000s but no clearly defined boom vs. counterfactual due to persistent volatility.
  - Angola, Chile, Yemen: lacked clearly defined counterfactual periods for different reasons:
    - Angola and Yemen: insufficient pre-boom data.
    - Chile: late-1990s low copper prices deemed too short to be a legitimate counterfactual.
  - Botswana: lacks a clear 2000s boom and counterfactual; long-running diamond export profile dating to late 1970s; non-mining GDP per-capita grew at 4.7 percent per annum over the period if considered a boom.
  - Angola: civil war (1975-2002) and socialist planning to ~1990 confound boom timing; non-oil economy boomed after 2000; by 2010 real per-capita GDP recovered to 1975 independence level—suggesting post-war bounce-back rather than overcoming the curse.
- Specific excluded-country illustrations:
  - Yemen (Figure 8): log GDP per-person whole economy vs. non-oil economy around boom start shows lack of faster non-oil growth during boom.
  - Chile (Figure 9): using 1998 as boom start, non-Copper GDP per person growth during boom was not faster than before.
- Lagged-impact argument:
  - Few cases show late acceleration in non-resource GDP during 2000s booms:
    - Angola: growth recovery plausibly tied to end of civil war.
    - Equatorial Guinea: boom started in 1994 with growth accelerating upward in 2005 (nine years after start).
    - Papua New Guinea: sharp recovery in 2006, long after its boom began.
  - Aggregate evidence: no general tendency for non-resource GDP growth to accelerate late in boom periods.

### Saving and investment
- Saving rates and current account:
  - For 16 countries with data, mean saving rates rose from 16 percent of non-resource GDP during counterfactual to 27 percent during boom (Table 5).
  - Current account shifted towards surplus by approximately 5 percentage points of GDP (Table 6).
  - Chad identified as an outlier; excluding Chad:
    - Mean saving rate rose from 0.17 to 0.32 (table context).
    - Current account shifted from a deficit of 0.06 to a surplus of 0.04.
- Investment rates:
  - Mean investment rates (16 countries) rose from 22 to 27 percent of GDP during booms compared to counterfactuals (Table 7).
  - Public investment share rose slightly from a mean of 9 percent of GDP during counterfactuals to 10 percent during booms (Table 8).
  - Private investment rose from 14 to 18 percent of GDP.
  - Two Gulf States (Saudi Arabia and UAE) show slight decline in investment ratio; Malaysia shows a larger decline.
- Interpretation:
  - Despite higher saving and sizable current account surpluses (more saving held offshore), domestic investment effort remained strong or rose.
  - Data suggest low returns to investment in the domestic (non-resource) economy, rather than lack of saving or investment effort, as explanation for weak non-resource growth.
- Comparison with 1970s booms:
  - Six countries with booms in both periods (Bolivia, Libya, Oman, Saudi Arabia, Trinidad and Tobago, United Arab Emirates):
    - Mean domestic investment shares of GDP approximately the same across periods: 23 percent (2000s) vs. 21 percent (1970s).
    - Public investment share in GDP: 11 percent in both periods.
  - Conclusion: investment rates did not decline in the 2000s booms relative to counterfactuals or to 1970s booms; higher saving rates in 2000s imply more saving held offshore (sovereign wealth funds), but domestic investment remained sufficiently high to expect some positive impact that has not materialized.

### Emirates and Qatar: labor productivity and population/employment dynamics
- United Arab Emirates and Qatar: rapid real GDP growth in non-hydrocarbon economy, but:
  - Rapid population growth due to labor migration reduces output per person gains.
  - Employment growth exceeded population growth (limited employment data), causing value-added per worker to grow even more slowly than GDP per-capita.
- UAE specific facts (table 10 summary):
  - Non-Oil annual real GDP growth during boom period (2002 to 2011): 5.8 percent.
  - Number of workers grew: 9.5 percent per year.
  - Labor productivity change implied: -3.7 percent average annual rate.
- Qatar specifics (available 2006–2012 data):
  - Non-Oil GDP growth: 14.2 percent per year.
  - Employment growth: 12.9 percent per year.
  - Labor productivity growth in non-oil sector: 1.1 percent per year.
- These labor productivity outcomes complement Table 9 results showing slow growth in GDP per-person in these countries.

*Source: IMF working paper content provided in the supplied PDF excerpt.*

### 2.2  Post Soviet Countries

### 2.2  Post Soviet Countries

### U-shaped transition pattern and empirical validation
- The u-shaped profile of total GDP is described as a natural outcome of a two-sector model in which one sector declines sharply (the state sector) while another rises gradually from a small base (the new private sector).
- The paper first tests, and confirms, that the evidence supports the common u-shape for the path of GDP over the transition period.
- Empirical specification used to confirm the u-shape:
  - Equation (14): ln() = 0 + =2010 X =1994   + , where  is non-resource GDP measured in constant-price local currency, normalized with 1994=100, and  are year dummies.
  - Estimation for eleven post-soviet economies yields coefficients that trace out a U-shape (reported in table 11 panel A).
  - A more parsimonious quadratic representation is used:
    - Equation (15): ln() = 0 + 1  + 2 2 + , achieving similar explanatory power (table 11 panel B).
- Empirical validation is based on the similarity in the adjusted R2’s between the year-dummy regression and the quadratic specification.

### Method for testing impact of resource booms
- Focus: growth in the non-resource economy (non-resource GDP), using pre-boom periods as counterfactuals.
- Post-soviet countries identified as experiencing resource booms: Azerbaijan, Kazakhstan, Russia, and Turkmenistan.
- Testing approach:
  - Allow 2 to be country-specific by adding country-specific dummy variables interacted with the quadratic term.
  - Interpret a higher estimated 2 for a particular country as faster non-resource GDP growth relative to comparator countries, consistent with successful use of resource rents to raise productivity in the non-resource economy.
  - Estimating equation used:
    - Equation (16): ln() = 0 + 1  + 2 2 + 2  2 + .
- Note: non-resource GDP is pegged at 1994=100 for all countries; country-specific 2 indicates deviation in growth path from comparators.

### Findings on resource booms in post-Soviet economies
- The results (table 12) indicate that the five resource intensive countries experienced slower growth during their resource boom relative to resource-poor countries.
  - Growth was statistically significantly slower than resource-poor countries for all except Azerbaijan.
- Conclusion drawn: little evidence that the resource booms served to accelerate non-resource GDP growth above levels experienced by other post-soviet economies.
- The paper states: "Based on this evidence it is difficult to claim that the resource booms served to raise the path of GDP above what it would have been without the booms."

### Broader conclusions across booming economies (method and outcomes)
- Purpose: present a method for testing whether newly booming economies overcome the slow-growth syndrome known as the curse of natural resources by focusing on non-resource GDP growth and using pre-boom periods as counterfactuals.
- Implementation: data deemed sufficiently clean to conduct tests for 18 booming economies.
  - Of these 18 cases:
    - 7 showed higher average (non-resource) growth during the boom than before.
    - 11 showed lower growth.
  - None were found in which the economy had overcome the curse in the strong sense of having statistically significantly higher growth during, compared to before, the boom period.
  - In one case, Bolivia, growth was significantly lower.
  - Equatorial Guinea and perhaps Papua New Guinea may be exceptions on year-by-year data, but growth was too short lived to be significant in the statistical tests.
- Post-soviet economies examined separately: it was found that none of thefive resource-rich countries showed significantly higher growth during their resource booms.
- Additional evidence and considerations:
  - The 1970s oil-boom data suggest the non-resource economy on average boomed during the Oil boom; median GDP per-capita in the non-Oil economy grew by only 1.48 percent per year over 1986-2000.
  - Many post-2000 boom periods have already lasted ten years, with no evidence within these episodes of higher growth overall at the end than the beginning.
  - Two countries sometimes cited as exceptions (Equatorial Guinea and Papua New Guinea) are not hugely convincing due to unusually volatile GDP data. Botswana is noted as a long-standing success case, but its fast non-resource GDP growth in the 2000’s is not a new phenomenon unique to the recent boom.
- Dominant finding: no change in growth rates of non-resource GDP per-capita during the recent boom years, contrary to expectations given large resource revenues and rising investment rates (including public investment) in several countries.

### Caveats, critiques, and welfare considerations
- Potential selection bias in country choice or counterfactual date choice was considered case by case; no evidence was found that excluded countries were systematically different on measurable items.
- Possible critique addressed: expectation that the non-resource economy might slump during a boom and recover afterwards. The paper argues:
  - Although some traded sectors might slump, non-traded sectors could boom, so no presumption of an overall slump.
  - Historical evidence from the 1970s does not support a generalized slump hypothesis.
- Further critique: non-resource growth might eventually accelerate if more time passed; the paper finds no historical support for such reversal (1970s experience and post-2000 evidence).
- Welfare versus GDP growth:
  - Growth in per-capita GDP is not identical to growth in welfare.
  - In several countries, boom revenues have been invested abroad, funded social expenditures, subsidized education/transportation/health spending, or used for indirect transfers to citizens.
  - Valuing these expenditures is part of a complete assessment, but non-resource per-capita growth remains a central metric given policy goals to diversify economies beyond rentier outcomes.

* _wp15237 - 2.2  Post Soviet Countries_*

### References

### References

### Bibliography
- Hartwick J. M. (1977) "Intergenerational Equity and the Investing of Rents from Exhaustible Resources" American Economic Review 66: 972-74
- Berg, Andrew, Rafael Portillo, Shu-Chun Yang and Luis-Felipe Zanna (2012), "Public Investment in Resource Abundant Low-Income Countries", International Monetary Fund.
- Gelb, Alan (2010), "Economic Diversification in Resource Rich Countries" Paper for IMF High-Level Seminar, Algeirs November 2010.
- Gelb, Alan H. and Associates, (1988) Oil windfalls: blessing or curse? New York: Oxford University Press.
- Arezki, Rabah, Kirk Hamilton and Kazim Kazimov (2011) "Resource Windfalls, Macroeconomic Stabilityand Growth: The Role of Political Institutions" IMF Working Paper WP/11/142.
- Ploeg, F. van der (2011), Natural Resources: Curse or Blessing?, Journal of Economic Literature, 49(2), 366-420
- Ploeg, F. van der and Venables Anthony J., (2012) Natural Resource Wealth: The Challenge of Managing a Windfall, Annual Review of Economics, 2012.3: 325-337.
- Sachs, J. and A. M. Warner. 1995. “Natural Resource Abundance and Economic Growth.” NBER Working Paper No. 5398, Cambridge, MA: NBER.
- Ismail, K. K. (2011), The Structural Manifestation of the ‘Dutch Disease’ In Oil-Exporting Countries, mimeo, International Monetary Fund.
- Brahmbhatt, Milan, Otaviano Canuto, and Ekaterina Vostroknutova. 2010. "Dealing with Dutch Disease." Economic Premise 16. Washington, D.C.: World Bank.
- Gylfason, Thorvaldur, Tryggvi Thor Herbertsson, and Gylfi Zoega, (1999) "A Mixed Blessing: Natural Resources and Economic Growth." Macroeconomic Dynamics, 3(2): 204-25.
- Bravo-Ortega, Claudio, and Jose de Gregorio. 2005. "The Relative Richness of the Poor? Natural Resources, Human Capital, and Economic Growth." World Bank Policy Research Working Paper 3484.
- Ndulu, O’Connell, Bates Collier and Soludo (2008) eds., The Political Economy of Economic Growth in Africa, 1960-2000, Volume 1, Cambridge: Cambridge University Press

### Sample selection and exclusions
- List of countries that pass the first screening (total entries shown) and reasons for elimination from the final sample:
  - Countries listed (with elimination reason columns indicated in source): Algeria, Angola, Bahrain, Bolivia, Chad, Chile, Rep Congo, Dem Rep Congo, Ecuador, Egypt, Equatorial Guinea, Guinea, Guinea Bissau, Iran, Kuwait, Lao PDR, Libya, Malaysia, Mauritania, Mongolia, Mozambique, Oman, Papua New Guinea, Qatar, Saudi Arabia, Sudan, Togo, Trinidad and Tobago, United Arab Emirates, Venezuela, Yemen, Zambia.
  - Totals reported: Implausible Data = 6; Inadequate Counterfactual Period = 9; Final Sample = 18

### Counterfactual and Boom periods (dates)
- Country-specific counterfactual and boom periods (period ranges as stated):
  - Algeria: Counterfactual 1986-1997; Boom 1998-2011
  - Bolivia: Counterfactual 1987-1998; Boom 1999-2011
  - Chad: Counterfactual 1990-2002; Boom 2003-2011
  - Ecuador: Counterfactual 1979-1997; Boom 1998-2011
  - Equatorial Guinea: Counterfactual 1980-1991; Boom 1992-2011
  - Lao PDR: Counterfactual 1991-1999; Boom 2000-2011
  - Libya: Counterfactual 1986-1998; Boom 1999-2010
  - Malaysia: Counterfactual 1993-2001; Boom 2002-2011
  - Mauritania: Counterfactual 1987-2002; Boom 2003-2011
  - Mongolia: Counterfactual 1992-2005; Boom 2006-2011
  - Mozambique: Counterfactual 1993-1998; Boom 1999-2011
  - Oman: Counterfactual 1986-1997; Boom 1998-2011
  - Papua New Guinea: Counterfactual 1976-1990; Boom 1991-2011
  - Qatar: Counterfactual 1987-1997; Boom 1998-2011
  - Saudi Arabia: Counterfactual 1986-2001; Boom 2002-2011
  - Trinidad and Tobago: Counterfactual 1983-1997; Boom 1998-2011
  - United Arab Emirates: Counterfactual 1995-2001; Boom 2002-2011
  - Zambia: Counterfactual 1997-2002; Boom 2003-2011

### Adjustments to counterfactual periods (reasons)
- Country-specific adjustments and rationale (as stated):
  - Algeria: Period with previous booms (through 1985) not included in counterfactual
  - Bolivia: Period with previous booms (through 1986) not included in counterfactual
  - Chad: Counterfactual period limited to post-1990 period of relatively stable policy under Idriss Deby
  - Ecuador: Beginning of counterfactual period is 1979 - chosen to avoid inclusion of earlier 1970’s boom
  - Equatorial Guinea: Counterfactual period excludes period of state breakdown prior to 1979
  - Lao PDR: Counterfactual period excludes socialist state planning prior to 1991
  - Libya: Period with previous booms (through 1986) not included; implausible trade data in 1960-1970
  - Malaysia: Beginning of counterfactual period is 1993 - chosen to avoid inclusion of earlier boom
  - Mauritania: Counterfactual period limited to post 1986 regime of Ould Taya
  - Mongolia: Counterfactual period excludes socialist state planning prior to 1991
  - Mozambique: Counterfactual limited to post-civil war, “syndrome-free” regime of Joaquim Chissano, 1992-
  - Oman: Counterfactual excludes Dhohar rebellion prior to 1976 and boom through 1986
  - Papua New Guinea: Counterfactual starts with independence in 1976
  - Qatar: Period with previous booms (through 1986) not included; implausible trade data in late 1960s
  - Saudi Arabia: Period with previous booms (through 1986) not included
  - Trinidad and Tobago: Earlier boom periods through 1982 not included in counterfactual
  - United Arab Emirates: Period with previous booms (through 1986) not included in counterfactual
  - Zambia: Counterfactual limited to post 1992 "syndrome free" period, and period of low commodity exports post-1997

### Increase in Natural Resource Rents during the Boom (value-added share of GDP, period averages)
- Country values: Counterfactual Period | Boom Period | Difference (Boom - Counterfactual)
  - Algeria: 0.214 | 0.319 | 0.105
  - Bolivia: 0.050 | 0.110 | 0.060
  - Chad: 0.001 | 0.294 | 0.293
  - Ecuador: 0.075 | 0.119 | 0.044
  - Equatorial Guinea: 0.156 | 0.815 | 0.659
  - Lao People's Dem.Rep: 0.051 | 0.103 | 0.051
  - Libya: 0.275 | 0.505 | 0.230
  - Malaysia: 0.126 | 0.172 | 0.046
  - Mauritania: 0.191 | 0.326 | 0.135
  - Mongolia: 0.015 | 0.114 | 0.099
  - Mozambique: 0.007 | 0.163 | 0.156
  - Oman: 0.335 | 0.422 | 0.087
  - Papua New Guinea: 0.044 | 0.228 | 0.184
  - Qatar: 0.382 | 0.492 | 0.110
  - Saudi Arabia: 0.296 | 0.383 | 0.087
  - Trinidad and Tobago: 0.200 | 0.379 | 0.179
  - United Arab Emirates: 0.238 | 0.291 | 0.053
  - Zambia: 0.111 | 0.189 | 0.079
- Mean values reported: Counterfactual 0.154 | Boom 0.301 | Difference 0.148

### Saving rates (fraction of GDP, period averages)
- Country values: Counterfactual Period | Boom Period | Change in saving ratio
  - Algeria 0.28 | 0.45 | 0.17
  - Bolivia 0.09 | 0.19 | 0.10
  - Chad -0.01 | -0.52 | -0.51
  - Ecuador 0.15 | 0.21 | 0.06
  - Equatorial Guinea 0.10 | 0.38 | 0.28
  - Lao People's Dem.Rep n.a. | n.a. | n.a.
  - Libya 0.19 | 0.56 | 0.36
  - Malaysia 0.36 | 0.34 | -0.03
  - Mauritania 0.16 | 0.16 | 0.00
  - Mongolia 0.22 | 0.40 | 0.18
  - Mozambique 0.14 | 0.09 | -0.05
  - Oman 0.13 | 0.31 | 0.18
  - Papua New Guinea n.a. | n.a. | n.a.
  - Qatar 0.08 | 0.51 | 0.43
  - Saudi Arabia 0.13 | 0.37 | 0.24
  - Trinidad and Tobago 0.18 | 0.32 | 0.14
  - United Arab Emirates 0.31 | 0.29 | -0.02
  - Zambia 0.03 | 0.19 | 0.17
- Mean values: Mean 0.16 | 0.27 | 0.11
- Mean excluding Chad: ex-Chad 0.17 | 0.32 | 0.15

### Current Account (fraction of GDP, period averages)
- Country values: Counterfactual Period | Boom Period | Change in current account
  - Algeria 0.02 | 0.19 | 0.17
  - Bolivia -0.06 | 0.03 | 0.09
  - Chad -0.16 | -0.80 | -0.64
  - Ecuador -0.01 | 0.01 | 0.02
  - Equatorial Guinea -0.44 | -0.31 | 0.13
  - Lao People's Dem.Rep n.a. | n.a. | n.a.
  - Libya 0.07 | 0.26 | 0.18
  - Malaysia 0.01 | 0.13 | 0.11
  - Mauritania -0.05 | -0.19 | -0.14
  - Mongolia -0.01 | 0.01 | 0.02
  - Mozambique -0.07 | -0.12 | -0.05
  - Oman -0.03 | 0.06 | 0.09
  - Papua New Guinea n.a. | n.a. | n.a.
  - Qatar -0.16 | 0.22 | 0.38
  - Saudi Arabia -0.06 | 0.20 | 0.26
  - Trinidad and Tobago -0.01 | 0.13 | 0.14
  - United Arab Emirates 0.07 | 0.09 | 0.02
  - Zambia -0.13 | -0.03 | 0.11
- Mean values: Mean -0.06 | -0.01 | 0.06
- Mean excluding Chad: ex-Chad -0.06 | 0.04 | 0.10

### Investment rates (fraction of GDP, period averages)
- Country values: Counterfactual Period | Boom Period | Change in investment ratio
  - Algeria 0.26 | 0.26 | 0.00
  - Bolivia 0.15 | 0.16 | 0.01
  - Chad 0.15 | 0.28 | 0.13
  - Ecuador 0.16 | 0.20 | 0.04
  - Equatorial Guinea 0.54 | 0.69 | 0.15
  - Lao People's Dem.Rep n.a. | n.a. | n.a.
  - Libya 0.12 | 0.30 | 0.18
  - Malaysia 0.35 | 0.21 | -0.14
  - Mauritania 0.21 | 0.35 | 0.14
  - Mongolia 0.23 | 0.39 | 0.16
  - Mozambique 0.21 | 0.21 | 0.00
  - Oman 0.16 | 0.25 | 0.09
  - Papua New Guinea n.a. | n.a. | n.a.
  - Qatar 0.24 | 0.29 | 0.05
  - Saudi Arabia 0.19 | 0.17 | -0.02
  - Trinidad and Tobago 0.19 | 0.19 | 0.00
  - United Arab Emirates 0.24 | 0.20 | -0.04
  - Zambia 0.16 | 0.22 | 0.06
- Mean values: Mean 0.22 | 0.27 | 0.05
- Mean excluding Chad: ex-Chad 0.23 | 0.27 | 0.05

### Public investment rates (period averages)
- Country values: Counterfactual Period | Boom Period | Change in Public Investment
  - Algeria 0.07 | 0.12 | 0.04
  - Bolivia 0.04 | 0.07 | 0.03
  - Chad 0.00 | 0.06 | 0.06
  - Ecuador 0.00 | 0.05 | 0.05
  - Equatorial Guinea 0.30 | 0.17 | -0.13
  - Lao People's Dem.Rep n.a. | n.a. | n.a.
  - Libya 0.10 | 0.20 | 0.10
  - Malaysia 0.12 | 0.11 | -0.01
  - Mauritania 0.06 | 0.07 | 0.01
  - Mongolia 0.07 | 0.07 | 0.00
  - Mozambique 0.12 | 0.11 | -0.01
  - Oman 0.11 | 0.13 | 0.02
  - Papua New Guinea n.a. | n.a. | n.a.
  - Qatar 0.06 | 0.07 | 0.01
  - Saudi Arabia 0.07 | 0.07 | 0.00
  - Trinidad and Tobago 0.05 | 0.09 | 0.03
  - United Arab Emirates 0.10 | 0.08 | -0.03
  - Zambia 0.10 | 0.05 | -0.04
- Mean values: Mean 0.09 | 0.10 | 0.01
- Mean excluding Equatorial Guinea: ex-Equatorial Guinea 0.07 | 0.09 | 0.02

### Country-by-country regression estimates: Non-resource growth during the Boom
- Regression specification (stated): ln(Yit) − ln(Yit−1) = θ1i + θ2i d1i + Oi
- Reported country results (selected coefficients and significance as listed for Growth Before the Boom (a1) and Change in Growth During Boom (a2)):
  - Algeria: Growth Before the Boom (a1) -2.122 (t-stat -1.619); Change in Growth During Boom (a2) 2.970 (t-stat 1.635); Observations 25; R-squared 0.104
  - Bolivia: (a1) 2.384*** (t-stat 4.955); (a2) -2.496*** (t-stat -3.670); Observations 24; R-squared 0.380
  - Chad: (a1) 0.704 (t-stat 0.240); (a2) -1.560 (t-stat -0.376); Observations 16; R-squared 0.010
  - Ecuador: (a1) 0.503 (t-stat 0.557); (a2) 0.156 (t-stat 0.110); Observations 32; R-squared 0.000
  - Equatorial Guinea: (a1) -3.681 (t-stat -0.222); (a2) 20.958 (t-stat 0.895); Observations 24; R-squared 0.035
  - Lao PDR: (a1) 3.261*** (t-stat 4.485); (a2) 1.428 (t-stat 1.456); Observations 20; R-squared 0.105
  - Libya: (a1) -2.469 (t-stat -0.572); (a2) -1.762 (t-stat -0.276); Observations 24; R-squared 0.003
  - Malaysia: (a1) 3.731* (t-stat 1.816); (a2) -1.686 (t-stat -0.580); Observations 18; R-squared 0.021
  - Mauritania: (a1) -0.393 (t-stat -0.237); (a2) -2.338 (t-stat -0.778); Observations 23; R-squared 0.028
  - Additional country panels (Mongolia through Zambia) report analogous (a1) and (a2) values, observations, and R-squared as tabulated in the source.
- Notes: t-statistics in parentheses; significance denoted *** p<0.01, ** p<0.05, * p<0.1

### Labor productivity growth: United Arab Emirates and Qatar
- United Arab Emirates (GDP at constant 2007 Prices, Million Dirhams)
  - 2002 Oil GDP 265728; Non Oil 402889; Total 668618
  - 2011 Oil GDP 315410; Non Oil 667316; Total 982725
  - Ave. Annual Growth (Oil 1.9%, Non Oil 5.8%, Total 4.4%); Workers growth (Oil 7.8%, Non Oil 9.5%, Total 9.5%); Labor Productivity growth (Oil -5.9%, Non Oil -3.7%, Total -5.1%)
  - Source: Department of Economic Statistics
- Qatar (GDP at constant 2004 Prices, Million Qatari Riyal)
  - 2006/7 Oil GDP 77662; Non Oil 93089; Total 170751
  - 2012 Oil GDP 147726; Non Oil 192919; Total 340646
  - Ave. Annual Growth (Oil 12.4%, Non Oil 14.2%, Total 13.4%); Economically Active (15+ years) Workers growth (Oil 16.7%, Non Oil 12.9%, Total 13.2%); Labor Productivity growth (Oil -3.7%, Non Oil 1.1%, Total 0.2%)
  - Source: Ministry of Development Planning and Statistics and Qatar Statistics Authority

### Regressions on Post-Soviet countries: U-shaped path and resource-rich comparison
- Panel A (log of GDP in non-resource economy, Index, 1994=100): Year-specific coefficients from 1994 through 2010 reported (e.g., year 1994 -0.178; year 1995 -0.344; ... year 2010 0.524). N = 193; Adj R-squared = 0.58; F(17,175) = 16.52; Prob > F = 0.00
- Panel B: Quadratic specification (year and year^2) shows year coefficient -0.187 and year^2 coefficient 0.0038; Constant 6.884; N = 193; Adj R-squared = 0.57; F(2,190) = 135.19; Prob > F = 0.00
- Regression showing resource-rich Post-Soviet countries did not experience faster economic growth (Dependent Variable: log of GDP in non-resource economy, Index, 1994=100):
  - year coefficient -0.184 (Std. Err. 0.044; T -4.140)
  - year^2 coefficient 0.00376 (Std. Err. 0.0007; T 5.380)
  - Interaction terms year^2* country dummies (AZE, KAZ, TAJ, RUS, TUR) show negative coefficients for KAZ, TAJ, RUS with associated t-statistics as listed.
  - Constant 6.838 (Std. Err. 0.688; T 9.930)
  - N = 193; Adj R-squared = 0.62; F(7,185) = 45.3; Prob > F = 0.00

### Appendices: Graphical and panel summaries
- Appendix 1: Graphs showing boom periods and counterfactual periods for the 18 countries retained for analysis. Plotted lines show natural resource exports as a share of GDP (USD), raw and smoothed. Shaded areas denote boom periods; remaining area denotes counterfactual periods. Notations in graphs include events such as "Socialist Planning Abandoned" and "Civil War Ended".
- Separate graphs shown for 9 countries not used in the analysis: Angola, Bahrain, Chile, Venezuela, Yemen, Sudan, Egypt, Iran, Kuwait.
- Appendix 2: Further detail on total and non-resource GDP per person, before and after the boom years. Panels report indexed values (=100 in year 0) and median values across approximately 15 Countries for Total GDP and Non-resource GDP.

*Source: _wp15237 - References (PDF chapter/section).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15237.pdf_