## wpiea2023090-print-pdf

## Source details

**Canonical URL:** [wpiea2023090-print-pdf](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023090-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023090-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023090-print-pdf.pdf.json)

---

### Overview and context
- Guinea faces significant development and social inclusion challenges while preparing to transition back to democratic elections.
- Guinea consistently ranks among the lowest 12 countries worldwide in the World Bank’s Human Capital Index (HCI) without meaningful improvements over the past decade.
- In health and education, Guinea lags behind the average Sub-Saharan African country and behind the average Low-Income Country (LIC).

### Mining sector and fiscal potential
- Mining accounted for 84 percent of Guinea’s total exports and 21 percent of its GDP as of end-2021.
- Guinea has an estimated 7.4 billion tons of bauxite reserves, the world’s largest.
- Short-run revenue potential: more transparent pricing schemes by multinationals can generate substantial revenue.
- Long-run revenue potential: applying the existing mining code to new contracts while limiting corporate income tax exemptions for mining producers would raise substantial revenues.
- Laffer-curve and revenue findings:
  - Using Caldara et al. (2016) top elasticity estimate 0.21 → γ_x = 0.1749.
  - Maximum revenue in general equilibrium occurs at effective tax rate between 64.4 and 74.4 percent; mining sector size falls from 15 percent of GDP to roughly 9 percent of GDP.
  - Steady-state output shrinks by between 5 and 6 percent (effect concentrated in mining).
  - Baseline reform (effective rate at 34.4 percent) yields additional revenue ≈ 2.6 percent of GDP and implies steady-state output drop of less than 2 percent.
  - Effective mining tax rate used in calibration: 0.1440 (statutory 30 percent combined with tax gap 52 percent).
  - Labor share of mining calibrated at 8 percent.

### Development policies analyzed
- Three policy areas quantified when financed by additional mining-derived fiscal space:
  - Human capital formation through early and basic education.
  - Inclusive infrastructure investments (transportation, communications, energy) that impact TFP for all productive activities, including informal work.
  - Social protection via lump-sum cash transfers to all households (targeting left for future work).
- Educational policies are a tractable stand-in for other human capital policies (child mortality, nutrition, health, longevity).
- Infrastructure modeled as inclusive and productivity-enhancing (panel-regression-based assumptions); contrasted with “white elephant” outcomes.
- Cash transfer policy modeled as lump-sum transfers to all households (no means-tested targeting modeled).

### Model structure and household behavior (high-level)
- Infinite-horizon economy with heterogeneous urban and rural households and idiosyncratic labor productivity histories.
- Consumer types and population shares:
  - Urban households (j=1), Rural households (j=2), Farmtrepreneurs (j=3).
  - Population shares (μ1, μ2, μ3) = (0.33, 0.6463, 0.0237).
- Annual consumption basket c = (c_f, c_sv, c_m); utility u(c).
- Households choose s (share of time in formal labor), c (consumption), a′ (savings).
- Dynamic programming problem (location l):
  - V_l(a,z) = max_{s,c,a′} [u(c) + β E[V_l(a′,z′) | z]]
  - Budget constraint: a(1 + r) + y(z,l,w_l,s) + y*(z,l,w_l,1−s) − T(l,c,y,p) ≥ p·c + a′.
- Formal labor share closed-form (for positive formal work):
  - s_l(z,a) = max ( 0, 1 − 1/z ( α ε p_l / [ w_l (1 − ˆτ_income) ] )^{1/(1−α)} )  (Equation 3)
  - s increases with wages w_l and skill z; s decreases with effective tax rate and informal price p_l; corner solution s = 0 possible.

### Farmtrepreneurs (section 1.3)
- Farmtrepreneurs differences and roles:
  - No idiosyncratic risk.
  - Can hire rural labor to produce either informal food or a capital-intensive export commodity.
  - Cannot borrow capital; accumulate capital over time to produce export commodity.
- Dynamic optimization:
  - Value function: Vf(k) = max_{c,k′} [u(c) + β Vf(k′)]  (Equation (4))
  - Budget constraint: k(1−δ) + π(w_rural) + π∗(k,w_rural) − Tf(R∗1,R∗2) ≥ c · p + k′  (Equation (5))
- Profit functions:
  - π(w_rural) = max_{n1} R1(n1) − w_rural n1, with R1 = p_f ε1 n1^{α1}.
  - π∗(k,w_rural) = max_{n2} R2(k,n2) − w_rural n2, with R2 = p∗ ε2 (k^{ζ} n2^{1−ζ})^{α2}.
- Taxation on farmtrepreneurs:
  - Tf(R∗1,R∗2) = ˆτf R∗1 + ˆτfx R∗2 − g.
- Calibration context:
  - Share of farmtrepreneurs among rural population = 2.37 percent.
  - Average sizes: Informal farms = 1.38 hectares; Formal farms = 238 hectares.

### Calibration: interest rate, debt, and taxes (section 3.2)
- Real interest rate set to 1.6 percent (13.4 percent average 90 day t-bills and bonds minus annual inflation 11.78 percent).
- Domestic government debt = 15.2 percent of GDP (end-2021).
- Statutory rates (KPMG 2019):
  - VAT = 18 percent; CIT (non-mining non-oil) = 25 percent; PIT (labor income + social security) = 29 percent; mining statutory CIT = 30 percent.
- Empirical tax collection averages (2015-19):
  - Mining tax collection = 2.16 percent of GDP.
  - Indirect tax collection = 6.02 percent of GDP.
- Tax gap calculations:
  - Mining tax gap = 100(1−2.16/4.5) = 52 percent (expected collection at 30 percent statutory rate and mining ≈ 15 percent of GDP → 4.5 percent of GDP).
  - VAT tax gap = 100(1−6.02/14.4) = 58.2 percent (estimated VAT base = 80 percent of GDP → statutory VAT revenue = 14.4 percent of GDP).
  - Other tax gaps approximated midpoint = 53.10 percent.

### Calibration: preferences, outputs, government, trade, inequality, informality
- Preferences:
  - u(c) = c^{γ1} / f + γ2 log(c_sv) + γ3 log(c_m).  (Equation 6)
  - Calibration targets from Guinea’s 2018/2019 household survey: (1.05,0.42,0.12).
- Sectoral output share calibration targets (agriculture, services, commodity exports, mining, manufactures):
  - (0.23,0.39,0.008,0.15,0.22).
- Government and trade:
  - Average public consumption (2015-2019) = 17 percent of GDP; assumed 11 percent of GDP manufactured goods initially, adjusted to 4 percent to balance current account in long run.
  - Total investment = 10 percent of GDP.
  - Net imports: manufactured goods = 9 percent of GDP; agricultural goods net imports = 13 percent of GDP → total imports = 22 percent of GDP.
  - Total exports: mining = 15 percent of GDP; agricultural commodities = 0.8 percent → total exports = 15.8 percent of GDP → initial permanent current account deficit = 7 percent of GDP.
- Inequality calibration targets:
  - Poverty rate target = 43.7 percent.
  - Gini coefficient target = 50.
  - Wealth share of top 10 percent target = 90 percent.
- Informality calibration:
  - α_rural = α_urban = 0.3.
  - Goal: approximate total share of work time devoted to informal work = 95 percent.

### Fixed and equilibrium parameters (selected)
- Pre-determined parameters (selected):
  - Urban Share of Population = 0.33
  - Rural Share of Population = 0.6463
  - Large Farm Population = 0.0237
  - Govt. Service Consumption to GDP = 0.06
  - Price Elasticity of Food Imports = 2.5
  - Baseline Public Infrastructure Stock to GDP = 0.5
  - World Real Interest Rate = 0.016
- Pre-determined tax parameters (selected statutory rates and tax gaps):
  - Agriculture VAT: Statutory Rate 0.18, Tax Gap 0.582
  - Manufacture VAT: Statutory Rate 0.18, Tax Gap 0.582
  - Food CIT: Statutory Rate 0.25, Tax Gap 0.531
  - Export CIT: Statutory Rate 0.25, Tax Gap 0.531
  - Manufacture CIT: Statutory Rate 0.32, Tax Gap 0.531
  - Labor PIT: Statutory Rate 0.25, Tax Gap 0.531
- Parameters calculated in equilibrium (selected):
  - CES Curvature in Utility of Food = 1.0556
  - Weight of Log Utility of Services = 0.8856
  - Weight of Log Utility of Manufactures = 0.2570
  - Discount Factor = 0.92
  - Idiosyncratic Shock Variance Rural = 0.2666
  - Idiosyncratic Shock Variance Urban = 0.1736
  - Idiosyncratic Shock Persistence (Rural and Urban) = 0.93
  - TFP of Manufacturing = 1.4142
  - TFP of Mining = 0.5573
  - Poverty Line = 0.6444
  - World Price of Food = 0.6965
  - GDP = 2.527

### Human capital formation (section 4.1)
- Return to one-year expenditure:
  - Defined: 1 + θ ≡ w0 ̄z1(e) − w0 ̄z(0) / e.
  - Baseline lifetime rate of return θ = 0.13.
  - z1 = z0 + (1 + θ) / w0 e.
- Multi-year investment per student (t years):
  - z1 = z0 + t (1 + θ) / w0 e.
  - Total annual cost in a region with cohort size S: E = S t e.
- Rural–urban allocation rule:
  - e_rural / w_rural = e_urban / w_urban.
  - e_rural = w_r E / S t (w_u μ_u + w_r μ_r).
  - e_urban = w_u E / S t (w_u μ_u + w_r μ_r).
- Parameterization:
  - Share of population ages 0–14 ≈ between 0.4 and 0.47 historically → cohort size ≈ 47/15 = 3.13 percent.
  - Focus on early and primary education: t = 10.
- Long-run simulation outcomes:
  - Increasing education investment in steps from zero to 2 percent of GDP.
  - At 2 percent investment government’s long run surplus = 0.52 percent of GDP (down from initial 2.57 percent of GDP).
  - Assuming impact spread over 30 years:
    - Annual educational investments of 0.5 percent of GDP can generate 0.41 percent annual growth over 30 years.
    - Each additional 0.5 percent of GDP invested in education initially boosts annual growth by roughly 40 basis points, with diminishing marginal effects thereafter.
  - Distributional effects at 2 percent of GDP:
    - Poverty rate can fall to 11 percent in the long run.
    - Gini drops from 0.55 to 0.47.
    - Rural–urban income gap remains near initial level.
- Formalization effects:
  - Urban formalization increases; rural formalization falls under higher education investment.
  - Mechanism: increases in education can lower average return to education in urban area while increasing it in rural area due to relative price movements across sectors.

### Essential dynamics and policy interactions (section 5.1)
- Driving processes:
  - Expansion of labor supply via skill accumulation from education.
  - Productivity increases from infrastructure capital accumulation.
- Timing and growth impacts:
  - Education reforms have quicker turnaround than infrastructure; impact on annual growth is larger for education but decays sooner.
  - Combined education + infrastructure policy is complementary: combined policy produces more real GDP than sum of separate policies.
- Inequality transition:
  - Transition for inequality indicators expected to be monotonic and gradual based on similar numerical experiments.
- Policy conclusions and recommendations:
  - Mining revenues, if properly collected and invested, could have transformative impacts in Guinea.
  - Policies combining education, infrastructure, and social transfers yield the most balanced impacts on GDP, public revenue, poverty, and inequality.
  - Specific findings:
    - Education and infrastructure are complementary.
    - Infrastructure-only policies of the magnitude studied do not substantially reduce poverty or inequality alone but are highly effective when combined with education.
    - Self-financing nature of policies allows distribution of extra revenue as social transfers.
      - Experiment assumes a flat transfer (not targeted) delivered without leakage; these transfers are highly effective in reducing poverty and inequality.
      - Well-designed targeted transfers could be even better.
  - Policy implication: tradeoffs between investing mining-generated resources into hard infrastructure versus combining with social infrastructure that builds human capital; the latter can have larger long-term impacts on growth, revenue, poverty and inequality.

### Relevant human capital and health indicators cited
- Guinea consistently ranks among the 12 countries with the greatest human capital challenges worldwide.
- Guinea ranks the seventh lowest worldwide in human capital outcomes for women.
- Health:
  - Ten in 100 children will die before his or her fifth birthday.
- Education:
  - Children complete an average of 7 years of education, but learning outcomes align with only 4.6 years of high-quality education.
  - Primary school enrollment progress flattened since 2016 and worsened across gender dimensions.

*Source: wpiea2023090-print-pdf*

### Introduction

### wpiea2023090-print-pdf - Introduction

### Overview and context
- Guinea faces significant development and social inclusion challenges while preparing to transition back to democratic elections.
- Guinea consistently ranks among the lowest 12 countries worldwide in the World Bank’s Human Capital Index (HCI) without meaningful improvements over the past decade.
- In health and education, Guinea lags behind the average Sub-Saharan African country and behind the average Low-Income Country (LIC).

### Mining sector and fiscal potential
- Mining accounted for 84 percent of Guinea’s total exports and 21 percent of its GDP as of end-2021.
- Guinea has an estimated 7.4 billion tons of bauxite reserves, the world’s largest.
- Short-run revenue potential: more transparent pricing schemes by multinationals can generate substantial revenue.
- Long-run revenue potential: applying the existing mining code to new contracts while limiting corporate income tax exemptions for mining producers would raise substantial revenues.

### Development policies analyzed
- The paper quantifies impacts of plausible development policies financed by additional mining-derived fiscal space:
  - Human capital formation through early and basic education.
  - Inclusive infrastructure investments (transportation, communications, energy) that impact TFP for all productive activities, including informal work.
  - Social protection via lump-sum cash transfers to all households (targeting left for future work).
- Educational policies are used as a tractable stand-in for other human capital policies (child mortality, nutrition, health, longevity); precise health-investment analysis is left for future work.
- Infrastructure assumptions are based on panel regressions employing broad year and country samples and contrast with “white elephant” outcomes; infrastructure is modeled as inclusive and productivity-enhancing.
- Cash transfer policy is modeled as lump-sum transfers to all households; means-tested targeting is not modeled here (analysis is conservative regarding ease of means-testing).

### Outcome indicators
- Policies are evaluated using indicators of economic performance, social inclusion, and development:
  - Poverty rates, poverty depth, and the Gini coefficient of household-level consumption expenditures.
  - GDP per capita, government surplus and tax revenue.
  - Rural/urban inequality.
  - Sectoral reallocation and diversification, including reliance on food imports.

### Methodology and scope
- Approach: perform stylized policy reforms within a calibrated model of Guinea’s economy and use outcomes to generate lessons for policymakers; lessons are framed by the model and are not direct policy prescriptions.
- Main messages:
  - Revenues from mining can be crucial for long-run development.
  - Education and infrastructure policies are complementary:
    - Without education, infrastructure policy has a small and ambivalent impact on poverty and inequality.
    - Without infrastructure, increased education can erode returns from education and may lead to more informality in rural areas.

### Related literature and contribution
- Builds on quantitative macro models with heterogeneity applied to macro-inequality in LICs.
- Comparisons:
  - Gibson and Rioja (2020): quantify infrastructure effects on growth, inequality and welfare; do not model foreign extractive industries or link education and infrastructure.
  - Mendes-Tavares et al. (2021): analyze fiscal revenue generation in a similar model; this paper adds interactions between development policy and revenue generation.
  - Fabrizio et al. (2017): examine distributional effects of tax and investment reforms in LIDCs; highlight differences across tax policies and expenditure-side impacts.
  - Buffie et al. (2020): explore complementarity among government policies including education and infrastructure in representative-agent dynamic macro models.
- Two principal contributions claimed:
  1. Provide a framework to quantitatively analyze the connection between foreign extractive sectors, growth and distribution.
  2. Quantitatively address interplay among education, infrastructure and transfers in a general equilibrium semi-open economy macro-inequality framework.

### Model structure (high-level)
- Infinite-horizon economy with heterogeneous urban and rural households experiencing idiosyncratic labor productivity histories.
- Three consumer types: urban households (type j=1), rural households (type j=2), and “farmtrepreneurs” (type j=3); each type comprises a share μ_j of the population with Σ μ_j = 1.
- Agents face independent idiosyncratic productivity shocks (modeled as AR(1) discretely approximated).
- Sectors:
  - Foreign mining sector: employs low-skilled workers, imports capital services, exports all output, and pays taxes.
  - Manufacturing sector: rents capital services (domestic or foreign), hires urban labor, produces and pays taxes.
  - Government services sector: uses government capital and urban labor to produce services competing with informal household services.
  - Informal production: households supply informal food (rural) and services (urban) using non-formal work time.

### Household behavior and choices
- Annual consumption basket c = (c_f, c_sv, c_m) of food, services, and manufactured goods; utility denoted u(c).
- Households choose s (share of time in formal labor), c (consumption basket), and a′ (savings for next year).
- Savings from previous year and return: a(1 + r) denotes household financial wealth at beginning of year (including interest r).
- Dynamic programming problem (for location l):
  - Value function: V_l(a,z) = max_{s,c,a′} [u(c) + β E[V_l(a′,z′) | z]]  (Equation 1)
  - Time constraint: 0 ≤ s ≤ 1.
  - Budget constraint: a(1 + r) + y(z,l,w_l,s) + y*(z,l,w_l,1−s) − T(l,c,y,p) ≥ p·c + a′.
- Informal production interprets non-formal time 1−s as informal production (or idle time used attempting to “make ends meet”).

### Income and tax formulation
- Taxes: flat consumption tax (VAT) and flat labor income tax (PIT), identical across regions; effective taxes include tax gaps.
- Effective taxes collected from households:
  - T(l,c,y,p) = τ_income (1−gap_income) y + τ_cons (1−gap_cons) c_formal · p − g  (Equation 2)
  - g denotes government transfers.
- Income definitions:
  - Formal labor income: y(l,z,s,w_l) = w_l z s.
  - Informal income: y*(l,z,1−s,p) = p_l ε [z(1−s)]^α  with 0 < α < 1 and ε > 0.
- Net labor income with shorthand ˆτ for effective income tax rate τ_income (1−gap_income):
  - y* + y − yˆτ = p_l ε [z(1−s)]^α + z w_l s (1 − ˆτ).
- Intratemporal problem for s (auxiliary problem):
  - max_s p_l ε [z(1−s)]^α + z w_l s (1 − ˆτ), subject to 0 ≤ s ≤ 1.
  - Marginal informal income: α ε z^α (1−s)^{α−1} (strictly decreasing).
  - Marginal formal income: z w_l (1 − ˆτ) (fixed at the margin).
- Closed-form expression for optimal formal labor share s (for positive formal work):
  - s_l(z,a) = max ( 0, 1 − 1/z ( α ε p_l / [ w_l (1 − ˆτ_income) ] )^{1/(1−α)} )  (Equation 3)
  - Implications:
    - s increases with wages w_l and skill z.
    - s decreases with the effective tax rate and the informal price p_l.
    - Corner solution s = 0 is possible.
    - Economy-wide changes in skills z can have ambiguous labor-supply effects due to general-equilibrium wage and price responses.
- Model allows differential reactions across rural and urban areas to reforms.

*Source: wpiea2023090-print-pdf - Introduction*

### 1.3    Farmtrepreneurs

### 1.3    Farmtrepreneurs

### Role and key distinctions
- Farmtrepreneurs constitute the third type of consumer and share the same preferences for consumption as households.
- Two key differences from rural households:
  - Farmtrepreneurs do not face idiosyncratic risk.
  - Farmtrepreneurs can hire labor in rural areas to produce either: (i) the same food produced informally by rural households, or (ii) an export commodity.
- Producing the export commodity requires capital and labor; producing food requires only labor (and, implicitly, land).
- Farmtrepreneurs cannot borrow capital in financial markets and therefore gradually accumulate capital to produce the export commodity—highlighting limited depth of financial markets in many LIDCs.

### Dynamic optimization problem
- Value function and Bellman problem:
  - Vf(k) = max_{c,k′} [u(c) + β Vf(k′)]  (Equation (4))
- Budget constraint:
  - k(1−δ) + π(w_rural) + π∗(k,w_rural) − Tf(R∗1,R∗2) ≥ c · p + k′  (Equation (5))
  - R∗1 and R∗2 are the optimal revenues from food and export commodity sales of the farmtrepreneur.
  - π and π∗ are the optimal profits from farming (food) and production of agricultural export commodities, respectively.
- Every year the farmtrepreneur chooses consumption basket c and next-year capital stock k′.
- Optimal profits depend on accumulated capital k and wages w_rural per unit of effective labor.

### Profit maximization for production activities
- Profits from food production:
  - π(w_rural) = max_{n1} R1(n1) − w_rural n1
  - R1 = p_f ε1 n1^{α1}
- Profits from export commodities:
  - π∗(k,w_rural) = max_{n2} R2(k,n2) − w_rural n2
  - R2 = p∗ ε2 (k^{ζ} n2^{1−ζ})^{α2}
- Parameters introduced:
  - ε1 and ε2: productivity constants
  - α1 and α2: controlling returns to scale
  - ζ: controlling capital and labor shares
- For simplicity, all of the farmtrepreneur’s capital is assumed to be used for production of export commodities so they cannot save in the financial market.

### Taxation and transfers
- Taxes on optimal revenues:
  - Tf(R∗1,R∗2) = ˆτf R∗1 + ˆτfx R∗2 − g
  - ˆτf and ˆτfx are effective revenue tax rates on food and export commodity revenues.
  - g are lump sum government transfers received by farmtrepreneurs.

### Model implications and properties
- The model specification ensures farmtrepreneurs face a trade-off between labor-only food production and capital-intensive export production, with capital accumulation constrained by no-borrowing.
- Taxation enters directly on revenue streams R∗1 and R∗2, affecting the farmtrepreneur’s budget constraint and optimal choices.
- Farmtrepreneur decisions feed into market clearing conditions for agricultural commodities and aggregate GDP through their production and export choices.

### Calibration and population shares (relevant context)
- Share of farmtrepreneurs among the rural population is calculated as 2.37 percent.
- Population shares (μ1, μ2, μ3) = (0.33, 0.6463, 0.0237) for urban, rural, and farmtrepreneur groups respectively.
- Average sizes:
  - Informal farms: 1.38 hectares
  - Formal farms: 238 hectares

*Source: wpiea2023090-print-pdf - 1.3    Farmtrepreneurs*

### 3.2    Interest Rate and Debt

### 3.2    Interest Rate and Debt

### Interest Rate and Debt
- Real interest rate set to 1.6 percent, calculated as 13.4 percent average rate on 90 day t-bills and bonds (March 2022) minus annual inflation rate of 11.78 percent (Central Bank of Guinea).
- Domestic government debt set to 15.2 percent of GDP (reported by Guinea’s Ministry of Economy, Finance, and Planning at end-2021).

### Tax System
- Statutory rates (KPMG Fiscal Guide for Guinea, KPMG 2019):
  - Value added (VAT): 18 percent
  - Corporate taxes on non-mining non-oil sectors (CIT): 25 percent
  - Labor income taxes (wage tax + social security, PIT): 29 percent
  - Mining sector statutory corporate tax rate: 30 percent
- Empirical tax collection averages (2015-19):
  - Mining tax collection: 2.16 percent of GDP
  - Indirect tax collection: 6.02 percent of GDP
- Mining tax gap calculation:
  - Assuming mining contributes approximately 15 percent of GDP, expected collection at 30 percent statutory rate = 4.5 percent of GDP.
  - Tax gap = 100(1−2.16/4.5) = 52 percent.
- Indirect (VAT) tax gap calculation:
  - Estimated tax base for indirect taxes = 80 percent of GDP.
  - Statutory VAT revenue = 18 percent × 80 percent of GDP = 14.4 percent of GDP.
  - VAT tax gap = 100(1−6.02/14.4) = 58.2 percent.
- Tax gaps for other taxes approximated as midpoint between mining and VAT gaps = 53.10 percent.

### Preferences
- Period utility function specified as:
  - u(c) = c^{γ1} / f + γ2 log(c_sv) + γ3 log(c_m). (Equation 6 as given)
- Calibration targets from Guinea’s 2018/2019 national household survey (Institut National de la Statistique, 2020, Table 4.26) after manual reclassification:
  - (i) Ratio of food share: Bottom 20 percent vs. Average = 1.05
  - (ii) Share of service expenditure on total expenditure = 0.42
  - (iii) Share of manufacturing expenditure on consumption = 0.12
- Targets represented as (1.05,0.42,0.12).

### Output Shares
- Productivity controls and parameters:
  - ε_rural, ε_urban control informal productivity; ε_1, ε_2 control productivity of agricultural and export commodities for farmtrepreneurs; A_x, A_m control productivity of foreign extractive sector and formal manufacturing.
  - Assumption: ε_rural = ε_1.
- Calibration targets for sectoral output shares (as share of GDP), in order agriculture, services, commodity exports, mining, manufactures:
  - (0.23,0.39,0.008,0.15,0.22).
- Targets obtained by averaging value added by sector from national accounts (2015-2019) and reclassifying into five model categories. Export commodity share adjusted so non-mining imports = 5 percent of total exports.

### Government Consumption and Trade
- Average public consumption (2015-2019) = 17 percent of GDP; assumption that 11 percent of GDP of this is manufactured goods.
- Total investment set at 10 percent of GDP (equal to private and public capital consumption to GDP).
- Assumption: all investment composed of manufactured goods → net imports of manufactured goods = 9 percent of GDP.
- Services:
  - Output = 39 percent of GDP; private consumption = 36 percent of GDP → government consumption of services set to 6 percent of GDP → net imports of services = 0.
  - Net imports of services maintained at 0 throughout analysis.
  - Resulting total government consumption = 17 percent of GDP.
- Mining exports set equal to mining output = 15 percent of GDP.
- Agricultural goods: net imports = 13 percent of GDP.
- Total imports: 9 percent (manufactured) + 13 percent (agricultural) = 22 percent of GDP.
- Total exports: 15 percent (mining) + 0.8 percent (agricultural commodities) = 15.8 percent of GDP → initial permanent current account deficit = 7 percent of GDP.
- To balance long-run current account, government consumption of manufactured goods adjusted from 11 percent of GDP to 4 percent of GDP.

### Inequality Indicators
- Calibration targets for inequality set by adjusting poverty line and shock process (variance, persistence, superstar shock):
  - Poverty rate target = 43.7 percent
  - Gini coefficient target = 50 (noted as above 33.7 reported in WDI)
  - Wealth share of top 10 percent target = 90 percent
- Lack of Guinea wealth data led to guidance from South Africa (Chatterjee, Czajka and Gethin, 2021) for top 10 percent wealth share target.

### Informality
- Economies of scale parameters set arbitrarily:
  - α_rural = α_urban = 0.3
- Goal: approximate total share of work time devoted to informal work = 95 percent (national household survey, Institut National de la Statistique 2019).

### Mining Sector Calibration
- Elasticity of mining output to foreign mining price:
  - Caldara et al. (2016) range: 0.021 to 0.081; top estimate 0.212.
  - Adopt highest estimate of 0.21 for conservative revenue potential.
- This elasticity yields γ_x = 0.1749 (parameter controlling decreasing returns to scale).
- Effective tax rate for mining from statutory 30 percent combined with tax gap 52 percent → effective tax rate = 0.1440 (used to back out Cobb-Douglas share).
- Labor share of mining chosen conservatively = 8 percent (from Feyrer et al. 2017 range 8–24 percent).
- Cobb-Douglas share parameter computed as:
  - α_x = 1 − 0.08 / (γ_x (1−ˆτ)) = 0.466
- Opportunity cost of capital for foreign miners set to 12.8 percent (average return to oil investors from Damodaran ROE by Sector database) used to pin down capital input K*.

### Fixed and Equilibrium Parameters (selected values from tables)
- Pre-determined parameters (selected):
  - Urban Share of Population = 0.33
  - Rural Share of Population = 0.6463
  - Large Farm Population = 0.0237
  - Govt. Service Consumption to GDP = 0.06
  - Price Elasticity of Food Imports = 2.5
  - Baseline Public Infrastructure Stock to GDP = 0.5
  - World Real Interest Rate = 0.016
- Pre-determined tax parameters (selected):
  - Agriculture VAT: Statutory Rate 0.18, Tax Gap 0.582
  - Manufacture VAT: Statutory Rate 0.18, Tax Gap 0.582
  - Food CIT: Statutory Rate 0.25, Tax Gap 0.531
  - Export CIT: Statutory Rate 0.25, Tax Gap 0.531
  - Manufacture CIT: Statutory Rate 0.32, Tax Gap 0.531
  - Labor PIT: Statutory Rate 0.25, Tax Gap 0.531
- Parameters calculated in equilibrium (selected):
  - CES Curvature in Utility of Food = 1.0556
  - Weight of Log Utility of Services = 0.8856
  - Weight of Log Utility of Manufactures = 0.2570
  - Discount Factor = 0.92
  - Idiosyncratic Shock Variance Rural = 0.2666
  - Idiosyncratic Shock Variance Urban = 0.1736
  - Idiosyncratic Shock Persistence (Rural and Urban) = 0.93
  - TFP of Manufacturing = 1.4142
  - TFP of Mining = 0.5573
  - Poverty Line = 0.6444
  - World Price of Food = 0.6965
  - GDP = 2.527

### Laffer Curve and Revenue Findings
- Laffer curve for mining is highly sensitive to price elasticity of supply:
  - Using overall elasticity from Caldara et al. (2016) (e.g., 0.081) → top of Laffer curve near effective tax rate of nearly 90 percent.
  - Using top estimate 0.212 → top of Laffer curve at effective tax rate slightly above 70 percent.
- In general equilibrium experiments:
  - Maximum revenue from mining sector model occurs at effective tax rate between 64.4 and 74.4 percent.
  - At these rates, mining sector size declines from 15 percent of GDP to roughly 9 percent of GDP.
  - Steady-state output shrinks by between 5 and 6 percent (effect concentrated in mining sector).
  - Sectoral and inequality indicators remain basically unchanged except mild increase in rural informality and mild drop in food price (2 percent for largest tax increase).
- Baseline reform (setting effective rate at 34.4 percent) achieves roughly half of maximum revenue potential:
  - Additional revenue ≈ 2.6 percent of GDP.
  - Implies steady-state output drop of less than 2 percent.

### Functioning of the Calibrated Economy
- Formal vs informal work allocation driven by ratio of informal goods prices to wage rate.
- Rural households choose among informal agriculture, formal agroindustrial employment, or mining employment; urban households face higher wages and thinner skill distribution tails.
- Formal labor participation:
  - Rural area characterized by relative scarcity of capital → lower wages, higher informality; only highest-ability rural households work formally.
  - Urban area: lower-skill households find formal work profitable due to higher wages relative to informal service prices.
- Heterogeneity in skills → heterogeneity in savings rates and consumption expenditures; precautionary saving motives prominent.
- Poverty measured using food consumption; model uses calibrated poverty line and distribution of food consumption.
- Rural-to-urban mean consumption ratio = 0.2 (Table 4); majority of poor households are rural with deeper poverty depth.
- Aggregate saving regulated by discount factor β and interest rate r (assumed fixed); savings channelled to national productive enterprises, government loans, and foreign assets.
- Social transfers to poor households would disproportionately increase food demand.

*Italic: Content derived from wpiea2023090-print-pdf - 3.2    Interest Rate and Debt*

### 4.1    Human Capital Formation

### 4.1    Human Capital Formation

### Definition and per-year return to education
- Return to an annual expenditure e is defined as: 1 + θ ≡ w0 ̄z1(e) − w0 ̄z(0) / e.
- Relationship between one-year investment and skill: z1 = z0 + (1 + θ) / w0 e.
- Baseline lifetime rate of return of expenditures specified as θ = 0.13.
- Notes on empirical context:
  - θ = 0.13 is taken from Garcia, Heckman and Laef (2016) for best targeted resources using best practice interventions in the U.S. assuming constant wages.
  - Other empirical findings: higher rates for small programs in Guinea Bassu (Fazzio et al. 2020); Psacharopoulos and Patrinos (2018) find approximately 9.3 percent return per additional year of schooling in developing countries.

### Multi-year investment per student and cost accounting
- Multi-year (t years) investment in a single student, abstracting from compounding and assuming constant per-year return:
  - z1 = z0 + t (1 + θ) / w0 e, where e is the annual investment and θ is the per-year return over t years.
- Total annual cost in a region with cohort size S: E = S t e.
  - Interpretation: S = number of students in each grade, t = number of grades covered, e = expenditure per student per grade.

### Rural–urban allocation rule
- Equal skill impact per student per grade across locations imposed by: e_rural / w_rural = e_urban / w_urban.
- Given aggregate education expenditure E per year, per-student expenditures are:
  - e_rural = w_r E / S t (w_u μ_u + w_r μ_r)
  - e_urban = w_u E / S t (w_u μ_u + w_r μ_r)

### Parameterization for Guinea and policy design
- Empirical cohort sizing:
  - Share of population ages 0–14 in Guinea fluctuated between 0.4 and 0.47 since 1960 (WB WDI).
  - Implies size of an age cohort ≈ 47/15 = 3.13 percent of the population.
- Focus on early and primary education: set t = 10.

### Distributional impact of identical returns (Figure 4)
- Because the reform yields the same return for all households, proportional impact is larger for low-skilled households, enabling substantial improvements at the bottom of the income distribution.

### Long-run economy reaction to increasing education investment
- Simulation setup: increase investment in education in steps from zero to 2 percent of GDP and recalculate equilibrium.
- Key macro outcomes:
  - At 2 percent investment the government’s long run surplus is 0.52 percent of GDP, down from an initial level of 2.57 percent of GDP.
  - Assuming total impact of the reform is spread over 30 years:
    - Annual educational investments of 0.5 percent of GDP can generate 0.41 percent annual growth over 30 years.
    - Each additional 0.5 percent of GDP invested in education initially boosts the annual growth rate by roughly 40 basis points, with diminishing marginal effects thereafter.
- Inequality and poverty:
  - Increasing investment in education to 2 percent of GDP can reduce the poverty rate to even 11 percent in the long run while the Gini drops from 0.55 to 0.47.
  - Rural–urban income gap remains near its initial level.
- Formalization:
  - Urban formalization increases while rural formalization falls under higher education investment.
  - Urban effect: public service and manufacturing absorb additional labor as wages fall and government enterprise capital rises.
  - Rural effect: increased demand for food, without matching productivity gains in formal food producers, leads to higher food prices, larger informality, and larger informal food production.
- Mechanism insight:
  - In the multisector model, additional educational investment can decrease the average return to education in the urban area while increasing it in the rural area, because relative price movements across sectors can offset wage movements expected in one-sector models.

*Italic: Source — 4.1 Human Capital Formation, wpiea2023090-print-pdf*

### 5.1    Essential Dynamics

### 5.1    Essential Dynamics

### Driving processes and stock dynamics
- The two driving processes in the economy are:
  - the expansion of labor supply through the accumulation of skills that follow the educational reform; and
  - the increase of productivity that comes from the accumulation of infrastructure capital.
- Calculations (presented in Figure 9) are performed by calculating the stock of educated workers and the stocks of infrastructure over time.
- These stocks increase because of the annual investments imposed in the analysis.

### Timing and growth impacts
- Reforms to education have a quicker turnaround than those in infrastructure, due to:
  - the gradual accumulation of infrastructure as compared to the universal, albeit limited, educational reform.
- The impact on the annual growth rate, considering the particular policy analysis, is larger for education, but decays sooner.
- The combined policy interaction: education and infrastructure are complementary in generating additional economic activity; the combined policy is more powerful in terms of real GDP than the sum of the policies applied separately to the model.

### Inequality transition
- Uncovering the nature of the transition for inequality variables is beyond the scope of the calculations because it would require the computation of equilibrium over time, adding substantial complexity.
- Anticipated properties of inequality transitions:
  - The transition for inequality indicators would be monotonic (i.e. free of zigzagging) and gradual.
  - Although there are no theoretical guarantees of this conclusion, existing numerical experiments have resulted in gradual and monotonic transitions for very similar models (see Mendes-Tavares et al. (2021)).

### Key conclusions and policy recommendations
- Main conclusion: mining revenues, if properly collected and invested, could have transformative impacts in Guinea.
- Policies that combine spending on education with infrastructure investment and social transfers have the most balanced impact on GDP, public revenue, poverty, and other measures of inequality.
- Specific findings:
  - First: education and infrastructure are complementary; the combined policy produces more real GDP than the sum of separate policies.
  - Second: infrastructure-only policies of the magnitude studied do not produce substantial improvements in poverty reduction and inequality relief, but can be highly effective when combined with education policy.
  - Third: the self-financing nature of the policies allows distribution of any extra revenue as a social transfer.
    - In the experiment a flat transfer (not targeted) is assumed that can be delivered without leakage.
    - These transfers are highly effective in providing further reductions in poverty and inequality.
    - Well-designed targeted transfers, if available, could be even better.
- Policy implication: policymakers must consider tradeoffs between investing resources generated by mining revenues only into “hard” infrastructure (i.e. roads) versus combining investments in hard infrastructure with social infrastructure that builds human capital, which could have larger long-term impacts on growth, revenue, poverty and inequality.

### Relevant human capital and health indicators cited
- Guinea consistently ranks among the 12 countries with the greatest human capital challenges worldwide (World Bank’s Human Capital Index).
- Guinea ranks the seventh lowest worldwide in human capital outcomes for women.
- Health indicators:
  - Ten in 100 children will die before his or her fifth birthday, the fifth lowest child survival rate worldwide.
  - Guinea’s high under-five mortality rate outpaces the averages for Sub-Saharan Africa and low-income countries.
  - The 2014-2016 Ebola outbreak led to a permanent decline in certain child immunization rates, which likely worsened during the COVID-19 pandemic.
- Education indicators:
  - Children complete an average of 7 years of education, but the suboptimal quality of education results in learning outcomes in line with receiving only 4.6 years of high-quality education.
  - Guinea has made progress in increasing primary school enrollment since 2000, but this progress has flattened since 2016 and worsened across gender dimensions.

*Source: 5.1 Essential Dynamics (from wpiea2023090-print-pdf - 5.1    Essential Dynamics)*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023090-print-pdf.pdf_