## MOZAMBIQUE’S NATURAL GAS RESOURCES: TRADEOFFS AND OPPORTUNITIES

## Source details

**Canonical URL:** [MOZAMBIQUE’S NATURAL GAS RESOURCES: TRADEOFFS AND OPPORTUNITIES](https://www.imf.org/-/media/files/publications/cr/2019/1mozea2019003.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2019/1mozea2019003.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2019/1mozea2019003.pdf.json)

---

### Background
- Around 150 trillion cubic feet (Tcf) of proven natural gas reserves in Area 1 and Area 4 in the offshore Rovuma Basin (discovered in 2010).
- Mozambique ranking: third largest holder of natural gas reserves in Africa (after Nigeria and Algeria) and twelfth globally.
- Global context: as of January 1, 2017, global natural gas reserves were estimated at 6,923 Tcf; Russia, Iran, and Qatar accounting for around 50 percent of the total.
- LNG trade growth: in 2018, global trade in LNG increased by 3.2 billion cubic feet per day (Bcf/d) to 41.3 Bcf/d, an 8 percent increase from the previous year.
- LNG infrastructure and investment requirements:
  - Developing liquefaction plants (trains) takes around 4-5 years and involves large investments.
  - The five trains being developed in Mozambique will require total investments of around US$55 billion (the equivalent to four-times Mozambique’s 2018 GDP).
  - Total planned capacity from the five trains: around 30 million tons per annum (MTPA) (2.4 Tcf) — equivalent to 17 percent of total 2017 LNG trade or 7.5 percent of projected LNG trade by 2026.
  - Initial production expected in late 2023; full capacity by 2026.
- Project-level highlights:
  - Offshore Area 1 Golfinho/Atum: two onshore trains, 6.44 MTPA per train, total investment US$22 billion. SPAs secured: 9.5 MTPA. FID expected by mid-2019; first train production expected by end-2024 and second by mid-2025.
  - Area 4 Coral South: offshore floating train, 3.4 MTPA, total investment around US$8 billion. FID reached in June 2017. Production expected by late 2023; entire LNG production sold to BP over 20 years.
  - Area 4 Mamba: two onshore trains, 7.6 MTPA per train, total investment estimated at around US$25 billion. FID planned by mid-2019; first train production expected by August 2024 and second by mid-2025.
  - Additional potential: investments could add another 27 MTPA by 2032 (12 MTPA from Area 1 Prosperidade field and 15.2 MTPA from overlapping fields between Area 1 and Area 4).

### Macroeconomic Impact of Natural Gas Production (FARI model outputs)
- Methodology: Fiscal Analysis of Resource Industries (FARI) model aggregates project-level information to derive resource revenues, production value added and exports.
- Real GDP:
  - Projected to increase in 2023-24 with onset of LNG production.
  - Non-LNG economic activity conservatively assumed to grow at 4 percent per year over the long-term.
- Fiscal:
  - LNG fiscal revenue starts to flow by 2023.
  - Primary fiscal balance would improve and turn into surplus reaching around 13¼ percent of GDP by 2038, under the assumption that (i) recovery of LNG development costs occurs over initial four-to-six years of production, (ii) LNG fiscal revenue would account for almost half of total fiscal revenue, and (iii) all LNG fiscal revenue would be saved.
- External sector:
  - As LNG exports pick up, current account deficits are projected to turn into surpluses by 2027.

### The DIGNAR Model and Public Investment Approaches
- Model structure:
  - Three production sectors: an exogenous LNG production sector and firms producing tradeables and non-tradeables via a Cobb-Douglas production function combining private and public capital and labor.
  - Two household types: optimizing (use financial markets to smooth consumption) and non-optimizing (no access or do not use markets to smooth consumption).
  - Public sector raises taxes from private sector including LNG sector to finance recurrent expenditures, infrastructure investment, and service debt.
- Two public investment approaches analyzed:
  - Spend-As-You-Go (SAYG):
    - LNG fiscal revenues in each period fund public infrastructure investment, with recurrent spending fixed.
    - No accumulation in a stabilization fund; taxes adjust endogenously to keep debt sustainable.
    - Highly procyclical: public investment and macro variables follow LNG fiscal revenue dynamics.
  - Delinked:
    - Public investment is scaled up gradually and determined exogenously (delinked from LNG revenues).
    - Investment financed by combination of LNG fiscal revenues, debt issuance and non-LNG fiscal revenues while allowing buildup of savings in a stabilization fund.
    - Stabilization fund drawn down in revenue shortfalls to maintain planned investment path.
    - Gradual scaling up allows time to improve absorptive capacity and public investment efficiency and to build fiscal buffers.

### LNG Production and Price Scenarios
- LNG production path: follows FARI model; assumed to start in 2023 and peak in 2032 based on production from a total of ten LNG trains.
- Price scenarios:
  - Baseline LNG prices generated by random simulations ranging between US$2 to US$9 per million cubic feet (MCF), reflecting observed 20-year price range and capturing volatility.
  - Adverse scenario: a 20 percent negative LNG price shock applied to baseline scenario prices.
  - Justification for 20 percent shock: in the last 45 years there were 16 negative variations of LNG prices (36% of total observations), of which 11 (69% of total negative variations) were up to 20 percent.

### Results and Key Findings
- General result: prudent and gradual investment scaling-up is preferable to aggressive, front-loaded investments because of absorptive capacity constraints and private sector crowding-out effects.
- Stabilization fund benefits:
  - Allows accumulation of buffers to smooth macroeconomic volatility from volatile and potentially declining LNG production and prices.
  - Under the Delinked approach, gradual investment scaling-up enables buildup of fiscal buffers through a stabilization fund; under SAYG no such savings accumulate, making the economy more susceptible to volatility.
- Investment efficiency under SAYG versus Delinked:
  - Under SAYG, public investment follows a volatile trend as it is linked to LNG revenues.
  - Aggressive investment scaling up under SAYG leads to a huge decline in investment efficiency relative to the Delinked approach, mainly due to absorptive capacity constraints (lack of planning and coordination, and lower capital budget execution ratios).
  - Under an adverse scenario, public investment efficiency improves in relative terms under SAYG, but inefficiencies remain higher than in the Delinked approach.
  - In a negative price shock:
    - Under the Delinked approach, the government would be able to maintain the same public investment path because this path is delinked from LNG revenues.
    - Under SAYG, the government would be forced to implement public investment cuts as LNG revenues decline.
    - Because there are no accumulated savings under SAYG, the government must resort to additional taxes and/or debt financing to satisfy its budget constraint.
- Effects on private investment, consumption, non-LNG output, and taxes:
  - Higher LNG revenues lead to other taxes being lowered, which stimulates private investment, private consumption, and non-LNG output.
  - Higher LNG revenues and higher GDP (taxation base) result in lower tax rates needed to satisfy the budget constraint.
  - Over the medium term, lower tax rates stimulate private consumption, private investment and non-LNG output.
  - Under SAYG: the tax rate is initially lower, but over the long term, as LNG revenues decline, other taxes need to increase to satisfy the budget constraint. As a result, private investment, private consumption and non-LNG output become higher under the Delinked approach over the long term.
- Real exchange rate and Dutch disease pressures:
  - LNG production and exports will lead to real exchange rate appreciation.
  - Appreciation is more pronounced under SAYG because all foreign exchange proceeds from LNG exports are immediately channeled to the economy.
  - Under the Delinked approach, accumulation of resources in a stabilization fund mitigates Dutch disease effects by containing, to some extent, real exchange rate appreciation pressures.
- Public debt and debt sustainability:
  - LNG fiscal revenues can contribute to a more sustainable debt path.
  - As LNG fiscal revenues increase, the government needs to resort less to debt accumulation, leading to a decline of public debt as a share of GDP over the medium term.
  - In the long term, due to decreasing LNG fiscal revenues, public debt-to-GDP ratios tend to increase as additional borrowing would be required.
  - Under the Delinked approach, in both the baseline and adverse scenarios: debt levels are lower than under the SAYG approach and the government’s ability to service its debt obligations is relatively high due to accumulated savings.
- Main synthesized findings:
  - Mozambique is poised to become one of the world’s largest LNG exporters over the medium term.
  - LNG exports will generate significant fiscal revenues that the government can use to address infrastructure gaps and other social needs, fostering economic development and significantly reducing poverty.
  - Volatility in LNG fiscal revenues and absorptive capacity constraints create challenges to macroeconomic management, requiring the government to find the right balance between public investment, investment efficiency and macroeconomic stability.
  - The DIGNAR model assessment shows that gradually scaling up investment gives Mozambique time to improve absorptive capacity and public investment efficiency while building fiscal buffers to prevent disruptions when a negative LNG price shock occurs (accumulation of savings in an actual or virtual stabilization fund would prevent the need for sizable investment cuts).
  - Gradual scaling up contains macroeconomic volatility, including to output and real exchange rate appreciation pressures, and is therefore more conducive to private sector led economic diversification.

### Calibration: selected parameters and assumptions
- Long-run GDP growth rate (in percent): 6.00
- Exports/GDP*100 (in percent): 33.00
- Imports/GDP*100 (in percent): 65.00
- Government consumption/GDP*100 (in percent): 21.00
- Government investment expenditures/GDP*100 (in percent): 7.50
- Private investment/GDP*100 (in percent): 17.00
- Mining value added (natural resource production)/GDP*100 (in percent): 3.00
- Government wealth fund/GDP*100 (in percent) ' (external savings): 1.00
- Share of tradables in government expenditures (in percent): 60.00
- Share of tradables in private consumption (in percent): 60.00
- Government domestic debt / GDP*100 (in percent): 8.81
- Private foreign debt/GDP*100 (in percent): 49.18
- Concessional debt/GDP*100 (in percent): 4.71
- Government external commercial debt/GDP*100 (in percent): 48.13
- Grants/GDP*100 (in percent): 5.09
- Annualized domestic net real interest rate: 10.00
- Annualized foreign net real interest rate earned by the stabilization fund: 2.70
- Annualized net real interest rate paid on concessional debt: 0.00
- Annualized net real risk-free rate: 2.40
- Annualized net real interest rate paid on government external commercial debt: 6.00
- Labor income share in non-traded sector (in percent): 45.00
- Labor income share in traded sector (in percent): 60.00
- Elasticity of output wrt public capital: 0.25
- Capital depreciation rate in non-tradable sector (in percent): 10.00
- Capital depreciation rate in traded sector (in percent): 10.00
- Depreciation rate of public capital (in percent): 7.00
- Learning by doing externality in the traded sector: 0.10
- Persistence in TFP in traded sector: 0.10
- Investment adjustment cost parameter in the non-traded sector: 25.00
- Investment adjustment cost parameter in the traded sector: 25.00
- Steady-state efficiency of public investment (share of investment turned into actual capital) (in percent): 50.00
- Inverse of the Frisch elasticity of labor supply for optimizers: 10.00
- Inverse of the Frisch elasticity of labor supply for rule of thumb consumers: 10.00
- Inverse of the intertemporal elasticity of consumption: 2.00
- Elasticity of substitution between the two types of labor (in tradables and nontradables): 1.00
- Measure of optimizers in the economy in percent (non-credit-constrained households): 0.60
- Elasticity of substitution between traded and non-traded goods: 0.44
- Home bias for additional government spending: 0.50
- Elasticity of portfolio adiustment costs: 0.00
- Royalty tax rate on natural resources (in percent): 20.00
- User fees of public infrastructure (in percent of recurrent costs): 50.00
- Labor income tax rate: 2.51
- Consumption tax rate: 7.90
- Tax rate on the return on capital: 5.86
- Elasticity of sovereign risk: 0.00
- Severity of public capital depreciation when not maintained: 1.00
- Severity of absorptive capacity constraints: 20.00
- Thresholds of investment scaling up beyond which absorptive capacity constraints start binding (in percent): 50.00
- Persistence of efficiency of public investment: 0.80
- Floor for the sovereign wealth fund (in percent of GDP): 0.00

### Neutral real interest rate (NRIR) estimates for Mozambique — methods and key results
- NRIR estimates range between 4.6 percent and 7.8 percent.
- Consumption-based CAPM approach (no habit and with habit persistence):
  - NRIR with CRRA preferences (no habit): between 6.4 percent and 10.7 percent. Mean = 8.6 percent.
  - NRIR with habit formation: between 4.9 percent and 7.8 percent. Mean = 6.4 percent.
  - Authors select results with habit formation as more plausible for Mozambique.
- Uncovered Interest Parity (UIP) approach:
  - Calibration inputs: r* = 0.6; RER̂ = 1.7; ρ = [4.0, 5.3].
  - Derived NRIR r = [6.3, 7.6] (Percent).
- Dynamic approaches — Statistical filters:
  - Filters applied: U.S. Census, Hodrick-Prescott, Baxter-King, Christiano-Fitzgerald (symmetric and asymmetric).
  - NRIR estimates across filters fall in range 4.6 percent to 5.6 percent. Average across filters = 4.9 percent.
- Savings-Investment semi-structural macroeconomic model (Laubach and Williams style):
  - Quarterly data 2008–2018; controls include HP-filter deviations of real exchange rate, oil-price inflation, commodity prices.
  - Results: NRIR around 4.7 percent (average for 2012-18). NRIR of 4.9 percent over the projection period.
  - Time profile: increase from 4 percent (2012) to 5.2 percent (mid-2014); decline from 2016 to Q1-2017 to 4 percent; recovery thereafter to average level.
- Main synthesis across methods:
  - NRIR clustered within a relatively narrow band: 4.6 percent to 7.8 percent.
  - Assuming inflation expectations anchored around 5.5 percent, corresponding neutral nominal interest rate (NNIR) range = 10.1 percent to 13.2 percent.
  - Method-specific central estimates:
    - Consumption-smoothing model with habit formation: mean NRIR = 6.4 percent; range [4.9 percent, 7.8 percent].
    - Uncovered interest parity: NRIR = [6.3 percent, 7.6 percent].
    - Statistical filters average: 4.9 percent (range 4.6 percent to 5.6 percent).
    - Laubach–Williams semi-structural model: average NRIR 2012–18 = 4.7 percent; projection period mean = 4.9 percent.
  - Key interpretations:
    - Width of range reflects methodological uncertainty and the roles of country risk premium and international financial conditions.
    - Structural breaks (e.g., April 2017 regime shift) complicate estimation and may affect the signaling properties of standing lending rate as proxy for policy rate.
    - NRIR is time-varying for Mozambique, exhibiting an increase in 2012–2014 and a decline around 2016–2017 before recovering.
    - Current assessment: Mozambique’s monetary policy stance is somewhat tight and in a process of normalization amid a negative output gap, expected single-digit inflation, and domestic and external risks.

*Source: Mozambican authorities and IMF staff estimates and projections; Author’s calculations.*

### References ____________________________________________________________________________ 17

### MOZAMBIQUE’S NATURAL GAS RESOURCES: TRADEOFFS AND OPPORTUNITIES

### Background
- Significant natural gas resources discovered in Mozambique in 2010: around 150 trillion cubic feet (Tcf) of proven natural gas reserves in Area 1 and Area 4 in the offshore Rovuma Basin.
- Mozambique ranking: third largest holder of natural gas reserves in Africa (after Nigeria and Algeria) and twelfth globally.
- Global context: as of January 1, 2017, global natural gas reserves were estimated at 6,923 Tcf, with Russia, Iran, and Qatar accounting for around 50 percent of the total.
- LNG trade growth: in 2018, global trade in LNG increased by 3.2 billion cubic feet per day (Bcf/d) to 41.3 Bcf/d, an 8 percent increase from the previous year.
- LNG infrastructure and investment requirements:
  - Developing liquefaction plants (trains) takes around 4-5 years and involves large investments.
  - The five trains being developed in Mozambique will require total investments of around US$55 billion (the equivalent to four-times Mozambique’s 2018 GDP).
  - Total planned capacity from the five trains: around 30 million tons per annum (MTPA) (2.4 Tcf) — equivalent to 17 percent of total 2017 LNG trade or 7.5 percent of projected LNG trade by 2026.
  - Initial production expected in late 2023; full capacity by 2026.
- Project-level highlights:
  - Offshore Area 1 Golfinho/Atum: two onshore trains, 6.44 MTPA per train, total investment US$22 billion. SPAs secured: 9.5 MTPA. FID expected by mid-2019; first train production expected by end-2024 and second by mid-2025.
  - Area 4 Coral South: offshore floating train, 3.4 MTPA, total investment around US$8 billion. FID reached in June 2017. Production expected by late 2023; entire LNG production sold to BP over 20 years.
  - Area 4 Mamba: two onshore trains, 7.6 MTPA per train, total investment estimated at around US$25 billion. FID planned by mid-2019; first train production expected by August 2024 and second by mid-2025.
  - Additional potential: investments could add another 27 MTPA by 2032 (12 MTPA from Area 1 Prosperidade field and 15.2 MTPA from overlapping fields between Area 1 and Area 4).

### Macroeconomic Impact of Natural Gas Production (FARI model outputs)
- Methodology: Fiscal Analysis of Resource Industries (FARI) model aggregates project-level information to derive resource revenues, production value added and exports.
- Real GDP:
  - Projected to increase in 2023-24 with onset of LNG production.
  - Non-LNG economic activity conservatively assumed to grow at 4 percent per year over the long-term.
- Fiscal:
  - LNG fiscal revenue starts to flow by 2023.
  - Primary fiscal balance would improve and turn into surplus reaching around 13¼ percent of GDP by 2038, under the assumption that (i) recovery of LNG development costs occurs over initial four-to-six years of production, (ii) LNG fiscal revenue would account for almost half of total fiscal revenue, and (iii) all LNG fiscal revenue would be saved.
- External sector:
  - As LNG exports pick up, current account deficits are projected to turn into surpluses by 2027.

### The DIGNAR Model and Public Investment Approaches
- Model structure:
  - Three production sectors: an exogenous LNG production sector and firms producing tradeables and non-tradeables via a Cobb-Douglas production function combining private and public capital and labor.
  - Two household types: optimizing (use financial markets to smooth consumption) and non-optimizing (no access or do not use markets to smooth consumption).
  - Public sector raises taxes from private sector including LNG sector to finance recurrent expenditures, infrastructure investment, and service debt.
- Two public investment approaches analyzed:
  - Spend-As-You-Go (SAYG):
    - LNG fiscal revenues in each period fund public infrastructure investment, with recurrent spending fixed.
    - No accumulation in a stabilization fund; taxes adjust endogenously to keep debt sustainable.
    - Highly procyclical: public investment and macro variables follow LNG fiscal revenue dynamics.
  - Delinked:
    - Public investment is scaled up gradually and determined exogenously (delinked from LNG revenues).
    - Investment financed by combination of LNG fiscal revenues, debt issuance and non-LNG fiscal revenues while allowing buildup of savings in a stabilization fund.
    - Stabilization fund drawn down in revenue shortfalls to maintain planned investment path.
    - Gradual scaling up allows time to improve absorptive capacity and public investment efficiency and to build fiscal buffers.

### LNG Production and Price Scenarios
- LNG production path: follows FARI model; assumed to start in 2023 and peak in 2032 based on production from a total of ten LNG trains.
- Price scenarios:
  - Baseline LNG prices generated by random simulations ranging between US$2 to US$9 per million cubic feet (MCF), reflecting observed 20-year price range and capturing volatility.
  - Adverse scenario: a 20 percent negative LNG price shock applied to baseline scenario prices.
  - Justification for 20 percent shock: in the last 45 years there were 16 negative variations of LNG prices (36% of total observations), of which 11 (69% of total negative variations) were up to 20 percent.

### Results and Key Findings
- General result: prudent and gradual investment scaling-up is preferable to aggressive, front-loaded investments because of absorptive capacity constraints and private sector crowding-out effects.
- Stabilization fund benefits:
  - Allows accumulation of buffers to smooth macroeconomic volatility from volatile and potentially declining LNG production and prices.
  - Under the Delinked approach, gradual investment scaling-up enables buildup of fiscal buffers through a stabilization fund; under SAYG no such savings accumulate, making the economy more susceptible to volatility.
- Macroeconomic responses (qualitative from figures and model calibration):
  - SAYG is more susceptible to macroeconomic volatility driven by LNG fiscal revenue variability.
  - Delinked approach supports maintaining investment and smoothing shocks through fund drawdowns and buildup, reducing the likelihood of boom-bust cycles and Dutch disease pressures when combined with external savings.

*Sources: Mozambican authorities and IMF staff estimates and projections.*

### 13. Excessively quick public investment scaling up would lead to investment

### 13. Excessively quick public investment scaling up would lead to investment inefficiencies

### Investment efficiency under SAYG versus Delinked (gradual) scaling up
- Under SAYG, public investment follows a volatile trend as it is linked to LNG revenues.
- Aggressive investment scaling up under SAYG leads to a huge decline in investment efficiency relative to the Delinked approach, mainly due to absorptive capacity constraints (lack of planning and coordination, and lower capital budget execution ratios).
- Under an adverse scenario, public investment efficiency improves in relative terms under SAYG, but inefficiencies remain higher than in the Delinked approach.
- In a negative price shock:
  - Under the Delinked (gradual scaling up) approach, the government would be able to maintain the same public investment path because this path is delinked from LNG revenues.
  - Under SAYG, the government would be forced to implement public investment cuts as LNG revenues decline.
  - Because there are no accumulated savings under SAYG, the government must resort to additional taxes and/or debt financing to satisfy its budget constraint.

### Effects on private investment, consumption, non-LNG output, and taxes
- Higher LNG revenues lead to other taxes being lowered, which stimulates:
  - private investment,
  - private consumption,
  - non-LNG output.
- Higher LNG revenues and higher GDP (taxation base) result in lower tax rates needed to satisfy the budget constraint.
- Over the medium term, lower tax rates stimulate private consumption, private investment and non-LNG output.
- Under SAYG:
  - The tax rate is initially lower, but over the long term, as LNG revenues decline, other taxes need to increase to satisfy the budget constraint.
  - As a result, private investment, private consumption and non-LNG output become higher under the Delinked approach over the long term.

### Real exchange rate and Dutch disease pressures
- LNG production and exports will lead to real exchange rate appreciation.
- Appreciation is more pronounced under SAYG because all foreign exchange proceeds from LNG exports are immediately channeled to the economy.
- Under the Delinked approach, accumulation of resources in a stabilization fund mitigates Dutch disease effects by containing, to some extent, real exchange rate appreciation pressures.

### Public debt and debt sustainability
- LNG fiscal revenues can contribute to a more sustainable debt path.
- As LNG fiscal revenues increase, the government needs to resort less to debt accumulation, leading to a decline of public debt as a share of GDP over the medium term.
- In the long term, due to decreasing LNG fiscal revenues, public debt-to-GDP ratios tend to increase as additional borrowing would be required.
- Under the Delinked approach, in both the baseline and adverse scenarios:
  - Debt levels are lower than under the SAYG approach.
  - The government’s ability to service its debt obligations is relatively high due to accumulated savings.

### Main findings (summarized)
- Mozambique is poised to become one of the world’s largest LNG exporters over the medium term.
- LNG exports will generate significant fiscal revenues that the government can use to address infrastructure gaps and other social needs, fostering economic development and significantly reducing poverty.
- Volatility in LNG fiscal revenues and absorptive capacity constraints create challenges to macroeconomic management, requiring the government to find the right balance between public investment, investment efficiency and macroeconomic stability.
- The DIGNAR model assessment shows:
  - Gradually scaling up investment gives Mozambique time to improve absorptive capacity and public investment efficiency while building fiscal buffers to prevent disruptions when a negative LNG price shock occurs (accumulation of savings in an actual or virtual stabilization fund would prevent the need for sizable investment cuts).
  - Gradual scaling up contains macroeconomic volatility, including to output and real exchange rate appreciation pressures.
  - Gradual scaling up is therefore more conducive to private sector led economic diversification.

### Calibration: selected parameters and assumptions (as presented)
- Long-run GDP growth rate (in percent): 6.00
- Exports/GDP*100 (in percent): 33.00
- Imports/GDP*100 (in percent): 65.00
- Government consumption/GDP*100 (in percent): 21.00
- Government investment expenditures/GDP*100 (in percent): 7.50
- Private investment/GDP*100 (in percent): 17.00
- Mining value added (natural resource production)/GDP*100 (in percent): 3.00
- Government wealth fund/GDP*100 (in percent) ' (external savings): 1.00
- Share of tradables in government expenditures (in percent): 60.00
- Share of tradables in private consumption (in percent): 60.00
- Government domestic debt / GDP*100 (in percent): 8.81
- Private foreign debt/GDP*100 (in percent): 49.18
- Concessional debt/GDP*100 (in percent): 4.71
- Government external commercial debt/GDP*100 (in percent): 48.13
- Grants/GDP*100 (in percent): 5.09
- Annualized domestic net real interest rate: 10.00
- Annualized foreign net real interest rate earned by the stabilization fund: 2.70
- Annualized net real interest rate paid on concessional debt: 0.00
- Annualized net real risk-free rate: 2.40
- Annualized net real interest rate paid on government external commercial debt: 6.00
- Labor income share in non-traded sector (in percent): 45.00
- Labor income share in traded sector (in percent): 60.00
- Elasticity of output wrt public capital: 0.25
- Capital depreciation rate in non-tradable sector (in percent): 10.00
- Capital depreciation rate in traded sector (in percent): 10.00
- Depreciation rate of public capital (in percent): 7.00
- Learning by doing externality in the traded sector: 0.10
- Persistence in TFP in traded sector: 0.10
- Investment adjustment cost parameter in the non-traded sector: 25.00
- Investment adjustment cost parameter in the traded sector: 25.00
- Steady-state efficiency of public investment (share of investment turned into actual capital) (in percent): 50.00
- Inverse of the Frisch elasticity of labor supply for optimizers: 10.00
- Inverse of the Frisch elasticity of labor supply for rule of thumb consumers: 10.00
- Inverse of the intertemporal elasticity of consumption: 2.00
- Elasticity of substitution between the two types of labor (in tradables and nontradables): 1.00
- Measure of optimizers in the economy in percent (non-credit-constrained households): 0.60
- Elasticity of substitution between traded and non-traded goods: 0.44
- Home bias for additional government spending: 0.50
- Elasticity of portfolio adiustment costs: 0.00
- Royalty tax rate on natural resources (in percent): 20.00
- User fees of public infrastructure (in percent of recurrent costs): 50.00
- Labor income tax rate: 2.51
- Consumption tax rate: 7.90
- Tax rate on the return on capital: 5.86
- Elasticity of sovereign risk: 0.00
- Severity of public capital depreciation when not maintained: 1.00
- Severity of absorptive capacity constraints: 20.00
- Thresholds of investment scaling up beyond which absorptive capacity constraints start binding (in percent): 50.00
- Persistence of efficiency of public investment: 0.80
- Floor for the sovereign wealth fund (in percent of GDP): 0.00

### Neutral real interest rate (NRIR) estimates for Mozambique (summary)
- The NRIR estimates range between 4.6 percent and 7.8 percent.
- Interpretation: Given current and expected inflation for Mozambique, these results indicate that monetary policy is still somewhat tight and thus there is room to continue with the ongoing process of normalization.

*Source: IMF staff calculations; Republic of Mozambique, INTERNATIONAL MONETARY FUND.*

### 7. This approach uses the consumption-based capital asset pricing model (CAPM). We use

### 1mozea2019003 - 7. This approach uses the consumption-based capital asset pricing model (CAPM). We use

### Consumption-based CAPM approach (no habit and with habit persistence)
- Framework:
  - Follows Cochrane (2001) and Campbell and Cochrane (1999) to compute the NRIR by solving the Euler equation for reasonable parameter values. Introduction of habit persistence allows for lower yield estimates that accommodate higher risk premia.
  - Endowed economy populated by representative agents maximizing intertemporal utility with access to an asset yielding a gross real return that smooths consumption.
- Assumptions:
  - Constant risk aversion utility function (CRRA).
  - Aggregate consumption equals the endowment; growth rate is log-normal distributed.
- Equilibrium real interest rate (NRIR) without habit persistence (from Euler equation):
  - lnR_t = r_t = −lnβ + γ E_t ∆ln y_{t+1} − ( (γ^2 / 2) ) Var_t(∆ln y_{t+1})    (equation (1))
  - Notation: β is subjective discount/time preference factor; γ is coefficient of relative risk aversion; ∆ is the difference operator; VAR(.) is variance operator; y_t is endowment at time t proxied by output; r_t is equilibrium real interest rate (NRIR).
- Habit persistence specification:
  - Consumption above habit level; agent risk aversion varies with consumption relative to habit x_t.
  - Habit evolves slowly: x_t = λ ∑_{j=0}^{∞} φ c_{t−j}.
  - Surplus consumption ratio follows AR(1) dynamics.
  - NRIR with habit formation:
    - lnR_t = r_t = −lnβ + γ E_t ∆ln y_{t+1} − (1/2) γ (1−φ)    (equation (2))
    - First two terms same as equation (1); last term related to precautionary savings implying lower equilibrium real interest rate the more volatile is income.

### Calibration for Mozambique (consumption-smoothing models)
- Medium-term inputs:
  - Median potential per capita output growth rate for 2004-15: 4.37 percent.
  - Standard deviation: 1.05 percent.
- Habit formation calibration:
  - φ calibrated following Fuentes and Gredig (2007, p.5).
  - Example calibration: using potential per capita output growth of 4.37 percent, a real interest rate of 6.39 percent, with β=0.978 and γ=1.5, yields φ=0.973 (assuming zero output gap and interest rate close to neutral in 2013).
- NRIR estimates (Table 1 / Table 2 summary):
  - With CRRA preferences (no habit): NRIR between 6.4 percent and 10.7 percent for the range of discount factors and risk aversion coefficients. Mean = 8.6 percent.
  - With habit formation: NRIR between 4.9 percent and 7.8 percent. Mean = 6.4 percent.
  - Authors select results with habit formation as more plausible for Mozambique.

### Uncovered Interest Parity (UIP) approach
- Real term UIP expression after substitution:
  - r = i − π = i* − π* + RER̂ + ρ = r* + RER̂ + ρ    (equation (3))
  - Where r* is foreign real rate; RER̂ is expected real effective exchange rate depreciation; ρ is country-specific risk premium.
- Calibration for Mozambique:
  - r* taken as average NRIR for U.S. estimated by Holston, Laubach and Williams (2017) for 2008-18.
  - RER̂ uses average real effective exchange rate depreciation from BM historical series for 2011-18.
  - ρ uses alternatives including Mozambican Association of Banks (AMB) country risk component and Damodaran computations across comparators.
- UIP-based results (Table 3):
  - r* = 0.6 (Value; comment: Holston, Laubach and Williams (2018))
  - RER̂ = 1.7 (Average real effective exchange rate depreciation)
  - ρ = [4.0, 5.3] (Average of high and low estimates from Mozambican Banking Association (ABM), and Damodaran (2018))
  - Derived NRIR r = [6.3, 7.6] (Percent)

### Dynamic approaches — Statistical filters
- Methods applied:
  - U.S. Census, Hodrick-Prescott, Baxter-King, Christiano-Fitzgerald (symmetric and asymmetric) filters on policy rate series.
  - Monthly policy rate series deflated by 12-month inflation.
  - Forecast real policy rate over 84 months using Census X-13 and SARIMA to minimize end-of-sample bias.
- Results (Table 4 summary):
  - NRIR estimates across filters fall in range 4.6 percent to 5.6 percent.
  - Average across filters = 4.9 percent.
  - These projections used in dynamic estimation to minimize end-of-sample bias.

### Savings-Investment semi-structural macroeconomic model (Laubach and Williams style)
- Model structure:
  - Backward-looking IS equation relating output gap and NRIR; Phillips curve relating inflation to output gap.
  - Equilibrium: output gap closed, inflation at target, real interest rate equals NRIR.
- IS equation (output gap representation):
  - (y_t − y_t^*) = ∑ α_s^y (y_{t−s} − y_{t−s}^*) + ∑_{v=1}^{S} α_v^r (r_{t−v} − r_{t−v}^*) + V_{v=1}^T x_{1,t}^T α + ε_t^y    (equation (4))
- Phillips curve (inflation gap representation):
  - (π_t − π_t^*) = ∑ β_p^π (π_{t−p} − π_{t−p}^*) + ∑_{q=1}^{P} β_q^y (y_{t−q} − y_{t−q}^*) + Q_{q=1}^T x_{2,t}^T β + ε_t^π    (equation (5))
- Potential output and NRIR dynamics:
  - y_t^* = y_{t−1}^* + g_{t−1}    (6.1)
  - g_t = g_{t−1} + ε_t^g, ε_t^g ~ N(0, σ_g^2)    (6.2)
  - r_t^* = c g_t + ε_t^r, ε_t^r ~ N(0, σ_r^2)    (7)
- Estimation details:
  - Quarterly data 2008–2018; forecast to mitigate end-sample bias.
  - Seasonally-adjusted national inflation and quarterly GDP.
  - Controls: HP-filter deviations of real exchange rate, oil-price inflation, commodity prices (smoothing parameter 200).
  - One lag used to eliminate serial correlation in residuals where indicated.
- Results:
  - NRIR around 4.7 percent (average for 2012-18).
  - NRIR of 4.9 percent over the projection period.
  - Time profile: increase from 4 percent (2012) to 5.2 percent (mid-2014); decline from 2016 to Q1-2017 to 4 percent; recovery thereafter to average level.
- Uncertainty and caveats:
  - Monetary policy regime shift in April 2017 may hinder accuracy because pre-April 2017 standing lending facility rate may not capture prior regime stance that relied on monetary aggregates.
  - Combined shocks in 2015-16 may not be captured well.
  - Estimating QPM or DSGE models and explicitly modelling regime shift could help assess uncertainty.

### Main findings and policy-relevant implications
- NRIR range across methods:
  - Clustered within a relatively narrow band: 4.6 percent to 7.8 percent.
  - Assuming inflation expectations anchored around 5.5 percent, corresponding neutral nominal interest rate (NNIR) range = 10.1 percent to 13.2 percent.
- Method-specific central estimates:
  - Consumption-smoothing model with habit formation: mean NRIR = 6.4 percent; range [4.9 percent, 7.8 percent].
  - Uncovered interest parity: NRIR = [6.3 percent, 7.6 percent].
  - Statistical filters average: 4.9 percent (range 4.6 percent to 5.6 percent).
  - Laubach–Williams semi-structural model: average NRIR 2012–18 = 4.7 percent; projection period mean = 4.9 percent.
- Key interpretations:
  - Width of range reflects methodological uncertainty and the roles of country risk premium and international financial conditions.
  - Structural breaks (e.g., April 2017 regime shift) complicate estimation and may affect the signaling properties of standing lending rate as proxy for policy rate.
  - NRIR is time-varying for Mozambique, exhibiting an increase in 2012–2014 and a decline around 2016–2017 before recovering.
  - Current assessment: Mozambique’s monetary policy stance is somewhat tight and in a process of normalization amid a negative output gap, expected single-digit inflation, and domestic and external risks.

---


_Source: https://www.imf.org/-/media/files/publications/cr/2019/1mozea2019003.pdf_