## wpiea2023088-print-pdf — Introduction

## Source details

**Canonical URL:** [wpiea2023088-print-pdf — Introduction](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023088-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023088-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023088-print-pdf.pdf.json)

---

### Research question and scope
- Core question: "Do Fiscal Rules Foster Fiscal Discipline in Resource-Rich Countries?"
- Sample and period:
  - 57 resource-rich countries (with 28 classified as oil exporters).
  - Annual data from 1976–2021.
- RRC classification rule: natural resources exceed 20 percent of total export earnings, on average between 2016 and 2020.

### Data, main variables, and estimation strategy
- Dependent variables:
  - Real expenditure growth (to test procyclicality).
  - Non-resource primary balance to non-resource GDP (to test fiscal discipline).
- Term-of-trade (ToT): change in the logarithm of the term of trade; lagged by one year for procyclicality tests and both lagged and current for discipline tests.
- Fiscal rule indicator: dummy = 1 if a country placed a numerical constraint on fiscal aggregates (budget balance, spending, debt or revenue) at the national and/or subnational level; enters models lagged by one year.
- Controls (selected): presence of an IMF program; real non-resource GDP growth; for discipline model additionally real GDP per capita and debt-to-non-resource-GDP ratio.
- Two empirical models:
  - Model 1 (procyclicality): OLS with country and time fixed effects; tests whether public spending is procyclical with ToT and whether fiscal rules reduce that procyclicality.
  - Model 2 (fiscal discipline): OLS with fixed effects and Blundell and Bond two-step system-GMM dynamic estimator; dependent variable is non-resource primary balance (or components).
- Identification and robustness:
  - Selection-bias tested via pre-adoption trends (1–5 years); reported results support rejection of selection bias.
  - Lagged ToT used to strengthen exogeneity; omitted variable bias mitigated with controls and fixed effects.
  - GMM implementation: instrument proliferation handled by collapsing instrument matrix and limiting lags to three; Windmeijer (2005) finite-sample correction applied.

### Key empirical findings — headline
- Procyclicality and deficit bias:
  - A rise in term of trade is associated with an increase in real public expenditure growth and a drop in non-resource primary balance.
  - Government expenditures are procyclical with ToT and commodity prices; procyclicality impact stronger in oil exporting countries.
- Fiscal rules’ effects:
  - Fiscal rules reduce procyclicality of spending following ToT rises, but only in oil exporting countries.
  - Fiscal rules improve non-resource primary balances across all resource-rich countries, with larger impact in oil-exporting countries.
  - Decomposition: effects are essentially driven by primary expenditure behavior.

### Effectiveness by rule type (ER, RR, BBR, DR)
- Sample limitations: small number of observations for some rule types; interpret with caution.
- Procyclicality:
  - Full sample: expenditure rules (ER) appear to mitigate fiscal procyclicality.
  - Oil exporters: revenue rules (RR) appear relevant while ER do not show significant effect (scarcity of ER in oil exporters noted).
- Fiscal discipline (OLS results):
  - Full sample: all rules except revenue rule (RR) are effective in improving primary balances after ToT increases.
  - Oil exporters: RR, BBR, and DR are effective; no evidence for ER in OLS.
- GMM robustness:
  - Balanced budget rule (BBR) and debt rule (DR) remain statistically significant in both samples.
  - ER and RR significance attenuates in some GMM specifications.

### State-dependence: upturns vs downturns
- Definitions:
  - Upturn (downturn) periods defined when the one-period lagged ToT growth is positive (negative).
  - Alternative tests using contemporaneous ToT also reported.
- Findings:
  - Procyclicality stronger during price decreases (downturns), where spending cuts are more pronounced.
  - Fiscal rules’ disciplinary effect more pronounced during ToT upturns: rules restrain relaxation of fiscal stance in good times and preserve room for maneuver during bad times.
  - GMM Table 2 interaction: lag (∆term of trade * rule) coefficients example: 0.196 ***, 0.150 ** (significant in upturn panels).

### Annex II — GMM estimation highlights (selected numeric outputs)
- Lag dependent variable examples (Table 1):
  - Full sample: 0.558 ***, 0.554 ***, 0.527 ***, 0.551 ***.
  - Oil exporters: 0.563 ***, 0.568 ***, 0.558 ***, 0.581 ***.
- Lag2 dependent variable examples:
  - Full sample: -0.030, -0.061 **, -0.016, -0.022.
  - Oil exporters: -0.063, -0.120 ***, -0.048 *, -0.054 **.
- Lag ∆term of trade examples:
  - -0.071 ***, -0.044 **, -0.084 ***, -0.082 ***.
- Interaction examples (lag (term of trade growth*rule)):
  - Lag (term of trade growth*BBR): 0.087 ***, 0.109 **.
  - Lag (term of trade growth*DR): 0.082 ***, 0.114 **.
- Diagnostics and samples (examples):
  - Observations: 621, 693, 693, 693, 350, 422, 422, 422.
  - AR(1)/AR(2): examples 0.013/0.153; Hansen test p-values examples 0.259, 0.153, 0.118.
- Term-of-Trade Rise and Fall (Table 2):
  - Lag dependent variable examples: 0.818 ***, 0.803 ***, 0.535 ***, 0.543 ***.
  - Lag ∆term of trade (examples): -0.159 ***, -0.146 **.
  - Interaction lag (∆term of trade * rule): 0.196 ***, 0.150 **, 0.111, 0.137.
  - Observations: 408, 265, 285, 157.
  - Hansen test p-values examples: 0.029, 0.163, 0.048, 0.157.
- Estimation notes:
  - Standard errors clustered by country.
  - Significance convention: *** p<0.01, ** p<0.05, * p<0.1.

### Country case studies — key practical insights and statistics
- Botswana:
  - Policy challenge: balance saving diamond revenue vs spending for development and diversification.
  - Gross public debt around 24 percent of GDP in 2021–22.
  - Government Investment Account at the Bank of Botswana fell from 45 percent of GDP in 2007/08 to 6 percent of GDP in 2021/22.
  - Design issues: ceiling on gross debt exists but no specific guidance on net financial position; spending targets not directly linked to net financial assets.
  - Recommendations: consider anchoring on PIH, implement transition period, cap non-resource primary fiscal balance and set a complementary floor on net financial assets.
- Mongolia:
  - FSL (2013) adopted three rules: structural deficit rule, expenditure rule, and debt rule.
  - Implementation weaknesses:
    - Structural deficit ceiling met only twice since 2013 and delayed until 2025.
    - Expenditure rule intended from 2013 but implemented only from 2017.
    - 2015: debt definition narrowed from public debt to general government debt excluding certain SOE debts and guarantees.
    - 2016: debt ceiling raised from 40 percent of GDP to 60 percent of GDP (in NPV terms).
  - Outcomes: non-compliance, frequent revisions, quasi-fiscal spending by DBM and BOM, volatile expenditure, sharp increase in public debt.
  - Recommendation highlight: make rules more binding and improve compliance and monitoring.
- Timor-Leste:
  - Petroleum Fund (PF) established 2005; all petroleum revenue goes into PF; Fund can only be used to finance the budget.
  - ESI methodology: Estimated Sustainable Income set at 3 percent of total government wealth (financial assets in the Fund plus net present value of future oil revenue); transfers in excess of ESI allowed only after government justification approved by Parliament.
  - Complementary constraints: ceiling on cost of external debt at 3 percent per year; benchmark external borrowing costs against average PF investment returns.
  - Key statistics:
    - PF net assets: USD 371 million in 2005 to USD 4.2 billion in 2008 (390 percent of non-oil GDP) [also reported elsewhere as 647 percent of non-oil GDP for 2008 in summary].
    - Withdrawals from PF averaged 5.2 percent of government wealth from 2010 to 2020 (IMF estimate), above the ESI benchmark.
    - Government deficit averaged 30.9 percent of non-oil GDP over the same period.
    - PF stock in 2020: USD 19 billion, about 12 times higher than non-oil GDP.
    - PF level in 2021 reported as USD billion 19.6 (about 1,100 percent of non-oil GDP) in one summary passage.
  - Risks: expected depletion of oil revenue from active fields as early as end-2023; need to balance development financing and fiscal sustainability.
  - Policy implication: strong institutions, medium-term strategy, and investment efficiency improvements required; Timor-Leste estimated to lose about 54 percent of potential value in the public investment process versus 24 percent average for emerging economies.

### Lessons and policy recommendations
- Design and anchors:
  - Anchor long-term fiscal policy on adapted versions of the permanent income hypothesis (PIH) focusing on net financial assets (existing financial assets + net present value of future resource revenue − public debt).
  - Consider transition periods when implementing PIH-style anchors to avoid unnecessary upfront adjustment and allow for diversification and development objectives.
- Buffers and escape clauses:
  - Accumulate liquid financial assets as fiscal buffers.
  - Define clear escape clauses to handle unexpected shocks without ad-hoc suspension of rules.
- Institutional support and compliance:
  - Consider relying on an independent fiscal council to supervise implementation of fiscal rules.
  - Develop supporting institutions and fiscal watchdogs to monitor rule implementation (none of the three case-study countries had such a fiscal watchdog).
- Integration and transparency:
  - Articulate fiscal rules with medium-term fiscal frameworks (MTFF) to bridge annual budgets and long-term objectives, guide transitions for resource exhaustion, and improve transparency and coverage to avoid quasi-fiscal activities outside the rules’ scope.
- Balance development and saving:
  - Ensure rules are consistent with development objectives while saving for future generations and shocks.
  - Support economic diversification, mobilize non-resource revenues, and improve public investment efficiency.

### Limitations and open questions
- Data limitations: inadequate data to test compliance or measure rule stringency across countries; empirical results rely on presence/absence of numerical rules.
- Heterogeneity: institutional differences across countries complicate some identification approaches.
- Small samples for specific rule types limit inference on comparative effectiveness.
- Open research question: assess fiscal rules’ comprehensive impact on fiscal sustainability against a long-term fiscal anchor that accounts for exhaustibility of natural resources and long-term challenges including climate change and demand/production risks for fossil-fuel exporters.

_International Monetary Fund — IMF Working Papers Do Fiscal Rules Foster Fiscal Discipline in Resource-Rich Countries? (Introduction section)_

### Introduction ...........................................................................................................

### wpiea2023088-print-pdf - Introduction

### Introduction and scope
- The chapter addresses the question: "Do Fiscal Rules Foster Fiscal Discipline in Resource-Rich Countries?"
- It frames common challenges affecting the conduct of fiscal policy in resource-rich countries and sets out the analytical approach used to assess fiscal rules.

### Stylized facts: common challenges
- Presents "Stylized Facts: Common Challenges Affecting the Conduct of Fiscal Policy in Resource-Rich Countries".
- Includes diagnostics on commodity price co-movement and volatility:
  - Figures listed: "Commodity Prices Co-Movement", "Difficulty in Forecasting Commodity Prices: The Case of Oil Prices", "Oil Exporters are More Dependent on Commodity Revenue than Other Commodity Exporters", and country-specific illustrations for Nigeria ("Co-Movement of Term of Trade and Commodity Prices, 2000–21" and "Expenditure Procyclicality, 2000–21").

### Data and estimation methodology
- Sections titled "Description of Data Used" and "Estimation Methodology".
- Two empirical models outlined:
  - Model 1: "Procyclicality of Public Expenditures".
  - Model 2: "Fiscal Discipline".

### Key empirical results (as reported in chapter headings)
- "Model 1 Confirms the Procyclicality of Expenditures and the Counter-Cyclical Role of Fiscal Rules."
- "Model 2 shows that the Presence of Fiscal Rules Improve Non-Resource Primary Fiscal Balances."
- Additional analyses reported:
  - "Investigating the Effectiveness of Fiscal Rules Across Different Types."
  - "Investigating if Fiscal Rules Are More Effective During Term-of-Trade Upturns."

### Tables and quantitative outputs referenced
- Main tables listed (titles preserved exactly):
  - Table 1. Fiscal Policy Procyclicality and Fiscal Rules
  - Table 2. Fiscal Discipline and Fiscal Rules: Fixed Effects OLS Estimations
  - Table 3. Fiscal Discipline and Fiscal Rules: GMM Estimations
  - Table 4. Fiscal Policy Procyclicality and Fiscal Rules Design
  - Table 5. Primary Balance and Term-of-Trade Index Growth: Effects by Rules Design
  - Table 6. Real Expenditure Growth During Term-of-Trade Index Rise and Fall
  - Table 7. Non-resource Primary Balances During Term-of-Trade Index Rise and Fall

### Country case analysis and practical insights
- Country cases section with subpoints indicated by headings:
  - "Rules Are Usually Set Up as a Mix of Formal and Indicative Ones."
  - "Rules Can Have a Positive Impact on Fiscal Discipline Even if they Are Not Fully Complied With, But Non-Compliance and Design Flaws Can Undermine the Sustainability of the Fiscal Framework."
- Additional country case materials listed in Annex IV:
  - Botswana: "Making Fiscal Rules More Effective Guides of Fiscal Policy"
  - Mongolia: "Making Rules More Binding"
  - Timor-Leste: "Balancing development needs financing and fiscal sustainability"
  - Figures for Timor-Leste listed: "Fund Revenues and Withdrawals" and "Total Government Balance"

### Lessons and policy recommendations (section headings preserved)
- "Some Lessons: Strengthening the Design of Fiscal Rules and Incentives for Compliance."
- Final synthesis phrased as a policy question: "Conclusion: Should RRC Adopt Fiscal Rules?"

### Methodological and supplementary material
- Annexes and robustness checks listed:
  - Annex I. Testing Selection Bias
  - Annex II. Results From GMM Estimations
  - Annex III. Data
  - Annex IV. Additional Information on Country Cases
- Glossary of acronyms reproduced (entries preserved exactly), including:
  - BBR Budget Balance Rules
  - BOM Bank of Mongolia
  - DBM Development Bank of Mongolia
  - DR Debt Rule
  - ER Expenditure Rules
  - ESI Estimated Sustainable Income
  - FSL Fiscal Stability Law
  - GDP Gross-Domestic Product
  - GMM Generalized Method of Moments
  - IMF International Monetary Fund
  - OLS Ordinary Least Squares
  - PF Petroleum Fund
  - PFM Public Financial Management
  - PIH Permanent Income Hypothesis
  - RCC Resource-Rich Countries
  - RR Revenue Rules
  - SBI Sustainable Budget Index
  - ToT Term-of-Trade
  - WEO World Economic Outlook

*Source: wpiea2023088-print-pdf - Introduction*

### Introduction

### wpiea2023088-print-pdf - Introduction

### Research question and scope
- Core question: Could fiscal rules bring fiscal discipline to resource-rich countries (RRC)?
- Sample and period: 57 resource-rich countries (with 28 classified as oil exporters), using annual data from 1976–2021.
- RRC classification rule: natural resources exceed 20 percent of total export earnings, on average between 2016 and 2020.

### Key empirical findings (headline)
- A rise in trade term is associated with increase in real public expenditure growth and a drop in non-resource primary balance, confirming procyclicality of public spending and pressures for running lower primary balances.
- Fiscal rules help reduce procyclicality following the rise in term of trade, but only in oil exporting countries.
- Fiscal rules improve primary balances across all resource-rich countries, with a larger impact in oil-exporting countries.

### Stylized facts on resource-rich countries
- RRCs face large, persistent and unpredictable commodity price shocks and difficulty in forecasting commodity prices over 1976–2021.
- In 2019, commodity export revenues represented 55 percent of government revenues in oil exporting countries on average compared to 11 percent for mining exporters.
- Historical behavior: fiscal policy in many RRCs is correlated with the commodity cycle; expenditures increase sharply during booms and are stickier during downturns, producing a deficit bias.

### Fiscal rules adoption patterns
- Fiscal rules adoption rose from 1 country, or (2 percent of all RRC) in 1985 to 30 countries (53 percent) in 2021.
- Popularity by rule type: Budget balance rule (BBR) remains the most popular, followed by debt rule (DR); expenditure rule (ER) saw a surge end of 2000s; revenue rule (RR) adoption stabilized since the end of the last decade.

### Data and main variables
- Dependent variables: real expenditure growth (to test procyclicality) and non-resource primary balance to non-resource GDP (to test fiscal discipline).
- Term-of-trade (ToT): defined as the change in the logarithm of the term of trade; ToT variable used lagged by one year for procyclicality tests and both lagged and current for discipline tests.
- Fiscal rule indicator: dummy = 1 if in a given year a country placed a numerical constraint on fiscal aggregates (budget balance, spending, debt or revenue) at the national and/or subnational level; enters models lagged by one year.
- Controls (selected): presence of an IMF program; real non-resource GDP growth; for discipline model additionally real GDP per capita and debt-to-non-resource-GDP ratio.

### Estimation strategy and identification
- Two complementary models:
  - Model 1 (procyclicality): OLS with country and time fixed effects testing whether public spending is procyclical with ToT and whether fiscal rules reduce that procyclicality.
  - Model 2 (fiscal discipline): OLS with fixed effects complemented by Blundell and Bond two-step system-GMM dynamic estimator to address dynamic panel endogeneity; dependent variable is non-resource primary balance (or its components).
- Identification considerations:
  - Selection bias tested via trends in real expenditures growth in the 1–5 years before rule introduction; authors report results consistent with rejection of selection bias, supporting OLS.
  - Reverse causality from fiscal stance to ToT considered unlikely for most individual commodity exporters given limited market power; use of lagged ToT strengthens exogeneity.
  - Omitted variable bias mitigated with controls and fixed effects; system-GMM used to handle dynamic panel endogeneity and instrument proliferation addressed by collapsing instrument matrix and limiting lags to three; Windmeijer (2005) finite-sample correction applied.

### Detailed empirical results (summarized)
- Model 1 (procyclicality)
  - Evidence confirms that government expenditures are procyclical with ToT and commodity prices.
  - The procyclicality impact is stronger in oil exporting countries.
  - Fiscal rules reduced procyclicality of spending in oil exporters; impact not statistically significant in the full sample of commodity exporters.
  - Business cycle control (real non-commodity GDP growth) matters but is of smaller magnitude than commodity price effects; IMF programs do not significantly tame procyclicality in these regressions.

- Model 2 (fiscal discipline)
  - Presence of deficit bias: ToT increases associated with deterioration of non-resource primary balances (via higher primary expenditures and lower non-resource revenues).
  - Fiscal rules mitigate the deficit bias and improve non-resource primary fiscal balances.
  - The disciplinary effect of fiscal rules is larger for oil exporting countries.
  - Decomposition shows effects are essentially driven by primary expenditure behavior.
  - The IMF program indicator and level of debt appear to influence fiscal discipline in OLS results; these controls are not always significant in GMM estimates.
  - Inertia of fiscal stance confirmed through significant first-order lag of dependent variable (one period).

### Effectiveness by rule type (ER, RR, BBR, DR)
- Sample limitations: small number of observations for some rule types; results should be interpreted with caution.
- Procyclicality (full sample): expenditure rules (ER) appear to mitigate fiscal procyclicality.
- Procyclicality (oil exporters): revenue rules (RR) appear relevant while expenditure rules do not show significant effect (noting scarcity of ER in oil exporters).
- Fiscal discipline (full sample, OLS): all rules except revenue rule (RR) are effective in improving primary balances after ToT increases.
- Fiscal discipline (oil exporters, OLS): revenue (RR), balanced budget (BBR), and debt rules (DR) are effective; no evidence found for ER in oil exporters in OLS.
- GMM robustness: balanced budget (BBR) and debt rules (DR) remain statistically significant in both samples; ER and RR significance attenuates in some GMM specifications.

### State-dependence: upturns vs downturns
- Procyclicality and rule effectiveness are state-dependent.
- The impact of ToT on procyclicality of public spending is especially relevant during periods of price decreases (price downturns), where spending cuts are more pronounced.
- The disciplinary effect of fiscal rules on the deficit bias is more pronounced during terms-of-trade upturns: fiscal rules are particularly effective in restraining relaxation of the fiscal stance in good times and preserving room for maneuver during bad times.
- Operational definitions: upturn (downturn) periods defined when the one-period lagged ToT growth is positive (negative); alternative tests using contemporaneous ToT show fiscal rules' effects significant during contemporaneous upturns.

### Country case studies and lessons
- Case studies: Botswana, Mongolia, and Timor-Leste illustrate mixed experiences and complement empirical analysis where data on compliance and design are limited.
- Common lessons:
  - Fiscal rules often combine formal (binding) and indicative (non-binding) elements.
  - Even when not fully complied with, fiscal rules can lead to some degree of fiscal discipline.
  - Frequent revisions, absence of long-term fiscal strategy, lack of compliance, or low stringency can significantly hamper effectiveness and undermine fiscal sustainability.
  - Fiscal rules should be framed within a broad fiscal framework focusing on financial assets, the non-resource fiscal balance, and supported by adequate Public Financial Management (PFM) systems.

- Selected country rule designs (examples):
  - Botswana: mix of indicative and formal rules including a Sustainable Budgeting Index (SBI) expenditure indicator, a composition target of development spending at least 30 percent of total spending, an indicative target of non-negative fiscal balance (2003), and a formal debt cap introduced in 2005 to be kept below 40 percent of GDP.
  - Mongolia: Fiscal Stability Law (FSL) adopted in 2010 defining binding rules from 2013, including a ceiling on the structural deficit of 2 percent of GDP based on a 16-year moving average of mineral prices; compliance challenges noted (structural deficit ceiling met only twice since 2013 and delayed until 2025).

### Limitations noted by the authors
- Lack of adequate data to test compliance with rules or fully assess design features across countries; empirical results rely on presence/absence of numerical rules rather than measures of compliance or stringency.
- Heterogeneity of institutional settings across countries complicates some identification approaches (e.g., treatment effects).
- Small sample sizes for specific rule types limit inference on comparative effectiveness of rule designs.

_Italic line: IMF Working Papers — Do Fiscal Rules Foster Fiscal Discipline in Resource-Rich Countries? (Introduction section)_.

### 2016. The debt ceiling has been subject to numerous changes as well. In 2015, the definition of the debt rule was narrow

### wpiea2023088-print-pdf - 2016. The debt ceiling has been subject to numerous changes as well. In 2015, the definition of the debt rule was narrowed from public debt to general government debt, excluding the debt of state-owned enterprises, state contributions to mining, energy, and railway projects, and state guarantees that are fully backed by government securities. In 2016, the debt ceiling was raised from 40 percent of GDP to 60 percent of GDP (in NPV terms).

### Changes to fiscal and debt rules
- 2015: The definition of the debt rule was narrowed from public debt to general government debt, excluding:
  - debt of state-owned enterprises,
  - state contributions to mining, energy, and railway projects,
  - state guarantees that are fully backed by government securities.
- 2016: The debt ceiling was raised from 40 percent of GDP to 60 percent of GDP (in NPV terms).
- Explicit additional constraints applied:
  - A ceiling on expenditure growth tied to the non-mineral GDP growth to prevent new mines from mechanically increasing the space for expenditure growth.
  - A ceiling on public debt (amended each year over the period 2014–21) as a secondary constraint to ensure policy consistency with sustainable debt.

### Timor-Leste: petroleum fund rules and fiscal limits
- Petroleum Fund (PF) established in 2005:
  - All petroleum revenue goes into the PF; the Fund can only be used to finance the budget (no direct spending authority).
- Transfers from PF guided by maintaining the real value of government wealth using an ESI methodology:
  - Estimated Sustainable Income (ESI) set at 3 percent of total government wealth (defined as financial assets in the Fund plus the net present value of future oil revenue).
  - Transfers in excess of the ESI are allowed only after government justification approved by Parliament.
- Complementary political commitment:
  - Maintain a ceiling on the cost of external debt at 3 percent per year.
  - Requirement to benchmark external borrowing costs against the average rate of PF’s investment returns.

### Country experiences and key statistics
- Botswana:
  - Gross public debt (including ...) was only around 24 percent of GDP in 2021–22.
  - The Government Investment Account at the Bank of Botswana fell from 45 percent of GDP in 2007/08 to 6 percent of GDP in 2021/22.
  - Diamond revenue sustained at high level; expenditures generally above levels recommended by rules; net financial assets fell during the pandemic.
- Timor-Leste:
  - Net assets of the Petroleum Fund grew from USD 371 million in 2005 to USD 4.2 billion in 2008 (647 percent of non-oil GDP).
  - Since 2009, the government started withdrawing from the PF in excess of the ESI to finance large infrastructure projects.
  - The PF still reached a level of USD billion 19.6 in 2021 (about 1,100 percent of non-oil GDP).
  - Fiscal sustainability challenged with active oil fields nearly depleted, limited progress in diversification, and sizable development needs.
- Mongolia:
  - Despite more formal rules, compliance was not better than Botswana or Timor-Leste.
  - Frequent revisions of rules often guided by spending policy concerns rather than fiscal prudence.
  - Quasi-fiscal spending by the Development Bank of Mongolia (DBM) and the Bank of Mongolia (BOM) exacerbated consolidated budget deficits and debt.
  - Non-compliance and frequent revisions undermined credibility, led to volatile expenditure and sharp increase in public debt.

### Empirical findings
- Econometric analysis (controlling for potential endogeneity) shows:
  - Fiscal rules reduce the procyclicality of real public expenditures with terms-of-trade in oil exporting countries.
  - Fiscal rules improve non-resource primary balances, especially during terms-of-trade upturns.
  - The design of rules matters: expenditure rules have the largest impact in the full sample of commodity exporters.
  - Revenue rules can be effective in oil exporting countries, but results require caution due to limited observations.
- Selection-bias testing (Annex I):
  - Interaction between leads of fiscal-rules adoption and term-of-trade index not statistically significant, suggesting absence of pre-treatment trends and supporting the use of Ordinary Least Squares (OLS).
  - Table 1: Observations 819, 819, 840, 859, 878; R2 0.205, 0.205, 0.205, 0.207, 0.205.
  - Table 2: Observations 477, 477, 492, 506, 520; R2 0.542, 0.542, 0.543, 0.543, 0.538.

### Lessons and policy recommendations
- Improve rule design and alignment with long-term sustainability:
  - Anchor long-term fiscal policy on adapted versions of the permanent income hypothesis (PIH) that focus on net financial assets (existing financial assets + net present value of future resource revenue − public debt).
  - Consider transition periods when implementing PIH-style anchors to avoid unnecessary upfront adjustment and allow for diversification and development objectives.
- Build buffers and plan for shocks:
  - Accumulate liquid financial assets as fiscal buffers.
  - Define clear escape clauses to handle unexpected shocks without ad-hoc suspension of rules.
- Institutional support to foster compliance:
  - Consider relying on an independent fiscal council to supervise implementation of fiscal rules.
  - Develop supporting institutions and fiscal watchdogs to monitor rule implementation (noting that none of the three countries studied had such a fiscal watchdog).
- Integrate fiscal rules with broader frameworks:
  - Articulate fiscal rules with medium-term fiscal frameworks (MTFF) to bridge annual budgets and long-term objectives, guide transitions for resource exhaustion, and improve transparency and coverage to avoid quasi-fiscal activities outside the rules’ scope.
- Balance between development and saving:
  - Fiscal rules should be consistent with broader development objectives, striking a balance between investing now (infrastructure, diversification) and saving for future generations and shocks.
  - Support economic diversification, mobilize non-resource revenues, and ensure public investment efficiency.

### Conclusions and further research
- Fiscal rules support fiscal discipline by reducing expenditure procyclicality and improving non-resource primary balances, but effectiveness depends on design, compliance, and institutional support.
- Practical weaknesses observed: lack of compliance, frequent rule revisions, loopholes, and missing medium-term strategy undermined rule efficiency in Botswana, Mongolia, and Timor-Leste.
- Recommended essentials: (i) comprehensive fiscal framework anchored on PIH (covering net financial assets), (ii) clear escape clauses, (iii) independent fiscal council, and (iv) a broader fiscal strategy addressing development needs and buffer accumulation.
- Open research question: Assessing fiscal rules’ comprehensive impact on fiscal sustainability against a long-term fiscal anchor that accounts for exhaustibility of natural resources and long-term challenges including climate change and demand/production risks for fossil-fuel exporters.

*International Monetary Fund — IMF WORKING PAPERS Do Fiscal Rules Foster Fiscal Discipline in Resource-Rich Countries?*

### Annex II.   Results From GMM Estimations

### Annex II.   Results From GMM Estimations

### Table 1 — Fiscal Discipline and Rules Design: GMM Estimations (Non-commodity primary balance)
- Lag dependent variable:
  - Full sample columns: 0.558 ***, 0.554 ***, 0.527 ***, 0.551 ***
  - Oil exporters columns: 0.563 ***, 0.568 ***, 0.558 ***, 0.581 ***
  - Standard errors (examples): (0.1663), (0.1660), (0.1426), (0.1380), (0.1358), (0.1713), (0.1273), (0.1228)
- Lag2 dependent variable:
  - Full sample entries: -0.030, -0.061 **, -0.016, -0.022
  - Oil exporters entries: -0.063, -0.120 ***, -0.048 *, -0.054 **
  - Standard errors (examples): (0.0251), (0.0256), (0.0275), (0.0271), (0.0466), (0.0305), (0.0266), (0.0239)
- ∆term of trade and lag ∆term of trade:
  - ∆term of trade (selected entries): -0.011, -0.027, -0.018, -0.018, -0.029, -0.048, -0.044, -0.046
  - Lag ∆term of trade (selected entries): -0.071 ***, -0.044 **, -0.084 ***, -0.082 ***, -0.068 *, -0.019, -0.082 **, -0.083 **
  - Standard errors (examples): (0.0335), (0.0199), (0.0200), (0.0193), (0.0913), (0.0390), (0.0415), (0.0427)
- Interaction terms (lag (term of trade growth * rule) variants):
  - Lag (term of trade growth*ER): 0.037, 0.134 (std. errors (0.0873), (0.5640))
  - Lag (term of trade growth*RR): 0.020, 0.010 (std. errors (0.0416), (0.1448))
  - Lag (term of trade growth*BBR): 0.087 ***, 0.109 ** (std. errors (0.0249), (0.0477))
  - Lag (term of trade growth*DR): 0.082 ***, 0.114 ** (std. errors (0.0255), (0.0461))
- Other covariates (examples):
  - Log real gdppc: -4.166, -2.766, -2.365, -2.121, 2.028, 1.709, -2.327, -0.022 (std. errors shown per column)
  - Real non commodity GDP growth: 0.094, 0.062, 0.097, 0.100, 0.097, -0.001, 0.013, 0.017
  - IMF program: -0.150, 0.132, -0.066, -0.057, -0.418, 1.054, 2.120, 2.345
  - Lag debt to non commodity GDP: -0.023, -0.015, -0.009, -0.006, 0.008, -0.018, -0.017, -0.002
- Constants (examples): 31.081, 16.690, 12.833, 10.781, -36.947, -30.900, 8.930, -13.431
- Sample sizes and diagnostics:
  - Observations: 621, 693, 693, 693, 350, 422, 422, 422
  - AR(1)/AR(2): 0.013/ 0.153, 0.014/ 0.197, 0.020/0.201, 0.016/0.200, 0.043/0.238, 0.031/0.202, 0.030/0.224, 0.029/0.220
  - Hansen test p-value: 0.259, 0.153, 0.118, 0.112, 0.209, 0.212, 0.125, 0.121
  - Country/instrument: 35/20, 36/20, 36/20, 36/20, 19/18, 20/20, 20/20, 20/20
- Estimation notes:
  - Standard errors clustered by country in brackets.
  - Significance: *** p<0.01, ** p<0.05, * p<0.1.

### Table 2 — Term-of-Trade Rise and Fall: GMM Estimations (Non-commodity primary balance)
- Lag dependent variable:
  - Full sample (price increase / price decrease sets): 0.818 ***, 0.803 ***, 0.535 ***, 0.543 ***
  - Standard errors (examples): (0.1131), (0.0900), (0.1604), (0.1750)
- Lag2 dependent variable:
  - Entries: -0.012, -0.029, -0.069 **, -0.093 **
  - Standard errors (examples): (0.0546), (0.0659), (0.0353), (0.0403)
- ∆term of trade and lag ∆term of trade:
  - ∆term of trade (examples): -0.055, -0.067, -0.016, -0.004
  - Lag ∆term of trade (examples): -0.159 ***, -0.146 **, -0.033, -0.029
  - Standard errors (examples): (0.0559), (0.1205), (0.0220), (0.0645), (0.0536), (0.0708), (0.0509), (0.0583)
- Interaction: Lag (∆term of trade * rule)
  - Coefficients: 0.196 ***, 0.150 **, 0.111, 0.137
  - Standard errors (examples): (0.0610), (0.0693), (0.0824), (0.1170)
- Other covariates and results (examples):
  - Log real gdppc: -0.255, -2.737, -10.649 *, -12.000
  - IMF program: 1.962 **, 2.876, 2.419, 5.661
  - Lag debt to non commodity GDP: -0.031, -0.062, -0.071, -0.170
  - Constant examples: 0.908, 24.386, 93.057 *, 112.498
- Sample sizes and diagnostics:
  - Observations: 408, 265, 285, 157
  - AR(1)/AR(2): 0.016/0.556, 0.027/0.681, 0.096/0.164, 0.293/0.619
  - Hansen test p-value: 0.029, 0.163, 0.048, 0.157
  - Country/instrument: 35/20, 19/20, 36/20, 20/20
- Estimation notes:
  - Price increase vs Price decrease panels shown.
  - Standard errors clustered by country in brackets.
  - Significance: *** p<0.01, ** p<0.05, * p<0.1.

### Key empirical findings (from Tables 1–2)
- The lagged dependent variable is consistently positive and often highly significant (examples: 0.558 ***, 0.818 ***).
- Second lag often displays negative coefficients, with some statistically significant entries (examples: -0.061 **, -0.120 ***).
- Lagged changes in the terms of trade tend to have negative effects on the non-commodity primary balance in several specifications (examples: -0.071 ***, -0.159 ***).
- Interaction terms between lagged term-of-trade growth and specific rule types (BBR, DR, and the rule dummy) show positive and in several cases statistically significant coefficients (examples: 0.087 ***, 0.082 ***; 0.196 *** for lag (∆term of trade*rule)).
- Hansen test p-values and AR diagnostics are reported for each specification to assess instrument validity and serial correlation.

### Estimation details
- Standard errors clustered by country.
- Sample composition varies across specifications (Full sample vs Oil exporters; price increase vs price decrease).
- Country/instrument counts and observations vary by specification as reported above.

### Annex III — Data definitions (variables used)
- Term of trade index: Ratio of export prices to import prices. Data from World Economic Outlook (WEO).
- Real government expenditure growth: Growth rate of general government expenditure in local currency adjusted by GDP deflator. Data from WEO.
- Non-resource primary fiscal balance: Non-resource primary balance in percentage of non-resource GDP. Data from WEO and United Nations Statistical Database.
- Non-resource primary expenditure: Non-resource primary expenditure over non-resource GDP. Data from WEO and United Nations Statistical Database.
- Non-resource revenue: Non-resource revenue to non-resource GDP. Data from WEO and United Nations Statistical Database.
- Fiscal rules: Dummy = 1 if country i at year t has a numerical rule on expenditure, revenue, budget balance or revenue and 0 elsewhere. Data from Davoodi et al. (2022).
- Expenditure rule: Dummy = 1 if country i at year t has a numerical rule on expenditure and 0 elsewhere.
- Revenue rule: Dummy = 1 if country i at year t has a numerical rule on revenue and 0 elsewhere.
- Budget balance rule: Dummy = 1 if country i at year t adopts a budget balance rule and 0 elsewhere.
- Debt rule: Dummy = 1 if country i at year t has a numerical rule on debt and 0 elsewhere.
- Real Non-resource GDP growth: Real non-resource GDP growth rate. Non-resource GDP computed by multiplying GDP from WEO by the non-mining sector’s share in total GDP from the United Nations Statistical Database.
- IMF program: Dummy = 1 if country i at date t is under the IMF program and 0 otherwise. Data from MONA.
- Real GDP per capita: GDP, constant prices, PPP 2017 international dollars, per capita. Data from WEO.
- Debt to Non-resource GDP: General government gross debt (WEO) as percentage of non-resource GDP. Both variables in current currency.

### Annex IV — Additional information on country cases (high-level summaries and key numbers)
- Botswana: Making Fiscal Rules More Effective Guides of Fiscal Policy
  - Policy challenge: Balance between saving diamond revenue for future generations and buffers vs spending for development and diversification.
  - Design issues noted: ceiling on gross debt exists but no specific guidance on net financial position (total financial assets minus debt); spending targets not directly linked to net financial assets.
  - Recommendation highlights:
    - Consider anchoring strategy on the permanent income hypothesis (PIH) model to turn net financial wealth into a constant stream to finance a sustainable non-resource deficit.
    - Implement a transition period if adopting PIH to avoid large upfront adjustments and preserve development objectives.
    - Consider a fiscal rule that caps the non-resource primary fiscal balance and a complementary floor on net financial assets.

- Mongolia: Making Rules More Binding
  - Background: 2013 Fiscal Stability Law (FSL) adopted three rules: structural deficit rule, expenditure rule, and debt rule.
  - Implementation weaknesses and timeline issues:
    - The ceiling on the structural deficit has been met only twice since 2013 and has been delayed until 2025.
    - The expenditure rule was intended from 2013 but implemented only from 2017.
    - Debt rule definition and ceiling changes: in 2015 the debt definition narrowed from public debt to general government debt excluding certain state-owned enterprise debts and guarantees; in 2016 the debt ceiling was raised from 40 percent of GDP to 60 percent of GDP (in Net Present Value (NPV) terms).
  - Outcomes and challenges:
    - Implementation assessed as inadequate by the IMF.
    - Non-compliance and frequent revisions undermined credibility, leading to volatile expenditure and a sharp increase in public debt.
    - Quasi-fiscal spending by DBM and BOM exacerbated consolidated budget deficits and debt.

- Timor-Leste: Balancing development needs financing and fiscal sustainability
  - Historical fiscal context:
    - Bayu-Undan production began 2004; fiscal framework adopted 2005 enabled accumulation of sizable savings.
    - Petroleum Fund net assets: USD 371 million in 2005 to USD 4.2 billion in 2008 (390 percent of non-oil GDP).
    - From 2010 to 2020, withdrawals from the Petroleum Fund averaged 5.2 percent of government wealth (IMF estimate), above the ESI benchmark.
    - Over the same period government deficit averaged 30.9 percent of non-oil GDP.
    - Petroleum Fund stock in 2020: USD 19 billion, about 12 times higher than non-oil GDP.
  - Forward risks and policy trade-offs:
    - Expected depletion of oil revenue from active fields as early as end-2023 and uncertain development prospects for new fields.
    - Need to balance fiscal sustainability with development and diversification needs.
  - Policy implications:
    - Fiscal rules in resource-rich countries need supporting institutions and fiscal strategies (including domestic revenue mobilization and spending efficiency improvements).
    - Investment efficiency concerns: Timor-Leste estimated to lose about 54 percent of potential value in the public investment process, compared with an average loss of 24 percent for emerging economies.

*IMF Working Paper — Do Fiscal Rules Foster Fiscal Discipline in Resource-Rich Countries? Working Paper No. WP/2023/88*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023088-print-pdf.pdf_