Policy Research Working Paper 9402 Designing Oil Revenue Management Mechanisms An Application to Chad Benoît Campagne Markus Kitzmuller Silvana Tordo Macroeconomics, Trade and Investment Global Practice September 2020 Policy Research Working Paper 9402 Abstract Oil resources usually play a significant role in oil-rich coun- and drawing down on the savings during difficult periods. tries, in gross domestic product and government revenues. Using the macro-structural model MFMod, the paper pres- High dependence of government revenues on oil can con- ents, evaluates, and discusses the strengths and weaknesses tribute to severe recession following an adverse commodity of different oil revenue management mechanisms applied price shock, such as in 2014. This paper examines the extent to the specific case of Chad. The scenarios demonstrate that to which a fiscal rule or stabilization fund could translate a well-designed management rule can successfully insulate into a less pro-cyclical fiscal policy, with the government the public budget from the oil price cycle, resulting in a saving part of its oil revenues during periods of high prices significant reduction in the volatility of the economy. This paper is a product of the Macroeconomics, Trade and Investment Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at bcampagne@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Designing Oil Revenue Management Mechanisms: An Application to Chad Benoît Campagne, Markus Kitzmuller, Silvana Tordo JEL Codes: C50, E62, Q32 Keywords: economic modeling, fiscal rule, oil revenue management, Chad Disclaimer: This analysis was written and finalized in 2019 prior to the COVID-19 crisis. Although, the baseline assumptions on the trajectory of the Chadian economy have crucially changed since then, the analytical results presented in this paper on the relative performance of different oil revenue management rules still hold. Acknowledgments: the authors would like to thank Ana Maria Jul, Fulbert Tchana Tchana, Olanrewaju Malik Kassim, Lars Moller as well as the IMF Chad and fiscal team for the fruitful discussions that benefited this analysis. Contents 1. Introduction ...............................................................................................................................3 2. The macro-econometric model MFMod-TCD ...............................................................................6 3. Baseline scenario and “business as usual” behavior of the Chadian economy ...............................8 i. Oil sector and price modelling .......................................................................................................... 8 ii. Modelling the « standard » behavior of the public expenditures .................................................. 11 iii. Baseline scenario ............................................................................................................................ 14 4. Outturns under alternative revenue management rules ............................................................17 i. The CEMAC fiscal framework .......................................................................................................... 17 ii. Price smoothing mechanism ........................................................................................................... 22 iii. CEMAC+ rule ................................................................................................................................... 24 iv. Comparison of outcomes ................................................................................................................ 32 5. Conclusion................................................................................................................................34 References .......................................................................................................................................36 Appendix .........................................................................................................................................37 A. Baseline response of the Chadian economy to an oil price shock.................................................. 37 B. Functioning of the implemented oil revenue management ........................................................... 40 2 1. Introduction In many resource-rich economies (see Herrera et al. 2019), procyclical fiscal policy, high dependence on the resource sector and the absence of oil revenue management mechanism left countries vulnerable to volatility and exogenous commodity price shocks. Indeed, low savings rates and boom-bust cycles have plagued economic development in a wide range of resource-rich developing countries (IMF, 2012). For resource-rich economies, the exhaustible character of the resource and capacity constraints that restrict the speed with which investments can be efficiently made and absorbed give rise to difficult intertemporal trade-offs in deciding how much of the wealth and revenues generated today should be consumed and invested today versus saved and invested or consumed tomorrow. The decisions made have implications for intergenerational equity, as well as fiscal and external sustainability. Especially in resource-rich developing countries, the volatility and disproportionate size of revenues compared with other sectors of the economy may imply different short and medium-term trade-offs and additional precautionary savings. Carneiro and Kouame (2019) argue that the use of fiscal rules in resource-rich countries has been associated with better fiscal/debt outcomes in the countries that have adopted them. Countries that have adopted fiscal rules experience a lower procyclical bias in fiscal policy as well as a positive effect on the risk premia of their sovereign bonds. Yet, the design of fiscal rules is crucial in order to obtain concrete, sustainable and efficient benefits. The fiscal rule that works for one country may well not be appropriate for another. A consensus in the literature (notably, see Kopits and Symansky, 1998) shows that a good fiscal rule is (i) simple: characterized by clearly stated, transparent and with realistic targets; (ii) flexible: allowing for course correction with clear escape clauses and, for resource-rich countries, allows for cyclical adjustments to the business cycle (or booms and busts); and (iii) enforceable: with clear and credible mechanisms to monitor and oversee the implementation of the rule. In addition, the literature is clear that an optimal resource management system allows current spending to be increased in a sustainable way, while scaling up investment in a way that can sustain a higher public capital stock (IMF, 2012). Precisely what that optimal system might be is less clear, different management frameworks and rules have been proposed. These include: • Non-resource primary balance rules or expenditure growth rules that aim to limit the procyclicality of expenditures, by setting a limit or a constraint to the growth of government spending (in nominal or real terms, or as a percent of non-resource GDP). When resource revenues are large (i.e. in “good” times), the cap set to expenditure growth forbids the government to immediately spent all available resource revenues and de facto forces the intertemporal transfer of some of those revenues to future periods. • The Hartwick rule which earmarks all revenues stemming from the exploitation of a depletable asset for the financing of capital expenditures only, which will generate returns in the future. This rule forbids the government to finance recurrent and therefore non-productive expenditures using a transient stream of revenue. • Permanent income hypothesis rules that impose that the resource-funded consumption level should not exceed the (implicit) return on the net present value of future natural resource revenue. During the extraction phase, most of the revenues are saved to build up non-resource 3 capital, in order for the return on these assets to compensate for the fall in revenue after extraction has ended. All of these fiscal frameworks for resource-rich countries share a common characteristic: they focus on constraining public expenditures to prevent governments from overspending resource revenues on non- productive expenditures. This approach naturally derives from a long-term view on how to convert depletable resources into long-term non-resource capital capable of generating revenues after the resource is depleted. Although theoretically desirable, constraining expenditures, and in particular earmarking revenues to non-recurring expenditures, is a politically sensitive topic and might be unfeasible in some contexts, for instance where urgent and unavoidable security-related expenditures are a recurring competitor for funding. A second option is to focus on the revenue side and attempt to decouple revenues as much as possible from the commodity cycle, leaving full freedom to the government for the allocation of expenditures. This corresponds to addressing the inherent volatility of resource revenues and their impact on budget planning at the source/intake. Cangiano et al. (2013) argue that volatility should be accommodated through strengthened public financial management practices such as the implementation of multiannual budgets or the design of long-term strategy plans. Yet, such plans require both a strong institutional capacity and a strong commitment capacity to elaborated plans that might not be feasible in low-income countries. Alternatively, volatility can be addressed directly at the source either by the creation of natural resource funds, in most cases for stabilization and saving purposes, or by market-based financial insurances (like the large hedging program set in place in Mexico, see Arezki et al., 2012). The latter option however can prove costly and again requires strong institutional capacities to supervise and manage sophisticated financial instruments. Cangiano et al. stress that in countries lacking strong budget planning systems, reliable internal controls and expenditure tracking systems (and most generally in institutionally fragile states), the creation of special natural resource funds can be adequate. An application to the case of Chad In Chad, the oil sector represents around 15% of real GDP, 85 % of exports and 40 % of public revenues in 2018. During the last 15 years, Chad did not manage to successfully exploit its oil dividend to foster long- term growth and the low diversification of the economy outside of the oil sector has actually been identified as one of the structural constraints weighting on Chad’s growth potential (Kitzmuller et al. 2018). Funding available for crucial structural reforms is repeatedly destabilized by volatile oil prices, unexpected oil-production shortages and competing priorities, notably those related to security. The economy’s dependency on the oil sector was strongly illustrated in the aftermath of the 2014 commodity price shock. The large loss in oil revenues for the government forced important budget cuts notably in social and capital expenditures with important spillovers to the real sphere. GDP declined by 6% in 2016, the budget deficit rose to 5% of GDP and public debt rose from 40% of GDP in 2014 to 55% in 2016. Prior to the COVID-19 outbreak, Chad had been gradually recovering from this severe recession and the country was pursuing a gradual fiscal consolidation in line with an IMF program that granted access to US$310.8 million in credits. Despite a tight spending envelope and improvements in the mobilization of non-oil revenues, the fiscal balance in the medium-term will still be driven largely by oil revenues. Oil revenues are projected to represent about 7 % of non-oil GDP over the medium term, partly reflecting 4 expected increases in oil production as new extraction technologies are implemented and new fields start to produce. Production volumes are expected to reach a peak in 2023 and stagnate afterwards. The 2018 restructuring agreement with an international commercial creditor - Glencore – markedly improved Chad’s liquidity position and restored debt sustainability. The agreement included a significant maturity extension, a large reduction in restructuring fees, a lower interest rate and contingency mechanisms that allow for reduced external debt service to Glencore when oil prices are lower. Since 2016, public debt dropped from 55 % of GDP to 48 % in 2018 and was estimated to be about 44 % of GDP at the end of 2019. While Chad’s debt is less vulnerable to oil price fluctuations now and appears to be on a sustainable path, the risk of external and overall debt distress remains high. The COVID-19 crisis is expected to exacerbate these sensitivities with the Brent oil price having fallen to 40 dollars per barrel in June 2020 and the Chadian economy projected to contract by 5.7 percent in 2020. All in all, given the significant role of oil in GDP and government revenues in Chad, and its weak institutional framework, it would therefore seem that Chad could be a strong candidate for a saving fund- based fiscal rule that allows to save part of oil revenues in the good days and to draw on them in difficult times (countercyclically). The absence of a functional fiscal rule or stabilization fund was particularly felt during the fall in commodity prices in 2014 (not that some attempts with no success where made in the past, see Box 1). With no fiscal buffers available, the resulting recession and shortfall in revenues put severe strains on public finances, ultimately rendering the government illiquid and public debt unsustainable. Box 1: Chad’s history with resource-revenue management Past attempts in Chad at managing oil revenues failed to deliver fiscal stability. In 1999, at the outset of the Chad-Cameroon pipeline project, an oil revenue management law was designed. The independent oil revenue control (Collège de Contrôle et Surveillance des Revenus Petroliers, CCSRP) was tasked with monitoring of expenditures financed by oil revenues. The mechanism was built around expected direct oil revenue, namely royalties from the sale of oil, and dividends from the government’s participation in the two transportation companies (Chad Oil Transportation Company - TOTCO and Cameroon Oil Transportation Company - COTCO) that own and manage the Chad-Cameroon pipeline. In 2007, indirect oil revenue from resource rent tax and corporate rent tax were added to the scope of application of the oil revenue management law and the role of the CCSRP was expanded accordingly. In 2014, triggered by the tightening fiscal space, a new revenue management law was enacted, but without any stabilization or saving function, the law simply became an earmarking mechanism to channel oil revenue to a list of priority sectors. In September 2018, the CCRSP was dissolved. The remainder of this paper therefore explores the extent to which alternative revenue management frameworks on the revenue side would succeed in supporting countercyclical expenditure policies in the face of oil price and production volatility in the particular case of Chad. Frameworks will be evaluated on the extent to which boom-bust cycles emanating from resource revenues are minimized and the consequences that this has in yielding a more or less smooth path for consumption. The analysis will focus on the capacity of those mechanisms to reduce the volatility of public expenditures (and hence of growth) both from of deterministic perspective in face of determined oil price trajectories, and from a stochastic perspective in face of volatile and unpredictable oil price paths. The following section 2 gives a presentation of the macro-structural model MFMod-TCD used for the analysis. Section 3 describes the key assumptions used in the simulations, while Section 4 present the alternative fiscal mechanisms examined and the results of simulations. Section 5 concludes. 5 2. The macro-econometric model MFMod-TCD In order to design and assess alternative oil revenue management mechanisms, this paper uses the macro- structural model The Macroeconomic and Fiscal Model for Chad (MFMod-TCD). MFMod-TCD is a country- specific structural econometric model of the Chadian economy, derived from the MFMod model framework of the World Bank (see Burns et al., 2019). It corresponds to a standard macro-structural, with significant customizations to reflect the special characteristics of the Chadian economy. Structural macro-econometric models are important tools for forecasting and policy analysis due to their ability to incorporate a detailed mix of judgement and the rich set of transmission channels they embody, which allows policy impacts on the broader economy to be followed at a deeper level of disaggregation than in other macroeconomic models, notably DSGE-type of models (Blanchard, 2018). MFMod-TCD is well suited to the present study. It incorporates the necessary flexibility to model alternative and complex oil revenue management mechanisms while accounting for the distinctive features of the Chadian economy, including: very specific debt-repayment rules for its restructured debt that imply a highly non- linear repayment mechanism; a country-specific government behavior that reflects national priorities; and a highly constrained fiscal space. Most importantly, MFMod-TCD includes a detailed and internally consistent description of both national income and fiscal accounts, with public expenditures affecting the real sphere of the economy, which in turn has an impact on both revenues and debt for the government. Those linkages are essential to assess the performance of a revenue management mechanism. In contrast, a pure fiscal-focused model would omit key components of fiscal sustainability, such as the direct relationship between savings from oil revenues on capital accumulation and the country’s sustainable growth path. The presence of such feedbacks in MFM-TCD allows the model to capture the trade-offs between lower volatility of public revenues and medium-term growth. Model characteristics In the MFMod framework, country-specific models are designed to reproduce the flow of funds across the whole economy by mapping out the main identities of the national accounts, balance of payments, labor markets and financial sectors (Burns et al., 2019 provides a detailed description of the model structure and equations). Structural relationships are consistent with both economic theory and the observed dynamics of the economy. The short-run dynamics are data-driven, with estimated parameters reflecting the actual behavior of the economy. Most equations are estimated using error-correcting models and parameters were estimated using, or calibrated to, Chadian data. In the MFMod framework, the overall equilibrium growth path of the economy is determined by potential output, which is calculated using the production-function methodology which is a function of total factor productivity, the level of capital and labor. Deviations from equilibrium, as well as convergence to equilibrium, are determined by the historical behavior of the Chadian economy. The data used in MFMod-TCD were obtained from the World Bank WDI database for national accounts data, the IMF for fiscal and balance of payments data and the ILO for labor market data. Detailed data on the fiscal accounts were sourced from the Ministry of Finance, and include: • Oil revenues, grants and other non-oil-related revenues. • Expenditures on goods and services, public wages and compensations of employees, interest payments, transfers and capital expenditures. 6 • A careful modeling of the debt-restructuring mechanism related to Glencore debt agreed in 2017. In MFMod-TCD, changes in the oil price or production transmit to the rest of the economy through the government’s expenditure decision. In the absence of a revenue stabilizing mechanism, as public oil revenues increase or fall, expenditures of the government will adapt accordingly, therefore affecting the real economy. Model simulations suggest that a permanent 10 USD increase in the price of oil leads to a 10 % increase in public expenditures, itself leading to a 0.5 % increase in real GDP (see appendix A). Different expenditures of the government have different transmission channels and therefore different economic effects (or fiscal multipliers): • Goods and services expenditures directly boost demand for goods (both domestic and imported) • Public wages and compensations directly boost GDP through public consumption but also indirectly as it increases the income of households • Transfers also contribute to an increase the income of households • Finally, capital expenditures boost demand for investment goods (both domestic and imported) and contribute to raise the capital stock, and subsequently potential production. In the end, the difference in the transmission channels of each expenditure shocks leads to relatively strong short-term multipliers (see Figure 1) around 0.5 to 0.6 for wages and transfers, as well as for capital expenditures (around 0.4). In the short-term the increase in demand leads to a widening of the output gap. As such, prices will adjust and progressively rise to restore market equilibrium, counteracting most of the initial increase in activity after 5 years. The capital expenditures multiplier is different from zero in the medium to long-term as the increase in investment contributes to a permanent increase in the capital stock and therefore in potential output. The limited size of estimated fiscal multipliers for Chad mainly reflects the high import content of some expenditures. Figure 1: real GDP response following a permanent increase of 1% of GDP in public expenditures (% deviation from baseline) 7 3. Baseline scenario and “business as usual” behavior of the Chadian economy A baseline scenario was generated as a point of reference from which to compare the impact of different mechanisms under alternative oil price trajectories. This baseline reflects the historical behavior of the economy, and notably of the Chadian government, in the absence of any revenue management mechanism. The baseline and analysis reported here was generated prior to the onset of the COVID-19 outbreak so that it can be interpreted as a no-COVID counterfactual. The model could also be used to explore the impacts of COVID on the Chadian economy, but that would be the subject of a separate paper. Box 2 outlines the assumptions regarding productivity, population growth and labor market behavior that are central to the long-run growth path of the Chadian economy incorporated in MFMod-TCD. In addition to these critical assumptions for the baseline include those for the baseline behavior of the government, the oil sector and the composition of spending in the “business-as-usual” . Box 2: Structural growth assumptions Central to the definition of the steady state of the model, the following structural global assumptions are made: • Total factor productivity is set to grow exogenously at 1.1 % -- its average growth rate over the period 2007-2017. • Working age population grows at 3.5 %, in line with the UN Population Division growth projection over the period 2020-2030. • Inflation expectations are set to the BEAC inflation target of 3 %. In addition, in order for a balanced growth path to exist, in the global world commodity prices are also assumed to grow at the same pace. • Real global growth is set to 3.9 % corresponding to the average forecast of the World Bank for 2019-2021. • As a member of the CEMAC monetary union, the EUR/CFAF exchange rate is held constant, while the EUR/USD exchange rate is held constant in both real and nominal terms. i. Oil sector and price modeling As a main driver of the Chadian economy and the focus of the revenue management mechanism, assumptions pertaining to oil are crucial for the quantitative simulation results. Baseline oil production and price forecast First, in line with IMF forecasts, the following assumptions are made concerning the output of the oil sector (also see Figure 2): • Up until 2023, oil production is aligned with the forecast in the IMF’s medium-term fiscal framework (IMF, 2019, Fourth Review of the Extended Credit Facility). After 2023, production is assumed to be constant, reflecting a balance between reducing oil reserves and additional light crude oil production expected to come on stream. 8 • The Brent oil price is also set to be consistent with the IMF medium-term framework up until 2023 and to grow at 0.74% after 2023 in line with World Bank forecasts. The discount on the DOBA oil price is also aligned with the MTFF and held constant at 4 USD after 2023. Figure 2: oil price and production baseline forecast Oil price volatility Although volatility in public oil revenues stems from both the production level and the oil price, the main source of volatility is oil prices. As such, the focus in this paper will be limited to price volatility. The nature of results would not be changed if the volatility arose from the production side. While the above assumptions drive the baseline or median scenario results, in order to assess the dynamic behavior of different revenue management mechanisms over the commodity price cycle, a data generating process for the deviation of the oil price from this baseline is required. Figure 3: Brent price (USD/bbl) Annual frequency Monthly frequency Historically, the price of oil has tended to be much less stable than assumed in the baseline, with large intra-annual volatility (see Figure 3). Changes in the price of oil are positive in 65% of the cases, leading to a mean increase of 2.8 USD per year over the period 2000-2018, with the trend growth reflecting the continuous increase in the price of oil from 25 USD to 110 USD between 2002 and 2012, notably in the context of an increased demand for oil from China. Since 2005 however, there has been no statistically significant trend in oil prices. In order to preserve the volatility of oil prices observed on monthly data, alternative price scenarios are generated at a monthly frequency before being averaged into annual data. To smooth outliers and 9 discontinuities in the empirical distribution of month-to-month changes in oil prices, the data generating process is obtained by fitting a known distribution over this empirical distribution. This distribution will then be used to draw alternative series of price changes over the simulation period. Due to the strong kurtosis of the empirical distribution (see Figure 4), fitting a normal distribution would be inappropriate (a Jarque-Bera test suggests a full rejection of the null hypothesis of a normal distribution). A Student distribution provides a much better approximation of the empirical distribution. including its fat-tail property. Figure 4: density of the month-to-month change in the price of the Brent oil in USD for 2000-2018 .12 Histogram Student's t .10 Normal .08 Mean: 0.135 Median: 0.859 Density .06 Std. Dev.: 5.620 .04 Skewness: -1.174 Kurtosis: 5.984 .02 .00 -30 -20 -10 0 10 20 30 Overall, the data generating process for the monthly price of oil that will be used in the following sections will assume that the price of the Brent oil follows a random walk with a deterministic drift: = −1 (1 + ) + avec ϵt ~ σ ∗ T(ν) The is calibrated to the long-term growth of nominal oil prices forecast by the World Bank of 0.74% per year. The innovation of the process is assumed to be of mean zero (as opposed to the positive mean observed over the period) and to follow a Student distribution calibrated over the period 2000-2018. The best fitted distribution is obtained for = 4 and = 5. The monthly price trajectories are then averaged to obtain annual price trajectories. With this data generating process, alternative oil price scenarios lead to the simulated distribution presented in Figure 5. 10 Figure 5: generated distribution of the BRENT oil price over 2020-2030 Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. ii. Modeling the « standard » behavior of the public expenditures Central to the evaluation of the benefits and disadvantage of the revenue management mechanism, the unconstrained (or “business as usual”) behavior of the Chadian government needs to be defined. Understanding and defining the historical pattern of expenditures from the government is necessary to see how the fiscal rules and mechanisms to be simulated interact and restrict this pattern. This is equivalent to defining the government spending reaction function. Taking the large fall in oil prices and revenues during 2014-2016 as a benchmark, fiscal consolidation was achieved through cuts in all expenditures, with the exception of wages (and partly goods and services, see Figure 6). As a result, expenditures are classified into two categories: • Inelastic expenditures, namely wages and compensations as well as interest payments. These expenditures are assumed to be paid regardless of the level of actual revenues. This corresponds to the prioritization of payment of wages of public servants by the government of Chad (notably military wages) and to the compulsory nature of interest payments – and has a rough equivalence to the notion of “entitlements” in the US fiscal literature – although entitlements in the US mainly concern pensions, Medicare and other transfers mandated by law. • Elastic (or discretionary) expenditures, in the Chadian context include goods and services, transfers and capital expenditures. In the modeling, these expenditures are assumed to adjust (with lag) to any fall or increase in disposable revenues, defined as revenues less the payment of inelastic expenditures. In terms of modeling, the following assumptions are made for the business as usual behavior of the government: 1. Tax rates and non-tax revenue’s share of GDP are held constant. As a result, government revenues are determined by fluctuations in the various tax base for each revenue type. 2. Nominal interest rates on the government debt are held constant: as a result, interest payments vary with fluctuations in the outstanding public debt. 11 3. Nominal wages are assumed to grow at their average post-2014 growth rate of 3.1%, roughly corresponding to constant real public wages (at unchanged public employment level). 4. Elastic expenditures , (goods and services expenditures, transfers and capital expenditures) are modeled as reacting to changes in net revenues (revenues after inelastic expenditures) as below: − , , = ∗ ( ) − − + − , − , = ∗ � + � − , Where is the government's net borrowing limit including Glencore repayments, are interest payments on debt, and correspond to public wages. are total revenues of the government, whereas are disposable revenues after payment of wages and interest payments. In other words, any increase (or decrease) in revenues or in the fiscal space (i.e. borrowing limit) will be fully passed on to expenditures. Under this regime, the government uses all its available fiscal space. The weight − , / − , fixes the expenditure shares – i.e. the compositional shares do not shift. Figure 6: behavior of expenditures during 2014-2016 (% growth) 12 Figure 7: share of inelastic and elastic expenditures (% of GDP) For the period through 2023, Chad’s borrowing limit is effectively dictated by its IMF program (linked to the Extended Credit Facility). Under the terms of this agreement, the Chadian government is constrained in its access to both concessional and non-concessional (net) borrowing. After 2023, the government is assumed to have only limited access to financial markets and spending will be constrained by the capacity to borrow. In the model, the borrowing limit is first defined and computed from the Medium-Term Fiscal Framework of the IMF program (MTFF) as the sum of loan disbursements, treasury bills and bonds as well as relief under the Heavily Indebted Poor Countries initiative, minus amortizations scheduled in the fiscal framework. After 2023, the borrowing limit is extended assuming that additions to gross debt are constant in nominal terms, and that amortizations decline proportionately with falling debt levels. The modeling treats Glencore and non-Glencore debt amortization differently. In line with existing agreements, the Glencore principal and interest repayments depend upon oil prices. Glencore-related amortizations are endogenous and dynamically determined according to the restructuring agreement rules (i.e. the redemption schedule follows the contractual mechanism based on the level of oil prices). Thus, across different oil price scenarios, Glencore repayments will vary. Non-Glencore debt-servicing and amortization are assumed to be unchanged across scenarios. 13 Figure 8: decomposition of the borrowing limit (bn of CFAF) Note: If the borrowing limit is < 0, the budget must be in surplus, the amortizations being larger than the issuance of new debt. Numerically, in the baseline scenario, the computed borrowing limit stabilizes around 50 bn CFAF after 2024 (see Figure 8). The borrowing limit notably improves largely after 2024, as Glencore-related repayments are expected to come to an end, easing the constraint for the government by as much as 1.5 % of GDP. iii. Baseline scenario The assumptions presented above allow to trace the path of the economy in the absence of any revenue management mechanism (see Figure 9). Note that for the period 2019-2021, the forecast is aligned with both the World Bank Macro Poverty Outlook for April 2019 and the concurrent IMF fiscal framework. In the baseline scenario, after 2023 Chad's dependence on oil decreases over time, principally because oil production stabilizes at 220,000 barrels per day, while the economy keeps growing at around 4% per annum and because the non-oil price of domestic production rises (c. 3% per annum) faster than oil prices. 14 Figure 9: baseline scenario Real GDP (% growth) Budget balance (% of GDP) Public debt (% of GDP) Oil revenues (% of GDP) On the fiscal side, until 2023, the government is assumed to run a budget surplus to meet scheduled debt repayments (amortizations planned in the MTFF). In 2024, the end of the Glencore repayments decreases the budget constraint imposed to the government, which in line with the reaction function described above, increases spending bringing the fiscal balance back toward zero (see Figure 10). Figure 10: fiscal balance and borrowing limit (in bn of CFAF) Note: the budget constraint is defined as the opposite of the borrowing limit presented in Figure 8. A positive budget constraint as over the whole sample imposes that the overall budget must be in surplus, required amortizations being larger than the allowed issuance of new debt. 15 Simulations of alternative oil trajectories under the historical fiscal reaction functions Assuming the expenditure functions laid out in the preceding section, and a range of possible outcomes for oil prices (drawn from the historical behavior of oil prices), economic outcomes are more volatile than suggested by the baseline scenario (see Figure 11). These scenarios are generated by running 5,000 simulations each drawing an oil price trajectory from the oil-price probability distribution each year. In 4,500 (90%) of these scenarios oil revenues lie between 3 % and 12 % of GDP in 2023 (the frontier defined by the light blue part of the distribution). There is a 70% probability of revenues ranging between 5 and 11 percent of GDP (area bound by the darker blue shading), and a 60 percent chance of revenues ranging between 7 and 10 percent of GDP –a significant level of uncertainty. Even the darkest blue area, which accounts for only 20 percent of cases, spans a relatively wide range of between 7 to 9 percent of GDP. After 2023, both the median revenue as a percent of GDP and the range of possible outcomes are expected to decline as the economy diversifies away from oil, due to faster growth in the non-oil economy. Despite the declining trend, oil plays a crucial role in the Chadian economy and will continue to do so in the near future. In this “business-as-usual” scenario, the procyclical spending behavior described in the previous section reduces the volatility of the fiscal balance (deficit/surplus), but contributes to the volatility of GDP – especially in the early years where the debt repayments required under the Glencore deal constrains the total spending of the government. Depending on oil prices observed, GDP growth outcomes could vary by as much as 4 percentage points during the period 2020-2025, and up to 2 p.p. in the longer term. 16 Figure 11: expected volatility of the economy Real GDP (% growth) Oil revenues (% of GDP) Budget balance (% of GDP) Public debt (% of GDP) Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. 4. Outturns under alternative revenue management rules Having identified the large volatility of growth in Chad over the commodity price cycle, this section now focuses on the evaluation of three alternative revenue management mechanisms aimed at reducing this volatility: (i) the CEMAC macro-surveillance rule, (ii) an oil price smoothing mechanism and (iii) an extended and refined version of the CEMAC rule, named CEMAC+ below. Each mechanism is evaluated in a systematic manner along two axes: (i) their behavior in face of deterministic oil price trajectories (called deterministic performance analysis below), and (ii) their behavior in face of stochastic oil prices (called stochastic performance analysis below). Those two approaches complement each other and help assess how each mechanism responds to a well-defined oil price shocks as well as how they perform on average over the unobserved probabilistic commodity price cycle. i. The CEMAC fiscal framework In August 2017, the Central African Economic and Monetary Community (CEMAC) adopted a new multilateral surveillance framework (see the Programme des Réformes Economiques et Financières de la Cemac (PREF-CEMAC), p. 36-37). This framework introduced a budget rule taking into account the 17 importance of oil revenues for the CEMAC members. As a legally binding framework for the Chadian government designed to address the question of commodity-linked revenue management, it is natural to start with the examination of the extent to which this rule reduces Chad’s vulnerability to oil price volatility. Definition The CEMAC rule imposes that the Reference Budget Balance () should be above -1.5 % of GDP. The reference budget balance is defined as follows: RBBt OBt − FSOR t = GDPt GDPt FSOR t OR t OR t 1 OR t−1 OR t−2 OR t−3 = 0.2 ∗ + 0.8 ∗ � − ∗( + + )� GDPt GDPt GDPt 3 GDPt−1 GDPt−2 GDPt−3 OR t 1 OR t−1 OR t−2 OR t−3 = − ∗( + + ) GDPt 3 GDPt−1 GDPt−2 GDPt−3 Where is the overall budget balance, are financial savings from oil revenues, and total collected oil revenues in year . In practice, the financial savings from oil revenues corresponds to the deviation (measured as a % of GDP) of oil revenues from its (3-year) trend. Effectively the rule requires a government to save “surprise” or “excess” revenues (revenues that exceed the backward looking 3-year moving average of oil revenues). In this setting, the reference budget balance is the balance calculated based on the 3-year average of oil revenues, whereas the overall balance is the actual revenues. If spending is not adjusted and “excess” revenues are not spent, they will correspond to savings from oil revenues for the government. Under this formulation, savings are inherently asymmetrical as negative “surprise” revenues will still automatically generate a cut in expenditure. Moreover, the rule is not binding as the framework does not require that “excess” revenues be explicitly saved. A supplementary budget can be passed authorizing their expenditure. The CEMAC surveillance framework also includes a 70% of GDP debt ceiling over 25 years, limiting the speed of debt accumulation. In practice, this ceiling is non-binding in all simulations due to the strong downward trajectory of Chadian debt. It is therefore ignored in the following discussion. Prior to 2013, the CEMAC deficit rule was systematically violated with the reference deficit regularly exceeding the CEMAC target (Figure 12). Since 2013/2014, following the decline in oil prices and the large fiscal consolidation that it engendered the rule has been met (see Figure 12). Prior to 2013, the average deviation from the target was 3.3 pp of GDP. In 2018, savings from oil revenues () were 2.3% of GDP. 18 Figure 12: reference budget balance (% of GDP) Stochastic performance analysis Over the commodity price cycle (i.e. over a large number of simulated oil price trajectories), the recent fiscal consolidation efforts and its limited borrowing capacity implies that, given the current structure and behavior of expenditures, the rule is rarely binding (see Figure 13). Chad runs large budget surpluses in all simulated scenarios. The deficit target is only binding in the most extreme cases (when oil prices fall below the second standard deviation). Thus, in the bulk of cases public expenditures do not behave differently than in the absence of a rule. Indeed, if the deficit target does not bind, financial savings from oil revenues are re-integrated in the general budget and are spent (with unchanged total oil revenues). Overall, the CEMAC rule helps stabilize public finances but does not contribute to the stabilization of the economy in simulated scenarios (see Figure 14). Only in the case of strong upward oil price shocks does spending behavior change by enough to have real economic consequences, reducing the range of outcomes when prices rise above the 70th percentile of the distribution. While this diminishes the amplitude of booms, busts are unaffected. Figure 13: reference budget balance with and without the rule No CEMAC rule With CEMAC rule Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. 19 Figure 14: volatility of the economy with and without the rule No CEMAC rule With CEMAC rule GDP growth (in %) Overall budget balance (% of GDP) Public debt (% of GDP) Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. Deterministic performance analysis During “good” times (when the oil price above its trend), financial savings from oil revenues defined by the rule increase and tend to reduce the reference budget balance. If the increase is large enough to bring the reference balance below the deficit target, the rule will bind, and expenditure cuts will be required to comply with the -1.5% deficit target (see Figure 15). 20 During “bad” times (when the oil price below its trend) financial savings from oil revenues required by the rule decrease and the reference budget balance increases – alleviating the constraint imposed by the rule. In an unconstrained borrowing environment, this might allow the government to conduct a counter- cyclical policy (increase expenditures during downturns) while remaining compliant with the CEMAC surveillance framework. However, when access to credit is constrained the borrowing to finance the additional spending may not be possible, and lower oil revenues might also lead to lower expenditures (see Figure 15). Overall, the CEMAC rule only weakly reduces the volatility of the economy and does so by reducing the amplitude of booms (lower peak growth even for large increase in oil prices), without mitigating negative price shocks – which still induce large expenditure cuts. This result is the direct consequence of the limited borrowing capacity of the Chadian government. Were that constraint to be relaxed (either by the creation of a rainy-day fund into which surplus revenues were stored during good times) or if the government had better access to financial markets the rule would allow for some counter-cyclical spending during bad times. However, in a constrained borrowing environment as for Chad, this require an explicit savings account to be able to access current surpluses at a later date. Figure 15: Response of the economy to oil price shocks under the CEMAC budget rule Public expenditures (% deviation from balance) Budget balance (pp difference from baseline) GDP (% deviation from balance) 21 ii. Price smoothing mechanism Having stressed the importance of building accessible surpluses to circumvent the borrowing constraint of the government, the second mechanism under scrutiny explicitly introduces the creation of a dedicated savings account. This mechanism was suggested by the IMF as part of its Fourth review of the Extended Credit Facility in order to strengthen the design and the preparation of spending plans in face volatile of oil prices using a simple price smoothing rule. Definition The suggested price smoothing mechanism consists of using a smoothed (or structural) oil price when preparing the budget. This smoothed oil price is computed as the moving average of oil prices. For instance, a (5,1,2) rule will imply a reference oil price equal to the moving average of the past five years, the current year and next two years projections – 8 years in total. When actual (realized) oil revenues are higher than projected, excess revenues are accumulated in a stabilization buffer. Conversely, when actual oil revenues are lower than projected by the reference price, the deficit is financed by drawing down resources previously accumulated in the stabilization buffer. In this setting, the calibration of the moving average used to compute the reference price is central to the stabilization capacity of the mechanism. There is a trade-off between a more reactive (short moving average) and a more persistent (longer moving average) reference price (see Figure 16). If too long lags are used, the mechanism may not help be beneficial as deficits will be persistently too high as the system slowly adjusts. If too short a period, then the rule ceases to be binding (at the limit a lag of zero is the equivalent of no rule at all). Figure 16: calibration of the moving average BRENT oil price (in USD/bbl) A longer moving average implies that any contemporary increase in the oil price will be “assessed” by the rule as transitory. The trend, i.e. the estimated structural price will only be revised slowly in the face of a persistent shock. As such, the smoothing mechanism will favor an accumulation of savings in the face of a persistent rise in oil prices, and the accumulation of debt during a persistent downturn. If shocks are transitory, the rule will help to stabilize spending and reduce volatility. Because of the difficulty of assessing the transient or permanent character of a given shock, a moving average of intermediate length (3 to 5 years) was proposed by the Fund. Retrospectively, this mechanism would have successfully mitigated the decline in revenues in 2014 (see Figure 17). Using a three-year backward moving average in the spirit of the CEMAC rule, the mechanism 22 leads to the accumulation of savings over most of the period 2005-2014, and to dissaving between 2015- 2017 during the commodity price shock. Going in more details, had the mechanism been in place before 2008, the fund could have largely made up for the loss of oil revenues (up to a cumulated amount of 8 % of GDP) before being exhausted. Figure 17: retrospective size of the stabilization fund Note: for comparison purposes with the CEMAC framework, the moving average is calibrated here to a 3-year backward moving average (3,0,0). Stochastic performance analysis Under simulation, the rule serves to reduce volatility – partly because the median price of oil is assumed to be stable. If it were rising or falling over time, the rule could generate the kind of misallocations described above. Under simulation the smoothing effect of the rule is muted initially because it takes time for the stabilization fund to accumulate money to be spent during bad times. As a result, the real payoff in terms of increased stability only begins to be seen after 2025 as the stabilization buffer ramps up to a median size of 5 % of GDP (see Figures 18 and 19). Figure 18: volatility of the economy with and without the smoothing mechanism Without smoothing mechanism With smoothing mechanism GDP growth (in %) Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. For comparison purposes with the CEMAC framework, the moving average is calibrated here to a 3-year backward moving average (3,0,0). 23 Figure 19: savings with a smoothing mechanism bn of CFAF % of GDP Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. For comparison purposes with the CEMAC framework, the moving average is calibrated here to a 3-year backward moving average (3,0,0). iii. CEMAC+ rule Although explicitly designed to account for the volatile nature of oil revenues, the CEMAC fiscal framework falls short of stabilizing activity in Chad. The first reason was the absence of an explicit savings account to which financial savings from oil revenues are allocated, and from which they could be disbursed in downturns. The second reason related to the -1.5 % of GDP target for the fiscal balance, a target that was too low for Chad. A CEMAC+ rule might be imagined that introduced an explicit stabilization account. This stabilization account would be similar to that imagined in the IMF price smoothing mechanism but would be based on oil revenues rather than prices and thus would cover fluctuations in production as well as in prices. In addition, the deficit target will be adjusted to be tighter and reflect both the current state of public finances in Chad as well as its limited access to financial borrowing. Finally, while introducing additional modifications detailed below to the management of the stabilization account, the CEMAC+ rule also remains fully compliant with the original CEMAC rule. Definition The reference budget balance is still defined as the overall balance (OB) less financial savings (FS) from oil revenues (OR): RBBt OBt − FSOR t = GDPt GDPt FSOR t OR t OR t 1 OR t−1 OR t−2 OR t−3 = 0.2 ∗ + 0.8 ∗ � − ∗( + + )� GDPt GDPt GDPt 3 GDPt−1 GDPt−2 GDPt−3 OR t 1 OR t−1 OR t−2 OR t−3 = − ∗( + + ) GDPt 3 GDPt−1 GDPt−2 GDPt−3 However, the deficit target now imposes a balanced budget: ≥ 0. Oil revenues available in the budget are then defined as: ORst 1 OR t−1 OR t−2 OR t−3 = 0.8 ∗ ∗ ( + + ) GDPt 3 GDPt−1 GDPt−2 GDPt−3 24 In addition, oil savings are now explicit and saved/dissaved into a stabilization account (). While savings are not bounded upwards, dissaving is limited by the amount already available into the account. In addition, the rule allows for an adjustment in the case of persistent negative oil revenue shocks. 100% of the revenue shortfall is covered in the first year of occurrence of a shock, 66 % in the second year, 33 % in the third and 0 % in the fourth. Any residual adjustment is absorbed by an adjustment of public expenditures. The flow of oil savings () for a shock occurring at is given by: �ORt0 − ORs s t0 � ≥ 0 ℎ OSt = �ORt0 − ORt0 � 0 = − min�ORs t0 − ORt0 , FAt0 −1 � ⎧ ⎪ 2 �ORt0 − ORs 0 +1 = − ∗ min�ORs t0 +1 − ORt0 +1 , FAt0 � t0 � < 0 ℎ 3 ⎨ 1 ⎪ s ⎩ 0 +2 = − 3 ∗ min�ORt0 +2 − ORt0 +2 , FAt0 +1 � Under this revenue management scheme oil savings are explicit and taken out of the general budget, and the reference budget balance corresponds to the budget balance computed using the budgeted oil revenues. As such, the zero-deficit target for the reference balance simply ensures that the planned budget is balanced. This ensures the ex-ante compliance of the rule with the CEMAC framework. Note that the suggested modifications to the original CEMAC rule presented above are not the only possible ones. The selected version of the rule acts as a general framework for discussion, yet multiple sophistications or adjustments can be added (see Box 3). As such changes do not alter the spirit of the rule, those additions are not discussed here. Box 3: Further refining the CEMAC+ rule The current formulation of the rule is highly adjustable and customizable. The parameters and mechanisms can be adjusted according to the preference of the government choosing to implement a revenue management rule. The following parameters/lines can easily be adjusted to fit specific needs and preferences: • Deficit target: the deficit target can be adjusted to allow for more stringent or looser fiscal policy. • Level of savings: the strict adherence to the CEMAC rule can be loosened and tighter or stronger savings accumulation rule can be designed. This can allow the rule to better reflect political preferences and financing priorities in the economy. • Adjustment schedule and coverage: the length and size of insurance coverage can be adjusted to allow for shorter/longer period of coverage with more/less gradual phasing out of the coverage. • Use of interest income: interest income can be used to protect the value of the fund against inflation or can be used externally to finance investment or the general public budget. • Ceiling to the stabilization fund size: the size of the fund might be constrained to a ceiling value to avoid over-accumulation of savings beyond the desired level of risk coverage. 25 • Minimum shortfall threshold for triggering: if self-insurance is targeted to large shortfall in revenues, a minimal shortfall threshold might be implemented to avoid triggering dissaving from the stabilization fund in face of minor shocks. 1 • Exemption and escape clauses: the mechanism might be suspended, or dissaving might be authorized in the event of adverse shocks such as climate disaster or security issues. This would allow to preserve financial flexibility in stress situations. Note that the modification of the rule might sever the compliance of the CEMAC+ rule with the CEMAC macro-surveillance framework. If this rule has been in place from 2005, the budgeted revenues (see Figure 20) would have been smoother than actual revenues, savings would have accrued, which would have allowed the stabilization mechanism to kick in during the negative commodity price shock of 2014-2016. If the rule had been implemented before 2012, oil savings during the oil price boom of the 2000s and the early 2010 would have provided sufficient budgetary support to cover revenue losses (see Figure 21 and Table 1). This would have in turn allowed for a more gradual adjustment of public expenditures in 2014 and 2015. Figure 20: actual vs. budgeted oil revenues bn of CFAF % of GDP 1 In the context of the decreasing coverage schedule exposed above, the introduction of a threshold would avoid triggering the 3-year count for small shocks. For instance, in 2013, oil revenues would have fallen below the budgeted revenues allowed by the CEMAC+ rule and would have therefore led to the start of the 3-year coverage period. As a result, the larger shocks in 2014 and 2015 would have been covered at only 66 % and 33 % of the shortfall respectively. 26 Figure 21: retrospective hypothetical value of the stabilization account Table 1: decomposition of loss coverage Loss in oil Losses covered by revenues Remaining the stabilization compared to adjustment % coverage fund (provided budgeted through public enough past revenues expenditures savings) (bn of CFAF) 2014 230 100% 230 0 2015 382 66% 252 130 2016 193 33% 64 129 Total bn of CFAF 805 - 546 259 % of 2014 GDP 11,7 % - 7,9 % 3,8 % Note: this calculation assumes that the adjustment mechanism is only triggered in 2014. In fact, a small fall in the oil price in 2013 would have triggered the coverage mechanism and 2014 and 2015 shocks would have only been covered by 66 % and 33 % respectively. The possibility to define a minimum triggering threshold for shocks is explored below. Stochastic performance analysis In the baseline scenario, where oil prices follow the assumed central tendency, the value of the stabilization fund rises fairly quickly to around 9.5% of GDP in 2023, before stabilizing and falling gradually (see Figure 22). Debt and budget balance trajectories are unchanged, with the government exploiting all available margins. However, the switch to the new framework requires a downward adjustment of public spending in the first year as expenditures are adjusted downwards in accordance with the new revenue rule as the stabilization fund accumulates. 27 Figure 22: baseline value of the stabilization fund Real GDP growth (in %) Budget balance (% of GDP) Public debt (% of GDP) End-of-year stabilization account (% of GDP) Under stochastic simulation, the rule helps to smooth the impact of oil shocks (see Figure 23 and 24). Growth is being adversely affected in the short term due to the adjustment in public spending (around 1pp lower growth in 2019). Once this adjustment is made, the volatility of the economy decreases as the stabilization account reaches a value large enough to cover negative revenue shocks. Figure 23: GDP growth (in %) Without rule With CEMAC+ rule Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. 28 Figure 24: volatility of the economy under the CEMAC+ framework Public debt (in % of GDP) Budget balance (in % of GDP) Stabilization account (bn of CFAF) Stabilization account (in % of GDP) Note: stochastic simulations for 5000 draws. Confidence bands displayed at the 20 %, 40 %, 70 % and 90 % level. Deterministic performance analysis Looking more closely at specific oil price scenarios, the CEMAC+ rule helps smooth the impact of oil shocks on the Chadian economy (see Figure 25 and 26). In response to a transitory negative oil price shock of 20 USD in 2025, in the absence of a rule, public expenditures have to immediately adjust downwards to match the decline in revenues. This behavior reflects both the observed historical behavior of the Chadian government and its limited access to financial markets, which preclude the government going to financial markets to raise additional debt to cover the loss in revenue. With the CEMAC+ mechanism, initially there are only limited savings available in the stabilization account allow, so not much stabilization is realized, but over time this builds and volatility is reduced. 29 Figure 25: response of the economy to a transitory 20 USD two-year oil price shock in 2025 (% deviation from baseline) Real GDP Public expenditures Note : 2025 is chosen as the base year for the shock to allow the stabilization account to reach a positive and non-negligible value in the baseline scenario. If the account was to be empty, the CEMAC+ rule cannot offer any insurance against adverse shocks. Note that the persistence of the adjustment comes from the formula used in the model to simulate GDP forecasting for the upcoming year when the budget plan is being defined (ie. how is being forecasted real time to determine FSOR t ). In the present simulation, the GDP forecast is anchored on potential GDP and does not anticipate on the transitory or permanent nature of the oil price shock. Therefore, due to the adverse transitory oil price, the growth forecast is revised downwards for the current and following years as investment and subsequently potential output temporarily falls. This reduces budgeted oil revenues persistently and therefore associated expenditures by the Chadian government, with a persistent impact on GDP growth. Yet, if the shock is temporary and known to be so, it can be expected that an actual growth forecast using a less naïve method will lead to a more optimistic result. In the case of an adverse oil price shock, a more optimistic forecast will lead to higher budgeted revenues and therefore higher public expenditures. In other words, a more optimistic forecast would mitigate the persistent negative impact observed in Figure 26. In the case of a permanent decrease in the price of oil, the mechanism behaves in a similar manner. As the price of oil permanently adjust downwards, the mechanism smooths the adjustment towards a new state of the world where oil prices are permanently lower. It does not, however, eliminate the need to reduce spending to reflect the new “normal”. The role of the mechanism is therefore crucial in the transition but does not prevent the necessary structural and long-term changes in reaction to the lower price of oil. 30 Figure 26: response of the economy to a permanent 20 USD oil price shock in 2025 (% deviation from baseline) Real GDP Public expenditures Note : 2025 is chosen as the base year for the shock to allow the stabilization account to reach a positive and non-negligible value in the baseline scenario. If the account was to be empty, the CEMAC+ rule cannot offer any insurance against adverse shocks. Interest paid on the stabilization fund In addition to the mere accumulation of savings, one can also consider the fact that these savings will generate interest revenues. Depending on rules of use of those interest revenues, part might be returned to the government budget. A standard rule is to first use the revenues to protect the nominal value of the fund (i.e. cover real losses linked to inflation), and to allow the free use of the remaining revenues (typically, re-investment within the account or payment of dividends). In practice, following this rule, if the inflation rate is 3 %, the first 3 % paid in interests are accumulated within the account so that its value remains constant in real terms, including in the absence of other transfers into the account. If returns now exceed the inflation rate, the difference is returned to the general budget of the government. Simulations assuming a real return of 2 % on savings (i.e. a nominal return of 5 %) suggest interest revenues paid into the government budget (i.e. after nominal value protection deductions) of about 0.1% of GDP in 2025, and just over 0.2% of GDP in 2030. Yet, the return on savings will depend on the financial management of assets. Because the stabilization account should be highly liquid to allow for withdrawal in face of negative oil revenue shocks, the return on the assets might be lower than presented in the simulations. In addition, for Chad, the CEMAC monetary union rules impose that stabilization accounts of this nature should be housed within the central bank. In this case, the return on assets is marginal. 31 iv. Comparison of outcomes Simulations reveal that the unamended CEMAC framework is by and large not binding in the Chadian context of a combined borrowing constraint and a budget balance close to equilibrium today. It only allows to reduce growth volatility compared to a no-rule case by compressing growth at the upper bound, that is when oil prices are particularly high (see Figure 27). This inability to reduce volatility despite a rule targeted at smoothing oil revenues for the government over the commodity cycle stems from the incapacity for Chad to borrow on financial markets and therefore to use markets to transfer wealth/revenues inter-temporally in the absence of explicit savings. Introducing a stabilization account into which excess revenues are saved in good times and dissaved in bad times as in the price smoothing mechanism and the CEMAC+ rule allow a substantial reduction in volatility. At the 70 % confidence interval, both rules reduce the width of the GDP growth volatility interval by as much as 0.8 pp. In the short term (until 2021-2022), the price smoothing mechanism performs better than the CEMAC+ rule because it imposes lower savings for the constitution of the stabilization account and a smaller short- term cut in public expenditures. Over the medium term (2022-2026), the CEMAC+ rule performs by reducing the amplitude of downturns as the larger funds allows more gradual and persistent disbursements from the stabilization fund than the price smoothing mechanism (see Figure 28). In the long term, both mechanisms perform comparably and achieve a large reduction in GDP growth volatility compared to both the no rule and CEMAC rule cases. The preference given to the CEMAC+ rule resides in two elements: (i) its compliance with the CEMAC macro-surveillance framework, and (ii) its reliance on oil revenues (prices and volumes) rather than oil prices solely. Yet, both mechanisms operate in a very similar fashion with oil revenues being saved in good times in a stabilization fund and dissaved in bad times. Figure 27: real GDP growth volatility under different rules and mechanisms Confidence interval at 20% Confidence interval at 70% Note: Stochastic simulations for 5000 draws. 32 Figure 28: real GDP response to 20 USD oil price shocks in 2025 under different rules and mechanisms Transitory two-year shock Permanent shock 33 5. Conclusion Unexpected revenue shortfalls in resource-rich countries following adverse commodity price shocks can lead to severe fiscal and economic crisis, including illiquidity, debt distress, and severe recession. In the absence of any stabilizing revenue management mechanism in place, governments usually have no choice but to absorb the full extent of the shock through large expenditure cuts and arrears accumulation. To avoid such costly adjustments and manage commodity price and production volatility more efficiently, the key objective for a revenue management mechanism is to introduce a countercyclical dimension into public financing. To identify and calibrate an adequate and effective resource revenue management mechanism, a macro-structural model such as MFMod allows to explore various options to stabilize public expenditure in face of revenue volatility and shocks taking into account the interactions between variables. In the present paper, alternative oil revenue management mechanisms in the specific case of Chad were compared on five main dimensions: (i) minimizing GDP volatility, (ii) ensuring fiscal sustainability, (iii) the transparency and simplicity of the rule, (iv) the compliance with existing rules (i.e. the CEMAC macro- surveillance framework) and fiscal frameworks (the IMF Medium Term Fiscal Framework), as well as (v) with country immediate economic and social needs. For Chad, a smoothing mechanism alone does not significantly reduce volatility because the limited borrowing capacity of the government prevents it from using financial markets to supply savings during bad times. Introducing an explicit stabilization account with identified rules for savings/dissaving creates a pool of savings from which to draw during downturns and at the cost of slower growth in the near-term significantly reducing GDP volatility. Effectively, the revenue management mechanism acts as a commitment device that forces the government to save during good times, permitting dis-savings during bad. Both the simple price smoothing mechanism and the enhanced CEMAC framework achieve similar reduction in volatility. The CEMAC+ rule is marginally better because it is specifically tailored to Chad’s needs. It is compliant with the CEMAC macro-surveillance framework, smooths revenue volatility from both price and production shocks, and allows for a gradual adjustment in public expenditures in the context of structural changes in revenues. Based on the present analysis and with the support of the World Bank, the Government of Chad designed and enacted in November 2019 an oil revenue management mechanism for fiscal stabilization purposes. Economic and fiscal modeling has informed the determination of fiscal policy objectives and allows to assess the strengths and weaknesses of different mechanisms under various scenarios as well as the fiscal performance of the selected mechanism in a medium term macro-fiscal framework. Although the new mechanism differs in its formulation from those presented in this paper, its core structure is deeply rooted in the conclusions of the analysis. The key objective of the new mechanism is to set aside sufficient oil revenues to cushion the fiscal impact of large, unexpected oil revenue shortfalls. 2 The present analysis was conducted before the COVID-19 crisis, and the reformed revenue mechanism enacted by the government was put in just before the crisis. Because the saving mechanisms had only begun, neither the CEMAC+ nor the actual system will provide much in the way of relief to the revenue falls currently being experienced by the Chadian authorities. However, over time as the economy recovers 2 A detailed description and illustration of the functioning of this oil revenue management mechanism is provided in appendix B. 34 and savings are accumulated, it should help arm the government with the kind of reserves that will allow it to mount a more robust response to future shocks. All in all, an oil revenue management mechanism alone is not enough to bolster strong, sustainable and balanced growth in a resource-rich country. This is particularly true in the event of extreme price shocks such as illustrated by the COVID-related crisis, with associated loss in revenues most likely exceeding the value of any (hypothetical) stabilization account. Prudent fiscal management as well as increased economic diversification remain key elements to face crises of this magnitude. 35 References Arezki, R., Pattillo, C., Quintyn, M., & Zhu, M. 2012. Chapter 15. Mexico’s Oil Price–Hedging Program. Commodity Price Volatility and Inclusive Growth in Low-Income Countries. International monetary fund. Burns, Andrew; Campagne, Benoit Philippe Marcel; Jooste, Charl; Stephan, David Andrew; Bui, Thi Thanh. 2019. The World Bank Macro-Fiscal Model Technical Description (English). Policy Research working paper; no. WPS 8965. Washington, D.C. : World Bank Group. Blanchard, Olivier. 2018. On the future of macroeconomic models. Oxford Review of Economic Policy, Volume 34, Issue 1-2, Spring-Summer 2018, Pages 43–54. Cangiano, M., Curristine, T., and Lazare, M. 2013. Chapter 13: Public Financial Management In Natural Resource-Rich Countries. Public Financial Management and Its Emerging Architecture. International Monetary Fund. Carneiro, Francisco G. and Kouame, Wilfried A. 2020. Procyclicality trap in resource-rich countries and how to escape from it. Future Development Blog; Brookings Institution. CEMAC. 2017. Programme des Réformes Economiques et Financières de la CEMAC (PREF-CEMAC) Herrera Aguilera, Santiago; Kouame, Wilfried Anicet Kouakou; Mandon, Pierre Jean-Claude. 2019. Why Some Countries Can Escape the Fiscal Pro-Cyclicality Trap and Others Cannot ? (English). Policy Research working paper; no. WPS 8963. Washington, D.C. : World Bank Group. International Monetary Fund. 2012. Macroeconomic Policy Frameworks For Resource-Rich Developing Countries. Policy Papers. International Monetary Fund. 2019. Chad : Staff Report for the 2019 Article IV Consultation, Fourth Review under the Extended Credit Facility Arrangement, Request for Modification of Performance Criteria-Press Release; Staff Report; and Statement by the Executive Director for Chad. Country Report No. 19/258. Kitzmuller, Markus; Matta, Samer Naji; Kassim, Olanrewaju Malik; Nyman, Sara; Koschorke, Julian Alexander; Gebregziabher, Fiseha Haile. 2018. Escaping Chad’s growth labyrinth : disentangling constraints from opportunities and finding a path to sustainable growth (French). Washington, D.C. : World Bank Group. Kopits, M. G.; Symansky, M. S. A. 1998. Fiscal policy rules. No. 162. International monetary fund. World Bank. 2019. Chad - Second Programmatic Economic Recovery and Resilience Development Policy Financing Project (English). Washington, D.C. : World Bank Group. 36 Appendix A. Baseline response of the Chadian economy to an oil price shock In response to a permanent increase in the oil price of 10 USD in 2025 (see Figure A1), the only direct impact corresponds to increased revenues for the government. Under the business-as-usual behavior of the government, this increase in revenues leads to a corresponding increase in expenditures. This increase in expenditures will affect the real sphere through increased goods and services expenditures, transfers and capital expenditures. Increased public spending leads to an increase of 0.5 % in real GDP. The response is symmetric and roughly linear for larger increases or decreases in the price of oil. If the shock is implemented before 2025, the response of the economy differs as rising or falling oil prices will accelerate or slow Glencore-related repayments (see Figure A2). The presence of an oil price- dependent restructuring mechanism affects the fiscal space available to the government for non-debt management needs. With the Glencore mechanism, debt repayments rise in the presence of higher oil prices effectively reducing the funds available for spending on capital goods, current consumption and transfers. Symmetrically, but to a lesser extent, lower oil prices lead to reduce Glencore-related repayments allowing the government reducing the extent to which non debt-repayments spending must be cut from what it would have been otherwise. For example, faced with a 40USD oil price drop, in 2020 spending would have to fall by closer to 1 %, versus a 2 % drop had the price dependency mechanisms not been in place. 37 Figure A1: response of the economy following a permanent oil price shock in 2025 Real GDP (% deviation from baseline) Public revenues (% deviation from baseline) Public expenditures (% deviation from baseline) Budget balance (pp deviation from baseline) Debt to GDP ratio (pp deviation from baseline) 38 Figure A2: response of the economy following a permanent oil price shock in 2019 Real GDP (% deviation from baseline) Public revenues (% deviation from baseline) Public expenditures (% deviation from Budget balance (pp deviation from baseline) baseline) Debt to GDP ratio (pp deviation from baseline) 39 B. Functioning of the implemented oil revenue management The self-insurance mechanism aims to provide coverage against unexpected oil revenue shortfalls beyond 10 percent of budgeted (i.e. anticipated) oil revenues by constituting a stabilization fund. Based on the historical distribution of oil price fluctuations, such shortfalls roughly correspond to oil price reductions greater than US$5/bbl (ceteris paribus), an event that occurs with an average probability of 20 %. Yet, about half of all oil price reductions have historically been greater than US$5/bbl. The revenue management mechanism passed by the authorities in November 2019 consists of three elements (i) a saving rule; (ii) a withdrawal rule; (iii) a formula for estimating oil revenue in the Budget. Those elements are as follows: i. Saving Rule (i.e. inflows) 1. An annual amount of CFAF10 billion shall be paid into the Stabilization Fund – through quarterly payments. 2. In addition, if petroleum revenues exceed budgeted revenues, 20 percent of this difference shall be paid into the Stabilization Fund up to a maximum of CFAF 10 billion. Therefore, the minimum inflow per year is CFAF 10 billion and the maximum is CFAF 20 billion. 3. The maximum balance of the Stabilization Fund is capped at CFAF 40 billion . In absence of withdrawals, the Fund will reach full capacity over a period of minimum two years and maximum four years. 4. The maximum balance of the Stabilization Fund can be increased after two years of implementation by the Minister of MFB. ii. Withdrawal rule (i.e. outflows) 1. Withdrawals from the Fund occur automatically when actual oil revenues fall short from budgeted oil revenues by 10 percent or more. 2. Oil revenue shortfalls up to 10 percent of budgeted oil revenues will be accommodated through expenditure adjustment. 3. Any shortfall beyond 10 percent of budgeted oil revenues will be compensated subject to availability of resources in the Fund. 4. The Stabilization Fund may only be used to finance expenditure budgeted in a given fiscal year. The Fund may not be utilized for the satisfaction of any sovereign or commercial debt of the government, and no legal or beneficial interest in the Fund may be created. iii. Formula to budget oil revenues Budgeted oil revenues are estimated using conservative assumptions: Future oil prices are estimated with a US$3/barrel discount on the forecast published in the World Economic Outlook by the IMF. The forecasted volume of production is set 10 % below the projected production volumes estimated by petroleum companies operating in Chad. Illustrated behavior of the mechanism The baseline scenario assumes a constant oil price of oil of US$60. Figures B1 and B2 below illustrate how the oil revenue management mechanism for stabilization works under three alternative scenarios: • Scenario 1 corresponds to a fall in the price of oil of US$10 for one year. • Scenario 2 corresponds to a US$20 fall for one year. • Scenario 3 entails a US$10 increase in the price of oil. 40 Note that the revenue shortfalls under Scenarios 1 and 2 occur immediately, i.e. in year 1 and with a Fund balance of only CFAF 10 billion. Under the baseline, if the price of oil stays at US$60, the stabilization account is expected to be filled by the end of the fourth year with an average savings flow of 10 bn to CFAF 13 bn during the first three years. In Scenario 1, in the case of a fall in the oil price by US$10 during the first year, the shortfall in revenues (actual revenues – budgeted revenues) is expected to be 8 bn CFAF. This represents less than 10 % of the budgeted revenues of CFAF 337 bn. As such, dissaving from the account is not activated and the transfer into the account is set at its minimum value of CFAF 10 bn for this year. In Scenario 2, the fall in the oil price in the first year generates a shortfall in oil revenues of CFAF 74 bn, enough to trigger dissaving from the account. In this case, expected revenues were CFAF 33 bn so that only CFAF 40 bn can be dissaved from the account (a 74 bn shortfall minus 10 %of budgeted oil revenues, i.e. 34 bn). As the value of the account is insufficient to withdraw 40 bn, 10 bn is withdrawn and the account is fully depleted at the end of the year. Of course, if the oil revenue shortfall occurred after 4 years, the fund would be at full capacity (CFAF 40 bn) and able to accommodate the full amount under the mechanism (in this scenario exactly CFAF 40 bn). Finally, in Scenario 3, the increase in the price of oil generates an acceleration in the ramp up of the fund. As actual oil revenues are of CFAF 460 bn, this corresponds to excess revenues of CFAF 123 bn. Of this amount, 20 % corresponds to CFAF 25 bn and therefore exceeds the minimum threshold for savings of CFAF 10 bn. It also exceeds the maximum flow of CFAF 20 bn so that the savings for this year is set to the maximum of CFAF 20 bn. In this scenario the account reaches full capacity by the end of the third year. Figure B1: oil price scenarios Price of oil (USD/bbl) The baseline corresponds to a price of oil constant at 60 USD. Scenario 1 corresponds to a fall in the price of oil of 10 USD for one year. Scenario 2 corresponds to a 20 USD fall for one year. Scenario 3 corresponds to a 10 USD increase in the price of oil for one year. 41 Figure B2: illustration of the oil revenue management Savings into the account End-of-year value of the stabilization account (bn of CFAF) (bn of CFAF) Budgeted oil revenues Actual oil revenues (bn of CFAF) (bn of CFAF) The baseline corresponds to a price of oil constant at 60 USD. Scenario 1 corresponds to a fall in the price of oil of 10 USD for one year. Scenario 2 corresponds to a 20 USD fall for one year. Scenario 3 corresponds to a 10 USD increase in the price of oil for one year. 42