Content uploaded by Babatunde S. Omotosho
Author content
All content in this area was uploaded by Babatunde S. Omotosho on Oct 13, 2024
Content may be subject to copyright.
Working Paper
Mitigating the Impact of Fuel Subsidy Removal in an Oil-
Producing Emerging Economy
Norges Bank Research
Authors:
Junior Maih
Babatunde S. Omotosho
Bo Yang
Keywords
Fuel subsidy, DSGE model,
Regime switching, Policy analysis,
Nigeria
15 | 2024
Norges Bank Working Paper 1
Working papers fra Norges Bank, fra 1992/1 til 2009/2 kan bestilles på e-
post: servicesenter@norges-bank.no
Fra 1999 og senere er publikasjonene tilgjengelige på www.norges-bank.no
Working papers inneholder forskningsarbeider og utredninger som vanligvis ikke
har fått sin endelige form. Hensikten er blant annet at forfatteren kan motta
kommentarer fra kolleger og andre interesserte. Synspunkter og konklusjoner i
arbeidene står for forfatternes regning.
Working papers from Norges Bank, from 1992/1 to 2009/2 can be ordered
by e-mail: servicesenter@norges-bank.no
Working papers from 1999 onwards are available on www.norges-bank.no
Norges Bank’s working papers present research projects and reports (not
usually in their final form) and are intended inter alia to enable the author to
benefit from the comments of colleagues and other interested parties. Views
and conclusions expressed in working papers are the responsibility of the
authors alone.
ISSN 1502-8143 (online)
ISBN 978-82-8379-335-2 (online)
Mitigating the Impact of Fuel Subsidy Removal in an
Oil-Producing Emerging Economy
*
Junior Maih
Babatunde S. Omotosho
Bo Yang
§
August 9, 2024
Abstract
This paper examines the implications of fuel subsidy removal in an oil-producing econ-
omy, focusing on the central bank’s response to volatile oil prices. Using a Markov-switching
dynamic stochastic general equilibrium model, we analyze the welfare effects of this policy
change under different regimes of oil price volatility and monetary policy. Our empirical
findings, based on data from Nigeria (2000:2 - 2021:4), reveal time-varying switches in oil
price fluctuations and monetary policy adjustments that synchronize with states of high
oil price volatility. We also find that subsidy removal has welfare-reducing and heteroge-
nous effects on households, especially when implemented in an environment of heightened
volatility. The efficacy of monetary policy in mitigating the impacts of subsidy removal de-
pends on the ability of the central bank to design a flexible framework capable of adapting
to economic shifts, while balancing its stabilization objectives. Furthermore, the observed
monetary policy switching endogenous to different states of oil price shocks suggests a need
for the central banks of oil-producing emerging economies to consider the prospects of a
dual-mandate regime.
JEL Classification: C32; E37; Q43
Keywords: Fuel subsidy; DSGE model; Regime switching; Policy analysis; Nigeria
*
This Working Paper should not be reported as representing the views of the African Development Bank or
Norges Bank. The views expressed are those of the authors and do not necessarily reflect those of the African
Development Bank or Norges Bank. This research was supported by the British Academy [grant number SG2122-
210952]. Earlier versions of this paper were presented at the Economic Policy Directorate Seminar, Central Bank
of Nigeria, 28 August 2023, the CReMMF Workshop on Macroeconomic Policy and Growth, Swansea University,
15 September 2023, the 99th Annual WEAI Conference, 29 June - 3 July, 2024, and the 2nd RISE Workshop,
25 - 26 July, 2024. We acknowledge comments by participants at these events. Thanks to the Central Bank of
Nigeria for their hospitality and stimulating discussions.
Norges Bank, Oslo, Norway. E-mail: junior.maih@norges-bank.no
Statistics Department, African Development Bank, Abidjan, Cote d’Ivoire. E-mail: b.omotosho@afdb.org
§
Department of Economics, School of Social Sciences, Swansea University, United Kingdom. E-mail:
bo.yang@swansea.ac.uk
Contents
1 Introduction 2
2 Related literature 5
2.1 Nonlinearities of oil price shocks ........................... 5
2.2 Episodic switches in DSGE frameworks ....................... 6
2.3 SOE models for oil-importers and -exporters ..................... 7
2.4 Macroeconomic effects of fuel subsidy reforms .................... 7
2.5 Outline of contributions ................................ 8
3 The Regime-switching DSGE model 8
3.1 Oil production and pricing ............................... 9
3.2 Fuel subsidy and fiscal policy ............................. 10
3.3 Monetary policy and the switching Taylor rule ................... 12
3.4 Perturbation solution of the model .......................... 12
4 Parameterization and filtration implications 13
4.1 Data and calibration .................................. 13
4.2 Bayesian estimation .................................. 14
4.3 Smoothed transition probabilities ........................... 15
5 Macroeconomic implications 17
5.1 Responses of fuel subsidy removal ........................... 18
5.2 Economic performances under counterfactual scenarios ............... 19
6 Macroeconomic stabilization and optimal policy 20
6.1 The central bank’s role ................................. 20
6.2 Optimized Taylor rules ................................. 22
6.3 Unconditional standard deviations .......................... 23
6.4 Volatility implications of optimized rules ....................... 24
6.5 Impulse responses to a negative oil price shock ................... 24
7 Statistical validation and sensitivity analysis 27
8 Implications for macroeconomic policy 28
9 Concluding remarks and future research 31
1 Introduction
In the context of global debates surrounding fossil fuel consumption subsidies, particularly in
resource-rich emerging economies which are exposed to high volatility of international commod-
ity prices, this paper addresses two critical research questions: How should monetary policy
respond to the macroeconomic implications of fuel subsidy removal in an oil-producing emerg-
ing economy, considering regime-switching dynamics in oil price volatility? How should central
bank policy rules adapt to mitigate the adverse effects of subsidy removal?
Fuel consumption subsidies, a widespread global phenomenon, have been implemented to
stimulate production, reduce inequality, alleviate energy poverty, and stabilize domestic prices
(Estache and Leipziger,2009;Taylor,2020). In 2017, a total of 191 countries accounted for
global fossil fuel subsidies amounting to USD 5.2 trillion, representing 6.5% of global GDP
(Coady et al.,2019). In line with the continued relevance of oil as a source of energy for
households and firms, the amount of energy subsidies increased to USD 6 trillion (7% of global
GDP) in 2020, with the top three subsidizers being China (USD 2.2 trillion), the US (USD
0.66 trillion), and Russia (USD 0.52 trillion). Interestingly, oil-producing developing countries
accounted for about 48.3% of total pre-tax global subsidies in 2017, highlighting the significance
of fossil fuel subsidies for those countries (Estache and Leipziger,2009).
However, concerns have emerged regarding the fiscal costs and negative externalities of fuel
consumption subsidies, prompting calls for reforms.1Indeed, fuel subsidies involve sizeable fiscal
costs which hinder growth, especially in oil-producing developing countries; thus, limiting the
capacity of such countries to mobilize the needed fiscal resources for addressing their develop-
mental challenges.2More importantly, ineffective fuel consumption subsidies distort domestic
price signals and complicate the monetary policy environment. The complete removal of fuel
subsidies in Nigeria in May 2023 has intensified policy debates, reflecting the ongoing challenge
of balancing macroeconomic stability with the imperative for reform.
Historically, subsidy reforms have been among the most challenging fiscal adjustments for
governments, primarily due to the complex political economy surrounding such reforms (In-
chauste and Victor,2017). The reluctance to implement reforms stems from potential socio-
political implications, especially in the absence of safety nets for vulnerable populations.3In
Nigeria, while the calls for subsidy removal have been largely driven by the inefficiencies in its
implementation, the associated fiscal costs, and its roles in widening the already existing wealth
disparity and worsening the country’s debt profile have indeed intensified the debate (Omotosho,
2019b).4While there is uncertainty as to the macroeconomic implications of subsidy removal,
one thing that is clear is that the debate on the appropriateness of necessary complementary
1Fuel consumption subsidies have been documented to be associated with the issues of widening inequality,
triggering inefficient energy consumption, negative environmental effects, market distortions, crowding-out effects
in public spending, as well as balance of payments and fiscal imbalances (Clements et al.,2013;Coady,2015;
Taylor,2020).
2For example, the size of fuel subsidies in Nigeria posed a significant fiscal burden on the distribution of
resources to the state governments.
3For instance, reforms that could result in higher retail energy prices could have non-trivial socio-political
implications, especially in situations where there are no safety nets for the poor and vulnerable.
4Several attempts made under different administrations to abolish the country’s fuel subsidy program have
been unsuccessful. For instance, the attempt in January 2012 to introduce total subsidy removal was met with
fierce resistance by the citizens, causing the decision to be rescinded within two weeks.
2
policies to support the reform continues to evolve.
To explore this issue, we choose Nigeria, which is a large net oil exporter, for two main
reasons. First, Nigeria characterizes the typical resource-rich emerging markets. According
to the Central Bank of Nigeria (CBN) Statistical Bulletin, since 2000, the oil and gas sector
in Nigeria has accounted for about 40% of GDP, 72% of government revenue, and 95% of
exports earnings. Nigeria continues to depend heavily on oil with less diversification towards
non-oil exports and industrialization. Thus, the economy is considerably vulnerable to oil price
fluctuations. The mono-product nature of exports and fiscal revenue makes macroeconomic
outcomes susceptible to the vagaries of oil prices. Second, our focus is on monetary policy.
Resource abundance can be a major cause of weak institutions and economic mismanagement
since the fiscal authorities often face the political incentives generated by a resource endowment.
Given the political expediencies faced by Nigeria’s fiscal authority, combined with the country’s
low level of export diversification, policy for maintaining macroeconomic stability and managing
aggregate demand may largely lie with the monetary authority.
Furthermore, large external price shocks have been a major force behind the recent inflation-
ary pressures in Nigeria, given the openness of the small open-economy (SOE) and the country’s
fuel imports requirements. The recent episodes of heightened volatility in relative price move-
ments which get factored into inflationary expectations have made it difficult for central banks
to achieve price stability while maintaining economic growth. Indeed, Kilian and Vigfusson
(2017) and Hwang and Zhu (2024) find that monetary policy behaves differently depending on
how inflation is affected by oil price shocks. Meanwhile, in the face of increasing uncertainty
about future prices which appears to be emanating from the supply side, a major challenge
before policymakers remains the need to maintain macroeconomic stability and a stable fis-
cal profile while discounting volatile oil sector developments so that the effect of fuel subsidy
reforms can be transformed into development gains.
Many studies have focused on assessing the macroeconomic implications of subsidy reforms
on the assumption of Gaussianity of the shocks that characterize the domestic economy.5How-
ever, relatively little exists on the optimal interactions of monetary policy and the economic
consequences of oil price volatility using micro-founded macroeconomic analysis for an emerg-
ing SOE and no consensus has emerged in the literature. Additionally, one of the issues that
remains open to debate is whether the welfare consequences and stabilization incentives are
expected to be different as regime-switching occurs in oil price fluctuations. Existing studies
primarily focus on the macroeconomic response to subsidy reforms but often overlook the dy-
namic nature of economic conditions, particularly the role of stochastic regime shifts, which are
essential considerations in the design of subsidy policies.
Indeed, past research in the empirical literature studying the consequences of fuel subsidy
reforms is relatively silent on the mechanisms through which oil price shocks affect the economy
through, for example, the connection between regime-switching and monetary policy. It leaves
unanswered, the more fundamental question of what leads the policymaker to behave differently
over time. Answering this question requires a model with richer dynamics, accounting for both
the key features of resource-rich emerging economies and the possibility of the variance of shocks
5See Omotosho (2019b) and Omotosho (2022) for an overview of this literature.
3
changing over time.
To fill this gap, we incorporate regime-switching dynamics into a dynamic stochastic gen-
eral equilibrium (DSGE) model tailored to the Nigerian context. Within this framework, we
model the heteroskedasticity of oil prices in order to account for the distinct periods of oil
price volatility, and thus the considerable uncertainty about the nature of the shock faced by
policymakers. Our approach, directly linking the significant increase in price volatility and the
behavior of policymakers responding to these dynamics, plays a central role in our analysis and
is completely novel in the context of an estimated two-bloc regime-switching model. This study
aims at gauging the empirical relevance of fuel subsidies and how they affect macroeconomic
volatility and monetary policy. Our primary objectives include evaluating optimal monetary
policy responses to an oil price shock, assessing welfare implications of alternative policy sce-
narios, assessing the historical implications of fuel subsidy policies on economic performances,
and understanding the potential impacts of reforms on macroeconomic stability.
To achieve our objectives, we employ empirical analysis based on data from 2000:2 - 2021:4.
The model parameterization considers the distinctive features of emerging economies, partic-
ularly oil-producing ones, to capture the dynamic interactions among oil price innovations,
economic variables, and policy interventions. The model features an oil producer and a fis-
cal authority that governs the level of fuel subsidies. The subsidy program determines the
pass-through effect of international oil price into domestic prices. The monetary authority fol-
lows a simple Taylor rule responding gradually to aggregate inflation, domestic output and the
exchange rate.6To examine the interactions between oil price volatility and monetary pol-
icy adjustments, our model incorporates stochastic regime shifts in the oil price volatility and
monetary policy rule coefficients, highlighting the link between switching dynamics and policy
implications. Thus, we make contributions to the existing literature by discussing a range of
practical and theoretical implications, shedding more light on the underlying mechanism that
guides policy choices for a successful fuel subsidy reform. The main findings of this paper are
summarized as follows.
We start by examining the implications for dynamic monetary policy response. The paper
finds substantial empirical support for time-varying switches between major volatile episodes
in oil prices and monetary policy adjustments that synchronize with states of high oil price
volatility. The results show that the major volatile episodes are observed during 2008-2009,
2014-2016, and 2020-2021, corresponding to the US credit crisis of 2008, the supply-driven
collapse in oil prices that led to an economic recession in Nigeria in 2016, and the more recent
drop in oil demand starting from 2019Q4 triggered by the pandemic, respectively.
Next, we examine the estimation results and discuss the extent to which the central bank
adjusts its behavior in response to oil price volatility. First, during highly volatile oil price
periods with heightened uncertainty faced by policymakers, the central bank is less sensitive to
movements in headline inflation, focuses less on exchange rate stabilization, but places greater
emphasis on the output gap and adjusts interest rates more quickly. Second, we compare the
economic performances under the estimated policy rule and the counterfactual scenario where
6The modeling of monetary policy is consistent with the operations of the CBN which currently adopts a
monetary targeting framework, uses the Monetary Policy Rate as a key instrument for signalling monetary
policy stance, and has recently moved to a regime of market-based exchange rate determination.
4
subsidies were not in place historically. The macroeconomy would have been more volatile with
the realized shocks and policy rule. With the additional income available, consumption rises
initially, but low-income consumers experience comparatively smaller increases.
To further understand the implications of removing fuel subsidies, we solve the model for
given policy and then evaluate welfare using this solution. We find that a complete, one-off
removal is welfare-reducing in an environment with recurrent periods of exogenous, volatile oil
price shocks. Counterfactual simulations reveal that subsidy removal leads to higher macroe-
conomic instabilities and welfare cost of the business cycle. The effects of subsidy removal on
private consumption and future retail energy prices are possible explanations for the different
welfare consequences, implying that this policy change has non-trivial socio-political implica-
tions.
Finally, the optimal policy operation that accommodates oil price volatility has interesting
implications for macroeconomic policy. In the regime of high oil price volatility, the optimal
monetary response prescribes a more aggressive inflation response compared to the regime
characterized by low price volatility. The impulse responses of the interest rate to economic
shocks based on the realized rule differ from those generated in an economy when the central
bank adopts optimal policy. When shocks are large and volatile, the best response prescribes
an initial cut to interest rates, which is in contrast with the results generated by the estimated
rule. In addition, monetary policy faces a less severe trade-off between price stability and output
stabilization in the absence of fuel subsidy but a regime of high oil price volatility is associated
with worsening policy trade-offs.
The paper is organized as follows. Section 2reviews related literature to provide context
for our study. Section 3outlines the baseline model and introduces the Taylor-type monetary
policy rules. Section 4presents the model parameterization and filtration implications. Section
5discusses the historical and counterfactual implications of fuel subsidy policies on economic
performances. While Section 6.1 details the central bank’s role and its linear-quadratic problem,
Sections 6.2 and 6.5 discuss results and analysis, including optimal monetary policy, welfare
assessment, and impulse responses. Section 7explores the statistical validation of the model
and robustness of the results based on posterior simulations. Section 8discusses the implications
of our results for macroeconomic policy. Finally, Section 9concludes, summarizing key findings
and suggesting avenues for future research.
2 Related literature
There are four strands of literature related to our paper, which are discussed in this section.
Additionally, we discuss the contrast between these relevant pieces of literature and our approach
to further clarify the contribution of this paper.
2.1 Nonlinearities of oil price shocks
The first strand is a largely econometrics literature studying oil price shocks, which have been
known to generate macroeconomic instability in many resource-rich countries. This strand
of literature explores the econometric aspects of oil price shocks, dissecting their asymmetric
5
output effects and state dependence. Notable studies such as Barsky and Kilian (2004), Kilian
(2009) and Ramey and Vine (2011) delve into the sources of these shocks, while others like
Rahman and Serletis (2010), Holm-Hadulla and Hubrich (2017) and Hwang and Zhu (2024)
investigate central banks’ responses, accounting for time-varying impacts.
The major discussion under this strand of literature is around the changing estimated effects
of oil price shocks, and the use of alternative methods for studying these changing effects. This
motivates us to take account of time-dependent nonlinearities in modeling the impact of oil
shocks. However, studying episodic nonlinearities of macroeconomic shocks is difficult and
requires nonlinear techniques. SVAR-based studies that explicitly allow for nonlinearities find
mixed evidence but the existence of a VAR representation can be compromised due to non-
invertibility/-fundamentalness.7
Although useful for understanding the degree of macroeconomic co-movements, nonlinear
multivariate models and dynamic factor models do not provide much information about the
mechanisms through which oil price shocks affect the macroeconomy. Furthermore, to be useful
for optimal policy design, we must require a data-based DSGE model that provides the struc-
tural investigation from the richer dynamics and model-implied moments behind the estimated
parameters. Invariably, the contradicting findings earlier mentioned allude to the importance
of a need to ascertain the appropriate transmission channel(s) through which oil price shocks
affect both the oil and non-oil sectors of the economy in order to proffer optimal policies using
open-economy DSGE models.
2.2 Episodic switches in DSGE frameworks
The second piece of literature focuses on episodic switches in DSGE frameworks, introducing
Markov chains for macroeconomic volatility and structural parameters. Empirical studies, in-
cluding Schorfheide (2005) and Liu et al. (2011), employing regime-switching DSGE models,
reveal substantial evidence of structural shifts.8
A number of recent papers are related to the present paper. Chen and Macdonald (2012)
focus on UK monetary policy and study its stabilization properties in a DSGE model that
is subject to several regime shifts. Bjornland et al. (2018) construct and estimate a regime-
switching model that studies the roles of oil prices and monetary policy in the US economy
for the timing of the Great Moderation. Best and Hur (2019) evaluate the role of monetary
policy with time-varying volatilities of non-policy shocks. Maih et al. (2021) investigate the
implications of asymmetric monetary policy rules for the Euro area and the US based on a
sample that encompasses the Great Recession and periods of financial distress.
In our paper, we study the macroeconomic implications under different policy scenarios
that may be affected by changing oil price volatility and modeling uncertainties based on a
model tailored to incorporate unique economic features for an oil-rich emerging economy. The
paper elects to stay closer to the current consensus on synchronized-switching models, but
offers a number of innovations that address the new challenges. In contrast to the previous
7See, for a detailed discussion, Levine et al. (2022) and Levine et al. (2019).
8Recent applications also include Liu and Mumtaz (2011), Bianchi (2013), Davig and Doh (2014), Bianchi
and Ilut (2017) and Bianchi and Melosi (2017), among others.
6
papers, the stochastic switch in our model assumes that responses of the real economy depend
on the volatility and persistence of oil price shocks.9In addition, we implement a hybrid,
flexible framework to bridge the behavior of policy and the heteroskedasticity of oil prices.
This is particularly relevant to the behavior of monetary policy adjustments that is affected by
significant increases in oil price volatility.
2.3 SOE models for oil-importers and -exporters
The third strand of literature explores small open-economy (SOE) DSGE models investigating
the macroeconomic impacts of oil price shocks and policy design, particularly in oil-importing
and -exporting nations. Omotosho (2019a) estimates an open-economy model using Nigerian
data and focuses on sources of macroeconomic fluctuations and inflation dynamics. A com-
prehensive review by Omotosho (2022) provides insights, covering various monetary policy re-
sponses, exchange rate regimes, and levels of oil intensity dependence. The variations in findings
are also explained and discussed in Medina and Soto (2005), Allegret and Benkhodja (2015),
Ferrero and Seneca (2019), Bergholt and Larsen (2016) and Algozhina (2022).
Different from the previous SOE studies, our paper focuses on the connection between
regime-switching dynamics and policy implications, and assessing the role that the central bank
can play in mitigating the consequences of fuel subsidy removal. Against inherent uncertainties,
it is evident that the determination of crude oil prices exhibits a volatile process which poses
considerable challenges to macroeconomic stability and demand management in oil-exporting
emerging economies. Our benchmark form of an open-economy model applies to oil-producing
emerging market economies. We fit the oil-exporting regime-switching model to Nigerian data
using Bayesian methods. The latter provides an empirical assessment of how different stabiliza-
tion actions affect the macroeconomic outcomes.
2.4 Macroeconomic effects of fuel subsidy reforms
Empirical evidence initiated by Hamilton (2003) and recent studies, such as Clements et al.
(2013), Siddig et al. (2014), Dennis (2016), Rentschler et al. (2017) and Coady et al. (2019),
delves into the macroeconomic effects of fuel subsidy reforms. Notably, Fan and Wang (2022)
assess the net social welfare effect of China’s petroleum pricing mechanism reform. Siddig
et al. (2014) study the effect of subsidy reduction on consumption, income and fiscal planning
in Nigeria. Much of the empirical studies finds non-trivial implications for the response and
volatility of macroeconomic variables. The main predictions of these studies show that fuel
subsidy reforms could cause inflation, reduce economic welfare, distort fiscal planning, reduce
household income, and worsen the problem of inequality. However, research within a general
equilibrium framework remains limited.
Building on the earlier contributions by Omotosho (2019b), this paper departs from pre-
vious studies and focuses on evaluating Nigeria’s potential subsidy reforms by incorporating
fuel subsidy into a DSGE model and is closely related to the work of Omotosho (2019b) and
9In line with the findings in Omotosho and Yang (2024), our assumption is based on the observation that the
monetary authority in Nigeria has responded differently to different oil price shocks in the past, depending on
the size and persistence of the shock and the state of the economy prior to the shock.
7
Omotosho and Yang (2024). These papers examine the pass-through effects of oil price shocks
on domestic fuel prices, investigating macroeconomic volatility and dynamics. However, this
literature lacks an exploration of policy formulation and welfare implications in the context of
potential subsidy reforms, considering dynamic interactions between stochastic regime shifts in
the nature of oil prices and time-varying central bank adjustments, and subsidy reforms – a gap
addressed by this paper.
2.5 Outline of contributions
In the light of this review, our paper makes the following three main contributions to the
literature. First, our paper emphasizes the crucial importance of the heteroskedasticity of oil
prices in studying the behavior of monetary policy adjustments. Second, our paper constructs
and estimates a benchmark form of an open-economy micro-founded macroeconomic model that
captures key features of oil-producing emerging market economies. Third, in our application,
we study simple Taylor-type monetary policy rules that are ‘operational’, in the sense that they
are easy for the public to monitor, whilst approximating the stabilizing properties of complex
optimal rules. Indeed, our empirically based theoretical approaches have properties that make
them particularly suitable for policy analysis and the design of monetary policy frameworks
should exhibit flexibility and adaptability. In addition to providing the important insights into
optimal policy responses and welfare consequences drawn from Nigeria’s subsidy removal, our
paper offers the core guiding/controlling principles to implement fuel subsidy reforms for similar
economies facing similar challenges.
3 The Regime-switching DSGE model
As in Omotosho and Yang (2024), the model features the SOE and the foreign economy and
presents the regime-switching monetary policy. There are four categories of firms operating in
the economy: the final goods firm, the intermediate goods producing firms, the foreign goods
importing firms, and the oil-producing firm. The economic environments in which the first three
categories of firms operate are standard. For the oil sector that is owned by the government
and foreign investors, there are three departures from the standard open-economy model that
lead to interesting results. First, oil enters firms’ production technology and results in a direct
impact of oil shocks on the supply side (Medina and Soto,2005 and Ferrero and Seneca,2019).
Second, in the oil market, the government sells the imported fuel based on a fuel pricing rule
that connotes an implicit subsidy regime as in Allegret and Benkhodja (2015). Third, there
are frictions in the financial markets facing households as in Gabriel et al. (2023) and in the
form of non-Ricardian consumers to capture credit constraints,10 and an inefficient financial
sector as in Smets and Wouters (2007). Furthermore, we allow for the law of one price (LOP)
gap in imports and by implication assume incomplete exchange rate pass-through into import
10The presence of rule of thumb consumers who have no access to formal financial services and credit to
smooth out consumption should improve the model fit for an emerging economy in the type of volatile economic
environment that has been described so far (see Gabriel et al.,2016 and Gabriel et al.,2023, among others).
8
prices as in Monacelli (2005) and Senbeta (2011).11 We assume that our model can switch
exogenously between regimes of oil price volatility and the monetary policy rule over time. The
main elements of the model are as follows.12
3.1 Oil production and pricing
The oil firm’s profit maximization problem is similar to that of Ferrero and Seneca (2019) and
Algozhina (2022). The firm is owned by the government and foreign investors and combines
materials sourced from the domestic economy, Mt, and oil-related capital, Ko,t , to produce
oil output, Yo,t, which is exported to the rest of the world at a price, P∗
o,t, determined in the
international crude oil market, using the following Cobb-Douglas technology
Yo,t =Ao,tKαk
o
o,t Mαm
o
t(1)
where Ao,t represents the oil technology. αk
oand αm
o∈(0,1) represent the elasticities of oil
output with respect to Ko,t and Mt, respectively. The former is accumulated by foreign direct
investment (FDI), F D I∗
t, as follows
Ko,t = (1 −δo)Ko,t−1+F D I∗
t(2)
where δois the depreciation rate. The intuition of (2) follows closely the assumption made
in Melina et al. (2016) and Algozhina (2022). The natural resource sector in oil-exporting
developing and emerging countries attracts capital inflows from the rest of the world in the form
of FDI. Melina et al. (2016) argue that the decisions for resource production and developments in
these countries typically happen via negotiations between governments and foreign multinational
firms. As a result, FDI can be thought as the outcome of these negotiations and is accumulated
to create Ko,t used in (1).13 F DI ∗
tinflows to the oil sector respond to the real international
price of oil, P∗
o,t, as follows
F DI ∗
t=F DI ∗
t−1ρfdi P∗
o,t1−ρf di (3)
where ρfdi measures the extent of inertia in the accumulation of F DI ∗
t.
The oil firm receives its revenues net of royalties levied by the government on production
11As supported by empirical literature on Nigeria, there is incomplete exchange rate pass-through of imports
to domestic prices. Various studies have estimated the level of exchange rate pass-through for Nigeria (see, for
example, Oyinlola and Adetunji,2009 and Adebiyi and Mordi,2012).
12As is standard in most DSGE models, we assume that wages as well as prices of domestically produced
goods are sticky. Also, an investment adjustment cost is incorporated into the model to generate hump-shaped
investment response to shocks. For simplicity, parts of the model associated with the wage setting dynamics,
behaviors of non-oil firms, interaction between the SOE and the foreign economy, and analogous ‘foreign’ variables
are largely omitted in the exposition. See Omotosho and Yang (2024) for details.
13The role of FDI inflows aimed at oil and gas resource production is important in countries and regions that
are new to resource development and may have weak institutions and tax systems. Previous literature that
investigates the importance of FDI in the natural resource sector finds a significant correlation between FDI and
resource endowments and studies the role and regulations of host governments in supporting large-scale resource
developments with foreign investments. See, for example, Asiedu and Lien (2011), Teixeira et al. (2017) that
apply the dynamic panel data analysis, and Goldwyn and Clabough (2020) that use the narrative approach based
on case studies.
9
quantity at a rate τas follows
Πo,t = (1 −τ)εtP∗
o,tYo,t (4)
where εtis the nominal exchange rate. The oil firm’s profits are fully taxed. It is clear that a
shock to P∗
o,t affects the firm’s production and demand for capital and materials.
We assume that P∗
o,t and Ao,t evolve according to the following AR (1) processes
P∗
o,t =P∗
o,t−1ρP∗
o(svol
t)exp σP∗
o(svol
t)ξP∗
o
t, Ao,t = (Ao,t−1)ρAoexp σAoξAo
t(5)
In order to capture possible nonlinearities in the response of the resource-rich economy to
oil price instabilities, we allow the volatility of the oil price shock, σP∗
o(svol
t), to change from
one regime to another as follows
svol
t∈ {High, Low}(6)
To take into consideration the possibility where the responses of the real economy may
depend also on the persistence of the shock, we also restrict the persistence parameter, ρP∗
o(svol
t),
to follow a Markov chain that switches at the same time, but not necessarily in the same
direction.
3.2 Fuel subsidy and fiscal policy
We assume that the government respects a budget constraint given by
T Xt+ORt+Bt=Pg,t Gc,t +OSt+Bt+1
Rt
(7)
where (7) shows that an increase in government expenditure, Gc,t , consisting of imported goods
and domestically produced goods, can be financed either by increasing per-capita lump-sum
taxes levied on households, T Xt, generating more oil revenues collected from oil royalties, ORt,
or issuing more debt, Bt. On the payment side of (7), Bt+1
Rtrepresents the interest payments
on Bt. When the need arises, the government makes refined oil subsidy payments, OSt, within
a framework that allows for the stabilization of domestic fuel price. Pg,t is the deflator of
government expenditure.
As in Medina and Soto (2007), we assume that the government consumption basket consists
of imported goods, Gf,t, and domestically produced goods, Gh,t
Gc,t ="(1 −γg)
1
ηgG
ηg−1
ηg
h,t +γ
1
ηg
gG
ηg−1
ηg
f,t #
ηg
ηg−1
(8)
where ηgis the elasticity of substitution between Gf,t and Gh,t.γgis the share of foreign goods
in the consumption basket.
Standard results from the government optimal intra-temporal decisions subject to (8) and
the usual Dixit-Stigitz aggregation yield the demand functions for Gh,t and Gf,t, respectively
Gh,t = (1 −γg)Ph,t
Pg,t −ηg
Gc,t, Gf,t =γgPf,t
Pg,t −ηg
Gc,t (9)
10
and the government consumption price index given by
Pg,t =h(1 −γg)P1−ηg
h,t +γgP1−ηg
f,t i1
1−ηg(10)
Following Allegret and Benkhodja (2015), we assume that aggregate refined oil, Ot, is pro-
duced abroad and imported into the SOE at a landing price, Plo,t, by the government. In turn,
the government sells the imported fuel to households and domestic firms, at a regulated price,
Pro,t, based on a fuel pricing rule given by
Pro,t =P1−ν
ro,t−1Pν
lo,t (11)
where 0 ≤ν≤1 governs the extent to which the government subsidizes fuel consumption.
When ν= 1, the implicit subsidy regime ceases to exist whereas ν= 0 implies complete
price regulation. Plo,t, which is the prevailing landing price of refined oil expressed in domestic
currency, is given by14
Plo,t =εt
P∗
o,t
P∗
t
Ψo
t(12)
where P∗
o,t is the foreign currency price of oil abroad, Ψo
tis the LOP gap associated with the
import price of fuel, and P∗
tis aggregate consumer price index of the foreign economy.
Thus, the implicit fuel subsidy payment is given by the difference between the value of fuel
imports expressed in domestic currency and the amount realized from fuel sales in the domestic
economy as follows
OSt= (Plo,t −Pro,t )Ot(13)
where total imported fuel (Ot) comprises fuel consumption by households, Co,t, and consumption
by domestic firms, Oh,t.
On the revenue side of the budget constraint, (7), the amount of oil revenues, ORt, accruing
to the government are given by
ORt=τ εtP∗
o,tYo,t (14)
Following Algozhina (2022), we consider backward looking fiscal policy reaction functions
that allow government consumption and taxes to respond to lagged debt, ORtand OSt
Gc,t
¯
G=Gc,t−1
¯
Gρg"Yo,t
¯
Yoωyo Bt−1
¯
B−ωbORt
OR ωor #1−ρg
exp (σgcξgc
t) (15)
T Xt
T X =Gc,t
¯
GφgBt−1
¯
BφbOSt
OS φos ORt
OR −φor
exp σtxξtx
t(16)
where ρg∈[0,1] represents the degree of smoothing in the government spending rule. ωyo ,
ωband ωor are the government consumption feedback coefficients with respect to oil output,
lagged domestic debt and ORt, respectively. In (16), lump-sum taxes respond to government
consumption, lagged debt, OStand ORtwith the feedback parameters, φg,φb,φos and φor,
respectively. The tax shock, ξtx
t, and government spending shock, ξgc
t, are given by an AR (1)
exogenous process.
14This is similar to the specification in Poghosyan and Beidas-Strom (2011).
11
3.3 Monetary policy and the switching Taylor rule
In setting the short-term nominal interest rate, Rt, the central bank follows a simple time-
varying Taylor rule by gradually responding to aggregate inflation, πt=Pt
Pt−1, domestic output,
Yh,t, and the exchange rate depreciation, ∆εt
Rt
¯
R=Rt−1
¯
Rρr(svol
t)"πt
¯πωπ(svol
t)Yh,t
¯
Yhωy(svol
t)∆εt
∆εωε(svol
t)#1−ρr(svol
t)
exp (σrξr
t) (17)
where ρr∈[0,1] is the interest rate smoothing parameter capturing monetary policy inertia.
ωπ,ωyand ωεare the policy coefficients chosen by the central bank with respect to inflation,
domestic output and the exchange rate, respectively. These policy parameters are assumed to
be governed by the same Markov process and switch together with σP∗
o(svol
t).
To bridge the behavior of policy and the heteroskedasticity of oil prices, we implement a
hybrid framework
ρr(svol
t) = ¯ρr+ ˆρr(svol
t) (18)
ωx(svol
t) = ¯ωx+ ˆωx(svol
t) (19)
where x=π, y, ε. This specification postulates that the behavior of policy responses is made
up of a systematic component, ¯ρrand ¯ωx, and a regime-dependent component, ˆρr(svol
t) and
ˆωx(svol
t). The setup is very flexible and nests as special cases the structures which characterize
the systematic response of monetary policy that is consistent regardless of the regimes of oil price
volatility and those which are typically thought to motivate the assumption of regime shifts in
policy stance. In doing so, it not only gives an explicit role to oil price volatility, capturing the
increasing uncertainty faced by policymakers and making these variances affect the behavior of
policy directly, but also allows us to take an agnostic approach in regime-switching of volatility
and policy adjustments.
3.4 Perturbation solution of the model
The generic problem of our rational expectations nonlinear DSGE model with Markov-switching
can be written as
Et
h
X
st+1=1
pst,st+1 (It)fstxt+1 (st+1), xt(st), xt−1, θst, θst+1 , ϵt= 0 (20)
where Etis the expectation operator. pst,st+1 (It) is the transition probability of going from
state stin the current period to state st+1 in the next period. fstis a vector of (potentially)
nonlinear functions. xt(st) is a vector of all the endogenous variables in the current regime st.
θstcontains the parameters in the current regime. ϵt∼N(0, I) is a vector of stochastic shocks.
In this paper, we follow Maih (2015), which computes the solution using the Newton al-
gorithm and starts out with the general Markov-switching framework set out in (20). This
approach also develops a perturbation solution technique that allows us to approximate the
decision rules and is suitable for large systems such as our model. The exact solution can be
12
returned in the form of a first-order VAR, utilizing the idea of a minimum state variable solution
of the form
xt=Tst(xt−1, ϵt) (21)
As the solution also depends on the regime st, a p-order perturbation of xt=Tst(zt) yields
the following solution that approximates the decision rule in (21)
Tst(zt)≃ T st(¯zst) + Tst
z(zt−¯zst) + 1
2!Tst
zz (zt−¯zst)⊗2+... +1
p!Tst
z(p)(zt−¯zst)⊗p(22)
where zt≡hx′
t−1χ ϵ′
ti′
is a vector of state variables, ¯zstis the steady state values of the
state variables in st, and χis the perturbation parameter.
4 Parameterization and filtration implications
Next, we turn to Bayesian methods for estimating the parameters in the model and explain
the data and calibration used in the quantitative analysis. We compute a first-order solution
of (22) and estimate the parameters using the RISE toolbox in Matlab which also includes the
procedure that we use for filtering the regime-switching model.15
4.1 Data and calibration
The model is estimated by Bayesian methods for the Nigerian economy using 87 quarterly
observations on 15 selected macroeconomic variables over the sample period of 2000:2 - 2021:4.16
While Nigeria represents the SOE in our model setup, the rest of the world consists of Nigeria’s
major trading partners of the Euro area, the US, and India.17
The domestic variables include real GDP growth (∆yh,t), real consumption growth (∆ct),
real investment growth (∆ino,t), real effective exchange rate (qt), headline CPI inflation (∆pt),
core CPI inflation (∆pno,t), the nominal interest rate (Rt), oil output (∆yo,t ), growth rate of
government debt (∆bt), change in tax revenue (∆txt) and government consumption growth
(∆gc,t). The foreign variables are trade-weighted real GDP growth (∆y∗
t), aggregate CPI in-
flation (∆p∗
t), and the interest rate (R∗
t), as well as log growth in the international oil price
(∆p∗
o,t).18
We fix a subset of parameters by a calibration and the values of the calibrated parameters
are presented in Table 1. The steady state ratios reported in Table 2are derived using data
for the Nigerian economy spanning the last three decades. The parametrization is done to
15See Maih (2015) for further details.
16The data were provided by the Statistics Department of the CBN, the Federal Reserve Bank of St. Louis,
and the International Financial Statistics. This choice of the estimation sample was based on data availability
for Nigeria, encompasses periods of financial distress and oil price fluctuations, but excludes any observations
since the outbreak of the Russo-Ukrainian conflict.
17These three regions account for about 65% of Nigeria’s total external trade over the last two decades. In
the normalized weights for the computation of the foreign variables, the Euro area is predominant with a trade
weight of 0.39 while the weights for the US and India are 0.36 and 0.25, respectively. Details about data sources
and transformations are consistent with those given in Omotosho and Yang (2024).
18To set up the model for policy analysis, it is solved by linearizing about the steady state. The lower case
variables denote the deviations of these variables from their steady state.
13
fit quarterly data with values borrowed from those assumed by Omotosho and Yang (2024)
for Nigeria and by Gali and Monacelli (2005), Romero (2008), Wolden-Bache et al. (2008),
Hove et al. (2015), Ferrero and Seneca (2019), Iklaga (2017), Allegret and Benkhodja (2015),
Algozhina (2022), Ncube and Balma (2017) and Hollander et al. (2018) for standard calibration
for small open-economies and resource-rich emerging economies.
Parameter Definition Symbol Value Source
Discount factor β0.990 Allegret and Benkhodja (2015)
Depreciation rate in both the oil and non-oil sectors δh=δo0.025 Allegret and Benkhodja (2015); Algozhina (2022)
Share of imports in household’s consumption γc0.400 Gali and Monacelli (2005)
Share of fuel in household’s consumption γo0.085 National Bureau of Statistics
Share of imports in household’s investment γi0.200 National Bureau of Statistics
Calvo - wages θw0.750 Hollander et al. (2018)
Elasticity of domestic output with respect to capital αk
h0.330 Rasaki and Malikane (2015); Algozhina (2022)
Elasticity of domestic output with respect to oil αo
h0.120 1 −αk
h−αn
h
Elasticity of domestic output with respect to labor αn
h0.550 Ncube and Balma (2017)
Elasticity of oil output with respect to capital αk
o0.700 Algozhina (2022)
Elasticity of oil output with respect to materials αm
o0.300 Ferrero and Seneca (2019); Algozhina (2022)
Share of imports in government’s consumption γg0.120 Algozhina (2022)
Elasticity of substitution between foreign & domestic goods - Govt. ηg0.600 Hollander et al. (2018)
Share of household fuel consumption in total fuel imports γco 0.750 Hollander et al. (2018)
Persistence in oil sector foreign direct investment process ρfdi 0.300 Algozhina (2022)
Foreign relative risk aversion σ∗1.000 Gabriel et al. (2023)
Foreign habit formation ϕ∗
c0.500 Gabriel et al. (2023)
Intra-temporal elasticity in foreign demand ηc∗
h0.790 Medina and Soto (2005)
Coefficient of inflation in Taylor Rule - foreign economy ωπ∗1.500 Bhattarai et al. (2021); Gabriel et al. (2023)
Coefficient of output in Taylor Rule - foreign economy ωy∗0.500 Bhattarai et al. (2021); Gabriel et al. (2023)
Table 1: Calibrated Parameters
4.2 Bayesian estimation
The joint posterior distribution of the estimated parameters is then obtained in two steps. First,
the posterior mode and the Hessian matrix are obtained via standard numerical optimization
routines. Second, we carry out a sensitivity analysis that explores the robustness of the results
based on posterior distributions.
Table 3reports the parameter estimates, summarizing the prior and posterior distributions
of the estimated parameters and 90% high posterior density intervals (HPDI). Overall, the
parameter estimates are plausible. Our estimation delivers that, based on the posterior estimate
of 1 −γR, about 20% of the households are liquidity-constrained. These households in Nigeria
do not trade on financial markets and consume entirely their wage income each period. This is
slightly below the prior value and those usually found in earlier empirical studies. The estimate
of νimplies that there is about 52% pass-through of international oil price to domestic oil prices
and the government subsidizes about half of the consumption of fuel in the domestic economy.
Before focusing on monetary policy rules in the following section, the role of fiscal policy
needs to be briefly discussed here in the context of the present paper as the implicit subsidy
expenditure determined by the fuel pricing rule is closely associated with the conduct of fiscal
policy and planning. The feedback parameter with respect to oil output, ωyo, defines the
cyclicality of government spending. The estimate suggests evidence of counter-cyclical fiscal
policy. In other words, the fiscal policy is ‘active’ for demand stabilization, implying that
the country is able to sustain increased government spending even during the periods of oil
price falls. The tax policy response is, on the other hand, ‘passive’ so that the fiscal authority
14
Ratio Symbol Value
Domestic consumption to domestic output ¯
Ch
/¯
Yh0.690
Investment to domestic output ¯
Ino
/¯
Yh0.150
Domestic materials to domestic output ¯
M/¯
Yh0.010
Government consumption to domestic output ¯
Gc
/¯
Yh0.070
Non-oil export to output ¯
C∗
h/¯
Yh0.070
Import to domestic output IM/¯
Yh0.150
Share of non-oil export in aggregate export 1−¯
Yo
/EX 0.050
Share of oil in GDP ¯
Yo
/¯
Y0.260
Fiscal debt to oil revenue ¯
B/OR 0.700
Taxes to oil revenue T X/OR 0.050
Government consumption to oil revenue pg¯
Gc
/OR 0.700
Non-oil imports to total import ¯
Cf/IM 0.400
Fuel import to total import ¯
O/IM 0.300
Oil sector foreign direct investment to net exports qF DI/NX 0.300
Exports to net exports EX/NX 0.600
Imports to net exports 1−EX/NX 0.400
Foreign debt service payments to net exports q¯
B∗¯
R∗
/NX 0.020
Foreign debt to net exports q¯
B∗
/NX 0.3112
Oil profit repatriation to net exports (1−τ)q¯
P∗
o¯
Yo
/¯
NX 0.600
Public goods imports to total import ¯
Gf/IM 0.050
Fuel subsidy payments to oil revenue OS/OR 0.200
Fuel sales value to fuel subsidy payments ¯
Pro ¯
O/OS 0.300
Domestic debt service payments to oil revenue ¯
B/ ¯
R
/OR 0.020
Fuel import value to fuel subsidy payments q¯
P∗
o¯
O/OS 0.300
Table 2: Steady State Ratios
Notes: The implied steady state ratios are derived based on the National Accounts Statistics, National Bureau of Statistics
and the Balance of Payments Statistics, Central Bank of Nigeria.
strongly adjusts lump-sum taxes in order to ensure debt stability in the regime that passively
accommodates the monetary authority.
It is interesting to note that the estimated values of the key switching parameters are very
different between the various regimes. The standard deviation of the oil price shock, ξP∗
o
t, is
estimated to be over two times higher in the high volatility regime than in the low volatility
regime. The probability of moving from the low volatility regime to the high volatility regime
is higher than the probability of switching from high to low volatility but the periods of major
oil price fluctuations do not tend to be long-lasting. The difference clearly suggests evidence of
distinct oil price movements that are time-varying in our model-implied dynamics.
To understand the central bank behavior, our empirical analysis evaluates the extent of its
adjustments that interact with the varying oil price volatility. There is no prior information
about the regime-dependent policy parameters which are assumed to be normally distributed
and centered at 0 with standard deviations of 0.25. Our posterior maximization identifies
several observed switches in monetary policy. During highly volatile oil price periods (s2) with
heightened uncertainty faced by policymakers, the central bank is less sensitive to movements
in headline inflation, focuses less on exchange rate stabilization, but places greater emphasis on
the output gap and adjusts interest rates more quickly.
4.3 Smoothed transition probabilities
There is ample evidence in favor of stochastic regime switches in the parameters. Figure 1
plots the smoothed state probabilities for being in the high volatility regime (s2) in the model
(based on the posterior mode) and clearly shows recurrent spikes in the oil price movements
15
Parameter Prior distribution Posterior distribution
Density Mean SD/DoF Mode Median 90% HPDI
Ricardian consumers γRB0.60 0.10 0.811 0.807 [0.739: 0.877]
Labour supply elasticity φG1.45 0.10 1.424 1.397 [1.252: 1.547]
Relative risk aversion σIG 2.00 0.40 1.271 1.304 [1.021: 1.549]
External habit ϕcB0.70 0.10 0.388 0.400 [0.281: 0.503]
Investment adjustment cost χG4.00 3.00 16.866 18.050 [17.043: 18.619]
Fuel pricing parameter νB0.30 0.10 0.526 0.522 [0.387: 0.625]
Oil-core consumption elasticity ηoG0.20 0.10 0.151 0.189 [0.059: 0.318]
Foreign-domestic consumption elasticity ηcG0.60 0.20 0.560 0.586 [0.435: 0.784]
Foreign-domestic investment elasticity ηiG0.60 0.20 0.561 0.586 [0.257: 0.807]
Calvo - domestic goods θhB0.70 0.10 0.616 0.621 [0.557: 0.678]
Calvo - imported goods θfB0.70 0.10 0.691 0.664 [0.498: 0.822]
Calvo - exports goods θhf B0.70 0.10 0.716 0.717 [0.522: 0.867]
Monetary policy: systematic
Taylor rule - inflation ¯ωπG1.50 0.25 3.492 3.234 [2.831: 3.719]
Taylor rule - output ¯ωyG0.125 0.05 0.108 0.115 [0.051: 0.186]
Taylor rule - exchange rate ¯ωεG0.125 0.05 0.177 0.199 [0.087: 0.341]
Interest rate smoothing ¯ρrB0.50 0.25 0.162 0.146 [0.020: 0.272]
Monetary policy: regime-dependent
Taylor rule - inflation (L) ˆωπ(s1)N0.00 0.25 0.609 0.632 [0.378: 0.923]
Taylor rule - inflation (H) ˆωπ(s2)N0.00 0.25 0.206 0.011 [-0.360: 0.302]
Taylor rule - output (L) ˆωy(s1)N0.00 0.25 -0.077 0.008 [-0.144: 0.200]
Taylor rule - output (H) ˆωy(s2)N0.00 0.25 0.161 0.029 [-0.151: 0.258]
Taylor rule - exchange rate (L) ˆωε(s1)N0.00 0.25 0.866 1.071 [0.844: 1.327]
Taylor rule - exchange rate (H) ˆωε(s2)N0.00 0.25 0.363 0.152 [-0.167: 0.433]
Interest rate smoothing (L) ˆρr(s1)N0.00 0.25 0.002 0.030 [-0.090: 0.131]
Interest rate smoothing (H) ˆρr(s2)N0.00 0.25 -0.092 -0.009 [-0.191: 0.189]
Fiscal policy
Government consumption - output ωyo N0.40 0.50 -0.388 -0.388 [-0.440: -0.322]
Government consumption - fiscal debt ωbN0.30 0.50 0.079 0.086 [0.000: 0.178]
Government consumption - oil revenue ωor N0.80 0.50 0.778 0.758 [0.603: 0.929]
Government consumption smoothing ρgcB0.50 0.25 0.310 0.307 [0.161: 0.466]
Tax - fiscal debt φbN0.40 0.50 0.205 0.207 [0.006: 0.384]
Tax - government consumption φgN0.95 0.50 0.630 0.680 [0.507: 0.853]
Tax - subsidies φos N0.10 0.50 0.531 0.604 [0.239: 0.809]
Tax - oil revenue φor N0.30 0.50 0.077 0.130 [0.007: 0.241]
Standard deviation and persistence of shock
Oil price (L) σP∗
o(s1)IG 0.10 4.00 0.100 0.126 [0.114: 0.137]
Oil price (H) σP∗
o(s2)IG 0.01 4.00 0.226 0.325 [0.214: 0.475]
Oil price (L) ρP∗
o(s1)B0.50 0.28 0.994 0.957 [0.907: 0.999]
Oil price (H) ρP∗
o(s2)B0.50 0.28 0.548 0.587 [0.371: 0.888]
Transition probability
[L, H] pvol
12 B0.50 0.28 0.045 0.043 [0.006: 0.084]
[H, L] pvol
21 B0.50 0.28 0.178 0.280 [0.103: 0.476]
Table 3: Prior and Posterior Distributions
Notes: Two chains of 100,000 draws are generated and the first half of these draws is discarded. The variance-covariance
matrix of the perturbation term for the jumping distribution in the Metropolis-Hastings algorithm is adjusted so that an
acceptance rate of 0.2469% is obtained. In the estimation the number of draws that we choose is sufficient to allow for
convergence. To formally check the convergence of the parameters, we use the convergence indicators such as the scale
reduction factor statistic recommended by Brooks and Gelman (1998) and Gelman et al. (2004).
16
and the ensuing policy responses that are time-varying and contribute to their nonlinear effect
on the real economy. The economy has stayed in the high oil price volatility regime with a high
probability of occurrence which is mostly responsible for the macroeconomic instability that we
observe and coincides with monetary policy adjustments.
Interestingly, major volatile episodes in oil prices are observed during 2008-2009, 2014-2016,
and 2020-2021. These high oil price volatility states are mostly related to historical events and
are clearly triggered by a plunge in oil prices in these distinct periods. The first period of a
huge price swing coincides with the US credit crisis of 2008. The second episode of heightened
volatility saw crude oil spot prices drop from as high as USD115 a barrel in June 2014 to a low
of USD45 in January 2015.19 This may be explained by the booming US shale oil production
causing the sharp and persistent price fall starting from the third quarter of 2014 that culminated
into an economic recession in Nigeria in 2016.20 The third break date is estimated to occur in
2019/2020. We find that oil prices have displayed a more recent surge of volatility since then,
coinciding with the demand-driven collapse in oil prices starting from the fourth quarter of 2019
due to the COVID-19 outbreak.
2003Q1 2005Q4 2008Q3 2011Q2 2014Q1 2016Q4 2019Q3 2021Q4
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1High oil price volatility
Probability of high oil price volatility
Real oil price inflation
Figure 1: Smoothed Probability of High Volatility Regime (s2)
Notes: The figure presents the smoothed probabilities for being in the high oil price volatility regime in the model that
allows synchronized switching in the standard deviations of the oil price shock and in the monetary policy parameters.
Their priors are assumed to be inverse gamma with (0.1,4) and normal with (0,0.25), respectively, between the high and
low volatility regimes. Observed data: log growth in the international oil price.
5 Macroeconomic implications
Before conducting our policy exercises to quantify the time-varying optimized monetary policy
response that depends on the volatility regimes, we simulate the model based on the posterior
estimates (mode) under a condition of complete pass-through of international oil price to the
retail price of fuel to assess how the economy would have responded in the absence of the subsidy
program.
19Source: U.S. Energy Information Administration.
20Beyond the oil price movements in 2014-2016, oil revenue in Nigeria was subject to heightened volatility due
to production shutdowns and supply disruptions, and was affected by the peculiar market demand conditions
associated with the shale oil and gas revolution. As a result, the country recorded reduced patronage of its crude
oil, and more so, increased disinvestment of oil-related capital by foreign investors.
17
5.1 Responses of fuel subsidy removal
Based on (11), the parameter ν∈[0,1] governs the level to which the government subsidizes
fuel consumption. A value of ν= 1 provides us an alternative case where there is complete
pass-through of international oil prices into the retail price of fuel and the subsidy regime ceases
to exist. In Figure 2, we simulate the model based on the parameterization in Section 4and
report the impulse responses (henceforth IRFs) comparing the transmission mechanism of a
negative oil price shock when the subsidy program is in place or ‘turned off’.21
5 10 15 20
-0.025
-0.02
-0.015
-0.01
-0.005
Output
5 10 15 20
0
5
10
15
10-3
Aggregate cons.
5 10 15 20
0.005
0.01
0.015
0.02
R cons.
5 10 15 20
-5
0
5
10-3 NR cons.
5 10 15 20
-4
-2
0
2
4
10-3
Non-oil output
5 10 15 20
-8
-6
-4
-2
010-3 Total inf.
5 10 15 20
-8
-6
-4
-2
0
10-3 Dom. inf.
5 10 15 20
-5
0
5
10-3 Core inf.
5 10 15 20
-10
-5
0
10-3 Interest rate
5 10 15 20
0
0.01
0.02
0.03
0.04
Exchange rate
5 10 15 20
-0.025
-0.02
-0.015
-0.01
-0.005
0
Marg. cost
5 10 15 20
-0.03
-0.02
-0.01
Hours worked
s1
s2
No subsidy, s1
No subsidy, s2
Figure 2: Responses to a Negative Oil Price Shock
Notes: The figure plots the response corresponding a negative one standard deviation of the oil price shock’s innovation.
Each response is for a 20 period (5 years) horizon. s1→Regime 1: Low oil price volatility; s2→Regime 2: High oil price
volatility. Subsidy: ν= 0.526; No subsidy: ν= 1.
The results are briefly discussed here for the alternative economy with ν= 1. The projected
outcomes for GDP and private consumption are in line with the findings of Siddig et al. (2014).
Following a negative real oil price shock, the contraction of aggregate GDP is less severe in
the short run. Private consumption rises more due to more resources becoming available to
consumers (and an improvement in the terms of trade), compared to the case in which the
subsidy policy is in place. The increased aggregate demand arising from additional income
that is available to consumers and the associated lower real marginal cost owing to the full
21A negative oil price shock (an unanticipated fall in oil prices) can be strongly correlated with domestic output
and capital inflows. The frailties in emerging market countries can lead to sudden and sharp reversals of capital
inflows during a negative oil price shock (the ‘sudden stops’ highlighted in Calvo,1998).
18
pass-through effect ameliorates the shock’s contractionary effects on non-oil GDP.
The behaviors of different measures of inflation are qualitatively and quantitatively consis-
tent with those of Omotosho (2019b). Domestic inflation declines more as domestic prices are
less sticky in their downward adjustment, while the response (decline) of aggregate inflation is
less pronounced in the benchmark which can be explained by the price rigidity implied by the
subsidy program. Reacting to aggregate inflation, the monetary authority is able to cut interest
rates, in the aftermath of a large, negative oil price shock, ameliorating the short-run output
contraction. The real exchange rate depreciates more and the inflationary effects of such depre-
ciation on headline inflation are more than offset by the reduction in domestic inflation. Finally,
labor demand decreases with an decreased cost of production (firms wishing to substitute away
from labor).
More importantly, there are marked differences in IRFs when the economy moves from s1
to s2. Overall, the effects are generally stronger and more persistent in s2. In particular, the
decline in firms’ real marginal cost is more fully reflected by the removal effects causing more
downward adjustment in the prices of domestically produced goods which is exacerbated if the
shock is large and volatile. Given an oil price shock in the high volatility period (s2), the
alternative economy (ν= 1) is associated with a large decline in total inflation and, facing
no inflationary pressures following a fall in oil prices, its central bank would cut interest rates
in the bid to boost aggregate demand. As a result, it would ameliorate the contractionary
effects of the negative shock and produce better outcomes for consumption. Compared to the
low-variance state, the output contraction on impact in the benchmark economy is more severe
in s2. Interestingly, the model also predicts that there is an amplification effect in the output
dynamics in s2where the policy reacts strongly to output stabilization.
Our simulations so far also reveal another interesting finding. As our model features Ricar-
dian and Non-Ricardian households (rich and poor), we depict the responses of consumption
of both types of consumers (R and NR, respectively). When we remove the subsidy program
in the alternative economy, this introduces more amplified responses into the model, drives a
bigger wedge between the responses of the two economies in aggregate consumption and the
consumption of R households than those in the consumption of NR households, in the sense that
the increase in aggregate consumption is mostly associated with that from the R consumers,
arising from additional income that is available to the latter. Additionally, the magnitude of
the increase is larger in s2where the shock is more volatile. Clearly, this suggests that potential
fuel subsidy reforms would have mainly benefited the upper-income households exacerbating
wealth inequality between these households and the working poor that are credit-constrained.
These results are useful in helping us understand the issues of different welfare consequences in
different policy scenarios.
5.2 Economic performances under counterfactual scenarios
What would the economy look like if the central bank does not change its policy when the fuel
subsidy is removed? We look at the same counterfactual economy as assumed in Section 5.1 in
which we impose ν= 1. Table 4corresponds to the standard deviations of observed macroe-
conomic variables under this alternative assumption and the estimated benchmark model, re-
19
spectively. With the estimated Taylor rule, this leads to higher volatility in most of the key
economic indicators. The alternative economy is associated with higher volatility in headline,
core, domestic and imported inflation. Table 5reports the cross-correlations of the observ-
able variables vis-`a-vis output. The alternative economy also performs poorly in capturing the
countercyclicality of inflation, generating the wrong sign.
std. dev. ∆yh,t ∆ct∆ino,t πtRtqtπc,t ∆yo,t ∆bt∆txt∆gc,t
Benchmark 0.146 0.182 0.092 0.109 0.429 0.143 0.107 0.457 0.551 0.444 0.233
ν= 1 0.147 0.186 0.092 0.113 0.432 0.159 0.109 0.458 0.550 0.454 0.235
Table 4: Standard Deviation of Domestic Observables
cross-corr. ∆yh,t ∆ct∆ino,t πtRtqtπc,t ∆yo,t ∆bt∆txt∆gc,t
Benchmark - 0.324 0.033 0.015 -0.069 0.241 0.042 0.543 0.610 0.152 0.444
ν= 1 - 0.341 0.034 -0.020 -0.081 0.272 0.057 0.543 0.608 0.176 0.455
Table 5: Co-Movement of Domestic Observables
What would be the stabilization properties of the estimated policy had the subsidy removal
been implemented over the last two decades? Figure 3provides an additional comparison of
the economic performances under the counterfactual scenario where subsidies were not in place
historically. To this end, we plot the simulated time series from the benchmark model based on
the smoothed shocks and compare those with the simulated economy from the counterfactual
scenario. With the realized monetary rule, Figure 3provides some interesting result that shows
that the central bank had done well and achieved better performances in terms of stabilizing
inflation and exchange rate movements and smoothing out fluctuations in output, especially
during the early periods of the sample during which oil prices were relatively stable, in the
presence of subsidies. In the absence of subsidies (ν= 1), the macroeconomy (particularly
in terms of the three policy target variables) would have been more volatile with the realized
shocks and policy rule.
6 Macroeconomic stabilization and optimal policy
The estimated structural model set out above is well-suited for the study of policy options.
In this section, we move to optimal monetary policy exercises. The policymaker may be con-
strained to simple rules even with commitment, thus, for transparency, information available
for communications, and ease of implementation, we focus on the optimized simple Taylor-type
commitment rule that minimizes the expected inter-temporal loss as given by (23) at time t.
6.1 The central bank’s role
The central bank sets out to maximize a general discounted welfare criterion subject to the
constraints of the DSGE model. In a no-subsidy economy, the reason why monetary policy is
more important for stabilizing economic activity is two-fold. First, we do not know whether the
20
2000Q2
2003Q1
2005Q4
2008Q3
2011Q2
2014Q1
2016Q4
2019Q3
-0.1
-0.05
0
0.05
0.1
0.15
0.2 Core inflation
Realized rule and benchmark Realized rule and =1
2000Q2
2003Q1
2005Q4
2008Q3
2011Q2
2014Q1
2016Q4
2019Q3
-0.05
0
0.05
0.1
0.15
0.2
Headline inflation
2000Q2
2003Q1
2005Q4
2008Q3
2011Q2
2014Q1
2016Q4
2019Q3
-1.5
-1
-0.5
0
0.5
1Interest rate
2000Q2
2003Q1
2005Q4
2008Q3
2011Q2
2014Q1
2016Q4
2019Q3
-0.2
-0.1
0
0.1
0.2
0.3 Exchange rate
2000Q2
2003Q1
2005Q4
2008Q3
2011Q2
2014Q1
2016Q4
2019Q3
-0.1
-0.05
0
0.05
0.1 Output
2000Q2
2003Q1
2005Q4
2008Q3
2011Q2
2014Q1
2016Q4
2019Q3
-0.2
-0.1
0
0.1
0.2 Oil output
Figure 3: Simulated Economies with Realized Rule and Shocks
central bank has behaved optimally, in terms of a Taylor rule model. Second, agents are more
vulnerable to oil price fluctuations which can be exacerbated by fuel subsidy removal.
There are generally two approaches to optimally evaluate policies for welfare analysis in
DSGE models. The welfare loss function can be either utility-based or derived through a stan-
dard ad-hoc quadratic loss function. We examine the potential consequences of removing fuel
subsidies in Nigeria from the viewpoint of a central banker, focusing on their role in managing
the adverse effects of oil price shocks and the removal. Therefore, we opt for the latter ap-
proach and evaluate monetary policy rules based on the linearized model of (20) with a simple
quadratic loss function that penalizes variability in an observed subset of key macroeconomic
variables (i.e. welfare-relevant variables).
The estimated structural parameters of the model, other than the monetary policy parame-
ters, are used to seek optimized simple monetary policy rules, based on the time-varying Taylor
interest rate rule set out in Section 3.3, that can accommodate the Markov-switching parame-
ters. We consider the standard ad-hoc quadratic period loss function in deviation form which
is given by
Ω0= (1 −β)E0"∞
X
t=0
βt(λππ2
t+λyy2
h,t +λr∆R2
t+λε∆ε2
t)#
≃λπvar(πt) + λyvar(yh,t) + λrvar(∆Rt) + λεvar(∆εt) as β→1 (23)
where the variances above are unconditional variances of the target variables and the period
utility Ω0is an unconditional welfare loss function where β→1. We also carry out a search
21
over a grid on a range of different weights on the variances in terms of (1, λy, λr, λε).
We use the estimated structural parameters of the model to derive optimized simple mon-
etary policy rules that can accommodate the Markov-switching behavior that we set out in
our model, assuming that, like agents in the model, the central bank can observe the different
regimes (i.e. they observe st). We compare the result with counterfactual simulations in terms
of their abilities in stabilizing inflation, output and exchange rates.
6.2 Optimized Taylor rules
To optimize the policy parameters in (17), we set bounds (priors) on the parameters to discipline
the process with the wide 90% ranges within which the optimization searches for the parameters.
Our focus is on linear-quadratic problems that are available for different forms of policy, and
we use the ad-hoc approach for the (central bank’s) loss as the welfare criterion assuming that
they dislike inflation, output gap, and exchange rate movements to assess the welfare-reducing
effects implied by the model features. Table 6uses the same prior densities as in Table 3but
imposes more prior uncertainty and uses quantiles of the distributions, allowing such priors to
have more diffuse distributions.
We carry out the policy simulations that compare the expected inter-temporal losses and
macroeconomic volatilities for periods of varying economic conditions (changing variances of
oil prices), and for cases of zero or partial subsidy. In a sense, we have designed and derived
‘robust’ simple policy rules with respect to exogenous uncertainty incorporated into the AR(1)
shock process of oil price for the estimated model and a counterfactual (i.e. ν= 1).
Parameter Prior distribution Posterior mode
Density Lower quartile Upper quartile Estimated rule OSR ν= 0.526 OSR ν= 1
¯ωπG1 10 3.492 6.782 6.897
¯ωyG0.1 4 0.108 0.153 0.144
¯ωεG0.1 4 0.177 0.824 0.801
¯ρrB0.5 0.95 0.162 0.783 0.781
Density Mean SD
ˆωπ(s1)N0 0.5 0.609 0.060 0.060
ˆωπ(s2)N0 0.5 0.206 0.023 0.024
ˆωy(s1)N0 0.5 -0.077 -0.036 -0.047
ˆωy(s2)N0 0.5 0.161 -0.008 -0.012
ˆωε(s1)N0 0.5 0.866 0.115 0.114
ˆωε(s2)N0 0.5 0.363 0.044 0.043
ˆρr(s1)N0 0.5 0.002 -0.345 -0.351
ˆρr(s2)N0 0.5 -0.092 -0.232 -0.235
Ω00.0417 0.0251 0.0262
Table 6: Estimated and Optimized Simple Rule Coefficients
Notes:s1→Regime 1: Low oil price volatility; s2→Regime 2: High oil price volatility. ν= 0.526: estimated partial
subsidy; ν= 1: zero subsidy. We assume that λπ= 1, λy= 0.2, λr= 0.1 and λε= 0.1 in (23) for the policymaker’s
linear-quadratic problem. Following Chen and Macdonald (2012), we carry out a search over a grid on a range of different
weights. We allow (λy, λr) to vary over a grid of [0,1] and compute the optimized simple rules and the unconditional
variances of the target variables which compares the output-inflation volatilities associated with each set of (λy, λr) used.
Our result shows that a conservative banker’s policy preference increases the variance of output and clearly faces a policy
trade-off which moves to the upper-left corner when λrincreases. To derive an optimal monetary policy rule, we also
choose the above parameter configuration in the loss function that generates a low level of exchange rate volatility, whilst
keeping the output-inflation volatility low at the same time.
In Table 6, we benchmark the optimized simple rules (OSR) against the estimated policy
22
rule. The OSR prescribes larger responses to all three target variables and has a much higher
degree of interest rate smoothing compared to the realized rule. These results suggest that, had
the central bank acted optimally, it would more aggressively respond to fluctuations in inflation
and exchange rates and be much more inertial whether we use subsidies or not. Such a rule
would achieve the best welfare outcome under the subsidy program (i.e. we use Ω0= 0.0251 as
our performance metric). In addition, the OSR responses are more symmetric between regimes
and under the OSR the central bank behaviors are relatively systematic regardless of regime
shifts, except for stabilizing exchange rates. In the no-subsidy economy, optimal monetary
policy is more focused on preserving price stability and anchoring inflation expectations.
6.3 Unconditional standard deviations
Under different optimized rules, aggregated standard deviations of the key domestic variables
are computed by re-weighting the system with appropriate probabilities (i.e. the ergodic distri-
bution of the regimes) and presented in Table 7with one exception: “s2” indicates the scenario
where the regime-specific volatilities are computed. It shows the level of macroeconomic in-
stabilities conditional on regimes, for example, if the economy had stayed in the high-variance
state (s2).
sd(∆yh,t) sd(∆ct) sd(∆ino,t) sd(∆yno,t ) sd(πt) sd(πc,t) sd(πd,t) sd(πf,t ) sd(Rt) sd(∆εt) sd(cR
t) sd(cNR
t)
Estimated rule 0.146 0.182 0.092 0.190 0.109 0.107 0.111 0.091 0.429 0.204 0.298 0.215
OSR ν= 0.526 0.142 0.178 0.092 0.186 0.068 0.070 0.088 0.064 0.388 0.168 0.285 0.207
OSR ν= 1 0.144 0.183 0.092 0.186 0.070 0.072 0.088 0.064 0.392 0.177 0.288 0.209
OSR ν= 1, λy= 0.5 0.144 0.182 0.092 0.186 0.072 0.073 0.089 0.064 0.391 0.158 0.287 0.208
OSR ν= 1, s20.145 0.181 0.092 0.183 0.079 0.079 0.089 0.065 0.383 0.194 0.245 0.198
Table 7: Standard Deviation of Macroeconomic Variables
Notes:ν= 0.526: estimated partial subsidy; ν= 1: zero subsidy. s2→Regime 2: High oil price volatility. λy= 0.5 is
the weight on the output gap variance. The variables include real GDP growth (∆yh,t), real consumption growth (∆ct),
real investment growth (∆ino,t ), non-oil output growth (∆yno,t), headline inflation (πt), core inflation (πc,t ), domestic
inflation (πd,t), imported inflation (πf,t), nominal interest rate (Rt), real effective exchange rate (∆εt), consumption by
Ricardian household (cR
t), and consumption by Non-Ricardian household (cNR
t).
The first two rows of Table 7compute the cost of following the estimated rule relative to the
optimal rule in the benchmark economy (OSR ν= 0.526). In line with the results that compare
the welfare losses, which represent the expectation of all future outcomes, the computed OSR in
the alternative economy (OSR ν= 1) is able to more effectively stabilize the economy, compared
to the estimated rule in the benchmark economy, with the exception of aggregate consumption.
Table 7also presents a scenario under which we can achieve better policy outcomes than the
estimated rules when ν= 1. To do so, the central banker sets the weight on the output gap
variance to λy= 0.5. This result shows the potential welfare gains from eliminating business
cycle fluctuations in the alternative economy (ν= 1). Putting a higher weight on controlling
output volatility increases inflation variance. Nevertheless, this policy scenario generates an
optimized Taylor rule that can achieve better policy outcomes than the estimated rules in the
ν= 1 economy. Finally, we investigate the level of macroeconomic instabilities that is regime-
specific, that is, we compute the model-implied moments conditional on the alternative economy
staying in the high-variance state (OSR ν= 1, s2). The OSR in this economy generates higher
volatility for inflation and exchange rate movements.
23
6.4 Volatility implications of optimized rules
As in Section 6.3, we allow λyto vary and compute the optimized simple rules and the un-
conditional variances of the target variables in Table 8which compares the output-inflation
volatilities associated with λy= 0.1 and λy= 0.5, respectively. By increasing λy, we show that
the central banker’s policy preference decreases the variance of output growth, whilst facing a
clear policy trade-off which generates a higher level of inflation volatility. For this exercise, our
focus is on comparing the following four cases: the benchmark economy (Bench: ν= 0.526),
the alternative economy (Alter: ν= 1), the benchmark economy staying in the high-variance
state (Bench in s2:ν= 0.526), and the alternative economy staying in the high-variance state
(Alter in s2:ν= 0.526). Our results show that the monetary policy trade-offs can be less severe
under optimized rules with an aggressive response to inflation in the alternative economy when
subsidy is removed but are amplified in states of highly volatile oil prices.
λy= 0.1 sd(∆yh,t) sd(πt)λy= 0.5 sd(∆yh,t) sd(πt)↓sd(∆yh,t)↑sd(πt)
Bench 0.14226 0.06838 Bench 0.14210 0.06950 0.00016 0.00112
Alter 0.14423 0.07027 Alter 0.14397 0.07171 0.00026 0.00143
Bench in s20.14387 0.07690 Bench in s20.14377 0.07780 0.00010 0.00090
Alter in s20.14496 0.07930 Alter in s20.14478 0.08051 0.00018 0.00121
Table 8: Output-Inflation Volatility for Optimized Simple Rules
Notes:ν= 0.526: estimated partial subsidy; ν= 1: zero subsidy. s2→Regime 2: High oil price volatility. λy= 0.1,0.5
is the weight on the output gap variance. The policy variables are real GDP growth (∆yh,t ) and headline inflation (πt).
6.5 Impulse responses to a negative oil price shock
Insights into the working of optimal policy and of the transmission mechanism can be obtained
by deriving posterior IRFs following an unanticipated 1% negative international oil price shock.
The aim of this exercise is two-fold. First, we are interested in comparing the transmission of the
shock when the subsidy program is ‘turned on’ and ‘turned off’. This way, we assess the impact
of imposing/removing the program on different model dynamics under different monetary policy
rules. Second, we investigate the importance of shocks to the endogenous variables of interests
in order to gain a better understanding of the model uncertainties faced by policymakers and
the source of welfare differences. In Figures 4and 5, we plot the IRFs for the low (s1) and high
volatility regimes (s2), respectively. The policy rules presented are the estimated rule under the
benchmark model, an optimized simple rule (OSR) derived based on the posterior mode of the
model, and an OSR when we remove the subsidy program in the same model.
Qualitatively, the IRFs are broadly similar under the different monetary policy rules and
under the volatility-switching assumption. Only the qualitative responses of total inflation and
interest rate differ depending on the subsidy regulation. Following a negative oil price shock,
output immediately falls and domestic inflation falls. This effect in turn leads to a reduction
in equilibrium labor. The supply-side shock results in a fall in the marginal cost, and the fall
is larger in the absence of subsidies. Domestic consumption rises due to the depreciating real
exchange rate and the resultant improvement in the terms of trade.
The optimal policy (when the subsidy is in place) is to raise the interest rate a little initially
24
5 10 15 20
-12
-10
-8
-6 10-3 Output
5 10 15 20
0.006
0.008
0.01
0.012
0.014
0.016 Aggregate cons.
5 10 15 20
0.008
0.01
0.012
0.014
0.016
R cons.
5 10 15 20
-2
0
2
4
6
8
10-3 NR cons.
5 10 15 20
0
2
4
10-3
Non-oil output
5 10 15 20
-2
-1.5
-1
-0.5
10-3 Total inf.
5 10 15 20
-4
-3
-2
-1
010-3 Dom. inf.
5 10 15 20
0
1
2
3
4
510-3 Core inf.
5 10 15 20
0
1
2
3
4
10-3Interest rate
5 10 15 20
0.014
0.016
0.018
0.02
0.022
Exchange rate
5 10 15 20
-8
-6
-4
-2
10-3 Marg. cost
5 10 15 20
-12
-10
-8
10-3
Hours worked
Estimated rule
OSR benchmark
OSR no subsidy
Figure 4: Responses to a Negative Oil Price Shock with s1
5 10 15 20
-0.025
-0.02
-0.015
-0.01
-0.005
Output
5 10 15 20
0
5
10
15
10-3
Aggregate cons.
5 10 15 20
0.005
0.01
0.015
0.02
R cons.
5 10 15 20
-5
0
5
10-3 NR cons.
5 10 15 20
-4
-2
0
2
4
610-3
Non-oil output
5 10 15 20
-6
-4
-2
0
10-3 Total inf.
5 10 15 20
-8
-6
-4
-2
0
10-3 Dom. inf.
5 10 15 20
-5
0
5
10-3 Core inf.
5 10 15 20
-10
-8
-6
-4
-2
0
10-3Interest rate
5 10 15 20
0
0.01
0.02
0.03
0.04
Exchange rate
5 10 15 20
-0.025
-0.02
-0.015
-0.01
-0.005
0
Marg. cost
5 10 15 20
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
Hours worked
Estimated rule
OSR benchmark
OSR no subsidy
Figure 5: Responses to a Negative Oil Price Shock with s2
25
to contain inflation (the headline or core measure), but then to commit to a sharp monetary
relaxation before gradually returning to the steady state. The same trajectory is depicted
by the estimated policy. Contrary to the situation in which the government subsidizes the
consumption of fuel, but similarly to Figure 2, the OSR predicts an initial cut in the interest
rate in the counterfactual case in response to the falling headline inflation rate, ameliorating
the contractionary effects of a negative oil price shock on aggregate GDP. As is consistent with
Table 6, the responsiveness of the initial monetary expansion to inflation is stronger in the case
when the subsidy is removed (if we compare the magnitude of their responses under the two
optimal rules).
As expected, the reaction to the shock is less aggressive in s1(low volatility). The three
policy rules generate less contractionary effects on output and are better at containing inflation
and mitigating the real exchange rate depreciation in this regime. It should be noted that,
while the magnitude of the interest rate responses under the OSR are slightly larger in s2, the
target variables exhibit generally much larger responses to the shock in s2compared to in s1
(especially those of output and aggregate inflation). This suggests that the optimal policy has
a more aggressive reaction function to variations in these variables in s1– a result consistent
with our Table 6above.
As oil is an input to both production and consumption, our results can also reveal more
evidence for the central bank’s trade-off, i.e., in its objective to stabilize prices and fluctuations
in output, and that the severity of this dilemma depends on the price volatility (uncertainty) and
the impact of subsidy removal. For example, our simulations imply that, when the central bank
tends to respond to increased uncertainty about future prices during an episode of persistent
inflationary pressures (Figure 5) by raising the interest rate (a small rate increase), the OSR
and realized policy rule predict a more severe decline in output than under the low volatility
regime. Similarly, as noted, the responsiveness of the initial monetary expansion to inflation
is stronger in the case when the subsidy is removed, from a central bank whose objective is
to smooth out fluctuations in output, thus this helps prevent the economy from contracting
drastically following the oil price shock.
Finally, as discussed initially in Section 5.1, the policy IRFs provide more interesting results.
With the additional recourses available to consumers (which, as discussed, would mainly benefit
the R consumers), under the OSR, consumption rises initially but the NR (poor) consumers
see this happening with a much smaller increase in their consumption. Qualitatively, these
consumption responses are similar regardless of the exogenous switching between the volatility
states. The responses of real variables - output, hours and consumption - differ considerably
between the benchmark OSR and the OSR in the no-subsidy economy, and between the low and
high volatility regimes, following the shock, which explains the large welfare differences (for all
shocks combined). Furthermore, the OSR policy derived in both regimes of price volatility and
in the no-subsidy economy sees a larger increase in both output and consumption on impact and
in the short run. This is a major source of the expected welfare differences noted previously.
26
7 Statistical validation and sensitivity analysis
Our analysis is based on a particular parameterization of the model which generates some
interesting findings. In this section, we explore the robustness of our results based on the
posterior distribution. The posterior density is approximated by using the Monte-Carlo Markov
Chain Metropolis-Hastings (MCMC-MH) algorithm, starting from the posterior kernel mode,
with two parallel chains of 100,000 random draws from the posterior density,22 with the variance-
covariance matrix of the perturbation term in the algorithm being adjusted in order to maintain
an acceptance ratio of 25%. These draws are then used for validating the main results.
One of the great advantages of adopting a Bayesian approach is that it facilitates a formal
comparison of different models through their posterior marginal likelihoods. The differences in
marginal data densities (MDD) or the posterior odds ratio (Bayes Factors) are important as
they help to provide decisive evidence for choosing a particular model over others. In Table 9, we
compare two approximations of log MDD based on the posterior distribution.23 The benchmark
model with subsidy clearly wins the likelihood ranking and attains the highest posterior odds,
thus providing the most comprehensive form of assessment that suggests that our model with
switching dynamics is empirically relevant and statistically improves the fit to the Nigerian
data.24 Our finding clearly rejects the models with time-invariant parameters and complete
pass-through in terms of explaining the data for Nigeria over the sample period. Indeed, the
benchmark model produces tight posterior estimates.
MDD Non-switching Switching (benchmark) Switching (ν= 1)
Meng and Wong’s Bridge 919.26 922.20 899.31
Modified Harmonic-Mean 915.07 918.30 893.62
Table 9: Bayesian Model Comparison
Next, we draw 10,000 random parameters from the posterior simulation above and repeat
the exercises in Sections 4.3 and 5.1. In Figure 6, we compute the smoothed probabilities for
being in the high oil price volatility state. The figure shows the median response, together
with the shaded areas that correspond to the 90% credibility interval. The simulation identifies
qualitatively the distinct periods of oil price volatility similar to the finding in Section 4.3. In
line with the previous result, there are synchronized switches to a responsive policy state that
are triggered by the recurrent volatility episode with a high probability. It also shows that the
economy mostly stays in the more active policy regime (ˆωπ(s1)).
However, we identify a switch to a less responsive monetary state that coincides with the
first and third oil price volatility episodes with a high probability, albeit for a relatively brief
period. As explained by some of the previous studies,25 inflation in Nigeria was brought under
22We burn-in the first 25% of the chain to remove any dependance from the initial conditions. The number
of replications is sufficiently large to explore the whole parameter space and asymptotically move to its ergodic
distribution.
23Our computational methods for estimating the MDD are based on the Geweke (1999) Modified Harmonic-
Mean estimator and the Meng and Wong (1996) Bridge Sampling estimator.
24Based on Kass and Raftery (1995), a Bayes factor of 10 −100 or a log data density range of [2.30,4.61] is
“strong to very strong evidence”.
25Similar responsive states are also detected in Omotosho and Yang (2024) based on posterior simulations
27
control during the 2008-09 global financial crisis, although monetary policy seems to have played
a role in stabilizing output during this distinct episode. The central bank appears to have been
less sensitive to movements in headline inflation since the financial crisis as the Monetary Policy
Rate becomes more persistent during the period, reflecting a transition to a multiple-mandate
regime.
High oil price volatility
2000Q2 2003Q1 2005Q4 2008Q3 2011Q2 2014Q1 2016Q4 2019Q3
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
90% credibility interval
Probability of high oil price volatility
Figure 6: Smoothed Probability of High Volatility Regime (s2)
Notes: The figure presents the smoothed probabilities for being in the high oil price volatility regime in the model that
allows synchronized switching in the standard deviations of the oil price shock and in the monetary policy parameters.
Their priors are assumed to be inverse gamma with (0.1,4) and normal with (0,0.25), respectively, between the high and
low volatility regimes.
Finally, in Figures 7-10, we conduct 10,000 simulations of 20 quarters each, and plot the
median responses of the key variables, along with the 90% credibility intervals, that can be
compared with the impulse response simulations reported in Section 5.1 above. This robustness
exercise presents similar dynamics compared with those generated by the posterior mode, with
the magnitude of responses under the latter generally lying within the credibility intervals of
median responses. We uphold these results with both the volatility and subsidy scenarios.
Under the no-subsidy regime, when shocks are small, the central bank responds by increasing
the interest rate, but when shocks are large, the simulated response predicts an initial cut to the
interest rate, loosening monetary policy – a result consistent with the policy response prescribed
by the OSR.
8 Implications for macroeconomic policy
By generating important dynamics in the data, the results presented in the paper can be
used to derive some concrete policy recommendation for an oil-producing emerging economy.
On the design of monetary policy frameworks, we find that, for a welfare-maximizing central
bank, the best policy rules are aggressive on inflation and exchange rates and are much more
inertial whether we use fuel subsidies or not. But the trade-offs faced by the central bank in
balancing its objectives such as price stability and output stabilization are less severe when the
which clearly shows that monetary responses are sensitive to the oil price variance.
28
Output
1 4 7 10 13 16 19
-0.016
-0.014
-0.012
-0.01
-0.008
-0.006
Aggregate cons.
1 4 7 10 13 16 19
0.005
0.01
0.015
0.02
R cons.
1 4 7 10 13 16 19
0.005
0.01
0.015
0.02
NR cons.
1 4 7 10 13 16 19
-5
0
5
10
10-3
Non-oil output
1 4 7 10 13 16 19
-2
0
2
4
6
8
10-3 Total inf.
1 4 7 10 13 16 19
-3
-2
-1
0
10-3 Dom. inf.
1 4 7 10 13 16 19
-5
-4
-3
-2
-1
0
10-3 Core inf.
1 4 7 10 13 16 19
0
1
2
3
10-3
Interest rate
1 4 7 10 13 16 19
-2
0
2
4
10-3 Exchange rate
1 4 7 10 13 16 19
0.005
0.01
0.015
0.02
0.025
Marg. cost
1 4 7 10 13 16 19
-8
-6
-4
-2
0
10-3 Hours worked
1 4 7 10 13 16 19
-15
-10
-5
10-3
Figure 7: Responses to a Negative Oil Price Shock with s1
Output
1 4 7 10 13 16 19
-0.05
-0.04
-0.03
-0.02
-0.01
0Aggregate cons.
1 4 7 10 13 16 19
0
0.01
0.02
0.03
0.04
R cons.
1 4 7 10 13 16 19
0
0.01
0.02
0.03
0.04
NR cons.
1 4 7 10 13 16 19
-0.01
0
0.01
0.02
Non-oil output
1 4 7 10 13 16 19
-5
0
5
10
15
10-3 Total inf.
1 4 7 10 13 16 19
-6
-4
-2
0
10-3 Dom. inf.
1 4 7 10 13 16 19
-10
-5
0
10-3 Core inf.
1 4 7 10 13 16 19
-5
0
5
10
15
10-3
Interest rate
1 4 7 10 13 16 19
-0.01
0
0.01
Exchange rate
1 4 7 10 13 16 19
0
0.02
0.04
0.06
Marg. cost
1 4 7 10 13 16 19
-0.03
-0.02
-0.01
0
Hours worked
1 4 7 10 13 16 19
-0.04
-0.03
-0.02
-0.01
0
Figure 8: Responses to a Negative Oil Price Shock with s2
29
Output
1 4 7 10 13 16 19
-0.016
-0.014
-0.012
-0.01
-0.008
-0.006
Aggregate cons.
1 4 7 10 13 16 19
0.005
0.01
0.015
0.02
R cons.
1 4 7 10 13 16 19
0.005
0.01
0.015
0.02
NR cons.
1 4 7 10 13 16 19
-5
0
5
10
10-3
Non-oil output
1 4 7 10 13 16 19
0
2
4
6
8
10-3 Total inf.
1 4 7 10 13 16 19
-6
-4
-2
0
10-3 Dom. inf.
1 4 7 10 13 16 19
-6
-4
-2
0
10-3 Core inf.
1 4 7 10 13 16 19
-1
0
1
2
3
4
10-3
Interest rate
1 4 7 10 13 16 19
-2
-1
0
1
10-3 Exchange rate
1 4 7 10 13 16 19
0.005
0.01
0.015
0.02
0.025
Marg. cost
1 4 7 10 13 16 19
-15
-10
-5
0
10-3 Hours worked
1 4 7 10 13 16 19
-0.02
-0.015
-0.01
-0.005
Figure 9: Responses to a Negative Oil Price Shock with s1and ν= 1
Output
1 4 7 10 13 16 19
-0.04
-0.03
-0.02
-0.01
Aggregate cons.
1 4 7 10 13 16 19
0
0.01
0.02
0.03
0.04
R cons.
1 4 7 10 13 16 19
0
0.01
0.02
0.03
0.04
0.05
NR cons.
1 4 7 10 13 16 19
-0.01
0
0.01
0.02
Non-oil output
1 4 7 10 13 16 19
-5
0
5
10
15
10-3 Total inf.
1 4 7 10 13 16 19
-15
-10
-5
0
10-3 Dom. inf.
1 4 7 10 13 16 19
-20
-15
-10
-5
0
10-3 Core inf.
1 4 7 10 13 16 19
-0.01
0
0.01
0.02
Interest rate
1 4 7 10 13 16 19
-0.025
-0.02
-0.015
-0.01
-0.005
0
Exchange rate
1 4 7 10 13 16 19
0
0.02
0.04
0.06
0.08
Marg. cost
1 4 7 10 13 16 19
-0.05
-0.04
-0.03
-0.02
-0.01
0
Hours worked
1 4 7 10 13 16 19
-0.06
-0.04
-0.02
0
Figure 10: Responses to a Negative Oil Price Shock with s2and ν= 1
30
subsidy is removed. They are however larger in the regime of high oil price volatility.26 Due to
increased economic volatility, the challenges of the central bank intervention in mitigating the
impacts of subsidy removal when responding to contractionary supply shocks lie in designing a
flexible framework capable of adapting to economic shifts while balancing inflation and output
stabilization.
In terms of building capacity around crisis response in times of uncertainty, disentangling
the impact responses from shocks whose variances are changing over time is helpful for de-
veloping the policy coordination scenarios in the context of exploring the potential benefits
of coordination between monetary and fiscal authorities in achieving macroeconomic stability,
particularly in periods of stress and high volatility. Our results show that monetary policy
becomes more expansionary in a no-subsidy economy when oil prices are large and volatile.
This is useful for policymakers to give careful consideration to the consequences of combina-
tions of fiscal and monetary policy for jointly stabilizing the economy through scenario analysis.
Furthermore, the observed switches in monetary policy appear to suggest a need for a policy
transition to a multiple-mandate regime.27 Fiscal operations (quasi-fiscal activities) aimed at
boosting output would positively impact the supply-side drivers of inflation. It is clear that
the best policy framework required to effectively respond to abrupt changes in global economic
conditions should exhibit flexibility and adaptability.
Finally, in terms of the socio-political implications of fuel subsidy reforms, our empirical
results and policy simulations show that the vast majority of the subsidy and benefits of its
removal goes to better-off households. This explains why, especially during periods of high oil
price volatility, subsidy reductions could widen the wealth and income gap, and be a major
source of the different welfare consequences previously noted in the analysis.28 Potential ad-
justments to interest rates, reserve requirements or other policy tools that could be used to
counteract the contractionary effects on income and wealth distribution could be helpful to
inform future policy design and implementation, particularly in navigating regime-switching
dynamics in uncertain environments.
9 Concluding remarks and future research
We estimated a regime-switching DSGE model of the Nigerian economy. In this model, the
fiscal authority sells the imported fuel using a pricing rule that connotes an implicit subsidy
program and the monetary authority responds to stochastic regime shifts in shocks to oil prices.
We studied the impact of fuel subsidy removal on central bank behavior and optimal monetary
policy. We have three major findings.
A general finding is that monetary policy adjustments are time-varying and synchronize with
high oil price volatility states. In the high volatility state, the central bank policy rule delivers a
26This result is in line with the finding in Natal (2012) who makes a similar argument that oil price volatility
operates as a source of monetary policy trade-off amplification.
27See, for more discussions, Omotosho and Yang (2024) who discuss about a transition of the Central Bank of
Nigeria that has been involved in quasi-fiscal operations aimed at boosting output and may explain the observed
switch in the monetary policy regime after the financial crisis.
28The similar results can be found from previous empirical studies, such as Siddig et al. (2014), that have
focused on examining the implications of fuel subsidy for the Nigerian economy.
31
lower degree of interest rate smoothing and places a greater emphasis on the output gap relative
to inflation, compared to the low volatility state. Our study strengthens the connection between
regime-switching dynamics and monetary policy responses in times of uncertainty.
A second finding is that a complete, one-off removal of subsidy may lead to welfare losses due
to increased macroeconomic volatility, highlighting the need for careful policy consideration.
The optimal monetary rules are aggressive on inflation and exchange rates regardless of the
subsidy arrangement in place, and prescribe an initial monetary expansion in the absence of
fuel subsidy and in high-variance states. Based on our counterfactual simulations, we find that
the economy would have experienced increased macroeconomic instability with the realized
shocks in the absence of subsidy. The result suggests that possible central bank interventions
may be crucial in mitigating the impacts of subsidy removal, underscoring the importance of
coordinated policy responses capable of adapting to economic shifts.
Third, there are marked differences between the model-implied simulations of the two alter-
native economies suggesting that subsidy removal would play a significant role in affecting the
business cycle dynamics and economic performances. The effect on private consumption is very
different across the different types of households; thus offering an explanation for its potential
impact on widening the wealth and income gap, and the different welfare consequences. These
effects are more pronounced when the economy is in a high volatility environment where the
severity of the output contraction is amplified, thus emphasizing the importance of proactive
policy measures to manage economic volatility and safeguard welfare.
We provide a flexible, novel framework for policy analysis that can be general and geared
towards applications in resource-producing emerging countries. There are a number of possible
avenues for future research. An issue that certainly deserves further attention is informality, in
particular for emerging market economies. The heterogeneity of the model could be enriched
by considering a two-sector economy allowing for informality. Gabriel et al. (2012) study such
a model for India and find empirical evidence of a sizeable informal, low-skilled labor intensive
sector. Given the significant presence of informality in resource-producing developing countries,
the adverse impact of subsidy removal could be felt more by the informal sector. Incorporating
an informal sector should stylize the nature of productive activity in these economies. Monetary
policy actions would impact differently depending on the sector’s access to financial wealth.
References
Adebiyi, M. and Mordi, C. (2012). A dynamic stochastic general equilibrium (dsge) model of
exchange rate pass-through to domestic inflation in nigeria. ECOMOD Conference 2012.
Algozhina, A. (2022). Monetary policy rule, exchange rate regime, and fiscal policy cyclicality
in a developing oil economy. Energy Economics,112, 106126.
Allegret, J. P. and Benkhodja, M. T. (2015). External shocks and monetary policy in an oil
exporting economy (Algeria). Journal of Policy Modeling,37(4), 652–667.
Asiedu, E. and Lien, D. (2011). Democracy, foreign direct investment and natural resources.
Journal of International Economics,84(1), 99–111.
32
Barsky, R. B. and Kilian, L. (2004). Oil and the macroeconomy since the 1970s. Journal of
Economic Perspectives,18(4), 115–134.
Bergholt, D. and Larsen, V. (2016). ’Business cycles in an oil economy: Lessons from Norway’.
Technical report, Norges Bank.
Best, G. and Hur, J. (2019). Bad luck, bad policy, and learning? a markov-switching approach
to understanding postwar u.s. macroeconomic dynamics. European Economic Review,119,
55–78.
Bhattarai, K., Mallick, S. K., and Yang, B. (2021). Are global spillovers complementary or
competitive? need for international policy coordination. Journal of International Money and
Finance,110, 102291.
Bianchi, F. (2013). Regime switches, agents’ beliefs, and post-world war ii u.s. macroeconomic
dynamics. Review of Economic Studies,80(2), 463–490.
Bianchi, F. and Ilut, C. (2017). Monetary/fiscal policy mix and agents’ beliefs. Review of
Economic Dynamics,26, 113–139.
Bianchi, F. and Melosi, L. (2017). Escaping the great recession. American Economic Review,
107(4), 1030–1058.
Bjornland, H. C., Larsen, V. H., and Maih, J. (2018). Oil and macroeconomic (in)stability.
American Economic Journal: Macroeconomics,10(4), 128–151.
Brooks, P. and Gelman, A. (1998). General methods for monitoring convergence of iterative
simulations. Journal of Computational and Graphical Statistics,,7(4), 434–455.
Calvo, G. A. (1998). Capital Flows and Capital-Markets Crises. Journal of Applied Economet-
rics,1(1), 35–54.
Chen, X. and Macdonald, R. (2012). Realized and optimal monetary policy rules in an estimated
markov-switching dsge model of the united kingdom. Journal of Money, Credit and Banking,
44(6).
Clements, B. J., Coady, D., Fabrizio, S., Gupta, S., Alleyne, T. S. C., and Sdralevich, C. A.
(2013). Energy subsidy reform: lessons and implications. International Monetary Fund.
Coady, D. (2015). The unequal benefits of fuel subsidies revisited. International Monetary
Fund.
Coady, D., Parry, I., Le, N. P., and Shang, B. (2019). Global fossil fuel subsidies remain large:
An update based on country-level estimates. IMF Working Papers,19(89), 39.
Davig, T. and Doh, T. (2014). Monetary policy regime shifts and inflation persistence. Review
of Economics and Statistics,96(5), 862–875.
Dennis, A. (2016). Household welfare implications of fossil fuel subsidy reforms in developing
countries. Energy Policy,96, 597–606.
33
Estache, A. and Leipziger, D. (2009). Stuck in the middle: is fiscal policy failing the middle
class? Brookings Institution Press.
Fan, W. and Wang, Z. (2022). Whether to abandon or continue the petroleum product price
regulation in china? Energy Policy,165.
Ferrero, A. and Seneca, M. (2019). Notes on the underground: Monetary policy in resource-rich
economies. Journal of Money, Credit and Banking,51(4), 953–976.
Gabriel, V., Levine, P., Pearlman, J., and Yang, B. (2012). An Estimated Model of the Indian
Economy. In C. Ghate, editor, Handbook of the Indian Economy. Oxford University Press.
Gabriel, V., Levine, P., and Yang, B. (2016). An Estimated Open-Economy DSGE Model of the
Indian Economy with Financial Frictions. Chapter in Monetary Policy in India: A Modern
Macroeconomic Perspective, 2016, Springer Verlag .
Gabriel, V. J., Levine, P., and Yang, B. (2023). Partial dollarization and financial frictions in
emerging economies. Review of International Economics,31(2), 609–651.
Gali, J. and Monacelli, T. (2005). Monetary policy and exchange rate volatility in a small open
economy. Review of Economic Studies,72(3), 707–734.
Gelman, A., Carlin, J. B., Stern, H. S., and Rubin, D. B. (2004). Bayesian Data Analysis.
Chapman and Hall, 2nd edition.
Geweke, J. (1999). Using simulation methods for Bayesian econometric models: inference,
development, and communication. Econometric Reviews,18, 1–127.
Goldwyn, D. and Clabough, A. (2020). Confronting the resource curse: Advice for investors
and partners. Center for Energy Studies Working Paper.
Hamilton, J. D. (2003). What is an oil shock? Journal of Econometrics,113, 363–398.
Hollander, H., Gupta, R., and Wohar, M. E. (2018). The impact of oil shocks in a small open
economy new-keynesian dynamic stochastic general equilibrium model for an oil-importing
country: The case of south africa. Emerging Markets Finance and Trade, pages 1–26.
Holm-Hadulla, F. and Hubrich, K. (2017). Macroeconomic implications of oil price fluctuations:
a regime-switching framework for the euro area. ECB Working Paper Series No 2119.
Hove, S., Mama, A. T., and Tchana, F. T. (2015). Monetary policy and commodity terms of
trade shocks in emerging market economies. Economic Modelling,49, 53–71.
Hwang, I. and Zhu, X. (2024). State-dependent oil price shocks on inflation and the efficacy of
inflation targeting regime. Journal of International Money and Finance,144, 103077.
Iklaga, F. O. (2017). Estimated and optimised monetary policy rules for resource-rich emerging
economies. Ph.D. thesis, University of Surrey.
34
Inchauste, G. and Victor, D. G. (2017). The political economy of energy subsidy reform. World
Bank Publications.
Kass, R. E. and Raftery, A. E. (1995). Bayes Factors. Journal of the American Statistical
Association,90, 773–795.
Kilian, L. (2009). Not all oil price shocks are alike: Disentangling demand and supply shocks
in the crude oil market. American Economic Review,99(3), 1053–1069.
Kilian, L. and Vigfusson, R. J. (2017). The role of oil price shocks in causing u.s. recessions.
Journal of Money, Credit and Banking,49(8), 1747–1776.
Levine, P., Pearlman, J., Wright, S., and Yang, B. (2019). Information, VARs and DSGE
Models. University of Surrey, School of Economics Discussion Papers 1619.
Levine, P., Pearlman, J. G., Volpicella, A., and Yang, B. (2022). The Use and Mis-Use of
SVARs for Validating DSGE Models. University of Surrey, School of Economics Discussion
Papers 0522.
Liu, P. and Mumtaz, H. (2011). Evolving macroeconomic dynamics in a small open economy:
An estimated markov switching dsge model for the uk. Journal of Money, Credit and Banking,
43(7), 1443–1474.
Liu, Z., Waggoner, D. F., and Zha, T. (2011). Sources of macroeconomic fluctuations: A
regime-switching dsge approach. Quantitative Economics,2(2), 251–301.
Maih, J. (2015). Efficient perturbation methods for solving regime-switching dsge models.
Norges Bank Working papers 01/2015.
Maih, J., Mazelis, F., Motto, R., and Ristiniemi, A. (2021). Asymmetric monetary policy rules
for the euro area and the us. ECB Working Paper Series No 2587.
Medina, J. P. and Soto, C. (2005). Oil shocks and monetary policy in an estimated DSGE
model for a small open economy. Central Bank of Chile Working Paper,353.
Medina, J. P. and Soto, C. (2007). The chilean business cycles through the lens of a stochastic
general equilibrium model. Central Bank of Chile Working Papers,457.
Melina, G., Yang, S.-C. S., and Zanna, L.-F. (2016). Debt sustainability, public investment,
and natural resources in developing countries: The dignar model. Economic Modelling,52,
630–649.
Meng, X. L. and Wong, W. H. (1996). Simulating ratios of normalizing constants via a simple
identity: a theoretical exploration. Statistica Sinica,6, 831–860.
Monacelli, T. (2005). Monetary policy in a low pass-through environment. Journal of Money,
Credit and Banking, pages 1047–1066.
Natal, J. (2012). Monetary policy response to oil price shocks. Journal of Money, Credit and
Banking,44(1), 53–101.
35
Ncube, M. and Balma, L. (2017). Oil shocks, public investment and macroeconomic and fis-
cal sustainability in Nigeria: Simulations using a DSGE model. QGRL Working Paper,
(2017/01).
Omotosho, B. S. (2019a). Business cycle fluctuations in nigeria: Some insights from an estimated
dsge model. MPRA Paper 98351, University Library of Munich, Germany.
Omotosho, B. S. (2019b). Oil price shocks, fuel subsidies and macroeconomic (in)stability in
nigeria. CBN Journal of Applied Statistics,10(2), 1–38.
Omotosho, B. S. (2022). Oil price shocks and monetary policy in resource-rich economies: Does
capital matter? Journal of Economic Dynamics and Control,143, 104479.
Omotosho, B. S. and Yang, B. (2024). Oil price shocks and macroeconomic dynamics in resource-
rich emerging economies under regime shifts. Journal of International Money and Finance,
144, 103082.
Oyinlola, M. and Adetunji, M. (2009). A bound testing analysis of exchange rate pass-through
to aggregate import prices in nigeria: 1980-2006. Journal of Economic Development,34(2).
Poghosyan, T. and Beidas-Strom, S. (2011). An estimated dynamic stochastic general equilib-
rium model of the jordanian economy. IMF Working Paper,11(28).
Rahman, S. and Serletis, A. (2010). The asymmetric effects of oil price and monetary policy
shocks: A nonlinear var approach. Energy Economics,32(6), 1460–1466.
Ramey, V. A. and Vine, D. J. (2011). Oil, automobiles, and the us economy: How much have
things really changed? NBER Macroeconomics Annual 2010,25, 333–367.
Rasaki, M. G. and Malikane, C. (2015). Macroeconomic shocks and fluctuations in african
economies. Economic Systems,39(4), 675–696.
Rentschler, J., Kornejew, M., and Bazilian, M. (2017). Fossil fuel subsidy reforms and their
impacts on firms. Energy Policy,108, 617–623.
Romero, R. E. (2008). Monetary policy in oil-producing economies. CEPS Working Paper,
(169).
Schorfheide, F. (2005). Learning and monetary policy shifts. Review of Economic Dynamics,
8(2), 392–419.
Senbeta, S. (2011). A small open economy new keynesian dsge model for a foreign exchange
constrained economy. Technical report, University of Antwerp, Faculty of Applied Economics.
Siddig, K., Aguiar, A., Grethe, H., Minor, P., and Walmsley, T. (2014). Impacts of removing
fuel import subsidies in nigeria on poverty. Energy Policy,69, 165–178.
Smets, F. and Wouters, R. (2007). Shocks and frictions in us business cycles: A bayesian dsge
approach. American economic review,97(3), 586–606.
36
Taylor, M. (2020). Energy subsidies: Evolution in the global energy transformation to 2050.
International Renewable Energy Agency, Abu Dhabi.
Teixeira, A. A., Forte, R., and Assun¸c˜ao, S. (2017). Do countries’ endowments of non-renewable
energy resources matter for fdi attraction? a panel data analysis of 125 countries over the
period 1995–2012. International Economics,150, 57–71.
Wolden-Bache, I., Brubakk, L., and Maih, J. (2008). ’Estimating a DSGE model for Norway
with optimal monetary policy’. Technical report, Working Paper, Norges Bank.
37
