I. Introduction

The role of monetary policy and inflation dynamics in the wake of an oil price shock have motivated a large body of research, including Hamilton (1996), Bernanke, Gertler, and Watson (1997), Leduc and Sill (2004), and Bodenstein, Erceg, and Guerrieri (2008). However, less attention has been paid to the origins of oil price changes, for example, Kilian, Rebucci, and Spatafora (2007), Hamilton (2008) and Kilian (2008b). More prominently, Kilian (2009) argues that there are three main sources of oil price shocks that lead to oil price fluctuations: oil supply shocks; world aggregate demand shocks, and the shifts in the precautionary demand for oil due to uncertainties about future oil supply.

Our paper contributes to the literature by showing the impacts of oil price movements caused by various sources. We develop a sticky-price DSGE model through which we analyze the effects of various shocks studied in Kilian (2009), namely, the increase in aggregate demand, unexpected oil supply disruption and the precautionary oil demand on an (oil importing) small open economy (SOE) and on the rest of the world (ROW). In particular, we assume that the world economy is composed of a domestic SOE and a continuum of other small open economies. Effectively, an SOE has a negligible effect on the world economy. Hence, the ROW is regarded as a closed economy and oil demand and price are determined in the ROW. Oil supply is assumed to follow an exogenous process.²

We show that, although both productivity and fiscal policy shocks can be regarded as “demand” shocks for the oil market, leading to a rise in the world aggregate demand and a subsequent surge in the real price of oil, their impacts on inflation are different. In the case of a productivity increase, the marginal cost of the firms decline, which leads to lower CPI inflation. On the other hand, an increase in government spending directly creates additional demand for the consumption goods, while putting upward pressure on the marginal cost of the firms and the aggregate price level. We also compare the effects of expected and unexpected oil supply shocks. In the case of an expected future oil supply shortage, a precautionary oil demand raises the current real price of oil. Our results suggest that distinguishing between the causes of oil price shocks is very important, if not crucial, in both identifying the impact of oil price shocks on key macroeconomic variables and determining the appropriate policy responses to deal with them.

There are a few features of our model that are worth emphasizing. First, we incorporate oil demand in a relatively simple manner as in Blanchard and Gali (2008), while it is this simplicity that allows us to get analytical insights into both the causes and the international transmission mechanism of the oil price shocks. Contrary to their analysis, however, oil price is endogenously determined in our model. Since oil is included in the production function, oil demand and supply shocks in the ROW create oil price fluctuations. Given that oil price changes are not resilient to the global macroeconomic developments, it is more appropriate that oil prices are determined inside the model.

Second, we model the SOE and the ROW together, so that we are able to analyze the impacts of these shocks both in an SOE and a closed economy. In the SOE framework, ROW is traditionally modelled by a few exogenous processes, ignoring possible (and important) channels through which shocks affecting the ROW are transmitted to the SOE.³ Although it is implicitly assumed in the existing literature that the effects of an oil shock can be analyzed by the introduction of an exogenous oil price process, this method is misleading as it does not consider the causes or effects of oil price changes, which eventually affects the SOE through real exchange rate and trade channels. Our results confirm the contribution of this detailed modelling exercise as each shock considered in the paper influences the real price of oil and the economies differently and has specific transmission channels.

Third, our model reveals that a precautionary oil demand shock is likely to cause sharp fluctuations in the price of oil. If there is an expectation of lower future oil supply, the current spot prices will adjust immediately, causing expectations-linked oil price volatility in the world oil market.

Recently, the rapid increases in the world oil prices have stimulated a renewed interest on the macroeconomic effects of oil price shocks. The fact that the oil price increases started in the early 2000s have led to very different outcomes than the 1970s in particular has received much attention. Nordhaus (2007) and Segal (2007) propose that the main reason for a smaller impact of oil shocks is the change in the transmission mechanism which helps to impede oil prices feed through to core inflation, and therefore enables a less aggressive monetary policy. Blanchard and Gali (2008) suggest that more flexible labor markets, more credible and stronger anti-inflationary stance of monetary policies and declining oil intensities in the major economies have helped to curb the effects of higher oil prices.

As Woodford (2007) argues, however, these explanations are not convincing enough as they ignore the endogenous responses of the real price of oil to the global economic conditions. In this line of thought, Elekdag and Laxton (2007) and Kilian (2008a, 2008b) argue that one of the reasons for the oil price increases of the 2000s is the increased world aggregate demand, possibly due to the world productivity increases, contrary to the oil price shocks mainly driven by supply side changes in the 1970s.⁴ Relative to this literature, we take a step further by modelling different causes of oil price changes for domestic economy, as well as the world economy with a clear consideration of possible demand and supply side developments. The remainder of the paper is organized as follows. In Section 2, the structure of the model is laid out. The oil market equilibrium and the equilibrium conditions are derived in Section 3. We discuss the findings in Section 4. The conclusions are outlined in Section 5.

II. The Small Open Economy Model

We develop an open economy sticky price DSGE model with a representative household, producers, a government and a monetary authority. The model shares its basic features with many new Keynesian SOE models, including the benchmark models of Gali and Monacelli (2005) and Clarida, Gali, and Gertler (2001), although several differences remain. In order to capture oil shocks, we follow Blanchard and Gali (2008) by introducing an oil input in the production function.

Oil market equilibrium is determined in the ROW. In order to highlight our interest on a SOE and its interlinkages with the foreign economy (ROW), variables without superscripts refer to the home economy, while variables with a star indicate the foreign economy variables. Small letters denote percentage deviations of the respective variables from their steady-state levels. We briefly sketch the model and present the log-linearized equations in Table 1, while the details of the model are provided in the Appendix.

Table 1.

Model in Log-Linearized Form: Behavioral Equations for the SOE

Table 1.

Model in Log-Linearized Form: Behavioral Equations for the SOE

${\pi }_{H,t}=\beta E_t\{{\pi }_{H,t+1}\}+\lambda {\widehat{mc}}_t$	Phillips curve
${\widehat{mc}}_t=\left[\frac{\eta \phi }{1+(1-\eta )\phi }+\frac{{\Omega }}{1-\alpha }\right]x_t$	Marginal cost
$\begin{array}{l}{y}_t=E_t\left\{y_{t+1}\right\}-\frac{(1-G_y)}{\sigma }(r_t-E_t\left\{{\pi }_{t+1}\right\})\dots \\ \dots -G_y{E}_t\left\{{\Delta }{g}_{t+1}\right\}-\psi E_t\left\{{\Delta }{q}_{t+1}\right\}\end{array}$	IS curve
$\begin{array}{l}\overline{y_t}=\frac{1}{\left[{{\Psi }}_3+\frac{{\Omega }}{1-\alpha }\right]}(-\mu +{{\Psi }}_1{a}_t+{{\Psi }}_4{b}_t\dots \\ \dots -\left[\frac{{{\Psi }}_2}{1-G_y}-\frac{{\Omega }}{1-\alpha }\right](y_t^{*}-G_y{g}_t^{*})+\frac{{\Omega }}{1-\alpha }{G}_y{g}_t-{{\Psi }}_4{\tilde{P}}_{o,t}^{*}\end{array}$	Flexible price output
$x_t=y_t-{\overline{y}}_t$	Output gap
$q_t={\Omega }(y_t-G_y{g}_t)-{\Omega }(y_t^{*}-G_y{g}_t^{*})$	Real exchange rate
rt =ϕππt +ϕxxt	Monetary policy rule
${\pi }_t={\pi }_{H,t}+\frac{\alpha }{1-\alpha }{\Delta }{q}_t$	CPI inflation
$e_t=p_t-p_t^{*}+q_t$	Nominal exchange rate
${\tilde{p}}_{o,t}={\tilde{p}}_{o,t}^{*}+q_t.$	Real oil price

View Table

Table 1.

Model in Log-Linearized Form: Behavioral Equations for the SOE

${\pi }_{H,t}=\beta E_t\{{\pi }_{H,t+1}\}+\lambda {\widehat{mc}}_t$	Phillips curve
${\widehat{mc}}_t=\left[\frac{\eta \phi }{1+(1-\eta )\phi }+\frac{{\Omega }}{1-\alpha }\right]x_t$	Marginal cost
$\begin{array}{l}{y}_t=E_t\left\{y_{t+1}\right\}-\frac{(1-G_y)}{\sigma }(r_t-E_t\left\{{\pi }_{t+1}\right\})\dots \\ \dots -G_y{E}_t\left\{{\Delta }{g}_{t+1}\right\}-\psi E_t\left\{{\Delta }{q}_{t+1}\right\}\end{array}$	IS curve
$\begin{array}{l}\overline{y_t}=\frac{1}{\left[{{\Psi }}_3+\frac{{\Omega }}{1-\alpha }\right]}(-\mu +{{\Psi }}_1{a}_t+{{\Psi }}_4{b}_t\dots \\ \dots -\left[\frac{{{\Psi }}_2}{1-G_y}-\frac{{\Omega }}{1-\alpha }\right](y_t^{*}-G_y{g}_t^{*})+\frac{{\Omega }}{1-\alpha }{G}_y{g}_t-{{\Psi }}_4{\tilde{P}}_{o,t}^{*}\end{array}$	Flexible price output
$x_t=y_t-{\overline{y}}_t$	Output gap
$q_t={\Omega }(y_t-G_y{g}_t)-{\Omega }(y_t^{*}-G_y{g}_t^{*})$	Real exchange rate
rt =ϕππt +ϕxxt	Monetary policy rule
${\pi }_t={\pi }_{H,t}+\frac{\alpha }{1-\alpha }{\Delta }{q}_t$	CPI inflation
$e_t=p_t-p_t^{*}+q_t$	Nominal exchange rate
${\tilde{p}}_{o,t}={\tilde{p}}_{o,t}^{*}+q_t.$	Real oil price

III. Rest of the World and the Oil Market

A. Equilibrium in the Rest of the World

Apart from being asymmetric in size, SOE and ROW share the same preferences, technology, market structure for the consumption goods sector, and same structures for the monetary and fiscal policies. The price of oil is determined according to the macroeconomic developments in the ROW, which is regarded as a closed economy. In Table 2, we present the log-linearized equations for the ROW.

Table 2.

Model in Log-Linearized Form: Behavioural Equations for the ROW and the Oil Market

Table 2.

Model in Log-Linearized Form: Behavioural Equations for the ROW and the Oil Market

${\pi }_t^{*}=\beta E_t\left\{{\pi }_{t+1}^{*}\right\}+\lambda {\widehat{mc}}_t^{*}$	Phillps curve
${\widehat{mc}}_t^{*}=\left[\left(\frac{\sigma }{1-G_y}\right)+\frac{1+\phi -\eta }{\eta }\right]x_t^{*}$	Marginal cost
$\begin{array}{l}{y}_t^{*}=E_t\left\{y_{t+1}^{*}\right\}-\frac{(1-G_y)}{\sigma }(r_t^{*}-E_t\left\{{\pi }_{t+1}^{*}\right\})\dots \\ \dots -G_y{E}_t\left\{{\Delta }{g}_{t+1}^{*}\right\}\end{array}$	IS curve
$\begin{array}{l}{\overline{y}}_t^{*}=\frac{1}{\left[(\frac{\sigma }{1-G_y})+\frac{1+\phi -\eta }{\eta }\right]}(-\mu +(1+\phi )a_t^{*}\dots \\ \dots +\frac{(1-\eta )(1+\phi )}{\eta }{b}_t^{*}+\frac{\sigma G_y}{1-G_y}{g}_t^{*}+\frac{(1-\eta )(1+\phi )}{\eta }{o}_t^s\end{array}$	Flexible price output
$x_t^{*}=y_t^{*}-{\overline{y}}_t^{*}$	Output gap
$r_t^{*}={\varphi }_{\pi }{\pi }_t^{*}+{\varphi }_x{x}_t^{*}$	Monetary policy rule
${\tilde{p}}_{o,t}^{*}={{\Gamma }}_1{y}_t^{*}-{{\Gamma }}_2{g}_t^{*}-{{\Gamma }}_3{a}_t^{*}-{{\Gamma }}_4{b}_t^{*}-{{\Gamma }}_5{o}_t^s$	Real oil price

View Table

Table 2.

Model in Log-Linearized Form: Behavioural Equations for the ROW and the Oil Market

${\pi }_t^{*}=\beta E_t\left\{{\pi }_{t+1}^{*}\right\}+\lambda {\widehat{mc}}_t^{*}$	Phillps curve
${\widehat{mc}}_t^{*}=\left[\left(\frac{\sigma }{1-G_y}\right)+\frac{1+\phi -\eta }{\eta }\right]x_t^{*}$	Marginal cost
$\begin{array}{l}{y}_t^{*}=E_t\left\{y_{t+1}^{*}\right\}-\frac{(1-G_y)}{\sigma }(r_t^{*}-E_t\left\{{\pi }_{t+1}^{*}\right\})\dots \\ \dots -G_y{E}_t\left\{{\Delta }{g}_{t+1}^{*}\right\}\end{array}$	IS curve
$\begin{array}{l}{\overline{y}}_t^{*}=\frac{1}{\left[(\frac{\sigma }{1-G_y})+\frac{1+\phi -\eta }{\eta }\right]}(-\mu +(1+\phi )a_t^{*}\dots \\ \dots +\frac{(1-\eta )(1+\phi )}{\eta }{b}_t^{*}+\frac{\sigma G_y}{1-G_y}{g}_t^{*}+\frac{(1-\eta )(1+\phi )}{\eta }{o}_t^s\end{array}$	Flexible price output
$x_t^{*}=y_t^{*}-{\overline{y}}_t^{*}$	Output gap
$r_t^{*}={\varphi }_{\pi }{\pi }_t^{*}+{\varphi }_x{x}_t^{*}$	Monetary policy rule
${\tilde{p}}_{o,t}^{*}={{\Gamma }}_1{y}_t^{*}-{{\Gamma }}_2{g}_t^{*}-{{\Gamma }}_3{a}_t^{*}-{{\Gamma }}_4{b}_t^{*}-{{\Gamma }}_5{o}_t^s$	Real oil price

Table 3.

Model in Log-Linearized Form: Parameters

Table 3.

Model in Log-Linearized Form: Parameters

$\lambda =\frac{(1-\theta )(1-\beta \theta )}{\theta }$

$\psi =(1-G_y)\alpha [\frac{\gamma }{1-\alpha }+\gamma -\frac{1}{\sigma }]$

${\Omega }=1/\left(\frac{1-G_y}{\sigma }+\psi \right)$

${{\Psi }}_1=\frac{\eta (1+\phi )}{1+(1-\eta )\phi },{{\Psi }}_2=\frac{\eta \sigma }{1+(1-\eta )\phi }$

${{\Psi }}_3=\frac{\eta \phi }{1+(1-\eta )\phi },{{\Psi }}_4=\frac{(1-\eta )(1+\phi )}{1+(1-\eta )\phi }$

${{\Gamma }}_1=\frac{\sigma }{1-G_y}+\frac{(1+\phi )}{\eta },{{\Gamma }}_2=\frac{\sigma G_y}{1-G_y}$

${{\Gamma }}_3=1+\phi ,{{\Gamma }}_4=\frac{(1-\phi )(1-\eta )}{\eta },{{\Gamma }}_5=\frac{1+(1-\eta )\phi }{\eta }.$

View Table

Table 3.

Model in Log-Linearized Form: Parameters

$\lambda =\frac{(1-\theta )(1-\beta \theta )}{\theta }$

$\psi =(1-G_y)\alpha [\frac{\gamma }{1-\alpha }+\gamma -\frac{1}{\sigma }]$

${\Omega }=1/\left(\frac{1-G_y}{\sigma }+\psi \right)$

${{\Psi }}_1=\frac{\eta (1+\phi )}{1+(1-\eta )\phi },{{\Psi }}_2=\frac{\eta \sigma }{1+(1-\eta )\phi }$

${{\Psi }}_3=\frac{\eta \phi }{1+(1-\eta )\phi },{{\Psi }}_4=\frac{(1-\eta )(1+\phi )}{1+(1-\eta )\phi }$

${{\Gamma }}_1=\frac{\sigma }{1-G_y}+\frac{(1+\phi )}{\eta },{{\Gamma }}_2=\frac{\sigma G_y}{1-G_y}$

${{\Gamma }}_3=1+\phi ,{{\Gamma }}_4=\frac{(1-\phi )(1-\eta )}{\eta },{{\Gamma }}_5=\frac{1+(1-\eta )\phi }{\eta }.$

Table 4.

Model in Log-Linearized Form: Exogenous Processes

Table 4.

Model in Log-Linearized Form: Exogenous Processes

Small open economy
	at = ρaat−1 + εa,t	Labor productivity
	$a_t={\rho }_a{a}_{t-1}+{\epsilon }_{a,t}+0.3{\epsilon }_{a^{*},t}$	Labor productivity with spillover effects
	bt = ρbbt−1 + εb,t	Efficiency of oil use
	gt = ρggt−1 + εg,t	Government spending
Rest of the world and oil market
	$a_t^{*}={\rho }_{a^{*}}{a}_{t-1}^{*}+{\epsilon }_{a^{*},t}$	Labor productivity
	$b_t^{*}={\rho }_{b^{*}}{b}_{t-1}^{*}+{\epsilon }_{b^{*},t}$	Efficiency of oil use
	$g_t^{*}={\rho }_{g^{*}}{g}_{t-1}^{*}+{\epsilon }_{g^{*},t}$	Government spending
	$o_t^s={\rho }_o{o}_{t-1}^s+{\epsilon }_{o,t}$	Oil supply

View Table

Table 4.

Model in Log-Linearized Form: Exogenous Processes

Small open economy
	at = ρaat−1 + εa,t	Labor productivity
	$a_t={\rho }_a{a}_{t-1}+{\epsilon }_{a,t}+0.3{\epsilon }_{a^{*},t}$	Labor productivity with spillover effects
	bt = ρbbt−1 + εb,t	Efficiency of oil use
	gt = ρggt−1 + εg,t	Government spending
Rest of the world and oil market
	$a_t^{*}={\rho }_{a^{*}}{a}_{t-1}^{*}+{\epsilon }_{a^{*},t}$	Labor productivity
	$b_t^{*}={\rho }_{b^{*}}{b}_{t-1}^{*}+{\epsilon }_{b^{*},t}$	Efficiency of oil use
	$g_t^{*}={\rho }_{g^{*}}{g}_{t-1}^{*}+{\epsilon }_{g^{*},t}$	Government spending
	$o_t^s={\rho }_o{o}_{t-1}^s+{\epsilon }_{o,t}$	Oil supply