Asymmetric and nonlinear oil price pass-through to economic growth in Croatia: Do oil-related policy shocks matter?

Highlights

• The pass-through of oil price shocks to economic activity is tested in a small and open economy.

• Oil price indicators are expressed in linear, asymmetric, and asymmetric and nonlinear form.

• The asymmetric, nonlinear and direct short-run effect of oil price shocks on economic activity is confirmed.

• Oil-related policy shock dummy variables are not important determinants of real GDP growth.

Abstract

Recent turbulent developments in the world oil market have increased the volatility of oil prices and rekindled interest in studying their impact on macroeconomic activity. Using the vector autoregression framework, this paper investigates the impact of oil price shocks, expressed in linear, asymmetric and nonlinear form, on real gross domestic product (GDP) growth in an oil-importing small open economy, i.e., Croatia, for the period 1995:Q1-2019:Q4. The results corroborate the presence of the asymmetric, nonlinear and direct short-run effect of oil price shocks on real GDP growth. A decrease in oil prices leads to a statistically significant increase in real GDP and its more volatile behavior, while an increase in oil prices causes a considerable drop in economic activity only in the fourth quarter. The oil-related policy shock variables, fuel market liberalization and fuel taxation, turned out to be insignificant predictors of GDP growth in an asymmetric and nonlinear model specification.

Introduction

Oil is (still) the most globally traded commodity and the volatility of its prices significantly affects world economies (Van Eyden et al., 2019). The literature on the effects of oil prices on macroeconomic variables is quite extensive with an ongoing debate on the relationship between oil prices and economic activity. Starting with the oil crises in the 1970s and the pioneering work of James Hamilton (1983), the nature of this relationship has received significant attention from researchers with their scientific productivity being positively related to real oil price increases. Hamilton's conclusions (1983), indicating that oil price increases had preceded almost all post-war economic recessions in the United States (US) strongly affecting output levels, were widely accepted and confirmed by many studies (for a recent discussion of the most influential research, see Herrera et al., 2019).

However, as recent studies have shown, the oil price-gross domestic product (GDP) relationship weakened or even disappeared in the 1980s, 1990s and early 2000s (Lee et al., 1995; Hooker, 1996, 2002; Herrera and Pesavento, 2009; Blanchard and Riggi, 2013; Herrera et al., 2019). This called into doubt the conventional wisdom, i.e., Hamilton's conclusion, and raised a series of questions about the nature of that relationship. Some authors explained the lack of a strong relationship by several factors such as monetary policy adjustment, sectoral restructuring, more flexible labor markets or increased energy efficiency (Blanchard and Riggi, 2013). Some authors constructed alternative oil price indicators to identify possible alterations (mainly nonlinearities) in the oil price-GDP relationship, namely: i) the asymmetric response of GDP (Mork, 1989; Hamilton, 1996, 2003; Bernanke et al., 1997; Davis and Haltiwanger, 2001; Jiménez-Rodríguez and Sánchez, 2005), ii) the importance of oil price volatility (Lee et al., 1995; Hooker, 1996), and iii) the importance of the surprise effect of the oil price change (Lee et al., 1995; Ferderer, 1996; Hamilton, 1996, 2010).

Scholars have devoted considerable effort to explaining the channels through which macroeconomic variables, namely real GDP, nonlinearly and asymmetrically respond to oil price shocks. As Herrera et al. (2015) pointed out, considerable attention has been paid to the reallocation channel (Hamilton, 1988; Davis and Haltiwanger, 2001; Herrera et al., 2015), the operating cost channel (Hamilton, 1988; Kilian, 2014), and the uncertainty channel (Plante and Traum, 2014). According to the former, due to oil price shocks, lower consumption of energy-intensive durable goods leads to sectoral reallocations of labor and capital towards more energy-efficient sectors and activities (Hamilton, 1988; Edelstein and Kilian, 2007). A decline in spending may be exacerbated by an increase in operating costs of energy-intensive durables, such as automobiles (Hamilton, 1988; Kilian, 2014). An increase and decrease in oil prices causing these disturbances would intensify or attenuate movements in real output, respectively, creating an asymmetry. The latter suggests a negative reaction of investment to oil price increases with increasing uncertainty about future oil prices, resulting in postponing or even abandoning investments (Ferderer, 1996; Elder and Serletis, 2010). Uncertainty about oil prices also affects the postponement of durable goods purchases, while a possible increase in precautionary savings further reduces consumption and adversely affects real output. Since a decrease in oil prices does not have such strong effect on uncertainty about the future and thus on real output, there is an asymmetry.

Oil price volatility is gaining more and more popularity over real oil prices in academic studies (Van Eyden et al., 2019). However, most of the literature on the impact of oil prices on the economy has been focused on the US (for a review, see Herrera et al., 2019). Accordingly, Narayan et al. (2014) concluded that outside of the US, very little is known about the impact of oil prices on economic growth. This observation is especially true for small and open economies (SMOPEC) that are highly dependent on oil imports, and thus sensitive and vulnerable to oil price shocks. For example, Dybczak et al. (2008) analyzed the case of the Czech Republic, while Papapetrou (2001) analyzed the case of Greece, as did Miguel et al. (2005) and Cuñado and Pérez de Gracia (2003), along with other European Union (EU) member states. Not only are the cases of SMOPEC underrepresented in the literature, but the rare existing studies indicate variations in their results, implying the necessity of further research. This is particularly related to the asymmetric and nonlinear effects of oil price shocks in oil-importing SMOPECs, whereas an increase in oil prices usually, but not always, leads to worsening of macroeconomic performance, while the effect of an oil price decrease remains open to debate.

This paper investigates the pass-through of oil prices on economic growth, as measured by GDP growth, within the vector autoregressive (VAR) framework in Croatia for the period 1995:Q1-2019:Q4. To that end, it uses linear and several nonlinear oil price indicators. In addition to GDP and different oil price indicators, the created VAR model incorporates the industrial production index (IPI), the real effective exchange rate (REER) and a crisis dummy variable.

The main novelty of this paper lies in expanding this basic model with a set of additional oil-related policy shocks variables, aiming to explore the nature of the relationship of interest in a heavily and increasingly oil-dependent SMOPEC such as Croatia. Policy shocks are related to fuel market liberalization and fuel taxation. The former refers to the liberalization processes encouraged by deregulation and privatization of the largest Croatian national fuel company called INA (Industrija nafte), while the latter refers to the changes in excise duties on fuel. To the best of our knowledge, a multivariate VAR model that combines asymmetric and nonlinear impacts of oil prices with oil-related policy shock variables has not been developed in the related literature. The paper also adds to the literature by increasing the understanding of the influence of oil price volatility and policy shocks on economic activity not only for Croatia, but also for other European SMOPECs that are heavily reliant on oil imports. A deeper understanding should better guide policy decision-makers to achieve sustainable economic growth.

The next section provides a conceptual background based upon which the multivariate oil price-GDP VAR model is constructed. Section 3 describes the method and the data used to analyze the relationship of interest, while Section 4 displays and discusses the results. Section 5 concludes the paper.