Elsevier

Energy Economics

Volume 43, May 2014, Pages 41-47
Energy Economics

Oil price uncertainty and manufacturing production

https://doi.org/10.1016/j.eneco.2014.02.004Get rights and content

Highlights

  • The effect of oil price uncertainty on manufacturing production

  • Use bivariate GARCH-in-mean VAR simultaneously.

  • Find a negative and significant impact.

  • Asymmetric response to positive and negative oil shocks

Abstract

Given the rapid rise and volatility of oil prices, the paper investigates the effect of oil price uncertainty on the South African manufacturing production using monthly observations covering the period 1974:02 to 2012:12. In addition, we quantify the responses of manufacturing production to positive and negative oil price shocks. We examine the dynamic relationship using a bivariate GARCH-in-mean VAR simultaneously estimated with a full information maximum likelihood technique. The conditional standard deviation of the forecast of the growth of US crude oil imported acquisition cost by refiners is used as a measure of oil price uncertainty. Our results show that oil price uncertainty negatively and significantly impacts on South Africa's manufacturing production. We also find that the responses of manufacturing production to positive and negative shocks are asymmetric.

Introduction

The world has seen sharp increases in oil price that largely impacted on economic growth of many emerging economies in recent years. It is argued that since the 1980s, oil prices have increased in real terms which adversely affected GDP, particularly that of oil importing economies. The rapid rise and volatility in oil prices, which started during the early 2000s significantly created great concerns amongst investors over macroeconomic variables such as real gross domestic product (GDP), savings/investment capacity and employment (Nkomo, 2006a). Kohler (2006) warns that such persisting sharp increases in oil prices will adversely affect economies across the world. Elder and Serletis (2010) present the theoretical channels through which oil price shocks affect economic activities. They first identify the real balances and monetary policy channel through which an increase in oil price tends to increase the overall level of prices leading to a reduction in real money balances held by households and firms and ultimately aggregate demand. They also argue that the second channel – the income transfer channel – emphasizes the transfer of income from oil importing countries to oil exporting countries associated with oil price increases.

The 2008/09 global financial crisis resulted in severe structural inefficiencies on many African economies. During this period, South Africa's manufacturing sector contracted with 10.4%, losing about R31 billion in GDP (2005 constant prices). Furthermore, the sector shed over 200,000 job opportunities. Positive oil shocks were also witnessed during the crisis which compelled a number of developing economies to adjust their structural reforms, particularly domestic petroleum pricing system which formed a critical component of macroeconomic policies. The South African manufacturing sector relies heavily on energy and oil which is largely imported from Middle East and West African countries. It is against this view that spikes in oil prices would likely impact on manufacturing production hence the overall GDP. Nour (2011) argues that too much dependence on oil may pose challenges to policy makers, through uncertainty in domestic growth given a very volatile international oil market.

Historical data indicate that prior to the 1970s, the price of oil had been fairly stable. However in the 1970s, there was a budge towards a sharp increase in oil prices, shooting to over $40 a barrel and by end of the decade, as it became even more volatile, rising above $100 a barrel (at current prices). The price of oil continued to be volatile but significantly declined to as low as $20 in the 1980s and these low prices were observed till the end of 2001. However, towards the end of 2007 the price of oil started accelerating again, rising sharply in the onset of the 2008–2009 recession recording an all-time high of $145 a barrel (Hamilton, 2009). The discourse on volatility in oil prices having an effect on macroeconomic components such as real GDP and inflation has been a pertinent issue. The literature on the effect of oil price shocks on economic activities has grown quite large in recent years and early work by Hamilton (1988), Mork (1989), Lee et al. (1995), and Hooker (1996), indicate a negative relationship between oil price shocks and economic activities.

Their results are also confirmed by Hamilton (2009), who investigated the effects of these oil price fluctuations on the US macroeconomic components during the oil shock period of 2007–2008. Hooker (2002) also argues that oil shocks significantly contributed to the U.S. core inflation and productivity before 1981. Such studies are also supported by Barsky and Kilian, 2002, Barsky and Kilian, 2004 and Edelstein and Kilian, 2007a, Edelstein and Kilian, 2007 who further extended their empirical study by investigating oil price shocks on nonresidential fixed investment. The researchers purport that oil price shocks may affect this particular economic activity through a “supply channel” in which an increase in the cost of production, driven by an increase in real oil prices in-turn decreases production (demand channel).

A similar conclusion was derived from a recent study by Elder and Serletis (2010), who estimated a model with disaggregated measures of investments and they found that oil price uncertainty negatively and significantly impacts on real output. They show that volatility in oil prices negatively affects manufacturing production given that firms' decisions to invest are clouded by doubts and uncertainties about future returns. In their study, they conducted an analysis on “real option” models where they analysed firms' expenditure patterns given the uncertainty of future returns and they found that volatility in oil prices tend to depress some components of aggregate investment. The real option theory states that firms' decisions to invest in an uncertain future return environment is either delayed or in some instances abandoned as uncertainty about the future returns intensifies. It is a widely accepted phenomenon that most firms fall under this category as their investment and production decisions are frequently influenced by potential future returns; for example, production levels in the automotive industry are driven by positive sales' outlook as they do not only incur non-recoverable human capital associated costs (hiring and training labour) but also physical capital expenditure (equipment and machinery). Hamilton (2009), states that the immediate effects of oil price changes significantly impact on purchases of motor vehicles which further result in the reduction of income for both buyers and sellers (manufacturers) of vehicles. Given these factors contributing to investment decisions, manufacturers will then choose to lower production rather than employing a complete exit strategy when faced with uncertainty of future returns. Such a stance is also echoed by research conducted by Bernanke (1983) and Pindyck (1991) who argue that volatility in oil prices creates uncertainty about future oil prices hence firms tend to irreversibly postpone their investment decisions. Their results have recently been backed up by Lee et al. (2011) who estimated an equation of oil price shocks on firms' investment decisions by exploring U.S. manufacturing data with over 3000 firms. Lee et al. (2011) results maintain that firm stock price volatility accompanied by future oil price uncertainty, adversely impacts firms' investment decisions for at least the first and second year of the initial shock.

A large number of studies on the impact of oil prices in South Africa have also been conducted by researchers such as Dagut (1978); Kantor and Barr (1986); Van der Merwe and Meijer (1990); Wakeford (2006). The authors mostly focused on the effects of the shocks on domestic inflation, the gold price and the terms of trade. A computable general equilibrium model study which simulated the economy-wide and sectoral impacts of an oil price hike has also been undertaken by PROVIDE (2005). Swanepoel (2006) then extended this analysis by employing a vector autoregressive (VAR) model to examine the impact of three external shocks that included oil prices on South African rates of import, producer and consumer inflation. According to Swanepoel (2006), oil shocks slightly, positively impacted on these prices. Bellamy (2006) also attempted a similar study using a VAR framework and he finds that the gold price played a significant role in controlling the negative effects from increasing oil prices. Furthermore, Fofana et al., 2007, Fofana et al., 2008, Nkomo, 2006a, Nkomo, 2006b, and Wakeford (2006) also analysed the South African economy and they contend that rising oil prices adversely impact on the country's economic performance through various activities such as a surge in total import bill, contraction in exports, acceleration in inflation and domestic interest rates.

As crude oil prices continue to accelerate, there is a renewed concern on its impact on emerging economies. It is argued that oil price shocks widely and negatively impact across sectors and industries (Essama-Nssah et al., 2007). It is worth noting that different (macro and micro-economic) approaches and methodologies have been employed in assessing the effects of oil price shocks—Richardson (1988), Faruqee et al. (1998) and IEA (2009). This paper overlaps with the previous studies by analyzing the effects of oil price shock specifically on manufacturing production which is considered the main driver of employment in South Africa. In this paper, we extend the analysis by Elder and Serletis (2010) and previous studies in South Africa by empirically investigating the response of a real oil price shock on manufacturing production in the South African context. Further, we consider the asymmetric effects of oil price uncertainty unlike the previous South African studies.

Previous studies that use the standard VAR framework do not account for volatilities. These studies implicitly assume that history must always repeat itself. In reality however, history has been proven to include different volatilities at different time intervals as well as time varying volatility (Fama, 1965, Orhan and Köksal, 2012). The GARCH model introduced by Bollerslev, 1986a, Bollerslev, 1986b is capable of handling such problems of clustering in time series data. Therefore, this study employs an empirical model that simultaneously estimates the parameters of interest in an internally consistent fashion, which is based on a structural VAR that is modified to accommodate bivariate GARCH-in-Mean errors, as described in Engle and Kroner (1995) and Elder, 1995, Elder, 2004. The model has the desirable property of encompassing homoskedasticity as a special case, so that if the true data generating process is homoskedastic, this will be reflected in the parameter estimates (Elder, 2004). The multivariate GARCH-in-Mean (MGARCH-in-Mean) models of which the bivariate one is a special case allow the conditional variance of one or more variables in a simultaneous equation system to impact the conditional mean of one or more other variables (Elder, 2003). Such feedback effects cannot be captured by the standard homoskedastic VAR mainly used in analysing the impact of oil prices, which assumes that the conditional variance is invariant over time. This is because an impulse-response function for the usual homoskedastic VAR will estimate the dynamic response of macroeconomic variables to an oil price shock, accommodating interaction between the conditional means of the variables in the system. However, suppose oil price displays evidence of GARCH, so that an oil price shock tends to increase current and future oil price volatility. If oil price volatility, in turn, affects the level of manufacturing production which is the variable of interest in this study, then this is another channel through which an oil price shock affects manufacturing production — a channel that can be accommodated by an MGARCH-in-Mean VAR. To summarise this point, an MGARCH-in-Mean VAR would accommodate the usual channel through the conditional means of the variables in the system — but it would also accommodate the effects of the oil price shock on oil price volatility, and, in turn, the effects of oil price volatility on manufacturing production or any real economic activity whereas the standard homoskedastic VAR will accommodate the first channel only. Further, the MGARCH-in-Mean VAR is less ad hoc than single-equation reduced forms and mitigates the effects of simultaneity and generated regressors prevalent in low order dynamic models and two-step estimation methodologies, which may lead to inefficient, inconsistent and/or biassed estimates of the parameters of interest (Elder, 2004).

Section snippets

The empirical model

The empirical model we use is a bivariate monthly model in the manufacturing production growth and the real price of oil growth following the primary model developed by Elder, 1995, Elder, 2004. The model is based on a structural VAR with modifications for conditional heteroskedasticity in the parametric form of multivariate GARCH-in-Mean. The main assumption lies in that the dynamics of the structural system can be summarised by a linear function of the variable of interest plus a term related

Data and empirical results

We use monthly observations covering the period 1974:02 to 2012:12 to estimate a two-variable simultaneous equation system. The paper also employed US crude oil imported acquisition cost by refiners which we obtain from the US Energy Information Administration as a measure for oil price. Oil price is then divided by CPI US to obtain real terms values since economic theory suggests that real rather than nominal oil price should be considered particularly when making economic decisions. This

Conclusion

This paper analyses the effects of oil price uncertainty on the manufacturing production of South Africa since an economic theory suggests that uncertainty about the price of oil tends to affect negatively real economic activity. This was achieved through the modified VAR model that accommodates bivariate GARCH-in-Mean errors as described by Elder, 1995, Elder, 2004. Oil price uncertainty is captured here by the conditional standard deviation of the one-period ahead forecast error of the change

References (47)

  • M.B. Dagut

    The economic effect of the oil crisis on South Africa

    S. Afr. J. Econ.

    (1978)
  • D.A. Dickey et al.

    Distribution of the estimators for autoregressive time series with a unit root

    Econometrica

    (1981)
  • Edelstein, P., Kilian, L., 2007a. Retail energy prices and consumer expenditures, Mimeo, Department of Economics,...
  • P. Edelstein et al.

    The response of business fixed investment to changes in energy prices: a test of some hypothesis about the transmission of energy price shocks

    B.E. J. Macroecon.

    (2007)
  • J. Elder

    Macroeconomic and Financial Effects of Monetary Policy and Monetary Policy Uncertainty

    (1995)
  • J. Elder

    Another perspective on the effects of inflation volatility

    J. Money Credit Bank.

    (2004)
  • J. Elder et al.

    Oil price uncertainty

    J. Money Credit Bank.

    (2010)
  • R.F. Engle et al.

    Multivariate simultaneous generalized ARCH

    Econ. Theory

    (1995)
  • B. Essama-Nssah

    Economy-wide and distributional impacts of an oil price shock on the South African economy

    The World Bank Africa Region, Office of the Chief Economist, September 2007. WPS4354 Public

    (2007)
  • E.F. Fama

    The behavior of stock market prices

    J. Bus.

    (1965)
  • H. Faruqee et al.

    Multimod mark III: the core dynamic and steady state model

    IMF Occasional Papers 164

    (1998)
  • I. Fofana et al.

    Oil prices and the South African economy: a macromeso-micro analysis

    Financial and Fiscal Commission Report

    (2007)
  • I. Fofana et al.

    Analysing impacts of alternative policy responses to high oil prices using an energy-focused macro–micro model for South Africa

    Financial and Fiscal Commission Report

    (2008)
  • Cited by (64)

    • Oil price uncertainty, workplace misconduct, and cash holding

      2023, International Review of Financial Analysis
    View all citing articles on Scopus
    View full text