Exogenous Oil supply Shocks in OPEC and Non-OPEC Countries.

• Author: Guntner, Jochen

1. INTRODUCTION

   Recent geopolitical events, such as the 2019 drone strikes on Saudi Arabian oil-processing facilities and growing tensions between the U.S. and Iran, have refueled concerns about oil supply disruptions. (1) At the same time, there is an ongoing debate about how to identify oil supply shocks in structural vector-autoregressive (VAR) models of the global oil market (see, e.g., Baumeister and Hamilton, 2019a,b; Caldara et al., 2019; Kilian and Zhou, 2019), calling for external instruments to assess the dynamic causal effects of oil supply disruptions. Kilian (2008) proposes a measure of global oil supply shocks based on arguably exogenous production shortfalls due to geopolitical events in major oil-producing countries relative to some country-specific counterfactual.

   In this paper, we expand Kilian's (2008) original time series along two dimensions. First, we extend the sample period to include production shortfalls in OPEC member countries during 2004:10-2019:12. Second, we also consider production shortfalls in major oil-producing countries outside of OPEC such as the U.S., Canada, and Norway, for example. While this merely adds three exogenous events to the original sample period of 1973:1-2004:9, a number of nontrivial production shortfalls in OPEC as well as non-OPEC countries occurred during 2004:10-2019:12. Moreover, we construct each country-specific counterfactual based on the historical rather than the current composition of OPEC.

   Our time series displays statistically significant contemporaneous correlation with state-of-the-art estimates of oil supply shocks based on structural VAR models. At the same time, it requires only a limited number of explicit assumptions about counterfactual oil production in the countries under consideration and provides thus an alternative to and instrument for the estimates based on structural VAR models. Given that the methodology in Kilian (2008) relies on the identification of arguably exogenous political events or natural disasters, the resulting time series predominantly reflects negative oil supply shocks. However, it is exactly this kind of exogenous production shortfall that is of primary interest to market participants and policymakers alike, as it is likely to cause an unpredictable oil price increase. (2)

   Following Kilian (2008), we use monthly data on country-level crude oil production provided by the U.S. Energy Information Administration (EIA) to detect oil supply disruptions and construct the counterfactual production levels during exogenous historical events. Each counterfactual is derived from the production behavior of a benchmark group of countries that adhered to the same global macroeconomic conditions--in particular the same oil price--and thus similar economic incentives, without being subject to the exogenous event in question or any other exogenous event affecting oil production at the time. Accordingly, the composition of the benchmark group varies both by country and episode, and the counterfactual is constructed by extrapolating the country's production level in the month prior to an event by the growth rate of production in the respective benchmark group.

   In his identification of exogenous oil supply disruptions, Kilian (2008) focuses on OPEC countries, while non-OPEC production data are only used for constructing a counterfactual during the Arab Oil Embargo of 1973/74. Bastianin and Manera (2018) extend this series of OPEC supply shocks until December 2013 and include thus the First Libyan Civil War. (3) Similar to Caldara et al. (2019), we furthermore consider exogenous events in oil-producing countries outside of OPEC such as the Norwegian oil strike of 1986 and the Ecuador earthquakes of 1987, for example. (4)

   The rest of this paper is structured as follows. Section 2 describes the crude oil production data. Sections 3 and 4 discuss the construction of counterfactuals during exogenous historical events in OPEC and non-OPEC countries, respectively. Section 5 presents our extended time series of exogenous oil supply shocks. Section 6 compares our series with alternative estimates of oil supply shocks based on structural VAR models. Section 7 concludes.

2. DATA

   All crude oil production data are from Tables 11.1a and 11.1b in the EIA's Monthly Energy Review (MER). (5) It is important to note that, in any given release of the MER, Total OPEC production corresponds to the sum of production in all countries that were an OPEC member in the most recent month featured. As a result, Total OPEC production as of today includes the output of current members that joined only recently (e.g. Equatorial Guinea), whereas it excludes the output offormer members that left during our sample period (e.g. Qatar). To address this inconsistency, we start by constructing a time series of OPEC production that reflects its historical composition at each point in time, as detailed in Table A.1 in the online appendix. The corresponding real-time series of Total non-OPEC production is then obtained by subtracting Total OPEC from World crude oil production. The benefit of this approach is that future changes in the composition of OPEC, such as Ecuador's suspension of membership in January 2020, have no effect on existing realizations, when we update our time series of exogenous oil supply shocks. (6)

3. OPEC COUNTERFACTUALS

   In this section, we replicate the counterfactuals for the exogenous oil production shortfalls in OPEC countries from Kilian (2008) for our extended sample period, 1973:1-2019:12, and consider three additional events in Ecuador, Libya, and the United Arab Emirates (UAE), respectively. The counterfactual for an event occurring in country i and period t extrapolates the level of production in t -1 by the growth rate of production in all OPEC countries that were not affected by the same or any other exogenous event at the time, i.e.

   [Please download the PDF to view the formula] (1)

   As in the original paper, these counterfactuals form the basis for our series of exogenous oil supply shocks in Section 5. The detailed composition of country-specific benchmark groups for OPEC events is summarized in Table A.2 in the online appendix.

   3.1 Arab Oil Embargo

   The bottom panel of Figure 1 in Kilian (2008) plots the sum of exogenous production shortfalls in Algeria, Iraq, Kuwait, Libya, Qatar, Saudi Arabia, and the UAE during the Arab Oil Embargo for 1973:11-1974:2. The corresponding counterfactual extrapolates these countries' total production in October 1973 by the growth rate of total non-OPEC production.

   Following the methodology in the original paper, we compute the total production shortfall in Algeria, Iraq, Kuwait, Libya, Qatar, Saudi Arabia, and the UAE associated with the Arab-Israeli War and the Arab Oil Embargo in 1973-1974. Since the resulting time series is qualitatively identical to the one in Kilian (2008), we do not show it separately in Figure 1, while it enters our final time series of exogenous oil supply shocks in Figure 3 and Figure A.1 in the online appendix.

   3.2 Counterfactual for Iran

   Kilian (2008) identifies the Iranian Revolution as an exogenous event and dates the beginning of adverse effects on the country's crude oil production to October 1978, when striking oil workers brought production to a near-halt. He further argues that, while the Shah abdicated and the Khomeini regime took over in January 1979, the effects on crude oil production continued at least until the end of his sample period in September 2004.

   Following this line of argument, we presume that Iranian production has not recovered until the end of our sample period in December 2019, as disputes over the country's nuclear program led the U.S. to impose sanctions on the Central Bank of Iran in December 2011, aggravated by the EU's import ban on Iranian oil and restrictions on transportation insurance starting in July 2012. (7) Due to the lack of progress in nuclear talks, the EU imposed further sanctions, and oil production dropped from 4,000 tbpd in December 2011 to 2,800 tbpd in October 2012.

   After an extended period of heightened political uncertainty and despite first signs of easing tensions in 2014, the Joint Comprehensive Plan of Action (JCPOA) was only ratified on July 14 of 2015. (8) The JCPOA subjected Iran's nuclear program to international control in exchange for a substantial easing of economic sanctions, in particular enabling the country to export crude oil without restrictions and accessing nearly $60 billion of foreign exchange reserves (Katzman, 2016, p. 59). On May 8 of 2018, the U.S. government announced its intentions to withdraw from the JCPOA and reintroduce sanctions unilaterally, inducing large companies to leave Iran (Katzman, 2019). Panel (a) of Figure 1 illustrates that the installment of sanctions in 2012 as well as in 2016 had substantial negative effects on Iran, reducing crude oil production levels by 1,000 tbpd within a few months on both occasions, while the lifting of nuclear-related sanctions after ratification of the JCPOA is characterized by a recovery of production of similar size.

   In forming a benchmark group for Iran, Kilian (2008) considers oil-producing countries with similar economic incentives that were not affected by the Iranian Revolution. Hence, the Iranian counterfactual is based on the growth rate of production in all OPEC countries but Iran, Iraq, and Saudi Arabia. No adjustment of this benchmark group is necessary at the start of the Iran-Iraq War in September 1980, while we account for the 1987 Ecuador earthquakes, the invasion of Kuwait by Iraq in August 1990, and the Venezuela oil lockout in December 2002 by temporarily excluding these countries from Iran's benchmark. As we extend the sample period to December 2019, we furthermore account for the Libyan civil wars starting in February 2011. Panel (a) of Figure 1 plots the actual and counterfactual production levels for Iran.

   3.3...