The effect of oil price shocks on asset markets: Evidence from oil inventory news

• Author: Jianjian Jin,Reinhard Ellwanger,Ron Alquist
• Date: 01 August 2020
• DOI: http://doi.org/10.1002/fut.22096
• Published date: 01 August 2020

J Futures Markets. 2020;40:1212–1230.wileyonlinelibrary.com/journal/fut1212

|

© 2020 Wiley Periodicals LLC

Received: 31 December 2019

|

Accepted: 31 December 2019

DOI: 10.1002/fut.22096

RESEARCH ARTICLE

The effect of oil price shocks on asset markets: Evidence

from oil inventory news

Ron Alquist

1

|Reinhard Ellwanger

2

|Jianjian Jin

3

1

Global Alternative Premia, AQR Capital

Management LLC, Greenwich,

Connecticut

2

International Economic Analysis, Bank of

Canada, Ottawa, Ontario, Canada

3

Investment Strategy and Risk, British

Columbia Investment Management

Corporation, Victoria, British Columbia,

Canada

Correspondence

Reinhard Ellwanger, International

Economic Analysis, Bank of Canada, 234

Wellington St W, Ottawa, ON K1A 0G9,

Canada.

Email: rellwanger@bankofcanada.ca

Abstract

We quantify the reaction of U.S. equity, bond futures, and exchange rate re-

turns to oil price shocks driven by oil inventory news. Across most sectors,

equity prices decrease in response to higher oil prices before the 2007/2008

crisis but increase after it. Positive oil price shocks cause a depreciation of the

U.S. dollar against a broad range of currencies but have only a modest effect on

bond futures returns. The evidence suggests that changes in risk premia help to

explain the time‐varying effect of oil price shocks on U.S. equity returns.

KEYWORDS

exchange rates, interest rates, news, oil price shocks, stock market returns

JEL CLASSIFICATION

E44; G14; G15; Q41; Q43

1|INTRODUCTION

Oil price fluctuations have important implications for the terms of trade, investment, output, and other macroeconomic

aggregates of both oil‐importing and oil‐exporting economies. Yet, even before oil price shocks are fully transmitted to the

real economy, the prices of financial assets adjust to reflect market expectations about the response of macroeconomic

fundamentals to such shocks. Recent empirical research has related oil price fluctuations to variation in equity market

returns (Gao, Hitzemann, Shaliastovich, & Xu, 2017;Kilian&Park,2009;Ready,2017), exchange rates (Akram, 2004;

Chen, Rogoff, & Rossi, 2010; Ferraro, Rogoff, & Rossi, 2015), and interest rates (Datta, Johannsen, Kwon, & Vigfusson, 2018;

Kilian & Zhou, 2019). However, because oil prices and asset prices are jointly determined with other macroeconomic

variables in general equilibrium (Akram, 2009;Chenetal.,2010;Frankel,2008;Hitzemann,2016; Kilian, 2009)identifying

the effects of oil price fluctuations on asset prices remains a significant challenge.

This paper uses weekly news about U.S. oil inventories to investigate and quantify the effect of oil price shocks on the

returns of different financial assets and the expectations about macroeconomic outcomes that the changes in returns reflect.

Changes in oil inventories are a fundamental feature of oil markets and play a central role in the intertemporal relationship

between current and future supply and demand conditions (Alquist, Bauer, & Diez de los Rios, 2014; Alquist & Kilian, 2010;

Kilian & Lee, 2014;Kilian&Murphy,2014). As such, news of higher‐than‐expected (lower‐than‐expected) U.S. oil in-

ventories leads to systematic decreases (increases) in oil prices in the minutes following the announcement (see, e.g., Halova,

Kurov, & Kucher, 2014). As showninFigure1, the response of oil prices to news about U.S. crude oil inventories is strong

The views expressed in this paper are those of the authors; no responsibility for them should be attributed to AQR Capital Management, the Bank of

Canada, or the British Columbia Investment Management Corporation. AQR Capital Management is a global investment management firm, which

may or may not apply similar investment techniques or methods of analysis as described herein.

and systematic, in a way that is consistent with economic theory. We use this variation in oil prices and a comprehensive,

high‐frequency data set of financial variables, including stocks, bonds, and exchange rates, to study how information about

oil market fundamentals is transmitted to asset prices and the broader economy. This approach differs from existing studies,

which often rely on low‐frequency exclusion restrictions in the context of structural models (Kilian & Park, 2009; Kilian &

Zhou, 2019;Ready,2017) or on indirect evidence (Datta et al., 2018; Gao et al., 2017) to establish causality between oil prices

and financial variables.

Our empirical results support existing evidence for a structural break in the relationship between oil and asset

returns around September 2008 (Aït‐Sahalia & Xiu, 2016; Datta et al., 2018; Foroni, Guérin, & Marcellino, 2017;

Lombardi & Ravazzolo, 2016). We document that before the 2007/2008 crisis, higher oil prices are associated with lower

equity market returns, while post‐crisis, higher oil prices are associated with higher equity market returns. This pattern

holds for aggregate equity market returns and is pervasive across different sectors, including those with limited direct

exposure to energy prices, such as health care. The estimates for bond returns follow the reverse pattern. Bond futures

returns tend to increase with higher oil prices before the crisis and to decrease with higher oil prices after it. While these

results suggest that nominal interest rates became increasingly aligned with oil price fluctuations, the estimates are

economically small and indicate that the effects of oil price changes on nominal interest rates are limited. Finally, we

show that higher oil prices are associated with a depreciation of the U.S. dollar against a broad range of currencies,

including those of major oil importers (such as the euro area).

Our empirical approach is based on instrumental variables (IV) estimation to capture the variation in oil prices

related to the predetermined oil inventory news. We interpret the news as reflecting mostly oil‐market‐specific in-

formation, which induces different comovement among oil and asset returns than more general macroeconomic news.

Accordingly, the IV estimates for the equity market returns indicate a consistently lower correlation between oil returns

and the equity market than ordinary least squares (OLS) regressions. This approach is consistent with the idea that

general macroeconomic news induces a positive correlation between these variables. A similar pattern is obtained for

exchange rates, where the depreciation of the U.S. dollar against other currencies is more pronounced for OLS

regressions. These results highlight that oil‐market‐specific news tends to have very different effects on the relationship

between oil prices and financial variables than more general macroeconomic news.

1

News about inventories reflects the availability of oil for future consumption and the willingness of market parti-

cipants to carry physical oil into the future. Thus, the oil price shocks associated with this news are different from the

structural oil supply and demand shocks identified in widely used structural vector autoregressive (SVAR) models. Still,

our results confirm several findings from this strand of literature, such as changes in the relationship between oil‐

market‐specific demand shocks and U.S. stock market returns (Datta et al., 2018; Foroni et al., 2017) and the

-10 -5 0 5 10 15 20

Minutes relative to announcement

-60

-40

-20

0

20

40

Cumulative return, bps

Lower than expected

As expected

Higher than expected

FIGURE 1 Response of oil futures returns to crude oil inventory news. Crude oil inventory news is computed as the actual change in

U.S. crude oil inventories as reported in the Weekly Petroleum Status Report minus the expected change according to the Bloomberg survey.

Lower than expected stands for the average cumulative return of crude oil futures for the lowest 33% of the news, As expected stands for the

average cumulative return of crude oil futures for the medium 33% of the news, and Higher than expected stands for the average cumulative

return of crude oil futures for largest 33% news. The sample period is 2003M10–2017M10. bps, basis points [Color figure can be viewed at

wileyonlinelibrary.com]

1

This interpretation is also consistent with Kilian and Vega (2011), who document that between 1983 and 2008, daily oil prices did not react to U.S. macroeconomic news. Instead, oil prices did react

strongly to oil inventory news, which highlights the oil‐market‐specific nature of such news.

ALQUIST ET AL.

|

1213