Comparing the Risk Spillover from Oil and Gas to Investment Grade and High-yield Bonds through Optimal Copulas.

• Author: Rahman, Md Lutfur

1. INTRODUCTION

   This paper examines the tail dependence and risk spillover from the oil and gas markets to the investment grade (IG) and high-yield (HY) bond markets. The motivation of this paper is three-fold. First, a large literature examines the impact of energy price shocks on stock markets (Driesprong et al., 2008; Ready, 2018). However, relatively little attention has been provided on examining the relationship between energy and bond markets (Kang et al., 2014; Gormus et al. 2018). While the impact of recent energy market volatility on aggregate stock markets (Mensi et al., 2017), Islamic equity markets (Shahzad et al., 2018), exchange rates (Ji et al., 2018), and commodity markets (Ji et al., 2018) has been examined, energy market's shock spillover to both HI and IG bond markets, particularly during extreme market conditions, is yet to be explored.

   Second, the global bond market is twice as big as the global stock market. The US bond market is worth of about US$40 trillion while the US stock market value is just under US$20 trillion (Bloomberg, 2016). Additionally, the global demand for bonds has been increasing due to an increase in investors' degree of risk aversion and shifting of investment from equity market to relatively safer bond market after the global financial crisis (GFC). A decline in interest rates after the GFC resulting from the quantitative easing may have attracted investors to a particular segment of the bond market (such as HY bond market). Given that the bond market constitutes a large portion of the overall financial market, it is worthwhile to understand the impact of energy price shock, a global risk factor, on the bond market.

   Third, with the financialization of commodity markets and cross-country integration of financial markets in the recent years, new information tends to diffuse quickly across markets. Further, the GFC in 2008 characterizes that (i) the entire financial market system can be vulnerable to external economic shocks, and (ii) cross-market linkage can be stronger during crisis periods due to financial contagion across markets (Liu et al., 2016). To this end, modelling the dynamic dependencies across markets is increasingly important.

   The link between bond and energy markets can arise from different channels. Since same underlying corporate assets represent claims against corporate bonds and stocks, energy price shocks affecting value of these corporate assets may have an impact on both stock and bond prices. This proposition is supported by a large literature showing a significant relationship between bond and stock markets (Hong et al., 2012; Tolikas, 2018). Additionally, energy price shocks may affect firms' future earnings, investors' required rate of return and eventually bond prices. It is argued that some bonds (particularly HY bond) behave like stocks, and bond prices show a trend in tandem to stock prices (Hong et al., 2012; Downing et al., 2009). Since there is significant evidence of energy price shock spillover to equity market, same spillover can be observed in bond market due to the latter's co-movement with stock market. Nonetheless, bond and stock markets may exhibit an asymmetric response to common factors since (i) bond market is dominated by sophisticated institutional investors while stock market is largely participated by individual investors (Tolikas, 2018), (ii) bond holders have priority over equity holders in a bankruptcy que, and (iii) bond holders' upside potential is limited.

   This paper makes several contributions to the literature. First, as indicated earlier, a very thin literature provides evidence of risk transmission between energy and bond markets. While Kang et al. (2014) focus on the aggregate US bond index and Gormus et al. (2018) concentrate on the HY bond market, we focus on the risk spillover from the oil and gas markets to the HY and IG bond markets. Theoretically, energy price shocks may have a differential impact on the HY and IG bond markets. Due to the cyclical nature of energy industry and banks' reluctance to provide long-term debt particularly after the GFC, energy firms typically use HY bonds to finance most of their expansion projects. (1) Since the HY bond portfolio includes a large amount of energy company debt (Gormus et al., 2018), energy market uncertainty is likely to have a negative impact on forecasted operating cashflows of energy firms and their HY bond prices. On the other hand, IG bonds are typically issued by large-cap and blue-chip companies which are less vulnerable to external shocks. Hence, IG bond prices may be less sensitive to energy price shocks compared to HY bond prices. (2)

   Second, Kang et al. (2014) and Gormus et al. (2018) respectively use variance decomposition approach and Lagrange multiplier (LM) based volatility transmission approach which are unable to capture dynamic tail dependence and systemic risk spillover across return distributions under different market conditions. In this paper, we use the time-varying optimal copula (TVOC) model. While the commonly-used static copulas and time-varying copula (TVC) parameters are unable to identify dynamic changes in the dependence structure between two time-series, the TVOC approach properly models asymmetric dependence across markets and identifies direction, intensity and nature of the dependence structure.

   Third, we quantify the upside and downside risk spillover between the energy and bond markets using value-at-risk (VaR), conditional VaR (CoVaR), and delta CoVaR ([DELTA] CoVaR) measures. While VaR is a widely used measure of downside risk of an investment (Girardi & Ergun, 2013), CoVaR reflects risk in investing in one market conditional on extreme movement in another market. [DELTA]CoVaR captures asymmetric upside and downside risk spillover across the markets. These methodologies have not been previously used to examine the dependence structure and risk spillovers between the bond and energy markets.

   On the whole, this study investigates the upside and downside risk spillovers between the energy and bond markets. The risk measures mentioned in the previous paragraph, firstly, represent a good tool for international investors to assess the extreme investment losses, and secondly, are relevant for safety-first investors who want to reduce the likelihood of extreme losses that may drive them out of business (Reboredo et al., 2016). Since financial and commodity markets tend to integrate more during crisis periods, exploring the dependence and risk spillover between the energy and bond markets particularly in the extreme conditions can be interesting. As indicated earlier, the HY bond portfolios typically hold a large amount of energy company debt. Therefore, HY bond prices are exposed to energy price shocks. However, any evidence relating to energy prices' impact on both the HY and IG bond markets can have significant implications for designing corporate debt policy, devising investors' portfolio decisions, and formulating bond and energy market stabilizing policies.

   We report several key findings. First, the degree and direction of dependence between the bond and energy markets is time-varying. For example, the HY bond returns exhibit a negative dependence on the oil returns from 2010 until 2015, however the direction turns out to be positive afterwards. Second, while the tail risk in the HY bond market is higher compared to that in the IG bond market during the oil-crash period (2014-2017), the IG bond market exhibits higher downside risk than HY bond market in the whole sample period. Moreover, the bond returns are typically more sensitive to risk shocks in the oil market compared to the gas market. Third, we find a co-movement in both the negative and positive tails for all the bond and energy pairs implying that the energy market is not a good hedge against the tail risk in the bond markets. However, an asymmetric upside and downside risk spillover is found between the markets.

   The rest of the paper proceeds as follows: section 2 provides a brief review of relevant literature; methodological aspects are described in section 3, data description and descriptive statistics are provided in section 4; section 5 presents empirical results and analysis; finally, a summary of the paper is provided in section 6.

2. LITERATURE REVIEW

   Since the seminal paper of Hamilton (1983), a large literature examines the relationship between oil prices and economic growth and the channels to which this relationship arises. Although Hamilton (1983) shows a significant negative relationship between oil prices and economic growth, Hamilton (2003) shows a non-linear relationship between them.

   Motivated by the conventional wisdom that shocks to energy prices particularly oil prices have a spillover effect on entire economy, researchers examine oil price changes' impact on financial markets and provide economic arguments for their findings. For instance, Driesprong et al. (2008) argue that oil shocks can lead to higher economic risk, higher required rate of returns and subsequent lower prices in stock market. Ready (2018) claims that a demand-driven oil price rise results in higher revenue for oil producing firms which ultimately leads to positive returns for their stocks. Kilian (2008) and Baumeister and Kilian (2016) indicate that energy price shocks can be transmitted either through supply (or cost) channel or demand (or consumption) channel. For instance, an increase in oil price increases cost of producing goods and services, and an inability to adjust output prices results in a reduction in firms' earnings. However, this proposition only holds for industries that use oil as a major factor of production such as transportation, chemical, plastic etc. The demand channel is dominant when majority of energy is consumed by final consumers (for example, in the USA). Oil price can also affect financial market prices through affecting consumer...