Forecasting Volatility Of The U.S. Oil Market – ValueWalk Premium
forecasting volatility

Forecasting Volatility Of The U.S. Oil Market

Forecasting Volatility Of The U.S. Oil Market

Erik Haugom

Lillehammer University College; Norwegian University of Science and Technology (NTNU)

Henrik Langeland

Norwegian University of Science and Technology (NTNU)

Peter Molnár

Norwegian University of Science and Technology (NTNU)

Sjur Westgaard

Norwegian University of Science and Technology (NTNU) – Department of Industrial Economics and Technology

January 29, 2014


We examine the information content of the CBOE Crude Oil Volatility Index (OVX) when forecasting realized volatility in the WTI futures market. Additionally, we study whether other market variables, such as volume, open interest, daily returns, bid-ask spread and the slope of the futures curve, contains predictive power beyond what is embedded in the implied volatility. In out-of-sample forecasting we find that econometric models based on realized volatility can be improved by including implied volatility and other variables. Our results show that including implied volatility significantly improves daily and weekly volatility forecasts, while including other market variables significantly improves daily, weekly and monthly volatility forecasts.

Forecasting Volatility Of The U.S. Oil Market – Introduction

Accurate volatility forecasts are crucial for portfolio optimization, options and derivatives pricing, value-at-risk modeling, and hedging. Forecasting volatility has traditionally been done using the generalized autoregressive conditional heteroscedasticity (GARCH) approach of Bollerslev (1986) and Engle (1982), also in energy commodity markets (see e.g. Marzo and Zagaglia (2010) and Wei et al. (2010)).

A breakthrough in volatility measuring was provided when Andersen and Bollerslev (1998) introduced realized volatility as the sum of squared intra-daily returns. This made volatility almost an observable variable which can be modeled straightforwardly with standard time-series techniques.

It has long been recognized that there are other sources of information about future volatility than realized volatility. A natural candidate is the market’s expectation of future volatility, commonly referred to as implied volatility (IV). Some previous studies (e.g. Lamoureux and Lastrapes (1993); Jorion (1995); Agnolucci (2009)) argue that forecasts obtained from implied volatility are both biased and inefficient.

Evidence that IV improves volatility forecasts has also been presented (e.g. Day and Lewis (1993); Szakmary et al. (2003); Doran and Ronn (2005); Agnolucci (2009)). According to Jorion (1995), a failure to unearth IV’s predictive power can only be interpreted in two ways; inefficient information processing in options markets or misleading test procedures. In highly liquid and transparent markets such as the WTI futures market the former is unlikely. Left is the latter, and in particular the discussion about the bias of the Black-Scholes (BS) formula (see e.g. Doran and Ronn (2005)). A way to avoid this possible problem (and several others) is to use a volatility index which is based on the market price of variance. Such an index was introduced for the WTI futures market in 2008 and is one of the main units of analysis in this paper. Volatility has also been linked to several other market variables. For instance, the relationship between volume and volatility is widely documented (e.g. Clark (1973) and Gallant et al. (1992)). In addition to possibly improve volatility forecasts, including additional variables in the analysis can increase our understanding of the market.

forecasting volatility

Even though realized and implied volatility in equity markets has been extensively studied (see e.g. Bollerslev et al. (2013) and references within), much less work has been done in this field for commodity markets. This is particularly suprising for the oil market,considering the market’s economic importance (Sadorsky, 2006). Wang et al. (2008) studied the realized correlation between oil and  gas markets and found the use of RV in energy markets to be highly appropriate, especially in areas such as volatility forecasting. Martens and Zein (2004) compared forecasts obtained from a long-memory model of RV with options-implied volatility for the WTI futures market. They found that both RV and IV contain useful information in volatility forecasting. Little work has been done regarding the WTI IV index, due to its recent inception. An exception is Padungsaksawasdi and Daigler (2013) who studied the return-IV relation, and concluded that IV increases with negative returns.

In this paper we examine the role of both volatility implied from the OVX and observable market variables when forecasting volatility for the WTI futures market. We apply the simple HeterogenousAutoregRessive (HAR) model of Corsi (2009) on realized volatility itself. Additionally, two fundamentally different types of variables are used in the model; the forward looking IV index and other exogenous market variables including volume, open interest, daily returns and the slope of the futures curve. The main findings can be summarized as follows. First, we find that including information from the OVX significantly improves the day-ahead and weekahead volatility forecasts. Second, the exogenous market variables improve volatility forecasts for daily, weekly and monthly horizons. Of the additional explanatory market variables, the daily returns is the most important factor to improve volatility forecasts.

forecasting volatility

forecasting volatility

See full PDF below.


Saved Articles

The top investors are reading ValuewalkPremium.

Click here to learn why