Tail Wags Dog: Intraday Price Discovery in VIX Markets

• Author: Robert E. Whaley,Nicolas P.B. Bollen,Michael J. O'Neill
• Published date: 01 May 2017
• Date: 01 May 2017
• DOI: http://doi.org/10.1002/fut.21805

Tail Wags Dog: Intraday Price Discovery

in VIX Markets

Nicolas P.B. Bollen, Michael J. O’Neill , and Robert E. Whaley*

Beginning with VIX futures in 2004, followed by VIX options in 2006 and VIX ETPs in

2009, the daily open interest in volatility contracts is now in the tens of billions of dollars.

Given this growth, it is important to develop a better understanding of price discovery and the

supply/demand dynamics in each market. Some of the price relations are linked by arbitrage.

Others are not. In particular, the relation between the VIX cash index and the VIX futures is

not arbitraged, and we show that, where once VIX changes led VIX futures price changes, the

VIX futures now leads. © 2016 Wiley Periodicals, Inc. Jrl Fut Mark 37:431–451, 2017

1. INTRODUCTION

More than twenty years ago, Whaley (1993) argued that the marketplace would benefit from

the introduction of derivative contracts written on the newly created CBOE Market

Volatility Index (VIX). The primary illustration centered on a stock index option market

maker who, in the process of fulfilling his duty to transact, at times accumulates a sizable

option position and a corresponding exposure to volatility risk. Whaley showed how volatility

derivatives could be used to hedge this risk, driving down the market maker’s inventory

holding costs and ultimately transaction costs for index option end users.

Over the next ten years, the VIX grew from an esotericmeasure of stock market risk to a

widely disseminatedand popularly followed gauge of investorfear regarding a downward jump

in the stock market. The interest in the VIX motivated the CBOE to introduce VIX futures

contracts in March 2004. Trading activity was meager until the launch of VIX options in

February 2006. VIX calls,in particular, were an immediate success, presumablybecause they

are a low cost mechanism for providing tail-risk insurance. The presence of an active options

market, as it turns out, spurredtrading activity in the VIX futures market. This is naturalin the

sense that VIX optionmarket makers, who are generally short calls fromsupplying insurance,

need to vega-hedge their inventories and the VIX futures are a convenient, low-cost

Nicolas P.B. Bollen is Frank K. Houston Professor of Finance in Vanderbilt University, Nashville, Tennessee.

Michael J. O’Neill is Adjunct Associate Professor in Bond University, Queensland, Australia. Robert E.

Whaley is Valere Blair Potter Professor of Finance and Director of the Financial Markets Research Center in

Vanderbilt University, Nashville, Tennessee. Seminar participants at the Auckland University of Technology,

the College of William and Mary, Southern Methodist University, and the University of Texas at Dallas

provided helpful suggestions, as did participants at Vanderbilt’s May 2015 Options and Volatility Conference.

Discussions with Henry Chien, Bernard Dumas, Gary Gastineau, Joanne Hill, Neil Ramsey, Jacob Sagi, Tom

Smith, Bill Speth, Hans Stoll, George Tauchen and Damon Walvoord and the research assistance of Daejin

Kim are gratefully acknowledged. This research was supported by the Financial Markets Research Center at

Vanderbilt University.

*Correspondence author, The Owen Graduate School of Management, Vanderbilt University, 401 21st Avenue

South, Nashville, TN 37203. Tel: 615-343-7747, Fax: 615-376-8214, e-mail: whaley@vanderbilt.edu

Received December 2015; Accepted July 2016

The Journal of Futures Markets, Vol. 37, No. 5, 431–451 (2017)

© 2016 Wiley Periodicals, Inc.

Published online 19 August 2016 in Wiley Online Library (wileyonlinelibrary.com).

DOI: 10.1002/fut.21805

mechanism for doing so.The next meaningful intervention in VIX tradingactivity occurred in

January 2009 with the launch of VIX exchange traded products (ETPs). The introduction of

VIX ETPs was motivated, at least in part, by un-satiated investor demand for volatility

exposure. Investors such as retail customers, too small or unsophisticated to trade in the

futures market, and institutions such as pension funds and endowments, barred from trading

in futures market, were able to trade market volatility for the first time.

Given the proliferation of VIX-related instruments, and the access they provide to

new classes of investors, a natural question to ask is whether volatility’s“price discovery”

has been affected. The purpose of this paper is to determine where information about

volatility first appears using intraday price movements in four related markets: S&P 500 index

options, VIX futures, VIX options, and VIX ETPs.

Prior literature studies the relation between the VIX, computed from S&P 500 index

option prices, and VIX futures prices. Zhang and Zhu (2006) develop a futures pricing

function based on Heston’s (1993) stochastic volatility model. They find that the large

deviations between model values and market prices are dramatically reduced when model

parameters are re-estimated annually, suggesting one or more model constants actually time

varies. Zhang, Shu, and Brenner (2010) present a similar model but specify a separate

diffusion for the long-term mean of the volatility process to accommodate the time variation

documented by Zhang and Zhu. Shu and Zhang (2012), who also study the lead/lag relation

between the cash VIX and VIX futures prices, find the two time-series are largely

contemporaneously related. Given the speed of today’s markets, however, daily changes

many not provide the granularity necessary to determine where traders trade first.

Our paper makes two main contributions. First, we use a more up-to-date sample, with

data through April 2013, and incorporate a more comprehensive set of VIX products,

including the most actively traded VIX ETP. These data provide additional insights regarding

the evolution of trading in the related markets. Second, we use intraday trade and quote data

to study lead/lag relations between the different VIX-related time-series. Frijns, Tourani-Rad,

and Webb (2016), in an important departure from prior work, also use intraday data in their

study of lead/lag relations between the VIX and VIX futures. Our study differs from theirs in

that we use all trade prices, as opposed to regularly spaced bid-ask midpoints, to identify the

transmission of information. In principle, new information enters the marketplace when

trades occur. Consequently, we also use different econometric techniques that are designed

explicitly to accommodate high frequency microstructure. We find that the lead/lag relations

between the prices of VIX futures and VIX options, and between the VIX ETPs and the VIX

futures, are short-lived as they are largely governed by arbitrage price relations. In contrast,

the deviations between VIX futures and the VIX cash index cannot be arbitraged due to the

high costs of trading the index options that comprise the cash VIX. One of the most

interesting results in the paper stems from this market friction: while the VIX futures price

lagged the VIX cash index in the first few years after it was launched, the VIX futures now

leads the cash index. Put differently, the VIX futures has become the “go-to”market for

hedging volatility risk in a timely manner.

The paper is organized into five sections. The first section describes the daily and

intraday data used in our analyses. Section II uses the daily data to sketch the landscape of the

volatility market during our sample period. Section III reviews the methodology we use for

assessing the intraday lead/lag relation between VIX and VIX products. The methodology is

relatively new in the sense that it does not involve sampling prices at fixed intervals during the

day. Instead, all trade prices changes are used whenever the time period between trades of

two prices series has overlap. Such a methodology is better able to determine the true nature

of the lead/lag relation. The main results of the paper are presented in Section IV. Section V

contains a brief summary of the study and our conclusions.

432 Bollen, O’Neill, and Whaley