Exchange Traded Funds and Stock Market Volatility

Published date01 December 2017
AuthorLiao Xu,Xiangkang Yin
Date01 December 2017
Exchange Traded Funds and Stock
Market Volatility*
International Institute for Financial Studies, Jiangxi University of Finance and
Economics, Nanchang, China and
Department of Economics and Finance, La Trobe University, Bundoora, Victoria,
This study investigates the relationship between the volatility of stock market
indexes and the trading volumes of their exchange traded funds (ETFs). Using
both ordinary least squares and generalized autoregressive conditional
heteroskedasticity approaches, we demonstrate that the contemporaneous
trading volume of S&P 500 ETFs is a key determinant of S&P 500 volatility at
both monthly and daily frequencies. Vector autoregressive estimation on the
other hand suggests a two-way Granger causality between S&P 500 volatility
and the trading of S&P 500 ETFs. A replication analysis of other market
indexes and the corresponding ETFs tracking these indexes conrms that
these ndings are robust.
JEL Codes: G12; G14
Much attention has been paid by academics, practitioners, and regulators over
the decades to the volatility of price and/or return in nancial markets. Research
on the issues associated to exchange traded funds (ETFs) has also grown rapidly
since the birth of the rst ETF in January 1993. It seems an intriguing study to
build a bridge across these two elds. The aim of this paper is to unveil the
relationship between the volatility of a market index and the trading volume of
the ETFs that track the index. This research not only is highly valuable to
academic analysts and market practitioners but also has signicant regulatory
implications because they are concerned about the role of ETFs in increasing
the volatility of their associated markets.
It is well known that trading volume is a substantial contributor to the
volatility of stock returns.
However, less is known about how the trading of ETFs
* The authors would like to thank Jae Kim and the seminar participants at La Trobe University for
their constructive comments.
1 See Ben-David et al. (2014) for a summary of regulatorsconcerns about ETFsrole in increas-
ing market volatility.
2 For an excellent survey,see Karpoff (1987).
© 2017 International Review of Finance Ltd. 2017
International Review of Finance, 2017
DOI: 10.1111/ir.12121
International Review of Finance, 17:4, 2017: pp. 525–560
DOI: 10.1111/irfi .12121
© 2017 International Review of Finance Ltd. 2017
as nancial derivatives affects the volatility of the market index they are tracking.
ETFs are related to their underlying index and the constituent securities of the
index primarily through two channels. First, ETF shares are created by the ETF
issuer who transfers its shares to its authorized participants (APs) in exchange
for the equivalent value of a portfolio of index constituents. Similarly, ETF shares
are redeemed when the ETF issuer trades a portfolio of the constituent securities
of the index with an AP in exchange for the equivalent value of ETF shares. Thus,
as demand and supply pressure in the ETF market induces the APs to create or
redeem ETF shares, it triggers them to sell and purchase index constituents and
in turn uctuates the index. Second, to keep an ETF closely tracking its target
index, the ETF sponsor announces the net asset value (NAV) under its manage-
ment every 15 s on all trading days. Therefore, if the price of an ETFs share
deviates from its NAV, arbitrageurs can reap prots from trading the ETF shares.
However, to hedge the risk in ETF trading, arbitrageurs must at the same time
trade in the underlying securities markets in the opposite direction to cover their
exposure in the ETF market. They hold their positions until the price of the ETF
share and its NAV converge, and then they close down the positions to realize
arbitrage prots. Thus, we argue that the trading volume of ETFs that track a
market index should be closely correlated with the price movement and
volatility of the tracked index. We study the S&P 500 index and its ETFs in detail
because this index covers the stocks with a market capitalization of 75% US
equities and the indexs ETFs have the longest history of all ETFs. We then extend
our analysis to other major stock indexes and their associated ETFs.
The study contributes to the literature on ETFs and market volatility in the
following ways. First, to explicitly relate ETF trading to the volatility of the under-
lying index, we quantitatively examine the effect of the trading volume of S&P
500 ETFs at both monthly and daily frequencies. When an ETF issuer announces
its NAV every 15 s, it is likely to provide an arbitrage opportunity if the ETF price
deviates considerably from the NAV. These high-frequency intraday activities
undertaken by arbitrageurs can be better captured by S&P 500 volatility at a daily
On the other hand, APs may trade less frequently in the constituent
markets to rebalance their portfolios, as the APscreation and redemption of the
ETF shares are usually (but not necessarily) through trading with the ETF issuer
near the close of a trading day or afterhours trading. Their less frequent trading
activities are probably more relevant to S&P 500 volatility at a monthly
Moreover, high-frequency intraday return data are only available
from 1996, while low-frequency data cover a longer period. Therefore, we
conduct our analysis using both monthly and daily time series. It is found that
the contemporaneous ETF trading volume in both absolute and relative terms
3 The daily time series of realized variance used in this paper are estimated on the basis of the 1-
min intraday returns of the index. We also estimated the realized variance using 15-s intraday
returns and have found no qualitative difference.
4 The monthly data of realized variance used in this paper are estimated on the basis of the daily
returns of S&P 500 index.
International Review of Finance
© 2017 International Review of Finance Ltd. 20172
International Review of Finance
526 © 2017 International Review of Finance Ltd. 2017
has an economically and statistically signicant effect on the volatility of index
returns. Moreover, the trading volume of ETFs that track S&P 500 index has
stronger power in explaining S&P 500 volatility relative to the control variables.
This thereafter suggests that ETF trading volume is a critical determinant of the
indexs volatility, in addition to the volatility determinants that have been found
in the existing literature.
Second, most existing studies analyzing the relationship between trading
activities and return volatility focus on the concurrent effect in the markets of
individual securities. This paper not only extends the analysis to the interaction
across two related markets but also examines the effects of lagged ETF trading and
S&P 500 volatility. Vector autoregressive (VAR) estimation using both monthly
and daily data presents evidence that the lagged ETF trading affects the volatility
of the index, whereas conversely, the lagged index volatility affects the trading
volume of ETFs. This means that while the trading of S&P 500 ETFs Granger
causes S&P 500 index to be more volatile, the indexs volatility also Granger
causes ETF trading volume to increase. Although the Granger causality does
not imply the actual causality,this nding provides additional evidence of strong
interaction and close correlation between the volatility of S&P 500 index and the
trading of its ETFs. The result of the two-way Granger causation is intuitive.
While a more volatile index is likely to be accompanied by a larger trading
volume of ETFs, the high volatility of the index implies that the ETF prices are
more likely to frequently deviate from the fundamental value of the underlying
index, which creates buying and selling pressure on the ETF market. This
two-way Granger causality is consistent with what Roll, Schwartz, and
Subranhmanyam (2007) nd about the two-direction interaction of price shocks
between the futures and equity markets.
In addition, the linear trend analysis of the monthly time series suggests that
the advent of S&P500 ETFs in January 1993 is associated with the acceleration of
S&P 500 volatility growth. The time series of S&P 500 volatility shows an upward
trend over August 1962 to December 2013. However, the trend is slightly steeper
after January 1993, which coincides with the introduction of the rst S&P 500
ETF. Empirical evidence also indicates that S&P 500 volatility had a structural
break in January 1993 and this break is statistically signicant by certain
measures. Consistent with this structural break, it is found that S&P 500 index
has experienced a greater persistence on volatility shocks since January 1993.
All these ndings imply that the introduction of ETFs is likely to generate a shock
to the underlying stock index, because a new ETF creates a new derivative market
with dedicated investors and arbitrageurs focusing on this market.
The strong interaction between the volatility of a market index and the
trading volume of its ETFs is not limited to the S&P 500 index. Extending our
examination to a small-capitalization index in the USA and other market indexes
around the world, we can draw a similar conclusion that the contemporaneous
trading volume of ETFs is a signicant determinant of their indexs volatility.
Furthermore, there is a two-direction Granger causality between an indexs
volatility and the trading volume of its associated ETFs.
Exchange Traded Funds and Stock Market Volatility
© 2017 International Review of Finance Ltd. 2017 3
Exchange Traded Funds and Stock Market Volatility
© 2017 International Review of Finance Ltd. 2017 527

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT