TIPS and the VIX: Spillovers from Financial Panic to Breakeven Inflation in an Automated, Nonlinear Modeling Framework
| Date | 01 April 2018 |
| DOI | http://doi.org/10.1111/obes.12218 |
| Published date | 01 April 2018 |
218
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 80, 2 (2018) 0305–9049
doi: 10.1111/obes.12218
TIPS and the VIX: Spillovers from Financial Panic
to Breakeven Inflation in an Automated, Nonlinear
Modeling Framework*
Josh R. Stillwagon
†Economics Division, Babson College, Institute for New Economic Thinking (INET)
Program on Imperfect Knowledge Economics (IKE), Westgate Hall 313, 231 Forest Street,
Babson Park, MA 02457, USA (e-mail: jstillwagon@babson.edu)
Abstract
This paper examines the determinants of the breakeven inflation rate (BEI) on U.S. Treasury
Inflation Protected Securities. After controlling for several measures of liquidity, inflation
expectations and inflation uncertainty; financial fear itself (proxied with theVolatility Index
or VIX) remains a primary influence on BEI. To delve into the mechanism underlying this
association, the VIX is decomposed, using intraday data, into conditional varianceand the
variance premium capturing risk aversion. Aside from the 2008 crisis, most of the effect
emanated from the variance premium. Following the crisis, indicators of bank insolvency
risk gain prominence as well. Lastly, an automated nonlinear model finds convex effects
of variance, and diminishing returns to insolvency risk and liquidity.
I. TIPS breakeven inflation and flights to liquidity
Since the inception of inflation indexed bonds, first in the UK in 1981 and then in the US in
1997, these guaranteed real interest rates have intrigued both researchers and policymakers
for their informational content.1The spread between traditional Treasury rates and those
on Treasury Inflation Protected Securities (TIPS), referred to as the breakeven inflation
rate (BEI), has often been invoked as a real time, market-based measure of inflation expec-
tations. This follows from the supposition that arbitrage should equalize the two expected
real returns. It has long been recognized however that inflation risk premia are important
drivers of nominal yields and can help to account for deviations from the expectations
JEL Classification numbers: E43, G12, G01, C22, C52.
*I thank INET and Trinity Collegefor financial suppor t of this research, and Marie Hoerova and Steve Furnagiev
for data. I also thank a number of people for very helpful comments including two anonymous referees, Anindya
Banerjee, Geert Bekaert, Jurgen Doornik, Neil Ericsson, Deniz Erdemlioglu, Roman Frydman, Michael D.Goldberg,
David Hendry, Chris Hoag, Sebastian Laurent, Bent Nielsen, Frank Strobel, Giovanni Urga, and other participants
of the 16th Oxmetrics Conference at Aix-Marseille University and seminars at Trinity College, the University of
New Hampshire, Sacred Heart University, Appalachian State University,and Babson College. Any errors of course
remain solely those of the author.
1Barr and Campbell (1997) provides an early study.D’Amico, Kim and Wei (2010) note examples of references
to TIPS in FOMC meetings and Fed speeches.
TIPS and the VIX: Nonlinear spillovers 219
hypothesis of the term structure.2Analogously, inflation risk premia may create a diver-
gence between inflation expectations and BEI, since nominal Treasuries are subject to
inflation risk in a way which TIPS are not.3More recent studies meanwhile have empha-
sized the role of liquidity premia in BEI, given the much smaller and newer market for
TIPS.4These studies have primarilyconcentrated on using alter nativemeasures of liquidity
to account for the discrepancy between BEI and expected inflation.
What this study finds is that BEI has not been driven primarily byinflation expectations,
inflation forecast dispersion, or several common measures of liquidity itself, but rather by
financial market panic. Financial panic is proxiedhere, using the ‘fear gauge’ of the Chicago
Board Options Exchange Volatility Index or VIX (Whaley,2000). The VIX is a model free
estimate of the implied volatility of the S&P500 over the next month backed out from
options’ prices assuming risk neutrality.5The VIX alone accounts for approximately 60%
of the variation in 5 and 10-year BEI from 1999:01 to 2014:11; while the other variables
add only 15% collectively.
More specifically, it is found that when the VIX rises, BEI tends to fall and this effect
is significant at the 0.1% level. This contradicts the sign posited by Adrian and Wu (2009).
Although not the primary focus of their paper, Adrian andWu suggest, through some simple
graphics, that when market volatility is high, 5–10-year forward, BEI tends to overpredict
their model implied expected inflation. Meanwhile, this paper observes that when the VIX
is high, BEI tends to further underpredict expected inflation.6
One may naturally wonderabout the theoretical under pinnings of this negative associa-
tion between BEI and theVIX. The VIX is an amalgam of the perception of risk (variance),
aversion to risk, and uncertainty. Prior theory suggests that all three of these may influ-
ence liquidity preference. For example, Caballero and Krishnamurthy (2008) highlight the
role of Knightian uncertainty in flight to quality episodes; while Vayanos (2004) develops
a model where fund managers’ liquidity preference depends on both risk aversion and
conditional variance.
To delve further into the connection between BEI and theVIX, we can exploit a recent
method in the literature which uses intraday data to decompose the VIX into conditional
variance and a variance premium embodying risk aversion and uncertainty (Bollerslev,
Tauchen and Zhou, 2009; Drechsler andYaron, 2011). While decomposing the VIX only
marginally enhances the model fit, it illuminates the relativecontribution of the two compo-
nents. Both the theorized direct impact on liquidity preference from changing risk aversion
and the indirect impact through expected variance appear to be important.7
2Pennacchi (1991), Buraschi and Jiltsov(2005), Ang, Bekaert and Wei(2008), Christensen, Lopez and Rudebusch
(2010), and Wright (2011).
3Technically,TIPS are subject to some inflation risk due to the three-month lag in indexation.
4G¨urkaynak, Sack and Wright (2009), D’Amico et al. (2010), Zeng (2013) and Pfleuger andViceira (2016).
5The assumption of risk neutrality means that the price of an option is equal to its expected value, so the expected
variance can be inferred. See Britten-Jones and Neuberger (2000).
6Adrian and Wu focus on forward BEI and conduct structural change tests in part to control for liquidity effects,
perhaps due to their desire to use daily data. More direct controls could be advantageous however.
7The significance of the VIX is robust to estimation excluding the financial crisis, although the effect primarily
derived from the variance premium. See online appendix section S2.
©2017 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial