Semivariance and semiskew risk premiums in currency markets
Published date | 01 March 2021 |
Author | José Da Fonseca,Edem Dawui |
Date | 01 March 2021 |
DOI | http://doi.org/10.1002/fut.22160 |
J Futures Markets. 2021;41:290–324.wileyonlinelibrary.com/journal/fut290
|
© 2020 Wiley Periodicals LLC
Received: 10 August 2020
|
Accepted: 11 August 2020
DOI: 10.1002/fut.22160
RESEARCH ARTICLE
Semivariance and semiskew risk premiums in currency
markets
José Da Fonseca
1,2
|Edem Dawui
2,3
1
Department of Finance, Business School,
Auckland University of Technology,
Auckland, New Zealand
2
PRISM Sorbonne, University of Paris 1
Pantheon ‐Sorbonne, Paris, France
3
The World Bank, Washington,
District of Columbia
Correspondence
José Da Fonseca, Department of Finance,
Business School, Auckland University of
Technology, Private Bag 92006, 1142
Auckland, New Zealand.
Email: jose.dafonseca@aut.ac.nz
Abstract
Using a model‐free methodology, variance, and skew swaps are extracted from
currency options for several foreign exchange rates. These swaps are decom-
posed into semivariance and semiskew swaps and can also be used to define
the variance‐skew swap. The decomposed “up”and “down”semivariance
swaps, the “down”semiskew swap and the variance‐skew swap explain well
the currency excess return. These properties remain valid when considering
the prediction of the currency excess return. Lastly, trimming these variables
does not affect the results implying that extreme values of the distribution
convey little information regarding the evolution of the currency excess return.
KEYWORDS
currency risk premium, semimeasures, skew risk premium, variance risk premium, variance‐skew
risk premium
JEL CLASSIFICATION
G11; G12; G13
1|INTRODUCTION
The explanation of the currency risk premium is an important problem in finance as it synthesizes how two economies
interact, Engel (2016). Beyond the usual macroeconomic variables used as determinant of the currency risk premium, the
option market offers an interesting route to investigate this problem. Since the seminal work of Bates (1991), many studies
have shown that options could be used to extract market participants' expectations for the currency risk premium.
Using a model‐free approach that has been extensively used during the recent years, a variance swap contract is extracted
from currency options and used to build factor models for the foreign exchange excess return. This contract enables the
computation of the variance risk premium that is a key quantity in asset pricing. Going one step farther than the literature, we
decompose the variance swap contract into up and down components, the former depending on the upper tail of the currency
distribution while the latter depends on the lower tail of the currency distribution. What is more, these semivariance swap
contracts allow us to compute the variance‐skew swap that captures the skewness of the currency distribution. To further
investigate the importance of the currency skewness distribution we follow Kozhan, Neuberger, and Schneider (2013)and
compute a skew swap value from foreign exchange rate options and, similarly to the variance swap contract, it is decomposed
into up and down semiskew swap contracts. This large set of option related variables allows us to build several factor models
for the currency excess return and assess to which extent they can explain and predict its evolution.
We show that decomposed variance and skew variables have higher explanatory power for the currency excess
return than undecomposed ones; that the variance‐skew swap is highly informative for the currency excess return; that
the down semiskew swap is as informative as the skew swap regarding the currency excess return; a combination of
down semivariance and semiskew swaps have higher explanatory power for the currency excess return than an up one;
for certain currencies third order moment related quantities, mainly down quantities, provide additional information to
second order moment related quantities. Trimming these explanatory variables, by removing extreme movements, and
re‐estimating the same factor models allow us to show that the information extracted from the far ends of the tails
contain no information on the evolution of the currency excess return. Lastly, the predictability of the 1‐,3‐, and
6‐month currency excess return is improved when semivariance and semiskew swaps are used in place of
undecomposed variance and skew swaps, it further illustrates the importance of the decomposition.
The paper is organized as follows. We present the key ingredients to obtain the quantities from option prices in
Section 2. A description of the empirical data used in our analysis is provided in Section 3. Regression tests and analysis
are performed in Section 4and Section 5concludes the paper.
2|ANALYTICAL RESULTS
The main purpose of this study is to analyze the variance, the variance‐skew and the skew risk premiums for the foreign
exchange option market using a model‐free methodology based on call and put foreign exchange rate options. To this
end,
ck()
tT,and pk()
tT,denote the European call and put option prices at time
t
with maturity
T
and strike
k
on the spot
foreign exchange rate denoted
s
t, which represents the value at time
t
in the domestic currency of one unit of the foreign
currency. It is often more convenient to use the forward foreign exchange rate (or forward price), so
f
t
T
,stands for the
forward price value at time
t
with maturity
T
that is related to the spot value through the standard equality
f
se=
tT trrTt
,(−)( −)
df , where
r
d
and
r
f
represent the risk free domestic and foreign rates, respectively. Lastly, it is con-
venient to define rf f=ln −ln
tT TT tT
,,,
, the log return of a position in the forward contract of maturity
T
for the period
tT
[
;]
. The availability of these derivative products enables the computation of a variance swap contract, a variance‐
skew swap contract as well as a skew swap contract along with the risk premiums associated with those contracts in a
model‐free way. In this study, we use the approach proposed by Kozhan et al. (2013). Extracting distribution
information from option prices, like higher moments, has a long history and has been performed in many works.
2.1 |The variance and semivariance risk premiums
A variance swap contract, payer of the fixed leg and receiver of the floating leg, is a contract between two counterparties
that involves the payment at maturity
tτ
+of an amount specified at the initiation date
t
, this amount is called the
variance swap rate, and receiving at maturity (i.e.,
tτ
+) of the swap contract the realized variance of a given asset
computed over the interval
tt τ
[
;+]
. The amount specified at the initiation date
t
is called the fixed leg as it is known
during the life of the contract while the amount received at maturity is called the floating leg of the swap as it is known
only at the end of the contract, so during the life of the contract that quantity fluctuates. Following the literature, it is
known that the variance swap rate is given by
var b
ck
kdk b
pk
kdk=2() +2()
,
tt τ
tt τf
tt τ
tt τ
ftt τ
,+
,+
+,+
2,+ 0
,+
2
tt τ
tt τ
,+
,+
∫∫
∞(1)
var var
=
+
,
tt τ
utt τ
d
,+ ,+(2)
with
b
tt
τ
,+
the zero‐coupon value at time
t
with maturity
tτ
+that is expressed in the domestic currency. The quantity
vartt
τ
u
,+ captures the second moment of the upper tail distribution of the underlying asset while vartt
τ
d
,+ captures the
second moment of the lower tail distribution. More precisely, vartt
τ
u
,+ involves only call options with strikes larger or
equal to the forward price
f
tt
τ
,+ and, as such, it depends on the second moment of the log return of the forward price
(i.e., rtt
τ
,+)conditional on the event that it is positive, thatis to say, conditional on
1
r{>0
}
tt τ,+ . Similarly, vartt
τ
d
,+ depends on
the second moment of the log return of the forward price (i.e., rtt
τ
,+)conditional on the event that itis negative, that is to
say, conditional on
1
r{<0
}
tt τ,+ . Also, as option prices are computed under the risk neutral probability, the variance swap
rate is also called the risk neutral variance and hereafter we use interchangeably these names.
Notice that Equation (1) involves a continuum of options and as in the market only a finite number of options are
available the two integrals are approximated by sums, see Equation (23) in Kozhan et al. (2013) for details.
DA FONSECA AND DAWUI
|
291
The floating leg of the variance swap is the realized variance of the underlying currency computed over
tt τ
[
;+]
using end‐of‐day values as it is the standard practice in the market and it is given by
rvarg r=(
¯)
,
tt τ
it
tτ
vii,+
=
+−1
,+1
∑(3)
with rf f
¯=ln −ln
ii itτit
τ
,+1 +1,+,+ with
i
ttt τ{,+1,…, +−1}∈the set of days covering the interval
tt τ
[
;+]
and
gx ex()=2( −1−
)
vx
. Performing a Taylor expansion of the exponential function the terms in the sum turn out to be
r
¯
ii,+1
2
, hence justifying the expression of realized variance for rvartt
τ
,+.The fixed and floating legs being defined, the
realization of a variance swap, payer of the fixed leg and receiver of the floating leg, is given by
vs rvar var=−
.
tt τtt τtt τ,+,+ ,+(4)
In practice this product is used to hedge against an increase of the currency's volatility. Indeed, at time
t
the amount
vartt
τ
,+ is specified (and paid at time
tτ
+) and if over the period
tt τ
[
;+]
the currency's volatility increases sub-
stantially, the quantity rvartt
τ
,+ received at time
tτ
+is larger than the amount specified at time
t
, the net value is
positive. Conversely, if the currency's volatility remains low or decreases over the period
tt τrvar
[
;+],
tt
τ
,+
is smaller
than vartt
τ
,+, overall the investormakes a loss. In this example, the point of view of a volatility protection buyer is taken,
the counterparty in that contract acts as a volatility protection seller and the cash‐flows are exactly the opposite.
Following the literature such as Carr and Wu (2009) or Ammann and Buesser (2013), taking the expectation (under
the historical probability measure) of Equation (4) leads to the variance risk premium, it is denoted
vs vs=[ ]
.
τtt τ,+
This quantity is negative in practice, it is interpreted as the amount an investor is willing to pay to hedge against volatility risk.
It is known that variance risk premiums for equity options and equity index options are negative, see Carr and Wu (2009).
It is of interest to compare an investment made in the variance swap contract with the return of an investment made
in a risky asset and therefore it is convenient to introduce the excess return of an investment made in the variance swap
contract, it is denoted by
xv rvar
var
=−1.
tt τ
tt τ
tt τ
,+
,+
,+
(5)
The qualifier “excess”follows from the fact that va
r
in Equation (1) involves at the denominator a zero‐coupon
bond, it makes that quantity a forward price, see Carr and Wu (2009) for details.
Similarly to the decomposition performed for the variance swap rate vartt
τ
,+ in Equation (2), the realized variance
can be decomposed into two components conditional on the return of the forward price
f
tt
τ
,+, it leads to
rvar grgr11=(
¯)+(
¯)
,
tt τ
it
tτ
vii r
it
tτ
vii r,+
=
+−1
,+1 {>0}
=
+−1
,+1 {<0}
tt τtt τ,+ ,+
∑∑ (6)
rvar rvar
=
+
.
tt τ
utt τ
d
,+ ,+(7)
Combining the decompositions for the risk neutral variance vartt
τ
,+ and realized variance rvartt
τ
,+,it is natural to
define the semivariance swap contracts
vs rvar var=−
,
tt τ
utt τ
utt τ
u
,+,+ ,+(8)
vs rvar var=−
,
tt τ
dtt τ
dtt τ
d
,+,+ ,+(9)
as these swaps involve “half”(roughly speaking) of the underlying currency distribution. Moreover, as vstt
τ
u
,+ depends
on the upper part of the underlying currency distribution, it is tempting to name it the up semivariance swap contract
while for obvious reasons vstt
τ
d
,+ is named the down semivariance swap contract.
292
|
DA FONSECA AND DAWUI
Get this document and AI-powered insights with a free trial of vLex and Vincent AI
Get Started for FreeStart Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting

Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
