Standard and optimized carry trades

• Date: 01 July 2018
• Author: Jurij‐Andrei Reichenecker
• Published date: 01 July 2018
• DOI: http://doi.org/10.1002/ijfe.1619

Received: 20 November 2017 Accepted: 13 December 2017

DOI: 10.1002/ijfe.1619

RESEARCH ARTICLE

Standard and optimized carry trades

Jurij-Andrei Reichenecker

Institute for Finance, Chair in Business

Administration, Banking and Financial

Management, Univrsity of Liechtenstein,

Fürst-Franz-Josef-Strasse, Vaduz9490,

Liechtenstein

Correspondence

Jurij-Andrei Reichenecker,Institute for

Finance, Chair in Business

Administration, Banking and Financial

Management, University of Liechtenstein,

Fürst-Franz-Josef-Strasse, Vaduz9490,

Liechtenstein.

Email: jurij-andrei.reichenecker@uni.li

JEL Classification: G10; G11

Abstract

Drawdown periods of standard carry trades are primarily the result of losses in

classic carry trade currencies. These periods coincide with an increased finan-

cial stress, such as the recent financial crisis. The introduced optimized carry

trades employ a dynamic weighting scheme for currencies, which incorporates

general risk components. Optimized carry trades are therefore less exposed to

losses under financial stress, and provide an enhanced risk-return profile over

the entire and second half of the sample period and during periods of volatile

markets. These results find robust statistical evidence. Furthermore, optimized

carry trades have a lower correlation with traditional asset classes than stan-

dard carry trades. Traditional models of risk are less successful in explainingthe

returns of optimized carry trades.

KEYWORDS

alternative asset class, carry trade, portfolio optimization

1INTRODUCTION

A carry trade is a long-short strategy,which purchases and

sells high- and low-yield currencies, respectively. Broad

empirical evidence exists that carry trades are profitable,

and therefore, violate uncovered interest rate parity (UIP;

Lothian & Wu, 2011). A standard carry trade selects an

even number of currencies and applies an equal weight-

ing scheme, so that each currency has the same absolute

weight.

The unique selection criterion of standard carry trades

is the interest rate of currencies. Risk components, such

as currency volatility or the correlation between curren-

cies, are disregarded. Therefore, optimized carry trades

are introduced, which extend the selection of currency by

risk components, and deviate from an equal weighting of

currencies.

Lustig, Roussanov, and Verdelhan (2014) and Barroso

and Santa-Clara (2015) suggest new types of carry trades.

These carry trades apply a weighing scheme that devi-

ates from the equal weighting scheme of standard carry

trades, and are able to outperform standard carry trades

in economic terms. The carry trade suggested by Barroso

and Santa-Clara (2015) doubles the Sharpe ratio relative to

standard carry trades.

We introduce two types of optimized carry trades. Both

outperform standard carry trades in an out-of-sample

framework and after transaction costs. This outperfor-

mance is driven primarily by a significant and robust

reduction in volatility. Optimized carry trades are less

exposed to traditional asset classes than standard carry

trades.

The economic motivation for optimized carry trades is

provided by Melvin and Shand (2017). They investigate

drawdown periods of standard carry trades and find that

the Australian dollar, New Zealand dollar, Japanese yen,

and Swiss franc exhibit the largest negative return contri-

bution during drawdowns. All these currencies are typical

carry trade currencies. Therefore, during drawdowns it

would be beneficial to sell (purchase) high- (low-) yield

currencies, which contradicts the investment rules of stan-

dard carry trades. The time-varying selection of curren-

cies provides the first motivation to investigate a dynamic

weighting scheme driven by market conditions.

Int J Fin Econ. 2018;23:329–344. wileyonlinelibrary.com/journal/ijfe Copyright © 2018 John Wiley & Sons, Ltd. 329

330 REICHENECKER

Furthermore, standard carry trades are also exposed to

rare economic disasters of high-yield currencies (Farhi &

Gabaix, 2015). The exposure can therefore lead to catas-

trophic losses and large drawdowns. Drawdowns of carry

trades appear primarily during periods of high volatility,

which is why the correlation between standardcarry trades

and traditional asset classes increases during such time

periods. Due to the dynamic weighting of optimized carry

trades, these carry trades have the opportunity to react to

the time-varying market conditions, and to reduce the cor-

relation to traditional asset classes during financial stress.

If the correlation between optimized carry tradesand tradi-

tional asset classes declines in high-volatility time periods,

then such a behaviour provides a strong indication of a

diversification effect. This provides the second motivation

to study a dynamic weighting scheme for carry trades.

Brunnermeier, Nagel, and Pedersen (2008), Bakshi and

Panayotov (2013), and many others study standard carry

trades with a varying number of involved currencies.1It

is reported that the risk-return profile of standard carry

trades is enhanced as more currencies are involved, that

is, the Sharpe ratio increases as more currency pairs are

selected. It therefore seems that a diversification effect

does exist within the currency market. Optimized carry

trades apply a quadratic optimization, as indicated by

Markowitz (1952). The currency weights are chosen under

optimal risk-return trade-off. This potential diversifica-

tion effect within the currency market provides the third

motivation.

It is the basis of these three motivations that we inves-

tigate the weighting scheme of standard carry trades.

The aim of a deviation from the weighting of standard

carry trades is to reduce the drawdown periods, to reduce

the exposure to rare economic disasters, and to improve

the general risk-return profile. A time-varying weighting

scheme therefore offers the opportunity to detect market

opportunities.

The two optimized carry trades pursue the following

optimization approaches. The tracking error (TE) carry

trade aims to minimize the TE to a predefined stan-

dard carry trade. The TE carry trade therefore tracks

the standard carry trade as closely as possible. However,

the carry trade has the freedom to choose any currency

weighting, with the view of enhancing the risk-return

profile. The minimal volatility (MinVol) carry trade is a

minimal-volatility portfolio consisting of currencies. One

constraint is that the minimal-volatility portfolio captures

a certain interest rate delta between the selected curren-

cies. This strategy aims to implement a low volatility carry

trade under the constraint of a baseline interest rate delta.

The empirical examination reveals that both optimized

carry trade strategies are able to outperform standardcarry

trades in economic terms. The MinVol carry trade'sability

to outperform standard carry trades is supported by robust

statistical evidence. An investigation into the relationship

between carry trades and traditional asset classes reveals

that standard and TE carry trades are more correlated with

the traditional asset classes than MinVol carry trades —

particularly during periods of financial stress.

In order to emphasize the differences between standard

and optimized carry trades, we examine the explanatory

power of equity and currency risk factors for carry trades.

We test the risk factor for global foreign exchange (FX)

volatility, the cross-sectional dollar return (suggested by

Lustig, Roussanov, & Verdelhan, 2011), equity volatility

(VIX index), illiquidity (TED spread2), and equity risk fac-

tors (Fama-French factors). Additionally, we investigate

the systematic risk of carry trades (Christiansen, Ranaldo,

& Söderlind, 2011). We find that the examined risk fac-

tors are able to explain to up 12% of the return variation of

standard and TE carry trades. All risk factors can explain

up to 2% of the MinVol carry tradevariation. Furthermore,

systematic risk factors do not increase the degree of expla-

nation. This analysis reveals that MinVol carry trade does

not have a significant sensitivity to any risk factors relat-

ing to the equity or bond market, and the positive return

compensates for non-examined risk factors. The exam-

ined risk factors are therefore not part of the MinVol carry

trade. Standard and TE carry trades have a positive and

significant beta to the market return, with the result that

a compensation for market risk is indicated. Overall, this

finding leads to the hypothesis that the MinVolcarry trade

is a good candidate to increase the diversification of a bond

and equity portfolio.

Our study is closely related to Lustig et al. (2014) and

Barroso and Santa-Clara (2015). Both empirical investi-

gations examine carry trades in which the selection of

currencies is motivated by economic measures and risk

variables. Moreover, Bakshi and Panayotov (2013) use the

predictive power of commodities to enhance the weighting

scheme of currencies, so that the related carry trade has an

improved risk-return profile.

However, our study is different, as we apply a classi-

cal Markowitz portfolio and a TE optimization to create

optimized carry trades. The main contribution to the exist-

ing literature is that the MinVol carry trades robustly

outperform standard carry trades after transaction costs,

and in an out-of-sample setting. Moreover, the deviating

time-varying behaviour indicates that the risk consump-

tion of MinVol carry trades differs from standard carry

trades. This empirical finding increases the forward pre-

mium puzzle, as the UIP does not seem to be fulfilled for

optimized carry trades.

The rest of the paper is structured in the following

way. Section 2 discusses the related literature. Section 3

introduces standard and optimized carry trades. Section 4