Journal of Asset Management

https://doi.org/10.1057/s41260-020-00161-6

ORIGINAL ARTICLE

Gauging the effectiveness of sector rotation strategies: evidence

from the USA and Europe
Constantinos Alexiou1 · Anshul Tyagi2

Revised: 3 April 2020

© Springer Nature Limited 2020

Abstract
This paper examines the performance of different sector rotation strategies for the US and European market spanning the
period 1999–2019. By utilising three different strategies, we shed further light on the effectiveness of interest rate, momen-
tum and Fama–French three- and five-factor alphas as switching signals to enter and exit a particular sector. The emerging
evidence suggests that within the European market, sector rotation strategies tend to produce returns above the average
benchmark, both during contractionary and expansionary monetary policy regimes, while excessive returns within both the
US and European markets are observed.

Keywords Sector rotation · Fama–French · Momentum · Interest rate · Business cycles

JEL Classification E32 · E52 · G11

Introduction (business models) among different sectors; (b) the existence

of psychological aspects associated with investor sentiments
Over the last decade, sector rotation—one of the most preva- about one sector in comparison with the future performance
lent and effective dynamic asset allocation strategies—has of other sector.
gained considerable attention from investors. Its rationale is While a significant number of studies have been con-
premised on the maintenance of a balance between passive ducted on the concept of sector rotation strategies, a com-
indexing strategies and active stock picking. It assigns asset mon recurring theme among them has been the focus on
allocation in accordance with the current stage of business majorly the US market and the use of investment tools such
cycle by taking a long stance on sectors which exhibit the as mutual funds. The obvious problem with this recurring
potential to perform well in the future and short on sec- theme is that the conclusions drawn from such a limited data
tors expected to perform poorly. In this context, more than source and the lack of diversity in factors considered may be
a thousand exchange traded funds (ETFs) throughout the subject to various biases. This inherently makes the practi-
world have been constructed to allow investors to limit their cal applicability of these conclusions somewhat restricted,
exposure to certain sectors, making sector rotation attractive regardless of their theoretical premise.
to investors. It has been over a decade since the Great Recession
In the extant literature, two potential reasons are proposed erupted in 2007/8, and Europe is still trying to recover
to explain the impact of economic cycles on different sec- from the shock and derailed many EU economies. The US
tors; (a) the existence of fundamentals, i.e. certain sectors economy recovered relatively faster and has been close to its
benefit/suffer in relation to other sectors during the boom/ pre-crisis levels since 2011. The difference in the recovery
bust cycle because of difference in underlying fundamentals and growth of the equity market in these two regions can to a
certain extent be attributed to the existing differences in their
regulatory and monetary policies adopted by their respective
* Constantinos Alexiou monetary institutions, i.e. the ECB and the Federal Reserve.
[email protected] The underlying different approaches as to how these regions
1
Cranfield University, Bedford, UK dealt with the global financial crisis in conjunction with the
2 performance of their respective equity markets provided the
Carlsquare, London, UK

Vol.:(0123456789)
 C. Alexiou, A. Tyagi

impetus for this study to explore the differences in the sector The majority of the existing literature is concentrated
rotation strategies, hence enabling us to gain a more insight- on the mutual fund performance and has paid relatively
ful understanding. very little attention to sectoral performance. For instance,
This paper aims to address these problems and break new Dellva et al. (2001) studied the performance of the 35 Fidel-
ground in the area by offering more robust conclusion. The ity sector mutual funds during the 1989–1998 time period.
contribution of the paper is threefold: the first contribution Throughout the sub-periods, the value of Jensen’s alpha was
of this paper lies within the range and diversity of data taken observed as positive and within the range of 24–33. The
into consideration. Unlike the majority of the previous stud- only exception was between the time periods 1994–1998, as
ies solely focusing on the US market, this paper also takes in this period the alpha values were in negative. Similarly,
into consideration the European market. This will enable us Kacperczyk et al. (2005) studied the relationship between
to identify if there exists a “universal” indicator that can per- industry concentration and the performance of actively man-
haps predict the right timing for switching between sectors aged US mutual funds from 1984 to 1999. The paper argued
within different economies. To the best of our knowledge, that mutual fund managers may benefit more from concen-
this is the first study trying to gauge the effectiveness of trating their holdings in specific sectors/industries than from
different sector rotation strategies in the European market. a well-diversified portfolio.
Moreover, the use of the three major indicators collectively In contrast, Dou et al. (2014) studied cross-regional and
in one study is another contribution to the existing literature, cross-sectoral asset allocation in both bullish and bearish
while at the same time we focuses on the use of ETFs as a market and reported net positive alpha for energy, technol-
tool to execute the strategies. ogy and healthcare sectors; and negative alphas for Durable,
The rest of the paper is organised as follows: “Related lit- Telecom and Manufacturing sectors. In a study to under-
erature” section reviews the existing literature, while “Meth- stand the impact of business cycles on stock returns, Chor-
odological framework” section elaborates on the methodo- dia and Shivakumar (2002) found that momentum-driven
logical framework utilised. “Results and discussion” section trading strategies were only able to statistically outperform
presents and discusses the results, and finally, “Conclusions” the market during periods of expansion; however, it failed
section provides some concluding remarks. to perform equally well during periods of recession. In this
context, it is argued that if we can predict the current phase
of the business cycle, then it would help to predict the sec-
Related literature tors capability to outperform or underperform the market
(Luk 2012).
Over the years, several studies have emerged in an attempt to
examine the effectiveness of sector rotation strategy. Jacob- Macroeconomic considerations
sen, et al. (2009) argued that investing money on sector
rotation was able to generate higher returns in comparison In the past several studies tried to explore the nature of rela-
with a passive investment strategy of “buy and hold”. Grauer tionship between macroeconomic variables and sectoral
et al. (1990) constructed a multi-period investment model performance in different economies including inter alia a
with 12 industry indices to create a sector rotation strategy study by Dedola and Lippi (2005) which analysed cross-
and found out that equally weighted portfolios were able to sectoral effects arising from monetary policy in five OECD
outperform value-weighted portfolios. In the same spirit, countries or Hayo and Uhlenbrock (1999) who estimated
Doeswijk and Vliet (2010) successfully predicted future cross-sectoral effects in Germany. Apart from them, most
sector returns using the momentum and season factor, and of the previous literature on this aspect of sector rotation
thereafter created a sector rotation strategy using these fac- has primarily focused on the US market include, Conover,
tors for a number of countries in which the strategy obtained et al. (2005).
an average annual return of 12.9% from 1970 to 2008. In an Kouzmenko and Nagy (2009) in a study on the associa-
attempt to predict future returns for various sectors, Rapach tion between business cycles and sectoral performance in
et al. (2015) constructed three econometric models the developed economies namely USA, Europe and Japan found
results of which were used to design three sector rotation that sectors showing high cyclicality were able to outperform
strategies. The generated average annual alpha returns from sectors which reflected the existence of strong defensive
1985 to 2012 were 7.82% and 12.28%, respectively. Bel- behaviour in the time of expansion and vice versa. Siegel
ler et al. (1998) using key macroeconomic variables found and Hoban (1991) presented an alternative sector rotation
that switching sectors based on mean–variance optimisation strategy where the author switched from equities to bonds
was able to outperform the benchmark and generate statisti- based on the stage of the business cycle resulting in excess
cally significant monthly returns of about 1.7% from 1981 returns. Parvez et al. (2002) found that using more than
to 1995. one category of stocks (e.g. mid-cap growth equity)-based
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

rotation strategy can enhance the performance of the port- due to investors “overreaction” during periods of poor
folio much better than by simply rotating across unilateral performance.
categories (e.g. mid-cap equity). Jegadeesh and Titman (1993) found out that strategies
Building upon their previous work, Conover et al. (2008) focusing on return reversals are transaction intensive & rely
established that cyclical sectors performance is enhanced by on short-term price volatility, therefore making the success
easing of monetary policy, while that of non-cyclical sectors of the strategy rather questionable and suggesting that these
is enhanced by tight monetary policy. This strategy was able abnormal returns may be because of short-term price pres-
to outperform the benchmark by 3.4% higher yield. sure or a lack of liquidity in the market rather than over-
Jensen and Johnson (1995) studied the daily returns after reaction. Robert (1967) stated that a trading strategy based
interest rate changes and found that monetary easing leads on buying stocks with current prices that are substantially
to positive abnormal returns for a period of time, while mon- higher than their average prices over the past 27 weeks real-
etary tightening leads to negative results. To understand the ises significant abnormal returns. However, Jensen and Ben-
cross-country and cross-sector impact of Federal monetary nington (1970) analysed the same strategy over an extended
policy, Ehrmann and Fratzscher (2004) conducted a study period of time, mostly outside Levy’s sample period and
and found out that cyclical sectors are relatively more sen- found contradictory results and hence attribute Levy’s result
sitive towards changes in interest rate ranging from a 7.4% to a selection bias.
response of the information technology sector to 1.6% for the Sorensen and Burke (1986) studied the relative strength
utilities sector. In this paper, it was established that certain index (RSI) of 43 sectors in the USA from 1972 to 1982 and
non-cyclical sectors such as food, agriculture and beverages found that a sector rotation strategy based on buying and
to be least sensitive to changes in interest rate. Bernanke and holding the best-performing sectors may enhance returns.
Kuttner (2005) studied the impact of changes in monetary Similarly, Cavaglia and Moroz (2003) while studying
policy on the equity prices and found out that an unexpected cross-sector and cross-country allocation framework used
25 basis point drop in the Federal funds rate can lead to 1% multi-factors such as price momentum, dividend yield, EPS
increase in the broad stock indexes. The paper also argued forecast and found that a strategy focused on investing in
that cyclical sectors respond more sensitively to changes in profitable, fairly priced, and with positive momentum sec-
interest rate, similar to Ehrmann and Fratzscher (2004). tors generated average annualised returns between 3 and
Even though it becomes apparent from the above that 4.5% from 1990 to 2001. Grinblatt and Titman (1989, 1992)
using regime changes in the interest rate as an indicator for studied a number of the mutual funds from 1975 to 1984 and
sector rotation is advantageous a number of limitations have found out that a majority of them preferred to buy stocks that
to be considered. Moe specifically, the nature of the rela- showed surge pricing over the previous quarter.
tionship between the central banks’ monetary policy, busi- Jegadeesh and Titman (1993) documented that going long
ness cycles and other macroeconomic conditions is highly on stocks with superior performance in past and going short
convoluted, hence making it hard to understand their causal on stocks with poor performance in the past generated sig-
dimension (see Durham 2005). nificant positive returns over 3- to 12-month holding peri-
ods. Additionally, the authors suggested the reason behind
Momentum trading strategy considerations abnormal returns is that the investors buying past winners
and selling past losers cause a temporary shift in the prices
The past research on the examination of time-series predicta- away from their original pathway values, which results in
bility of stock returns on historical basis research established price overreaction.
that the predictable patterns in stock returns can be used Although the current academic literature has preferred
to outperform the market. However, most of the traditional contrarian strategies over relative strength trading strategies,
risk-based asset pricing models have failed to explain the however, many practitioners still use relative strength as a
stock returns which led researchers to look for answers in criterion for stocks screening. But since the seminal work of
the behavioural theories that could explain the returns and Jegadeesh and Titman (1993), many other academicians and
the associated trends. In this context, Jegadeesh and Titman practitioners have tried to explore the impact of momentum
(1993) revealed the presence of positive autocorrelations effect (anomaly) on stock returns. However, unlike the long-
(momentum) in stock returns in short- and medium-term term return reversal, all the popular asset pricing models fail
horizon. to effectively explain the momentum effect, making it an
In line with the popular view held by many that indi- anomaly and is therefore subject to further analysis.
viduals tend to overreact to new information, De Bondt and Moskowitz and Grinblatt (1999) argued that most of
Thaler (1985) found that overreaction caused by new infor- the momentum-based returns can be captured by forming
mation have significant impact on stock prices. The authors industry-/sector-based portfolios. The authors utilised sec-
suggested that contrarian strategies achieve abnormal returns tor momentum to construct a sector rotation strategy. The
 C. Alexiou, A. Tyagi

strategy used was to use the front 6-month returns to catego- In view of the preceding analysis, we proceed to the
rise twenty US sectors, and then go long on the three best- empirical investigation whereby three methodological
performing sectors and go short on three worst-performing approaches are used to investigate the three different strate-
sectors. This strategy obtained an average monthly return of gies and shed further light on the effectiveness of interest
0.43% from 1963 to 1995. Continuing the work of Moskow- rate (economic indicator), momentum (technical indicator)
itz and Grinblatt (1999), O’Neal (2000) suggested the use of and Fama–French three- and five-factor alphas (performance
tradable assets based on distinct sectors, specifically sector- indicators) as switching signals to enter and exit a particular
based mutual funds, to generate superior returns arising from sector.
the momentum effect. Later, Swinkels and Tjong-A-Tjoe
(2006) proposed the use of exchange traded funds for the
same purpose. Methodological framework
Scowcraft and Sefton (2005) investigated cross-sec-
tor- and cross-country-based rotation strategies in order To investigate the effectiveness of macroeconomic factors,
to identify if the strategy could be used to outperform the i.e. changes in GDP and the interest rate, we collate monthly
market from 1980 to 2003. By estimating a linear factor returns of SPDR sector ETFs for both USA and Europe
model (LFM) based on industry, country and country-spe- spanning the period January 1999 to June 2019. The main
cific factors to better understand the driving force behind data sources were OECD and the Federal Reserve Bank
stock momentum the found that 58% of the momentum is of St. Louis. The changes were categorised into periods
attributed to industry factors and 8% to stock momentum in of growth and decline. For the interest rate shifts, the Fed
case of market value-weighted portfolio. However, 66% is Funds Target rate for the USA and Euro Main Refinancing
attributed to stock momentum in case of equally weighted Announcement Rate for European market have been used
portfolio. over the same period. Interest rate changes were categorised
into monetary easing or tightening. The main data providers
Multifactor asset pricing considerations were Bloomberg and DATASTREAM.
For the empirical investigation, 9 US sector ETFs and 10
Fama and French (1993)’s three-factor model (FF3), Carhart European sector ETFs are considered. The sectors for both
(1997)’s four-factor model and Fama and French (2015)’s the markets have been categorised into either cyclical or
five-factor model (FF5) are the most commonly used multi- non-cyclical sectors based on MSCI’s classification criteria
factor asset pricing models and performance evaluation tools on GICS code. Cyclical sectors, by definition, are more sen-
in empirical research. Despite the popularity of FF3 model, sitive to changes in market and other related macroeconomic
several papers suggested the limitations of FF3 model in variables than non-cyclical sectors.
explaining many capital market anomalies including the The monthly returns of the sector-based ETFs for the
cross-section of stock returns. USA and Europe are analysed and compared to the monthly
Although the FF3 model is an extension of CAPM, by returns of their benchmarks, namely S&P 500 index and
adjusting for outperformance tendencies observed in CAPM, Stoxx, respectively. To assess the effectiveness of mon-
it still failed to account for cross-sectional variation in etary policy as an indicator for sector rotation, this paper
expected returns primarily associated with investment fac- follows Conover et al. (2008), i.e. an equity portfolio was
tor and profitability factor. A number of papers supported constructed by investing in an equally weighted portfolio
this finding include inter alia Chen et al. (2011), Aharoni comprising of all the cyclical sector ETFs in periods of mon-
et al. (2013) and Walkshäusl and Lobe (2014). etary easing, and investing in an equally weighted portfolio
Other major anomalies that remain unanswered by FF3 comprising of all the defensive (non-cyclical) sector ETFs
model are positive relationship of average returns with sur- during periods of monetary tightening. In passing, it should
prises associated with momentum and earnings, negative be noted that transaction costs are assumed to be zero due to
relationship of average returns with financial hardship, idi- the complexities involved in allocating the transaction cost
osyncratic volatility, net equity issues and growth in assets. to each transaction as they are based on a number of factors
These anomalies have been studied in depth by many authors such as tracking error, ETF Bid/Ask spread and commission.
(see for example Spiess and Affleck-Graves, 1995 and Chan We then proceed to assess the second strategy by creating
et al. 1995). To counter such limitations, Fama and French a momentum-based portfolio with some modifications (see
(2015) proposed the five-factor model—i.e. extending on Jegadeesh and Titman 1993). To reflect the price momen-
the previous FF3 model—by further adding two extra fac- tum, this section uses previous “J” month’s returns to con-
tors—profitability and investment and found it to be more struct a portfolio and the performance is measured over the
effective than the FF3 model. next “K” months. For this section, 16 portfolios were created
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

for each economy with “J” and “K” taking values of 1, 3, 6 The model specification for the respective asset pricing
and 12 months. models is expressed as follows:
Using 1 month previous returns as “J” and holding period
of 1 month as “K” are an addition to Jegadeesh and Titman 1. Fama–French—three-factor model
(1993) as some other empirical papers such as Moskowitz ( )
and Grinblatt (1999) and Grundy and Spencer (2001) found Rit − Rft = 𝛼it + 𝛽1 RMt − Rft + 𝛽2 SMBt + 𝛽3 HMLt + 𝜀it .
the sector momentum to be strongest at 1 month time inter-
2. Fama–French—five-factor model
val. The momentum portfolios were created by taking only
long positions in the top 3 sectors with the highest momen-
( ) ( ) ( )
Rit − Rft = 𝛼it + 𝛽1 Rmt − Rft + 𝛽2 SMBt + 𝛽3 HMLt
tum for each value of “J”. ( ) (
+ 𝛽4 RMWt + 𝛽5 CMAt + 𝜀it
)
The Winner sectors in this section are ranked based on
the work of De Bondt and Thaler (1985) and Jegadeesh and where Rit denotes total return of a stock or portfolio i at
Titman (1993), which define the most commonly used meth- time t; Rft represents the risk-free rate of return at time t
ods at present. The criteria for determination of ranking of (proxied by 1-month US Treasury bill); RMt is the total
the sectors are based on the performance in the formation market portfolio return at time t; Rit − Rft is the expected
period. The most common methods of calculating cumula- excess return; RMtt − Rft represents the excess return
tive stock return for a time period are either arithmetic mean generated over the market portfolio; S ­ MBt denotes
or geometric mean. We have adopted the arithmetic mean the size premium (difference between small and big);
for calculating cumulative returns of the formation period ­HMLt is the value premium (difference between high
as Jacquier et al. (2003) reported that for short-term invest- and low); ­RMWt denotes the profitability factor (dif-
ment horizons the arithmetic mean gives results closer to ference in return of the most profitable firms and the
“unbiased compounding rate” and for long-term horizons the least profitable firms); CMAt represents the investment
geometric mean minimises bias and presents results closer factor (difference in return of firms investing conserva-
to “unbiased compounding rate”. tively & those investing aggressively); β1,….,β5 are the
The formula for calculating the return is as follows: factor coefficients; and it (alpha) represents the access
T return that remains unexplained by FF3 and FF5 models,
Arithmetic Mean (Ri,T ) =
1 ∑
∗ Return(i) respectively.
T i=1
In this direction, we use the S&P 500 composite index
where Arithmetic Mean (R i,t) denotes the buy-and-hold data which are taken from the DataStream and Robert Shill-
return for sector “i” within the formation period; Return er’s database. The US and European business cycle data are
(Ri) is the monthly return of sector “i” within formation taken from NBER and FRED database. This paper uses
(ranking) period; and T is the month of sector returns under ­SPDR® MSCI Europe UCITS ETFs as a proxy to European
consideration within formation period. sectors & Select Sector ­SPDR® Fund as a proxy to US sec-
Sectors are ranked based on the above criteria, and the top tors as an alternative to US Fama–French sector portfolios to
3 sectors are included in the winner portfolio on an equally analyse sector rotation strategies, something that an investor
weighted basis. After ranking, the zero-cost sector portfo- would most likely use in real-world scenarios. However, the
lio tagged as “Winner Portfolio” is bought for K months. trading period is shorter than Fama–French sectors because
The overall return of the winner portfolio here is the equally of the availability of the data. The ETFs data have been col-
weighted average of the cumulative return of different stocks lected from Bloomberg and DataStream from 1999 to 2019,
that constitute the portfolio within holding period. After consisting of 234 trading intervals (in months).
J months, the portfolio is rebalanced using the same steps This paper uses 36 months rolling FF3 and FF5 alphas
as mentioned above and the returns are calculated, and the as a signal for asset reallocation via long only and a long/
process continues till the end of the sample period. short strategy. The 36-month rolling FF5 alphas were cal-
In Table 16 in “Appendix”, a pictorial representation of culated for the period 1999–2019. The long-only strategy
sector rotation strategy based on previous “J” months per- buys the sector ETFs in “t + 1” month for sectors with posi-
formance and “K” months investment horizon with J = 6 and tive 36 month rolling FF3 and FF5 alpha, respectively, in
K = 3 is provided. month “t”. On the other hand, the long & short strategy
Finally, in exploring the risk-adjusted performance of invests in the sector ETFs in “t + 1” month for sectors with
the US and European equity sector portfolios, we utilise the positive 36 month rolling FF3 and FF5 alpha, respectively,
Fama and French five-factor sector rotation strategy based in month “t” and sells the sector ETFs in “t + 1” month for
on alpha as a performance indicator and switching signal as sectors with negative 36 month rolling FF3 and FF5 alpha,
a framework (see Sarwar et al. 2017).
 C. Alexiou, A. Tyagi

Table 1  The US monetary policy changes (1999–2019)

Policy period Policy Start date Start rate (%) End rate (%) No. of rate changes Total changes
in rates (%)

1 Tightening 30/06/1999 4.75 6.50 6 1.75

2 Easing 31/01/2001 6.50 1.00 12 − 5.50
3 Tightening 30/06/2004 1 5.25 17 4.25
4 Easing 28/09/2007 5.25 0.13 8 − 5.12
5 Tightening 31/12/2015 0.13 2.38 as on 28/06/2019 9 until 28/06/2019 2.25

Table 2  European monetary policy changes (1999–2019)

Policy period Policy Start date Start rate (%) End rate (%) No. of rate changes Total changes
in rates (%)

1 Easing 31/12/2001 3.75 2.00 4 − 1.75

2 Tightening 31/01/2006 2.00 4.25 9 2.25
3 Easing 30/11/2008 4.25 1.50 7 − 2.75
4 Tightening 31/05/2011 1.00 1.50 2 0.50
5 Easing 31/12/2011 0.05 0.00 as on 28/06/2019 8 until 28/06/2019 − 0.05

respectively, in month “t”. And the portfolio is rebalanced in monetary policy in the recent past and have implemented
every month using the same framework. This paper also uti- the restrictive policy from January 2016 and is continuing
lises the Sharpe ratio to compare the risks involved between with that until today.
different strategies along with the buy-and-hold strategy. The In Europe, however, the average discount rate during
particular ratio is used as it not only measures the return of tightening and easing monetary policy periods has been
the security with respect to the risk taken but also bench- 2.88% and 1.75%, respectively, from 1999 to 2019 with the
marks it with the returns generated from a risk-free asset, discount rate ranging from 0.00% in June 2019 to 4.25% in
which in true sense is a correct reward measurement matrix. January 2006 and November 2008. Unlike the USA, mon-
The buy-and-hold strategy represents an investment in the etary policy in the EU is dominated by monetary easing over
benchmark for respective markets, namely S&P 500 index the span of the sample. The table reflects the ECB’s initia-
for the USA and Euro Stoxx 50 for the European market. tive to deal with the financial crisis of 2008, and Euro debt
crisis at its peak in 2011 and their attempt to ease the credit
market conditions by using monetary easing in the respec-
tive periods following the crises. The ECB has implemented
Results and discussion the easing policy from January 2012 and is still continuing
with that till today with current interest rate still being at 0%
Sector rotation based on interest rate and GDP from April 2016.
Using the change in monetary policy from easing to
Tables 1 and 2 summarise the nature of monetary policy tightening and vice versa as an indicator for sector rotation
(easing and tightening) from June 1999 to June 2019 for the between cyclical and non-cyclical (defensive) sectors, the
USA and Europe, respectively. portfolio was rebalanced only 5 times in the last 19 years for
The average interest rate during easing and tightening both the USA and Europe.
monetary policy in the USA was 3.34% and 3.22%, respec- Table 3 presents the annual mean returns and standard
tively, from 1999 to 2019 with interest rate ranging from deviations of the sector ETFs under consideration for the
0.13% in December 2015 to 6.5% in January 2001. As it USA and Europe, respectively. The US sectors have shown
becomes apparent from the data monetary policy in the USA a wide range of returns from 4% in technology sector to
was mostly contractionary in the sample period. The table 17% in energy sector with a wide volatility ranging from
reflects the Feds initiative to overcome the financial crisis of 12% in consumer staples to 25% in energy. Similarly, Europe
2008 and ease the credit market conditions by using expan- has shown wide range in returns (9–16%) and volatility
sionary monetary policy from end of 2007 until December (15–28%). In general, cyclical sectors have shown to be
2015. Also, the Feds have reduced the frequency of changes more sensitive to market changes which can be observed
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

Table 3  Mean sector returns and standard deviations (1999–2019) Table 5  Europe—sectors performance by monetary periods (1999–
2019)
USA Europe
Expansionary Restrictive
Mean (%) SD (%) Mean (%) SD (%)
Mean (%) SD (%) Mean (%) SD (%)
Non-cyclical
Consumer staples 9 12 12 15 Non-cyclical
Energy 17 22 5 16 Utilities 7.42 12.97 − 0.61 7.52
Health care 5 13 7 15 Consumer staples 8.53 11.02 0.26 6.03
Utilities 9 16 9 19 Health care 7.75 11.50 − 1.25 5.82
Average for non-cyclical 10 16 8 16 Energy 6.82 15.11 − 0.97 10.24
Cyclical Average for non-cyclical 7.63 12.65 − 0.64 7.40
Consumer discretionary 8 20 10 22 Cyclical
Financial 11 22 6 25 Communication 2.90 15.35 0.14 7.41
Industrial 11 19 14 23 Consumer discretionary 11.11 16.00 − 3.02 9.64
Materials 11 21 16 28 Industrials 12.09 15.35 − 2.82 10.62
Technology 4 25 9 24 Technology 8.46 23.35 − 3.69 9.86
Communication – – 9 18 Materials 11.93 16.69 − 2.51 12.43
Average for cyclical 9 21 11 23 Financials 8.70 20.55 − 5.61 10.82
Average for cyclical 9.20 17.88 − 2.92 10.13

Table 4  The USA—sectors performance by monetary periods (1999–

2019) tightening monetary policy regimes in sample period. How-
Expansionary Restrictive ever, cyclical sectors were able to perform better than the
non-cyclical sectors during monetary easing regimes and
Mean (%) SD (%) Mean (%) SD (%)
non-cyclical sectors were able to perform better than the
Non-cyclical cyclical sectors during monetary tightening regimes. These
Consumer staples 2.39 9.32 2.47 7.66 results are in line with the previous studies conducted by
Energy 0.72 16.48 5.73 13.04 Kouzmenko and Nagy (2009) and Conover et al. (2005). The
Health care 4.88 11.24 2.04 8.18 standard deviation for cyclical sectors is higher than the non-
Utilities − 0.48 11.51 5.05 8.35 cyclical sectors during both monetary regimes indicating
Average for non-cyclical 1.88 12.14 3.82 9.31 that cyclical sectors are more sensitive to changes in inter-
Cyclical est rate; these results are similar to Ehrmann and Fratzscher
Consumer discretionary 6.11 15.01 2.78 9.89 (2004). Basically, simple reallocation between sectors in the
Financial − 0.04 17.87 3.85 11.07 USA based on changes in monetary policy as an indicator
Industrial 2.12 15.34 4.59 9.46 based on previous studies could have slightly improved the
Materials 3.12 16.94 3.23 10.88 returns for this particular sample period.
Technology 2.22 18.77 4.18 12.91 Table 5 suggests that the market in Europe reflected
Average for cyclical 2.71 16.79 3.73 10.84 similar results as the USA where cyclical sectors were able
to perform better than non-cyclical sectors during mon-
etary easing regime and non-cyclical sectors were able to
from high standard deviation in cyclical sectors as compared perform better during monetary tightening regimes in the
to non-cyclical sectors in both the USA and Europe, which sample period. Further, the differences in return between
is in line with the previous research. easing and tightening monetary policies are higher in
The table reflects that different sectors have generated cyclical sectors than compared to non-cyclical sectors
different returns and have shown different volatility from which reflects that the cyclical sectors are more sensi-
1999 to 2019 irrespective of the market, which further signi- tive to changes in interest rate (Ehrmann and Fratzscher
fies that a sector rotation strategy if implemented using the 2004). The average difference between the easing and
right entry and exit signal has the potential to substantially tightening periods average returns for cyclical sectors
outperform the market returns and also reduce volatility. in Europe is 12.12% (9.20% t − (− 2.92%)) & for non-
Tables 4 and 5 summarise the sector returns and their cyclical sectors the average difference in returns is 8.27%
standard deviations for both monetary easing and monetary (7.63%t − (− 0.64%)). However, the average standard devi-
tightening policies. Table 4 reflects that the market in the ation for cyclical sectors in easing as well as tightening
USA generated very weak results during both easing and monetary policy regime is higher than the non-cyclical
 C. Alexiou, A. Tyagi

Table 6  Sector rotation USA Europe

portfolio versus benchmark
(1999–2019) Mean (%) SD (%) Sharpe ratio Mean (%) SD (%) Sharpe ratio

Portfolio 6.73 16.60 0.406 11.14 15.24 0.73

Benchmark 7.04 14.70 0.48 8.53 21.76 0.39
Excess return − 0.31 – – 2.61 – –

Mean percentages denote annual returns

Fig. 1  Europe—momentum-
based sector rotation strategy
performance

sectors. Meaning, that simple reallocation between sec- this section, the portfolio was rebalanced only 5 times for
tors in Europe based on changes in monetary policy as an both USA and Europe.
indicator might have improved the returns but could also We decided against using the results that corresponds to
have increased the volatility. changes in GDP as an indicator for switching between sec-
Table 6 compares the mean return, standard deviation and tors because the historical data revealed that it leads to very
Sharpe ratio of the sector rotation portfolio & benchmark high volatility in the stock portfolio (see Tables 17 and 18
for both the USA and Europe, respectively. According to in “Appendix”). This might to some extent be attributed to
the table, the difference between (annualised) mean return the frequency of the data, i.e. quarterly, which fails to reflect
of the sector rotation strategy portfolio and the benchmark the true market sentiment over certain periods.
portfolio is − 0.31%, thereby making the strategy economi-
cally insignificant. For USA, the sector rotation portfolio Sector rotation based on momentum and consumer
generates lower returns than the benchmark along with sentiment
higher standard deviation, resulting in a poor Sharpe ratio
over the sample period. However, Table 6 also suggests that Figures 1 and 2 illustrate the performance of the strategy
the sector rotation portfolio based on European sector ETFs in monetary easing and monetary tightening regimes for
is able to generate superior returns than the benchmark port- Europe and the USA, respectively. The portfolios for both
folio with significantly lower standard deviation, making the the markets have managed to capture the trend during mon-
sector rotation an economically significant venture. Higher etary tightening policy regime and have continued to out-
mean returns and lower standard deviation results in a net perform the market during those, especially for Europe. The
higher Sharpe ratio of 0.73 for the sector rotation portfolio, strategy apart from increasing the returns roughly twofold,
which is roughly over two times more than that of bench- also reduces the standard deviation of the returns which can
mark portfolio. be noticed from less sharp turns in graph for returns of the
The added advantage of sector rotation strategy based sector portfolio as compared to the benchmark portfolio. An
on monetary policy as a switching indicator is that it does interesting trend to observe here is that right after announce-
not require to rebalance portfolio frequently which further ment of monetary tightening, most of the times the returns
reduces transaction costs. During the sample period used in start to decline for both the markets and after an announce-
ment of monetary easing there is a trend of short-term
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

Fig. 2  The USA—momentum-

based sector rotation strategy
performance

Table 7  The US winner K=1 K=3 K=6 K = 12 S&P 500

portfolio
J=1 Monthly return 0% 0% 0% 0% 0%
Annual return 3% 4% 0% − 2% 4%
p value 41% 18% 91% 55% 16%
SD 4% 4% 3% 4% 4%
Sharpe ratio (monthly) 0.05 0.09 0.01 − 0.04 0.09
J=3 Monthly return 0% 0% 1% 1% 0%
Annual return 6% 5% 6% 6% 4%
p value 9% 12% 8% 9% 16%
SD 4% 4% 4% 5% 4%
Sharpe ratio (monthly) 0.11 0.10 0.11 0.11 0.09
J=6 Monthly return 0% 0% 0% 0% 0%
Annual return 6% 5% 6% 6% 4%
p value 9% 11% 10% 11% 16%
SD 4% 4% 4% 5% 4%
Sharpe ratio (monthly) 0.11 0.10 0.11 0.10 0.09
J = 12 Monthly return 0% 0% 0% 0% 0%
Annual return 5% 5% 3% 5% 4%
p value 13% 15% 35% 16% 16%
SD 4% 4% 4% 4% 4%
Sharpe ratio (monthly) 0.10 0.09 0.06 0.09 0.09

increase in the returns. These results are in line with the More specifically, according to Table 8, the returns of
study conducted by Jensen and Johnson (1995). These trends the US sector rotation strategy based on momentum are not
could be attributed to the psychological behaviour of inves- significantly different from zero for all 16 portfolios (J = 1, 3,
tors who take monetary easing as a sign of growth and mon- 6, 12 and K = 1, 3, 6, 12). In other words, momentum effect
etary tightening as a sign of decline in economy. was not significant for the sample period and sectors under
Tables 7 and 8 use t tests to analyse the returns of the win- consideration in this paper for the USA and in contrast it
ner portfolio strategies in the USA and Europe, respectively. suggested towards the existence of a contrarian effect. How-
The table represents the monthly return, annual return, p ever, Table 8 shows that the returns of European Sector rota-
value associated with single sample t test, standard deviation tion strategy based on momentum are significantly differ-
and Sharpe ratio. ent from zero for the majority of portfolios. In other words,
On the basis of the evidence reported in Tables 7 and momentum effect has shown statistical significance for the
8, the following conclusions can be drawn for the US and sample period and sectors under consideration in this paper
European sector rotation strategy based on momentum. for Europe and in contrast it favoured the non-existence of
 C. Alexiou, A. Tyagi

Table 8  European winner K=1 K=3 K=6 K = 12 Euro Stoxx 50

portfolio
J=1 Monthly return 1% 1% 1% 0% 1%
Annual return 8% 7% 11% − 5% 8%
p value 5% 6% 0% 8% 15%
SD 4% 5% 4% 3% 6%
Sharpe ratio (monthly) 0.14 0.13 0.21 − 0.12 0.10
J=3 Monthly return 1% 1% 1% 0% 1%
Annual return 10% 8% 7% 6% 8%
p value 1% 3% 5% 9% 15%
SD 4% 4% 4% 4% 6%
Sharpe ratio (monthly) 0.19 0.16 0.14 0.12 0.10
J=6 Monthly return 0% 0% 1% 1% 1%
Annual return 4% 6% 7% 9% 8%
p value 3% 10% 5% 2% 15%
SD 2% 4% 4% 4% 6%
Sharpe ratio (monthly) 0.16 0.12 0.14 0.17 0.10
J = 12 Monthly return 0% 1% 1% 1% 1%
Annual return 4% 7% 7% 8% 8%
p value 4% 5% 5% 4% 15%
SD 2% 4% 4% 4% 6%
Sharpe ratio (monthly) 0.15 0.14 0.14 0.15 0.10

contrarian effect in European market. For distinct values of Momentum effect in the USA: Table 7 suggests that all
J, the following results were found: of the possible combinations of J = 1 have resulted in the
least Sharpe ratio in comparison with the benchmark and all
1. The high p value for J = 1 (formation period) for both the other values of J, meaning that the momentum effect is the
USA and Europe is in line with Jegadeesh (1990) and weakest at 1-month interval. This is in contradiction to the
Lehmann (1990) which suggest the existence of strong results of Moskowitz and Grinblatt (1999) which could be
contrarian effect for short holding/formation period. The attributed to either the use of different time frames or the use
average return for the US sector-based strategy for J = 1 of different assets in the two studies. However, at 3-month
and K = 1, 3, 6, 12 is mere 1.5% further suggesting the and 6-month interval, momentum effect has resulted in a net
weak impact of momentum effect in short term. higher Sharpe ratio. The returns have increased by more than
2. For formation period “J” = 3, the momentum effect has 2% annually with almost similar standard deviation (vola-
been the most prominent among all the formation peri- tility) for J = 3 and J = 6 with respect to the benchmark. At
ods followed by “J” = 6 months for both the USA and 12-month time interval, the momentum effect can still be felt
Europe. (net increase of 1% returns annually then the benchmark) but
3. Almost all the portfolios generated positive average can be considered dissipating when compared to the momen-
returns for the strategy with the exception of J = 1 and tum effects at 3- and 6-month time intervals.
K = 12 portfolio which generated negative returns for Momentum effect in Europe: Table 8 reports that 15 out
both the markets. Further emphasising on the failure of of 16 combination of sector rotation strategies based on
short-term returns to capture momentum effect. momentum have been able to generate a higher Sharpe ratio
than the benchmark (Euro Stoxx 50); however, most of this
No statistically significant pattern in the volatility of the credit could be attributed to the ability of these strategies
returns (standard deviation) for neither the USA nor Europe to reduce volatility than their ability to generate superior
was observed during the analysis. The momentum effect was returns as only 4 out of 16 of these combinations were able
statistically strong in the European market in comparison to generate abnormal returns. This could potentially reflect
with the US market and therefore resulted in higher aver- the absence of momentum in the European market.
age returns for all the portfolios and most importantly also Figures 7, 8, 9, 10, 11, 12, 13 and 14 in “Appendix” illus-
generated lower volatility (standard deviation), thereby gen- trate the performance of all these momentum-based strate-
erating more Sharpe ratio for all the “J” & “K” combination gies (J = 1,3,6 & 12 and K = 1, 3, 6 and 12) over the sample
portfolios. period under consideration.
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

Table 9  Sector rotation USA Europe

portfolio versus benchmark
(1999–2019) Mean (%) SD (%) Sharpe ratio Mean (%) SD (%) Sharpe ratio

Portfolio 6.70 60.73 0.11 5.06 59.81 0.08

Benchmark 7.04 14.70 0.48 8.53 21.76 0.39
Excess return − 0.34 – – − 3.47 – –

Mean percentages denote annual returns

Tables 11 and 12 report the descriptive statistics of the Table 10  Rolling regression summary for the US sectors
sector portfolios monthly excess returns ­(Ritt − Rft) from Sector Sector ETF Mean alpha (%) SD (%)
1999 to 2019. Out of all the 9 sectors in the USA only Mate-
rials reflected positive skewness and only technology and Consumer discretionary XLY US equity −6 36
financials out of the 10 sectors in Europe reflected positive Consumer staples XLP US equity − 16 41
skewness. This signifies that the chances of sector returns Energy XLE US equity 15 42
taking extreme values are mostly towards the left tail. In the Financial XLF U S equity − 42 50
USA—Consumer discretionary exhibited the highest mean Health care XLV US equity 14 32
monthly excess return of 0.76%, while in Europe—Indus- Industrial XLI U S equity − 56 63
trials exhibited the highest mean monthly excess return of Materials XLB US equity − 22 40
0.62%. In both markets, the technology sector appeared to Technology XLK US equity − 18 22
be the most volatile sector between 1999 and 2019. Utilities XLU US equity − 20 61
In this paper, as we also purport to gain further insights Net − 17 43
into the impact of consumer sentiments, which is believed
to drive momentum, we utilise OECD’s consumer confi-
dence indicator (CCI) as a timing tool to switch between
Table 11  Rolling regression summary for the European sectors
sectors. The CCI indicator reflects the potential of peoples’
consumption and savings based on their expected financial Sector ETF Mean alpha (%) SD (%)
situation, general consumer sentiment regarding the future Consumer discretionary STR FP equity 25 106
of the market, level of employment and savings potential. A Consumer staples STS FP equity 45 51
score over 100 reflects a bullish attitude among consumers Energy STN FP equity − 46 101
which results in less savings and more expenditure. How- Financials STZ FP equity 51 106
ever, a value below 100 reflects bearish attitude among con- Health care STW FP equity 75 83
sumers which results in more savings and less expenditure. Industrials STQ FP equity 48 57
Below, we provide an analysis of the monthly returns of Technology STK FP equity 56 122
the sector-based ETFs for the USA and Europe both from Materials STP FP equity 16 49
1999 to 2019 and compare it with the monthly returns of Communication STT FP equity 14 78
their benchmarks, namely S&P 500 index and Stoxx, respec- Utilities STU FP equity 18 50
tively. To evaluate the effectiveness of consumer sentiment Net 30 80
as an indicator for sector rotation, we adopt the following
methodological framework. In this direction, we construct an
equity portfolio by a) investing in an equally weighted port-
folio comprising of all the cyclical sector ETFs in periods hence making the strategy economically insignificant.
where the CCI was over 100 and b) investing in an equally For the USA, the sector rotation portfolio generates lower
weighted portfolio comprising of all the defensive (non- returns than the benchmark along with higher standard
cyclical) sector ETFs during periods where the CCI was deviation, resulting in a poor Sharpe ratio over the sample
below 100. period. Further, Table 11 also shows that the sector rota-
Table 9 compares the mean return, standard deviation and tion portfolio based on European sector ETFs generated
Sharpe ratio of the sector rotation portfolio & benchmark for even poorer results with the benchmark having an excess
both the USA and Europe, respectively. return of 3.47% annually over the portfolio with signifi-
In particular, Table 9 suggests that the difference cantly higher standard deviation, making the sector rota-
between (annualised) mean return of the sector rotation tion an economically insignificant venture.
strategy portfolio and the benchmark portfolio is − 0.34%, In both the cases, the benchmark was able to outper-
form the portfolio and had relatively far lower volatility.
 C. Alexiou, A. Tyagi

This would lead us to infer that the CCI might not be an 2.25

accurate indicator for switching between sectors in any of 1.25

the two economies.
0.25

Sector rotation based on rolling regression -0.75

-1.75
Rolling regression with a 36-month window based on Fama

2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
and French five-factor model was executed for 9 sector ETFs Consumer Discretionary Consumer Staples Energy
for the USA and 10 European Sector ETFs over the sample
period. Alpha here represents the returns that remain unex- Fig. 3  36-month rolling alpha trend—USA
plained by the 5 factors used in Fama and French five-factor
model. Tables 11 and 12 summarise the annualised mean of
rolling five-factor sector wise alphas and their correspond-
ing standard deviations for USA and Europe, respectively. 2.5
The number of 36 month rolling regression-based alphas for
each sector in USA and Europe is 211 and 176, respectively.
Table 10 suggests that the highest mean alpha observed
in USA is for energy sector (15%) and healthcare sector -2.5

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018
(14%) while the lowest mean alpha is for industrial sector
(− 56%) and Financial sector (− 42%). Within the 9 sectors Consumer Discreonary Consumer Staples Energy
under observation in this section, the overall mean alpha is Financials Health Care Industrials

negative (− 17%). This finding is contrary to that of Dellva

et al. (2001) where the authors found a net positive alpha for Fig. 4  36 month rolling alpha trend—Europe
35 Fidelity sector funds in USA. However, the difference
in results may be potentially attributed to two facts, firstly
the use of Sector ETFs in this paper unlike the specialised period. Energy sector was recorded with the highest posi-
mutual funds (sector funds) used in Dellva et al. (2001) and tive alpha during 2001–2003 period and has been declin-
secondly to the different sample period used for analysis, ing ever after, up until recently. Utilities sector seems to
where the aforementioned authors used the sample period have a cyclical alpha during the sample period with alter-
from 1989 to 1998, whereas this paper focuses on sector nate periods of positive and negative alphas. Healthcare
performance from 1999 to 2019. sector has performed relatively well during the sample
However, these findings are similar to relatively recent period with only brief periods with negative alphas.
literature focusing on sector performance and sector selec- Figure 4 shreds some light on the relative performance
tivity like the findings of Sarwar et al. (2017) and Dou of sector ETFs in Europe. Technology sector has seen
et al. (2014), both of which reported a net positive alpha exceptional performance with the highest recorded posi-
for energy and healthcare sectors; and negative alphas for tive alphas between 2004 and 2005 and has performed
consumer staples and industrial sectors. well during the sample period except from 2005 to 2010.
Table 11 reflects that the net overall rolling five-factor Financials sector has performed worst from 2007 to 2009
alphas for European sectors is 30%. The highest observed which is consistent with the 2008 financial crisis and has
mean alpha is for health care (75%) and the lowest mean recovered slowly over the time. Similar to USA, Utilities
alpha is for energy (− 46%). The sectors in Europe in com- sector in Europe seems to have followed a cyclical pat-
parison with those in the USA have been successful in gen- tern with alternate periods of positive and negative alphas.
erating net positive alpha for the given sample period with Communication, technology and healthcare sectors have
9 out of 10 European sector ETFs with net positive alpha in performed well from 2010 onwards.
comparison with only 2 out of 9 sectors with positive alpha The results so far point towards the difference between
in the USA risk-adjusted performance of sector ETFs as per FF5 and
In order to further understand and illustrate the perfor- FF3 model. The difference further suggests that both the
mance of each of the analysed sectors over time, charts asset pricing model incorporates and present different
have been plotted for rolling Fama and French 5 factor information about the asset performance.
alpha for USA and Europe over the sample period. As it Tables 13 and 14 summarise the ordinary least square
can be seen clearly in Fig. 3, the industrial and technol- (OLS) regression results for FF5 and FF3 models for both
ogy sectors appear to have performed poorly throughout the US and European sectors, respectively, for the period
the sample period except for 2004–2005 and 2007–2008 under consideration from 1999 to 2019. The tables below
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

Table 12  OLS regression estimates—US sectors

Model Alpha Market SMB HML RMW CMA R2
Beta Beta Beta Beta Beta

Consumer 5 Factor 0.348 0.668*** − 0.197 − 0.100 − 0.159 − 0.637*** 50.60%

(0.237) (0.000) (0.185) (0.605) 0.481 (0.007)
Discretionary 3 Factor 0.123 0.762*** − 0.127 − 0.286** – – 48.63%
0.655 (0.000) 0.386 (0.043) – –
Consumer 5 Factor 0.247 0.460*** − 0.293*** − 0.382*** 0.590*** 0.052 38.68%
0.258 (0.000) (0.008) (0.008) (0.001) 0.765
Staples 3 Factor 0.508** 0.417*** − 0.272*** − 0.637*** – – 35.19%
(0.014) (0.000) (0.014) (0.000) – –
Energy 5 Factor − 0.289 0.556*** − 0.422*** 0.701*** 1.204*** − 0.289* 38.55%
(0.365) (0.000) (0.009) (0.001) (0.000) (0.056)
3 Factor 0.115 0.544*** − 0.273 − 0.033 – – 30.17%
(0.712) (0.000) (0.102) 0.835 – –
Financials 5 Factor 0.261 0.682*** − 0.538*** 0.705*** − 0.968*** − 1.206*** 75.28%
(0.312) (0.000) (0.000) (0.000) (0.000) (0.000)
3 Factor − 0.477* 0.905*** − 0.438*** 0.632*** – – 67.86%
(0.078) (0.000) (0.003) (0.000) – –
Health care 5 Factor 0.480* 0.302*** − 0.432*** − 0.552*** − 0.191 − 0.015 24.80%
(0.057) (0.000) (0.001) (0.001) (0.322) 0.938
3 Factor 0.366 0.321*** − 0.448*** − 0.501*** – – 24.54%
0.113 (0.000) (0.000) (0.000) – –
Industrials 5 Factor 0.329 0.762*** 0.088 − 0.134 − 0.254 − 0.469*** 60.63%
(0.209 (0.000) (0.508) (0.437) (0.208) (0.025)
3 Factor 0.116 0.837*** 0.135 − 0.197 – – 59.38%
(0.635) (0.000) (0.302) (0.115) – –
Technology 5 Factor 0.932** 0.777*** − 0.515*** − 1.139*** − 1.955*** − 0.478 54.92%
(0.015) (0.000) (0.008) (0.000) (0.000) (0.112)
3 Factor − 0.086 0.973*** − 0.563*** − 0.751*** – – 44.04%
(0.824) (0.000) (0.007) (0.000) – –
Materials 5 Factor 0.035 0.825*** − 0.009 0.385** 0.451** − 0.798*** 60.07%
0.904 (0.000) (0.950) (0.047) (0.046) (0.001)
3 Factor 0.058 0.897*** 0.125 − 0.116 – – 57.01%
(0.835) (0.000) (0.403) (0.418) – –
Communication 5 Factor 0.483* 0.433*** − 0.851*** − 0.503*** − 0.800*** − 0.119 44.99%
(0.086) (0.000) (0.000) (0.007) (0.000) 0.592
3 Factor 0.058 0.897*** 0.125 − 0.116 – – 40.64%
(0.835) (0.000) (0.403) (0.418) – –
Utilities 5 Factor 0.085 0.511*** − 0.451*** 0.361** 0.399** − 0.307 48.79%
(0.720) (0.000) (0.000) (0.022) (0.029) (0.103)
3 Factor 0.169 0.529*** − 0.388*** 0.042 – – 46.95%
(0.447) (0.000) (0.001) (0.709) – –

*, **, *** denote significance at 10%, 5% and 1% level; p values are given in parentheses

capture the R-square, alphas, coefficients of the factors represents the p value which reflects the statistical signifi-
associated with the three-factor and five-factor asset pric- cance of each factor.
ing models namely market factor, SMB factor, HML fac- The following inferences can be drawn from Table 12
tor, RMW factor and CMA factor. The value in parenthesis for the USA:
 C. Alexiou, A. Tyagi

Table 13  OLS regression estimates—European sectors

Model Alpha Market SMB HML RMW CMA R2
Beta Beta Beta Beta Beta

Consumer 5 Factor − 0.091 1.138*** 0.054 0.180*** 0.362*** − 0.092 79.46%

(0.572) (0.000) (0.338) (0.010) (0.000) (0.346)
Discretionary 3 Factor 0.077 1.065*** − 0.098* 0.248*** – – 77.41%
(0.633) (0.000) (0.055) (0.000) – –
Consumer 5 Factor − 0.286* 0.702*** − 0.297*** − 0.160** 0.427*** 0.569*** 58.52%
(0.064) (0.000) (0.000) (0.016) (0.000) (0.000)
Staples 3 Factor 0.059 0.525*** − 0.390*** 0.146*** – – 47.11%
(0.722) (0.000) (0.000) (0.007) – –
Energy 5 Factor − 0.242 1.018*** 0.079 0.205 0.319** 0.170 44.29%
(0.444) (0.000) (0.477) (0.132) (0.030) (0.372)
3 Factor − 0.034 0.922*** − 0.034 0.378*** – – 43.10%
(0.912) (0.000) (0.722) (0.000) – –
Financials 5 Factor − 0.236 1.132*** − 0.334*** 1.014*** − 0.233*** − 0.225** 85.37%
(0.139) (0.000) (0.000) (0.000) (0.002) (0.020)
3 Factor − 0.415*** 1.219*** − 0.274*** 0.785*** – – 84.63%
(0.008) (0.000) (0.000) (0.000) – –
Health care 5 Factor 0.069 0.794*** − 0.190*** − 0.110 0.123 0.789 63.39%
(0.680) (0.000) (0.001) (0.126) (0.111) (0.432)
3 Factor 0.147 0.755*** − 0.229*** − 0.080 – – 63.20%
(0.355) (0.000) (0.000) (0.119) – –
Industrials 5 Factor − 0.282* 1.195*** − 0.016 0.206*** 0.302 0.140 83.43%
(0.056) (0.000) (0.763) (0.001) (0.000) (0.114)
3 Factor − 0.094 1.106*** − 0.116** 0.344*** – – 82.05%
(0.520) (0.000) (0.012) (0.000) – –
Materials 5 Factor − 0.426* 1.276 0.256 0.178* 0.426*** 0.255* 70.09%
(0.060) (0.000) (0.747) (0.068) (0.000) (0.062)
3 Factor − 0.143 1.144*** − 0.121* 0.406*** – – 67.87%
(0.519) (0.000) (0.086) (0.000) – –
Technology 5 Factor 0.111 1.171*** − 0.157** − 0.454*** − 0.284*** − 0.329*** 85.11%
(0.530) (0.000) (0.012) (0.000) (0.001) (0.002)
3 Factor − 0.120 1.279*** − 0.062 − 0.700*** – – 83.81%
(0.493) (0.000) (0.260) (0.000) – –
Utilities 5 Factor − 0.185 0.581*** − 0.275*** − 0.058 0.330** 0.434*** 29.55%
(0.441) (0.000) (0.001) (0.574) (0.030) (0.003)
3 Factor 0.082 0.447*** − 0.360*** 0.164** – – 25.47%
(0.725) (0.000) (0.000) (0.032) – –

*, **, *** denote significance at 10%, 5% and 1% level; p values are given in parentheses

Table 14  The USA—trading Trading strategy Mean return (%) SD (%) Sharpe ratio
strategy performance
Long only based on five-factor alpha 8.20 13.70 0.34
Long & short based on five-factor alpha − 7.60 61.10 − 0.18
Long only based on three-factor alpha 8.10 13.80 0.33
Long & short based on three-factor alpha − 10.60 60.10 − 0.23
Buy & hold S&P 500 6.80 14.10 0.23
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

• Table 12 suggests that 3 out of 9 US sector alphas the sample period. This suggests that like in case of USA,
(financials, materials and technology) are statistically five-factor model is also better suited to explain Euro-
significant for five-factor model, at 90% significance pean sector returns from a statistical significance point
level. However, only 1 sector alpha (consumer staples) of view.
is statistically significant for three-factor model, at 99% • However, here also large number of statistically insignifi-
significance level. cant alphas suggests that market, size, value, profitability
• In case of five-factor model for USA, majority of the and investment factors are not enough to explain some
sectors show underperformance (negative alphas) after sector returns.
accounting for market, size, value, investment and profit-
ability risk factors except for energy sector and healthcare We have also conducted unpaired t test to check if the
sector. However, in case of three-factor model for USA, mean of Fama–French five-factor sector alphas for both the
after incorporating market, size and value risk twice the USA and Europe are statistically different across different
number of sectors had positive alphas representing out- sectors. On the basis of the results (not presented here for
performance but none of them was statistically signifi- economy of space) in the ten sectors under consideration
cant. Moreover, majority of the betas for both the models in the Europe, 42 out of 55 pairs were found to be signifi-
in USA were found to be statistically significant. cantly different from each other, at 5% level of significance,
• The market, size and value factors were found to be sta- whereas Table 4 indicates that in the nine sectors under con-
tistically more significant in case of five-factor model sideration in the USA, 38 out of 45 pairs were found to be
as compared to three-factor model. Also, the adjusted significantly different from each other, at 5% level of signifi-
R-square of five-factor model was found to be higher than cance. To further assess whether the profitability (RMW)
that of three-factor model for each of the ten sectors in and investment (CMA) factors in the five-factor model are
the sample period. This suggests that five-factor model is statistically significant, the redundant variable test has been
a better suited to explain US sector returns from a statisti- performed.1
cal significance point of view. According to the results of the likelihood ratio test for
• However, the insignificant alphas suggest that market, the US sectors, 8 out of 9 sector portfolios have statisti-
size, value, profitability and investment factors are not cally significant profitability coefficients (betas) except the
enough to explain some sector returns. healthcare sector, whereas only 5 out of 9 sector portfolios
have statistically significant investment coefficients (betas),
The following inferences can be drawn from Table 13 for at 10% significance level. However, for hypothesis three, 8
Europe: out of 9 sector portfolios have statistically significant both
profitability and investment coefficients (betas) with the only
• Table 13 reflects that 3 out of 10 European sector alphas exception of healthcare sector. For the European sectors, 7
(health care, technology and communication) are sta- out of 10 sectors have statistically significant profitability
tistically significant for five-factor model at 90%, 95% coefficients (betas), whereas 6 out of 10 sectors have sta-
and 90% significance level, respectively. However, only tistically significant investment coefficients (betas), at 10%
2 sector alphas (consumer staples and health care) are significance level. However, for hypothesis three, 9 out of
statistically significant for three-factor model at 95% and 10 sector portfolios have statistically significant both prof-
90% significance level, respectively. itability and investment coefficients (betas) with the only
• In case of five-factor model for Europe, majority of the exception of healthcare sector in Europe.
sectors show outperformance after adjusting for mar- The model specification tests conducted to analyse the
ket, size, value, investment and profitability risk factors “goodness of fit” of RMW and CMA factors with higher
except for energy sector. However, in case of three-
factor model for USA, after incorporating market, size
and value risk twice the number of sectors had negative 1
We used the likelihood ratio test (LRT) for the redundant variable
alphas (Financials and Technology) representing under- test. Likelihood ratio test (LRT) is a hypothesis test that assesses the
performance but neither of them was statistically signifi- “goodness of fit” between two nested statistical models on the basis
cant. Moreover, majority of the betas for both the models of their likelihoods. “Nested models” refers to the fact that one model
is a special case of another model. The primary assumption that goes
in Europe were found to be statistically significant. into this diagnostic test is that three-factor model is a part of five-fac-
• The market, size and value factors were found to be sta- tor model, and further hypothesise that the coefficient of the variable
tistically more significant in case of five-factor model under consideration is zero and therefore can be eliminated from the
as compared to three-factor model. Also, the adjusted equation. The test statistics for the null hypothesis used on the redun-
dant variables is standard F statistic based on a sample from the F
R-square of five-factor model was found to be higher than distribution. (For economy of space, the results are not presented here
that of three-factor model for each of the ten sectors in but are available upon request).
 C. Alexiou, A. Tyagi

Table 15  Europe—trading Trading strategy Mean return (%) SD (%) Sharpe ratio
strategy performance
Long only based on five-factor alpha 5.50 14.13 0.20
Long and short based on five-factor alpha − 26.45 94.80 − 0.31
Long only based on three-factor alpha 6.96 14.03 0.31
Long and short based on three-factor alpha − 30.69 92.15 − 0.36
Buy & hold S&P 500 5.52 22.06 0.13

Fig. 5  The USA—sector rota- The U.S. - Sector Rotation Strategies Performances
tion strategies performances 350

300

250

200

150

100

50

0
Jan-02

May-03
Jan-04

May-05
Jan-06

May-07
Jan-08

May-09
Jan-10

May-11
Jan-12

May-13
Jan-14

May-15
Jan-16

May-17
Jan-18

May-19
Sep-02

Sep-04

Sep-06

Sep-08

Sep-10

Sep-12

Sep-14

Sep-16

Sep-18
FF3 - Long FF3 - Long & Short FF5 - Long FF5 - Long & Short S&P

statistically significant factors and higher R-square values of and short strategy returns in USA has been around 60% and
five-factor model over the three-factor model suggests that for Europe it has been around 93%, which are substantially
five-factor model is better suited to explain sector returns in higher than the volatility of the returns of the benchmark for
both USA and Europe. These results are in coherence with both the markets.
previous study conducted by Sarwar et al. (2017) which also Table 14 reflects that the long only strategy based on
reported FF5 to be a better fit for the USA over FF3 model. FF3 and FF5 factor models for US sectors generated 1.3%
Tables 14 and 15 provide a summary of the results of long and 1.4% higher returns, respectively, than the S&P 500
and short and only long trading strategies based on FF5 and along with slightly lower standard deviations in each
FF3 factor models for the USA and Europe, respectively. case. Table 15 reflects that in Europe the long only strat-
The tables include annualised mean return, standard devia- egy based on FF3 factor model generated 1.44% higher
tion and Sharpe ratio for all the strategies. These tables can returns than the benchmark; however, FF5 factor model-
be used to draw comparisons among different strategies and based strategy underperformed the benchmark by 0.02%.
between same strategies in different markets. Moreover, they Meanwhile both the long only strategies were successful
suggest that both the FF5 and FF3 long only strategies were in minimising the volatility of the returns considerably
able to outperform the benchmark in Europe, but only long from 22.06% in the benchmark to just close to 14% in both
strategy based on FF3 was able to outperform the benchmark the strategies. This led to relatively higher Sharpe ratio
in the USA. However, interestingly the annualised returns for both the strategies in comparison with the benchmark,
of long only strategies by both FF3 and FF5 models do not almost double in case of FF3 long only strategy.
show significant difference in the profitability, despite higher The effective results of the strategy based on FF3 and
explanatory power and statistical significance of FF5 factor FF5 indicators over the buy-and-hold strategy of bench-
model in explaining sector returns. This finding is similar to mark portfolios for both the USA and Europe can be
Sarwar et al. (2017). observed in Figs. 5 and 6. These figures represent the
We can also observe the failure of both FF5 and FF3 fac- performance of all the strategies outlined in Tables 14
tor models alphas as indicators for sector rotation in case and 15 over the time period under consideration and also
of long and short strategy. Long strategy and short strategy display the cumulative growth in the returns over time
based on FF5 and FF3 for both the USA and Europe show in comparison with the benchmark portfolios. As can be
negative returns along with abnormally high standard devia- observed from both the figures, sector rotation strategies
tion leading to negative Sharpe ratio. The volatility in long based on FF3 and FF5 alphas have managed to consistently
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

Fig. 6  Europe—sector rotation Europe - Sector Rotation Strategies Performance

strategies performance 350

300

250

200

150

100

50

FF3 - Long FF3 - Long and Short FF5 - Long FF5 - Long and Short Stoxx

outperform the market, which is in line with the findings This paper examined multiple macroeconomic factors
by Sarwar et al. (2017). ranging from as simple as changes in interest rates to con-
sumer sentiment in an attempt to establish whether there
exists a “universal” switching indicator that can perhaps
Conclusions predict the right timing to switch between sectors. How-
ever, in most of the cases, the results for both the economies
As mentioned previously, the purpose of this paper is to have turned out to be significantly different for almost all the
determine the performance of sector rotation strategies switching indicators. This perhaps leads one to believe that
under a range of different factors to determine its viability there might not be a true “universal” indicator that predicts
and practicality. Three indicators were used to understand the right timing to switch between sectors. It does, however,
their respective impact within both the US and the European make one think that there may be a combination of several
market in an attempt to establish consistent results. It would factors that potentially indicate the right timing to switch. In
be unrealistic to express an opinion with strong conviction view of the above, we urge future researchers to build upon
that sector rotation strategies come without any limitations our work and investigate this prospect.
and will generate abnormal returns every time. However, The main conclusion, and in turn the potential ques-
there are indications implying that under certain circum- tions this paper draws is that under certain circumstances
stances, sector rotation strategy holds its ground and can be and specific factors, the potential to earn excessive return
considered a viable investment approach. using sector rotation strategies may in fact be quite plau-
The economic indicator, which in this case is the interest sible, thus proving its reliability. However, it is essential
rate, shows as per the results of the analysis that within the to further understand what these potential factors may be.
European market, sector rotation strategies tend to produce Further in-depth analysis within different markets, various
returns above the average benchmark, both during contrac- individual factors and other potential indicators is required
tionary and expansionary monetary policy regimes. A simi- to truly understand the effectiveness of sector rotation strate-
lar impact can be seen for the technical indicator (momen- gies. If anything, upon completion, this paper may enable the
tum), where the results show the presence of higher returns, researchers to appreciate the complexity behind appreciat-
while in the case of the performance indicator, excessive ing sector rotation strategies and develop further questions
returns within both the US and European markets are towards the direction of improving their practical reliability.
observed. Future directions as to how to improve or build upon the
Additional factors such as the volatility have also been existing work as well as find ways to circumvent the exist-
used for robustness. More specifically, the significance of ing challenges may be considered in the following context.
the technical indicator with regards to the US market can Firstly, the lack of availability of data of ETFs for both the
be considered statistically negligible. In this context, exces- USA and Europe presents a challenge as the data is only
sive returns generated within the European market under the available post 1999, and during most of the time after that
performance indicator may actually be only due to the reduc- till today both the US and European market have faced major
tion in volatility rather than an increase in the real returns economic downturns and have mostly been in either reces-
of the strategy. sion or recovering phase up until recently. This could lead
to the data being biased and influence the empirical results.
 C. Alexiou, A. Tyagi

Secondly, the lack of sector ETFs for emerging countries Table 17  The US sector rotation portfolio versus benchmark (1999–
restricted the study to only developed economies. The study 2019)—GDP
of sector rotation strategy via sector ETFs in emerging coun- Mean (annual SD (yearly) (%) Return to risk
tries would have presented a more holistic idea about the return) (%)
effectiveness of the sector rotation strategy. Lastly, incor-
Portfolio 7.05 71.05 0.12
porating transaction costs in a future research study might
Benchmark 7.04 14.70 0.48
provide further insights, hence exposing additional aspects
Excess return 0.01
of sectoral rotation when the allocation process of transac-
tion costs is considered.

Table 18  European sector rotation portfolio versus benchmark

Appendix (1999–2019)—GDP
Mean (annual SD (yearly) (%) Return to risk
See Tables 16, 17 and 18 and Figs. 7, 8, 9, 10, 11, 12, 13 return) (%)
and 14.
Portfolio 4.60 64.19 0.09
Benchmark 8.53 21.76 0.39
Excess return − 3.93

Table 16  Pictorial Jun-18

representation of sector rotation
strategy Jul-18
Aug-18
m
tu
en

Sep-18
m
om

tu
en

Oct-18
M

m
om

tu
en

Nov-18
M

m
om

tu
en

Dec-18
M

m
om

tu
ld

en
Ho

Jan-19
M

m
om

tu
ld
&

en
Ho

Feb-19
M
y
Bu

om
ld
&

Ho

Mar-19 M
y
Bu

ld
&

Ho

Apr-19
y
Bu

ld
&

Ho

May-19
y
Bu

ld
&

Ho

Jun-19
y
Bu

&

Jul-19
y
Bu
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

Fig. 7  Sector rotation perfor- 250

mance based on momentum in Rebalance - Monthly Rebalance Quarterly
the USA—1-month momentum Rebalance Half Yearly Rebalance Annually
200

150

100

50

0
1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018
Fig. 8  Sector rotation perfor- 300
Rebalance - Monthly Rebalance Quarterly
mance based on momentum in
Rebalance Half Yearly Rebalance Annually
the USA—3-month momentum 250

200

150

100

50

0
1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018
Fig. 9  Sector rotation perfor- 300
Rebalance - Monthly Rebalance Quarterly
mance based on momentum in Rebalance Half Yearly Rebalance Annually
the USA—6-month momentum 250

200

150

100

50

0
1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018
 C. Alexiou, A. Tyagi

Fig. 10  Sector rotation perfor- 250 Rebalance - Monthly Rebalance Quarterly

mance based on momentum in Rebalance Half Yearly Rebalance Annually
the USA—12-month momen-
200
tum

150

100

50

Fig. 11  Sector rotation perfor- Rebalance - Monthly Rebalance Quarterly Rebalance Half Yearly
mance based on momentum in
Rebalance Annually Euro STOXX 50
Europe—1-month momentum
600

500

400

300

200

100

Fig. 12  Sector rotation perfor- 450 Rebalance - Monthly Rebalance Quarterly Rebalance Half Yearly
mance based on momentum in
400 Rebalance Annually Euro STOXX 50
Europe—3-month momentum
350

300

250

200

150

100

50

0
Gauging the effectiveness of sector rotation strategies: evidence from the USA and Europe

Fig. 13  Sector rotation perfor- 400 Rebalance - Monthly Rebalance Quarterly Rebalance Half Yearly
mance based on momentum in
Rebalance Annually Euro STOXX 50
Europe—6-month momentum 350

300

250

200

150

100

50

0
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Fig. 14  Sector rotation perfor- Rebalance - Monthly Rebalance Quarterly Rebalance Half Yearly
mance based on momentum in 350
Rebalance Annually Euro STOXX 50
Europe—12-month momentum
300

250

200

150

100

50

0
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

References Conover, C.M., G. Jensen, R. Johnson, and M. Mercer. 2008. Sec-

tor Rotation and Monetary Conditions. Journal of Investing 28
(1): 34–46.
Aharoni, G., B. Grundy, and Q. Zeng. 2013. Stock Returns and the
Doeswijk, R., and P. van Vliet. 2010. Global Tactical Sector Alloca-
Miller Modigliani Valuation Formula: Revisiting the Fama French
tion: A Quantitative Approach, Available at SSRN: https​://ssrn.
Analysis. Journal of Financial Economics 110 (2): 347–357.
com/abstr​act=17424​22 or http://dx.doi.org/10.2139/ssrn.17424​
Beller, K., J. Kling, and M. Levinson. 1998. Are Industry Stock Returns
22.
Predictable? Financial Analysts Journal 54 (5): 42–57.
De Bondt, W., and R. Thaler. 1985. Does the Stock Market Over-
Bernanke, B., and K. Kuttner. 2005. What Explains the Stock Mar-
react? Journal of Finance 40: 793–805.
ket’s Reaction to Federal Reserve Policy? Journal of Finance 60:
Dellva, W.L., A.L. DeMaskey, and C. Smith. 2001. Selectivity and
1221–1257.
Market Timing Performance of Fidelity Sector Mutual Funds.
Carhart, M. 1997. On Persistence in Mutual Fund Performance. Jour-
The Financial Review 36 (1): 39–54.
nal of Finance 52: 57–82.
Dedola, L., and F. Lippi. 2005. The Monetary Transmission Mecha-
Cavaglia, S., and V. Moroz. 2003. Cross-Industry, Cross-Country
nism: Evidence from the Industries of Five OECD Countries.
Allocation. Financial Analysts Journal. https​://doi.org/10.2469/
European Economic Review 49 (6): 1543–1569.
faj.v58.n6.2488.
Dou, P.Y., D.R. Gallagher, D. Schneider, and T.S. Walter. 2014. Cross-
Chan, L.K.C., N. Jegadeesh, and J. Lakonishok. 1995. Momentum
Region and Cross Sector Asset Allocation with Regimes. Account-
Strategies, NBER Working Paper No. w5375.
ing and Finance 54: 809–846.
Chen, L., R. Novy-Marx, and L. Zhang. 2011. An Alternative Three-
Durham, B. 2005. More on Monetary Policy and Stock Price Returns.
Factor Model. Working Paper. https:​ //doi.org/10.2139/ssrn.14181​
Financial Analyst Jounral 61 (4): 83–90.
17.
Ehrmann, M., and M. Fratzscher. (2004) Taking Stock: Monetary Pol-
Chordia, T., and L. Shivakumar. 2002. Momentum, Business Cycle,
icy Transmission to Equity Markets, ECB Working Paper No. 354.
and Time-varying Expected Returns. Journal of Finance 57:
Fama, E., and K. French. 1993. Common Risk Factors in the Returns
985–1019.
on Stocks and Bonds. Journal of Financial Economics 33: 3–56.
Conover, M., G. Jensen, R. Johnson, and J. Mercer. 2005. Is Fed Policy
Fama, E., and K. French. 2015. A Five-Factor Asset Pricing Model.
Still Relevant for Investors? Financial Analysts Journal 61 (1):
Journal of Financial Economics 116: 1–22.
70–79.
 C. Alexiou, A. Tyagi

Grauer, R.R., N.H. Hakansson, and F.C. Shen. 1990. Industry Rotation Robert, L. 1967. Relative Strength as a Criterion for Investment
in the US Stock Market: 1934–1986 Returns on Passive, Semi- Selection. The Journal of Finance 22: 595–610. https​://doi.
passive, and Active Strategies. Journal of Banking and Finance org/10.1111/j.1540-6261.1967.tb002​95.x.
14 (2): 513–538. Sarwar, G., C. Mateus, and N. Todorovic. 2017. US Sector Rota-
Grinblatt, M., and S. Titman. 1989. Mutual Fund Performance: An tion with Five-Factor Fama-French Alphas, Available at SSRN:
Analysis of Quarterly Portfolio Holdings. The Journal of Business https​://ssrn.com/abstr​act=29878​19 or http://dx.doi.org/10.2139/
62 (3): 393–416. ssrn.29878​19.
Grinblatt, M., and S. Titman. 1992. The Persistence of Mutual Fund Scowcraft, A., and J. Sefton. 2005. Understanding Momentum. Finan-
Performance. The Journal of Finance 47 (5): 1977–1984. cial Analyst Journal 61 (2): 64–82.
Grundy, B.D., and M. Spencer. 2001. Understanding the Nature of the Siegel, F., and J. Hoban. 1991. Measuring Risk Aversion: Allocation,
Risks and the Source of the Rewards to Momentum Investing. Leverage, And Accumulation. The Journal of Financial Research
Review of Financial Studies 14: 29–78. 14 (1): 27–35.
Hayo, B., and B. Uhlenbrock. 1999. Industry Effects of Monetary Sorensen, E.H., and T. Burke. 1986. Portfolio Returns from Active
Policy in Germany, ZEI Working Paper, No. B 14-1999, Rhei- Industry Group Rotation. Financial Analysts Journal 42 (5):
nische Friedrich-Wilhelms-Universität Bonn, Zentrum für 43–50.
Europäische Integrationsforschung (ZEI), Bonn. Spiess, D.K., and J. Affleck-Graves. 1995. Underperformance in Long-
Jacquier, E., A. Kane, and A. Marcus. 2003. Geometric or Arith- Run Stock Returns Following Seasoned Equity Offerings. Journal
metic Mean: A Reconsideration. Financial Analysts Journal of Financial Economics 38 (3): 243–267.
59: 46–53. Jacobsen, B, J. Stangl, and N. Visaltanachoti. 2009. Sector Rotation
Jensen, G., and R. Johnson. 1995. Discount Rate Changes and over the Business Cycle. Working Paper, Massey University,
Security Returns in the U.S., 1962–1991. Journal of Banking Department of Finance and Economics.
& Finance 19: 79–95. Swinkels, L., and L. Tjong-A-Tjoe. 2006. Can Mutual Funds Time
Jensen, M., and G. Bennington. 1970. Random Walks and Techni- Investment Styles? Available at SSRN: https​://ssrn.com/abstr​
cal Theories: Some Additional Evidence. Journal of Finance act=10031​27.
25 (2): 469–482. Walkshäusl, C., and S. Lobe. 2014. The Alternative Three-Factor
Jegadeesh, N., and S. Titman. 1993. Returns to Buying Winners Model: An Alternative beyond US Markets? European Financial
and Selling Losers: Implications for Stock Market Efficiency. Management 20 (1): 33–70.
Journal of Finance 48: 65–91.
Jegadeesh, N. 1990. Evidence of Predictable Behavior of Security Publisher’s Note Springer Nature remains neutral with regard to
Returns. Journal of Finance 45: 881–898. jurisdictional claims in published maps and institutional affiliations.
Kacperczyk, M., C. Sialm, and L.U. Zheng. 2005. On the Industry
Concentration of Actively Managed Equity Mutual Funds. Jour-
nal of Finance 60 (4): 1983–2012.
Kouzmenko, R., and Z. Nagy. 2009. Sector Performance Across Dr. Constantinos Alexiou is a Professor at Cranfield School of Manage-
Business Cycles. MSCI Research Bulletin. ment where he delivers business-related courses across the full range
Lehmann, B.N. 1990. Fads, Martingales and Market Efficiency. of the School’s postgraduate and post-experience executive manage-
Quarterly Journal of Economics 105: 1–28. ment programmes. He has authored over 100 publications in academic
Luk, P. 2012. Is There Value in Asia Ex-Japan Sector Rotation journals such as the International Journal of Finance and Economics,
Strategies? Index Research & Design. https​://us.spind​ices.com/ Journal of World Business and International Business Review. He is a
docum​e nts/resea​r ch/resea​r ch-is-there​- value​- in-asia-ex-japan​ frequent contributor to many international conferences.
-secto​r-rotat​ion-strat​egies​-20121​0.pdf.
Moskowitz, T., and M. Grinblatt. 1999. Do Industries Explain Mr. Anshul Tyagi is an Analyst in Carlsquare, London. He advises cli-
Momentum? Journal of Finance 54: 1249–1290. ents (businesses, shareholders and financial investors) on the imple-
O’Neal, E. 2000. Industry Momentum and Sector Mutual Funds. mentation of strategic targets in the area of corporate finance/M&A
Financial Analysts Journal 56 (4): 37–49. as well as on how to secure capital for companies looking to scale up
Parvez, A., L. Lockwood, and N. Sudhir. 2002. Multi-style Rotation their businesses or finance structural investments to increase share-
Strategies. Journal of Portfolio Management 28: 17–29. holder value.
Rapach, D., J. Strauss, J. Tun, and G.Zhou. 2015. Industry Interdepend-
encies and Cross-Industry Return Predictability, Working Paper,
Research Collection Lee Kong Chian School of Business. https​://
ink.libra​ry.smu.edu.sg/lkcsb​_resea​rch/4515.