The Myth of Sector Rotation

Alexander Molchanov Jeffrey Stangl1

Abstract
Conventional wisdom suggests that sectors/industries provide systematic performance and that
business cycle rotation strategies generate excess market performance. However, we find no
evidence of systematic sector performance where popular belief anticipates it will occur. At
best, conventional sector rotation generates modest outperformance, which quickly diminishes
after allowing for transaction costs and incorrectly timing the business cycle. The results are
robust to alternative sector and business cycle definitions. We find that relaxing sector rotation
assumptions and letting any industry excess return predict future returns of other industries
results in predictability not significantly different than what would be expected by random
chance.

Keywords: Sector Rotation, Return Predictability, Investments.

JEL Codes: G11, G14

1 Alexander Molchanov (corresponding author).QB2.55, Massey University, Auckland, 0630, New Zealand. Phone: +64 9
414 0800 ext. 43152. E-mail: [email protected]

Jeffrey Stangl. Massey University, Auckland, New Zealand. Phone: +64 9 414 0800 ext. 43163. E-mail:
[email protected]

The authors would like to thank Ben Jacobsen for helpful comments on the earlier version of this paper.

1
 1. Introduction

Sector rotation refers to a common investment strategy that targets investments in particular

economic sectors at different stages of the business cycle. Bodie, Kane, and Marcus (2009)

suggest the “way that many [financial] analysts think about the relationship between industry

analysis and the business cycle is the notion of sector rotation.” Similarly, Lofthouse (2001)

states that financial analysts “think in terms of stylized economic cycles, with different sectors

performing at different stages of the cycle.” Fabozzi (2007, pg. 581) acknowledges, “Sector

rotation strategies have long played a key role in equity portfolio management.”

The seemingly mythical belief that tactically timing sector/industry investments generates

systematic excess returns persists unabated with certain investors, as supported by the media.

Popular investment websites (Investopedia, Stockcharts, and Seeking Alpha) detail the sector

rotation strategy, while providing examples of practical application. Any number of “How to

Guides”, starting with “Sector Investing” (1996) to “Trading for Dummies” (2013) also provide

step-by-step instruction on timing sector investments with business cycles. While the largest

investment companies (iShares, Vanguard, and Fidelity), provide a suite of sector funds that

facilitate sector rotation application. Several direct sector rotation funds are available, including

the Sector Rotation ETF (XRO), Line Industry Rotation Portfolio Fund (PYH), and Sector

Rotation Fund (NAVFX). However, comparing NAVFX returns since inception (2010-2018)

with the S&P 500 Index over the same period reveals roughly 5% underperformance (7.34%

versus 12.23%). Raising the question, does investor belief in sector rotation outperformance

represent a myth or reality?

Our study tests the two fundamental assumptions of sector rotation. Do certain sectors provide

systematic performance across business cycles? Does sector rotation generate excess market

performance? Bodie, Kane, and Marcus (2009) comment that “sector rotation, like any other

2
form of market timing, will be successful only if one anticipates the next stage of the business

cycle better than other investors.” This study overcomes the obstacle of correctly timing

business cycles with a simple and intuitive approach. That approach gives sector rotation

investors the benefit of the doubt, by assuming investors can perfectly time business cycle

turning points. If the business cycle drives sector returns, then an investor who perfectly times

business cycle stages and rotates sectors following popular belief on sector performance should

generate excess returns. Our analysis begins with the assumption of a sector rotation strategy

that follows conventional guidance on sector performance. However, we acknowledge many

potential versions of sector rotation strategy implementation. Consequently, we relax any

assumptions of a specific sector rotation model, testing performance of all sectors across all

business cycle stages.

Investors can choose to implement sector rotation at sector, industry, or firm level. The choice

depends on how precisely an investor wants to target expected sector performance and the

desired level of diversification. A common approach to sector rotation is industry-level

implementation. Industries allow a targeted approach to sector exposure, while still maintaining

the benefits of diversification. For instance, the healthcare sector includes pharmaceutical,

healthcare provider, and medical equipment industries. A sector rotation investor might

outweigh pharmaceuticals relative to other healthcare industries, based on a specific view of

expected industry performance. Our initial analysis focuses on the Fama and French 49

industry portfolios. Expanded robustness analysis considers alternative sector and industry

groupings.

The initial analysis documents sector rotation outperformance – but only marginally so. The

analysis investigates industry performance over 10 National Bureau of Economic Research

(NBER) dated business cycles from 1948 to 2018. The NBER defines only broad phases of

economic expansion and recession. The analysis first divides broad NBER phases into

3
additional sub-periods. We then map industries to business cycle stages where popular belief

anticipates optimal performance will occur. With few exceptions, industries expected to

perform well in various stages show no systematic performance. The analysis next combines

industries across stages to analyse whether conventional sector rotation generates

outperformance. Investors, guided by popular belief on sector performance and with perfect

foresight in timing business cycle stages, achieve a risk-adjusted return of 0.11 percent per

month before transaction costs. While this may seem high, a simple market timing strategy that

invests continuously in the market except during early recession generates a 0.15 percent

outperformance. With transaction costs, sector rotation performance quickly dissipates.

The results are robust to variety of tests and specifications. The analysis investigates whether

the results differ when investors anticipate business cycles early or late. Alternatively, we

examine business cycle stages delineated by the Chicago Federal Reserve National Activity

Index (CFNAI). When considering alternative sector and industry groupings, the results remain

unaffected. The main results are also robust to various performance measures such as the

Sharpe ratio and Jensen’s alpha. The results remain the same whether measured by a single

index, Fama and French three-factor, or Carhart four-factor model.

Lastly, the study generalizes the analysis to allow for all variations of sector rotation. The initial

analysis follows a commonly accepted version of sector rotation, as defined by Stovall (1996)

in Table 3 and illustrated by Standard & Poor’s in Figure 1. However, there are other potential

versions of sector rotation. The results are thus subject to the criticism of being limited to a

specific sector rotation model. To counter such criticism, the analysis tests for systematic

performance of any sector across any business-cycle stage. Measuring statistically significant

outperformance, the generalized results align with a hypothesis of neither systematic nor

persistent differences in sector returns across business-cycle stages. The significance levels

observed are only marginally different from those expected to occur randomly, without any

4
systematic outperformance. The result suggests that no sector rotation variant provides

systematic outperformance, questioning the popular belief that timing sector investments with

business cycles generates excess market performance.

This study contributes to the literature as the first to question the underlying assumptions of

sector rotation: systematic sector performance and the opportunity for investors to profitably

time sector rotation with the business cycle. Elton, Gruber, and Blake (2011) and Avramov and

Wermers (2006), suggest the importance that sector rotation plays in mutual fund performance.

Apart from a return predictability perspective, this study provides additional insights. Sector

rotation generates order flows, which transmit information about asset fundamentals. For

instance, Beber, Brandt, and Kavajecz (2010) provide evidence that sector-order flows forecast

macroeconomic conditions. The evidence suggests that sector-order flows, however, do not

translate into systematic sector performance. Evidence in Avramov and Wermers (2006) finds

that switching industry investments across business cycles drives equity fund performance.

Jiang, Yao, and Yu (2007), similarly, conclude that industry rotation underlies mutual fund

timing strategies, where fund managers switch between cyclical and non-cyclical stocks. A

natural question to ask is whether mutual funds follow conventional sector rotation or

alternative timing strategies. The results suggest that mutual funds profit from the latter. This

study contributes to a renewed interest in the literature on rotation strategies and industry

allocation, providing additional insight into these questions, among others.

2. Background and Hypotheses

One can dismiss, within the framework of rational expectations and the efficient market

hypothesis, the idea that investors systematically profit from sector rotation. Sector prices

should instantaneously reflect all available information and fundamental value – irrespective

of business-cycle stages. Yet, the prominence of sector rotation in practice suggests that

investors profit from timing systematic sector performance with the business cycle.

5
The apparent ability to profit from sector rotation might be consistent with the Hong and Stein

(1999) gradual information diffusion hypothesis. Gradual information diffusion, as Hong and

Stein (1999) describe, involves two groups of traders (news watchers and arbitrageurs) and the

lead-lag relation of their responses to economic news. News watchers have a limited ability to

process news and consequently revise asset prices with a delay. Arbitrageurs, in contrast, fully

incorporate news in their price adjustments and devise simple trading strategies that generate

excess returns. Analogously, one can view sector rotation investors as arbitrage traders who

respond to economic news by profitably timing sector rotation.

Hong, Torous, and Valkanov (2007) empirically test the gradual information diffusion

hypothesis with U.S. industries. They conjecture that economic news affect industry

fundamentals differently, and that the information content in the performance of certain

industries diffuses slowly across asset markets. Related literature documents differences in the

informational content of economic news, dependent on business cycle conditions. McQueen

and Roley (1993) find that the S&P 500 decreases in value with news of economic growth

when the economy is strong and increases in value when the economy is weak. Boyd, Hu, and

Jagannathan (2005) find that the impact of unemployment news on equity returns depends on

whether the economy is in a period of expansion or recession. The empirical evidence thus

shows that the effect of economic news on expected sector performance depends not only on

the sector but also on current business-cycle conditions.

Empirical research provides evidence that fund managers time their sector investments with

business cycles and that their order flows coincide with conventional sector rotation. Lynch,

Wachter, and Boudry (2004) also note that fund manager performance varies over business

cycles. Avramov and Wermers (2006) show that predictable variation in fund performance

relates to a manager’s skill in timing industry rotation with NBER business-cycle turning

points. Jiang, Yao, and Yu (2007) also observe that fund managers adjust industry allocations

6
based on common business cycle proxies. In a related study, Beber, Brandt, and Kavajecz

(2010) conclude that active order flows, defined as flows in excess of market capitalization,

directly link to economic news. Notably, for the motivation of this study, Beber, Brandt, and

Kavajecz (2010) observe that aggregate sector rebalancing emulates a conventional sector

rotation strategy, one that exploits the relative outperformance of certain sectors at different

business-cycle stages. Moreover, and of further interest for this study, they find institutional

order flows into certain sectors predict economic direction. For instance, order flows into the

basic materials sector predict economic expansion while order flows into the

telecommunication, consumer discretionary, and financial sectors predict economic

contraction. Such investment flows also coincide with popular belief on the sequence of sector

performance.

An empirical examination of cyclical sector performance is topical for both financial

researchers and investors. According to Hong and Stein (1999), informed arbitrage traders can

generate excess returns with simple trading strategies based on the release of economic news.

Sector- and industry-level investing also constitutes a dynamic growth segment in financial

markets. Cavaglia, Brightman, and Aked (2000) and Conover, Jensen, Johnson, and Mercer

(2008) document the increased importance of industry-level versus country-level investing.

Kacperczyk, Sialm, and Zheng (2005) find that active managers with concentrated industry

positions generate the greatest outperformance. From a practitioner’s perspective, the

widespread availability of sector funds and ETFs makes sector allocation strategies more

feasible than ever. Nonetheless, there is an apparent absence of empirical research on sector

performance over business cycles.

Related literature does describe the performance of alternative business-cycle timing strategies.

For instance, Siegel (1991) illustrates the potential of profitably timing allocations between

7
equities and cash. The author documents 12 percent annual market outperformance switching

between equity and cash at NBER business-cycle turning points. Brocato and Steed (1998)

similarly observe market outperformance rebalancing portfolios at NBER turning points.

Further, Levis and Liodakis (1999) and Ahmed, Lockwood, and Nanda (2002) report

outperformance to rotation strategies based on firm characteristics (such as earnings, value,

and capitalization) conditioned by well-known business-cycle variables. Conover, Jensen,

Johnson, and Mercer (2008) show 3.4 percent annual outperformance to a strategy that times

investments in cyclical and non-cyclical stocks with Federal Reserve monetary policy. While

closely related, this study fundamentally differs from previous research, by thoroughly

analyzing sector and industry performance across different measures of business-cycle stages.

Additionally, as Fama and French (1997) and Lochstoer (2009) identify time-variant industry-

risk premiums related to business cycles, this study also evaluates industry performance using

different risk correction measures. Some more recent work on sector rotation profitability

includes Sarwar et al. (2018), Guris and Pala (2014), Chava et al. (2018).

The above discussion leads to a formal statement of this study’s null and alternative hypotheses.

H0: Industry returns are unrelated to the stage of a business cycle stage.

H1: There is a systematic relationship between industry performance and stages of the business

cycle.

H2: Rotating sector investments with business cycle stages generates systematic excess returns.

Answering these hypotheses tests the fundamental assumption of sector rotation investors that

timing industry allocations with the business cycle is a profitable investment strategy.

8
 3. Business Cycles

3.1.NBER business cycle dates

Our analysis covers 10 business cycles from January 1948 to July 2018.2 The official U.S.

Government agency responsible for dating business cycles is the NBER. While academics and

practitioners widely accept NBER cycle reference dates, other business-cycle measures are also

available.3 The NBER dates cycle peaks and cycle troughs that broadly define phases of

economic expansion and economic recession. Panel A of Table 1 reports business cycle

durations from business cycle peak to business cycle peak. The sample covers the 10 business

cycles enumerated in the far left column of Panel A. Each business cycle spans the first month

following a peak to the subsequent peak. Business cycles average 70 months over the sample.

Earlier business cycle durations are much shorter than recent cycles, particularly during phases

of economic expansion.4

3.2.Business cycle stages

While the NBER defines broad economic phases, researchers and investment practitioners

commonly divide expansions and recessions into more discrete stages. Investment

professionals and practitioner guides, such as Stovall (1996), commonly divide expansions into

three equal stages (early/middle/late) and recessions into two equal stages (early/late). Three

stages of expansion allow for the longer duration of expansions relative to recessions. Other

research, such as DeStefano (2004), divides both expansions and recessions into two equal

stages. Our analysis evaluates sector/industry performance across five business cycle stages.

Subsequent analysis further evaluates performance across two-stage and four-stage business

cycle partitions.

2 Eleventh business cycle is still ongoing, with the latest peak recorded on December 2017. However, the 11 th business cycle
already has all five stages – 9 months each of early and late contractions, 40 each of early and middle expansions, and 29 of
late expansion.
3 For a survey of business cycle dating methodologies, see Cover and Pecorino (2005).
4 Moore (1974) provides a detailed discussion of post-1948 differences in business cycle dynamics.

9
The analysis measures expansions from the first month following a cycle trough to the

subsequent cycle peak and recessions from the first month following a cycle peak to the

subsequent cycle trough. The analysis also delineates three equal stages of expansion and two

equal stages of recession. The five business cycle stages are early expansion (Stage I), middle

expansion (Stage II), late expansion (Stage III), early recession (Stage IV), and late recession

(Stage V). Panel B of Table 1 reports the duration of expansions, recessions, and stages over

10 business cycles occurring from 1948 to 2018. Recessions average approximately 10 months

and expansions approximately five years.

3.3. Evaluation of business cycle proxies

The analysis first investigates whether the five NBER delineated stages are consistent with

well-known business cycle proxies. The common business cycle proxies (BCP) in the literature

are term-spread, default-spread, dividend yield, unemployment, and industrial production.

Studies by Keim and Stambaugh (1986), Chen, Roll, and Ross (1986), Fama and French

(1989), Schwert (1990), Campbell (1987), Chen (1991), Jensen, Mercer, and Johnson (1996),

and Petkova (2006), among others, document the relation between these proxies and business-

cycle conditions.

Panel A of Table 2 provides a summary of expected business cycle proxy changes over the five

NBER delineated stages. For instance, term-spread, default-spread, and dividend yield are

smallest near economic peaks and largest near economic troughs (Fama and French (1989)).5

The expectation is that these variables will decrease across early, middle, and late stages of

expansion. Conversely, these same variables should increase across stages of early and late

recession. Other studies, such as Balvers, Cosimano, and McDonald (1990) and Chen (1991),

document a close link between business cycles and both unemployment rates and industrial

5 The term-spread, default-spread, and dividend yield data come from http://www.globalfinancialdata.com

10
production. Stock and Watson (1999) and Hamilton and Lin (1996) show, for example, that

industrial production peaks and unemployment rates bottom out as the economy enters

recession. Industrial production should increase across successive stages of expansion and

decrease across successive stages of recession. Conversely, unemployment rates should

decrease across early, middle, and late expansion, then increase across early and late recession.

Panel B of Table 3 reports proxy averages by business-cycle stage estimated with Equation 1,

where Ds is a dummy variable that takes the value of one or zero dependent on the current

business cycle stage.

5
BCPt   s Ds ,t  t (Eq. 1)
s1

Next, the table reports changes in business-cycle proxy values (s-s-1) between successive

business-cycle stages. Panel B establishes that changes in the selected business-cycle variables

track NBER delineated business-cycle stages and show the mostly expected signs as reported

in Panel A. For instance, the results should indicate a significantly negative default-spread

difference between early expansion and late recession. The analysis tests for statistical

significance using a simple difference in means test. Panel B reports p-values under the null

hypothesis of no difference in business-cycle proxies across successive stages, formally stated

as Ho: s = s-1. Failure to reject the null would indicate no statistically significant difference in

the business-cycle proxy across successive stages and would invalidate the stage delineations.

For example, there is an average -0.5% difference between early expansion and late recession.

The results document that changes in the business-cycle proxies across successive business-

cycle stages, with few exceptions, have the expected sign and are highly significant.

11
 4. Industry Performance across Business Cycles

4.1. Data Description

Monthly market, industry, and Treasury bill return data come from the Kenneth French

website. Market returns represent the total value weighted returns for all NYSE, AMEX, and

NASDAQ listed stocks. The analysis initially uses the Fama and French 49 industry portfolios.

Fama and French map firms to industry groupings based on their standard industrial

classification (SIC).6 Firms mapped to the “other” industry come from a variety of sectors and

industries. As such, the “other” industry holds no relevance in a sector rotation strategy.

Consequently, the analysis omits the “other” industry, leaving 48 of the original Fama and

French 49 industries.7 The one-month Treasury bill serves as a proxy for the risk-free interest

rate.

4.2. Popular guide on industry performance

Table 3 shows the particular stage of the business cycle where popular belief anticipates

industries will perform best. We follow the popular Stovall (1996) practitioner guide to sector

investing. Stovall (1996) divides all equities into 10 basic sectors. He then maps sectors and

sub-sector industry groups to one of five business cycle stages.8 For example, Stovall suggests

that the technology and transportation sectors provide early expansion performance, basic

materials and capital goods provide middle expansion performance, and so forth. As Table 3

illustrates, there are four technology sub-sector industries and two transportation sub-sector

industries. Conventional guidance suggests each industry in those sectors provides early

recession performance. Performance then shifts from sector to sector across business-cycle

6 See http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html for further detail on the data and the

formation of industry portfolios.
7 The “other” industry group represents approximately 3.5 percent of total firms listed on NYSE, AMEX, and NASDAQ.
8 Lofthouse (2001) traces a similar approach of mapping sectors to stylized stages of economic cycles back to Markese

(1986). There are also different variants of mapping sector performance to business-cycle stages. Salsman (1997) uses
dividend yield, short-term interest rates, and precious metal prices to map sector performance. The present study concludes
with the total relaxation of any assumed sector rotation model.

12
stages. The analysis maps each of the 48 industry portfolios to a corresponding sector, then

maps each sector to the business-cycle stage of anticipated sector performance.

4.3. Nominal industry performance

Table 4 provides industry descriptive industry statistics and nominal performance for the

business-cycle stage popular belief anticipates outperformance will occur. The table reports the

average number of firms, number of observations, mean returns, standard deviation of returns,

and single-index betas by the indicated stage. For comparison, Table 4 reports mean returns,

standard deviation of returns, and single-index betas for the full 1948–2018 sample. The table

also reports industry averages and market statistics beneath each business-cycle stage.

The second column of Table 4 reports the average number of firms in an industry.

Implementing sector rotation at industry level allows for more precise targeting of

performance. The wide variety of available industry funds and ETFs reflects the popularity of

industry-level investing. The increased precision targeting industry versus sector performance,

however, comes at the cost of reduced diversification benefits. The defense, tobacco, and coal

industries, for instance, comprise on average fewer than 10 firms. As such, investments in those

industries are subject to a high level of firm-specific risk. It is unlikely, however, that sector

rotation investors would invest in only one industry during a particular business-cycle stage.

For example, there are 12 industries, including defense, expected to provide middle expansion

performance. Overall, conventional sector rotation investors would thus hold a well-diversified

middle expansion portfolio.9

We initially measure nominal industry performance to determine whether significant

differences occur over business cycles. The analysis then observes whether industry

9For an overview of tradeoffs in implementing sector rotation strategies at sector, industries, and firm levels, see
http://us.ishares.com/portfolio_strategies/investment_strategies/sector_strategies.htm

13
performance coincides with popular belief. Computer software, for instance, should provide

early expansion performance and basic materials should provide middle expansion

performance. Table 4 also reports p-values from a Wald test under the null hypothesis that

industry returns are not significantly different across business-cycle stages. However, in most

cases, the p-values reject the null, indicating that industry performance varies across business-

cycle stages. Sector rotation investors would find this initial result encouraging. Failure to

reject the null hypothesis of equal returns would question the basic premise of sector rotation

from the start.

Table 4 also reports average market returns beneath each business cycle stage. The analysis

compares industry and market returns to provide a simple relative return metric. As an example,

Table 4 reports transportation industry returns 1.84 percent, compared with 1.30 percent

average monthly market returns for early expansion. The transportation industry thus provides

market outperformance, where conventional wisdom expects. However, the realization of

expected outperformance does not always occur. Out of the 48 industries, 29 have nominal

returns higher than market returns, in the stage of expected outperformance. Thus, 60 percent

of industries offer the expected higher nominal performance. Market outperformance,

however, comes at a price. All but one industry (communications) has higher return volatility

than the market. Observing average industry performance for two stages reveals surprising

results. The 1.12 percent average return for industries expected to perform well in early

expansion actually underperforms the market by 0.18 percent. Similarly, average returns for

industries expected to perform well in middle expansion earn 0.13 percent less than the market.

Based on the initial results, popular belief holds true in the remaining three stages. Industries

on average outperform the market, as expected, in late expansion, early recession, and late

14
recession. Nominal sector performance coincides only partially with popular expectations.

Moreover, industry standard deviations and betas indicate that risk-adjusted performance will

coincide even less with popular expectations. For instance, in early and middle expansion,

average industry underperformance coincides with average standard deviations higher than the

market.

The nominal industry performance results are not encouraging for sector rotation investors.

The next section investigates whether industries provide systematic risk-adjusted business-

cycle performance.

4.4. Risk-adjusted industry performance measures

Table 5 reports industry excess market returns, Jensen’s alphas, Fama and French (1992) three-

factor alphas, and Carhart (1997) four-factor alphas by business-cycle stage. The table reports

performance alphas estimated with Equations 2 to 5.

Equation 2 estimates excess market industry performance (m), with a regression of excess

market industry returns (ri-rm) on the five business-cycle dummy variables (Ds). The regression

coefficient  ism measures market outperformance for industry i during business cycle stage s.

The results show that 7 of 48 industries, approximately one in seven, generate statistically

significant excess market performance when expected. More than half of the significant excess

market performance occurs in early and late recession. Notably, excess market performance

comes before any adjustment for systematic exposure to known sources of risk, which the

following analysis takes into account next.

5
rit  rmt    ism Dst   it (Eq. 2)
s 1

15
Equation 3 estimates a Jensen’s alphas (J) attributable to each business-cycle stage with a

modified single-index model.

5 5
rit  rft    isJ Dst    m,is (rmt  rft ) Dst   it (Eq. 3)
s 1 s 1

Equation 3 runs a regression of industry returns in excess of the one-month Treasury bill (ri-rf)

on one of five business-cycle timing variables (Ds) and the conditional market risk premium

(rm-rf). The Fama and French market index represents the market proxy.

To ensure the results do not depend on exposure to other well-known risk factors, the analysis

also estimates Fama and French three-factor alphas and Carhart four-factor alphas. The Fama

and French alphas (F), estimated with Equation 4, control for size and value risk factors in

addition to market risk. Lastly, the Carhart four-factor alphas (C), estimated with Equation 5,

add a momentum factor to the Fama and French three-factor model.

5 5
rit rft  isF Dst ism(rmt rft )  issSMBt  isvHMLt Dst it (Eq. 4)
s1 s1

5 5
rit  rf t    isC Dst     ism (rmt  rf t )  iss SMBt  isv HMLt  isc MOM t Dst   it (Eq. 5)
s 1 s 1

Regardless of the risk-adjusted alpha performance measure, there is scant evidence of

statistically significant industry outperformance where popular belief would suggest. The

performance results strengthen the earlier findings reported for nominal returns. Based on

Jensen’s alphas, there are four industries with significant outperformance. Based on the Fama

and French three-factor model, and using the Carhart four-factor model, there are no industries

with significant outperformance – only the ones with significant underperformance.

16
 5. Sector Rotation Performance

Can conventional sector rotation still be profitable, despite limited evidence of systematic

industry performance? This section focuses on strategy implementation, observing the

performance of sector rotation across the last 10 business cycles. The strategy assumes

investors perfectly time NBER business-cycle stages and rotate the 48 Fama-French industries

following the conventional sector rotation strategy and compares the result with a simple

market investment. Panel A of Table 6 provides mean monthly returns, as well as strategy

Sharpe ratios and standard deviations.

Sector rotation outperformance amounts to an average of 0.11% per month, which, at first

glance appears economically large. However, in perspective, this number presents the

maximum outperformance. Only the investors who followed popular market wisdom over the

last 70 years, ignored transaction costs, and perfectly timed the last 10 business cycles would

have realized 0.11% per month outperformance. It is also important to note that sector rotation

strategy has a higher standard deviation, higher beta, and lower Sharpe ratio than the simple

market portfolio.

Siegel (1991) suggests a simpler market timing strategy, showing that shifting between equities

and cash at business cycle turning points generates significant outperformance. However,

Siegel (1991) also recognizes the difficulty in correctly timing business cycles. To provide

perspective on sector rotation outperformance, the results also report the performance of the

simpler market-timing strategy suggested by Siegel (1991). Here, the analysis assumes a

theoretical investor who correctly times NBER recessions and expansions. Such an investor,

shifting from equities to cash early recession and back to equities late recession, would have

realized 0.15 percent average monthly outperformance. That same investor would have also

held a more diversified market portfolio, subject to less industry-specific risk.

17
Under a more realistic assumption of transaction costs, the results for sector rotation strategy

become even bleaker. Transaction costs, both explicit and implicit, are difficult to estimate

precisely. Estimated transaction costs include commissions, bid-ask spread, and market impact.

Actual costs depend on the stock, where it trades, and when it trades.10 Estimates vary

considerably and change over the sample.11 As an allowance, we estimate transaction costs that

range between 0.5 and 1.5 percent. Sector rotation has 53 round-trip transactions and the

market-timing strategy has 21 round-trip transactions from 1948 to 2018. With the inclusion

of transaction costs, the base-case sector rotation outperformance decreases to between 0.07 to

0.01 percent and becomes statistically indistinguishable from zero. The alternative market-

timing strategy increases in relative outperformance, owing to fewer transactions.

Thus far, the results indicate only marginal sector rotation outperformance for sector rotation

implemented in accordance with popular wisdom, even if one assumes investors can correctly

time business cycles. Results for industries expected to perform well in early expansion and

middle expansion are particularly disappointing. Still, it would be premature to conclude that

sector rotation does not work. Investors may use different industry or sector classifications,

different business-cycle indicators, or different business-cycle stages. Alternatively, investors

may time business cycles in advance or with a delay, which could generate outperformance.

The robustness tests also investigate whether the results improve if investors anticipate changes

in business-cycle turning points earlier or later. In addition to NBER business cycles, the

analysis tests business-cycle stages constructed from the CFNAI. The analysis concludes with

10See for example Goyenko, Holden, and Trzcinka (2009) and Hasbrouck (2009).
11Estimates of total trading costs vary greatly depending on the study. For instance, Lesmond, Schill, and Zhou (2004)
estimate round-trip transaction costs of 1 to 2 percent for most large-cap trades while Keim and Madhavan (1998) estimate
total round-trip transaction costs as low as 0.2 percent.

18
the total relaxation of any specific sector rotation model, testing for the systematic performance

of any sector across any business-cycle stage.

6. Robustness Checks

The analysis thus far has focused on a fairly specific version of a sector rotation strategy – a

five stage, 48-industry model based on Stovall’s (1996) rotation logic. While this particular

model is widely used, it is one of potentially thousands of sector rotation models available for

an investor. We now gradually relax the assumptions. We start by considering alternative

industry groupings. We then consider alternative business cycle stage delineations, an

alternative way to measure the business cycle, as well as timing the cycle in advance or with a

delay. We then deviate from Stovall’s model and consider every possible form of a sector

rotation strategy. We then relax the assumptions even further, looking if any industry’s

performance can act as a predictor of any other industry’s returns irrespective of the business

cycle.

6.1. Alternative sector/industry groups

There are alternative sector and industry classifications available to sector rotation investors.

As such, our analysis might merely reflect a particular industry grouping. The following

analysis investigates the performance of two alternative sector and industry groups. The

analysis maps the original Fama and French 49 industries to 10 sector portfolios and 23 major

industry portfolios, as listed in Table A1. The 10 sector portfolios are constructed following

the Kacperczyk, Sialm, and Zheng (2005) mapping of the Fama and French 48 portfolios. The

additional computer software industry included in the Fama and French 49 industry portfolios

goes into the business equipment and services sector. Additionally, the analysis maps the Fama

and French 49 industries to one of 23 GICS major industry groups. The Global Industry

Classification Standard (GICS), first introduced in 1999, provides a widely accepted alternative

19
to SIC classifications.12 Bhojraj, Lee, and Oler (2003) report GICS classifications are superior

to alternative classification schemes.

The results are presented in Panel B of Table 6. Both 10-sector and 23-industry groupings

generate similar mean monthly sector rotation returns to the ones generated by 49 Fama-French

industries (1.03% and 0.99% vs. 1.00%). Neither grouping generates mean returns higher than

a simple market timing strategy. Sector rotation performance based on alternative industry

groupings is also inferior to market timing in terms of volatility, beta, and Sharpe ratio. This

leads us to believe that our results are not driven by a particular industry classification13.

6.2. Alternative business cycle stage delineation

Arguably, business cycle stage delineations are arbitrary. Although the five-stage analysis

follows a common approach, one can potentially construct any number of business cycle

partitions. As a result, the base-case results face criticism that they are specific to particular

delineation of business cycle stages. The NBER officially dates the U.S. business cycle peaks

and troughs, delineating one stage of expansion and one stage of contraction. DeStefano (2004)

further separates the NBER stages of expansion and contraction into two equal halves, four

stages in all. The following analysis considers both NBER two-stage and DeStefano (2004)

four-stage partitions, to verify that the results are robust to alternative business cycle stage

definitions. The two-stage analysis uses NBER cycle dates to delineate one stage of expansion

and one stage of recession. The two-stage analysis maps early, middle, and late expansion

industries into one stage expansion, and early and late recession industries into one stage of

recession. The four-stage analysis further divides expansions and recessions in halves. Early

12For details, see http://www2.standardandpoors.com/spf/pdf/index/GICS_methodology.pdf

13We have also produced tables similar to that of Tables 4 and 5 (descriptive statistics and risk adjusted performance
measures) for alternative industry classifications. The results are equally unimpressive and are not reported to save space.
They are available from the corresponding author upon request.

20
and late expansion stages last on average 302 and 314 months. Early and late recessions last

on average 53 and 51 months, identical to the five-stage analysis.

The results for sector rotation strategy performance based on two- and four-stage business cycle

delineations are reported in Panel C of Table 6.14 The strategy based on two-stage delineation

underperforms the market portfolio across all dimensions. Although the strategy based on four-

stage delineation marginally outperforms the five-stage strategy, it is still inferior to the market

timing strategy reported in Panel A, having lower outperformance (0.12% vs. 0.15%), higher

standard deviation (4.85% vs. 3.99%), higher beta (1.04 vs. 0.89), and lower Sharpe ratio (0.21

vs. 0.26). Overall, alternative specifications of business cycle partitions provide no

improvement on the base case and the previous results continue to hold.

6.3. Alternative way to measure the business cycle

This section considers the Chicago Federal Reserve National Activity Index (CFNAI) and

Conference Board Leading Indicator as alternatives to NBER cycle dates. As results for these

two indicators are similar, the analysis focuses on the CFNAI.15 In contrast to static NBER

defined phases of expansion or recession, the CFNAI provides a continuous measure of

business cycle conditions. The CFNAI incorporates 85 economic variables that cover four

broad categories: production and income; employment, unemployment, and hours; personal

consumption and housing; and sales, orders, and inventories. CFNAI construction follows the

methodology of Stock and Watson (1989), who create an index based on the first principal

components of a large number of variables that track economic activity. By construction, the

CFNAI has a zero mean and unit standard deviation. Positive (negative) CFNAI values indicate

14 Just as with alternative industry groupings, we have produced industry descriptive statistics and risk-adjusted performance
measures. Just as with base-case results, we find very limited evidence of systematic industry outperformance across
business cycles. The results are not reported to save space. They are available from the corresponding author upon request.
15 The CFNAI and detrended Conference Board leading indicator have a 78 percent correlation coefficient. Both indices thus

reveal similar business cycle information. The study focuses on the CFNAI because it is freely available to the public and
released monthly by the Chicago Federal Reserve Bank.

21
above (below) trend economic activity. Publication of the CFNAI began in 2001 with data

available from 1967.16 Figure 4 overlays the CFNAI on NBER delineated phases of economic

expansion and contraction (shaded area). The CFNAI closely tracks NBER cycle dates, with

some variation. The variation may better reflect investor uncertainty when attempting to

pinpoint real-time changes in business-cycle stages.

The analysis partitions CFNAI business cycles into five equal stages. CFNAI values of 0.702,

0.312, -.0113, and -.637 delineate stages of early expansion through late recession. We then

proceed as with the five-stage NBER business cycle delineation. Sector rotation strategy results

are presented in the last line of Table 6 (Panel C). Such a strategy underperforms across the

board. All performance characteristics are inferior to market, market timing, as well as all

previously reported sector rotation strategies.

6.4. Timing the business cycle in advance or with a delay

Investors might profit from consistently timing the business cycle incorrectly. Suppose that

investors consistently assume that turning points occur earlier or with a delay from actual

NBER business cycle dates. If so, the base-case scenario might underestimate actual sector

rotation outperformance. To explore that possibility, the analysis advances the implementation

of sector rotation by one month, two months, and three months prior to NBER business-cycle

turning points. Similarly, the analysis considers delays from one to three months. Table 7

presents results before transactions costs.

There appears to be some benefit to anticipating business cycle one and two months in advance

when it comes to sector rotation strategy. However, (1) the improvement is very marginal and

(2) strategy performance remains inferior to that of simple market timing.

16 More information is available at http://www.chicagofed.org/economic_research_and_data/cfnai.cfm

22
6.5. Analyzing all possible sector rotation strategies
While the preceding robustness checks have relaxed a number of assumptions, the basic model

is still based on the one described in Stovall (1996). Conventional sector rotation presupposes

the sequential performance of sectors across business cycle stages. For instance, Standard &

Poor’s sequencing in Figure 1 shows that performance in the technology sector follows the

performance in the financial sector, which in turn follows performance in the utilities sector.

Figure 1 further illustrates other representative sequential patterns of sector performance.

While it depicts largely congruent beliefs on sequential sector performance, other variations

are possible. After all, throughout the analysis we have assumed that agricultural sector,

however defined, performs better in expansions, however defined. This assumption is

reasonable, but it is an assumption nonetheless. One could come up with a plausible argument

that agricultural sector should outperform in recession.

We now explicitly address this by analyzing every possible combination of sector rotation

strategy using a 10-sector industry definition and a two-stage business cycle partition. This

gives us 1,022 possible strategies. Return distribution of all of the possible strategies are

presented in Figure 4.

The results provide even more discouragement for a potential sector rotation investor. Average

return of these strategies is 0.86% per month, actually lower than that of a buy-and-hold

strategy of 0.89% (coincidentally, a two-stage partitioning model based on Stovall (1996) also

provides a raw return of 0.89%). A market-timing strategy based on a two-stage business cycle

partition (invest in an index during booms and in T-bills during recessions) yields an average

monthly return of 0.91%

Not surprisingly, some sector rotation strategies outperform both a buy-and-hold and a market

timing strategy. Only 35 out of 1,022 strategies (3.4%) outperform the market timing strategy,

and only 132 out of 1,022 (12.9%) outperform the buy-and-hold. Accounting for transaction

23
costs will make these strategies even less attractive. The mean of sector rotation strategies is

significantly lower than that of a buy-and-hold, providing strong evidence that any

outperformance of sector rotation strategies is due to data snooping.

6.6. Sequential industry performance

Although our analysis considers alternative stages, the actual progression of sector

performance across business cycles may not fully align with those partitions. To overcome

such obstacles, we next relax any assumed pattern of sequential performance and completely

ignore business cycle stages. The analysis tests whether the excess market returns of one sector

predict future excess market returns of other sectors at different lags. The analysis examines

lags from one to 24 months, to allow for different performance sequencing and business cycle

stage durations.

Figure 5 illustrates the distribution of t-statistics for cross-sector predictability of excess sector

performance. First, the analysis maps the Fama-French 49 industries to 10 equally weighted

sector portfolios following Table A1. Next, the analysis runs individual regressions of excess

market sector returns on the excess market returns of the remaining sectors at lags from one to

24 months. In total, there are 2,160 (10 x 9 x 24) t-statistics, covering all possible combinations

of sectors and lags. Figure 5 compares the resultant t-statistic distribution against an expected

normal distribution. The figure illustrates that the distribution of t-statistics for excess market

predictability follows a normal distribution. Under a normal distribution and a 10 percent

significance level, the estimations should indicate 5 percent positive significance and 5 percent

negative significance – even in the absence of actual excess market predictability. In total, t-

statistics are significantly positive 6 percent of the time and significantly negative 5 percent of

the time. Most significant predictability occurs at a one-month lag, indicating some short-term

cross-sector momentum. Cross-sector predictability is only marginally higher than a normal

24
distribution. As such, the results suggest that cross-sector predictability occurs only randomly,

without indicating any real evidence of statistically significant sequential sector performance.

7. Conclusion

Despite thorough empirical tests, there is scant evidence that conventional sector rotation

across business cycles generates systematic excess returns. The analysis assumes that sector

rotation investors perfectly time business cycles and rotate sectors in accordance with popular

belief on sector performance. Even then, sector rotation generates, at best, 0.11 percent monthly

outperformance. Performance quickly diminishes with the introduction of transaction costs or

business cycle mistiming. In comparison, a similar investor, with perfect market timing ability,

would realize 0.15 percent monthly outperformance by simply switching to cash during early

recession.

The analysis generalizes the base case to allow for all possible sector rotation variations. The

analysis explores whether any industry provides systematic performance across any business-

cycle stage. The general results again provide limited evidence of systematic industry

performance over business cycles. The results suggest that no variation of sector rotation

provides systematic outperformance, questioning the popular belief that timing sector

investments with business cycles generate excess returns. The results do not necessarily

preclude investors from profiting through sector rotation. Different investments in sector and

industry funds, beyond the scope of this study, may outperform the market. The results simply

show that sectors fail to provide systematic performance across the business cycle and question

the viability of popular sector rotation.

25
References

Ahmed, P., L. Lockwood, and S. Nanda, 2002, Multi-style rotation strategies, Journal of
Portfolio Management 28, 17-29.

Avramov, D. and R. Wermers, 2006, Investing in mutual funds when returns are predictable,
Journal of Financial Economics 81, 339-377.

Balvers, R., T. Cosimano, and B. McDonald, 1990, Predicting stock returns in an efficient
market, The Journal of Finance 45, 1109-1128.

Beber, A., M. Brandt, and K. Kavajecz, 2010, What does equity sector order flow tell us about
the economy? Review of Financial Studies 24, 3688-3730.

Bhojraj, S., C. Lee, and D. Oler, 2003, What's my line? A comparison of industry classification
schemes for capital market research, Journal of Accounting Research 41, 745-774.

Bodie, Z., A. Kane, and A. Marcus, 2009. Investments (McGraw Hill Irwin)

Boyd, J., J. Hu, and R. Jagannathan, 2005, The stock market's reaction to unemployment news:
Why bad news is usually good for stocks, Journal of Finance 60, 649-672.

Brocato, J. and S. Steed, 1998, Optimal asset allocation over the business cycle, The Financial
Review 33, 129-148.

Campbell, J., 1987, Stock returns and the term structure, Journal of Financial Economics 18,
373-399.

Carhart, M., 1997, On persistence in mutual fund performance, Journal of Finance 52, 57-82.

Cavaglia, S., C. Brightman, and M. Aked, 2000, The increasing importance of industry factors,
Financial Analysts Journal 56, 41-54.

Chava, S., A. Hsu, and L. Zeng, 2018. Does history repeat itself? Business cycle and industry
returns. SSRN working paper.

Chen, N., 1991, Financial investment opportunities and the macroeconomy, Journal of Finance
46, 529-554.

Chen, N., R. Roll, and S. Ross, 1986, Economic forces and the stock market, The Journal of
Business 59, 383-403.

Conover, M., G. Jensen, R. Johnson, and J. Mercer, 2008, Sector rotation and monetary
conditions, Journal of Investing 34-46.

Cover, J. and P. Pecorino, 2005, The length of u.S. Business expansions: When did the break
in the data occur?, Journal of Macroeconomics 27, 452-471.

DeStefano, M., 2004, Stock returns and the business cycle, Financial Review 39, 527-547.

26
Elton, E., M. Gruber, and C. Blake, 2011, An examination of mutual fund timing ability using
monthly holding data, Review of Finance 3, 619-645.

Fabozzi, F., 2007. Fixed income analysis (Wiley).

Fama, E. and K. French, 1989, Business conditions and expected returns on stocks and bonds,
Journal of Financial Economics 25, 23-49.

Fama, E. and K. French, 1992, The cross-section of expected stock returns, Journal of Finance
47, 427-465.

Fama, E. and K. French, 1997, Industry costs of equity., Journal of Financial Economics 43,
153-193.

Goyenko, R., C. Holden, and C. Trzcinka, 2009, Do liquidity measures measure liquidity?,
Journal of Financial Economics 92, 153-181.

Guris, S., and A. Pala, 2014. Equity returns, firm-specific characteristics and sector rotation:
Evidence from Turkey. International Journal of Economics and Financial Issues 4,
264-276.

Hamilton, J. and G. Lin, 1996, Stock market volatility and the business cycle, Journal of
Applied Econometrics 11, 573-593

Hasbrouck, J., 2009, Trading costs and returns for U.S. Equities: Estimating effective costs
from daily data, Journal of Finance 64, 1445-1477.

Hong, H. and J. Stein, 1999, A unified theory of underreaction, momentum trading, and
overreaction in asset markets, Journal of Finance 54, 2143-2184.

Hong, H., W. Torous, and R. Valkanov, 2007, Do industries lead stock markets?, Journal of
Financial Economics 83, 367-396.

Jensen, G., J. Mercer, and R. Johnson, 1996, Business conditions, monetary policy, and
expected security returns, Journal of Financial Economics 40, 213-237.

Jiang, G., T. Yao, and T. Yu, 2007, Do mutual funds time the market? Evidence from portfolio
holdings, Journal of Financial Economics 86, 724-758.

Kacperczyk, M., C. Sialm, and L. Zheng 2005, On the industry concentration of actively
managed equity mutual funds, Journal of Finance 60, 1983-2011.

Keim, D. and A. Madhavan, 1998, The cost of institutional equity trades, Financial Analysts
Journal 54, 50-69.

Keim, D. and R. Stambaugh, 1986, Predicting returns in the stock and bond markets, Journal
of Financial Economics 17, 357-390.

Lesmond, D., M. Schill, and C. Zhou, 2004, The illusory nature of momentum profits, Journal
of Financial Economics 71, 349-380.

27
Lofthouse, S., 2001. Investment management (Wiley).

Levis, M. and M. Liodakis, 1999, The profitability of style rotation strategies in the United
Kingdom, Journal of Portfolio Management.

Lochstoer, L., 2009, Expected returns and the business cycle: Heterogeneous goods and time-
varying risk aversion, Review of Financial Studies 22, 5251-5294.

Lynch, A., J. Wachter, and W. Boudry, 2004, Does mutual fund performance vary over the
business cycle?, AFA 2004 San Diego Meetings.

Markese, J., 1986, The stock market and business cycles, AAII Journal 8, 30-32.

McQueen, G. and V. Roley, 1993, Stock prices, news, and business conditions., Review of
Financial Studies 6, 683.

Moore, G., 1974, Some secular changes in business cycles, The American Economic Review
64, 133-137.

Petkova, R., 2006, Do the fama-french factors proxy for innovations in predictive variables,
The Journal of Finance 61, 581-612.

Salsman, R., 1997, Using market prices to guide sector rotation, (CFA Institute).

Sarwar, G, C. Mateus and N. Todorovic, 2018. US sector rotation with five-factor Fama-French
alphas, Journal of Asset Management 19, 116-132.

Siegel, J., 1991, Does it pay stock investors to forecast the business cycle, Journal of Portfolio
Management Fall, 27-34.

Schwert, W., 1990, Stock returns and real activity: A century of evidence., Journal of Finance
45, 1237.

Stock, J. and M. Watson, 1989, New indexes of coincident and leading economic indicators,
NBER macroeconomics annual 351-394.

Stock, J. and M. Watson, 1999, Business cycle fluctuations in U.S. Macroeconomic time series,
Handbook of Macroeconomics (Elsevier), Chapter 1.

Stovall, S., 1996. Standard & poor's guide to sector investing (McGraw-Hill).

28
Figure 1. Popular guidance on sector rotation

http://personal.fidelity.com/products/funds/content/sector/cycle.shtml

http://www2.standardandpoors.com/spf/pdf/index/Global_Sector_Investing.pdf

29
Figure 2. Stylized business cycles with stage partitions
Figure 2 illustrates a stylized business cycle. The official government agency responsible for dating U.S.
business cycles is the National Bureau of Economic Research (NBER). The NBER publishes dates for business
cycle peaks and troughs. Phases of expansion run from the month following a trough to the next peak and
phases of recession run from the month following a peak to the next trough. Similar to Stovall (1996) and
common practice, the analysis divides expansions into three equal stages (early/middle/late) and recessions into
two stages (early/late).

NBER peak

Expansion Recession

Stage I Stage II Stage III Stage IV StageV

NBER trough NBER trough

Stages of Expansion Stages of Recession

Early Expansion (Stage I) 241 months Early Recession (Stage IV) 62 months
Middle Expansion (Stage II) 242 months Late Recession (Stage V) 60 months
Late Expansion (Stage III) 242 months

30
Figure 3. CFNAI delineated business cycle stages
Figure 3 illustrates the CFNAI economic indicator over the period 1968–20018. Shaded areas indicate NBER
defined periods of economic contraction. The analysis partitions the range of CFNAI values over 1968–2018 into
five equal periods of economic activity. The five periods correspond to early expansion (SI), middle expansion
(SII), late expansion (SIII), early recession (SIV), and late recession (SV). Business-cycle stage delineations are
at CFNAI values of 0.702, 0.312, -.0113, and -.637 for boundaries SI|SII, SII|SIII, SIII|SIV, and SIV|SV
respectively.
2.5

1.5

0.5

1968 1974 1979 1985 1990 1996 2001 2007

-0.5

-1.5

-2.5

-3.5

-4.5

NBER Recessions SI|SII SII|SIII SIII|SIV SIV|SV CFNAI

31
Figure 4. Distribution of all possible sector strategy returns
Figure 4 presents the distribution of returns of 1,022 sector rotation strategies formed by 10 sectors using a two-
stage business cycle partition.

160

140

120

100

80

60

40

20

32
Figure 5. Predictability of excess industry performance
Figure 5 illustrates the distribution of t-statistics for cross-sector predictability of excess market performance. The
analysis constructs sector rotation portfolios from the Fama and French 49 industries mapped to one of 10 GICS
sectors reported in Table A1. The analysis tests lags from one to 24 months to allow for the possibility of different
performance sequencing and business cycle stage durations. To illustrate, Figure 1 shows financial sector returns
should predict subsequent technology sector returns. There are 2,160 t-statistics, covering all possible
combinations of cross-sector predictability at up to 24 lags.

25.00%

20.00%

15.00%

10.00%

5.00%

0.00%
1 to 1.65
0 to 0.5

0.5 to 1
< -3.29

> 3.29
-1 to -0.5
-3.29 to -2.58

-2.58 to -1.96

-1.96 to -1.65

-1.65 to -1

-0.5 to 0

1.65 to 1.96

1.96 to 2.58

2.58 to 3.29

Actual distribution Normal distribution

33
Table 1. NBER reference business cycle dates and stage partitions
Panel A of Table 1 reports NBER published business cycle peak and trough reference dates. Periods of recession
run from the first month following a cycle peak to the subsequent trough, and periods of expansion run from the
first month following a cycle trough to the subsequent peak. The sample covers 10 business cycles from 1948 to
2007, enumerated in the first column. The last column reports the total months in a business cycle from one month
after a peak to the next peak. The last recorded NBER business cycle date is the economic peak dated December
2007. Panel B of Table 1 reports the duration in months for stages of expansion and recession that correspond
with the business cycles reported in Panel A. The analysis partitions NBER defined periods of expansion into
three equal stages (early, middle, and late) and NBER defined periods of recession into two equal stages (early
and late). The bottom of Panel B reports the average duration of each business cycle stage.
Panel A: NBER business cycle dates from Jan 1948 - Dec 2007
Business Peak Trough Peak Total
Cycle Date Date Date Months
1 11/48 10/49 07/53 56
2 07/53 05/54 08/57 49
3 08/57 04/58 04/60 32
4 04/60 02/61 12/69 116
5 12/69 11/70 11/73 47
6 11/73 03/75 01/80 74
7 01/80 07/80 07/81 18
8 07/81 11/82 07/90 108
9 07/90 03/91 03/01 128
10 03/01 11/01 12/07 81
Panel B: Number of months in NBER delineated business cycle stages
Business Periods of Recession Periods of Expansion
Early Stage Late Stage Total Early Stage Middle Stage Late Stage Total
Cycle Months Months Months Months Months Months Months
1 6 5 11 15 15 15 45
2 5 5 10 13 13 13 39
3 4 4 8 8 8 8 24
4 5 5 10 35 35 36 106
5 6 5 11 12 12 12 36
6 8 8 16 19 19 20 58
7 3 3 6 4 4 4 12
8 8 8 16 30 31 31 92
9 4 4 8 40 40 40 120
10 4 4 8 25 25 23 73
stage average: 5 5 10 20 20 20 60

34
Table 2. Business cycle proxies across business cycle stages
Panel A of Table 2 lists the expected change in business cycle proxies from one business cycle stage to the next.
Panel B of Table 2 reports business cycle proxy means by business cycle stage and changes in means from the
preceding stage estimated with Equation 1, where business cycle dummy variables (Ds) take the value of one or
zero depending on the current business cycle stage. The analysis then calculates the difference in proxy means
between successive business cycle stages. As an example, Panel B reports an average 0.3% difference in term-
spread between the stages of early expansion and late recession ( 1-5). Lastly, the analysis performs a simple
difference in means test, to verify the statistical significance of the difference in means between the current and
preceding stage. The table reports p-values under a null hypothesis of no difference in proxies across successive
business cycle stages, formally stated as Ho: s=s-1.
Panel A: Change Change Change Change Change
Early Expansion Middle Expansion Late Expansion Early Recession Late Recession
Term-spread negative negative negative positive positive
Default-spread negative negative negative positive positive
Dividend yield negative negative negative positive positive
Unemployment negative negative negative positive positive
Industrial production positive positive positive negative negative

Panel B: Stage I Stage II Stage III Stage IV Stage V

Early Expansion Middle Expansion Late Expansion Early Recession Late Recession
Factor Mean Change p-value Mean Change p-value Mean Change p-value Mean Change p-value Mean Change p-value
Term-spread 0.021 0.003 0.00 0.014 -0.007 0.00 0.006 -0.009 0.00 0.098 0.004 0.00 0.018 0.008 0.00
Default-spread 0.009 -0.005 0.00 0.007 -0.002 0.00 0.008 0.001 0.32 0.010 0.002 0.00 0.014 0.004 0.00
Dividend yield 0.033 -0.010 0.00 0.030 -0.003 0.02 0.030 0.000 0.38 0.041 0.010 0.00 0.043 0.002 0.23
Unemployment 1.934 -0.001 0.51 1.684 -0.250 0.00 1.482 -0.202 0.00 1.662 0.180 0.00 1.936 0.274 0.00
Industrial production -2.280 0.284 0.32 0.219 2.500 0.00 2.357 2.138 0.00 1.290 -1.064 0.14 -2.560 -3.850 0.00

35
Table 3. Business cycle stages of expected industry performance
Table 3 reports the business cycle stage of anticipated sector/industry outperformance following the Stovall (1996)
classification and the investment websites illustrated in Figure 1. The table divides the periods of expansion into
three equal stages (early/middle/late) and periods of recession into two equal stages (early/late). The Fama and
French 49 industry portfolios (excluding “other”) are mapped to corresponding sectors.
Period of Expansion Period of Recession
Early Expansion - Stage I Middle Expansion - Stage II Late Expansion - Stage III Early Recession - Stage IV Late Recession - Stage V
Technology: Basic Materials: Consumer Staples: Utilities: Consumer Cyclical:
Computer Software Precious Metals Agriculture Gas & Electrical Utilities Apparel
Measuring & Control Equip. Chemicals Beer & Liquor Telecom Automobiles & Trucks
Computers Steel Works Etc Candy & Soda Business Supplies
Electronic Equipment Non-Metallic & Metal Mining Food Products Construction
Transportation: Capital Goods: Healthcare Construction Materials
General Transportation Fabricated Products Medical Equipment Consumer Goods
Shipping Containers Defense Pharmaceutical Products Entertainment
Machinery Tobacco Products Printing & Publishing
Ships & Railroad Equip. Energy: Recreation
Aircraft Coal Restaurants, Hotels, Motels
Electrical Equipment Petroleum & Natural Gas Retail
Services: Rubber & Plastic Products
Business Services Textiles
Personal Services Wholesale
Financial:
Banking
Insurance
Real Estate
Trading

36
Table 4. Industry summary statistics by business cycle stages
Table 4 reports industry summary statistics for the business cycle stage popular belief anticipates outperformance
will occur, as listed in Table 3. The table also reports Wald p-values under a null hypothesis of equal industry
returns across all five business cycle stages. For comparative purposes, the table provides industry summary
statistics for the full sample 1948-2018. The table reports equally weighted industry averages and market returns
beneath each business cycle stage.
Business Cycle Stage Full Sample 1948:01-2007:12
no. no. Wald
Sectors/Industries firms obs. mean std.dev. beta p-value mean std.dev. beta
Early Expansion - Stage I:
Computers 90 241 1.17 6.48 1.34 0.00 0.97 6.75 1.22
Computer Software 175 169 0.27 9.08 1.48 0.01 0.38 11.11 1.63
Electronic Equipment 157 241 1.20 6.96 1.43 0.00 0.91 7.21 1.39
Measuring & Control 70 241 0.92 6.23 1.31 0.00 0.98 6.64 1.28
Shipping Containers 26 241 1.32 5.10 0.99 0.01 0.94 5.39 1.00
Transportation 93 241 1.84 4.93 1.00 0.00 0.85 5.54 1.07
Industry Averages 1.12 6.46 1.26 0.00 0.84 7.11 1.27
Market 241 1.30 3.90 1.00 0.00 0.89 4.23 1.00

Middle Expansion - Stage II:

Chemicals 79 242 0.91 4.88 1.14 0.00 0.87 5.30 1.06
Steel Works 78 242 0.83 6.06 1.22 0.00 0.62 7.04 1.33
Precious Metals 20 206 0.11 9.97 0.61 0.29 0.44 10.26 0.62
Mining 24 242 0.55 6.72 1.17 0.00 0.81 6.96 1.09
Fabricated Products 20 206 0.64 6.74 1.03 0.00 0.51 7.25 1.13
Machinery 149 242 1.16 5.16 1.20 0.00 0.86 5.85 1.21
Electrical Equipment 68 242 1.30 5.43 1.27 0.00 1.02 6.02 1.23
Aircraft 25 242 1.43 5.80 1.14 0.00 1.09 6.54 1.13
Shipbuilding & Railroad 11 242 0.77 5.65 1.17 0.01 0.82 6.67 1.07
Defense 7 205 1.42 5.66 0.99 0.03 0.99 6.53 0.84
Personal Services 38 242 0.91 6.01 1.18 0.00 0.65 6.48 1.06
Business Services 166 242 1.11 4.48 1.04 0.00 0.85 5.26 1.08
Industry Averages 0.93 6.05 1.10 0.03 0.79 6.68 1.07
Market 242 1.06 3.72 1.00 0.00 0.89 4.23 1.00

Late Expansion - Stage III:

Agriculture 11 242 0.89 6.29 0.81 0.00 0.71 6.29 0.89
Food Products 80 242 0.52 4.23 0.61 0.00 0.95 4.11 0.69
Candy & Soda 12 195 0.44 6.41 0.77 0.00 0.96 6.25 0.83
Beer & Liquor 15 242 0.78 5.35 0.80 0.00 0.98 4.89 0.77
Tobacco Products 9 242 1.07 5.73 0.39 0.22 1.12 5.66 0.63
Healthcare 74 165 0.79 8.75 1.17 0.00 0.70 8.18 1.13
Medical Equipment 87 242 1.07 4.87 0.85 0.02 1.08 5.43 0.92
Pharmaceutical 130 242 0.83 4.54 0.71 0.00 1.05 4.89 0.83
Coal 8 242 1.84 9.49 1.04 0.00 0.70 9.54 1.16
Petroleum & Natural Gas 136 242 0.96 5.07 0.75 0.00 0.95 5.23 0.84
Industry Averages 0.92 6.07 0.79 0.02 0.92 6.05 0.87
Market 242 0.65 4.06 1.00 0.00 0.89 4.23 1.00

37
Table 4 continued:
Business Cycle Stage Full Sample 1948:01-2007:12
no. no. Wald
Sectors/Industries firms obs. mean std.dev. beta p-value mean std.dev. beta
Early Recession - Stage IV:
Utilities 137 62 -0.32 4.92 0.79 0.02 0.85 3.77 0.53
Communication 43 62 -0.71 4.60 0.77 0.02 0.80 4.27 0.74
Industry Averages -0.52 4.76 0.78 0.02 0.83 4.02 0.64
Market 62 -1.58 4.65 1.00 0.00 0.89 4.23 1.00

Late Recession - Stage V:

Recreation 32 60 3.05 10.36 1.32 0.00 0.70 7.09 1.16
Entertainment 33 60 2.26 11.79 1.56 0.04 1.06 7.30 1.34
Printing & Publishing 32 60 2.87 8.81 1.22 0.00 0.85 5.81 1.09
Consumer Goods 82 60 2.39 6.70 0.94 0.00 0.89 4.58 0.83
Apparel 57 60 3.18 9.11 1.20 0.00 0.85 5.91 1.05
Rubber & Plastic 29 60 1.99 8.59 1.11 0.00 0.94 5.78 1.07
Textiles 46 60 1.81 13.18 1.67 0.00 0.77 6.64 1.12
Construction Materials 125 60 2.37 9.86 1.40 0.00 0.87 5.74 1.16
Construction 32 60 3.14 10.48 1.45 0.00 0.81 6.92 1.29
Automobiles & Trucks 65 60 1.94 10.77 1.37 0.00 0.79 6.37 1.15
Business Supplies 32 60 2.13 8.08 1.15 0.00 0.87 5.61 1.03
Wholesale 96 60 2.28 7.71 1.07 0.00 0.86 5.35 1.05
Retail 172 60 2.96 7.27 1.09 0.00 0.96 4.99 0.97
Restaurants & Hotels 48 60 2.95 8.14 1.09 0.00 0.98 5.81 1.02
Banking 151 60 1.98 9.34 1.30 0.06 0.90 5.60 1.02
Insurance 77 60 2.06 8.37 1.13 0.00 0.89 5.56 0.96
Real Estate 32 60 2.35 14.14 1.75 0.00 0.57 7.37 1.23
Trading 186 60 2.89 8.45 1.25 0.00 0.99 5.86 1.23
Industry Averages 2.48 9.51 1.28 0.01 0.86 6.02 1.10
Market 60 2.10 6.29 1.00 0.00 0.89 4.23 1.00

38
Table 5. Industry performance measures by business cycle stage
Table 5 reports industry excess market returns, Jensen’s alphas, Fama and French (1992) three-factor alphas, and
Carhart (1997) four-factor alphas for the business-cycle stages of expected outperformance listed in Table 3.
Equations 2-5 estimate excess market returns, Jensen’s alphas, Fama and French alphas, and Carhart alphas by
business-cycle stage. Emboldened alpha performance indicates 10 percent statistical significance estimated with
White (1980) heteroskedasticity consistent t-statistics.

Excess Market Jensen's alpha Fama-French alpha Carhart alpha

Industries m p-value J p-value F p-value C p-value

Early Expansion - Stage I:
Computers -0.0029 0.37 -0.0065 0.04 -0.0088 0.02 -0.0060 0.10
Computer Software -0.0052 0.45 -0.0091 0.20 -0.0117 0.12 -0.0114 0.14
Electronic Equip. -0.0015 0.64 -0.0060 0.07 -0.0084 0.04 -0.0063 0.11
Measuring & Control -0.0065 0.03 -0.0097 0.00 -0.0119 0.00 -0.0110 0.00
Shipping Containers -0.0004 0.87 -0.0003 0.89 -0.0050 0.12 -0.0051 0.12
Transportation 0.0080 0.00 0.0080 0.00 0.0022 0.45 0.0010 0.72
Industry Average: -0.0014 -0.0039 -0.0073 -0.0065
Middle Expansion - Stage II:
Chemicals -0.0018 0.35 -0.0029 0.15 -0.0081 0.01 -0.0069 0.02
Steel Works 0.0006 0.84 -0.0009 0.77 -0.0069 0.08 -0.0055 0.15
Precious Metals -0.0081 0.32 -0.0053 0.54 -0.0108 0.23 -0.0088 0.35
Mining -0.0060 0.13 -0.0072 0.08 -0.0139 0.01 -0.0129 0.01
Fabricated Products -0.0012 0.79 -0.0012 0.78 0.0000 0.99 0.0018 0.74
Machinery 0.0017 0.40 0.0002 0.90 -0.0059 0.06 -0.0048 0.11
Electrical Equip. 0.0016 0.48 -0.0003 0.87 -0.0060 0.06 -0.0064 0.04
Aircraft 0.0024 0.44 0.0013 0.68 -0.0042 0.26 -0.0057 0.13
Shipbuilding/Railroad -0.0030 0.35 -0.0043 0.18 -0.0057 0.01 -0.0095 0.02
Defense 0.0029 0.46 0.0029 0.48 -0.0009 0.84 -0.0024 0.59
Personal Services 0.0012 0.71 0.0001 0.97 -0.0052 0.15 -0.0045 0.22
Business Services 0.0013 0.47 0.0009 0.59 -0.0036 0.16 -0.0032 0.20
Industry Average: -0.0007 -0.0014 -0.0059 -0.0057
Late Expansion - Stage III:
Agriculture 0.0059 0.14 0.0064 0.11 -0.0006 0.87 0.0000 0.90
Food Products -0.0026 0.32 -0.0017 0.51 -0.0090 0.00 -0.0085 0.00
Candy & Soda -0.0021 0.60 -0.0011 0.77 -0.0119 0.02 -0.0111 0.03
Beer & Liquor 0.0007 0.83 0.0120 0.70 -0.0059 0.07 -0.0051 0.13
Tobacco Products 0.0026 0.55 0.0041 0.30 -0.0032 0.44 -0.0021 0.66
Healthcare 0.0019 0.76 0.0015 0.81 -0.0028 0.56 -0.0017 0.75
Medical Equipment 0.0032 0.23 0.0036 0.18 -0.0032 0.32 -0.0043 0.17
Pharmaceutical -0.0030 0.40 0.0010 0.68 -0.0056 0.06 -0.0069 0.03
Coal 0.0192 0.00 0.0191 0.00 0.0113 0.08 0.0073 0.26
Petroleum & Natural Gas 0.0036 0.24 0.0042 0.16 -0.0035 0.26 -0.0041 0.21
Industry Average: 0.0029 0.0049 -0.0034 -0.0037
Early Recession - Stage IV:
Gas & Electric 0.0139 0.00 0.0095 0.02 0.0020 0.62 0.0004 0.93
Communication 0.0104 0.01 0.0054 0.12 -0.0027 0.45 -0.0024 0.55
Industry Average: 0.0122 0.0075 -0.0004 -0.0010

39
Table 5 continued:
Jensen's alpha Fama-French alpha Carhart alpha

Industries J p-value F p-value C p-value

Late Recession - Stage V:
Recreation 0.0000 0.99 -0.0098 0.24 -0.0017 0.82 -0.0019 0.81
Entertainment 0.0122 0.15 0.0067 0.42 -0.0158 0.06 -0.0153 0.08
Printing & Publishing 0.0088 0.14 0.0049 0.32 0.0019 0.70 0.0020 0.68
Consumer Goods 0.0032 0.43 0.0041 0.38 -0.0008 0.87 -0.0011 0.82
Apparel 0.0121 0.08 0.0087 0.17 0.0020 0.68 0.0021 0.66
Rubber & Plastic -0.0017 0.79 -0.0035 0.59 -0.0091 0.15 -0.0088 0.17
Textiles -0.0018 0.88 -0.0134 0.20 -0.0145 0.05 -0.0137 0.06
Construction Materials 0.0031 0.64 -0.0038 0.55 -0.0085 0.16 -0.0083 0.18
Construction 0.0120 0.12 0.0041 0.56 -0.0048 0.47 -0.0048 0.47
Automobiles & Trucks -0.0005 0.95 -0.0069 0.44 -0.0091 0.31 -0.0084 0.36
Business Supplies 0.0006 0.90 -0.0020 0.65 -0.0059 0.22 -0.0057 0.24
Wholesale 0.0023 0.64 0.0012 0.80 -0.0057 0.18 -0.0057 0.19
Retail 0.0085 0.07 0.0084 0.07 0.0020 0.68 0.0017 0.71
Restaurants & Hotels 0.0082 0.16 0.0066 0.30 -0.0018 0.77 -0.0018 0.77
Banking -0.0014 0.83 -0.0066 0.37 -0.0074 0.18 -0.0075 0.19
Insurance -0.0004 0.95 -0.0026 0.71 -0.0055 0.44 -0.0057 0.42
Real Estate 0.0061 0.63 -0.0070 0.55 -0.0072 0.47 -0.0066 0.52
Trading 0.0075 0.10 0.0031 0.53 0.0030 0.51 0.0032 0.49
Industry Average: 0.0044 -0.0004 -0.0049 -0.0048

40
Table 6. Performance comparison of alternative investment strategies
Table 6 reports means, standard deviations, betas, and Sharpe ratios for market timing and sector rotation strategies
under different assumptions.

Panel A: Base-case specification

Sharpe
Strategy mean std.dev. beta ratio
Market 0.89% 4.23% 1.00 0.21
Sector rotation 1.00% 5.02% 1.04 0.20
Market-timing 1.04% 3.99% 0.89 0.26

Panel B: Alternative sector/industry groupings

Sharpe
Strategy: Sector rotation mean std.dev. beta ratio
10 Sectors 1.03% 5.36% 1.08 0.19
23 Industry groups 0.99% 5.09% 1.05 0.20

Panel C: Alternative business cycle stages

Sharpe
Strategy mean std.dev. beta ratio
2 NBER stages 0.89% 4.69% 1.05 0.18
4 NBER stages 1.01% 4.85% 1.04 0.21
5 CFNAI stages 0.73% 5.25% 1.05 0.14

41
Table 7. Comparison of strategy performance with different timing

Table 7 reports the performance of sector rotation and market timing with advanced or delayed strategy implementation at
business cycle stage turning points by the indicated months. The strategy rotates the Fama and French 49 industry portfolios
according to Table 3. Table 7 reports mean returns, standard deviations, and Sharpe ratios. Beta estimates come from a single-
index model. The reported performance results are before transaction costs.

Full Period 1948-2018

Sharpe
Strategy Implementation mean std.dev. beta ratio
Market 0.89% 4.23% 1.00 0.21

Sector Rotation:
- 3 months 0.97% 4.88% 0.99 0.20
- 2 months 1.02% 4.90% 1.00 0.21
- 1 month 1.01% 4.98% 1.02 0.20
at turning point 1.00% 5.02% 1.04 0.20
+ 1 month 0.98% 5.01% 1.03 0.20
+ 2 months 0.93% 5.07% 1.05 0.18
+ 3 months 0.93% 5.00% 1.04 0.19

Market Timing:
- 3 months 0.95% 3.90% 0.85 0.24
- 2 months 1.01% 3.94% 0.87 0.26
- 1 month 1.02% 3.95% 0.88 0.26
at turning point 1.04% 3.99% 0.89 0.26
+ 1 month 1.05% 3.98% 0.88 0.26
+ 2 months 1.04% 3.99% 0.89 0.26
+ 3 months 0.98% 4.01% 0.90 0.24

42
Table A1. Alternative industry definitions

Table A1 provides a mapping of the Fama and French 49 industry portfolios to 23 Global industrial Classification
Standard (GICS) industry groups and 10 sector classifications.

49 Fama-French Industries 23 GICS Major Industry Groups 10 Sectors

Code Industry Description Code Industry Group Description Code Sector Description
01 Agriculture 3020 Consumer Staples 01 Consumer Non-Durable
24 Aircraft 2010 Industrials 04 Manufacturing
10 Apparel 2520 Consumer Discretionary 01 Consumer Non-Durable
23 Automobiles & Trucks 2510 Consumer Discretionary 02 Consumer Durable
45 Banking 4010 Financials 10 Finance
04 Beer & Liquor 3020 Consumer Staples 01 Consumer Non-Durable
34 Business Services 2020 Industrials 08 Business Equipment & Services
39 Business Supplies 2020 Industrials 04 Manufacturing
03 Candy & Soda 3020 Consumer Staples 01 Consumer Non-Durable
14 Chemicals 1510 Materials 04 Manufacturing
29 Coal 1010 Energy 05 Energy
32 Communication 5010 Telecommunication Services 07 Telecom
36 Computer Software 4510 Information Technology 08 Business Equipment & Services
35 Computers 4520 Information Technology 08 Business Equipment & Services
18 Construction 2550 Consumer Discretionary 04 Manufacturing
17 Construction Materials 1510 Materials 04 Manufacturing
09 Consumer Goods 2530 Consumer Discretionary 02 Consumer Durable
26 Defense 2010 Industrials 04 Manufacturing
22 Electrical Equipment 2010 Industrials 04 Manufacturing
37 Electronic Equipment 4530 Information Technology 08 Business Equipment & Services
07 Entertainment 2540 Consumer Discretionary 01 Consumer Non-Durable
20 Fabricated Products 2010 Industrials 04 Manufacturing
02 Food Products 3010 Consumer Staples 01 Consumer Non-Durable
11 Healthcare 3510 Healthcare 03 Healthcare
46 Insurance 4030 Financials 10 Finance
21 Machinery 2010 Industrials 04 Manufacturing
38 Measuring & Control 4520 Information Technology 08 Business Equipment & Services
12 Medical Equipment 3510 Healthcare 03 Healthcare
28 Mining 1510 Materials 05 Energy
33 Personal Services 2530 Consumer Discretionary 01 Consumer Non-Durable
30 Petroleum & Natural Gas 1010 Energy 05 Energy
13 Pharmaceutical 3520 Healthcare 03 Healthcare
27 Precious Metals 1510 Materials 05 Energy
08 Printing & Publishing 2540 Consumer Discretionary 01 Consumer Non-Durable
47 Real Estate 4040 Financials 10 Finance
06 Recreation 2520 Consumer Discretionary 02 Consumer Durable
44 Restaurants & Hotels 2530 Consumer Discretionary 09 Wholesale & Retail
43 Retail 2550 Consumer Discretionary 09 Wholesale & Retail
15 Rubber & Plastic 2550 Consumer Discretionary 04 Manufacturing
25 Shipbuilding & Railroad 2010 Industrials 04 Manufacturing
40 Shipping Containers 2030 Industrials 04 Manufacturing
19 Steel Works 1510 Materials 04 Manufacturing
16 Textiles 2520 Consumer Discretionary 01 Consumer Non-Durable
05 Tobacco Products 3020 Consumer Staples 01 Consumer Non-Durable
48 Trading 4020 Financials 10 Finance
41 Transportation 2030 Industrials 04 Manufacturing
31 Utilities 5510 Utilities 06 Utilites
42 Wholesale 2550 Consumer Discretionary 09 Wholesale & Retail

43