ArticlePDF Available

Abstract and Figures

A debate about whether firms with superior customer satisfaction earn superior stock returns has been persistent in the literature. Using 15 years of audited returns, the authors find convincing empirical evidence that stock returns on customer satisfaction do beat the market. The recorded cumulative returns were 518% over the years studied (2000-2014), compared with a 31% increase for the S&P 500. Similar results using back-tested instead of real returns were found in the United Kingdom. The effect of customer satisfaction on stock price is, at least in part, channeled through earnings surprises. Consistent with theory, customer satisfaction has an effect on earnings themselves. In addition, the authors examine the effect of stock returns from earnings on stock returns from customer satisfaction. If earnings returns are included among the risk factors in the asset pricing model, the earnings variable partially mitigates the returns on customer satisfaction. Because of the long time series, it is also possible to examine time periods when customer satisfaction returns were below market. The reversal of the general trend largely resulted from short-term market idiosyncrasies with little or no support from fundamentals. Such irregularities have been infrequent and eventually self-correcting. The authors provide reasons why irregularities may occur from time to time.
Content may be subject to copyright.
Claes Fornell, Forrest V. Morgeson III, & G. Tomas M. Hult
Stock Returns on Customer
Satisfaction Do Beat the Market:
Gauging the Effect of a
Marketing Intangible
A debate about whether rms with superior customer satisfaction earn superior stock returns has been persistent in the
literature. Using 15 years of audited returns, the authors nd convincing empirical evidence that stock returns on
customer satisfaction do beat the market. The recorded cumulative returns were 518% over the years studied
(20002014), compared with a 31% increase for the S&P 500. Similar results using back-tested instead of real returns
were found in the United Kingdom. The effect of customer satisfaction on stock price is, at least in part, channeled
through earnings surprises. Consistent with theory, customer satisfaction has an effect on earnings themselves. In
addition, the authors examine the effect of stock returns from earnings on stock returns from customer satisfaction. If
earnings returns are included among the risk factors in the asset pricing model, the earnings variable partially mitigates
the returns on customer satisfaction. Because of the long time series, it is also possible to examine time periods when
customer satisfaction returns were below market. The reversal of the general trend largely resulted from short-term
market idiosyncrasies with little or no support from fundamentals. Such irregularities have been infrequent and
eventually self-correcting. The authors provide reasons why irregularities may occur from time to time.
Keywords: customer satisfaction, customer lifetime value, intangibles, stock portfolio returns, abnormal returns
Among the many potential performance outcomes for
marketing, there has been a rapid rise in interest in
measures of stock performance and customer sat-
isfaction over the past several years (e.g., Katsikeas et al.
2016). The literature on customer satisfaction and stock
returns has also become fairly extensive (Aksoy et al. 2008;
Anderson, Fornell, and Mazvancheryl 2004; Anderson and
Mansi 2009; Bell, Ledoit, and Wolf 2014; Fornell, Mithas,
and Morgeson 2009a, b; Fornell et al. 2006; Gruca and Rego
2005; Ittner and Larcker 1998; Ittner, Larcker, and Taylor
2009; Jacobson and Mizik 2009a, b; Lambert 1998; Luo,
Homburg, and Wieseke 2010; Malshe and Agarwal 2015;
Ngobo, Casta, and Ramond 2012; OSullivan, Hutchinson,
and OConnell 2009; OSullivan and McCallig 2009; Peng
et al. 2014; Tuli and Bharadwaj 2009). Although most of
these studies examine customer satisfaction using the same
data source (American Customer Satisfaction Index [ACSI];
www.theacsi.org) and nd predominantly positive risk-
adjusted stock returns, their conclusions differ with respect
to the statistical signicance of the abnormal returns and
whether there is evidence of mispricing.
Some researchers (Bell, Ledoit, and Wolf 2014; Jacobson
and Mizik 2009a; OSullivan, Hutchinson, and OConnell
2009) have argued that abnormal returns cannot be dis-
tinguished from random variation, or that there is no evidence
that customer satisfaction predicts long-term stock returns
(Ittner, Larcker, and Taylor 2009). These arguments and pre-
dictions can now be put to the test. If they are correct, then
customer satisfaction returns subsequent to these cited studies
should approximately equal market returns. The new empirical
evidence, however, points to anything but equal to market
returns. For the full 15-year time period (20002014), the
model-free audited cumulative gross returns on customer sat-
isfaction amounted to 518%.1By comparison, the Standard
& Poors [S&P] 500 grew by 31% over the same period
(20002014). On an annual basis, the customer satisfaction
portfolio outperformed the S&P in 14 out of 15 years. The
magnitude of such a return disparity over a period of 15 years
Claes Fornell is Chairman, CFI Group (e-mail: cfornell@cfigroup.com).
Forrest V. Morgeson III is Director of Research, American Customer
Satisfaction Index, LLC (e-mail: morgeson@theacsi.org). G. Tomas M. Hult
is Professor and Byington Endowed Chair, Eli Broad College of Business,
Michigan State University (e-mail: hult@msu.edu). The authors appreciate
the input, feedback, and assistance provided by Joshua Blechman,
Xueming Luo, and David Durfee.
1The lead author is the owner and founder of the U.S. fund studied
and the American Customer Satisfaction Index (ACSI). As regis-
tered under the Investment Advisers Act of 1940, the funds
investment manager is regulated by the SEC. This should not be
taken to imply a certain level of skill or training. The returns were
audited by an independent third-party accounting rm, unafliated
with the authors and registered with the Public Company
Accounting Oversight Board. The audited returns have also been
provided to the Editor in Chief as a part of the review process.
©2016, American Marketing Association Journal of Marketing
ISSN: 0022-2429 (print) Vol. 80 (September 2016), 92–107
1547-7185 (electronic) DOI: 10.1509/jm.15.022992
implies that known risk factors or sector biases are very unlikely
explanations because investors would have had enough time to
discover the anomalies and adjust to them accordingly.
These ndings do not mean, however, that the returns
cannot be mitigated by known factors from time to time. In-
deed, we nd that this is the case. Moreover, and similar to
the impact of employee satisfaction on stock prices (Edmans
2011), there is evidence that the effect of customer satisfaction
is channeled, at least to some extent, through earnings surprises.
That is, to the extent that the market does not react to news
about customer satisfaction, there is a reaction to its materi-
alization in corporate earnings and effect on earnings surprises.
We nd signicant relationships between customer satisfac-
tion and earnings and that customer satisfaction is predictive of
earnings surprises. In turn, stock returns on earnings mitigate
the effect on customer satisfaction returns. That is, some of the
effect on stock prices from customer satisfaction seems to be
absorbed by the effect of earnings on stock prices.
In addition, we identify time periods under which the
abnormal returns are negative (even though the absolute re-
turns remain positive), but they are too short in duration to be
statistically signicant and are overwhelmed by the predom-
inance of positive risk-adjusted alphas. Nevertheless, there
are lessons to be learned from the return reversals. They seem
to happen largely without assistance from a corresponding
reversal in fundamentals but as a reaction to price increases of
hitherto strong performing stocks in favor of less expensive
stocks (i.e., stocks with lower price-to-earnings [P/E] ratios).
The remainder of the study is structured as follows. We
begin with the theoretical impetus behind the proposition that
customer satisfaction is germane to rm value and demon-
strate how a small increase in customer retention can have a
large effect on equity value. This is of critical importance
because the equity value grows exponentially at higher levels
of customer retention, whereas the value of an increase in
customer retention from a low level to a somewhat higher
level is much smaller. The implication, which we test in this
study, is that changes in customer satisfaction are much more
likely to have an effect when they are accompanied by high
levels of customer satisfaction.
Next, we present the data from the ACSI and the stock
portfolio. We describe the relevant properties of eachthe
ACSI, with respect to the degree to which it is representative
of the U.S. stock market (S&P 500 index), and the stock
portfolio, with respect to how it is constructed with ACSI data
as input. In both cases, we take these data sets as they are. The
same is true with respect to the U.K. customer satisfaction data,
but not the U.K. portfolio returns. The U.K. portfolio returns
result from back testing thatwe performed. Because of the low
credibility of back testing due to the risk of data snooping (Lo
and MacKinlay 1990), we selected the simplest and most
transparent of trading rules, as we detail subsequently.
Following this, we turn to the analyses and ndings. We
employ the standard capital asset pricing model (CAPM) to
estimate the market risk premium and add the momentum
risk premium (Carhart 1997), the size risk premium, and the
growth risk premium (Fama and French 1993). The Barra
model, controlled for market returns, is used for additional
attribution analysis and shows that only 2 (technology and
retail) out of 16 style factors have a signicant effect, and
approximately 80% of the return variance is due to idio-
syncratic effects. Next, we estimate the effect of customer
satisfaction on earnings and earnings surprises as well as the
effect of earnings returns on customer satisfaction returns.
We nd that customer satisfaction has a signicant effect on
both earnings and earnings surprises. As we hypothesize,
changes in customer satisfaction have an effect on earnings
only if they are complemented by high levels of customer
satisfaction. We also use the new ve-factor FamaFrench
(2015) model to estimate the effect of earnings returns on
customer satisfaction returns, nding that returns on earnings
reduce returns on customer satisfaction to some degree.
Finally, we examine the extent to which the returns to
customer satisfaction can be generalized. Conrmatory evidence
from the United Kingdom shows results similar to those in the
United States. There is also disconrmatory evidence: over the
15-year time period, stocks of companies with strong customer
satisfaction did not always outperform the market. In particu-
lar, there was a time period in 20122013 marked by under-
performance. We examine market behavior during this period and
nd that it deviates substantially from the norm in the sense that
companies that have weak balance sheets, are heavily shorted,
and have low price-to-earnings ratios outperformed the market.
Theoretical Impetus
Major shifts in economic activity over the past several decades
have caused intangible assets to become a major force for value
creation, economic growth, and performance assessment (e.g.,
Katsikeas et al. 2016). These assets loom large in the modern
economy and are often valued higher than the assets on balance
sheets. They are generally not, however, capitalized like other
investments, thereby disconnecting the timing of income from
expenditure in nancial statements. For example, even though
the benets accrue in the future, investments in customer service
are usually fully expensed when they occur. As a result, they
might become visible in income statements as an input variable
without much information about the nature or timing of actual
outputs. Under such circumstances, it is difcult to ascertain the
real value contribution from having satised customers. For-
tunately, there is output information available, but even so, the
economic value of customer satisfaction still might not become
manifest until there is an impact on other outputs (e.g., earnings).
Perhaps more than any other intangible, satised cus-
tomers are essential for any seller in a competitive market
if repeat business is a signicant portion of total revenue.
Accordingly, customer satisfaction occupies a central place
in both micro and macro analysis. At the micro level, it is
a leading indicator of favorable (high level/low volatility)
net cash ows (e.g.,Gruca and Rego 2005). At the macro level,
it is related to economic growth through consumer spend-
ing and to the efciency by which capital is allocated (e.g.,
Fornell, Rust, and Dekimpe 2010). Allocative efciency, in
this sense, depends on the joint ability of consumer and equity
markets to punish (reward) rms that fail (succeed) in sat-
isfying their customers. That is, consumer markets reward
high-consumer-utility-producing rms with repeat busi-
ness and punish low-consumer-utility-producing ones by
Stock Returns on Customer Satisfaction / 93
defection. By rewarding high-consumer-utility-producing
rms and their shareholders with higher stock prices and
penalizing low-consumer-utility-producing rms with capital
withdrawal, capital markets would be in line with consumer
markets under the notion of allocative efciency. It is in this
sense that buyer satisfaction plays a vital role in the individual
companys ability to generate wealth at the micro level and in
allocative efciency at the macro level.
Most consumer markets are characterized by numerous
purchase alternatives and by repeat sales as a large portion of
rm revenue. High customer satisfaction, relative to competi-
tion, is associated with repeat purchase, market share protec-
tion, lower price elasticity, lower transaction costs, and lower
selling/marketing costs (Anderson, Fornell, and Lehmann 1994).
Satised customers are therefore important for earnings, return
on investments, return on assets, and cash ows (Aksoy et al.
2008; Anderson, Fornell, and Mazvancheryl 2004; Fornell et al.
2006; Gruca and Rego 2005; Tuli and Bharadwaj 2009). Be-
cause of its inuence on buyer loyalty, customer satisfaction is
also benecial for risk reduction. Systematic, idiosyncratic, and
downside risk are lower for rmswithstrongcustomersat-
isfaction (Fornell, Mithas, and Morgeson 2009a, b; Tuli and
Bharadwaj 2009). In addition, customer satisfaction is asso-
ciated with other benets, such as positive word of mouth,
higher reservation prices, more cross-buying, fewer consumer
complaints, lower warranty and eld service costs, and less
customer defection and employee turnover (Fornell et al. 2006).
Most of these effects have positive impacts on acceleration,
stability, size, andby implication (from loyal customers)
risk reduction regarding future cash ows.
Although there is substantial empirical support for many of
the aforementioned effects, the most fundamental nding has to
do with repeat business. There is a large literature on customer
lifetime value (CLV), or customer equity, whereby the eco-
nomic value of repeat business is determined by the discounted
net present value of future cash ows from current customers.
Consistent with the proposition that customer satisfaction has
a positive impact on stock price, research has shown that an
increase in CLV can lead to an increase in stock price (Kumar
and Shah 2009, 2011) and to higher future prots (Venkatesan
and Kumar 2004). Although there are several ways to measure
CLV (Holm, Kumar, and Rohde 2012; Kumar and George
2007), for our purposes and without loss of generalization,
assuming that prot margins are constant, the time period is
innite, and the customer retention rate does not vary over time,
we can theoretically express customer equity value as follows:
CEV =m
r
1+i-r
,(1)
where CEV =customer equity value, m =the prot margin
multiple, r =the proportion of retained customers, and i =the
discount rate. By way of example, consider a company in which
75% of customers return as future buyers and the discount rate
is 5%. Solving Equation 1 gives a margin multiple of 2.5. If
customer satisfaction increased and led to a growth in the
proportion of retained customers by, say, 5 percentage points
(r =.8), the corresponding margin multiple would be 3.2a
growth in equity value of 28%. Note that a fairly small increase
in retention (5 percentage points) leads to a much larger growth
in the value of customer equity (28%). The effect obviously
becomes more pronounced at higher levels of r, but even at
moderate levels, it is quite substantial and of relevance to
investors. Because customer satisfaction is a major contributor
to buyer retention, it follows that it too may affect stock price.
Practically, it may be useful to illustrate a few short examples
of relationships between customer satisfaction and share price
before turning to formal analysis. Consider four rmsApple,
Netix, Costco, and Home Depotover an identical ten-year
time period. In Figure 1, Panels AD, we depict the relationship
between ACSI and stock price for these companies.
The most dramatic relationship is the one for Apple. As its
customer satisfaction increased from a middle-of-the pack com-
pany to best in class, Apples stock price soared (Figure 1, Panel
A). Netix depicts a reversal of fortunes in both ACSI and stock
price (Panel B). In contrast to Apple, Netix already had a
very high level of customer satisfaction when the ACSI began
tracking it. As its ACSI score continued upward, its stock price
did too. However, Netix frustrated customers by increasing its
prices by 30%50% in 2011. Both its ACSI and stock price
plummeted. Costco illustrates yet another type of relationship
(Panel C). Here, it is change in the ACSI that is most relevant.
Even though Costco customer satisfaction is high, it is not as
high as that for many other companies. Finally, Home Depot
shows what can happen to a company that starts out with a
reasonable level of customer satisfaction, drops back signi-
cantly (to a low ACSI score of 66), and later not only recovers
but also achieves a higher ACSI score than it had before the
slump (Panel D). As customer satisfaction deteriorates, Home
Depots stock price falls. When customer satisfaction increases
sharply later on, so does the rms stock price. These com-
panies illustrate how ACSI and stock price move together in
different ways: from good to superb (Apple); from superb to
weak (Netix); from good to better (Costco); and from good, to
bad, to better (Home Depot). Although these are just examples
of individual companies, they do show a strong relationship
between customer satisfaction and stock price. Next, we ex-
amine the extent to which these examples can be generalized
in a portfolio of companies, selected on the basis of ACSI data.
Data
Customer Satisfaction Data
The customer satisfaction data used in this study come from
the ACSI, which provides annual customer satisfaction scores
from a latent variable structural equation model estimated from
survey data on customer satisfaction for approximately 300 of
the largest companies, across 45 distinct industries, in the U.S.
consumer market (e.g., Fornell et al. 2006).2Data are collected
on a quarterly basis for different industries, with approximately
2Customer satisfaction is measured for the largest companies in
each industry. Within each industry, those with the largest U.S.
market shares are covered. The sample of measured ACSI com-
panies includes mostly larger consumer goods rms. Although this
means that the ACSI is not perfectly representative of the broader
universe of publicly traded companies, it is in keeping with the prior
literature investigating the impact of satisfaction (and other intan-
gibles) on market performance (Edmans 2011).
94 / Journal of Marketing, September 2016
25% of the total annual probability sample of about 70,000
households interviewed each quarter, resulting in annual satis-
faction scores for each company.3Although the ACSI is obvi-
ously skewed toward consumer markets, it seems to represent the
overall stock market quite well.4
Audited Returns and Back-Tested Returns
The other major data source is audited stock portfolio returns
obtained from the same stock fund examined in Fornell et al.
(2006). Previous research has not detected an announcement
effect regarding ACSI news releases and press coverage
(Fornell et al. 2006; Raithel et al. 2011; Srinivasan and Hanssens
2009) or has found only a limited effect (Ittner, Larcker, and
Taylor 2009). Accordingly, the trading is based on both levels
and changes in a companys customer satisfaction score. Stocks
are purchased both before and after the ACSI release date, with
a blackout period of 48 hours surrounding the announcement.5
In general, access to audited returns data, which is what
this study relies on for its major analysis, is advantageous be-
cause data snooping bias is eliminated (Black 1993; Lo and
MacKinlay 1990; Merton 1987). Data snooping refers to the
reuse of data for model testing. For example, trying out various
rules for stock portfolio construction in a back-testing scenario
is almost always subject to data snooping biasleading to
stock-picking rules based on sample uniqueness but without
predictive superiority. There is substantial evidence that
FIGURE 1
Examples of the ACSIStock Price Relationship
A: Apple B: Netflix
spApple ACSI Apple tock rice
Year
$0
$100
$200
$300
$400
$500
$600
65
70
75
80
85
90
Stock Price
ACSI
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Netflix ACSI Netflix stock price
Year
$0
$40
$80
$120
$160
$200
65
70
75
80
85
90
Stock Price
ACSI
2007
2008
2009
2010
2011
2012
2002
2003
2004
2005
2006
Home Depot ACSI Home Depot stock price
Year
$0
$10
$20
$30
$40
$50
$60
$70
60
64
68
72
76
80
Stock Price
ACSI
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Costco ACSI Costco stock price
Year
$0
$20
$40
$60
$80
$100
$120
72
74
76
78
80
82
84
Stock Price
ACSI
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
C: Costco D: Home Depot
3The ACSI database contains approximately 3,200 rm-year
observations over a 20-year period, with an average customer sat-
isfaction score of 76.44 and an average annual standard deviation
of 2.43.
4For the time period 19972003, Fornell et al. (2006) examined
the top 20% ACSI rms and found that they outperformed the Dow
Jones Industrial Average (DJIA) by 21%40%, but that the bottom
80% ACSI rms had a return virtually identical to the DJIA (20.4%
and 21%, respectively). For this study, the correlation between the
returns of the ACSI universe and the S&P 500 (for the time period
20002014) was .88.The reason for the high correspondencebetween
the ACSI universe returns and S&P returns is probably that consumer
spending is such a large proportion of the gross domestic product and
that many rms covered by ACSI compete in both consumer and
nonconsumer markets.
5The blackout period is not imposed by regulation, but it is used
to remove any reasonable risk of trading on announcements. Even
though most prior research has not detected a signicant
announcement effect, its existence cannot be completely ruled out,
especially as the returns on customer satisfaction become more
widely known. For example, a recent study by Ivanov, Joseph, and
Wintoki (2013) nds that stock trading volume is 2.8% higher on
average during a ten-day period around the ACSI announcements
and that there is a market reaction to these announcements over six
trading days.
Stock Returns on Customer Satisfaction / 95
data-driven discovery leads to seriously distorted ndings and
that it is endemic in the literature (White 2000). The lack of
credibility is so serious that the SEC prohibits the use of back-
tested results in presentations of stock funds to investors. In
academia, however, back testing is common, presumably
because real returns data are difcult to obtain. The advantage
of back-tested results is thatthe portfolio constructionrules can
be made explicit, but that is of limited value if the process of
how these rules were established is not disclosed. If the rules
were established through trial and error, the predictive powers
of back-tested results are greatly curtailed.
However, even though actual returns are generally pref-
erable, they are not without limitations. They can almost never
be perfectly replicated because, in practice, stock trading rules
are almost never truly formulaic. Most professional inves-
tors require different risk exposure, stop-loss criteria, portfolio
diversication, leverage, and so on. Yet perfect numerical
replication is not required for generalization or for theory
testingverication and general principles are. Only audited
returns can provide true numerical verication.
The general trading principle can be succinctly stated as
follows: go long in rms with strong customer satisfaction
(i.e., high and rising levels of ACSI scores relative to other
rms in the same industry), and short in rms with weak
customer satisfaction (i.e., low and declining ACSI scores
relative to other rms in the same industry). This principle
may be operationalized in slightly different ways; the issue is
the extent to which different studies using the same theo-
retical principle, with different operationalizations, arrive at
similar results. Among the studies that have examined the
customer satisfactionstock return relationship, most have ob-
tained similar numerical results. Indeed, all previous studies
that have analyzed stock returns of rms with strong customer
satisfaction (high levels and positive changes) have reported
numerically large abnormal returns. When the regression esti-
mates, which typically reect monthly returns in these studies,
are expressed as annual returns, it is evident that the abnormal
returns are very large.
For example, Jacobson and Mizik (2009a) report risk-
ad just ed monthly abnormal returns of .007. Had they expressed
these returns in annual terms, it would have been obvious that a
monthly above-market return of .007 is very high, equaling a
market outperformance of 8.4% per year over ten years, even
after adjustments for additional risk factorsbetter than almost
any mutual fund over the same time period. Aksoy et al. (2008)
report abnormal annual returns of 10.6% over a different (but
somewhat overlapping) ten-year period. As we show herein,
these estimates are similar to our ndings over a time period
50% longer. We return to other studies subsequently and
discuss circumstances under which some of them reached
conclusions different to ours, but for now it is important to point
out that even though our main study does not include formulaic
trading rules, it does not follow that the general principle behind
the portfolio construction is insufcient for verication or that
the results cannot be approximated in a meaningful way.
In our case, then, we expect that the general principle of
going long in rms with strong customer satisfaction and
short in rms with weak customer satisfaction would lead
to similar returns even though operationalizations of the
principle may differ. We test this prediction by using data
from an additional market and country and, in so doing,
mitigate the strengths and weaknesses of real data versus
back-tested data. In this case, we back test a paper portfolio
based on customer satisfaction data in the United Kingdom
and compare its returns with the Financial Times Stock Ex-
change (FTSE) 100 index. Notwithstanding the limitations
of back testing, it serves as a complement to the actual and
audited returns data in the U.S. study. By combining actual
and back-tested returns, the weakness in one approach is
compensated by the strength of the other, and vice versa. The
estimated annual abnormal return (12%) in the U.K. study is
fairly similar to what we and others nd in the United States
(8.4%10.8%). The difference is most likely due to the fact
that there are more variables available for adjusting the
returns for risk in the United States and that the U.K. returns
do not take dividends into account. If dividends had been
included, the returns would have been reduced by approx-
imately 2%, bringing the U.K. returns in line with the U.S.
results.
Portfolio Returns
Let us begin with the returns on the U.S. customer satisfaction
portfolio described previously. Figure2 shows the cumulative,
model-free returns, expressed as the value of $100 invested
from April 2000 through June 2014. The investment grew to
$618 (+518%). Although it might appear that the 2009 re-
cession dip in the portfolio returns in Figure 2 was larger than
market, the opposite is actually the case in terms of percent-
age changes. This nding is consistent with Merrin, Hoffman,
and Pennings (2013), in that customer satisfaction is found to
act as a buffer after the bursting of price bubbles. It is also
evident that the returns are not due to spectacular performance
in a few years and underperformance in others, as is often the
case with mutual funds. As Figure 2 illustrates, the customer
satisfaction returns were higher than the returns for the S&P
500 in 14 of the 15 years, with 2013 as the sole outlier. In view
of the large difference between the S&P 500 and the customer
satisfaction portfolio performance over a long period of time, it
is very unlikelythat the result can be explained by known (risk)
factors or sector bias (whether retail, technology, or otherwise).
With a short time period, the impact of such factors could be
substantial, but this is not true in the long run. This is because
investors would have had sufcient time to react, and the
abnormal returns would have been eliminated. In the absence
of these factors, it would seem reasonable to conclude that the
results were essentially driven by customer satisfaction and
that investors in the main were unaware of its effect or its
manifestation. It may be possible, however, that known fac-
tors, to a limited extent, mitigate customer satisfaction returns.
This is what we examine next.
Let us begin the analysis of risk-adjusted returns by using
the standard SharpeLintner CAPM (Lintner 1965a, b; Sharpe
1965):
SATit =a
it +b1MKTt+e
it,(2)
where SAT
it
=the portfolio return minus the risk-free rate at
time t (month), a
it
=the estimate of risk-adjusted above-market
96 / Journal of Marketing, September 2016
returns at time t, MKT
t
(market risk premium) =the market
returns in excess of the risk-free rate at time t, and e=the error
term. We obtained the risk data for this and subsequent equa-
tions from Kenneth Frenchs website and data library (http://mba.
tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html).
The results (presented in Table 1) indicate a signicant alpha
of .009, equivalent to annualized market risk-adjusted returns of
10.8%, as well as a signicant market beta of .692.
Jacobson and Mizik (2009a) claim that that abnormal
returns from satisfaction are limited to the technology sector,
where they found adjusted abnormal annual returns of 32.4%
38.4% over ten years. No commercially available portfolio of
common stock has ever even been close to outperforming the
market by 32%38% annually over a period of ten years, so
there is reason to be somewhat skeptical about these num-
bers. This nding also seems inconsistent with other results
reported by Jacobson and Mizik. When technology (and
utilities) is removed from the portfolio, the abnormal returns
are reduced by only 1.0%3.3% for the full sample (depending
on the model they use), which must be statistically insignicant
if 6.4%8.4% is insignicantly different from zero (or nearly
so for the latter alpha), as they report in their study.
Even though we could not nd any theoretical or empirical
justication for treating technology differently, we did check
to determine whether the results changed under narrower mar-
ket indices, such as NASDAQ and the Dow Jones Industrial
Average (DJIA). The estimated annual alpha (adjusting for
dividends as an average of 2%) relative to NASDAQ is 11.2%,
slightly higher than the FamaFrench benchmark (MKT) at
10.8%, and the alpha with the DJIA is 10.0%, slightly lower.
If technology stocks were much more sensitive to customer
satisfaction, an even larger alpha would be expected.
FIGURE 2
Cumulative Returns on $100 Invested in Customer Satisfaction: Portfolio Versus the S&P 500 (April 2000
Through June 2014)
$617.59
$130.81
Mar 2000
Nov 2000
Jul 2001
Mar 2002
Nov 2002
Jul 2003
Mar 2004
Nov 2004
Jul 2005
Mar 2006
Nov 2006
Jul 2007
Mar 2008
Nov 2008
Jul 2009
Mar 2010
Nov 2010
Jul 2011
Mar 2012
Nov 2012
Jul 2013
Mar 2014
$0
$100
$200
$300
$400
$500
$600
$700
Portfolio S&P 500
TABLE 1
Risk-Factor Model Results
SharpeLintner
CAPM
FamaFrench
Three-Factor CAPM
Carhart Four-Factor
CAPM
Winsorized (95%) Carhart
Four-Factor CAPM
a.009*** .009*** .009*** .009***
a(%) 10.8% 10.8% 10.8% 10.8%
MKT .692*** .654*** .622*** .608***
SMB .171* .183** .072
HML .005 -.003 .028
MOM ——-.062* .050
*p<.05 (one-tailed).
**p<.01 (one-tailed).
***p<.001 (one-tailed).
Notes: MKT =monthly market returns excess of the risk-free rate; SMB =size risk-factor returns; HML =value risk-factor returns; MOM =momentum
risk-factor returns.
Stock Returns on Customer Satisfaction / 97
Other known factors cannot be expected to do much
better, but we examine the extent to which they mitigate the
returns. Adding risk factors proposed by Fama and MacBeth
(1973) and Fama and French (1993, 2004), we specify a three-
factor model:
SATit =a
it +b1MKTt+b2SMBt+b3HMLt+e
it,(3)
where the additional factors are SMB
t
(stock size risk
premium) =the size risk-factor returns at time t and HML
t
(stock growth risk premium) =the value risk-factor returns
at time t.
With respect to alpha, the result is identical to the original
CAPM (an alpha of .009, or 10.8% annual return). The mar-
ket beta is somewhat smaller at .654, as some of the effect is
picked up by the small stock risk factor (.171). The effect of
the growth risk premium is not signicant. We are not aware
of any theoretical basis for interpreting the size risk-factor
returns effect, even though it is statistically signicant at .05.
The ACSI is, by design, dominated by large cap stocks. The
returns of small stocks minus big stocks do covary with the
customer satisfaction returns (of large stocks), but that asso-
ciation is most likely a spurious one.
Finally, we add another risk factor, suggested by Carhart
(1997)MOM
t
(stock momentum risk premium) =the mo-
mentum risk-factor returns at time tbut this factor is not
signicant and the results do not change.
SATit =a
it +b1MKTt+b2SMBt+b3HMLt+b4MOMt+e
it:(4)
To summarize, the results show an annual four-factor alpha of
10.8% over a 15-year period from a stock fund long in strong
customer satisfaction stocks and short in weak customer
satisfaction stocks. The results are unchanged even when
more risk factors are added to the model, but the results are at
odds with the conclusions, though not necessarily with the
numerical results of Jacobson and Mizik (2009a), Bell,
Ledoit, and Wolf (2014), and Ittner, Larcker, and Taylor
(2009). There are several reasons for this discrepancy beyond
the longer time period we used.
First, consider Ittner, Larcker, and Taylor (2009),
who report that customer satisfaction has incremental
informational value beyond nancial information but nd
no abnormal returns or evidence of mispricing. However,
levels of customer satisfactionaredisregardedintheir
portfolio construction; only changes are deemed relevant.
The assumption is that customer satisfaction levels are
already impounded in stock pricesconsistent with the
efcient market hypothesis of zero alpha. Aside from
ignoring information implied by CLV (Equation 1),
which posits that (high) levels are critical because of
their exponential impact on value, there is the logical
challenge of assuming that levels of customer satisfaction
are already reected in share prices when that very
assumption (market efciency) is being tested. High serial
correlation notwithstanding, the importance of levels (in
combination with changes) is that rms with high levels of
customer satisfaction (and thus, retention) create much greater
customer equity value than rms with low levels of customer
satisfaction, even if the latter rms also have increasing
satisfaction.
Two other studies (Bell, Ledoit, and Wolf 2014; Jacobson
and Mizik 2009a) obtain large positive alphas but likewise
conclude that these abnormal returns cannot be distinguished
from chance. Jacobson and Mizik (2009a) also apply a con-
ditional model with time-varying parameters and nd an
alpha of zero.6However, such a result would require the
correlation between alpha and beta to be enormous(to cite
Lewellen and Nagel 2006). We nd that correlation to be no
greater than .36, but that may not matter because time has
passed and the prediction of zero alpha is now testable and
shown to be inaccurate.
The Earnings Effect
Systematic underpricing of customer satisfaction could be due
to multiple factors, but recent research has suggested earnings
surprises as a major factor (Ngobo, Casta, and Ramond 2012;
OSullivan and McCallig 2009). That is, investors do not re-
act to information about customer satisfaction per se either be-
cause they are not aware of it or because they place little trust
in it until it is substantiated by nancials. Instead, they react to
earnings surprises, which, in turn, may have been caused by
strengthening or weakening customer satisfaction. Thus, we
examine the impact of customer satisfaction on both earnings
and earnings surprises.
Over time, the portfolio has held substantially more longs
than shorts (920 vs. 88). In Table 2, the rst two mean scores
reect average earnings surprises for the long (3.3 cents per
share) and short (.9 cents per share) books, respectively. The
long positions had positive quarterly earnings surprises 60%
of the time. That may not, in itself, suggest anything out of the
ordinary, because most S&P stocks tend to beat earnings
estimates (Jakab 2012). However, only 37% of the longs had
negative earnings surprises. The short positions had negative
6Unlike the unconditional (or static) CAPM, the conditional
CAPM suggests that beta (or exposure to market risk) varies over
time, as rational investors anticipate future investment conditions
that could change adversely and thus alter their investment strategies
intertemporally. According to the logic of the conditional CAPM,
investor expectations lead to a correlation between beta and market
risk premium that also varies over time, and that can help better
explain contemporaneous market risk premium (or alpha). Either
short-window regressions testing for the existence of intertemporal
variation in beta (Adrian and Franzoni 2009; Lewellen and Nagel
2006) or a version of the CAPM that integrates lagged terms for
market risk premium and beta (alpha
t-1
·beta
t-1
) (Jagannathan and
Wang 1996) are used to test the conditional CAPM. Although results
from these studies also vary, and although a debate in the nance
literature over the usefulness of the conditional CAPM in better
explaining asset pricing anomalies remains active, most research has
found only a marginally better explanation of risk premiums through
the conditional CAPM. Some studies have found that the conditional
CAPM decreases the size of the market risk premium moderately or
slightly but without alpha falling to zero or becoming statistically
insignicant (Adrian and Franzoni 2009; Jagannathan and Wang
1996). Other studies have found virtually no impact of the condi-
tional CAPM, suggesting that the correlation between alpha and beta
over time would need to be enormousto explain away asset
pricing anomalies (Lewellen and Nagel 2006). Yet all agree that the
intertemporal covariance between beta and alpha needs to be very
large for the conditional CAPM to improve explanation of market
risk premiums from the CAPM.
98 / Journal of Marketing, September 2016
quarterly earnings surprises 49% of the time, much more
frequently than S&P 500 rms; they also had 49% positive
earnings surprises, less common than S&P 500 rms. At rst
glance, these ndings would seem to suggest that strong
customer satisfaction is more closely associated with posi-
tive earnings surprises than weak customer satisfaction with
negative earnings surprises. Yet looking at quarterly earnings
surprises in dollar terms (the second set of mean scores in
Table 2, with mean earnings per share [EPS] multiplied by
common shares outstanding), negative earnings surprises are
larger: approximately $62.4 million on average for the shorts
versus +$26.1 million for the longs.
Accordingly, rms with strong (weak) customer sat-
isfaction are more likely to have positive (negative) earnings
surprises, suggesting that investors react to tangible con-
sequences of customer satisfaction and not necessarily to
information about customer satisfaction itself. This effect
seems to overwhelm the potentially countervailing impact
from the accounting practice of expensing, rather than
capitalizing, investments in customer service improve-
ments. Unless the increased service improvement is due to
technology investment or some other capital expenditure,
expensing can lead rms with strong short-term earnings
growth and weak customer satisfaction to be overvalued by
the market, whereas rms with weak short-term earnings
growth but strong customer satisfaction would be under-
valued. It is easy to nd conditions under which this can
happen.
For example, on the one hand, employee turnover or cost
cutting in customer service may have a positive effect on
short-term earnings but a detrimental impact on customer
satisfaction, thus eroding future earnings power. On the other
hand, a rms investment in better customer service may
have a positive effect on customer satisfaction and lead to
greater future earnings potential, butif the investment is
expensed instead of capitalized over timeshort-term earn-
ings would be negatively affected. If, under these circum-
stances, investors pay attention to accounting information but
not to prior and subsequent customer satisfaction, the resulting
valuation bias might be considerable. Even if 37% of the
companies with strong customer satisfaction did indeed have
negative earnings surprises, that is not enough to offset the
cumulative positive earnings surprises for companies with
strong customer satisfaction.
In light of these results, it seems that there is a relationship
between customer satisfaction and earnings surprises, which
we conrm by combining long and short positions and esti-
mating the following regression:
SURitt+3=b0+b1CSit +b2DCSit +b3ðCSit ·DCSitÞ+e
it,(5)
where SUR is earnings surprise for company i at time t, mea-
sured each quarter following the inclusion of the stock in
either the long or the short book as quarterly EPS minus the
Institutional Brokers Estimate System quarterly median
analyst forecast; CS is the annual ACSI customer satisfac-
tion score; DCS is changes from the prior year in the annual
ACSI customer satisfaction score; and CS ·DCS is the
interaction effect between the two.7Table 3 presents the
results. The effect of ACSI and the interaction between ACSI
changes and levels are signicant. The ACSI changes alone
are not signicant. This result underscores the importance of
the standard customer value equation (Equation 1) in that
changes are conditional on high values to have an effect.
Unless there is a change from a level that is already high, there
is no effect on earnings surprises from a change in customer
satisfaction.
If customer satisfaction leads to earnings surprises in
the manner suggested by the estimates in Equation 5 and in
TABLE 2
Customer Satisfaction Portfolio Earnings Surprises
N M Min Max SD
Long Positions
EPS -median EPS forecast 920 .0330 -6.3867 3.9900 .5999
EPS -median EPS forecast ($ millions) 920 26.1132 -6,663.75 2,477.3910 432.8846
Short Positions
EPS -median EPS forecast 88 .0086 -2.1033 3.7200 .6029
EPS -median EPS forecast ($ millions) 88 -62.3988 -3,497.84 697.6367 447.2678
Notes: Earnings surprise =quarterly earnings per share -median earnings per share forecast (source: Bloomberg L.P.). A t-test of the difference
between long and the short positions shows that the long positions had signicantly more positive earnings surprises than the short positions
(p<.05, one-tailed).
TABLE 3
The Effect of Customer Satisfaction on
Earnings Surprises
Coefcient SE
a.030 .019
CS .006* .003
DCS -.795 .582
CS ·DCS .120* .062
*p<.05 (two-tailed).
Notes: CS =annual ACSI customer satisfaction score; DCS =changes
in the annual ACSI customer satisfaction score; CS x DCS =the
interaction effect between CS and DCS.
7Because ACSI scores are constant from the quarter they are
released until new results come out 12 months later, and because
earnings surprises are quarterly, the same independent variable is
applied to four different dependent variables further and further into
the future (current quarter, next quarter, two quarters ahead, etc.).
Stock Returns on Customer Satisfaction / 99
Tables 2 and 3, it might also have an impact on earnings
themselves, as we postulate in Equation 6:
EPSitt+3=b0+b1CSit +b2DCSit +b3ðCSit ·DCSitÞ+e
it,(6)
where EPS is earnings for company i at time t, measured each
quarter following the inclusion of the stock in either the long or
the short book as quarterly earnings per share; CS is the annual
ACSI customer satisfaction score; DCS is changes in the annual
ACSI score; and CS ·DCS is the interaction effect between the
two. Table 4 provides the results, which are consistent with the
effect on earnings surprises. The level of customer satisfaction
is signicant, as is the interaction between levels and changes.
Changes by themselves are not signicant. Unless changes are
accompanied with high levels of customer satisfaction, they do
not seem to have an impact on earnings.
We have examined earnings, earnings surprises, and how
they are affected by customer satisfaction. Next, we investi-
gate whether stock returns on earnings explain the effect of
customer satisfaction returns. If they do, the substantial alphas
shown previously might be reduced. A new ve-factor model
proposed by Fama and French (2015) enables us to estimate
the effect of earnings returns on customer satisfaction returns.
In the new model, earnings are dened as annual revenues
minus cost of goods sold; interest expense; and selling,
general, and administrative expenses at time t -1. The factor,
termed RMW by Fama and French, is the difference in
returns between diversied portfolios with robust versus
weak protability. In that sense, it is similar to the customer
satisfaction portfolio, which, by going long and short, respec-
tively, also becomes the difference in returns between robust
and weak customer satisfaction. The FamaFrench model
also includes an investment factor (CMA); however, we do
not discuss it further because it turns out to be irrelevant in
this circumstance. Accordingly, we specify the ve-factor
model as follows:
SATit =a
it +b1MKTt+b2SMBt+b3HMLt
+b4RMWt+b5CMAt+e
it:
(7)
An important nding is that earnings mitigate the effect of
customer satisfaction on stock returns. The estimate of abnormal
returns is reduced from .009 to .007, or from an annualized
return of 10.8%8.4%. However, it is not surprising that cus-
tomer satisfaction and protability are related. The literature
supports such a nding (Morgan and Rego 2006), though it is
not always the case that high customer satisfaction has a positive
impact on protability. Yet when buyers have choice and are
reasonably well informed, it is usually the case.
The Winsorized estimates in Table 5 are consistent with
the full sample results, but the effect of earnings is smaller
and the abnormal returns larger. This means that the outliers
are more strongly correlated to earnings returns (as mea-
sured by RMW). Again, this follows from the exponential
effects encapsulated in Equation 1: stocks with the highest
(lowest) return had the highest (lowest) correlation to earn-
ings returns.
In addition to the models common in the academic -
nancial literature, there are also commercially available mul-
tifactor models. As we have seen, when controlling for risk
factors such as market, size, value, protability, momentum,
and so forth, the customer satisfaction alpha survives. Does it
also survive when using an expanded multifactor model? The
most well-known of these types of models is the model
introduced by Rosenberg and McKibben (1973) and made
available by Barra Inc., until the rm was acquired by MSCI in
2004. Table 6 presents the results from applying the most
current commercially available Barra USE4 comprehensive
trading model, using customer satisfaction returns from July
2006 to November 2013. This analysis was performed by
Morgan Stanley.
Of these factors, the estimated factor loading betas for
technology and retail are signicant at the .05 level. This is not
surprising, because technology stocks in general did well during
this time period and because the ACSI tracks both consumer
technology and many retail rms. No other factors were sig-
nicant. Approximately 80% of the variance was unaccounted
for, taking all factors into account. Accordingly, there seems to
be no known factor that can explain the market outperformance;
it is due to either customer satisfaction or some yet unknown
factor that is correlated with customer satisfaction.
Generalizability
We are not aware of previous research reporting higher risk-
adjusted returns for an intangible asset other than those found
TABLE 4
The Effect of Customer Satisfaction on EPS
Coefcient SE
a.630*** .028
CS .032*** .005
DCS -.673 .882
CS ·DCS .249** .094
**p<.01 (two-tailed).
***p<.001 (two-tailed).
Notes: CS =annual ACSI customer satisfaction score; DCS =changes
in the annual ACSI customer satisfaction score; CS x DCS =the
interaction effect between CS and DCS.
TABLE 5
Five-Factor Model Results
FamaFrench
Five-Factor CAPM
Winsorized (95%)
FamaFrench
Five-Factor CAPM
a.007*** .008***
a(%) 8.4% 9.6%
MKT .718*** .626***
SMB .213** .124*
HML -.142 -.059
RMW .392** .234*
CMA .117 .059
*p<.05 (one-tailed).
**p<.01 (one-tailed).
***p<.001 (one-tailed).
Notes: MKT =monthly market returns excess of the risk-free rate;
SMB =size risk-factor returns; HML =value risk-factor returns;
RMW =protability risk-factor returns; CMA =investment risk-
factor returns.
100 / Journal of Marketing, September 2016
in this study. For example, annual abnormal returns from
research and development have been estimated at 4.6% (Lev
and Sougiannis 1996) and 3.5% (Edmans 2011) from employee
satisfaction. Because the customer satisfaction returns are so
different than the returns on other intangibles, they call for extra
efforts in validation. We are fortunate in the sense that we have
access to comparable data in which both settings and time
periods are different and in which we can use trading rules based
on the same principle but with a different operationalization.
Ultimately, observed empirical regularities need not only to be
conrmed in time and space but also to be understood in terms
of theory such that the conditions under which they occur can
be specied (competitive consumer markets, repeat purchases,
market failure, etc.) and under which they do not. Next, we
discuss the former condition.
Conrmatory Evidence
Do these results generalize across time and space? As to the
question of whether the results hold in a different stock
market and are robust with respect to simpler but perfectly
replicable trading rules, the analysis is extended to the United
Kingdom, where the returns are not obtained from audited
prots, but from back testing. Although customer satisfaction
data comparable to the United States do not exist for many
countries, there is a similar customer satisfaction index in the
United Kingdom: the National Customer Satisfaction Index
(NCSI-UK, www.ncsiuk.com). It uses the same latent struc-
tural equation model as the ACSI and it is also updated
quarterly. The U.K. data were available from 2007 to 2011
from the customer satisfaction universe of publicly traded
rms (74 companies). We obtained the returns with simple
trading rules that generated two portfolios: the rst portfolio
comprised the top 50% of companies in customer satisfac-
tion, and the second contained the bottom 50%, all equally
weighted.8We purchased stocks at the end-of-month closing
price for the quarter in which NCSI-UK results were announced
and held all stocks for one calendar year. Each quarter, we
examined the stocks with new customer satisfaction infor-
mation and adjusted the portfolios accordingly. If a stock was
no longer in the top 50%, we moved it from the rst portfolio
to the second portfolio, and vice versa. Figure 3 provides
the model-free returns (trading costs excluded) on the two
portfolios compared with returns on the FTSE 100.
Except for a short period following inception, the top
50% customer satisfaction portfolio outperformed both the
FTSE and the bottom 50% portfolio. From August 2007 to
April 2011, the top 50% portfolio earned a (model-free)
return of 59%. The bottom 50% portfolio returned 13%,
and the FTSE had a return of 6%, all model free. The
estimated CAPM alpha was .010 (12% annualized) for the top
50% portfolio, compared with .005 (6% annualized) for the
bottom 50% portfolio. The former was signicant at the .01
level; the latter was insignicant. The market beta coefcient
was signicant at the .001 level for both portfolios; it was very
high for the bottom 50% portfolio at 1.163 and substantially
lower for the top 50% portfolio at .759.
The estimated alpha for the top 50% customer satisfaction
portfolio is large and, at an annualized rate of 12.0%, similar
to the U.S. alpha at 10.8%. The alpha of the bottom 50%
customer satisfaction portfolio is not signicant, but it might
still seem odd that stocks of the weakest 50% customer sat-
isfaction companies did better than market. The reason is
probably an effect of beta (i.e., the regression coefcient for
the FTSE 100), as the market moved up sharply after the
20072008 recession. As we have mentioned, the beta for
the bottom 50% portfolio was 1.163, compared with .759 for
the top 50% portfolio. In other words, high customer sat-
isfaction reduced market exposure, whereas low customer
satisfaction exacerbated it. Overall, the real audited protab-
normal returns to customer satisfaction in the United States
correspond well to the back-tested paper prots returns in the
United Kingdom.
TABLE 6
Barra USE4 Model Results
Factor Coefcient
Technology .30*
Retail .24*
Small-large .20
Volatility large cap .12
OTM put .24
EM-dev. .10
Momentum .06
Financials -.13
Japan -.05
Energy -.07
Industrials .11
Health care -.05
Europe .06
Illiquidity .02
Credit .04
Market -.06
*p<.05 (two-tailed).
Notes: Technology =long technology growth stocks; Retail =long con-
sumer, developed market bias; Small-large =long small caps,
short large caps; Volatility large cap =long volume, large cap
exposure; OTM put =long tail protection, downside exposure
management; EM-dev. =emerging markets bias, commodity bias;
Momentum =long momentum stocks, trading style; Financials =
short nancials/real estate, long growth/short value; Japan =short
bias Japan, long carry trades; Energy =short global growth, long
transports/developed markets; Industrials =long cyclicals/long
deep value stocks; Health care =short health care biotech,
short cyclicals; Europe =long bias Europe, long dollar protection;
Illiquidity =long less liquid stocks, long credit; Credit =long credit
sensitive equities; Market =short global equity. The Barra model
results were prepared by Morgan Stanley.
8We assigned all subsidiaries covered by the NCSI-UK the stock
prices of their parent company and deleted all private or government-
owned companies from the data set prior to analysis. The U.K. data
include only one manufacturing industry (autos), so the concen-
tration bias that would result from taking the top 50% rms in the
ACSI does not exist. Even though data relating to the FamaFrench
risk factors are not readily available in the United Kingdom, there
is no a priori reason to suggest that these factors would be relevant
when they were not signicant in the U.S. study (see also Grifn
2002). Accordingly, we analyze the U.K. data with the standard
SharpeLintner CAPM.
Stock Returns on Customer Satisfaction / 101
Boundary Conditions
As for boundary conditions and disconrmatory evidence
relating to our ndings, let us examine the periods of time
when the results deviated from the overall time period.
Figure 4 shows that from September 2012 to August 2013,
the stocks of companies with strong customer satisfaction
underperformed the S&P 500. As a result, the satisfaction
portfolio returns for the full year of 2013 were lower than
market. In other words, strong customer satisfaction did not
produce a positive alpha during this period or during 2013.
Because this was the only year of underperformance, it might
be justied to consider it a chance incident. Nonetheless, the
time period seems long enough to warrant closer examina-
tion, albeit not persistent enough for statistical inference.
In 2013, the long-short customer satisfaction portfolio
returned 21%, its long book was up 25% (both model free), and
the S&P 500 was up 30%. According to Xydias (2013), not only
was the stock market rally one of the broadest in history, but
rms with weak balance sheets outperformed those with strong
balance sheets. There were other idiosyncrasies as well. The
most heavily shorted stocks dramatically outperformed the
market indices. Specically, the 100 most heavily shorted stocks
in the Russell 3000 index were up 34%, while the index itself
was up 18%. The best performing stocks also had lower cus-
tomer satisfaction than the average stock. The best 25 stock
performers in the ACSI universe had an exceptionally high
average return of 95%, but their average ACSI score was only
75, lower than the average of the longs in the portfolio (82) as
well as the overall ACSI average (77).
With little or no economic justication, there were also
massive reversals in stock returns, from high to low and vice
versa. During the 20112012 time period, the top 25 ACSI
universe stocks in terms of returnswith an average return of
95% in 20122013returned only 1.7% on average over the
previous 12 months. The S&P 500 stocks had a similar re-
versal. As Table 7 shows, the top 100 stock performers in the
12-month 20122013 period had dramatically greater returns
when compared with the prior 12-month 20112012 period
and greatly outperformed the S&P average. Specically, the
top 100 returned 60% in 20122013, but these stocks
returned only 8% in 20112012. The corresponding returns
for the remainder of the S&P were 16% and 13%, and for the
total S&P 500, 14% and 24%.
Given that there was a signicant reversal in stock returns
both for the ACSI universe and for the S&P 500, what do the
data suggest regarding the justication of these reversals on
the basis of revenue and earnings? As Table 7 shows, the top
25 ACSI universe stock performers in 20122013 had an
average trailing 12-month revenue growth of 6.7%. This was
lower than the long stocks in the portfolio, which had a
revenue growth of 7.8%, but slightly higher than both the
average ACSI stock (at 5.6%) and the average S&P 500
stock (at 5.7%). Accordingly, the higher share prices could
hardly be justied from revenue growth. Were they driven by
earnings growth? As Table 7 shows, the trailing 12-month
EPS growth for the top 25 ACSI stock performers was 33%
during the underperformance period, compared with 12% for
the ACSI universe, 11% for the long portfolio, 10% for the
average S&P 500 stock, and 17% for the top 100 S&P 500
stock performers in 20122013. That is, the top stock
performers did have substantially higher EPS growth.
However, the multiples for the EPS increase on stock price were
FIGURE 3
Cumulative Returns on £100 Invested in Customer Satisfaction: High NCSI-UK Portfolio, Low NCSI-UK
Portfolio, and the FTSE 100
£159.3
£112.7
£94.3
–60%
–40%
–20%
0%
20%
40%
60%
80%
Top 50% Bottom 50% FTSE
Apr 2007
Aug 2007
Dec 2007
Apr 2008
Aug 2008
Dec 2008
Apr 2009
Aug 2009
Dec 2009
Apr 2010
Aug 2010
Dec 2010
Apr 2011
Notes: The high NCSI-UK portfolio consists of the top 50% of measured companies in customer satisfaction; the low NCSI-UK portfolio consists of
the bottom 50% of measured companies in customer satisfaction.
102 / Journal of Marketing, September 2016
extraordinarily high. For example, the top 100 S&P 500 stock
performers had EPS growth 63% greater than the average S&P
rm, but a stock return 153% greater. The top 25 ACSI stock
performers had EPS growth 194% greater than the long stocks,
but a stock return 417% greater.
If increasing stock price cannot be justied by revenue
growth but rather is driven by a very large increase in multiples
of EPS growth, did these stock prices rise because they were
cheap? Table 8 provides some statistics answering this
question.
The top 25 ACSI stocks in 20122013 had an average P/E
ratio of 14.98 in 20112012. This compares with a P/E ratio
of 17.64 for the long book and a P/E ratio of 18.24 for the
S&P 500. Thus, if the P/E is interpreted as the relative price
of a stock, then the long stocks were notably more expensive
than the top 25 stock performers. Again, the S&P shows a
similar pattern. The average S&P 500 stock was 28% higher
in P/E ratio than the subsequent top 100 stock performers.
Overall, the evidence suggests that the market exhibited
unusual characteristics during the 20122013 time period.
In addition to an unusually broad rally and above-market
performance by stocks with weak balance sheets and by those
most shorted, there was a signicant reversal in stock returns
between 20112012 and 20122013 for both the ACSI
universe and the S&P 500. It is not that the stocks that did
well in 20112012 did poorly in 20122013, but they did not
do as well as the (previous) underperformers of 20112012.
There is no evidence to suggest that the reversal is due to
differences in revenue increases, but some of the reversal
seems to be due to differences in earnings increases. There
were lower P/E ratios for the stocks that gained the most
and evidence that the investors preferred lower-priced stock
even without relating price to earnings. In 20112012, the
long-book customer satisfaction portfolio stocks were priced
(P/E) 100% higher than the 20122013 top 25 ACSI stock
performers. Even though the underperforming period began
FIGURE 4
Customer Satisfaction Portfolio Versus the S&P 500 (September 2012August 2013)
3.3%
16.1%
10%
–5%
0%
5%
10%
15%
20%
25%
Satisfaction portfolio S&P 500
Sep 2012
Oct 2012
Nov 2012
Dec 2012
Jan 2013
Feb 2013
Mar 2013
Apr 2013
May 2013
Jun 2013
Jul 2013
Aug 2013
TABLE 7
Satisfaction Portfolio and S&P 500 Stock Returns, Revenue Growth, and EPS Growth
20112012
Avg. Ret. %
20122013
Avg. Ret. %
2013 T12M
EPS Growth
2013 T12M
Rev Growth
ACSI
Absolute D
ACSI %
D
ACSI
Mean
ACSI
Median
Satisfaction Long Portfolio
Top 25 in ACSI 1.7% 94.6% 33.2% 6.7% .63 .9% 75.0 77.0
Bottom 111 in ACSI 17.3% 15.3% 7.3% 5.3% .87 1.2% 78.8 80.0
Total ACSI 14.4% 29.9% 12.1% 5.6% .82 1.2% 76.9 79.0
Total satisfaction portfolio 17.0% 18.3% 11.3% 7.8% 2.24 2.9% 81.9 83.0
S&P 500
Top 100 S&P 500 8.4% 60.1% 16.5% 7.8% ——
Bottom 349 S&P 500 15.6% 13.4% 8.3% 5.1% ——
Total S&P 500 14.0% 23.8% 10.1% 5.7% ——
Notes: Avg. Ret. % =average stock return percentage; T12M EPS Growth =trailing 12-month revenue growth; T12M Rev Growth =trailing 12-month
revenue growth. For the satisfaction portfolio, among the 235+companies measured in ACSI 136 were both publicly traded and had data
available during the entire time period examined, resulting in a comparison between the Top 25 in satisfaction and the Bottom 111 in
satisfaction. For the S&P 500, 449 companies were available during the entire time period examined, resulting in a comparison between the
Top 100 in returns and the Bottom 349.
Stock Returns on Customer Satisfaction / 103
in the fourth quarter of 2012, the calendar year returns for
2012 were consistent with prior years, with an above-market
return of 12%.
In summary, the large number of market idiosyncrasies
during the 20122013 time period accounts for a large por-
tion of the customer satisfaction underperformance during
this time period. It was an unusual market in many ways:
stocks with weak balance sheets did better than stocks with
strong balance sheets; companies with weak customer sat-
isfaction did better than companies with strong customer
satisfaction; there was a massive reversal in stock price
growth from companies that previously had strong growth
to companies that had had weak growth; the most heavily
shorted stocks had much higher return than market; stocks
with low P/E ratios performed better than stocks with high
P/E ratios; and stocks with low absolute prices also did bet-
ter. The better-performing stocks did not have superior revenue
growth, but they did show greater percentage earnings growth.
However, they also appear to have been unusually well-rewarded
by investors for those earnings.
Even with the many oddities of the 20122013 stock
market, it does suggest that a stock-picking strategy based on a
rather limited universe (rms tracked by the ACSI) is not likely
to outperform the market all the time. Unless there is large
variance in temporal customer satisfaction and, thus, much
turnover in the portfolio, holdings will eventually be domi-
nated by high-priced stocks. The price differential, whether in
absolute price or relative to earnings, may well make lower-
priced stocks appear to be bargains, largely independent of
their fundamentals. At some point, however, fundamentals will
matter again and demand will shift back. It is not clear how
long that usually takes, but in view of the nding that the
portfolio produced higher-than-market returns for 14 years and
less-than-market returns for 12 months, the period is probably
rarely much longer than 1 year.
Discussion
The ndings presented in this study suggest that risk-adjusted
stock returns on customer satisfaction are signicantly above
market and that these abnormal returns are robust to a variety of
alternative explanations such as size, value, and momentum
risk factors as well as data snooping. Strategically, this places
emphasis on customer satisfaction as an important intangi-
ble marketing asset, also labeled as an operational customer
mindsetperformance variable by Katsikeas et al. (2016).
Importantly, it also highlights marketings value to the rm
(e.g., Feng, Morgan, and Rego 2015) and, by at least indirect
extension, supports the notion that chief marketing ofcers
matter (e.g., Germann, Ebbes, and Grewal 2015) if these rm
managers are aligned in their understanding (levels and drivers)
of the customerssatisfaction (Hult et al. 2016). The nding that
stock returns on customer satisfaction do beat the market is
robust in that the 15-year time period studied (20002014)
should be long enough to eliminate sector bias as an explanation.
As an additional check, we also estimated models with different
benchmarks, especially for technology and industrial sectors.
The difference in overperformance was marginal whether we
used NASDAQ or the DJIA.
There is a statistically signicant relationship between
customer satisfaction and lagged earnings surprises, and cus-
tomer satisfaction is largely without inuence on contempora-
neous share prices until its effects are manifested in earnings
reports. Customer satisfaction also has an effect on earnings
themselves. Perhaps even more notable is the nding that
earnings returns have an effect on customer satisfaction returns.
When lagged earnings returns are included among the risk
factors, the abnormal returns on customer satisfaction are
reduced from 10.8% per annum to 8.4%.
Equation 1 alludes to the expectation of major abnormal
returns because (1) equity value, expressed as the discounted
net present value of future cash ows from current customers,
can increase greatly even with modest growth in customer
retention, and (2) investors in general do not seem to be aware
of this (see Gupta, Lehmann, and Stuart 2004). Because the
market does not generally value customer satisfaction until
its effects show up in improved company nancials, it is
noteworthy that a marketing intangible can produce returns
much higher than many other types of intangibles, further
underscoring the conclusion that familiarity with marketing
information is not widely spread among equity market par-
ticipants. This may be particularly true with respect to the
huge leverage from loyal customers. For example, Gupta and
Lehmann (2005) nd that a 1% improvement in customer
retention has an effect on customer equity value that is 5 times
greater than a comparable gain in prot margin and 50 times
TABLE 8
Average Returns and P/E Ratios
20112012 Avg. Ret. % 20122013 Avg. Ret. % 20112012 P/E Ratio 20122013 P/E Ratio
S&P 500 14.9% 23.2% 18.24 18.57
ACSI 17.2% 23.9% 15.72 18.25
Longs 15.5% 18.9% 17.64 19.24
S&P Top 100 9.7% 54.4% 14.26 18.25
ACSI Top 25 7.5% 65.9% 14.98 20.28
S&P Bottom 349 16.4% 13.6% 19.46 18.66
ACSI Bottom 111 19.8% 12.6% 15.92 17.71
Notes: Avg. Ret. % =average return percentage (source: Compustat); P/E Ratio =price-to-earnings ratio (source: Compustat). As in Table 6, S&P
500=the total sample of 449 measures available during the period; ACSI=the 136 companies in the ACSI universe for which data were
available; Longs=the long positions within the satisfaction portfolio; S&P Top 100=top 100 performing companies in the S&P 500 (vs. the
Bottom 349); ACSI Top 25=top 25 performing companies in the ACSI (vs. the Bottom 111).
104 / Journal of Marketing, September 2016
greater than the same magnitude of improvement in customer
acquisition cost.
The returns are so different from expected returns that it
might not be appropriate to label them as yet another asset
pricing anomaly. According to Fama and French (2008), all
stock return anomalies are proxies for intangibles and for
net cash ows. The reason that they are considered anomalies
is that they cannot be explained by the CAPM. This is, of
course, true for customer satisfaction as well. It is a predictor
of future cash ows, represents an intangible asset, and is
not explained by the CAPM. However, the similarities end
there. The contribution of customer satisfaction to value
creationespecially if viewed as consumer utility in the
classical economics senseis a more momentous economic
variable relative to anomalies such as value, size, and momentum,
all of which originate from observation, followed by post hoc
conjecture (Fama and French 2004).
The analysis of customer satisfaction and stock returns,
by contrast, did not begin with empirical observation. The
reasoning is not post hoc, but ex ante. Its origins can be traced
to utility theory (Kahneman, Wakker, and Sarin 1997), for-
tied by the marketing models of customer equity (Gupta
and Lehmann 2005; Kumar 2008; Rust, Zeithaml, and
Lemon 2000; Villanueva and Hanssens 2007). It is not that
post hoc reasoning is omitted in its entirety, but empirical
testing has generally trailed theory, not the other way around.
Yet one might ask whether various intangibles and the
anomalies they give rise to could be correlated to each other
and therefore have a common cause. Because our study
does not control for all known anomalies, might omitted
variable bias have affected our results? This is possible, but
unlikely. Green, Hand, and Zhang (2013) identied more
than 300 anomalies and found that they were predominantly
orthogonal, a nding also conrmed by McLean and Pontiff
(2015).
Implications
The most obvious implicationfrom the nding that com-
panies that treat their customers well tend to produce better
returns to their investorsis that rms should generally try to
improve customer satisfaction along with the volatility of
future customer cash ow risks. This is hardly a revolutionary
idea. Indeed, it has been a normative xture in marketing
literature and education for more than 50 years. However, it has
taken on increasing exigency as a result of the proliferation of
the Internet and information and communications technologies
as well as global competition and the rise in global division of
labor. Yet paradoxically, marketing has not gained stature in
the business organization. Rather, the opposite seems to be the
case (Nath and Mahajan 2008; cf. Feng, Morgan, and Rego
2015; Germann, Ebbes, and Grewal 2015).
As suggested by the sheer size of the abnormal returns, the
reward for having satised customers is much greater than is
generally known. By the same token, the value of marketing
information (and specically information about customer sat-
isfaction), which generates abnormal stock returns of approx-
imately 10% per annum, seems underappreciated. According to
Edmans (2011), abnormal returns for other types of intangible
assets (e.g., research and development, advertising, quality of
organization, patents, human capital) range between 4% and
6%. For employee satisfaction, he estimates an alpha of 3.5%.
These numbers suggest that it would be benecial to allocate
more resources to the marketing function, especially because
many of the most effective ways to increase customer sat-
isfaction lie not in product quality but in targeting, market
segmentation, customer service and the management of cus-
tomer relationships, and CLV (e.g., Fornell 2007; Kumar et al.
2008).
If the economic benets of high customer satisfaction in
terms of improved consumer utility and shareholder value
seem apparent, why do many companies fail to improve cus-
tomer satisfaction? Indeed, overall customer satisfaction, on
average, has decreased over the past two years, according to
the ACSI. Social and portable media, online data collec-
tion, voice-recorded automated telephone interviewing and
computer-assisted telephone interviewing, Big Data, and so
forth provide companies with access to more data on cus-
tomers than ever before. The information and communica-
tions technologies revolution has led to extraordinary growth
in data storage, transmission, and displaying. Yet it is
probably safe to say that there has not been corresponding
progress in data processing. Instead, raw data are often used
to calculate the number of likes,or the proportion of cus-
tomers who say they will or will not recommend. Validity and
reliability are concepts rarely mentioned. Calibration toward
objectives is virtually unheard of in this context. The same is
true with corrections for sampling and measurement error,
which would be problematic not only in terms of measurement
accuracy but also with respect to identifying what companies
need to do to increase their customerssatisfaction. Along with
measures and analysis that may be too simplistic, it is also
possible that rms thereby invest too little in customer retention.
This could be particularly damaging because underspending on
customer retention tends to have a more harmful effect than
underspending on customer acquisition (Reinartz, Thomas, and
Kumar 2005).
However, it should be recognized that customer satisfaction
information is not without interpretational challenges. There-
fore, it would be unrealistic to expect that equity markets would
be frictionless with respect to such information. In addition
to the friction associated with arbitrage costs, imperfect infor-
mation, limitations on investorscognitive and reasoning skills,
and institutional rigidities that impair market efciency, cus-
tomer satisfaction is not included in the analysis models most
investors use (Gupta and Lehmann 2005). Consequently, it
would be difcult for equity markets to instantly incorporate
customer satisfaction information. It is also not the case that
strong customer satisfaction always leads to above-market
stock returns. As our data demonstrate, in 20122013 there
was a period when the returns, though strong in an absolute
sense, underperformed the market. Even though the market
exhibited several unusual characteristics during this time,
the somewhat limited stock universe provided by the ACSI,
combined with small temporal variance in customer satis-
faction, eventually produces a portfolio of highly priced stock
vulnerable to temporary reversals in relative fortune during
broad stock rallies. Nevertheless, we may well be at a point
Stock Returns on Customer Satisfaction / 105
when one can envision a debate between marketing and nance
about whether earnings or customer satisfaction belongs among
the risk factors in asset pricing models. Empirically, they are
correlated; therefore, one mitigates the effect of the other. But
which one should be mitigated? What comes rst, earnings or
customer satisfaction? We are not aware of research suggesting
that earnings per se cause customer satisfaction, but there is
ample evidence pointing in the other direction.
REFERENCES
Adrian, Tobias and Francesco Franzoni (2009), Learning About Beta:
Time-Varying Factor Loadings, Expected Returns, and the Con-
ditional CAPM,Journal of Empirical Finance,16(4),53756.
Anderson, Eugene W., Claes Fornell, and Donald R. Lehmann
(1994), Customer Satisfaction, Market Share, and Protability:
Findings from Sweden,Journal of Marketing, 58 (July), 5366.
———,———, and Sanal K. Mazvancheryl (2004), Customer
Satisfaction and Shareholder Value,Journal of Marketing,
68 (October), 17285.
——— and Sattar A. Mansi (2009), Does Customer Satisfaction
Matter to Investors? Findings from the Bond Market,Journal of
Marketing Research, 46 (October), 70314.
Aksoy, Lerzan, Bruce Cooil, Christopher Groening, Timothy L.
Keiningham, and Atakan Yalçin (2008), The Long-Term Stock
Market Valuation of Customer Satisfaction,Journal of Mar-
keting, 72 (July), 10522.
Bell, David R., Olivier Ledoit, and Michael Wolf (2014), A New
Portfolio Formation Approach to Mispricing of Marketing
Performance Indicators with an Application to Customer Sat-
isfaction,Customer Needs and Solutions, 1 (4), 26376.
Black, Fischer (1993), Beta and Return,Journal of Portfolio
Management, 20 (1), 818.
Carhart, Mark M. (1997), On Persistence in Mutual Fund Per-
formance,Journal of Finance, 52 (1), 5782.
Edmans, Alex (2011), Does the Stock Market Fully Value
Intangibles? Employee Satisfaction and Equity Prices,Journal
of Financial Economics, 101 (3), 62140.
Fama, Eugene F. and Kenneth R. French (1993), Common Risk
Factors in the Returns on Stocks and Bonds,Journal of
Financial Economics, 33 (1), 356.
——— and ——— (2004), The Capital Asset Pricing Model:
Theory and Evidence,Journal of Economic Perspectives,
18 (3), 2546.
——— and ——— (2008), Dissecting Anomalies,Journal of
Finance, 63 (4), 165378.
——— and ——— (2015), A Five-Factor Asset Pricing Model,
Journal of Financial Economics, 116 (1), 122.
——— and James D. MacBeth (1973), Risk, Return, and Equilibrium:
Empirical Tests,Journal of Political Economy,81(3),60736.
Feng, Hui, Neil A. Morgan, and Lopo L. Rego (2015), Marketing
Department Power and Firm Performance,Journal of Mar-
keting, 79 (September), 120.
Fornell, Claes (2007), The Satised Customer: Winners and
Losers in the Battle for Buyer Preference. New York: Palgrave
Macmillan.
———, Sunil Mithas, and Forrest V. Morgeson III (2009a), The
Economic and Statistical Signicance of Stock Returns on
Customer Satisfaction,Marketing Science, 28 (5), 82025.
———,———, and ——— (2009b), The Statistical Signicance
of Portfolio Returns,International Journal of Research in
Marketing, 26 (2), 16263.
———,———,———, and M.S. Krishnan (2006), Customer
Satisfaction and Stock Prices: High Returns, Low Risk,Journal
of Marketing, 70 (January), 314.
———, Roland T. Rust, and Marnik G. Dekimpe (2010), The
Effect of Customer Satisfaction on Consumer Spending Growth,
Journal of Marketing Research, 47 (February), 2835.
Germann, Frank, Peter Ebbes, and Rajdeep Grewal (2015), The
Chief Marketing Ofcer Matters!Journal of Marketing,79(May),
122.
Green, Jeremiah, John R.M. Hand, and X. Frank Zhang (2013), The
Supraview of Return Predictive Signals,Review of Accounting
Studies, 18 (3), 139.
Grifn, John M. (2002), Are the Fama and French Factors Global or
Country Specic?Review of Financial Studies,15(3),783803.
Gruca, Thomas S. and Lopo L. Rego (2005), Customer Sat-
isfaction, Cash Flow, and Shareholder Value,Journal of
Marketing, 69 (July), 11530.
Gupta, Sunil and Donald Lehmann (2005), Managing Customers
as Investments: The Strategic Value of Customers in the Long
Run. Upper Saddle River, NJ: Pearson Education.
———,———, and Jennifer Stuart (2004), Valuing Customers,
Journal of Marketing Research, 41 (February), 718.
Holm, Morten, V. Kumar, and Carsten Rohde (2012), Measuring
Customer Protability in Complex Environments: An Inter-
disciplinary Contingency Framework,Journal of the Academy
of Marketing Science, 40 (3), 387401.
Hult, G. Tomas M., V. Forrest Morgeson III, Neil A. Morgan, Sunil
Mithas, and Claes Fornell (2016), Do Managers Know What
Their Customers Think and Why?Journal of the Academy
of Marketing Science, (available electronically June 1), [DOI:
10.1007/s11747-016-0487-4].
Ittner, Christopher D. and David F. Larcker (1998), Are Non-
Financial Measures Leading Indicators of Financial Per-
formance? An Analysis of Customer Satisfaction,Journal of
Accounting Research, 36 (Supplement), 135.
———,———, and Daniel J. Taylor (2009), The Stock Markets
Pricing of Customer Satisfaction,Marketing Science, 28 (5),
82635.
Ivanov, Vladimir, Kissan Joseph, and M. Babajide Wintoki (2013),
Disentangling the Market Value of Customer Satisfaction:
Evidence from Market Reaction to the Unanticipated Com-
ponent of ACSI Announcements,International Journal of
Research in Marketing,30(2),16878.
Jacobson, Robert L. and Natalie Mizik (2009a), The Financial
Markets and Customer Satisfaction: Reexamining Possible
Financial Market Mispricing of Customer Satisfaction,Mar-
keting Science, 28 (5), 81019.
——— and ——— (2009b), Rejoinder-Customer Satisfaction-
Based Mispricing: Issues and Misconceptions,Marketing
Science,28(5),83645.
Jagannathan, Ravi and Zhenyu Wang (1996), The Conditional
CAPM and the Cross-Section of Expected Returns,Journal of
Finance, 51 (1), 353.
Jakab, Spencer (2012), Earnings Surprises Lose Punch,The Wall
Street Journal, (May 6), [available at http://online.wsj.com/
article/SB10001424052702304020104577384304200945934.
html].
106 / Journal of Marketing, September 2016
Kahneman, Daniel, Peter P. Wakker, and Rakesh Sarin (1997),
Back to Bentham? Explorations of Experienced Utility,
Quarterly Journal of Economics, 112 (2), 375406.
Katsikeas, Constantine S., Neil A. Morgan, Leonidas C. Leonidou,
and G. Tomas M. Hult (2016), Assessing Performance Out-
comes in Marketing,Journal of Marketing, 80 (March), 120.
Kumar, V. (2008), Customer Lifetime Value: The Path to Prot-
ability. Boston: Now Publishers, Inc.
——— and Morris George (2007), Measuring and Maximizing
Customer Equity: A Critical Analysis,Journal of the Academy
of Marketing Science, 35 (2), 15771.
——— and Denish Shah (2009), Expanding the Role of Mar-
keting: From Customer Equity to Market Capitalization,
Journal of Marketing, 73 (November), 11936.
——— and ——— (2011), Can Marketing Lift Stock Prices?
MIT Sloan Management Review, 52 (4), 2426.
———, Rajkumar Venkatesan, Tim Bohling, and Denise Beckmann
(2008), Practice Prize ReportThe Power of CLV: Managing
Customer Lifetime Value at IBM,Marketing Science,27(4),
58599.
Lambert, Richard A. (1998), Customer Satisfaction and Future
Financial Performance: Discussion of Are Nonnancial Mea-
sures Leading Indicators of Financial Performance? An Analysis
of Customer Satisfaction,Journal of Accounting Research,36
(Supplement), 3746.
Lev, Baruch and Theodore Sougiannis (1996), The Capitaliza-
tion, Amortization, and Value-Relevance of R&D,Journal of
Accounting and Economics, 21 (1), 10738.
Lewellen, Jonathan and Stefan Nagel (2006), The Conditional
CAPM Does Not Explain Asset-Pricing Anomalies,Journal of
Financial Economics, 82 (2), 289314.
Lintner, John (1965a), The Valuation of Risk Assets and the
Selection of Risky Investments in Stock Portfolios and Capital
Budgets,Review of Economics and Statistics, 47 (1), 1337.
——— (1965b), Security Prices, Risk, and Maximal Gains from
Diversication,Journal of Finance, 20 (4), 587615.
Lo, Andrew W. and A. Craig MacKinlay (1990), Data-Snooping
Biases in Tests of Financial Asset Pricing Models,Review of
Financial Studies, 3 (3), 43167.
Luo, Xueming, Christian Homburg, and Jan Wieseke (2010), Cus-
tomer Satisfaction, Analyst Stock Recommendations, and Firm
Value,Journal of Marketing Research, 47 (December), 104158.
Malshe, Ashwin and Manoj K. Agarwal (2015), From Finance to
Marketing: The Impact of Financial Leverage on Customer
Satisfaction,Journal of Marketing, 79 (September), 2138.
McLean, R. David and Jeffrey Pontiff (2015), Does Academic
Research Destroy Stock Return Predictability?Journal of
Finance, 71 (1), 532.
Merrin, Robert P., Arvid O.I. Hoffman, and Joost M.E. Pennings
(2013), Customer Satisfaction as a Buffer Against Sentimental
Stock-Price Corrections,Marketing Letters, 24 (1), 1327.
Merton, Robert C. (1987), A Simple Model of Capital Market
Equilibrium with Incomplete Information,Journal of Finance,
42 (3), 483510.
Morgan, Neil A. and Lopo L. Rego (2006), The Value of Different
Customer Satisfaction and Loyalty Metrics in Predicting Busi-
ness Performance,Marketing Science, 25 (5), 42639.
Nath, Pravin and Vijay Mahajan (2008), Chief Marketing Ofcers:
A Study of Their Presence in FirmsTop Management Teams,
Journal of Marketing, 72 (January), 6581.
Ngobo, Paul-Valentin, Jean-Francois Casta, and Olivier Ramond
(2012), Is Customer Satisfaction a Relevant Metric for
Financial Analysts?Journal of the Academy of Marketing Sci-
ence, 40 (3), 480508.
OSullivan, Don, Mark C. Hutchinson, and Vincent OConnell
(2009), Empirical Evidence of the Stock Markets (Mis)pricing
of Customer Satisfaction,International Journal of Research in
Marketing, 26 (2), 15461.
——— and John McCallig (2009), Does Customer Satisfaction
Inuence the Relationship Between Earnings and Firm Value?
Marketing Letters, 20 (4), 33751.
Peng, Chi-Lu (Edward), Kuan Ling Lai, Miao-Ling Chen, and An-
Pin Wei (2014), Investor Sentiment, Customer Satisfaction
and Stock Returns,European Journal of Marketing, 49 (5/6),
82750.
Raithel, Sascha, Marko Sarstedt, Sebastian Scharf, and Manfred
Schwaiger (2011), On the Value Relevance of Customer Sat-
isfaction: Multiple Drivers and Multiple Markets,Journal of the
Academy of Marketing Science, 40 (4), 50925.
Reinartz, Werner, Jacquelyn S. Thomas, and V. Kumar (2005),
Balancing Acquisition and Retention Resources to Maximize
Customer Protability,Journal of Marketing, 69 (January),
6379.
Rosenberg, Barr and Walt McKibben (1973), The Prediction of
Systematic and Specic Risk in Common Stocks,Journal of
Financial and Quantitative Analysis, 8 (2), 31733.
Rust, Roland T., Valarie A. Zeithaml, and Katherine N. Lemon
(2000), Driving Customer Equity: How Customer Lifetime Value
is Reshaping Corporate Strategy. New York: Free Press.
Sharpe, William F. (1965), Risk-Aversion in the Stock Market:
Some Empirical Evidence,Journal of Finance, 20 (3), 41622.
Srinivasan, Shuba and Dominique M. Hanssens (2009), Market-
ing and Firm Value: Metrics, Methods, Findings, and Fu-
ture Directions,Journal of Marketing Research, 46 (June),
293312.
Tuli, Kapil and Sundar G. Bharadwaj (2009), Customer Sat-
isfaction and Stock Returns Risk,JournalofMarketing,
73 (November), 18497.
Venkatesan, Rajkumar and V. Kumar (2004), ACustomer
Lifetime Value Framework for Customer Selection and Resource
Allocation Strategy,Journal of Marketing, 68 (October),
10625.
Villanueva, Julian and Dominique M. Hanssens (2007), Customer
Equity: Measurement, Management and Research Opportunities.
Boston: Now Publishers, Inc.
White, Halbert (2000), A Reality Check for Data Snooping,
Econometrica, 68 (5), 1097126.
Xydias, Alexis (2013), Junk Stocks Spur Broadest Equity Advance
Since 1995,Bloomberg.com, (May 20), [available at http://
www.bloomberg.com/news/2013-05-19/leverage-loved-in-equities-
spurs-broadest-u-s-rally-since-1995.html].
Stock Returns on Customer Satisfaction / 107
... Because customer feedback and experience with a firm can influence a firm's future sales, profitability, and cash flow, investors should consider customer metrics when making investment decisions (Anderson & Mansi, 2009;Fornell et al., 2016;Luo et al., 2014). Higher levels of customer satisfaction, measured as ACSI, lead to abnormal stock returns of approximately 10% per annum (Fornell et al., 2016). ...
... Because customer feedback and experience with a firm can influence a firm's future sales, profitability, and cash flow, investors should consider customer metrics when making investment decisions (Anderson & Mansi, 2009;Fornell et al., 2016;Luo et al., 2014). Higher levels of customer satisfaction, measured as ACSI, lead to abnormal stock returns of approximately 10% per annum (Fornell et al., 2016). ...
... To determine whether the SMPS is associated with the firm's market performance (e.g., Fornell et al., 2006Fornell et al., , 2016, the following empirical model is used: ...
Article
Full-text available
Customer loyalty and satisfaction increase product sales, protect market share, and lower marketing costs, which potentially leads to greater returns on investment and cash flows. Therefore, investors take customer feedback into account when making investment decisions. Online social media platforms such as Facebook and Twitter have emerged as alternative sources for obtaining customer feedback information promptly and at a low cost. This study develops a new measure, the Social Media Promoter Score (SMPS), which combines several indicators of customers’ attitudes toward a company derived from detailed sentiment and content analyses of social media. Using a semiparametric model of customer loyalty index based on the generalized additive model (GAM), we found that both positive and negative social media metrics about customers’ attitudes were significantly associated with the customer loyalty index. Importantly, SMPS was also significantly associated with an increase in firms’ market performance. These findings suggest that SMPS can be a valuable measure to complement the existing customer metrics such as the ACSI. Theoretical contributions to research on the marketing–finance interface and managerial implications are discussed.
... We obtain customer satisfaction data from the American Customer Satisfaction Index (ACSI) (e.g., Fornell et al., 2016), which has been developed and is maintained by the National Quality ...
... Second, our findings extend the literature on customer satisfaction (Anderson et al., 2004) by identifying an important antecedentperceived greenwashing -and the corresponding underlying mechanism (i.e., corporate hypocrisy) of customer dissatisfaction. In so doing, respond to calls in the literature to examine factors that increase or decrease customer satisfaction (e.g., Anderson et al., 2004;Fornell et al., 2016;Luo & Bhattacharya, 2006). In fact, our study makes an empirical contribution to the literature that has argued that hypocritical acts may sometimes be necessary or unavoidable (Brunsson, 2002;Christensen, Morsing, & Thyssen, 2013;March, 2007). ...
Article
Full-text available
We investigate the impact of perceived greenwashing on customer satisfaction. Unlike prior research that largely examines customer perceptions associated with irresponsible behavior, we focus on cases where firms overcommit and/or do not deliver on promised socially responsible actions. We theorize that this type of greenwashing is associated with lower customer satisfaction because customers perceive greenwashing through the lens of corporate hypocrisy. Using data from the American Customer Satisfaction Index (ACSI) for U.S. companies during the period 2008–2016, we document a negative link between perceived greenwashing related to green product innovation (GPI) and the ACSI index. We demonstrate that this effect is primarily triggered by corporate policies exceeding the corresponding implementation actions and not by lower levels of implementation. We also show that a firm’s capability reputation mitigates the negative effect of greenwashing on customer satisfaction. Moreover, we conduct an experiment and provide evidence confirming that GPI greenwashing is in fact perceived by customers as corporate hypocrisy.
... First, customer satisfaction has been identified as a consequence of PCSR (Bello, Jusoh, & Nor, 2020;He & Li, 2011;Lee, Han, Radic, & Tariq, 2020) and PFI (Kunz, Schmitt, & Meyer, 2011;Mahmoud, Hinson, & Anim, 2018). Second, customer satisfaction has been found to be a long-term antecedent of future financial performance (Fornell, Morgeson, & Hult, 2016;Gruca & Rego, 2005;Guenther & Guenther, 2020;Van Doorn, Leeflang, & Tijs, 2013). Thus, we expect customer satisfaction to link PCSR and PFI to firms' future financial earnings. ...
... The strong customer loyalty of a firm should be reflected in the firm's economic returns because it ensures a steady future cash flow (Reichheld & Sasser, 1990). Satisfied customers are essential for any seller in a competitive market if customers' repurchase is a significant portion of total revenue (Fornell et al., 2016). ...
Article
Full-text available
Previous research on corporate social responsibility (CSR) and firm innovativeness and their impact on financial performance has focused on firms’ actions (i.e., what firms do). However, how customers perceive these firm activities have not been fully explored; there is a lack of research particularly on the long-term effects of these actions. Consequently, the present study investigates the effects of customer-perceived CSR and firm innovativeness on financial earnings, both in the short and long term. Firm actions, if meaningful, should impact customer perceptions of a firm, which affect customer satisfaction and the firm’s earnings consequently. Using panel data from service firms, our analysis indicates that perceived firm actions positively influence future earnings through customers’ overall evaluations of a firm. Furthermore, the results reveal a carryover effect of perceived actions in the long term. The present research also indicates that customers’ positive perceptions of firm actions do not directly impact financial earnings; however, they do impact earnings through customer satisfaction. This emphasizes the importance of communicating innovation, and particularly CSR activities, to customers.
... More satisfied customers tend to remain more loyal to a retailer, purchase more frequently from the retailer, and tend to recommend it to others (Ha, 2006;Menidjel et al., 2019). Earlier research has pointed out that customer satisfaction is also a predictor of improved organisational performance and high stock returns (Fornell et al., 2016(Fornell et al., , 2020. ...
Article
Full-text available
Despite the growing attention on enhancing purchase intent, there is no coherent framework for determining retail customers’ purchase intent. This study aims to contribute to marketing theory and practice by proposing and testing a comprehensive theoretical model that incorporates direct, mediating, and moderating relationships among preeminent constructs to produce decisive insight into purchase intent. Partial least squares‐based structural equation modelling (PLS‐SEM) was deployed to test research hypotheses and analyse the data collected from 515 customers. The findings reveal that customer relationship proneness has a positive impact on customer satisfaction, customer trust, and purchase intent. In addition, customer satisfaction is found to have a positive effect on customer trust. Customer trust, in turn, positively affects customer purchase intent. Customer trust fully mediates the association of customer satisfaction and purchase intent, and partially mediates the association of relationship proneness and purchase intent. Of our seven selected moderators, only marital status and income moderate the association of customer relationship proneness and customer satisfaction. A customer’s tendency to engage in relationships with retailers or brands depends on various factors that are related to customer characteristics. To develop customer loyalty and cultivate long‐term relationships, retailers are recommended to segment their customers based on their respective relationship proneness and relevant demographic characteristics, in particular, marital status and income, and target each segment with customized relationship marketing activities.
Article
Customer satisfaction management is increasing in importance within the insurance industry. In particular, to define a customer-oriented strategy, installing digital applications based on technologies, e.g. including artificial intelligence or cloud computing, ranks among the major strategic challenges. Against this background, the aim of this paper is to take an integrated perspective on managing customer satisfaction and the digital transformation. Towards this end, we identify and assess a set of digital applications, as a result of a comprehensive review of 106 academic papers and publications of the industry and supervisory authorities. We illustrate the opportunities to increase customer satisfaction and emphasise their impact on insurers at four major customer touch points: contract conclusion, contract modifications, the event of damage and further contacts. Our results are strategic measures to strengthen the position for sales and marketing, to simplify standard processes and to increase efficiency and interaction with the customer.
Article
The main purpose of this study is to identify relationships between customers’ service evaluations embedded in online reviews and customer satisfaction and demonstrate the unique role of price and location in the restaurant industry. Approximately 250,000 online reviews of restaurants were analyzed to answer (1) How cognitive and affective evaluations written in a review influence customer satisfaction and (2) How price plays a moderating role in the relationship between text-based service evaluations and customer satisfaction. By incorporating spatial dependency, geographically weighted regression analysis was conducted to identify the pure impact of affective and cognitive evaluations of restaurant services and to develop a sophisticated satisfaction prediction model. This study found the significant impact of cognitive and affective service evaluations on customer satisfaction, which was moderated by price. Moreover, this study highlighted geo-spatial dependency by demonstrating that spatial auto-correlation matters only for restaurants in the lowest ($) and highest ($$$$) price groups.
Article
An increasing number of business-to-business (B2B) service firms have transitioned to recurring revenue-based solutions. These subscription B2B solutions are becoming increasingly common, yet offer challenges for long-term renewal if value is not consistently realized by the customer. To address this concern, customer success (CS) management has emerged. CS management is based on regular proactive action taken by the seller to (a) educate, prepare, and engage customers for value co-creation; (b) demonstrate the value delivered by the solution; and (c) provide a channel for advocacy on behalf of customers within the service-providing firm. Our findings highlight the under-researched topic of CS in B2B settings. Specifically, we propose the CS function and role as a structural alternative to within-person (i.e., cross-functional) ambidexterity and emphasize the ability of a CS focus by service firms to complement existing firm operations in value creation efforts. Our case study analysis provides a multilevel perspective (i.e., executive, functional role employees, and customers) via in-depth interviews that offer unique insights on “how parts of the service-sales system work together.” Overall, CS is growing as a practice that propagates value to the customer via ongoing success with solutions while improving service-firm renewal and growth of subscription business.
Article
Purpose The study aims to examine the moderating role of variety-seeking behaviour between customer engagement and its antecedents (customer satisfaction and customer value). Further, this study also tests the existence of the value-satisfaction-engagement behaviour chain. Design/methodology/approach The perception of 262 respondents was used to examine the hypothesis using the structural equation modelling approach. Findings The moderation effect of variety-seeking behaviour between customer satisfaction and customer engagement was found to be significant. Also, customer satisfaction fully mediated the relationship between perceived value and customer engagement, hence, empirically validating the value-satisfaction-engagement model in the retailing context. Research limitations/implications The findings draw managers' attention towards the segment of consumers who are more likely to be engaged, thus helping managers develop a more efficient and focussed strategy to achieve customer engagement. The result also suggests that variety-seeking buyers may not get engaged even after satisfaction. Originality/value This paper is among the first to empirically test the moderating role of variety-seeking behaviour to achieve customer engagement.
Article
Full-text available
Firms struggle to respond to product recalls and manage post-recall customer satisfaction. In three studies, we examine the impact of firms’ remedy choice on satisfaction and provide evidence that firms’ post-recall remedy efforts are often not optimal. In Study 1 (field study), we estimate the longer-term effects of remedy on different satisfaction metrics and show that offering full remedy is much more important for low and high (vs. medium) brand equity firms, especially when failure severity is high. In Study 2 (experiment), we find further evidence that the positive impact of full remedy on satisfaction is moderated by brand equity in a u-shaped fashion. Finally, Study 3 (experiment) provides further evidence that the relationship between remedy and brand equity is contingent on failure severity. The findings contribute to the literature on firms’ management of negative relationship events and provide managers with the empirically grounded 5R guidelines to make better remedy decisions in response to product recalls.
Article
Previous research has shown that visitor satisfaction varies across nationalities, that tourists from some countries are more loyal than others, and that visitor satisfaction enhances destination loyalty. However, the relationships between nationality, satisfaction, and loyalty have not been examined together in the past. In this study, we fill this research gap by examining the mediating effect of visitor satisfaction on the relationship between nationality and destination loyalty. For this purpose, we use a joint modeling regression approach. Statistical testing favors this estimation strategy, and the results uncover substantial direct effects of nationality on loyalty. Satisfaction-mediated, or indirect, impacts are not visible in most of the cases (with the exception of Danish travelers). The study offers some practical managerial implications.
Article
Full-text available
The ability of a firm’s managers to understand how its customers view the firm’s offerings and the drivers of those customer perceptions is fundamental in determining the success of marketing efforts. We investigate the extent to which managers’ perceptions of the levels and drivers of their customers’ satisfaction and loyalty align with that of their actual customers (along with customers’ expectations, quality, value, and complaints). From 70,000 American Customer Satisfaction Index (ACSI) customer surveys and 1068 firm (manager) responses from the ACSI-measured companies, our analyses suggest that managers generally fail to understand their firms’ customers in two important ways. First, managers systematically overestimate the levels of customer satisfaction and attitudinal loyalty, as well as the levels of key antecedent constructs such as expectations and perceived value. Second, managers’ understanding of the drivers of their customers’ satisfaction and loyalty are disconnected from those of their actual customers. Among the most significant “disconnects,” managers underestimate the importance of customer perceptions of quality in driving their satisfaction and of satisfaction in driving customers’ loyalty and complaint behavior. Our results indicate that firms must do more to ensure that managers understand how their customers perceive the firm’s products and services and why.
Article
Full-text available
Research in marketing has increasingly focused on building knowledge about how firms' marketing contributes to performance outcomes. A key precursor to accurately diagnosing the value firms' marketing creates is conceptualizing and operationalizing appropriate ways to assess performance outcomes. Yet, to date, there has been little conceptual development and no systematic examination of how researchers in marketing should conceptualize and measure the performance outcomes associated with firms' marketing. The authors develop a theory-based performance evaluation framework and examine the assessment of such performance outcomes in 998 empirical studies published in the top 15 marketing journals during the 1981–2014 period. The results reveal a large number of different performance outcome measures used in prior empirical research that may only be weakly related to one another, making it difficult to synthesize findings across studies. In addition, the authors identify significant problems in how performance outcomes in marketing are commonly conceptualized and operationalized. They also reveal several theoretically and managerially important performance areas in which empirical knowledge of marketing's impact is limited or absent. Finally, they examine the implications of the results, provide actionable guidelines for researchers, and also suggest a roadmap for more systematically improving research practice in the future.
Article
Full-text available
Research in marketing has increasingly focused on building knowledge about how firms' marketing contributes to performance outcomes. A key precursor to accurately diagnosing the value firms' marketing creates is conceptualizing and operationalizing appropriate ways to assess performance outcomes. Yet, to date, there has been little conceptual development and no systematic examination of how researchers in marketing should conceptualize and measure the performance outcomes associated with firms' marketing. The authors develop a theory-based performance evaluation framework and examine the assessment of such performance outcomes in 998 empirical studies published in the top 15 marketing journals from 1981 through 2014. The results reveal a large number of different performance outcome measures used in prior empirical research that may be only weakly related to one another, making it difficult to synthesize findings across studies. In addition, the authors identify significant problems in how performance outcomes in marketing are commonly conceptualized and operationalized. They also reveal several theoretically and managerially important performance areas in which empirical knowledge of marketing's impact is limited or absent. Finally, they examine the implications of the results, provide actionable guidelines for researchers, and suggest a road map for systematically improving research practice in the future.
Article
Full-text available
This study empirically investigates marketing department power in U.S. firms throughout 1993-2008 and assesses its impact on firm performance. Using a new objective measure of marketing department power and a cross-industry sample of 612 public firms in the United States, the results reveal that, in general, marketing department power increased during this time period. Furthermore, the analyses show that a powerful marketing department enhances firms' longer-term future total shareholder returns beyond its positive effect on firms' short-term return on assets (ROA). The findings also reveal that a firm's long-run market-based-asset-building and short-run market-based-asset-leveraging capabilities partially mediate the effect of a firm's marketing department power on its longer-term shareholder value performance and fully mediate the effect on its short-term ROA performance. This research provides new insights for marketing scholars and managers with regard to both marketing's influence within the firm and how investments in building a powerful marketing department affect firm performance.
Article
Not all firms choose to have a chief marketing officer (CMO) in their top management teams (TMTs). This research investigates factors associated with this choice and whether CMO presence/absence in the face of these factors affects firm performance. Findings based on a multi-industry sample of 167 firms over a five-year period (2000–2004) show that innovation, differentiation, branding strategy, diversification, TMT functional experience in marketing, and the chief executive officer being an outsider are associated with the likelihood of CMO presence in the TMT. Furthermore, the authors find that CMO presence in the TMT has neither a positive nor a negative impact on firm performance. The authors discuss the implications of these findings for theory and practice.
Article
Do investments in customer satisfaction lead to excess returns? If so, are these returns associated with higher stock market risk? The empirical evidence presented in this article suggests that the answer to the first question is yes, but equally remarkable, the answer to the second question is no, suggesting that satisfied customers are economic assets with high returns/low risk. Although these results demonstrate stock market imperfections with respect to the time it takes for share prices to adjust, they are consistent with previous studies in marketing in that a firm's satisfied customers are likely to improve both the level and the stability of net cash flows. The implication, implausible as it may seem in other contexts, is high return/low risk. Specifically, the authors find that customer satisfaction, as measured by the American Customer Satisfaction Index (ACSI), is significantly related to market value of equity. Yet news about ACSI results does not move share prices. This apparent inconsistency is the catalyst for examining whether excess stock returns might be generated as a result. The authors present two stock portfolios: The first is a paper portfolio that is back tested, and the second is an actual case. At low systematic risk, both outperform the market by considerable margins. In other words, it is possible to beat the market consistently by investing in firms that do well on the ACSI.
Article
Are there economic benefits to improving customer satisfaction? Many firms that are frustrated in their efforts to improve quality and customer satisfaction are beginning to question the link between customer satisfaction and economic returns. The authors investigate the nature and strength of this link. They discuss how expectations, quality, and price should affect customer satisfaction and why customer satisfaction, in turn, should affect profitability; this results in a set of hypotheses that are tested using a national customer satisfaction index and traditional accounting measures of economic returns, such as return on investment. The findings support a positive impact of quality on customer satisfaction, and, in turn, profitability. The authors demonstrate the economic benefits of increasing customer satisfaction using both an empirical forecast and a new analytical model. In addition, they discuss why increasing market share actually might lead to lower customer satisfaction and provide preliminary empirical support for this hypothesis. Finally, two new findings emerge: First, the market's expectations of the quality of a firm's output positively affects customers’ overall satisfaction with the firm; and second, these expectations are largely rational, albeit with a small adaptive component.
Article
This article examines whether country-specific or global versions of Fama and French's three-factor model better explain time-series variation in international stock returns. Regressions for portfolios and individual stocks indicate that domestic factor models explain much more time-series variation in returns and generally have lower pricing errors than the world factor model. In addition, decomposing the world factors into domestic and foreign components demonstrates that the addition of foreign factors to domestic models leads to less accurate in-sample and out-of-sample pricing. Practical applications of the three-factor model, such as cost of capital calculations and performance evaluations, are best performed on a country-specific basis.