Finance and Economics Discussion Series
Divisions of Research & Statistics and Monetary Affairs
Federal Reserve Board, Washington, D.C.
Consumer Sentiment, the Economy, and the News Media
Mark Doms and Norman Morin
2004-51
NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS)
are preliminary materials circulated to stimulate discussion and critical comment. The
analysis and conclusions set forth are those of the authors and do not indicate
concurrence by other members of the research staff or the Board of Governors.
References in publications to the Finance and Economics Discussion Series (other than
acknowledgement) should be cleared with the author(s) to protect the tentative character
of these papers.
Consumer Sentiment, the Economy, and the News Media
Mark Doms
Federal Reserve Bank of San Francisco
Norman Morin
Board of Governors of the Federal Reserve System
September 2004
Abstract
The news media affect consumers’ perceptions of the economy through three channels. First, the
news media convey the latest economic data and the opinions of professionals to consumers. Second,
consumers receive a signal about the economy through the tone and volume of economic reporting.
Last, the greater the volume of news about the economy, the greater the likelihood that consumers
will update their expectations about the economy. We find evidence that all three of these channels
affect consumer sentiment. We derive measures of the tone and volume of economic reporting,
building upon the R-word index of The Economist. We find that there are periods when reporting on
the economy has not been consistent with actual economic events, especially during the early 1990s.
As a consequence, there are times during which consumer sentiment is driven away from what
economic fundamentals would suggest. We also find evidence supporting that consumers update
their expectations about the economy much more frequently during periods of high news coverage
than in periods of low news coverage; high news coverage of the economy is concentrated during
recessions and immediately after recessions, implying that “stickiness” in expectations is
countercyclical. Finally, because the model of consumer sentiment is highly nonlinear, month-to-
month changes in sentiment are difficult to interpret. For instance, although an increase in the
number of articles that mention “recession” typically is associated with a decline in sentiment, under
certain conditions it can actually result in an increase in various sentiment indexes.
We wish to thank Margaret Macleod, Kristen Hamden, Niels Burmester, Mathew Wilson, Brian Rowe, and Ashley Maurier
for research assistance. We want to especially thank Jon Eller for his outstanding efforts in helping us to complete this
project. We also wish to thank Gwen Eudey, Chris Carroll, Spencer Krane, Jose Lopez, Martha Starr, Justin Wolfers, Joyce
Zickler, and numerous colleagues at the Board and FRB San Francisco, and seminar participants at the 2004 ASSA meetings
and the FRB Chicago for helpful comments and suggestions. We also thank Carol Corrado who encouraged us and helped
provide the resources for this project. The opinions expressed in this paper are those of the authors and do not necessarily
reflect the views of the Federal Reserve System.
I. Introduction
Results from the consumer sentiment surveys conducted by the Conference Board and the
University of Michigan are closely watched and widely reported. One reason for the wide coverage is
the public perception of a connection between sentiment and consumption; in fact, several studies
have found such a connection.
1
Over the past several years, consumer sentiment has received much
attention in light of 9/11, accounting scandals, war, and the recent recession.
Figure 1.1 shows the University of Michigan’s composite measure of consumer sentiment from
January 1978 through May 2003.
2
Although this and other consumer sentiment series receive a fair
amount of attention, the research on consumer expectations about the economy is quite limited, with
the exception of inflation expectations.
3
This paper attempts to close the gap between the amount of
attention that sentiment receives and the amount that is known about how sentiment is formed. In
analyzing sentiment, we draw on the growing literature on information theory and decision making,
especially the literature on “sticky expectations” and “rational inattention.” This literature
emphasizes the costs and constraints in obtaining data and making decisions with that data. For
instance, several papers extend Shannon’s (1948) seminal work in information theory (see, for
example, Sims (2003) and Mascaroni (2003)). Many other papers that model decision making use
different mechanisms that can generate “sticky expectations,” such as Reis (2003) and Carroll (2003
and 2004).
One of the more interesting aspects of our work, we believe, is our attempt to explicitly
incorporate the role of the media in influencing consumer sentiment about the economy.
4
This seems
only natural because consumers are likely to form expectations largely based on what they hear from
the media in addition to their own personal experiences.
5
We derive several measures of the tone and
volume of economic reporting and relate these to sentiment. Our models allow news sentiment
through three channels. The first is through the dissemination of economic statistics and opinions of
experts. Second, consumers receive a signal about the economy through the tone and volume of
1
Several authors have examined the relationship between sentiment and consumption, one of the most recent is
Souleles (2004). Souleles finds that consumer sentiment is related to consumption even after controlling for a
host of factors, similar to the results of Carroll, Fuhrer, and Wilcox (1994).
2
Most of the series in the graphs in this paper stop in May 2003, the last date for which we have data on media
reporting.
3
There is a vast and growing literature on how inflation expectations are formed, and the consequences of the
different models.
4
In this paper we use sentiment measures from the University of Michigan’s Survey of Consumers. Results
from using the sentiment measures from the Conference Board are similar.
5
The alternative would be for consumers to wade through government press releases about the economy, such
as the monthly employment report, industrial production, consumer prices, and so on. However, we do not
believe that a measurable portion of the population consistently gleans their information on the economy by
reading these sources directly.
1
economic reporting, and this signal may not be consistent over time with actual economic events. As
Sims (2003) shows, the information theory model provides an explanation why the tone and volume
of economic reporting affect sentiment above and beyond the economic information contained in the
reporting. For instance, the headline “Recession Possible” is likely to elicit a greater negative
response in people’s views about the economy than an article entitled “Economic Conference
Presents Diverse Views” in which the possibilities of a recession are discussed in the last paragraph.
This signal process may be why over the years public officials have expressed misgivings about using
the word “recession”; a famous example is Alfred Kahn’s remarks in which he substituted the word
“banana” for “recession.”
The final channel through which the media influence sentiment is by affecting the likelihood that
consumers will update their expectations, a tenet of the sticky expectations and rational inattention
literatures. As suggested by Carroll (2003) in his study of inflation expectations, the greater the
volume of news coverage, the more likely expectations will be updated. Expectations may be
updated more frequently during periods of high news coverage for several reasons. The first is that
the costs of acquiring information about the economy are likely to be lower when news coverage is
high; all else equal, lower costs will increase how frequently people sample information (as is
Mascaroni (2003) and Reis (2003)). Another reason for a link between the intensity of news coverage
and the likelihood of updating expectations stems from the models of Akerloff, Dickens, and Perry
(2001) and Gabaix, Laibson, and Moloche (2003): Consumers may be more likely to read articles
with headlines like “Recession Possible” because such headlines suggest the information in the article
may be related to their own financial futures. Regardless of the reasons, we find that expectations are
much less “sticky” during periods of high news coverage than in periods of low news coverage.
Given the potential importance of the media for consumer sentiment we investigate how closely
the media’s coverage of the economy tracks the state of the economy. Several possible reasons that
the reporting on the economy may deviate from the economic fundamentals include the effect on
economic news coverage of the political cycle, the relative importance of other news events, the
novelty of the economic data, and the incentive to make economic news more alluring to readers.
6
Our initial motivation to investigate the possibility of inconsistencies in how the media cover the
economy stems from two data sources. The first, shown in figure 1.2, is the R-Word index from The
Economist magazine, a quarterly index of the number of articles in the Washington Post and The New
6
With regard to this last point, a business reporter told us, in a bit of tongue in cheek, that what sells are articles
that either describe how everyone is going to get rich or articles that describe how everyone is going to become
poor.
2
York Times that mention “recession.”
7
What we found striking from this admittedly crude series was
the size of the spikes in the early 1990s compared with the 1980s when the economy was in much
worse shape. Similarly, as shown in figure 1.3, we were surprised by the response to the question
"During the last few months, have your heard of any favorable or unfavorable changes in business
conditions?" from the Michigan Survey. More respondents reported hearing “unfavorable” news than
“favorable” news in nearly all periods, although the 1980s and 1990s was marked by tremendous
overall economic expansion.
8
Also, we were surprised by how quickly the share of respondents
hearing unfavorable news could increase.
To measure the effect that economic reporting has on consumers’ perceptions of the economy, we
need to construct quantitative measures of the volume and tone of economic reporting. How negative
or positive is the article about the economy? What is the proper way to aggregate the tone of articles
within a newspaper and then across newspapers and television? Perhaps because the answers to
questions like these are daunting attempts to quantify the tone of economic reporting have been
limited. One of the objectives of this paper is to take a step toward quantifying how economic news
is reported. Our approach builds on of The Economist’s R-word index. We use 30 newspapers and
search for articles that contain certain words and phrases. We next filter the results based on a
number of criteria and then construct indexes weighted by circulation. This approach is relatively
simple, but we are able to demonstrate that our indexes indeed represent the volume and tone of
economic reporting. Additionally, the results suggest a strong correlation between the newspaper-
based indexes and various measures of consumer sentiment. Further, our indexes are also strongly
correlated with the series on “favorable” and “unfavorable” news heard in figure 1.3.
We created an R-word index like The Economist’s with some modifications. First, we require the
word “recession” or “economic slowdown” to appear in the headline or first paragraph of the article
because we found that these articles tended to portray some negative aspect of the economy, such as
high unemployment, budget difficulties, low profits, and how people cope when the economy sours.
As a further refinement, we purged all of the “recession” articles that contained references to foreign
economic activity in order to avoid the possibility of counting stories on recessions or slowdowns
abroad as bad news about the U.S. economy. We also, extended our index to include 28 papers in
addition to the Washington Post and The New York Times. Also, we were able to compute a similar
7
Similarly, Carroll (2003) collects the number of articles that mention the word “inflation” in front-page stories
from The New York Times and the Washington Post.
8
Just why unfavorable news is heard outweighs favorable news being heard is an interesting question. Gassner
(1999) argues that the media have a strong propensity, and history, to publish stories that instill fear, as would
articles that dwelled on the possibility of a recession.
3
