way how private investors can impact the stock market and discuss a convenient metric
of their behavior. This exercise might be helpful to portfolio managers, who are often
concerned about the ‘‘little man’s’’ actions, which are argued to be more susceptible to
swings of mood, especially in periods of market stress.
There are reasons to believe, frequently based on insights from behavioral finance,
that private investor demand is more attention-driven than a systematic investment
approach should be. Private investors tend to follow simple heuristics, like picking
stocks they have positive associations with or the ones recommended by friends or
neighbors. This does not necessarily imply (as some would be happy to believe) that
they will inevitably be driven out of the market. In a rather provocative experiment
Gigerenzer (2007) showed that asking random people on the street for names of stocks
that they know and subsequently investing in them can be a very successful strategy. But
even if simple investment strategies fail in the long run, the next generation of inexper-
ienced investors will readily replace their frustrated predecessors. The reliance on simple
heuristics of many investors implies, however, that looking at the typical array of a stock
analyst’s indicators, be it fundamental or technical, is not likely to say much about the
direction private investors are headed in, simply because it is not what they themselves
look at.
A sizable number of studies have attempted to address this issue. Barber and Odean
(2008) name extreme returns, trading volume, and news and headlines as suitable
indicators, which have been developed to a varying extent in the literature. News and
headlines especially proved to be very fertile ground for research, originating in numer-
ous event studies (Liu, Smith, and Syed, 1990; Barber and Loeffler, 1993; Ferreira and
Smith, 1999; Arena and Howe, 2008), through time-series and cross-sectional regres-
sions (Mitchell and Mulherin, 1994; Fang and Peress, 2009), and developing into the
kind of linguistics-based analysis presented in this volume (Tetlock, 2007). Other
authors examined factors derived more from a corporate finance point of view, such
as the size of the advertising budget (Grulon, Kanatas, and Weston, 2004; Dong, 2008;
Chemmanur and Yan, 2009).
None of the above, however, is a direct measure of attention; they are all proxies,
which run into the fundamental problem of distinguishing between active and passive
effects or, in marketing parlance, between push and pull.
To understand the difference, consider the following simple case of trying to predict
the number of guests at a party. One might take the number of invitations sent as a
(passive) estimate, but few would argue that the number of positive confirmations
(which involve an active response from the addressee) would do a much better job.
Certainly, proxies are ubiquitous in economics and finance, where many phenomena
are not directly observable at all, and they rest on the assumption (motivated by theory
or empirical findings) that active and passive effects are robustly correlated. In situations
largely depending on human psychology like attention or sentiment such correlations
might, however, prove illusory or unstable over time. Therefore, in such circumstances
direct measures are of particular value.
We argue that such a direct measure exists in the case of private investor attention,
based on internet usage. It is presently rather uncontroversial to assume that most
people rely on the internet for information, also concerning investment, and they get
to that information by using search engines. Tracking the flow of search queries thus
arguably brings one as close as it gets to what is on people’s minds. This is exactly the
262 News and risk