The Wiley Finance Series : Handbook of News Analytics in Finance

(Chris Devlin) #1

Dow Jones 30 index in the period from January 2007 to September 2009. The quantities
under consideration are the share of positive respectively negative news in the total for
the given day. Sentiment scores are taken from Newssift, an online tool powered by the
Financial Times. Interestingly, there is often a significant positive correlation between
theshareof negative news and thenumberof news items overall, and conversely higher
share of positive news is associated with a lower number of news items overall. Further-
more, this effect appears to be more pronounced for stocks, which have more news items
on average. Therefore, stocks that are more covered on average (without differentiating
between analysts and other media) are also more susceptible to the ‘‘negative news bias’’.
The ideal argument would thus go as follows: more news means predominantly more
bad news, which makes investor reactions more pronounced, when there is downward
pressure on prices.
International data allow media penetration to be measured, and it turns out that it is
strongly correlated with volatility asymmetry. However, media penetration is also
closely correlated with the level of market development and might not always be a good
proxy for stock market media coverage. Thus when including both GDP/capita and
media penetration in the same regression, the impact of the media seems to disappear.
This can of course be somewhat deceptive as a clear link between the development of the
country and the level of asymmetry is much harder to explain than the link between
the impact of the media and asymmetry. However, the impact of the media on
volatility would be much easier to capture within a market if we had reliable data on
news flow.
International data still allow us to further test the hypothesis of news having a
significant impact on volatility asymmetry: analysts are an important source of informa-
tion for investors and could potentially influence their sentiment. We would expect
analysts to discover the shortcomings of companies and the media to communicate
their discoveries. In the case of good news, analysts might not get the same media
attention as in the case of disappointing news. Thus we might expect to see the co-
influence of the media and analysts to volatility asymmetry. As there are usually more
analysts in developed markets, the conclusion also fits the finding of higher volatility
asymmetry in developed markets.
As already mentioned, our data show a significant positive correlation between
asymmetric volatility and analyst coverage. The effect is still present when controlling
for other factors (e.g., the level of GDP/capita and the media); see Table 11.1. We
conclude that better coverage of listed companies helps to draw more attention to
possible shortcomings in a firm’s operations in the case of bad news and helps to react
more quickly to the news. The finding is also supported by previous work of Hong, Lim,
and Stein (2000) who argue that stocks with low coverage tend to react less precisely to
bad news compared with high-coverage stocks.
Our results indicate that analysts and the media could cause volatility asymmetry but
this can only happen if they can persuade at least some investors to trade more
erratically during down moves. The question is: ‘‘Who might these investors be?’’
Hens and Steude (2009) suggest that volatility asymmetry can be caused by investors’
preferences. Shefrin (2005) proposes biased expectations as a possible explanation. Since
individual investors are more prone to be biased than institutional investors, we would
expect large volatility asymmetry in markets where the share of individual investors is
higher. This might be the situation for more developed markets.


260 News and risk

Free download pdf