sacrifice the initial 50 observations, which corresponds to around one year of data. We
are therefore left with 250 observations, or roughly 5 years, for the analysis. Running the
above procedure for all three themes returns 18 events for ‘‘recession’’, 14 for ‘‘oil price’’,
and 15 for ‘‘inflation’’.
We then investigate what happens to cumulative returns, realized volatility, and
implied volatility (as measured by the VIX) of the S&P 500 in the time window of
20 toþ60 days around each event. Figures 11.3–11.5 show the average development
for each theme, respectively. As can be seen, each event is on average preceded by a dip
in cumulative returns.
There are two factors to explain this effect. For one, private investors might be
expected to react with a lag. For other, the results published by Google Trends, and
consequently the rates of change we computed, relate to the week just ended, so the few
days preceding each event might already be influenced by intraweek activity of private
investors. Notwithstanding, there is an immediate further drop in the first days after the
event, followed by a negative drift for almost the remainder of the time window. The
impact on realized and implied volatility is basically the mirror image of the impact on
returns, consistent with the volatility asymmetry evidence. However, the scale of this
impact is considerably larger making it an even more interesting phenomenon.
11.3.2 Who’s in the market when it becomes volatile?
When observing volatile markets, the question arises, who is in the market when it
becomes volatile? According to the theory we have built so far, the increase in volatility
should be caused by private investors and thus we would expect to see them on the
market.
To check this we used a dataset of the Estonian stock market NASDAQ OMXT.
We study this dataset since it has a unique feature: it includesalltransactions on the
market and moreover allows us to identify all distinct investors in the market at different
times and distinguish between individual and institutional investors, as well as locals and
foreigners. We would expect to see more individuals trading on the market when the
market becomes more volatile.
The first task is to measure volatility asymmetry in the OMXT index for the period
we have the transaction data (i.e., 2004–2008). Surprisingly, we do not observe any
asymmetry for the period by using similar APARCH models to those we used for our
international comparison. Our previous data show that such cases exist especially in
emerging markets. Estonia is a small emerging market with a relatively short history of
stock exchange, so this observation does not contradict the findings of our international
study. Particularly, there is very low or sometimes practically non-existent analyst
coverage of listed companies; and the market is quite young (remember the increasing
trend of asymmetry). In any case we can still see who is in the market when it becomes
volatile.
As a reult of lack of a volatility index, we estimate volatility from an APARCH
model. We count the number of individual and institutional investors as well as new
investors who enter the market. We calculate the share of individual investors, the share
of trades done by individual investors, and the share of turnover generated by individual
investors compared with the market total.
As can be seen from the chart, individual investor participation remains quite
264 News and risk