recent developments in high frequency financial ... - Index of
recent developments in high frequency financial ... - Index of
recent developments in high frequency financial ... - Index of
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Modell<strong>in</strong>g f<strong>in</strong>ancial transaction price movements: a dynamic <strong>in</strong>teger count data model 185<br />
for HAL a richer GLARMA(2,2) yields the best goodness <strong>of</strong> fit <strong>in</strong> terms <strong>of</strong> the<br />
Schwarz criterion.<br />
For all estimates, the dispersion parameter κ −1/2 is significantly different from<br />
zero so that we have to reject the null <strong>of</strong> a truncated-at-zero Poisson distribution <strong>in</strong><br />
favor <strong>of</strong> a Negb<strong>in</strong> distribution. The parameters for the ARMA components are<br />
significant as well. For JBX the estimates for the roots <strong>of</strong> the AR components<br />
are 0.8753 (pla<strong>in</strong> model) and 0.9275 (flexible model), so that stationarity is<br />
guaranteed. For HAL the implicit estimates for the two roots 11 <strong>of</strong> the AR components<br />
are 0.8452 and −0.4631 (pla<strong>in</strong> model) and 0.9649 and 0.5409 (flexible<br />
model), so that stationarity is guaranteed as well. S<strong>in</strong>ce for both stocks roots are<br />
close to one, the GLARMA model predicts a strong persistence <strong>in</strong> the non-zero<br />
absolute price changes.<br />
For both data sets the specification allow<strong>in</strong>g for more flexible news impacts are<br />
superior to the ones with standard news impact curves. The coefficients on ∣ɛi−1∣<br />
are significantly positive, but also the fit <strong>of</strong> the observed dynamics improves over<br />
the specification ignor<strong>in</strong>g the possibility <strong>of</strong> the additional news impact component.<br />
For HAL the effect is considerably lowered by a negative effect on ∣ɛ i − 2∣.<br />
The value <strong>of</strong> Ljung-Box Q-statistics for the standardized residuals e i reported <strong>in</strong><br />
Table 2 <strong>in</strong>dicate that the GLARMA(1,1)-dynamics chosen for the JBX absolute<br />
price changes does not remove autocorrelation <strong>in</strong> the residuals completely. However,<br />
based on the Q(50)-statistic for the flexible GLARMA(2,2)-specification for<br />
HAL the null <strong>of</strong> no serial correlation <strong>in</strong> the residuals cannot be rejected at the 1%<br />
level.<br />
3 Transaction price dynamics and market microstructure<br />
One <strong>of</strong> the fundamental questions <strong>of</strong> the market microstructure theory <strong>of</strong> f<strong>in</strong>ancial<br />
markets is concerned with the determ<strong>in</strong>ants <strong>of</strong> the price process and the specific<br />
role <strong>of</strong> the <strong>in</strong>stitutional set-up. 12 Generally, the goal is to figure out how new price<br />
relevant <strong>in</strong>formation affects the price process. Approaches based on the rational<br />
expectation hypothesis typically assume some k<strong>in</strong>d <strong>of</strong> heterogeneity among the<br />
market participants with respect to their level <strong>of</strong> <strong>in</strong>formation. The correspond<strong>in</strong>g<br />
transaction process generally leads to successive revelations <strong>of</strong> price <strong>in</strong>formation.<br />
This leads to empirically testable hypotheses about the jo<strong>in</strong>t process <strong>of</strong> transaction<br />
price changes and other marks <strong>of</strong> the trad<strong>in</strong>g process, such as transaction <strong>in</strong>tensities<br />
and transaction volume.<br />
Provided that short-sell<strong>in</strong>g is <strong>in</strong>feasible, Diamond and Verrecchia (1987) <strong>in</strong>fer<br />
that longer times between transactions can be taken as a signal for the existence<br />
<strong>of</strong> bad news imply<strong>in</strong>g negative price reactions. The absence <strong>of</strong> a short sell<strong>in</strong>g<br />
mechanism prevents market participants from pr<strong>of</strong>it<strong>in</strong>g by exploit<strong>in</strong>g the negative<br />
<strong>in</strong>formation through correspond<strong>in</strong>g transactions. Therefore, one can expect that<br />
low transaction rates (longer times between transactions) are associated with<br />
<strong>high</strong>er volatility <strong>in</strong> the transaction price process and vice versa.<br />
11 Computed as z1;2 ¼ 1<br />
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi<br />
1 2<br />
2 2 1 þ 4 p<br />
2:<br />
12 See O’Hara (1995) for a comprehensive survey on the theoretical literature on market<br />
microstructure.