This informal regression proxies for the possible impact of the FOMC meeting on interest rate changes. We now turn to the examination of whether the probability of a jump is linked in any way to the FOMC meetings. We achieve this using a modification of our Poisson-Gaussian estimation model depicted in equation (3.3). In the estimation equation (3.3), we specify that the arrival probability of a jump, denoted by the parameter q, be a function of the Fed meetings (ft). It is possible that jumps in the interest rate are caused by Fed actions, and then the information on meetings would determine the probability of a jump taking place.

The equation above accommodates a base level of jump probability Ao, augmented by a Fed dependent attribute, Ai. For the ARCH-diffusion model, wc investigate whether the Fed meetings have an impact on conditional volatility by specifying the ARCH equation with an additional coefficient } on the Fed event, i.e. the variance will be a0 + axe^ + }/*. First, we examine the one-day meetings only.

The results are presented in Table 7. The 1-day meetings appear to have very little impact on the usual levels of jump probability, as seen in the jump-diffusion model. The parameter Xi is not significant. And in fact, the ARCH model evidences a decrease in volatility when a one-day FOMC meeting takes place. We now examine the information impact of the 2-day meetings in Table 8. As in the basic regression in Table 6, this dummy variable proves to be significant, i.e. it increases the probability of a jump. This probability more than doubles im magnitude. In the ARCH model,

We examine via ARCH and jump models whether the FOMC meeting results in a information surprise. The jump model is extended by qt = A0 + \\dayft where ft is the dummy variable for the FOMC meeting.

We examine via ARCH and jump models whether the FOMC meeting results in a information surprise. The jump model is extended by qt — A0 + A2 day ft where ft is the dummy variable for the FOMC meeting. The ARCH model is written the FOMC meeting appears to be of little consequence. Finally, we put both 1 day and 2 day meetings together in one model and ascertain the results in Table 9. The results here are an amalgamation of those from the prior two tables. The two-day meetings result in a sharp increase in the possibility of a jump. The one-day meetings in fact seem to predicate a reduction in conditional volatility. One might speculate that the two-day meetings do result in information surprises, whereas the one-day meetings confirm the market’s forecasts.