Identifying News Effects on Trades and Prices

One can think of the information in news as having two components. The first component
is a common-knowledge (or “mean”) part: all agents agree about the appropriate impact of this
first part on the exchange rate. In macro models of news in currency markets, this first part is the
whole story: it fully characterizes the instantaneous adjustment of exchange rates to news. The
second component is the part whose implication for the exchange rate is not common knowledge.
It is this second part that is impounded in exchange rates via induced trading. Suppose, for
example, that all agents do not have access to the same technology for transforming the macro
data into an exchange rate forecast. The resulting inferences drawn are not known by the
marketmakers a priori. How do marketmakers aggregate the information in these inferences? The
answer from microstructure theory is that they learn from the sequence of submitted orders over
time. In this case, price adjusts instantaneously to the marketmaker’s rational expectation of the
market’s interpretation (this is the first of the two components), and then goes through a period of
gradual adjustment caused by the sequence of transacted orders.8
With respect to response time, remember that the announcements that we address here are
scheduled, so participants can plan their responses (conditional on realizations) in advance. This
is likely to lower response lags considerably relative to unscheduled news. In this sense, our tests
for whether market responses to news are protracted are conservative tests. Note, too, that the
model that we present below is capturing average responses (i.e., the total average response,
including the induced order flow effects). This is distinct from second-moment effects, i.e.,
effects on volatility. In fact, in response to news, order flow and price are also (jointly) more
volatile, which adds a dimension to the first-moment analysis that we do here. Addressing
directly these second-moment effects would take us too far afield from our central question, so
we do not take it on within this paper.