doi: 10.1111/j.1467-6419.2009.00603.

THE INFLUENCE OF FUNDAMENTALS ON

EXCHANGE RATES: FINDINGS FROM
ANALYSES OF NEWS EFFECTS
Rafael R. Rebitzky
Leibniz Universität Hannover
Abstract. As we survey the literature of macroeconomic news in the foreign
exchange market, we can by now look back on nearly 30 years of research.
The first studies which analysed news effects on exchange rates were established
in the early 1990s (see, for example, Dornbusch). Almost at the same time
Meese and Rogoff published their influential paper, revealing the forecasting
inferiority in exchange rates of structural models against the random walk. This
finding has shocked the pillars of exchange rate economics and thus cast general
suspicion on research focusing on fundamentals in this field. The eventual rising
popularity of event studies can partly be attributed to the re-establishment of the
raison d’être of exchange rate economics. This work focuses on systematically
surveying this literature with particular respect to its primary goal, i.e. shedding
light on the analytical value of fundamental research. Thus, its major findings
are, first, fundamental news does matter, whereas non-fundamental news matters
to a lesser degree. Second, news influences exchange rates via two separated
channels, i.e. incorporating common information into prices directly or indirectly
based upon order flow. Third, with a few exceptions the impact of fundamental
news on exchange rates is fairly stable over time.
Keywords. Exchange rates; Fundamentals; Market efficiency; News effects

1. Introduction
‘Changes in exchange rates will come to provoke no more comment than changes
in the real price of an airline ticket’! These are the presidential words of Fischer
Black (1986, p. 530) addressing the outlook on the future of the foreign exchange
market in his paper ‘Noise’ (published in the Journal of Finance). Certainly,
Black’s statement is very provoking since he implicitly relates exchange rate
movements largely or even solely to occurring noise. Nonetheless, economic
science is confronted with the mission to link asset price changes to fundamental
causes; otherwise they remain ‘noise’ in public perception even if they are in fact
connected to macroeconomic conditions. One promising technique to accomplish
this nontrivial task is the use of event studies.1
After surveying the literature of empirical exchange rate research Frankel and
Rose (1995, p. 1699) conclude that the primary reason for exchange rate movements

Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704


C 2010 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main

Street, Malden, MA 02148, USA.

 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 681

is connected to unexpected changes in fundamentals. Thus, they attach great

importance to the role of (fundamental) news in the foreign exchange market.
As a matter of principle, the event study approach is a methodology which is
based upon the validity of semi-strong market efficiency (see Fama, 1991). Hence,
influences on financial prices arising from fundamental announcements ought to
be restricted to unexpected variable changes, i.e. news. More precisely, a piece of
news, N kt , comprises the surprise element in a fundamental announcement, which
is in fact the difference between the actual realization, f tk , of a specific variable
and its prior expectation, E t−1 [ f tk ], as follows:
 
Ntk = f tk − E t−1 f tk
Furthermore, fundamentals represent macroeconomic variables, which are
universally accepted as relevant for the specific variable of interest.2 Thus, an
event study presents a mean with which arising news effects in financial market
prices such as stocks, bonds or exchange rates can be quantified. All this can
be traced back to the following understanding: surprises in announcements of
macroeconomic variables influence investors’ mindsets, which in turn lead to
changes in respective expectations of future fundamental outcomes. Hence, if the
market works efficiently, agents’ new perceptions are reflected instantaneously in
accordant price changes and fully anticipated announcements do not lead to asset
price changes; otherwise the assumption of semi-strong market efficiency would
be violated (see Fama, 1991; Frankel and Rose, 1995).
The event study approach looks back on a long history. Dolley (1933) published
an analysis of equity splits and their effects on subsequent stock prices. However,
the event study methodology had not established itself in the stock market research
until the 1970s, when extensive analyses of stock reactions to fundamental news
emerged.3 But then again, it was not until the 1980s that event studies found their
way into the foreign exchange market research (see, for example, Mussa, 1976;
Dornbusch, 1980). This coincided with Meese and Rogoff’s (1983) influential
paper, which shows that out-of-sample exchange rate forecasts produced by
structural models are worse than those obtained by the random walk. Indeed,
this finding put the very basis of exchange rate economics into question, thus
rendering corresponding fundamental research useless. At the same time the asset
approach became widely accepted in the exchange rate literature, and thus exchange
rates were regarded as primarily driven by fundamental expectations (see Frankel
and Rose, 1995). For this reason and with the proliferation of the event study
methodology hope of a rebound of the relevance of fundamentals to the research
on exchange rates arose.4
Due to the disclosure of larger data sets via higher data frequency or the inclusion
of further variables, different event study techniques have alternated over the years.
Whilst in the beginning (rough) news models had dominated the scene, accordant
critics regarding the low frequency of corresponding data series quickly prevailed
and announcement studies emerged. The latter closely resembled the previous
event studies on the stock markets, since they only focus on those days which
follow respective fundamental announcements. In the late 1980s the disclosure of
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
682 REBITZKY

tick-by-tick data provided an enormous amount of data, leading to high-frequency

models. This approach was more similar to the one used in the finance literature
due to the fact that, in the latter field, high-frequency data were used right from the
beginning. However, to put this somewhat odd observation in perspective, unlike
stocks or bonds, exchange rates had been undeniably solely related to the field
of macroeconomics and thus assumed to be exposed only to long-term forces, i.e.
fundamentals. Furthermore, since the latter variables are only measurable monthly
or quarterly in principle, it can be concluded that high-frequency data sets had been
irrelevant to the foreign exchange market literature, although this changed at least
after the establishment of the asset approach and therewith the interest in analysing
shorter term influences on exchange rates. Although this had already been reflected
in the above mentioned announcement studies, the idea came to fruition when the
high-frequency model was used. Then in the 1990s, the hybrid approach, which
is basically based on two separate news channels, asserted itself by means of the
inclusion of order flow in the news analysis. Finally, in recent times joint news
analyses of several asset classes such as stocks, bonds and exchange rates have
emerged, which aim to reveal interdependent asset reactions by analysing their
simultaneous reactions on specific news items.
To this day the event study literature of the foreign exchange market can look
back on nearly 30 years of research. This paper aims at surveying this strand
of literature in order to bring to light its insights on the relationships between
exchange rates and their fundamentals. In doing so, we follow the above mentioned
developments of the event study literature of exchange rates, since the findings are
strongly related to the consistent refinement of the approach.

2. The News Model

The first event studies on the foreign exchange market are based on monthly or
quarterly data, which are known in the literature as news models (see Dornbusch,
1980; Frenkel, 1981; Bomhoff and Korteweg, 1983; Edwards, 1983).

2.1 The Baseline Model

The news model is formulated as follows:

st − E t−1 [st ] = α + β k · Ntk + εt (1)
k
 
with Ntk = f tk − E t−1 f tk and k = 1, . . . , K

The expectation error in the exchange rate, st − E t−1 [st ], is regressed on the
related errors in the fundamentals, f tk − E t−1 [ f tk ], as well as the disturbance
variable, εt , which is independent and identically distributed (iid). Assuming that
market participants are rational, the expectations are based on all relevant public
information up to time t − 1 (see MacKinlay, 1997, p. 13).5 Since exchange
rates disclose an asset-like behaviour, they are supposed to react first only to
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 683

unexpected news, i.e. surprises in fundamentals, and second, instantaneously,

without any related trading in the foreign exchange market (see Frenkel, 1981,
p. 686).
However, three issues prove to be essential for this model setting: first, the choice
of the considered fundamentals, i.e. the macroeconomic variable set which enters
into equation (1) in order to explain exchange rate surprises; second, the way in
which expectations of fundamentals are actually obtained; and third, the way in
which exchange rate surprises are generated at all.
So, with regard to the first issue, structural macroeconomic models based on
conventional fundamentals do indeed deliver respective variables. An alternative
approach relies on the rather ad hoc choice of popular variables, which are seen as
important at the time.
With regard to the way of modelling fundamental expectations, two possibilities
are at hand: generating statistical innovations via univariate time series analysis
(see, for example, Frenkel, 1981; Copeland, 1984a, b; MacDonald, 1985) or via
multivariate analysis (for example Edwards, 1983). Alternatively, one could again
rely on accordant macroeconomic models in order to produce corresponding fun-
damental surprises (for example Edwards, 1982a; MacDonald, 1983; MacDonald
and Taylor, 1992).
As to the third issue, several authors use the forward rate to proxy exchange
rate expectations (for example Frenkel, 1981; Edwards, 1982b; MacDonald, 1985).
Obviously, its indisputable advantage turns out to be the straightforwardness of
attaining these data. However, if the forward rate includes a related risk premium,
by using the forward rate one would implicitly assume that the exchange rate risk
premium is either non-existent or constant over time, as otherwise the risk premium
would be unstable, which in turn would severely bias upcoming estimation results.
However, the existence of a risk premium and its nature is more than controversial
(see, for example, Cavaglia et al., 1994). Accordingly, a constant forward premium
would be captured by the constant α in equation (1) and would therefore not bias
the estimated β parameters. Needless to say, the estimated parameters would also
be unbiased if a risk premium did not exist. Nevertheless, if the risk premia were
time varying, the exact modelling of the risk premia would have to be determined
first, before the inherent exchange rate expectation could be identified.6
Alternatively, by modelling the exchange rate process arising residuals could be
pulled up to proxy the unexpected exchange rate movements, since those increments
cannot be explained by the respective underlying exchange rate model. Copeland
(1984a, b) runs an ARIMA process in the same manner using corresponding
residuals as exchange rate surprises. In a second attempt, he pulls up exchange
rate returns in order to proxy respective news by implicitly attributing all exchange
rate movements to fundamental news. However, Dornbusch (1980) considers the
concept of uncovered interest parity and thus uses exchange rate excess returns, i.e.
subtracting corresponding interest differences from exchange rate returns (see also
Rose, 1984). What in turn speaks against these alternatives is that they depend on
the validity of the assumed underlying model, which proves to be a priori unknown.

Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704

C 2010 Blackwell Publishing Ltd
684 REBITZKY

In fact, if the true data generating process is different from the one assumed, the
specification is incorrect and the estimated parameters are severely biased.

2.2 Empirical Evidence

A bulk of papers applying the news model has come to the surface; however,
from today’s perspective several serious shortcomings distort the results. Even
though corresponding R-squares prove to be very high, it becomes apparent that
the parameters of the underlying fundamentals are insignificant (see Dornbusch,
1980; Frenkel, 1981; Edwards, 1982a). If one compares the results of related
studies to one another, parameter estimates and corresponding significances vary
considerably and apparently depend on the data period, the specific exchange rate
and the data frequency.
Thus, several authors devote themselves to the delayed news effects. From a
rational point of view, such occurrences should not exist since ‘past innovations
are not news any more’ (see Edwards, 1982a, p. 219). Nonetheless, if news
is announced with delays, corresponding lag structures in news effects appear
consistent with rationality. Indeed, MacDonald (1983) finds significant lags in
money-news on exchange rates. However, since a clear structure is missing, he
leaves the question open to what sources they are linked. Despite the fact that
Bomhoff and Korteweg (1983) do in fact find lagged news effects, they appear
too long to be connected either with delayed publications or with related data
revisions. Instead, they make central bank interventions responsible for causing
these belated effects. In fact, announcements are published with large delays due
to serious revisions, which even occur several months later. Market participants do
not make final macroeconomic announcements on which most of the news models
are based. Thus, Hoffman and Schlagenhauf (1985) focus on the announcement
values which actually were available to investors at the time. They cannot make
any meaningful improvements; nevertheless, they declare that the validity of the
results depends on the data set’s frequency. By analysing monthly and quarterly
data, respectively, they reveal more stable results on the former data. Finally, from
today’s perspective disregarding cointegration among the variables most certainly
helps to explain the generated high R-squares in news models to a certain degree.7
All in all, news models prove to be very instable due to questionable expectation
modelling, too low data frequency and perhaps the dependence on time-invariant
relationships between exchange rates and fundamentals. The latter again depends
on the low data frequency, which in turn requires long time spans to generate a
sufficient amount of observations (see Frankel and Rose, 1995, p. 1701).

3. Announcement Studies
Thus, keeping all due criticism in mind it is not surprising that late in the 1980s
the news model was replaced by a more accurate technique, namely announcement
studies. These studies use daily data and are consequently able to restrict their
focus to the specific days on which economic announcements actually take place.
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 685

Comparing the news model with announcement studies, the latter’s event-window
is considerably smaller, i.e. it decreases from a month (or a quarter, respectively)
in news models to a day in announcement studies. By this means, exchange rate
reactions to news can be much better separated from other influences than was
the case in previous news models such as those applied in Dornbusch (1980),
Frenkel (1981) and Edwards (1982a) – see this statement in Ito and Roley (1987,
p. 256). Announcement studies overcome the shortcomings connected to final
released fundamental announcement data by using original release data in order
to replicate occurring news effects of the time. Even though fundamental
announcements still provide the backdrop for several subsequent revisions, this does
not affect upcoming results because the focus exclusively lies on measuring how
markets actually react to news. In the same spirit, Faust et al. (2003) investigate
the forecasting performance of original release data of standard exchange rate
fundamentals, i.e. in a real-time forecasting framework, and argue in favour of
its major relevance. Nevertheless, since the dates of fundamental announcements
are known beforehand in most industrialized countries, news effects can now be
revealed much easier than via the rough news model (see Frankel and Rose, 1995).
So, due to the fact that exchange rate changes are regressed on the particular news
items, Hardouvelis (1988, p. 23) finds the application of announcement studies to
be ‘a different, more direct strategy’.
Another innovation connected to announcement studies is the implementation
of survey forecasts in order to replace the artificially generated fundamental
expectations in news models.8 This eliminates further sources of inaccuracy, such
as the rather probable case of sub-models turning out to be misspecified. However,
accordant surveys typically deliver forecasts from financial markets professionals,
whose day-by-day business partly depends on their forecasts’ accuracy. Thus, this
kind of data proves to be promising – for more details of the use of survey
data see Takagi (1991), MacDonald (2000) and further Pesaran and Weale (2006).
Therefore, most announcement studies fall back on consensus data in order to cover
fundamental expectations, since these are seen as most suitable to cover market
opinions (see Hakkio and Pearce, 1985, p. 627).9

3.1 The Baseline Model

When it comes to the description of an announcement study, we refer to MacKinlay
(1997), who describes the implementation of an event study in several different
steps. Above all, the success of an event study depends on the exact identification
of the time, i.e. when events actually take place. Once again, we can recognize
the asset-like character of exchange rates such that the foreign exchange market
continuously generates market prices, which proves to be very useful for an
event study (see Hardouvelis, 1988, p. 23). An announcement study is presented
below:

st = β kj · Nt−
k
j + εt (2)
k j
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
686 REBITZKY

   k 
st = α + β k1
j · Nt− j +
k
β k2
j · E t−1− j f t− j + εt (3)
k j k j
 k 
with st = st − st−1 , k
Nt− j = f t− j − E t−1− j f t− j
k

with k = 1, . . . , K and j = 0, 1, . . . , J

In the spirit of market efficiency, the exchange rate change, st , is regressed in
equation (2) on the surprises in the fundamentals, Nkt , as well as an iid disturbance
variable, εt . Thus, the exchange rate change replaces the expectation error in the
exchange rate which was part of the news model. However, since announcement
studies are typically only based on those days when at least one announcement
actually took place, corresponding exchange rate changes are interpreted as rational
surprises on the foreign exchange market – this is why every fundamental together
with its corresponding parameter can be assigned to an individual regression.
Given that in the short run exchange rates expectations are fully incorporated
into market prices, they become irrelevant in the model setting. Mussa (1976)
claims that approximately 90% of exchange rate movements are unanticipated.
Thus, using the return in exchange rates as a proxy of unexpected exchange rate
changes appears logical. Moreover, the surprises in fundamentals, N k , consist of the
differences between consensus expectations and corresponding originally released
announcements. Equation (3) additionally tests for market efficiency by including
the respective expectations on the fundamentals, E t−1 [ f tk ]. Accordingly, if the
foreign exchange market in fact proves to be efficient, the parameters α and β k2
have to be insignificantly different from zero; thus, equation (2) implicitly assumes
efficiency (see Cornell, 1983). Furthermore, both equations (2) and (3) allow for
lagged news effects which are captured by index j. Since the event-window is much
smaller than the one in the news model, it would stand to reason to additionally
examine delayed news effects that take effect beyond the specific announcement
day (see Hakkio and Pearce, 1985).10

3.2 Empirical Evidence

Engel and Frankel (1984) analyse news in monetary policy and show that exchange
rates appreciate after positive money surprises, i.e. a positive relationship between
the two variables with money supply being bigger than previously expected. Cornell
(1983) questions this finding, since he does not find such a relation before October
1979. Only after this date does the relation become significant; however, it is
contradicted by the finding that bond rates react positively in respect of money
news as well, since the latter relationship indicates a long-term inflation expectation.
Hardouvelis (1984) confirms Cornell’s findings and additionally reveals a positive
relationship between long-term exchange rate expectations and money news. In his
opinion, the Fed regained credibility in financial markets after its policy change
in October 1979, but some related uncertainty remained, which manifests itself in
an inflation premium. Hakkio and Pearce (1985), who use three exchange rates
per day, show a structural break in the relationship between the exchange rate and
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 687

money news in October 1979, when the Fed changed its policy orientation from
the Fed Funds rate to money (for further accordant studies see Roley, 1986; Tandon
and Ulrich, 1987; Hardouvelis, 1988).
Irwin (1989) denotes a positive relationship between the exchange rate and
current account deficit – but only after June 1984. At this point in time the
deficit hit a critical high, which alarmed financial market participants and switched
market attention to current account announcements. It is rather common for market
participants to anticipate central bank interventions after surprisingly negative
current account numbers have been published. Klein et al. (1991) analyse the
relationship between the Plaza Agreement in September 1985 and the US current
account news’ impact on exchange rates. Surprisingly, they find a first structural
break several months later – in March 1986. In their understanding, the market
participants’ initial scepticism concerning the credibility of the policy change
disappeared later. In addition to that, Aggarwal and Schirm (1992) take the
Louvre Accord of 1987 into consideration. They state that next to the central
banks’ strategies, international agreements in particular play an important role in
influencing current account news on exchange rates. However, if particular policy
acts prove to be creditable, financial markets react very strongly to them. Karfakis
and Kim (1995) examine Australian current account news for two periods, one
before 1989 and one thereafter. With regard to the first period, they find the
exchange rate to appreciate significantly after higher current account deficit news,
which they connect to anticipated central bank interventions. In contrast to that,
the relationship turns out to be insignificant thereafter. They attribute this to a
change of the Australian central bank’s monetary strategy in 1989 from restrictive
to more relaxed. As a result, market participants did not anticipate interventions on
the foreign exchange market anymore in response to negative Australian current
account news. Taken together, actual policy orientations and accordant strategies
influence the way in which particular fundamental news impacts exchange rates. But
next to an institution’s orientation and connected actions, the power of enforcement
and the will to act accordingly determine if and how news is incorporated into
exchange rates.
Further specific news issues have been analysed, for example lagged and
anticipated news effects, which contradict market efficiency at first glance. Hakkio
and Pearce (1985) show the existence of lagged effects in Canadian CPI and
money announcements and attribute them to central bank interventions of the Bank
of Canada. MacDonald and Torrance (1988), however, cannot reveal any lagged
news effects in British money. Furthermore, Goodhart (1988) expects exchange
rates to either undershoot or overshoot with respect to the news, due to vague
effects arising from interactions between short-term oriented speculators, sluggish
fundamentalists and central bankers following political targets. However, Goodhart
cannot show a particular lag structure and presumes that daily data are unsuitable
for revealing such interrelations if it is in fact an intra-daily phenomenon.
Beck (1993) finds anticipated news effects of two to three days in US public
deficits. He assumes the existence of heterogeneous information due to the fact
that a few market participants hold back some private information. Therefore,
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
688 REBITZKY

he claims that asymmetric reactions to news can be in line with efficiency of

the foreign exchange market. Karfakis and Kim (1995) separate their data into
good and bad current account news and show that negative news is followed by
significant exchange rate appreciations, i.e. the current account deficit is higher
than expected. They connect this finding to anticipated interventions of the central
bank, which obviously would not arise if current account news turns out to be
positive.11
Summed up, a lot of announcement studies have been applied to the foreign
exchange market focusing – in contrast to news model analyses – to a greater extent
on specific issues such as lagged news effects and heterogeneous reactions. Several
advantages of announcement studies are at hand: first, the use of survey expectations
renders artificially generated fundamental expectations obsolete; second, the event-
window decreases dramatically, which is why news effects can now be better
separated from other influences than before; third, exchange rate expectations
prove to be irrelevant, which removes another potential source of misspecification.
However, even though this model proves to be clearly superior, some shortcomings
still remain. Above all, the main critique concerns the event-window, which may
still be too wide to capture certain news effects properly.12 Furthermore, despite
a lot of significant fundamental news and an increased number of many more
consistent results between the different announcement studies, R2 values turn out
to be very low. Frankel and Rose (1995) confess ‘that (news) effects typically
diffuse rapidly in a stream of other information that the researcher is not able to
observe, so that statistical significance disappears when exchange rate changes are
measured a day or two late’ (p. 1701).

3.3 Recent Studies

As a matter of fact, two announcement studies have been published recently which
bring some further insight into the news literature of the foreign exchange market.
Since these studies do not conceptually belong to later approaches, we present them
here in a separate section.
Ehrmann and Fratzscher (2005) analyse daily changes of the US dollar/euro (until
1999 US dollar/D-mark, respectively) by using real-time news of macroeconomic
announcements during the period 1993–2003. They confirm that US news exerts a
stronger influence on exchange rates than respective news from the euro area (or
Germany). Furthermore, by pooling respective news for the USA and the euro area
in a composite variable, they analyse for the existence of several asymmetries in
news effects. In particular, they discover that news effects depend on certain market
conditions. News tends to have a heavier impact if, first, previous exchange rate
volatility is higher, second, a piece of news is relatively large or, third, a piece of
news turns out to be bad. So, in accordance with the Veronesi effect, the higher the
current uncertainty, the more extensive the impact arising from news effects (see
Veronesi, 1999). In a behavioural exchange rate model Manzan and Westerhoff
(2005) show that investors’ perception of fundamental shocks on exchange rates
varies, depending on the latter’s volatility. When the volatility is low, they tend to
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 689

underreact on fundamental news, whereas overreaction takes place when volatility

is high.
Another interesting issue arises with the use of proxies for macroeconomic
variables, which are considered as exchange rate fundamentals. Since the latter’s
data underlie some severe problems, i.e. low frequency, obsolete data, some
difficulty generating related expectations, the use of appropriate proxies seems
to promise to perform more accurate event studies.13 Following this notion, Fatum
and Scholnik (2006) attempt to discover whether US monetary policy matters for
D-mark/US dollar, GB pound/US dollar and JP yen/US dollar rates on a daily basis.
Hence, they use daily changes in the two-month ahead Federal Funds future rate to
extract market expectations for the future US monetary policy without relying on
too infrequent survey outcomes or questionable model specifications to generate
such expectations. Next to previously shown exchange rate responses on infrequent
actual policy changes, they show that the monetary policy in fact also matters for
daily exchange rate determination. In accordance with the efficiency of the foreign
exchange market, they show that exchange rates react without delay to day-to-day
changes in monetary policy expectations, the latter extracted from Federal Funds
future changes, and further, only to the unexpected elements of actual monetary
policy changes.14

4. The High-frequency Model

In the late 1980s studies surfaced that used tick-by-tick data, minimizing the event-
window even more. This high-frequency model has ever since dominated the event
study literature. However, due to the immense mass of data produced in the foreign
exchange market for highly-liquid currency pairs every day, this step was merely a
matter of time. Whereas the first attempts were restricted to the announcement data
only (see, for example, Goodhart and Figliuoli, 1991; Goodhart et al., 1993), later
studies would use related expectations again in order to generate surprise elements.

4.1 The Baseline Model

Since the event-window d ranges from just a minute or so to a maximum of a
few hours, de facto single regressions result, which are separated by the specific
fundamental news items. The high-frequency model arises as follows:
  g
st,d = αd + β kj,d · Nt−
k
j + φd · s(t,d)−g + εt,d (4)
k j g
 
with k
Nt− j = k
f t− j − E t−1− j f t−
k
j , k = 1, . . . , K , g = 1, . . . , G
with st,d = st − st−d and d = 1, . . . , D
In equation (4), the exchange rate change, st,d , is regressed on the specific
news item Nkt , on G lags of past exchange rate returns to account for potential
autocorrelation and on iid, a disturbance variable. By using several event-windows
with different lengths d (measured in for example five-minute units), the durations
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
690 REBITZKY

of the individual news effects can be analysed, whereas index j includes all sorts of
lag effects. However, if asymmetric news effects exist, corresponding parameters
would be instable. So by depending on the validity of the current economic state,
equation (5) provides the relevant news parameters. For illustrative purposes, the
focus rests on only one specific event-window and we do not consider any lag
effects:

stk = αr + βrk · Ntk + εt,d (5)
k

with α1 + β1k · Ntk if state 1
k

and α2 + β2k · Ntk if state 2
k

Alternatively, several studies analyse the influence of news effects on the absolute
value of exchange rate returns. In doing so, first of all the exchange rate volatility
has to be de-seasonalized in order to control for underlying intra-daily seasonal
patterns which would otherwise bias the upcoming estimation results (see Andersen
and Bollerslev, 1998). For the same purpose lagged volatilities Vst−g have to be
considered since the intra-daily exchange rate volatility Vst also reveals certain
volatility dependences.
  g
s
Vt,d = αd + β kj,d · Nt−
k
j+ λd · V(t,d)−g
s
+ ηt,d (6)
k j g

4.2 Empirical Evidence

Goodhart and Figliuoli (1991) analyse exchange rate reactions to news effects on the
basis of minute-by-minute data for two trading days. Even though they allow for lag
effects, they cannot reveal significant relationships. However, the study of the full
three-month data set could indeed reveal several meaningful price jumps in response
to news (see Goodhart, 1989). Nevertheless, both studies lack the implementation
of corresponding fundamental expectations, which questions the results as
such.
Goodhart et al. (1993) pursue the issue whether news effects on exchange rates
are permanent or temporary. Thus they analyse the US dollar/GB pound during
April and July in 1989 regarding one US trade balance announcement and in
addition a particular rise of the UK prime rate. They discover that both the mean
and the variance of the exchange rate react significantly to these news items –
however, only temporarily, because during the following trading week such effects
are reversed. Nevertheless, one should keep in mind that Goodhart et al. (1993)
only examine two individual events. Analysing five-minute intervals of the D-
mark/US dollar from January 1992 until December 1994, Almeida et al. (1998)
wonder how much time it takes for news effects to be fully processed into exchange
rates. In fact, by using MMS expectations to generate the surprising elements in
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 691

announcements, they show that in general the news effects are significant until a
two-hour horizon after release except for the payroll and consumer confidence
figures. The latter continue to be significant until 12 hours after respective releases,
thereby confirming ‘that the very short-term reaction to news is drowned very
rapidly in subsequent noise’ (p. 392). Furthermore, they reveal that US news
reaches its maximum impact after 15 minutes, while German news only does so
after approximately three hours; however, both significant levels reach their peak at
the 15-minute horizon. Almeida et al. (1998) attribute this to the fact that the dates
of US announcements are known in advance, whereas German announcements are
usually unscheduled, thus making some extra time necessary for market participants
to process the latter news. Andersen and Bollerslev’s (1998) study intraday volatility
of the D-mark/US dollar using tick-by-tick data from October 1992 to September
1993. They show that macroeconomic announcements induce the largest intraday
movements; however, taken together, the typical intraday volatility patterns clearly
continue to dominate on the whole.15 Andersen et al. (2003) analyse five-minute
intervals for a total of five exchange rates from 1992 to 1998 on a total of 41 German
and US announcements. They find that prices fully adjust to news within five
minutes, while volatilities adjust gradually with complete adjustment within about
an hour. Moreover, they use MMS expectations to generate corresponding surprises.
Indeed, a lot of US news is significant, for example economic cycle variables induce
the US dollar to appreciate. In contrast, German news is seldom relevant, which they
explain in accordance with Almeida et al. (1998). While US news often explains
30% and more of exchange rates’ variance (trade balance news achieves almost
60%), German news rarely contributes more than 15%. Furthermore, Almeida
et al. (1998) show that the timing of news matters. By dividing the news
variables into different groups, for example real activity, consumption and prices,
and arranging them chronologically according to the announcement date and the
respective time, they are able to calculate if the impact resulting from news differs
accordingly. As presumed, first announcements of each group reveal the highest
impact on subsequent exchange rate changes. The authors explain this finding with
the fact that later news often only confirms the former and therefore is not really
news anymore. Finally, by separating good from bad news, the authors look for a
specific source of parameter instability with respect to heterogeneous news effects
by means of equation (5). In fact, they find asymmetrical exchange rate reactions
to news, particularly to trade balance and payroll announcements. Furthermore, the
authors pursue the idea that bad news should have an unusually large impact in
‘good times’ and consult the standard deviation of exchange rate expectations as a
proxy for market uncertainty. Focusing on bad news, payroll employment, durable
goods orders and trade balance increase market uncertainty after bad news by about
30%, 6% and 12%.16
By and large, the major criticism on event studies has been dealt with. Thus, by
using tick-by-tick data, the event-window shrinks to only a few minutes, thereby
most likely removing potential disturbances which could dilute the respective news
effects. On the other hand, high-frequency studies only focus on the instantaneous
exchange rate changes after news. Therefore, it should not come as a surprise that
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
692 REBITZKY

R2 values of up to 60% arise, since the major part of exchange rate movements
is masked out. Furthermore, high-frequency studies allow for the required duration
(needed to incorporate news into market prices) to be analysed. However, pulling up
the first two statistical moments of the exchange rates, it is revealed that news effects
only last temporarily. Moreover, earlier preliminary findings in announcement
studies that show asymmetric reactions arising from good and bad news can be
confirmed via finer analyses (see, for example, Karfakis and Kim, 1995; Andersen
et al., 2003). However, one major point of criticism remains: high-frequency studies
only focus on these short-term exchange rate changes, which follow macroeconomic
news. Hence, by only considering a small part of the total of exchange rate
movements by construction, high-frequency studies leave the greater part of the
variance unconsidered. In fact, Dominguez and Panthaki (2006) state that high-
frequency models which do not include order flow as an additional regressor can
only explain less than 1% of the total of exchange rate movements.

5. The Hybrid Approach

Actually, short-run exchange rate movements do not correlate well with macro-
economic variables, which is why various related announcements do not appear
significant in explaining exchange rate changes. Today, a common consensus has
been reached in exchange rate literature regarding investor heterogeneity that arises
from either differing information bases, market uncertainty or diverse opinions
(see, for example, Flood and Rose, 1996; Bacchetta and van Wincoop, 2006; De
Grauwe and Grimaldi, 2006). Moreover, microstructure models, which rely inter
alia on investor heterogeneity, are seen to be relatively successful in explaining
exchange rate movements in the short run, for example via order flow. The
market microstructure classifies order flow as private information, for example
heterogeneous expectations, interpretations and information. Hence, order flow is
declared an elementary medium to process disagreement among market participants
into prices. However, in contrast to that, the traditional macroeconomic theory
insists on market homogeneity, which goes against influences arising from order
flow that affects market prices.
Having said that, the bulk of the exchange rate variance still remains unexplained;
news may affect exchange rates additionally via another channel. Andersen
et al. (2003, p. 59) note that the consideration of order flow appears promising, and
thus the microstructure variable order flow has found its way into high-frequency
news studies.

5.1 News, Exchange Rates and Order Flow

In fact, we already introduced the direct news channel in Section 2 without
mentioning this explicitly, since at that stage we did not consider trading as a
vehicle to process fundamental news into market prices. Hence, the direct channel
is restricted to news on which market participants fully agree, and is thus based upon
public information, which is consequently called common knowledge information
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 693

(see Evans, 2002). The latter reflects investor heterogeneity, which is in turn
responsible for speculative trades and can hence be observed in order flow.17
Hence, non-common knowledge information due to diverging expectations about
future fundamentals’ paths and their respective impact on exchange rates, i.e. private
information, is processed into market prices via order flow, and constitutes the
indirect news channel, which, in contrast to the direct channel, requires currency
trading.
Cai et al. (2001) reveal a strong influence stemming from customer order flow
on the volatility of the JP yen/US dollar during 1998 – even after taking US and
Japanese macroeconomic announcements into account. In fact, Evans and Lyons
(2002) find a significant relationship between exchange rates and related interdealer
order flow from May until August 1996. Using daily data, they explain 63% of the
variance in the US dollar/D-mark and 40% in the US dollar/JP yen. Considering the
disappointing results of empirical exchange rate modelling (see Meese and Rogoff
(1983) and in a later study Cheung et al. (2005)), Evans and Lyons’ results are
striking.
In their intraday analyses, Evans and Lyons (2008) additionally show that
order flow following macroeconomic news contributes more to US dollar/D-
mark movements than at other times. Trade intensity rises significantly after
announcements, accompanied by a higher order flow. It stands to reason that market
participants do not share common opinions with respect to the meaning of certain
news, and thus form different expectations, which finally lead to a corresponding
order flow. Hence, it does not come as a surprise that, when following news,
the order flow’s influence on exchange rates rises, since a growing amount of
uncertainty following the heterogeneous news’ interpretations has to be processed
into market prices (see Evans and Lyons, 2002).18 A further interesting aspect of
Evans and Lyons’s (2008) intraday analyses reveals that trade intensity in exchange
rates proves to be higher following scheduled news than in response to unscheduled
news. Since the order flow’s impact on exchange rates increases with more trade,
order flow proves to be at least as important in subsequent price reactions following
scheduled news as non-scheduled news. This observation is unexpected, since the
former news is standardized and is thus more transparent than unscheduled news.
So common knowledge information is transported directly and instantaneously
into market prices via the already introduced news items Nkt and has to be
distinguished from non-common knowledge information carried by previous order
flows, OF t−1 . Known as the hybrid approach in the literature, news influences
exchange rates via two dimensions (see Lyons, 2001, p. 16), which arise as
follows.
  g 
st,d = αd + β kj,d · Nt−
k
j + φd · s(t,d)−g + δl,d · OF (t,d)−l + εt,d
k j g l
  (7)
with k
Nt− j = k
f t− j − E t−1− j k
f t− j , k = 1, . . . , K , g = 1, . . . , G
with st,d = st − st−d , d = 1, . . . , D and l = 1, . . . , L

Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704

C 2010 Blackwell Publishing Ltd
694 REBITZKY

5.2 Empirical Evidence

By simultaneously analysing daily US dollar/D-mark rates, order flow and a broad
pool of different fundamental announcements, Evans and Lyons (2008) explain
approximately 36% of the exchange rate movements via the hybrid approach.
Furthermore, they show that the indirect news channel is twice as important as the
direct channel for the processing of corresponding news. Nevertheless, regarding
both news channels, German news explains more than double the amount of US-
American news. However, due to the fact that German news occurs four times
as often as US-American news in the data set, one news item of the latter has a
slightly larger effect on prices. Love and Payne (2008) analyse high-frequency data
comprising exchange rates of the euro, US dollar and GB pound. They confirm
that the indirect news channel is more relevant than the direct channel (the former
contributes two-thirds of the explained variance). Moreover, by increasing the event-
window up to 20 minutes, they show that the relevance of the indirect channel
increases following announcements. Furthermore, with respect to the question how
news effects spread out, this depends on the geographical origin of the respective
news; news that does not stem from one of the two countries of the regarded
currency-pair is processed exclusively via order flow. Thus, after analysing news
effects in the foreign exchange market, Love and Payne propose a multi-country
approach.
So, combining these results with the findings in Evans and Lyons (2002, 2008),
where order flow explains 60% of the US dollar/D-mark, Evans and Lyons
confess that approximately a third of this explained exchange rate variance is
related to indirect macroeconomic news effects, whereas the remaining two-thirds
share is not related to news. Furthermore, 15% of the unexplained exchange rate
variance of order flow stems from direct news effects, leaving only 25% to further
unknown sources (see Evans and Lyons, 2008). All in all, the authors claim
that currency traders primarily use order flow to interpret new macroeconomic
information correctly, and to subsequently process it accurately into market prices.
This corresponds to Frömmel et al. (2008), who analyse different trading groups of
the foreign exchange market and show that only order flow from banks and financial
customers is positively related to exchange rate volatility. Their interpretation,
namely that this result is a consequence of information aggregation in the market,
is confirmed by the observation that others’ order flow (i.e. from commercial
customers) is not related to volatility.
In sum, order flow establishes itself as an important element to incorporate
macroeconomic news into financial prices. However, the question arises how much
time this process requires from being complete. Love and Payne (2008) show
that indirect news effects arising from order flow which follow macroeconomic
announcements are processed within only two minutes; thus, the foreign exchange
market appears to be fairly efficient. In contrast, Evans and Lyons (2008) find that
order flow is autocorrelated up to 90 minutes after corresponding announcements,
defying Love and Payne’s view. Using tick-by-tick transaction data of the D-mark/
US dollar rate, Carlson and Lo (2006) focus on a particular rise of the interest rate
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 695

by the Deutsche Bundesbank on 9 October 1997. This unforeseen event triggered an

interesting intraday pattern, which can be used to study the reactions of currency
dealers to news in detail. However, even though Carlson and Lo reveal some
instantaneous speculations of professionals in response to this news item, they
also show that those destabilizing actions die out shortly after; trading volume
and volatility settle down to normal conditions after two hours. Further evidence
for an even longer-lasting information process is indicated in Evans and Lyons
(2007), who reveal highly persistent expectation errors with respect to exchange
rate fundamentals. In addition, by examining US-American and German money
and economic growth as well as inflation rates, they show that order flow forecasts
future fundamentals better than accordant exchange rates do. In fact, autocorrelation
still proves to be significant three months after the corresponding announcements.
Finally, by again using the exemplary 25% of unexplained variance in exchange
rates from news and order flow (see Evans and Lyons, 2002, 2008), it is possible for
potentially relevant but so far unconsidered news sources to be taken into account.
If this is true, the news pool would have to be extended and to some extent
reshaped. In fact, scheduled macroeconomic news only comprises past information
of a very low frequency, i.e. on a monthly or quarterly basis. Thus, the information
content appears to be somewhat outdated (see also Fatum and Scholnick, 2006).
Consequently, it stands to reason that the extension of the considered information
basis regarding unscheduled macroeconomic news (for example new economic
laws and central bank interventions) and non-fundamental information (for example
technical trade signals and market sentiment) overcomes all criticism of being too
low in (fundamental) news frequency. In any case, when conducting a questionnaire
among currency traders, Gehrig and Menkhoff (2004) disclose an increasing
popularity of non-fundamental analyses. In fact, these professionals assign more
relevance to technical analyses than to fundamental analyses.
Dominguez and Panthaki (2006) analyse high-frequency data using 20-minute
intervals to explain US dollar/euro and US dollar/GB pound returns via both news
channels.19 Considering an extensive set of additional variables, many novel news
items prove to be significant – even though R2 values only rise marginally (about
1% and 4% concerning the US dollar/euro and US dollar/GB pound, respectively).
Even though Evans and Lyons (2008) already consider a broad spectrum of news,
they restrict their analyses to macroeconomic announcements since they see other
news sources as being negligible. Nevertheless, if Dominguez and Panthaki split
the data sets in two separate regimes, depending on respective exchange rate
volatility, R2 values rise from 15% to 32% in respect of the US dollar/euro and from
6% to 21% regarding the euro/GB pound. We conclude that the authors confirm
that news effects spread out primarily via order flow, and state that the indirect
news channel dominates the foreign exchange market.20 Moreover, it follows that
next to (common) scheduled macroeconomic news, unscheduled macroeconomic
news and non-fundamental news also prove to be relevant, and can be used to
explain exchange rate movements as well. Nevertheless, among said news sources it
stands to reason that scheduled macroeconomic news primarily influences exchange
rates.
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
696 REBITZKY

6. The Multi-asset Approach

In this section we will assemble the findings regarding cross-market linkages
and dynamic responses to news of joint analyses of different financial variables
such as stocks, bonds and exchange rates. This work in particular aims to reveal
fresh findings concerning mutual news reactions in financial markets and related
interdependences.
First of all, by surveying the prior literature related to this approach, Hardouvelis
(1988) examines daily exchange rate and interest rate reactions to news in
macroeconomic announcements from October 1979 to August 1984. He points out
that markets primarily react to monetary news, but also on trade balance, inflation
and real economy news. In accordance with models stressing price rigidity and
absence of strong forms of purchasing power parity, he finds that for all news items
positive interest rate responses are accompanied by US dollar appreciations and
vice versa. Ederington and Lee (1993) analyse several assets following scheduled
macroeconomic announcements in order to disclose the durations of related news
processions in market prices. By analysing five-minute intervals for the Treasury
bond, Eurodollar and US dollar/D-mark, they find that most of the price adjustments
occur within one minute, whereas volatility remains considerably higher than
during normal times for another 15 minutes. Ederington and Lee (1995) adhere
to the analysis by analysing even finer intervals of only 10 seconds and reveal an
overreaction within the first 40 seconds after the release of news, which is then
corrected in the following one or two minutes. However, the latter two studies have
in common that they do not relate the bond and exchange rate findings to one
another.
Edison (1997) examines daily news reactions of the D-mark/US dollar and the JP
yen/US dollar as well as the responses of US, German and Japanese interest rates.
By using survey data, she extracts the news items of several US macroeconomic
announcements and finds that the exchange rates react systematically to real
economy news, but not to inflation. As to the interest rates, only US rates respond
more or less to all country-specific news (Japanese interest rates respond less
strongly and German rates do not respond at all). Again, the respective results of
this study are not related among the different assets.
Fair (2003) focuses on direct news effects on future prices of the S&P 500,
Treasury bond, US dollar/D-mark (US dollar/euro beginning 1999), US dollar/
JP yen and US dollar/GB pound, respectively. He uses accordant tick-by-tick
transaction data from April 1982 to March 2000 in order to study the intraday
reactions of these variables following announcements (his study is not restricted
to macroeconomic news only). Further, it should be noted that his procedure
differs from that of the others in that he first localizes significant intra-daily
price jumps after the investigated announcements in at least one of the considered
assets, those incidents which he then calls news.21 Once a news item is localized,
the relationships among the assets can be analysed in order to reveal systematic
interdependences. Among others, he shows that a great deal of the news can indeed
be attributed to US macroeconomic announcements. Furthermore, compared to
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 697

bond rates, exchange rates are less affected by price news than by monetary and
real news. However, even though Fair relates the different influences arising from
news to one another, the overall picture remains vague.
Faust et al. (2007) analyse joint movements of exchange rates as well as US and
foreign term structures following US macroeconomic announcements (the foreign
countries are the UK, the Euro Area and Germany until 1999, respectively). Using
high-frequency data for a considerably long period spanning about 15 years, they
are able to investigate time variations in announcement effects. Moreover, they use
survey data in order to extract the pure news elements. In contrast to Ehrmann and
Fratzscher’s (2005) findings, responses on real and nominal announcements show
a very consistent pattern over time. For this reason, unstable relationships between
the considered asset prices and respective news cannot be seen as evidence.22 In
sum, they show that positive surprises regarding real activity in the USA lead the
US dollar to appreciate and, further, raise short-term as well as long-term interest
rates, particularly in the USA. However, by consulting the uncovered interest parity
relationship supplemented by the risk premium, they reveal that a positive surprise
in US real activity, which is accompanied by a stronger US dollar, implies either an
expected weaker US dollar in the future, a lower risk premium for holding foreign
currency assets, or a combination of these. Furthermore, positive US surprises in
inflation leave the US dollar nearly unchanged, but imply either an increase in the
premium required for holding US dollar assets, an expected US dollar depreciation,
or once again a combination of the above.
Andersen et al. (2007) examine the simultaneous response of US, German and
British stocks, bonds and exchange rates to real-time US macroeconomic news.
They use accordant price data from July 1998 to December 2002 and additional
survey data in order to extract the news elements. Following a multi-market
analysis, Andersen et al. allow for dynamic news responses and cross-market,
as well as cross-country, linkages in the periods surrounding US macroeconomic
announcements. They show that all examined markets are linked to fundamental
news in a high-frequency horizon, while bond rates respond most strongly to
news. Equity market reactions actually depend on the current condition of the
real economy. Taking this fact into consideration, stocks and exchange rates react
almost equally strong to macroeconomic news. Moreover, whereas news regarding
inflation does not affect exchange rates in a systematic way, positive shocks related
to the real economy lead to exchange rate appreciations. In sum, stocks, bonds and
exchange rates show significant contemporaneous cross-market and cross-country
intraday linkages, even after controlling for news effects. Focusing on the exchange
rates reveals that negative bond returns (higher bond rates) and positive stock
returns – indicating a stronger real economy – induce corresponding exchange rate
appreciations.

7. Summary
The adoption of event studies on the foreign exchange market is driven ab initio
by the hope of revitalizing the exchange rate research on fundamentals, ever
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
698 REBITZKY

since Meese and Rogoff showed in 1983 that macroeconomic models fail to
forecast exchange rates more accurately than the random walk. Over the years the
event study approach in exchange rate research underwent several methodological
refinements, flanked by larger and finer data sets, which led so far on the one hand
to the hybrid approach, which separates a direct and an indirect news channel, and
on the other hand to the multi-asset approach, which examines simultaneous news
effects on different asset prices jointly.
The essential findings can be summed up as follows. First, news effects on
exchange rates do indeed work through two separate channels. While the direct
news channel incorporates common knowledge information in exchange rates, non-
common knowledge information is processed into prices via order flow. Second,
next to scheduled macroeconomic news, unscheduled macroeconomic news and
non-fundamental news have proven to be relevant for exchange rate movements, as
well. However, as Dominguez and Panthaki (2006) point out, the latter two groups
are of less importance. Third, there is the question of how quickly prices adjust to
news: on the one hand, direct news effects are considered as being processed after a
short time, whereas on the other hand no consensus has been established as yet with
respect to the duration of indirect news processing via order flow (see, for example,
Evans and Lyons, 2008; Love and Payne, 2008). As concerns the direct news effects,
Carlson and Lo (2006) show exemplarily that after excessive trading among traders
in the interdealer market, normal market conditions regarding trading volume and
volatility of the exchange rate are reestablished about two hours following the
macroeconomic surprise. Fourth, the timing of macroeconomic announcements
matters. Regarding news items attributed to one specific macroeconomic group,
first released announcements presumably impact exchange rates strongly, whereas
later news tend to confirm the former with accordingly lower impacts (see Andersen
et al., 2003). Fifth, asymmetrical exchange rate reactions to news exist in part
and are most likely dependent on whether the news is good or bad; bad news
in particular has an unusually large impact in good times and vice versa. This
can be traced to the market uncertainty at hand. Sixth, with the exception of
inflation, nonfarm payrolls as well as the trade balance, exchange rate reactions to
macroeconomic news hardly vary over time; however, this is not the case for
stock reactions, which depend on related economic conditions (see Andersen
et al., 2007). Seventh, comparing the magnitudes of news reactions between
stocks, bonds and exchange rates reveals that bonds respond most strongly to
macroeconomic announcements. If we consider the asymmetric reaction pattern
of stocks (conditional on real activity), stocks and exchange rates react equally
strong. Eighth, in principle, news indicating a stronger real economy leads the
related currency to appreciate, although expected related future pathways depend
on the market participants’ mindsets, which eventually become unobservable (see,
for example, uncovered interest parity considerations in Faust et al., 2007).
Looking back on 30 years of event studies on the foreign exchange market, it
appears plausible to conclude that exchange rates react to macroeconomic news,
which, in turn, confirms that exchange rates are indeed related to fundamentals.
Nonetheless, on reviewing the insights gained from the hybrid approach, it
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 699

follows that the success of event studies depends heavily on the (accurate)
integration of the underlying expectations; as Mussa (1976, p. 236) points out, the
relationship between fundamentals and exchange rates depends on the influence
of news on accordant expectations. Since the latter map the general perception
pattern, it is part of an individual’s experience. Hence, modelling the market’s
expectations in an even more comprehensive manner in future could, in turn, lead
to a better understanding of the relationship between the exchange rate and its
fundamentals.

Notes
1. Ironically, if Black’s statement (1985) does not hold true, he in turn ‘blame(s) it on
noise’ (p. 530).
2. Neely (2005) discusses time-series-based as well as other event studies in the context
of government interventions in the foreign exchange market.
3. For an excellent introduction into the event study methodology used in the finance
literature, see Campbell et al. (1997).
4. Further news applications are devoted to the general question of efficiency of the
foreign exchange market (see Fama, 1991) as well as the quantification of the risk
premium inherent in exchange rates (see Froot and Frankel, 1989; Cavaglia et al.,
1993a, b; Koedijk and Verschoor, 1994; Koedijk and Wolff, 1996; Verschoor and
Wolff, 2001; Bams et al., 2004).
5. This implies a  given relationship between the exchange rate and its fundamentals, for
example st = k β k  · f k + εt , known to the investors, and hence their expectations
add up toE t−1 [st ] = k β k · E t−1 [ f tk ].
6. By using survey data, Koedijk and Wolff (1996) show among other things evidence
for news effects of risk premia on exchange rates. However, they cannot connect
lagged interest differentials with time-varying risk premia.
7. In fact, studies relying on quarterly data generate even higher R2 values than
accordant studies on monthly data.
8. It is interesting to note that Dornbusch (1980) is the first to use expectation data in an
event study on the foreign exchange market. When running a news model he relies
on OECD forecasts to proxy current account and economic growth expectations.
9. Specific surveys are, for example, Consensus Economics (London, Great Britain),
Financial Market Survey (Mannheim, Germany) and Money Market Services
International (MMS).
10. Anticipated news effects could also be on hand. This would alter index j in equations
(2) and (3) by including lead effects. A few studies actually found indications of
news anticipation (see, for example, Hakkio and Pearce, 1985).
11. They admit that the insignificant effects arising from positive news could be
connected to anticipated future data revisions – at this time good news has quite
often turned out to be bad news after later data revisions.
12. Hakkio and Pearce (1985) already point out this issue, for which they use three
exchange rates per day and indeed generate more consistent results relative to
Cornell (1982) and Hardouvelis (1984).
13. Solnik (1987) follows a news-model-like approach using monthly (and quarterly)
data for the period July 1973 to December 1983 in order to explain real exchange
rate changes. In contrast to the traditional news model he uses financial market
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
700 REBITZKY

prices to proxy macroeconomic variables since corresponding news are real-time

and more directly measured. So he proxies changes in economic activity by means
of real stock returns and monetary shocks by means of interest rates, respectively.
Altogether, his results are fairly weak, which he explains partly by the use of poor
proxies. However, he emphasizes that using financial prices in lieu of traditional
macroeconomic time series proves to be a promising field of further research.
14. Faust et al. (2007) also use Federal Funds future prices in order to generate
corresponding surprises.
15. In particular, US announcements related to the real economy appear most significant,
while in Germany monetary announcements dominate – they link this to different
perceived central bank policies.
16. Good news in bad times is not considered since good news dominates over the
whole period of this data set.
17. The sum of these trades flows into order flow, whereas the latter is defined as the
overall difference between the volumes of buyer- and seller-initiated trades.
18. Following fundamental news, traders are confronted with uncertainty concerning
subsequent price setting. Though, with emerging order flow traders learn how to
interpret news, which, in turn, reduces market uncertainty.
19. In fact, Melvin and Yin (2000) first examine the role of (broad) public information
arrival on the D-mark/US dollar and the JP yen/US dollar. They show its relevance
for related volatilities and quote frequencies as well.
20. Regime 1 holds when current exchange rate volatility exceeds twice the standard
deviation of its average volatility, whereas regime 2 holds in all other cases. When
splitting the sample dependent on the amount of arising news, results prove to be
very similar.
21. An event is considered as news if at least one of the analysed assets shows a one- to
five-minute price change greater than 0.75% in absolute value. Under these terms,
Fair reveals 221 such news elements.
22. Exceptions are surprises in the producer price index, CPI, nonfarm payrolls and the
trade balance, whose responses tend to fall in magnitude and significance over time.

References
Aggarwal, R. and Schirm, D.C. (1992) Balance of trade announcement and asset prices:
influence on equity prices, exchange rates, and interest rates. Journal of International
Money and Finance 28: 347–361.
Almeida, A., Goodhart, C. and Payne, R. (1998) The effects of macroeconomic ‘news’
on high frequency exchange rate behaviour. Journal of Financial and Quantitative
Analysis 33: 383–408.
Andersen, T.G. and Bollerslev, T. (1998) DM–dollar volatility: intraday activity patterns,
macroeconomic announcements, and longer-run dependencies. Journal of Finance
53: 219–265.
Andersen, T.G., Bollerslev, T., Diebold, F.X. and Vega, C. (2003) Micro effects of macro
announcements: real-time price discovery in foreign exchange. American Economic
Review 93: 38–62.
Andersen, T.G., Bollerslev, T., Diebold, F.X. and Vega, C. (2007) Real-time price
discovery in stock, bond and foreign exchange markets. Journal of International
Economics 73: 251–277.
Bacchetta, P. and van Wincoop, E. (2006) Can information heterogeneity explain the
exchange rate determination puzzle? American Economic Review 96: 552–576.
Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704
C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 701

Bams, D., Walkowiak, K. and Wolff, C.C. (2004) More evidence on the dollar risk
premium in the foreign exchange market. Journal of International Money and Finance
23: 271–282.
Beck, S. (1993) The Ricardian equivalence proposition: evidence from foreign exchange
markets. Journal of International Money and Finance 12: 154–169.
Black, F. (1986) Noise. Journal of Finance 41: 529–543.
Bomhoff, E.J. and Korteweg, P. (1983) Exchange rate variability and monetary policy
under rational expectations: some Euro-American experiences 1973–79. Journal of
Monetary Economics 11: 169–206.
Cai, J., Cheung, Y.L., Lee, R.S. and Melvin, M.T. (2001) ‘Once-in-a-generation’ yen
volatility in 1998: fundamentals, intervention, and order flow. Journal of International
Money and Finance 20: 327–347.
Campbell, J.Y., Lo, A.W. and MacKinlay, A.C. (1997) The Econometrics of Financial
Markets. Princeton, NJ: Princeton University Press.
Carlson, J.A. and Lo, M. (2006) One minute in the life of the DM/US$: public news
in an electronic market. Journal of International Money and Finance 25: 1090–
1102.
Cavaglia, S., Verschoor, W.F.C. and Wolff, C.C.P. (1993a) Further evidence on exchange
rate expectations. Journal of International Money and Finance 12: 78–98.
Cavaglia, S., Verschoor, W.F.C. and Wolff, C.C.P. (1993b) Asian exchange rate
expectations. Journal of Japanese and International Economies 7: 57–77.
Cavaglia, S., Verschoor, W.F.C. and Wolff, C.C.P. (1994) On the biasedness of forward
foreign exchange rates: irrationality or risk premia? Journal of Business 67: 321–343.
Cheung, Y.W., Chinn, M.D. and Garcia Pascual, A. (2005) Empirical exchange rate
models of the nineties: are any fit to survive? Journal of International Money and
Finance 24: 1150–1175.
Copeland, L.S. (1984a) The pound sterling/US dollar exchange rate and the news.
Economics Letters 15: 109–113.
Copeland, L.S. (1984b) Oil, news and the petropound: some tests. Economics Letters 16:
123–127.
Cornell, B. (1982) Money supply announcement, interest rates, and foreign exchange.
Journal of International Money and Finance 1: 201–208.
Cornell, B. (1983) The money supply announcement puzzle: review and interpretation.
American Economic Review 73: 644–657.
De Grauwe, P. and Grimaldi, M. (2006) Exchange rate puzzles: a tale of switching
attractors. European Economic Review 50: 1–33.
Dolley, J.C. (1933) Characteristics and procedure of common stock split-ups. Harvard
Business Review 11: 316–326.
Dominguez, K.M.E. and Panthaki, F. (2006) What defines ‘news’ in foreign exchange
markets? Journal of International Money and Finance 25: 168–198.
Dornbusch, R. (1980) Exchange rate economics: where do we stand? Brookings Papers
on Economic Activity 1: 143–85, 87–205.
Ederington, L.H. and Lee, J.H. (1993) How markets process information: news releases
and volatility. Journal of Finance 48: 1161–1191.
Ederington, L.H. and Lee, J.H. (1995) The short-run dynamics of price adjustment to new
information. Journal of Financial and Quantitative Analysis 30: 117–134.
Edison, H.J. (1997) The reaction of exchange rates and interest rates to news releases.
International Journal of Finance and Economics 2: 87–100.
Edwards, S. (1982a) Exchange rates and news: a multi-currency approach. Journal of
International Money and Finance 1: 211–224.
Edwards, S. (1982b) Exchange rate efficiency and new information. Economic Letters 9:
377–382.

Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704

C 2010 Blackwell Publishing Ltd
702 REBITZKY

Edwards, S. (1983) Floating exchange rates, expectations and new information. Journal
of Monetary Economics 11: 321–336.
Ehrmann, M. and Fratzscher, M. (2005) Exchange rates and fundamentals: new
evidence from real-time data. Journal of International Money and Finance 24: 317–
341.
Engel, C. and Frankel, J. (1984) Why interest rates react to money announcements.
Journal of Monetary Economics 13: 31–39.
Evans, M.D. (2002) FX trading and exchange rate dynamics. Journal of Finance 57:
2405–2447.
Evans, M.D. and Lyons, R.K. (2002) Order flow and exchange rate dynamics. Journal of
Political Economy 110: 170–180.
Evans, M.D. and Lyons, R.K. (2007) Exchange rate fundamentals and order flow. NBER
Working Paper 13151.
Evans, M.D. and Lyons, R.K. (2008) How is macro news transmitted to exchange rates?
Journal of Financial Economics 88: 26–50.
Fair, R.C. (2003) Shock effects on stocks, bonds, and exchange rates. Journal of
International Money and Finance 22: 307–341.
Fama, E.F. (1991) Efficient capital markets: II. Journal of Finance 46: 1575–1617.
Fatum, R. and Scholnik, B. (2006) Do exchange rates respond to day-to-day changes
in monetary policy expectations? Evidence from the Federal Funds futures market.
Journal of Money, Credit and Banking 38: 1641–1657.
Faust, J., Rogers, J.H. and Wright, J.H. (2003) Exchange rate forecasting: the errors we’ve
really made. Journal of International Economics 60: 35–59.
Faust, J, Rogers, J.H., Wang, S.Y.B. and Wright, J.H. (2007) The high-frequency response
of exchange rates and interest rates to macroeconomic announcements. Journal of
Monetary Economics 54: 1051–1068.
Flood, R.P. and Rose, A.K. (1996) Fixes: of the forward discount puzzle. Review of
Economics and Statistics 128: 748–752.
Frankel, J.A. and Rose, A.K. (1995) Empirical research on nominal exchange rates.
In G. Grossmann and K. Rogoff (eds), Handbook of International Economics
(Vol. III, pp. 1689–1729). Amsterdam: North-Holland.
Frenkel, J.A. (1981) Flexible exchange rates, prices, and the role of ‘news’: lessons from
the 1970’s. Journal of Political Economy 89: 665–705.
Frömmel, M., Mende, A. and Menkhoff, L. (2008) Order flows, news, and exchange rate
volatility. Journal of International Money and Finance 27: 994–1012.
Froot, K.A. and Frankel, J.A. (1989) Forward discount bias: is it an exchange risk
premium. Quarterly Journal of Economics 104: 139–161.
Gehrig, T. and Menkhoff, L. (2004) The use of flow analysis in foreign exchange:
exploratory evidence. Journal of International Money and Finance 23: 573–
594.
Goodhart, C. (1988) The foreign exchange market: a random walk with a dragging anchor.
Economica 55: 437–460.
Goodhart, C. (1989) News and the foreign exchange market. Discussion Paper 71, LSE
Financial Markets Group.
Goodhart, C. and Figliuoli, L. (1991) Every minute counts in financial markets. Journal
of International Money and Finance 10: 23–52.
Goodhart, C., Hall, S.G., Henry, B. and Pesaran, B. (1993) News effects in a
high-frequency model of the sterling–dollar exchange rate. Journal of Applied
Econometrics 8: 1–13.
Hakkio, C.S. and Pearce, D.K. (1985) The reaction of exchange rates to economic news.
Economic Inquiry 23: 621–636.
Hardouvelis, G.A. (1984) Market perceptions of Federal Reserve policy and the weekly
monetary announcement. Journal of Monetary Economics 14: 225–240.

Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704

C 2010 Blackwell Publishing Ltd
 THE INFLUENCE OF FUNDAMENTALS ON EXCHANGE RATES 703

Hardouvelis, G.A. (1988) Economic news, exchange rates and interest rates. Journal of
International Money and Finance 7: 23–35.
Hoffman, D.L. and Schlagenhauf, D.E. (1985) The impact of news and alternative theories
of exchange rate determination. Journal of Money, Credit, and Banking 17: 328–
346.
Irwin, D.A. (1989) Trade deficit announcements, intervention, and the dollar. Economics
Letters 31: 257–262.
Ito, T. and Roley, V.V. (1987) News from the U.S. and Japan which moves the yen/dollar
exchange rate? Journal of Monetary Economics 19: 255–277.
Karfakis, C. and Kim, S.J. (1995) Exchange rates, interest rates and current account
news: some evidence from Australia. Journal of International Money and Finance 14:
575–595.
Klein, M.W., Mizrach, B. and Murphy, R.G. (1991) Managing the dollar: has the Plaza
agreement mattered? Journal of Money, Credit and Banking 23: 742–751.
Koedijk, K.G. and Verschoor, W.F.C. (1994) Asian interest expectations and exchange
rate dynamics. Pacific-Basin Finance Journal 2: 439–452.
Koedijk, K.G. and Wolff, C.C. (1996) Exchange rate returns, ‘news’, and risk premia.
Economics Letters 50: 127–134.
Love, R. and Payne, R. (2008) Macroeconomic news, order flow and exchange rates.
Journal of Financial and Quantitative Analysis 43: 467–488.
Lyons, R.K. (2001) The Microstructure Approach to Exchange Rates. Cambridge, MA:
MIT Press.
MacDonald, R. (1983) Some test of the rational expectations hypothesis in the foreign
exchange markets. Scottish Journal of Political Economy 30: 235–250.
MacDonald, R. (1985) ‘News’ and the 1920s experience with floating exchange rates.
Economics Letters 17: 379–383.
MacDonald, R. (2000) Expectations formation and risk in three financial markets:
surveying what the surveys say. Journal of Economic Surveys 14: 69–100.
MacDonald, R. and Taylor, M.P. (1992) The economics of exchange rates. International
Monetary Fund Staff Papers 39: 1–57.
MacDonald, R. and Torrance, T.S. (1988) On risk, rationality and excessive speculation
in the Deutschemark–US-dollar exchange market: some evidence using survey data.
Oxford Bulletin of Economics and Statistics 50: 107–123.
MacKinlay, A.C. (1997) Event studies in economics and finance. Journal of Economic
Literature 35: 13–39.
Manzan, S. and Westerhoff, F. (2005) Representativeness of news and exchange rate
dynamics. Journal of Economic Dynamics and Control 29: 677–689.
Meese, R.A. and Rogoff, K. (1983) Empirical exchange rate models of the seventies: do
they fit out of sample? Journal of International Economics 14: 3–24.
Melvin, M.T. and Yin, X. (2000) Public information arrival, exchange rate volatility and
quote frequency. The Economic Journal 110: 644–661.
Mussa, M. (1976) The exchange rate, the balance of payments and monetary and fiscal
policy under a regime of controlled floating. Scandinavian Journal of Economics 78:
229–248.
Neely, C. J. (2005) An analysis of recent studies of the effect of foreign exchange
intervention. Federal Reserve Bank of St Louis Review 87: 685–717.
Pesaran, M.H. and Weale, M. (2006) Survey expectations. In C. Granger and A.
Timmermann (eds), Handbook of Economic Forecasting (pp. 715–776). Amsterdam:
North-Holland.
Roley, V.V. (1986) U.S. monetary policy regimes and U.S.–Japan financial relations.
NBER Working Paper 1858.
Rose, A.K. (1984) Testing for ‘news’ in foreign exchange markets. Economic Letters 14:
369–376.

Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704

C 2010 Blackwell Publishing Ltd
704 REBITZKY

Solnik, B. (1987) Using financial prices to test exchange rate models: a note. Journal of
Finance 42: 141–149.
Takagi, S. (1991) Exchange rate expectations. International Monetary Staff Papers 38:
156–183.
Tandon, K. and Ulrich, T. (1987) International market response to announcement of U.S.
macroeconomic data. Journal of International Money and Finance 6: 71–83.
Veronesi, P. (1999) Stock market overreaction to bad news in good times: a rational
expectations equilibrium model. Review of Financial Studies 12: 975–1007.
Verschoor, W.F.C. and Wolff, C.C. (2001) Exchange risk premia, expectations formation
and ‘news’ in the Mexican peso/U.S. dollar forward exchange rate market.
International Review of Financial Analysis 10: 157–174.

Journal of Economic Surveys (2010) Vol. 24, No. 4, pp. 680–704

C 2010 Blackwell Publishing Ltd