What is green and falling fast?

Not a skydiving bullfrog, but the U.S. dollar’s value. You do not have to look far these days to find news of the dollar’s decline against currencies from around the world. But what is behind this trend? It may seem as if almost any real-world occurrence can shift exchange rates, but what economic, business, political, and social events truly drive the dollar down—or up? More specifically, what is the immediate impact of macroeconomic news—especially that of a surprising variety (i.e., when expectations and realizations are mismatched)—on the dollar’s value?

That exact question motivated Torben G. Andersen (Professor of Finance at the Kellogg School of Management) and co-authors Tim Bollerslev (Professor of Finance at the Fuqua School of Business, Duke University), Francis Diebold (Professor of Finance and Statistics at the Wharton School, University of Pennsylvania), and Clara Vega (Assistant Professor of Finance at the Simon School of Business, University of Rochester and Economist, Federal Reserve Board of Governors) to study the effects of fundamentals-related announcements on the exchange rates of the dollar against multiple currencies. As Andersen suggests, “We can see foreign exchange markets react immediately to macroeconomic news, so this provides a nice ‘laboratory’ for studying not only the effects of different news types, but also how these effects vary across the phases of the business cycle, giving firms, investors, and others important insights into the forces at work in the foreign exchange market.” In an empirical investigation published in 2003 in The American Economic Review, Andersen and his colleagues created a new data set using six years of real-time exchange-rate quotations, macroeconomic expectations (predictions about key economic indicators), and macroeconomic realizations (announcements related to those indicators) to gauge the impact—in terms of speed and pattern of adjustment—of economy-related announcements on the exchange rates of the U.S. dollar against the German mark, British pound, Japanese yen, Swiss franc, and the euro.

The authors show that rapid adjustments in exchange rates, effectively amounting to “jumps,” occur in response to announcements, and that an announcement’s impact depends on its timeliness regarding current economic conditions, as well as on whether the exact release time of the announcement is known in advance. Moreover, as evidence of a “sign effect,” bad news has greater impact than good news. Andersen’s study, although related to earlier work in this area, is differentiated by three features: (1) a focus on foreign exchange markets, as opposed to stock or bond markets; (2) a focus on the exchange rates themselves, as opposed to their volatility, and hence on the direction of the change rather than simply on its size; and (3) use of a new data set encompassing a long time period and a broad set of exchange rates and macroeconomic indicators.

Setting Up the Study
Andersen and his co-authors obtained from the Swiss firm Olsen & Associates a sample of continuously recorded five-minute returns—from January 1992 through December 1998—on the value of the dollar against the mark, pound, yen, franc, and euro. The initial sample covered 2,189 days and more than 630,000 foreign exchange return observations. After the exclusion of weekends, holidays, and days contaminated by lapses in the data feed, the final sample comprised 1,724 days of data and represented almost 500,000 return observations.

For economy-related announcements, the authors used Money Market Services International (MMSI) real-time data on expected (based on telephone surveys of forecasters) and realized macroeconomic fundamentals for the countries with the targeted currencies. The researchers defined “news” as “the normalized difference between expectations and realizations.” Among the fundamentals studied were statistics including GDP, personal income, consumer credit, housing starts, consumer confidence, and unemployment rates. Andersen developed a model allowing determination of how news related to these figures could affect exchange rates and their volatility.

The Effects of Economic News on Exchange Rates
Based on their analysis, the authors found several effects of announcements on exchange rates.

News announcements matter—immediately. U.S.-related news had a generally statistically significant effect on exchange rates, whereas unsurprising announcements did not. As the authors put it, “only unanticipated shocks to fundamentals affect exchange rates.” Among the U.S. indicators related to statistically significant news effects across currencies were payroll employment, durable goods orders, and consumer confidence. The general pattern was that news announcements resulted in an immediate initial jump in exchange rates, with little subsequent movement. Favorable U.S. “growth news” tended to promote dollar appreciation, with negative news driving depreciation.

In contrast, only two German macroeconomic indicators affected mark-U.S. dollar exchange rates. The authors suggest that the “disparity may be due to the . . . fact that the release times of U.S. macroeconomic indicators are known exactly (day and time) but only inexactly for Germany (day but not time).” These uncertain release times may result in less market liquidity and trading around the announcement times, creating smaller news effects at those times and likely producing a more gradual adjustment, “perhaps for a few hours after the announcements.”

Announcement timing matters. To evaluate the issue of whether more timely news within a given category has a greater effect than other news, Andersen and his colleagues grouped U.S. indicators into seven types (e.g., consumption and real activity) and sorted the announcements within each group in terms of their timeliness regarding current economic activity (i.e., by how close in time they were to the economic conditions they represented). The authors found that more timely announcements had the greatest impact, as expected.

Exchange-rate volatility adjusts to news gradually and reflects pure announcement effects. As suggested above, exchange rates adjusted to news immediately. In contrast, exchange-rate volatilities adjusted more gradually, with complete adjustment occurring only after about one hour. Andersen also found that the mere presence of an announcement, regardless of how surprising it was, boosted volatility. This was in line with earlier findings for bond markets.

Effects vary with the sign of the news. Next, the authors tested whether news effects varied with the sign of the surprise—that is, whether it was a positive or negative announcement. Their analyses revealed that the sign did modify the effect of news. Specifically, “negative surprises often have greater impact than positive surprises,” providing evidence of an asymmetric response. This was the first such finding in the foreign-exchange context. Note that the sign effect was more pronounced for the dollar’s exchange rate with the yen, mark, and franc than with other currencies.

Andersen and his coauthors related this asymmetry in exchange-rate response to two strands of economics literature that suggest that bad news in “good times” should have an unusually large effect. Because the authors’ sample took place only in good times—the 1992-1998 growth period in the U.S. economy—they were able to test only a more general prediction: “bad news should have unusually large effects.” They tested this hypothesis by using the spread of forecasters’ predictions following the arrival of bad news in good times and found that analyst forecast dispersion was indeed greater following bad news about key economic indicators, reflecting increased uncertainty about the economy.

Future Research Directions
Andersen and his coauthors conclude with several suggestions for research that would build on their findings. For example, based on recent work that reveals order flow as a predictor of exchange-rate movements, the authors suggest that future work could examine whether news affects exchange rates via order flow or more directly. More broadly, investigators in this area could examine issues of directionality by testing whether news effects are a function of state uncertainty; this would involve looking at the simultaneous effects of news and general economic uncertainty. A third research area involves scrutinizing the effects of non-scheduled, qualitative “headline news,” as opposed to the regularly scheduled quantitative macroeconomic announcements of focus here. The authors also suggest that future work could consider which specific forms of news have the greatest effects and whether scenarios exist in which seemingly positive news has negative effects. Finally, the authors look forward to “characterizing the joint responses of the foreign-exchange, stock, and bond markets to real-time news surprises.”

Clearly there remains much fertile research ground to explore in this area, but the findings of Andersen and his colleagues help us understand what lies behind daily rises and falls in the value of a dollar.