Macroeconomic news and exchange rates

Abstract

This paper examines the relationship between macroeconomic news and the dollar–Mark and dollar–Yen exchange rates. We employ high-frequency observations for a 10-year period. We investigate whether exchange rate observations need to be sampled at a high frequency in order to detect significant effects from news announcements on mean returns and volatility. We examine the linearity and symmetry of the responses to news and also allow the effects of the news announcements to vary across states of the economy. We find that news indicating a stronger U.S. economy causes an appreciation of the U.S. dollar, that the responses are essentially complete within 5 min, and that measuring the responses over 6-h intervals eliminates the statistical significance of the news. The effects of news appear linear and symmetric but there is some evidence that the effects depend on the state of the economy.

Introduction

According to the efficient-markets view of asset price determination, asset prices move whenever new, relevant information arrives. The full response to such news should be quick, perhaps within minutes, assuming that the financial markets are open at the time the news is announced. Numerous studies have investigated this prediction, often with stock price data.¹ Many macroeconomic announcements in the U.S. are, however, made when the stock market and some debt markets are closed; presumably to let market participants digest the news. This makes it more difficult to estimate how quickly news is reflected in price changes. Foreign exchange markets, however, do not close so that exchange rate responses are a natural variable to test the speed with which the market responds to relevant news.

Foreign exchange rates have another advantage over stock prices and bond prices in understanding the responses of asset prices to economic news. Economic announcements may trigger changes in expected inflation or changes in the expected policy moves of the central bank. For example, an announcement of stronger-than-expected economic activity may cause economic agents to increase their expected inflation rate or to increase their expectation that the central bank will raise real interest rates in the near future. Either response should cause bond prices to fall and are likely to cause stock prices to fall. With foreign exchange rates, however, higher expected inflation should cause a depreciation whereas higher expected real interest rates should cause an appreciation.

Although there have been several studies on exchange rate responses to macroeconomic news, they generally have not used exchange rate changes over very short time intervals. Given the noisy nature of exchange rate changes, the typical finding that many news announcements have no perceptible effect on exchange rates might be due to too-long time intervals over which responses are measured.² Almeida et al. (1998) address this question by using high-frequency data on the dollar–Mark exchange rate and find that the length of the time interval does affect the significance of the response to some announcements.³ A recent paper by Andersen et al. (2003) uses similar high-frequency data and again finds a significant and very fast response of exchange rates to several U.S. macroeconomic news announcements.

This paper, which also employs high-frequency data on exchange rates, extends previous work in several ways. First, we examine a longer time period than previous studies. Second, we test whether the responses of the dollar–Mark and dollar–Yen exchange rates to U.S. macroeconomic news are the same. The responses could differ, for example, if the central banks of Germany and Japan were expected to react differently to U.S. news.⁴ Third, we investigate whether there are threshold effects in the responses to news and whether the effects of news depend on the “sign” of the news.⁵ Fourth, we allow the state of the economy to affect the size of the responses.⁶ If Almeida et al. (1998) and others are correct in attributing the responses to announcements, as largely due to the anticipated policy response of the Federal Reserve, it is plausible that the expected reaction of the Federal Reserve to any announcement about the economy may depend on whether the economy is in an expansion or a recession. Moreover, the effect of news on expectations of future economic developments may be state dependent. News of higher-than-expected inflation may affect future expectations of inflation differently if the economy is in a recession than it would if the economy is in an expansion. Hence, we test whether the effect of news differs across states of the economy. Given previous papers that have found that asset price volatility is also affected by news, we analyze news effects on both changes in exchange rates and their volatility.⁷