Do forecasters of major exchange rates herd?☆
Introduction
The recent global financial crisis has shown that the dynamics in financial markets have persistent and substantial effects on the economy. Hence, studying excessive volatility and reasons for a misalignment in financial markets is important for the understanding of economic developments. This is one reason why economists have studied financial market behavior, in particular the expectation formation process. To adequately study the effects, we focus on expectations of individual financial market participants. In particular, we study whether market participants are influenced by the behavior of other market participants. With respect to financial markets, this has led to a new direction in research examining the question of whether market participants display herding behavior when forecasting economic variables. Broadly defined, herding means that market participants tend to move their opinion towards the general market sentiment. By contrast, anti-herding means that market participants deliberately move their forecasts away from the general consensus.
We contribute to the discussion of herding behavior in financial markets by studying the forecasts of three of the most important exchange rates. More specifically, we examine the U.S. dollar exchange rate of the British pound, the euro, and the Japanese yen. A novelty of our analysis is that we employ a common data set for the forecasts of these three exchange rates. The fact that the forecasts are submitted for different exchange rates and time horizons at the same time allows us to compare the forecast behavior in the respective markets. Another novelty is that we use a very comprehensive and unique data set covering 36,000 forecasts from both financial and non-financial institutions over 20 years and include a variety of forecast horizons. Moreover, we also investigate, whether the results about the herding behavior for the different exchange rates have changed over time. Our findings suggest that forecasters in the foreign exchange markets we study in this paper do not primarily herd, but instead display significant anti-herding behavior.
The academic literature on whether herding behavior is prevalent in financial markets is not conclusive. In an analysis of forecaster behavior, Welch (2000) finds strong evidence for herding behavior in securities markets, when analyzing databases for financial market intermediaries. The author concludes that the recommendations of security analysts are being used as a yardstick for the subsequent market forecasts. The seminal paper of Bernhardt et al. (2006) on herding behavior has initiated a new strand of research focusing on forecasts of various macroeconomic variables. While Pierdzioch and Rülke (2013) investigate herding behavior in metal price forecasts, Frenkel et al. (2012) focus on herding behavior in forecasting current account balances. These authors find some evidence of anti-herding, although the degree is fairly different. In addition, Rülke et al. (2016) find mixed evidence of anti-herding behavior when studying the forecasts of business cycles.
The academic literature discusses reasons for herding in financial markets and suggests several potential drivers for this behavior. Some papers argue that career considerations as well as the imitation of superiors may lead analysts to under-weigh private information and to focus on consensus forecasts (Graham, 1999; Scharfstein and Stein, 1990). Other papers find inexperienced analysts as a reason for herding, pointing out that these analysts fear that delivering bold forecasts away from the consensus may more likely terminate their work contracts (Menkhoff et al., 2006; Mitchell and Pearce, 2007). More specifically, Hong et al. (2000) show that younger analysts tend to herd more when compared to senior forecasters.
With respect to the foreign exchange market, most papers focus on one or very few exchange rates and use fairly limited time periods. They present mixed evidence. Whilst Fritsche et al. (2015) find anti-herding in the markets of the Brazilian real and the Mexican peso, Tsuchiya and Kato (2015) cannot find such behavior for the South African rand. Pierdzioch and Stadtmann (2011) provide evidence of anti-herding using a limited sample of exchange rates and forecasts. Building on this, Pierdzioch et al. (2012) report evidence for anti-herding for exchange rate forecasters in several emerging markets.
In this paper, we extend the academic literature on herding behavior of forecasters by providing evidence on three of the most important exchange rates. This is the most comprehensive study of these exchange rates using around 36,000 point forecasts coming out of the same dataset. The data set offers monthly, institute-specific point forecasts over a period of nearly 20 years. To the best of our knowledge, no study has so far analyzed major exchange rates using a comparable data set.
The remainder of this paper is structured as follows: Section 2 briefly describes the data set. Section 3 explains the methodology applied in our study. Section 4 outlines the results on estimating the herding behavior and the changes over time. Section 5 offers some robustness tests. Finally, section 6 summarizes the key results and gives a brief outlook on further research.
Section snippets
The data set
In order to evaluate herding behavior in exchange rate forecasting, we use the Consensus Economics Foreign Exchange Forecasts data set. The survey data provide forecasts for the U.S. dollar (USD) exchange rates of the euro (EUR) and the Japanese yen (JPY), which are published monthly, as well as forecasts for the USD exchange rates of the British pound (GBP), which are published bi-monthly. In total, the data set contains around 36,000 forecasts of 67 difference forecasters from financial and
How to identify herding behavior in foreign exchange markets
With the availability of comprehensive data on exchange rate forecasts, an interesting question is whether other forecasters influence an expert, who delivers a forecasts. The results can have important implications for the interpretation of published forecasts. We investigate this question by studying whether the consensus forecast, i.e., the average of all individual forecasts, influences the individual forecasters. In case an individual forecaster, being aware of the consensus of other
Empirical evidence on herding behavior in the foreign exchange market
We estimate herding behavior for the forecasted exchange rates of the British pound, the euro, and the Japanese yen vis-à-vis the U.S. dollar and use the forecast horizon k of 1, 3, 12, and 24 months. At each point in time t, the data set covers a survey of up to 36 forecasters for each exchange rate.4
Robustness tests
In this section, we provide some robustness tests in order to get a better understanding of our results. Specifically, we investigate whether forecaster herding changes in times of financial crises as defined by the indicators of Reinhart-Rogoff and the NBER crisis indicators.7 We then test the herding results for time consistency and forecaster
Summary and conclusions
This paper aims at shedding more light on the characteristics of individual forecasts in foreign exchange markets by examining the features of herding and anti-herding of forecasters. We present the first study that uses a very comprehensive data set of more than 36,000 observations looking at forecasts of three of the most important exchange rates and at several forecast horizons. We find forecasters to exhibit significant anti-herding behavior in their exchange rate forecasts of the British
References (27)
- et al.
Forecasting ECB monetary policy: accuracy is a matter of geography
Eur. Econ. Rev.
(2009) - et al.
Geography, skills or both: what explains Fed watchers' forecast accuracy of US monetary policy?
J. Macroecon.
(2011) - et al.
Who herds?
J. Financ. Econ.
(2006) - et al.
Herding and anti-herding: a model of reputational differentiation
Eur. Econ. Rev.
(2001) - et al.
Forecasting the Brazilian real and the Mexican peso: asymmetric loss, forecast rationality, and forecaster herding
Int. J. Forecast.
(2015) - et al.
The impact of institutional trading on stock prices
J. Financ. Econ.
(1992) Macroeconomic forecasts and microeconomic forecasters
J. Econ. Behav. Organ.
(2002)- et al.
The impact of experience on risk taking, overconfidence, and herding of fund managers: complementary survey evidence
Eur. Econ. Rev.
(2006) - et al.
Professional forecasts of interest rates and exchange rates: evidence from the Wall Street Journal's panel of economists
J. Macroecon.
(2007) - et al.
Do inflation targets anchor inflation expectations?
Econ. Modell.
(2013)
Herding among security analysts
J. Financ. Econ.
How useful are the forecasts of intergovernmental agencies? The IMF and OECD versus the consensus
Appl. Econ.
Interval estimation for a binomial proportion
Stat. Sci.
Cited by (0)
- ☆
We are grateful to the editor and two anonymous referees for valuable comments and suggestions.