Analyst Forecasts and Currency Markets

Abstract

I examine the forecasting performance, directional accuracy, rationality and economic value of analyst forecasts and characteristics of investment portfolios built from these forecasts for 30 currency pairs from 2006 to 2020. My results show that analyst forecasts perform worse than forecasts based on a random walk and forward rates and that they are biased and do not provide significant economic value to investors. Forecasts from global systemically important banks do not differ from non-systemically important banks in terms of forecasting ability. Median forecasts may strongly deviate from market expectations, while analyst forecast dispersion is positively associated with future currency returns. Portfolios built from analyst forecasts tend to strongly underperform the dollar factor, value, carry and momentum portfolios and are spanned by them. My findings indicate that expected returns extracted from analyst forecasts are negatively related to realized excess returns in FX markets and thus contribute to the literature on survey-based returns in asset pricing.