(with Eric van Wincoop)
Journal of International Economics
Abstract: We use Twitter opinions about the Euro-Dollar exchange rate to estimate the private information model of Bacchetta and van Wincoop (2006) and investigate the disconnect between the exchange rate and macro fundamentals over both short and long horizons. We simulate the model with the estimated parameters and replicate the methodology of three studies that document the disconnect empirically. The model is consistent with the findings of the empirical literature, though for a different reason over short than long horizons. Over short horizons private information generates a true disconnect between exchange rates and macro fundamentals that accounts for empirical findings. Over long horizons the theory shows that exchange rate changes are mostly driven by observed fundamentals, but empirical limitations in identifying this long-run relationship often lead to an appearance of disconnect in the data
Exchange Rates and Information about Future Fundamentals
R&R, Journal of International Economics
Abstract: According to disperse information models, asset prices aggregate private information of agents about future fundamentals. In this paper, I empirically measure informativeness of exchange rates about future macro fundamentals. The measure of informativeness is the contribution of private information to the variance of adjusted exchange rate. The results show that the median exchange rate informativeness is between 0.02 and 0.07 for various macro variables. The median informativeness of exchange rate about the dominant macro fundamental is 0.11. Exchange rates of countries with high inflation are more informative than those of low inflation countries about future fundamentals. Moreover, exchange rate informativeness is higher for high-income countries.
Journal of Empirical Finance
Abstract: The paper introduces a daily index for expectations of returns based on tweets that express a directional prediction about the stock market index. I develop a dictionary that includes lexicon of traders to identify and classify opinionated tweets. The results show that (1) the Twitter Expectations of Returns Index (TERI) is positively correlated with weekly changes in net long position of investment managers, (2) expectations index of high followers accounts predicts stock market returns, and (3) private information is the primary source of return predictability.