Exchange rate prediction using MILA currencies with Google Trends

Abstract

The purpose of this research is to evaluate the predictive capacity of the Google Trends tool to predict the exchange rates of the countries that make up the MILA market: Chile, Peru, Colombia and Mexico. The economic and financial argument behind this research is that the search frequency provided by Google Trends is a proxy for the expectations that market agents have about the performance of the economy. The Harvard Dictionary is used to recognize words with positive and negative connotations regarding the state of the economy of each country. In-sample and out-of-sample analyzes are performed to assess the predictive ability of Google Trends. In addition, there is recent literature that uses this tool for exchange rate forecasts in the countries of Japan and the United States, but little is known about the Latin American economies of MILA countries. Consistent with previous evidence, the predictability results are unstable and not very robust. This study is important for investors, portfolio managers and/or agents who must manage exchange rate risk.