Operationalizing the New York Times to Forecast Economic Indicators

This paper builds upon previous literature on nowcasting, and attempts to use words from New York Times articles to predict Consumer Sentiment and the Unemployment Rate. The results are mixed: Consumer sentiment proves most predictable, while models predicting the unemployment rate perform poorly. Despite these results, this paper investigates the many models and methods to harness the predicting power of textual data, offering a road map for future researchers.