Dr. Steven Cassou | Working Papers

"Stratagy for a low-cost uncertainty index: A method for sparse data environments"

Abstract: This paper describes a low-cost strategy for generating an uncertainty index. The method is useful in sparse data settings. Several investigations are undertaken which illustrate its value. First, it is shown to work well in rich data settings. Using U.S. data from 1960 onward, it is shown that a variety of simple uncertainty measures can be generated which exhibit properties, such as high peak uncertainty, high correlation and goodness of fit, that aligns well with indices that require significantly higher data requirements, such as the approach proposed by Jurado et al. (2015). Second, an illustration using sparse historical U.S. data shows that this approach can generate sound uncertainty measures when rich data sources are unavailable. Historical uncertainty measures offer valuable insights into the evolution of the business cycle. They also enable comparisons of uncertainty during temporally dispersed economic events.