Deciphering Indian inflationary expectations through text mining: an exploratory approach

Abstract

Inflationary forecasts tend to play a crucial role in macroeconomic and financial decision/policy making. In particular, in an inflation-targeting framework, it is of paramount importance. While traditionally, model-based and survey-based inflation expectations are being used, in recent times, a literature has emerged to forecast various macro-aggregates using text-based sentiment estimates. Taking a cue from this approach, in this paper we attempt to decipher inflationary sentiments using text mining from two leading financial dailies, viz., the Economic Times and Business Line. We consciously avoid using social media news due to severe challenges and high noise-to-signal ratio. In our algorithm we aggregate CPI basket level (viz., food, fuel, cloth & miscellaneous) sentiment into an overall index of inflation, adapting techniques from natural language processing. Our results from this text-based model indicate significant success in tracking actual inflation.