Assessment of inflationary pressures using newspaper text analysis

Mirko Ðukic
Additional contact information
Mirko Ðukic: National Bank of Serbia

Working Papers Bulletin from National Bank of Serbia

Abstract: In this paper we present the results of a dictionary-based text analysis of articles from the economic sections of four Serbian daily newspapers, caried out to estimate if those articles contain useful information for assessing inflationary pressures. We analyzed 117,113 economic articles in total for the period 2007–2022, by counting terms related to inflation and price rises and price falls, or counting texts containing those terms. Measures of newspaper inflation sentiment, obtained in this way, were found to be highly correlated with inflation, mainly driven by the periods of large inflation swings. During the period of stable inflation, the correlation is significantly lower. Causal relationship clearly goes from the newspaper inflation sentiment to inflation, and simple inflation models with the sentiment as an explanatory variable beat benchmark AR model in the out-ofsample forecast. We conclude that newspaper inflation sentiment can be used as an indicator of inflationary pressures, especially during periods of high inflation volatility.

Keywords: inflation forecasting; text analysis (search for similar items in EconPapers)
JEL-codes: C13 C55 E31 E37 E52 (search for similar items in EconPapers)
Pages: 26 pages
Date: 2022-09
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.nbs.rs/export/sites/NBS_site/documents ... bulletin_09_22_2.pdf Full text (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:nsb:bilten:12

Access Statistics for this paper

More papers in Working Papers Bulletin from National Bank of Serbia Contact information at EDIRC.
Bibliographic data for series maintained by Marko Miseljic ().