Nowcasting using news topics. Big Data versus big bank

Leif Thorsrud

No 2016/20, Working Paper from Norges Bank

Abstract: The agents in the economy use a plethora of high frequency information, including news media, to guide their actions and thereby shape aggregate economic fluctuations. Traditional nowcasting approches have to a relatively little degree made use of such information. In this paper, I show how unstructured textual information in a business newspaper can be decomposed into daily news topics and used to nowcast quarterly GDP growth. Compared with a big bank of experts, here represented by official central bank nowcasts and a state-of-the-art forecast combination system, the proposed methodology performs at times up to 15 percent better, and is especially competitive around important business cycle turning points. Moreover, if the statistical agency producing the GDP statistics itself had used the news-based methodology, it would have resulted in a less noisy revision process. Thus, news reduces noise.

Keywords: Nowcasting; Dynamic Factor Model (DFM); Latent Dirichlet Allocation (LDA) (search for similar items in EconPapers)
JEL-codes: C11 C32 E37 (search for similar items in EconPapers)
Pages: 60 pages
Date: 2016-12-21
New Economics Papers: this item is included in nep-mac
References: Add references at CitEc
Citations: View citations in EconPapers (21)

Downloads: (external link)
http://www.norges-bank.no/en/Published/Papers/Working-Papers/2016/202016/

Related works:
Working Paper: Nowcasting using news topics Big Data versus big bank (2016)
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:bno:worpap:2016_20

Access Statistics for this paper

More papers in Working Paper from Norges Bank Contact information at EDIRC.
Bibliographic data for series maintained by ().