Abstract:

The main aim of this paper is to build a real time economic sentiment indicator (RT-ESI) for Spain, based on text mining and deep learning from Twitter and Google Trends, that can anticipate GDP and household consumer behaviour. This work contributes to the literature, firstly by carrying out a sentiment analysis with a set of selected keywords that are related to emotions and expectations, then we apply a factor analysis to create the composite indicator, next we use a descriptive analysis to highlight the main associations between indexes, and finally we employ fractional integration and cointegration techniques (ARFIMA and FCVAR) to assess the RT-ESI behaviour against the European Commission´s consumer confidence indicator and the GDP. The results show that the GDP (YoY) presents the same behaviour as ourleading indicator, finding mean reversion. The behaviour of the CCI series differs from the others.

Keywords:

Economic Sentiment Indicator, Business Cycle, Text Mining, Twitter, Google Trends, Fractional Cointegration VAR.

JEL Classification:

E32, E37.

How to Cite:

Manuel Monge, Ana Lazcano and Carlos Poza. A Proposal of a Real Time Economic Sentiment Indicator Based on Twitter and Google Trends for the Spanish Economy. [ref]: vol.20.2022. available at: https://refpress.org/ref-vol20-a4/