Abstract
With a novel data set, we test whether sentiment in TV news can be used as a proxy for consumer sentiment in order to explain changes in private consumption growth in the United States. The University of Michigan Index of Consumer Sentiment (ICS) is taken to compare its explanatory power with TV sentiment in classical consumer behaviour models. We find that TV sentiment can be used at least as good a proxy for consumer sentiment as the ICS, while TV sentiment can best explain private consumption behaviour with personal income and savings.
Acknowledgements
Thanks go to Jan-Egbert Sturm and two anonymous referees for helpful comments and suggestions. I thank MediaTenor for the TV sentiment data set. All errors are entirely my own.
Notes
1 See MediaTenor. Human Analysis vs. Software. Available at http://www.mediatenor.com/mca_brain_vs_software.php (accessed 1 March 2011).
2 For a more detailed description of MediaTenor's methodology, go to http://www.mediatenor.com/mca_methodology.php (accessed 1 March 2011).
3 See Federal Reserve Bank of St. Louis. ArchivaL Federal Reserve Economic Data. Available at http://alfred.stlouisfed.org/ (accessed 15 September 2010).
4 See http://www.sca.isr.umich.edu/ (accessed 15 January 2010).
5 See US Bureau of Economic Analysis. Personal Income and Outlays. Available at http://www.bea.gov/national/index.htm# personal (accessed 24 June 2010).
6 Short-term 3-month USD LIBOR interest rates and consumer price index data were obtained from Thomson Reuters Datastream.