Public opinion as nowcast: consistency and the role of news uncertainty

ABSTRACT

The purpose of this note is to consider the effect of perception noise when voters form public opinions. We provide a simple theoretical framework that will form the basis to investigate empirically the effect of news uncertainty on voters’ attentiveness when forming public opinion, or nowcasts. An attentive voter will consistently update their information set. Therefore, if voters’ nowcasts are consistent, any revision of the nowcasts must only reflect new information. We specifically consider how news uncertainty may affect voter attentiveness. The paper focuses on US presidential competence and popularity indices. We find that the nowcasts are consistent during periods of low news uncertainty but highly persistent when news uncertainty is high.

KEYWORDS:

• Public opinion polls
• perception noise
• voters’ nowcasts
• random walk
• non-linear time series analysis

Notes

¹ Fan and Cook (2003), following an established tradition of using economic news reported in the press, predict the time trend of US sentiments index based on variables used in discussions of the economy in the media.

² This is consistent with other recent assessments of informational uncertainty (see, for example, Bracha and Weber (2012) and Smithson (2008)) .

³ Permanent innovations can be both negative and positive.

⁴ The economic variables and targets by which voters can assess incumbent competence maybe clear but as Lott and Hassett (2014) analysis clearly shows they are subject to bias media reporting.

⁵ It is important to note that while the source of the information may be subjective and bias, the perception noise is not.

⁶ This is also congruent with the data used for empirical analysis in the next section.

⁷ When using a micro-founded macro-modeling approach this common way to determine the average agent’s forecast of macroeconomic variables (see, for example, Coibion & Gorodnichenko, 2015, p. 2650).:

⁸ We assume: $E({\kappa }_t\left[C_t-F_{t-1}{C}_{t-1})\right])=0$, this is consistent with our earlier definition that individual perception noises (${\omega }_{it}$) (which also includes common perception noise ${\kappa }_t$) are not serially correlated.

⁹ Please note that we use the term ‘popularity’ as the public’s, or voters,’ perception of the incumbent’s overall competence or administrative ability. This will be clearer when we describe and discuss the data in the next section.

¹⁰ Further details pertaining to the SRC series can be obtained at the respective websites; http://www.sca.isr.umich.edu.

¹¹ The Gallup measure and methodology are available at: https://news.gallup.com/interactives/185273/presidential-job-approval-center.aspx.

¹² An extensive discussion of the construction of this index can be found in Bloom et al. (2015).