The national unemployment rate and GDP growth indicators tell the same story: the US economy is improving. At the same time, there is remarkable variation in Americans’ perceptions of the state of the economy with some polls showing as many as 60% reporting that “economic conditions in the country as a whole are getting worse” according to Gallup.1 Why is there so much variation in economic perceptions and how might we square the generally positive outlook reported in national economic indicators with the more diverse economic evaluations of Americans?
Academics have long studied the political roots of economic evaluations (e.g. Duch, Palmer and Anderson 2000; Bartels 2002). While objective economic conditions certainly play a role in shaping individuals’ perceptions of of the economy, perceptions are also colored by ideology and party allegiances. Partisans opposed to Barack Obama and his policies will tend to have more negative evaluations of the economy than similarly situated supporters of the president. But even when we account for partisanship, the variability of economic perceptions remains; as if some voters are systematically exposed to different economic conditions.
Our recent article suggests that this is indeed the case. The “national” economy is really the sum of 50 state economies (or a few thousand counties, or several thousand cities. . . ) and each citizen in each state is exposed to their own local economy as well as their own national economy. More specifically, national economic indicators (GDP, national unemployment, etc.) simply aggregate the 50 state economies, sometimes weighted by population, sometimes just summed. But economic discourse, the things that people talk about when they discuss the economy, doesn’t necessarily combine that information in the same way.
For instance, Texans have more in common with Oklahomans (economically, politically, etc.), than they do with Hawaiians, so the performance of Oklahoma’s economy will bear more weight in Texas’ economic discourse than Hawaii’s economy. This means that each state not only has its own local economy, but it also has its own distinct picture of the national economy. Our analysis shows that these localized national economic measures explain Americans’ perception of the economy much better than the typical measures we use, like GDP, and also account for a substantial amount of the variation we observe in economic perceptions.
So what drives economic discourse and its variation across states? We find that the media play an integral role in shaping local economic views. When the media report on the economy in a particular state, that reporting typically comes with some degree of comparison to other states (“…unemployment in Texas fell by 1%, while Louisiana saw its unemployment grow by 0.5%…”). We find that the rate at which a particular state is compared to all other states produces the best set of “weights” for those states in creating the local picture of the national economy.2 A crude example: if reports of Texas’ well-being come with comparison to Oklahoma half the time, Louisiana a quarter of the time, and California a quarter of the time, then Texas’ picture of the “national” economy should be about 50% Oklahoma, 25% Louisiana, and 25% California.3
Figure 1 maps each state’s rate of comparison to Texas and therefore each state’s contribution to the typical Texan’s view of the national economy—states in darker green were mentioned more in reporting on the Texas economy than those in lighter shades.
Figure 1: Connections Among Local Economies Derived from Media Mentions in Economic News (Texas).
We used these comparison rates to construct a different “national” economy for all 50 states (weighting a state’s economic performance by its rate of comparison to the state in which the respondent lived) and then evaluated how well these media derived national economies predicted voters’ perceptions of economic well-being. We test our expectations using survey data from the Cooperative Congressional Election Study (2006, 2008-2012). Using spatial econometrics (spatial-X ordered logit) we model responses to the survey question “would you say that over the past year the nation’s economy has (1) gotten better, (2) stayed the same, or (3) gotten worse.”
Figure 2 shows the increased likelihood of a respondent identifying the economy as having improved (y-axis) following an increase from 1% to 3% in Gross Domestic Product (GDP) per capita (a measure of national growth), Gross State Product (GSP) per capita (a measure of state growth), and our contextualized state index (state performance weighted by media mentions).
Figure 2: Change in the Predicted Probability of Selecting “Better” given an Increase (1% to 3%) in Gross State Product Per Capita and Gross Domestic Product Per Capita.
Note: Vertical lines represent 95% confidence intervals. Dashed line depicts the null hypothesis test.
The data suggest that the localized national economies we constructed from media analysis for each state are substantially more predictive of Americans’ economic perceptions than the usual national indicator. In fact, the improvement in economic perceptions following an increase in contextualized state conditions (+0.14) is nearly three times as big as the improvement following a similar increase in national conditions (+0.05). We believe that this is because our measure is a closer approximation of the economic discourse in a particular state. That is, the economy is not a point, it’s a distribution and just as some parts of the economy improve while others may suffer, some parts may be salient to some parts of the country while others are irrelevant. These differences across states drive differences in the distribution of economic messages to which citizens are likely to be exposed. It’s natural that people in Louisiana have a weaker than average view of the economy — not only is Louisiana’s local economy suffering because of low oil prices, but the picture of the “national” economy presented by local media is dominated by similarly struggling states.
To sum up, respondents have different opinions of the strength of the national economy because they are looking at different pictures of the same national economy. The distribution of economic information that respondents draw from varies across contexts and scholars who fail recognize this variability limit our understanding of how individuals acquire economic information, and use that information to evaluate the economy.
- See Kayser and Peress (2012)
- These connections were derived from an automated content analysis of state-level economic news articles obtained from Lexis-Nexis for 2000 to 2015. Articles that referenced the state’s governor in the headline and any economic keywords in the headline or body of the text were collected for each state. In total 11,578 articles were collected and analyzed for mentions of potential economic comparisons.
- Bartels, Larry. 2002. “Beyond the Running Tally: Partisan Bias in Political Perceptions.” Political Behavior 24:117–150.
- Clarke, Kevin A. 2003. “Nonparametric Model Discrimination in International Relations.” Journal of Conflict Resolution 47(1):72–93.
- Clarke, Kevin A. 2007. “A Simple Distribution-Free Test for Nonnested Model Selection.” Political Analysis 15(3):1–17.
- Duch, Raymond M., Harvey D. Palmer & Christopher Anderson. 2000. “Heterogeneity in Perceptions of National Economic Conditions.” American Journal of Political Science 44(4):635–652.
- Duch, Raymond M. & Randolph T. Stevenson. 2008. Voting in Context: How Political and Economic Institutions Condition the Economic Vote. Cambridge: Cambridge University Press.
- Kayser, Mark & Michael Peress. 2012. “Benchmarking across Borders: Electoral Accountability and the Necessity of Comparison.” American Political Science Review 106(3):661–684.