Helluva Drug...

Partisanship’s a Helluva Drug — And It’s Reshaping the Economy

Is the economy in good shape? Your answer is influenced by partisanship and has real-world consequences.

By MASHA KRUPENKIN, DAVID ROTHSCHILD and SHAWNDRA HILL

Since his inauguration, President Donald Trump has tweeted about the economy, stock market, unemployment and/or consumer confidence at least 172 times. Read those messages, and you’ll find one consistent theme: The notion that Americans’ sky-high confidence in the economy is evidence of Trump’s popularity and success.

“Our Economy is doing better than ever,” he said in a fairly representative tweet last week. “Money is pouring into our cherished DOLLAR like rarely before, companies earnings are higher than ever, inflation is low & business optimism is higher than it has ever been.” Or, as he wrote last November: “Consumer Confidence is at Record High. I guess somebody likes me (my policies)!”

He might not have known it at the time, but with that tweet, the president stumbled onto a truth: Whether or not you support his politics affects whether or not you think the economy is good. And that, in turn, affects whether you make big purchases, like a home or a car—meaning that partisanship has a real impact on the economy.

In October 2016, only 28 percent of Republicans believed that the economy was improving. By May 2017, that number rose to 73 percent. Among Democrats, the inverse happened: In October 2016, 65 percent thought the economy was improving; by May 2017, only 35 percent did.

What happened in between that so dramatically altered how Democrats and Republicans each saw the economy? Donald Trump’s election and inauguration. Did Republicans truly believe that Trump’s presidency immediately and fundamentally strengthened the economy? And did Democrats really conclude that after Trump’s inauguration, the American economy was crushed almost overnight?

It seems as though the economy’s performance is just another one of those facts that Republicans and Democrats can’t agree on, like whether Russia interfered in the 2016 election. But unlike that issue, scholars, finance industry professionals and politicians alike regularly use consumer confidence as a benchmark for economic performance.

And that raises some interesting questions: What if the answers people give when asked about their confidence in the economy don’t reflect what’s actually happening in the economy? And if people’s partisan allegiances affect how they see the economy, what impact does that have on the economy? We are a team of researchers from Microsoft Research, and decided to dig in and try to figure out the answers.

Prior studies have shown that in some cases, Democrats and Republicans do engage in “expressive reporting” behavior, responding to surveys in ways that reflect their partisanship rather than answering a question honestly. Two academic papers published in 2013 showed that giving survey respondents financial incentives (i.e., paying them) to accurately answer factual questions about the economy (e.g. “What is the unemployment rate?”) dramatically reduced partisan differences in responses. Furthermore, partisans have been shown to engage in expressive reporting on non-economic topics, such as their post-election mental well-being.

To test whether partisans used surveys to express their true beliefs about the economy or instead engaged in expressive reporting after the 2016 presidential election, we relied upon online search data from Bing in conjunction with survey data from MSN to estimate partisan purchasing behavior both before and after the election (all data used in this study was both anonymous and opt-in).

Time and again, search data have been shown to be strongly associated with real-world behavior, especially purchasing behavior. It’s a powerful predictor of economic activity—used to accurately predict a variety of economic indicators, including the stock market, automobile sales and housing prices—and is a significantly better predictor of consumption than consumer confidence indices.

For the purposes of our study, though, search is particularly useful because it assesses whether partisans actually changed their behavior, not just their opinions, during the post-election period. If partisan survey responses are just expressive reporting, there should be no change in searches for houses or cars. If, on the other hand, partisan survey responses represent genuine perceptions of the economy, this should be reflected in partisans’ search patterns.

Our study found that after Trump’s election, Democrats, as members of the losing party, became less likely to purchase cars and invest in real estate, as demonstrated by search patterns, even when controlling for geographic and demographic variation. And despite Republicans’ extremely high reported confidence in the economy, our study did not find a meaningful change in their purchasing patterns—all of which suggests that the positive consumer confidence numbers Trump likes to cite may not accurately represent many Americans’ economic experiences or behavior.

This meaningful shift in an indicator for purchasing behavior suggests that Democrats’ negative responses to questions about the economy accurately represented their beliefs, not simply their opposition to Trump’s policies. Whether or not people shift their beliefs due to legitimate concerns about policy, partisan exposure to news or other reasons both economic or partisan, the point is that they actually do shift their real-world consumption.

The decrease in Democrats searching for housing is not trivial—it is comparable to the annual seasonal drop in housing purchases (people are significantly more likely to buy houses in June than in January). The year-over-year decrease in housing searches among Democrats after Trump’s election is equivalent to approximately 42 percent of the annual seasonal decrease between June and January. A significant proportion of Democrats who would otherwise have been in the market for a house in 2017 decided to sit out the year.

At the same time, one could assume that if Republicans were legitimately concerned about the economy under President Barack Obama—as they suggested they were in surveys—a positive change in their economic outlook would lead to a boom in spending. But we do not see that in the data. Republicans’ increase in purchases (if any) was not as pronounced as Democrats’ decrease. There are several possible reasons why.

First, survey results suggest that the losing party tends to react more strongly to a change in the presidency than the winning party. After 2000, 2008 and 2016, members of the winning party perceived the economy more or less accurately, while members of the losing party almost uniformly took an extremely negative view, despite the dramatic differences in actual economic conditions across those three elections.

Second, as predicted by prospect theory, humans have more fear of loss than desire for gain. If someone believes that the country is on the cusp of a recession, they will modify their behavior much more drastically than if they expect an economic boom. This makes a certain amount of sense: A worker who finds out they are at risk of losing their job has a stronger change in their financial behavior than a worker who finds out they might get a yearly bonus.

Third, the average everyday consumer is more affected by very negative economic outcomes than very positive ones. The majority of economic gains in the past half-century have gone to the superwealthy, while Americans across the economic spectrum suffered losses during the Great Recession. Even if Republicans genuinely believe the economy has gotten better since Trump was elected, they may not expect this to actually affect their personal economic situation. Democrats, on the other hand, may have responded more seriously to fears of another recession.

Finally, Republicans are, statistically, generally wealthier than Democrats, and have higher search rates for both cars and houses. A variety of factors go into the decision to buy a house or car—perception of the economy is just one of many—and it is possible that there were fewer Republicans near the “threshold” at which subtle factors like as economic expectations are likely to make a difference in purchasing behavior.

Economists and politicians often rely on survey estimates of public opinion to make important decisions. If survey responses aren’t an accurate guide of true, underlining sentiment about the economy—if Republicans say the economy because they support Trump, and Democrats wring their hands the economy because they oppose him—then people ought to be wary of interpreting the consumer confidence numbers President Trump likes to cite as anything other than more evidence of partisanship.

Even so, it is exciting and meaningful to show that, to an extent, partisans do act upon their incredibly acute shifts in economic confidence around elections, regardless of their true underlining sentiment.