Graham Beattie


4901 Posvar Hall
Pittsburgh, PA 15260

Research Interest

Environmental Economics, Public Economics, Political Economy, Applied Microeconomics

Job-market Paper

The social cost of "fake news": The effect of media coverage about climate change on behavior

Biased or inaccurate media coverage receives substantial criticism for distorting public beliefs. In this paper, I quantify this effect by examining media coverage of climate change, which has been shown to be biased towards skepticism. Skeptical viewpoints receive considerable coverage despite representing a minority of scientific views. I investigate the consequences of this bias
by studying the effect of media coverage of climate change on individual driving behaviors, which are a major source of the greenhouse gas emissions that cause climate change. I construct a measure of the tone of newspaper coverage about climate change based on comparisons of environmental and skeptical texts. I use this measure, along with detailed information about driving patterns, to show that households that have recently received more environmental coverage make travel decisions that reduce their emissions. These effects are strongest within one to two weeks of coverage and when there are lower-emissions substitutes available. By comparing the effect of environmental coverage to the effect of more skeptical coverage about climate change, I demonstrate that bias in media coverage about climate change has real effects on environmental behaviors. I find that not printing an environmental article about climate change or replacing it with a skeptical article has a social cost of at least 3¢ per subscriber.

Other Works

Advertising, media capture, and public opinion: The case of climate change

This paper analyzes the divergence between the scientific consensus and public opinion about climate change through the lens of media capture. I develop an objective measure of the tone of climate change coverage by creating an index that builds on Gentzkow and Shapiro (2010), using phrase frequency analysis to compare newspaper text with the UN's IPCC reports and the Heartland Institute's skeptical response. I also develop a measure of potential advertising from the auto industry, using national shocks to advertising at the vehicle-make level weighted by previous vehicle-make advertising at the newspaper level. The empirical analysis shows that this potential advertising allows firms to capture newspapers: within-newspaper, the potential for advertising from the auto industry both reduces the quantity of coverage and shifts the tone toward skepticism. I also find suggestive evidence that changes in coverage of climate change are correlated with changes in public opinion. The results indicate that, during the period 2005-2008, media capture led more than 1.2 million Americans to prioritize economic growth over  environmental protection.

Advertising spending and media bias: Evidence from news coverage of car safety recalls

Do news media bias content in favor of advertisers? We examine the relationship between advertising by auto manufacturers in U.S. newspapers and news coverage of car safety recalls. This context allows us to separate the influence of advertisers, who prefer less coverage, from that of readers, who demand more. Consistent with theoretical predictions, we find that newspapers provide less coverage of recalls by their advertisers, especially the more severe ones. Competition for readers from other newspapers mitigates bias, while competition for advertising by online platforms exacerbates it. Finally, we present suggestive evidence that lower coverage increases auto fatalities.

Thrivers and divers: Using non-academic measures to predict academic success 

We collect a comprehensive set of non-academic characteristics for a representative sample of incoming freshman to explore which measures best predict the wide variance in first-year college performance unaccounted for by past grades. We focus our attention on student outliers. Students whose first-year college average is far below expectations (divers) have a high propensity for procrastination–they self-report cramming for exams and wait longer before starting assignments. They are also considerably less conscientious than their peers. Divers are more likely to express superficial goals, hoping to 'get rich' quickly. In contrast, students who exceed expectations (thrivers) express more philanthropic goals, are purpose-driven, and are willing to study more hours per week to obtain the higher GPA they expect. A simple seven-variable average of these key non-academic variables does well in predicting college achievement relative to adding more variables or letting a machine-algorithm choose. Our results, descriptive in nature, warrant further research on the importance of non-linearities for the design and targeting of successful interventions in higher-education.

What sets college thrivers and divers apart? A contrast in study habits, attitudes, and mental health 

Students from 4-year colleges often arrive having already done very well in high school, but by the end of first term, a wide dispersion of performance emerges, with an especially large lower tail. Students that do well in first year (we call the top 10 percent Thrivers) tend to continue to do well throughout the rest of their time in university. Students that do poorly (we call the bottom 10 percent Divers) greatly struggle and are at risk of not completing their degree. In this paper we use a mandatory survey with open ended questions asking students about their first-year experience. This allows us to explore more closely what sets Thrivers and Divers apart, in terms of study habits, attitudes, and personal experiences. We find that poor time management and lack of study hours are most associated with poor academic performance, and that those who struggle recognize these weaknesses. Divers also report feeling more depressed and unhappy  with their lives. We posit an 'academic trap', whereby initial poor performance is related to poor time management which in turn lowers expectations, which in turn leads to lower study time, and so on. Thrivers, in contrast, study significantly more and meet with course instructors.

Learning about climate change: Social signals and solar panel adoption 

Despite overwhelming scientific consensus that climate change is human induced, public perception varies considerably. Exploiting the rapid growth of rooftop solar panels, we explore the extent to which households learn about climate change from the actions of their peers. Using a large repeated cross section and an instrumental variables strategy, we find evidence that, within postcode, increased solar adoption has a small positive effect on the likelihood that households believe climate change is human induced. However, we also find that higher solar penetration reduces concern about the impacts of climate change which may dampen additional abatement effort. Our results suggest that peers and government policies have important and subtle effects on the beliefs and actions of neighbors and constituents.

Biased media in an unbiased market 

It is well documented that biased media coverage can arise because of biases on the part of consumers (Gentzkow and Shapiro, 2010) or media outlets (Baron, 2006). I present a model in which biased coverage can occur even if neither media outlets nor their consumers are biased.

Topic analysis of climate change coverage in the UK 

The UK newspaper market is dominated by large national newspapers that compete for the same set of readers. I use topic modelling techniques to analyze how newspapers compete with each other through their coverage of climate change. Coverage across newspapers is highly correlated and driven by news events, but newspapers differentiate themselves by having areas of focus and increasing coverage more significantly when events related to those areas arise.

Job Market Paper Link