A groundbreaking study, Political Language in Economics, reveals how political ideology shapes economic research by analyzing academic writing patterns using machine learning techniques. The researchers found that economists’ political leanings, predicted from their writing style, correlate with their empirical findings on policy-relevant issues: conservative economists tend to find evidence supporting free-market policies. In contrast, liberal economists discover data backing government intervention. They also documented clear ideological sorting across different fields, with labor economics attracting more liberal scholars while finance and macroeconomics drawing more conservative ones. This correlation between ideology and research findings flows through even technical academic papers that are supposedly objective.
Does this mean economics is just subjective pseudo-science? No! As philosophy of science showed us, there is no such thing as truly objective science – as Kuhn and Feyerabend demonstrated, even physics contains subjective elements. The dismal science is no exception. While this study offers valuable insights, it could benefit from more modern topic modeling approaches and the rich toolkit of corpus linguistics to deepen our understanding of how ideology manifests in economic discourse.
The researchers developed a sophisticated machine learning approach that analyzed language patterns in academic papers and connected them to economists’ political behaviors, such as campaign contributions and petition signatures. Their analysis revealed that economists’ political leanings could be predicted from their academic writing style, despite the supposedly neutral nature of academic papers. This finding suggests that partisan preferences fundamentally influence how economists discuss their research. Furthermore, they discovered distinct ideological patterns across different fields of economics, with labor economics attracting more liberal scholars while finance, macroeconomics, and industrial organization tend to draw more conservative ones. Business school faculty also showed a tendency toward more conservative viewpoints compared to their economics department counterparts.
Most significantly, they found that an economist’s predicted political ideology correlates with their empirical findings on policy-relevant issues. This was particularly evident in tax policy analysis, where economists’ political leanings significantly influenced their findings about optimal tax rates – conservative economists typically found results supporting lower rates, while liberal economists found support for higher rates.
While the study’s methodology is sound within its framework, it could be enhanced in two major areas. First, the authors rely on relatively simple topic modeling approaches like Latent Dirichlet Allocation and Correlated Topic Model, when modern natural language processing offers more sophisticated tools such as transformer-based modeling, neural topic models, and dynamic topic models that could better capture how economic discourse evolves over time. Second, the rich methodological toolkit of corpus linguistics could provide more sophisticated approaches to exploring contextual meaning beyond simple word counts and better integration of qualitative and quantitative methods.
These methodological considerations aside, the study’s findings align with what philosophers of science have long understood about the nature of scientific knowledge. As Kuhn showed us, a scientific paradigm isn’t merely a theoretical framework – it’s an entire scientific culture that includes shared exemplars, practices, institutional structures, and ways of seeing the world. While physics and most natural sciences typically operate within a single dominant paradigm at any given time (like quantum mechanics and general relativity in their respective domains), social sciences, including economics, often maintain multiple competing paradigms simultaneously. In economics, these various paradigms – from neoclassical to Post-Keynesian to institutional approaches – are perpetuated through their own textbooks, graduate training programs, peer review processes, and professional rewards systems. The dominant neoclassical paradigm, for instance, isn’t just a set of theories about markets and rational agents – it’s embedded in how mainstream economics is taught, how research is evaluated, what gets published in top journals, and what kinds of questions are considered legitimate for investigation. This paradigmatic pluralism in economics, while sometimes creating methodological and theoretical tensions, might actually be beneficial for understanding complex social phenomena that resist single, unified explanations.
This institutional aspect of paradigms helps explain why changing fundamental assumptions in economics is so difficult. The 2008 financial crisis, for instance, might have seemed to invalidate many mainstream economic models, but the basic paradigm proved remarkably resistant to fundamental change – precisely because paradigms are social and institutional constructs, not just intellectual frameworks that can be falsified by evidence.
Understanding the role of ideology in economic research doesn’t diminish its scientific value – rather, it helps us better appreciate the complex nature of social scientific inquiry. Just as physics progresses through competing theories and paradigms, economics benefits from the tension between different perspectives. The key is not to eliminate ideology – an impossible and perhaps undesirable goal – but to acknowledge it transparently and understand how it shapes our understanding of economic phenomena. This study takes an important step in that direction.
Leave a Reply