Economics avoids a liberal or conservative bias because it is grounded in timeless established principles and the objectivity of math.
In a recent study, researchers sought to see if the content of academic papers from economists indicated their political preference. Starting with 53,000 economists, 62,888 research articles and 17,503 working papers, the data was voluminous. Oversimplifying what they describe in a 60+ page paper, we can just know that they first created a list of partisan words generated from a control group of 2,000 economists. With those words they developed an algorithm that was then applied to a new group of economists’ papers. The goal was to see if the terms signified the political bias of the writer.
And yes, 74.1 percent of the time, the algorithm accurately identified an economist’s politics. Those politics, though had an effect beyond writing bias. They also shaped where a person specialized. To the right, we would find the macroeconomists, financial economists and business school professors. On the left, meanwhile, are the labor economists.
These are some of the words left-and right-leaning economists use when writing about macroeconomics:
Long ago, both applying the same basic theory of rent, Thomas Malthus and David Ricardo sent Parliament conflicting advisory letters on the Corn Laws of 1815. With Malthus supporting more domestic production and Ricardo advocating free trade, each gentleman used seemingly objective economic analysis. Each though echoed his political preference.
Similarly, today, when economists provide expert counsel to politicians, their substantiating data and recommendations could reflect their bias. As the authors of “Political Language in Economics” state, “…with predicted liberals reporting elasticities that imply policies consistent with more interventionist ideology” and the reverse for predicted conservatives, economic advice reflects political bias.
Our Bottom Line: Confirmation Bias
When economic experts or investment analysts or pharmaceutical researchers generate conclusions with data that supports what they already believe, we can ask whether they are displaying confirmation bias.