Economic theory has long been at the intersection of prediction and the quest for understanding deeper truths behind market phenomena. Recent work by Kevin J. Lansing (2025) on “Improving the Phillips Curve with an Interaction Variable” shows that even long-standing models can still be enhanced. Lansing’s study demonstrated that by including an interaction variable in the regression model, the explanatory power of the Phillips Curve increased markedly—from accounting for 21% to 37% of the variance in the dependent variable. This significant improvement not only highlights advancements in predictive modeling but also invites a reexamination of the theoretical underpinnings of economic analysis. Lansing’s work indirectly validates Friedman’s (1953) assertion that the ultimate goal of positive science is to develop “valid and meaningful (i.e. not truistic) predictions about phenomena not yet observed”.
While predictive ability is important, I argue that a theory’s deeper value lies in its capacity to guide us toward a better understanding of economic truth—an aspect that surpasses mere prediction. Hausman (2008) describes the core purpose of theorizing as binding together relevant elements into a coherent pattern and idealizing complex phenomena to better understand causality. The ability to predict is just one dimension of a theory’s worth—a perspective that might challenge Friedman’s instrumentalist stance.
Firstly, the Phillips Curve theory originated from a realist perspective, aiming to capture the relationship between unemployment and wage inflation in the United Kingdom over nearly a century. Initially, Phillips’s approach was less about prediction and more about understanding patterns in the data. Early iterations of the model relied on strict assumptions—for example, a stable and predictable relationship between inflation and unemployment—which were later challenged by the phenomenon of stagflation during the 1970s, when high inflation overlapped with high unemployment.
Over time, the Phillips Curve has evolved; Lansing’s introduction of an interaction variable represents a crucial update that enhances the model’s predictive power. Despite these improvements, it is important to remember that the original Phillips Curve was not intended to serve as a flawless predictor. Rather, it was designed to function as a core theory—a starting point from which economists could deepen their understanding of complex economic processes.
Even Friedman (1953) acknowledges that the theory of perfectly competitive markets—a framework originally developed by Marshall—is “most useful” as a tool for outlining a general causal pattern across various contexts. However, the idealized assumptions underlying this theory render it less effective when directly applied to real-world predictions. The assumptions must be “de-idealized” one by one to mirror actual economic conditions more accurately. This necessary modification does not diminish the inherent value of the theory; instead, it reinforces its role as a steppingstone toward better, more nuanced predictive models. The insights collected from these theories remain invaluable for understanding the complexities of economic interactions.
Lastly, Maki (2009) argues that Friedman’s work can also be seen as an expression of realism. Maki believes that “F53 could be rewritten as an unambiguous and consistent realist manifesto” (p. 113). While many view Friedman’s arguments as representative of instrumentalism—suggesting that the worth of a theory is measured solely by its predictive success, Maki offers an alternative reading. Maki argues that Friedman implicitly encourages economists to assess the realism of their assumptions; that is, to consider whether the assumptions are sufficiently grounded in reality to yield correct predictions. Such an interpretation bridges the gap between the instrumentalist focus on prediction and a more realist commitment to uncovering the true causal mechanisms underlying economic phenomena.
In summary, while predictive accuracy remains a vital component of economic theory, it is not the sole measure of a model’s worth. Economic theories must also offer a realistic portrayal of complex market dynamics, grounding their assumptions in the nuances of real-world behavior. Instead of adhering solely to Friedman’s instrumentalist approach, I propose a balanced perspective that equally values robust predictive capabilities and the depth of realistic assumptions. This dual emphasis not only enriches our understanding of economic phenomena but also lays the groundwork for developing more comprehensive and adaptable models.
References:
- Kevin J. Lansing (2025). “Improving the Phillips Curve with an Interaction Variable.”
- Friedman, M. (1953). The Methodology of Positive Economics. In Essays in Positive Economics (pp. 3–43).
- Maki, U. (2009). Unrealistic assumptions and unnecessary confusions: rereading and rewriting F53 as a realist statement. In U. Mäki (Ed.), The Methodology of Positive Economics: Reflections on the Milton Friedman Legacy (pp. 90–116). Cambridge: Cambridge University Press.
- Caldwell, B. (1992). Friedman’s predictivist instrumentalism – A modification. Research in the History of Economic Thought and Methodology, 10, 119–128.
Leave a Reply