Image Source: Pexels

Macroeconomics is at times unfavorably compared to weather forecasting. But weather forecasts have in fact become more accurate over time: a forecast for four days inthe future, made today, today is as accurate as a one-day forecast 30 years ago. Can macroeconomics make at least a broadly similar claim to improvement? Olivier Blanchard makes the argument in “Convergence? Thoughts about the Evolution of Mainstream Macroeconomics over the Last 40 Years” (Peterson Institute for International Economics, Working paper 25-8, May 2025). He writes:

Let me state my two main conclusions. First, starting from sharply different views, there has been substantial convergence, both in terms of methodology and in terms of architecture. Second, this convergence has been mostly in the right direction, allowing future research to build on the existing conceptual structure. Put strongly, macroeconomics may have a claim to calling itself a mature science. … As macroeconomists, we should stop self-flagellating and not accept flagellation from others.

As in all arguments about what constitutes “science,” the way in which an author defines terms matters. Blanchard uses a definition of “mature science” that includes factors like whether there is an established theoretical framework, in which researchers agree about standars of evidence, such that knowledge can be refined and can accumulate over time. A mature science needs to provide practical knowledge for addressing real-world problems. For example, pretty much every central bank in the high-income countries now uses a New Keynesian model for understanding and forecasting changes in the economy.

The specific model on which agreement has been reached, according to Blanchard is called “New Keynesian.” On one side, this model allows for households and firms to react to incentives, and to shifts in expectations about the future, and thus is built upon macroeconomic behavior. But on the other side, the model does not require that these markets involve perfect competition or smooth outcomes: that is, prices and wage can be slow to adjust, imperfect competition and imperfect information can play substantial roles, technology can shift, and more. (For those who would like more detail, Jordi Galí offers an overview of this approach in the Summer 2018 issue of the Journal of Economic Perspectives, where I work as Managing Editor.)

Blanchard argues that the virtue of the basic New Keynesian model is its flexibility: that is, it allows analyzing the effects oif a wide array of topics. \

To mix metaphors, I see the minimalist model as the basic unit in an erector set. By itself, the basic unit is not extremely useful, but you canplug into it a whole set of extensions. You can extend it to introduce myopia … and reduce the role of expectations. You can replace rational expectations with other expectation formation mechanisms. You can extend it to include borrowing constraints, which lead to a more important role for current variables and more realistic consumption dynamics. You can extend it to more than one country. You can extend it to introduce various forms of heterogeneity and derive aggregate implications. In short, it provides a common and generally understood structure from which to start and organize research and discussion.

Blanchard readily admits that the predictive power of the New Keynesian macroeconomics is limited, but I do not view that as a fatal flaw. After all, every new discovery in any field of science suggests that the previous prediction made by the subject was wrong in some way. Social sciences like economics have the additional problem that they built on some ever-shifting combination of people, institutions, and events, rather than on an understanding of fascinating but personality-free subjects like chemistry, biology, or properties of matter and energy. As a social science, economics also suffers from a feedback mechanism where improvements in macroeconomic thinking will alter the behavior of central banks and large firms, as well as affecting other policymakers and households, which in turn willl require additional improvements in macroeconomic thinking.

In short, when Blanchard refers to a “mature science,” he is talking about a framework that has proven useful and flexible for analysis, not making a claim that economists possess a crystal ball about the future. Greg Mankiw recently wrote a letter to the Wall Street Journal that encapsulated some of this perspective. Mankiw wrote:

I always find it amusing when people assert that economics isn’t a science. Such statements suggest that they don’t know what scientists do. Here’s a reminder: Scientists observe the world. They develop theories that aim to explain what they see. They collect data to test their theories and reject those that don’t conform to the data. They try their best to put aside ideological preferences and preconceived notions. Most important, they always remain open to changing their minds when presented with better theories or new data. This approach can be applied whether one is studying apples falling from a tree or gross domestic product fluctuating over time. As Albert Einstein put it: “The whole of science is nothing more than a refinement of everyday thinking.”

More By This Author: