Research Topics

This paper offers a retrospective view of the key pillar of Solow’s neoclassical growth model, namely the aggregate production function. We review how this tool came to life and how it has survived until today, despite three criticisms that undermined its raison d’être. They are the Cambridge Capital Theory Controversies, the Aggregation Problem, and the Accounting Identity. These criticisms were forgotten by the profession, not because they were wrong but because of the key role played by Robert Solow in the field. Today, these criticisms are not even mentioned when students are introduced to (neoclassical) growth theory, which is presented in most economics departments and macroeconomics textbooks as the only theory worth studying.

A Caveat Emptor for Regional Scientists

Over the last 20 years or so, mainstream economists have become more interested in spatial economics and have introduced largely neoclassical economic concepts and tools to explain phenomena that were previously the preserve of economic geographers. One of these concepts is the aggregate production function, which is also central to much of regional growth theory. However, as Franklin Fisher, inter alios, has shown, the conditions necessary to aggregate microproduction functions into an aggregate production function are so stringent that in all probability the aggregate production function does not exist. This paper shows that the good statistical fits commonly found empirically are solely due to the use of value data and an underlying accounting identity. The result is that the estimates obtained cannot be regarded as providing evidence of the underlying technological structure of the spatial economy, including the aggregate elasticity of substitution, the degree of returns to scale, and the rate of technical progress.

Further Reflections on Temple’s Criticisms and Misunderstandings

In a reply to Felipe and McCombie (2010a), Temple (2010) has largely ignored the main arguments that underlie the accounting identity critique of the estimation of production functions using value data. This criticism suggests that estimates of the parameters of aggregate production functions cannot be regarded as reflecting the underlying technology of the industry. While Temple concedes some points, he erroneously believes that the critique holds only under some ad hoc assumptions. As a consequence, he argues that the critique works only “part-time.” This rejoinder discusses Temple’s arguments and demonstrates that the critique works full-time.