Business Strategy and Firm Location Decisions: Testing Traditional and Modern Methods
Abstract
For nearly a century, economists have relied upon the neoclassical principle of a profit-maximizing firm. Challenges to this principle have recently arisen: the theory of the value-maximizing firm, and machine learning methods. We make use of an unusual natural experiment, and extensive data, to empirically compare the predictive power of both traditional and modern methods.We proceed with:
1. Outline competing models of business decisions from both traditional, and modern, approaches: expert judgment; an income model of a profit-maximizing firm; a suite of machine learning models; and a recursive model of a value-maximizing firm.
2. Assemble data on costs, productivity, workforce, transit, and other factors for over 50 large North American cities.
3. Empirically test these approaches against each other, to determine which best explains the selection of 20 cities by Amazon Inc. for its HQ2.
We observe that expert judgment, of the type traditionally performed by business economists, outperformed a suite of machine learning models--even though these supervised learning models benefited from data unavailable to the experts. Indeed, some machine learning models performed worse than a coin flip. Second, we found that the novel model of a value-maximizing firm slightly outperformed an income model using exactly the same underlying data, and captured valuable insights that the traditional model missed. Based on these results, we recommend that business economists consider value methods for business strategy decisions. We also warn against the naive reliance on machine learning methods, particularly when the potential costs of errors are high. Finally, for forensic economists, we note frontier areas where the growth of machine learning methods will produce challenges in future practice.