This essay examines the argument given by Richard Dawid (2013, 2019) for the viability of non-empirical assessment of scientific theories. Dawid's argument is supposed to show that trust in a scientific theory can be justified without any direct empirical testing of the theory. This view is fundamentally different from what will be called the classical paradigm of theory assessment. The classical paradigm holds that only empirical testing can justify belief in a theory. It is argued in this essay that Dawid's argument does not provide sufficient reasons for claiming that non-empirical assessment can be seen as a valid form of justification of scientific theories. However, it is further argued that non-empirical assessment still can play an important role when evaluating the status of a theory that cannot yet be tested empirically.

<p>Principal agent theory is widely used to model supply chain relationship, in which a supplier is the agent and a manufacturer is the principal. Both the manufacturer and supplier can influence product quality and consequentially share costs of product failures. Rich theoretical results under the principal agent model framework have been accumulated in the last two decades, but empirical evidence on whether the Stackelberg’s leadership game truly imitates practical supply chain relationship remains unfound. We study the domestic automobile industry in the last decade and provides to our best knowledge the first empirical evidence to assess the validity and practicality of principal agent theory and draw the implications of principal agent theory on supply chain relationship costs. Our empirical results suggest that Japanese OEMs behave more like principal agent theory suggests than the US OEMs in general and thus gain significant benefits in terms of marginal effort costs in motivating suppliers’ quality improvement behaviors and reducing overall manufacturer’s quality costs. Specifically, Toyota behaves closest to the optimal solution in the principal agent theory and therefore has the lowest manufacturer effort costs in improving product quality and achieves the overall lowest manufacturer’s quality costs in supply chain. Honda and Nissan are ranked 2^nd and 3^rd in terms of principal agent behaviors, but their marginal quality improvement effort costs are 33% and 61% higher than Toyota, and their total manufacturer’s quality costs are both around 17% higher compared to industrial leader Toyota by our estimate. US OEMs GM, Ford and Chrysler are believed to behave inconsistent to principal agent theory suggest, and consequently suffer a much higher marginal effort cost in motivating supplier’s quality improvement than Toyota as well as the overall manufacturer’s quality costs. GM and Ford are estimated doubled marginal effort costs than Toyota, and Chrysler is even higher at 1.6 times. GM’s overall manufacturer’s quality cost is 24% higher than Toyota, Ford is around 31% higher and Chrysler is around 48% higher. Our analysis gives a new perspective from principal agent theory to explain why Japanese OEMs especially Toyota has a better supply chain quality costs than US OEMs as literature and consensus suggested. In addition, we contribute in literature by linking the principal agent theory with automotive industrial data and first ever empirically validate the legitimacy of principal agent theory in modeling manufacturer-supplier relationship and quantitatively derive practical conclusions on marginal effort costs and manufacturer’s total supply chain quality cost implications. To guarantee the robustness of the empirical results, various sensitivity analyses are conducted and our main conclusions remain unchanged. </p>