The formal representation of the history-friendly models presented some notable issues, first of all because of the huge amount of variables and parameters defining the models: some of these elements were common or at least analogous across the models, while others referred to completely different domains. In order to reduce the number of the main symbols to a manageable size, we adapted from computer programming languages the idea of overloading notation: a main symbol can have slightly different meanings according to the presence or absence of further details, such as superscripts and subscripts. For example, the symbol T indicates the total number of periods of a simulation, Tk indicates the period of introduction of technology k, and TI the minimum number of periods a firm has to stay integrated after its decision to switch to internal production of components. In general, we use as subscripts the indices for elements (products, firms, markets, technologies) that take different values, without changing the meaning of the main symbol. Instead, we use as superscripts further identifiers of the main symbols that are not instances of a general category: for example, PT is the symbol of patents and E is the symbol of exit. In a very limited number of cases an identifier can be used both as a subscript identifier (TR and MP in most of the cases are used as instances of component technology k) and as a superscript identifier (TR and MP are used as superscripts of the main symbol α, as they refer to different parts of the same equation).

Upper and lowercase letters are considered as different, although whenever it is possible they take related meanings: for example, i indicates the propensity to integrate and I the corresponding probability. The symbols used for specific variables and parameters are not used across models, unless these variables and parameters have the same or a very similar meaning and role in the different models. The values that parameters take and the range of values that heterogeneous parameters and variables can take are indicated in the tables in the Appendices.

INTRODUCTION

This chapter presents a “history-friendly” model of the evolution of the pharmaceutical industry, and in particular of the so-called golden age. This industry is an ideal subject for such an analysis, especially because it has characteristics and problems that provide both contrasts and similarities with the computer industry. Like computers, the pharmaceutical industry has traditionally been a highly R&D and marketing-intensive sector and it has undergone a series of radical technological and institutional “shocks.” However, despite these shocks, the core of the leading innovative firms and countries has remained stable for a very long period of time. Entry by new firms has been a rather rare occurrence until the advent of biotechnology. However, while the evolution of computers coincides largely with the history of very few firms, that of pharmaceuticals involves at least a couple of dozens of companies. Further, the degree of concentration has been consistently low at the aggregate level and the industry has never experienced a shakeout of producers.

We argue that the observed patterns of the evolutionary dynamics were shaped by three main factors, related both to the nature of the relevant technological regime and the structure of demand:

Having sketched the domain of our inquiry, we turn now to the task of bringing new light to that domain. While the empirical phenomena discussed in the first chapter are complex and variegated, the broad subject matter of innovation and industrial evolution is familiar to a great many economists. Most of our readers know what it is about. On the other hand, the promise of “history-friendly modeling” as an approach to that subject matter is less widely appreciated. A principal purpose of this book is to lay out this new methodological approach and to demonstrate its usefulness. In this chapter, we develop the basic argument for this approach, which we then apply in the remainder of the book. We explain our reasons for believing that history-friendly modeling is a valuable addition to the economists’ analytic tool kit. In doing so, we discuss considerations ranging from the extremely general – “what forms of economic theory help us to understand the world” – to the very specific – “how does one construct a history-friendly model?” We follow the indicated order, from the general to the specific.

COMPLEMENTARY APPROACHES TO MODELING: “FORMAL” vs. “APPRECIATIVE” THEORY

Our development of the concept and technique of history-friendly modeling reflects our belief that present notions of what constitutes “good theory” in economics are inadequate to the task of analyzing and illuminating the kind of phenomena this book addresses. Theorizing about complex phenomena involves, first of all, discriminating between the most salient facts that need to be explained, and phenomena that can be ignored as not relevant or interesting, and second, identifying the most important determining variables, and the mechanisms or relationships between these and the salient phenomena to be explained. This requires screening out or “holding constant” variables and mechanisms that seem relatively unimportant, at least given the questions being explored. There is probably little disagreement about these general points. Regarding the principles for assessing theories, there is more controversy.

The illumination provided by a theory can be assessed on at least two grounds. The criterion most recognized by economists today is its power to suggest hypotheses that will survive sophisticated empirical tests, where confirmation is found in the statistical significance levels achieved in the tests and in measures of “goodness of fit.”