We analyze a model of vertical (dis)integration between manufacturing and design in a monopolistically competitive market. Specialized input manufacturers can serve multiple design firms and the manufacturer-designer pairs negotiate a non-binding contract to share input customization cost and production surplus. Hand-collected data on 387 product lines from 118 semiconductor firms are used to predict the firm’s decision to outsource manufacturing. We find that, for instance, the use of design tools that facilitate collaboration and process technologies that facilitate learning are both positively associated with outsourcing, consistent with the model’s prediction.

Introduction

With the technological advancement in the 20th century and the establishment of such pioneering firms as Fairchild Semiconductor, practicable semiconductor devices, especially in the form of integrated circuits (ICs, or ‘chips’), have shaped the evolution of electronics industries. Originally, semiconductor devices were produced by the so-called integrated device manufacturers (IDMs) having both the capacity to design and manufacture semiconductor devices. A well-known example is Intel, which developed the first microprocessor in 1971 and still dominates the microprocessor product market today. By the late 1980s, however, the so-called foundry model came into being. The foundry model refers to the separation of chip design and fabrication process into different business entities.1 Design firms are often called ‘fabless’ and specialized manufacturers are called ‘foundries.’ Concurrently, most of the foundries were established in Asia, such as Taiwan Semiconductor Manufacturing Company (TSMC, founded in 1987), and there are numerous fabless companies today in many parts of the world specializing in chip designs for a broad range of applications from mobile phones to internet-of-things to self-driving cars. Although a number of industry reports (such as McKinsey & Co’s) and newspaper articles have talked about what might have given rise to the popularity of the foundry model, there have been relatively few studies that rigorously investigate such arguments theoretically and/or empirically, especially using data from the 21st century (see below). This paper fills this gap by presenting an industry equilibrium that explains the conditions under which a stable outsourcing equilibrium may arise and empirically showing that the factors that are highlighted in the model as the drivers can indeed explain a significant variation in the data.