Antecedents and performance implications of contracting for knowledge workers: evidence from information technology services.

This paper develops a theory that predicts why firms organize their knowledge workers as employees versus independent contractors and predicts the performance implications of this choice. It then empirically examines this organizational choice--which our theory predicts will be driven by contracting difficulties arising from expropriation concerns, measurement costs, and interdependence--and its implications for profitability for 190 information technology service projects. Using a two-stage switching regression model, our analysis shows that projects aligned according to our theory are on average more profitable than misaligned projects and that firm capability impacts organizational choice but not profitability.

Key words: knowledge workers; employment relation; performance; information technology

**********

The use of nonstandard employment arrangements such as independent contractors, job sharing, part-time workers, temporary workers, and other flexible staffing arrangements have been increasing over the past two decades (e.g., Kalleberg 2000). Of particular importance has been the trend toward the use of independent contractors to perform tasks previously assigned to employees (Lepak and Snell 1999). Independent contractors are frequently used in construction (Eccles 1981), entertainment (Baker and Faulkner 1991), defense, information technology (IT) services, and other industries in which activities are performed in discrete projects that involve forming a team for a task and disbanding the project team when the task is complete. Understanding the use of independent contractors is important because not only is their use rising in these industries, but also they are increasingly being used for technical and complex tasks (e.g., Slaughter and Ang 1996).

In this article, we examine the project (or task)-level factors that influence the use of independent contractors and the performance consequences of this choice, an issue that has received little theoretical or empirical attention. The choice of whether to use independent contractors or employees can be viewed as a governance decision. While transaction cost theory (Williamson 1985, 1991) has frequently been employed to analyze governance decisions regarding firm boundaries, specific investments in physical assets--the typical concern of transaction cost scholars--are largely absent in many high-technology and project-based industries. Thus, following Gulati and Singh (1998), we examine the impact of appropriability concerns and the degree of interdependence among workers and show that both types of contracting difficulties lead to a preference for employees over independent contractors. In addition, we draw on agency theory (Holmstrom 1979) to argue that increases in the cost of verifying project quality also lead to the choice of employees over independent contractors. Thus, we predict that projects involving any of these three contracting difficulties are more likely to be organized internally, while those that are not are more likely to be assigned to independent contractors.

While researchers have examined the determinants of variation in governance theoretically (e.g., Williamson 1985, 1991) and empirically (e.g., Anderson 1985, Gulati and Singh 1998) in a variety of contexts, much less attention has been paid to the performance implications of governance choice (for exceptions, see Masten et al. 1991, Walker and Poppo 1991, Silverman et al. 1997, Poppo and Zenger 2002). Hence, we develop a theory that predicts how governance choice for a project impacts its profitability. We argue that projects organized as we predict realize performance benefits relative to projects that are not.

We use a unique data set from the IT services industry to empirically examine the choice and performance implications of using employees versus contractors at the project level. Our data consist of 190 projects of a large IT services firm (hereafter fictitiously known as Compustar) in Silicon Valley and include a variety of transaction attributes, measures of Compustar's relative capabilities, choice of employment relation, and profitability for each project. This field survey is conducted within a particular organizational context, which allows us to control for a number of firm-level variables and capture detailed data on individual projects; it does raise concerns of generalizability that we address in the discussion.