Random walks and organizational mortality

Administrative Science Quarterly,  Sept, 1991  by Daniel A. Levinthal

• E-mail
• Print
• Link

A simple model is developed here that relates organizational age and mortality rates. The critical insight of the analysis is that surviving organizations will tend to be organizations that were, in prior periods, successful, and this prior success, in turn, will buffer them from subsequent selection pressures. The model is then tested using data on the Argentinian and Irish newspaper industries. Further empirical analysis examines the degree of heterogeneity in these two populations of organizations, and a subsequent parametric analysis provides some insight as to the sources of this heterogeneity. (*)

The stream of research on ocological perspectives of organizational change has generated tremendous excitement in the community of organizational theory researchers (for reviews, see Carroll, 1984; Singh and Lumsden, 1990). Several core issues have been explored. Among them are the relationship between organizational age and organizationl mortality (Carroll and Delacroix, 1982; Freeman, Caroll, and Hannan, 1983; Carroll, 1983; Singh, Tucker, and House, 1986), the role of population density on selection and birth rates (Britain and Freeman, 1980; Hannan and Freeman, 1988), and the nature of competitive interaction across organizational forms (Barnett and Carroll, 1987; Brittain and Wholey, 1988). The question of age dependence in the rate of organizational mortality is the feature of organizational ecology that has been most extensively examined empirically. The basic empirical regularity that emerges is that the risk of mortality tends to decline with organizational age. However, despite its prominence in ecological analysis, there has been little explicit modeling of time-varying organization-level processes which, in turn, lead to predictions of population-level rates of mortality.

Why do older organizations have a lower rate of failure? This is the basic question posed by the empirical analysis of age dependence in the pattern of organizational mortality. One obvious explanation for the relationship between organizational age and organizational mortality is that organizations tend to become more effective with age. This process is conventionally though of in the context of a learning curve and a firm's production costs, but it is also consistent with richer and more general characterizations of the accumulation of skills and knowledge over time (Nelson and Winter, 1982). Greater competence over time should lead to a lower risk of mortality. Hannan and Freeman (1984, 1989) proposed a more subtle effect of learning, suggesting that the variation in performance, as distinct from its expectation, declines with time. They argued that selection processes operate within a population so as to eliminate those organizations with low reliability or accountability. Reliability is, in turn, enhanced by the development of highly standardized routines, which form the basis of continuity in an organization's behavior over time (Nelson and Winter, 1982). Thus, Hannan and Freeman suggested that organizational mortality declines with age due to the increasing reliability of organizational behavior over time. Stinchcombe's (1965) argument for the liability of new organizations is based on the instability of new social relations, social relations both within a new organization and between the new organization and external actors. This hypothesis has been interpreted to imply that the likelihood of organizational failure declines monotonically with time.

These explanations of the temporal pattern of organizational mortality are arguments concerning organization-level processes. However, the link between these organization-level processes and population-level rates of mortality have not been explicitly developed. This paper develops a simple model of organizational performance and explores the implications of such a process for the pattern of organizational mortality and, in turn, aggregate rates of mortality. In developing such a model, one is confronted by the question of what factors account for an organization's performance and how performance is linked to organizational survival. If one focuses attention on business organizations, the latter question can, at a general level, be readily answered. A business fails to survive when it can no longer meet its financial obligations to debt holders, employees, or suppliers and resorts to or is forced into bankruptcy or liquidation. Such a failure is typically viewed as a liquidity crisis: the firm is not generating sufficient cash flow, nor has it sufficient readily saleable assets to meet its immediate cash demands.

Conversely, this characterization of failure suggests that firms are buffered from failure either by strong current performance or a large stock of assets. However, organizations tend not to accumulate large amounts of financial assets. Such "slack" resources (Cyert and March, 1963) tend to be dissipated through a variety of means--dividends to shareholders, diversification and acquisition activity, increased salaries, or underutilization of management and labor. Nonetheless, many firms, as a result of past success, appear to be greatly buffered from immediate selection pressures. Firms such as Xerox may endure a number of periods of poor financial performance and yet face no threat to their survival. In part, this buffering is the result of their extensive financial assets; but, in addition, it reflects the strength of their market position associated with their extensive distribution system, vast manufacturing infrastructure, and technological capabilities. These various kinds of assets, both financial and nonfinancial are labeled here organizational capital. (1)