Structural embeddedness and the liability of newness among nonprofit organizations

Abstract

Ecological studies have consistently reported that younger organizations are more likely to close or disband than older organizations. This article uses neo-institutional theory and social capital theory to explore this finding. We derive hypotheses from these perspectives and test them on a panel of nonprofit organizations in Minneapolis-St Paul (USA) using event history analysis. We find that larger organizations and organizations more dependent upon private donations are less likely to close, and government funding reduces the age effect on mortality; that is, older and younger publicly funded organizations are equally likely to survive or fail. However, among older organizations, not having government funding increases chances of survival. In contrast, volunteer staffing accentuates the age effect. Older organizations that were more dependent on volunteers had a lower likelihood of closure than younger organizations dependent on volunteers, while age had no effect on closure for organizations not dependent on volunteers. We conclude by examining our findings in light of the extant thinking on the liability of newness and the role of institutional and network embeddedness on the chances of organizational survival.

Keywords:

• Institutional theory
• nonprofit organizations
• organizational ecology
• structural embeddedness

ACKNOWLEDGEMENT

The National Science Foundation (USA), the Northwest Area Foundation (USA) and the Nonprofit Sector Research Fund (USA) provided funding for this research. Send inquiries to the first author at 2100 M Street, NW, Washington DC, 20037, USA; or [email protected]

Notes

More recently, German scholars (Udo 1989; Brüderl and Schüssler 1990; Brüderl et al. 1992; Jungbauer-Gans 1994) have shown that organizations are not necessarily at an immediate risk of death after they start out. Rather, they are bolstered by initial stocks of resources and reserved judgment from supporters and decision makers that allow for a ‘honeymoon period’. Once this stock of resources is depleted, however, risk of demise increases. This thesis is termed the ‘liability of adolescence’ argument.

Hannan and Freeman (1984) argue further that change in established organizations can disrupt internal processes and external relationships such that the changed organization now faces the same liabilities that a new organization might face. Amburgey et al. tested this thesis and found that ‘change reset the liability-of-newness clock and increased the risk of failure above what it would have been otherwise’ (1993: 68).

See Galaskiewicz and Bielefeld (1998) for a discussion of organizational types that were excluded from the sampling frame.

Data collection in each wave took several years to complete. Consequently, references to 1980 data collection should most properly refer to the range 1980 – 2. Nonetheless we resort to a shorthand that corresponds with the year in which most questions in the interview schedule refer, which we refer to as a ‘primary year of interest’. That is, whether data were collected in 1980, 1981 or 1982, interview questions centered on events in 1980. Similarly, references to data collected in 1984 were collected in 1984 – 5; 1988 data were collected in 1988 – 9; 1992 data were collected in 1994 – 5; and 1994 data were collected in 1996.

The data were gathered by project research assistants who were assigned the role of tracking down the organizations for a given year. They ascertained the fate of the organization based on information provided by former personnel of the organization.

Bielefeld (1994) provided initial descriptive data on organizations that had exited the panel between 1980 and 1988, with special attention to the industries to which these organizations belonged. He noted high levels of exit in the educational and housing industries.

Often organizations received a proportion of their income from several sources and/or cited more than one activity as ‘primary’. In these cases we allocated a fraction of that organization to a given cell depending upon the proportion of its funding from that source and its investment in a given activity area.

Medicaid and Medicare payments were not coded as government funding, but rather as program service revenue.

We included dues among donated revenues because some nonprofits in the sample labeled individual gifts as dues and informed us that this is the way they should be treated in our analysis.

We conducted a preliminary analysis with age, expenditures, density, density squared and the dummy for the missing expenditure items. The squared term was significant at the .09-level. Since density is centered and the squared term is marginally significant, this suggests that the odds of disbanding were less in very sparse niches and very dense niches. This is the opposite of what we hypothesized.