Social Capital, Management Capacity and Public Service Performance

Abstract

Many scholars and policy-makers contend that social capital and management capacity are associated with better public services. It is also likely that organizations with the capacity to manage effective co-production are better able to realize these benefits. To test these assumptions, we explore the independent and combined effects of social capital and management capacity on the performance of major public services in the US states using Robert Putnam's index of social capital and the Government Performance Project's index of state management capacity.1 We find that social capital is associated with higher performing public services, and that strong management capacity enhances its positive effects.

Key words:

• Social capital
• management capacity
• state government
• public service
• performance
• United States

Notes

The social capital index comprises indicators covering five broad dimensions of civic culture: community organizational life, engagement with public affairs, community volunteerism, informal sociability and social trust. The management capacity index is composed of assessments of five management systems: financial management, capital management, human resources, managing for results and IT. These systems are assessed quantitatively and qualitatively, and scores are tracked over time.

Our indicator of state corrections performance (parole violators as a percentage of prison admissions) is taken in 1999 as comprehensive data on this item are not publicly available at the state level between 2001 and 2003.

These service indices exhibited strong discriminant validity. Principal components analysis using varimax rotation revealed that the variables composing the indices for each public service loaded onto six distinct factors corresponding to each service.

An alternative version of this overall score weighting each service index by the US average state budget composition was also constructed, but the statistical results from the ensuing analysis were essentially the same.

Population and total general expenditure displayed non-normal distributions (skew test results of 2.25 and 1.95), so logged values of these variables were used in the statistical analysis.

We are grateful to an anonymous reviewer for this important insight.