Does Power Corrupt? An Empirical Study of Power, Accountability, and Performance Triangle in Public Administration

ABSTRACT

Foundational work in public administration has considered the relationship between administrative power, accountability, and performance in public organizations. Even with the vast literature addressing power, accountability, and performance, scholars are still theorizing on how they influence one another. This study proposes and empirically tests a theoretical model of the relationships between each of these three constructs. Data are analyzed from an original survey of a national sample of US local government public administrators—city managers. Using structural equation modeling, results show that there are positive relationships between power and accountability, power and performance, and accountability and performance. The article discusses theoretical and practical implications of these findings.

KEYWORDS:

• Power
• accountability
• performance

Notes

1. The term “administrative power” is used because the data is collected from chief appointed administrative officers. Studies on power in related fields make similar choices such as CEO power, consumer power, executive power, or political power when empirical studies are conducted in particular contexts and on particular issues. Administrative power as a term better reflects the focus of the present empirical research.

2. This likely applies even to possible exceptions as school districts, recently subjected to rigorous demands to achieve certain test score benchmarks, considering the ongoing disagreement over the nature of the product (academic excellence), and the validity and the reliability of the prioritized measures, and the fact that no single indicator “can capture all that schools are trying to accomplish,” as acknowledged even by school privatization ideologues (e.g., Chubb & Moe, 1990).

3. The more ambiguous the goals are, the poorer “organizational performance” generally is (Chun & Rainey, 2005). Alternatively, if conflicting goals or outcomes are taken for granted, as the normal state of organizations, then organizations by definition cannot be effective, as it is inherently impossible to maximize conflicting goals (Steers, 1975), and in principle there is no possibility for stable consensus about the nature of the goals (Yuchtman & Seashore, 1967).

4. In SEM, this practice is known as parceling. A parcel is an aggregate-level indicator comprising the sum of two or more items, responses, or behaviors. For a thorough discussion of this practice, see Little, Cunnigham, Shahar, and Widaman (2002). When the purpose of test is to explore the relationships between constructs rather than the relationship of each manifest variable to the construct, this method is particularly recommended. Composite variables were decided based on theoretical reasons. In order to ensure factorial validity, however, a confirmatory factor analysis was conducted with all the individual variables included to see if the individual variables load on their respective factors. Convergent and discriminant validity were achieved. Fit indices confirmed factorial validity of the performance and accountability latent constructs.

5. Although the survey does not have a sufficient number of variables to capture the degree of professionalism among city managers, there are some variables and diagnostic findings that can help model building efforts in the future. In the survey, city managers were asked the extent of their agreement to various statements on their consideration of the preferences of elected officials, citizens, the reactions of the groups affected, and knowledge from the profession when responding to routine matters during policy implementation. The mean scores, on a 5-point Likert scale, are 3.29 for elected officials, 3.32 for groups affected, 3.50 for citizens, and 3.87 for knowledge from the profession. This limited finding shows the importance of professional considerations for city managers in making decisions. A larger research project, however, with a more comprehensive set of variables measuring professionalism and its impact on accountability (and performance) is needed to fully assess the role of professionalism as moderating variable.