Teacher's PAT? Multiple‐role principal–agent theory, education politics, and bureaucrat power

Abstract

This article aims to contribute to current debates about political power and agency relationships in education and other public sectors. In a recent clarion call for a major redirection of political principal–agent theories (PAT), Terry Moe has argued that standard information asymmetries ought no longer to be regarded as the sole foundation of bureaucrat power. According to Moe, current theories largely overlook the direct electoral power of agents and their unions (EPA) in voting for their own bureaucratic principals. Therefore, they are biased systematically towards underestimating agent power. We critically address both Moe's theoretical arguments, and his empirical applications to Californian school board elections. We conclude that Moe overestimates the power consequences of EPA on both counts. We outline a more balanced version of ‘multiple‐role’ PAT and of its potential implications for our understanding of the political power of public school teachers and bureaucrats more generally.

Keywords:

• bureaucratic agency
• Californian education
• democratic governance
• political power
• rational choice theory
• school board elections
• teacher unions

Acknowledgements

For helpful comments on earlier versions of this article, we are grateful to Rotem Bresler‐Gonen, Keith Dowding, Andrea Monika Herrmann, Maoz Rozenthal, and seminar participants at Haifa University's Faculty of Education, and Division of Public Administration & Policy. Pieter Vanhuysse also thanks the Max Planck Institute for the Study of Societies in Cologne for providing an excellent research environment during the writing of this article.

Notes

1. On PAT, see, for example, Moe (1984, 1987). On American schooling, see Chubb and Moe (1988, 1990).

2. The main theoretical arguments are made in Moe (2005, 2006a). The applications to education politics are in Moe (2001, 2003a, 2003b, 2006a, 2006b, 2009).

3. More generally, a large economics literature on the effectiveness of education inputs has reported that, across a wide range of contexts, inputs favored by teachers (e.g. higher wages and smaller classrooms) are less often statistically significant in improving educational outcomes than are inputs lower on teachers' priority lists, such as more or better textbooks, classroom equipment and other physical facilities, writing materials and other instructional materials, libraries, and software (Hanushek, 2003).

4. For applications to public sector settings, see, for instance, Ferejohn (1986), Huber and Shipan (2002) and Miller (1992, 2005).

5. These assumptions position the theory close to the ‘thick’ end of the rational choice theory spectrum, which includes relatively comprehensive behavioral conceptions (Korobkin & Ulen, 2000, pp. 1060–1061).

6. On the problem of teaching to the test, see, for instance, Firestone, Monfils, and Schorr (2004).

7. See Moe (2005, pp. 228, 215), and the references therein to the extensive rational choice literature criticized by Moe.

8. On teacher unions, see Brimelow (2003), Lieberman (1997) and Moe (2009). More generally, in Italy, Britain, the USA and Japan, the most dramatic post‐war strikes that were organized, ostensibly, against large‐scale job losses were never primarily intended to protect the unions' rank‐and‐file, but rather to safeguard the union organization and its most senior members (Golden, 1992). Across a wide range of political–institutional settings, union leaderships in post‐communist Europe have promoted their own interests often at the cost of those of their members (Vanhuysse, 2007). In the US, in a seeming application of Michels' iron law of oligarchy, working‐class gains achieved through disruptive protests have often been diluted subsequently as a result of formal hierarchical organization (Piven & Cloward, 1977; also see Levi, 2003, p. 51).

9. On theories of social capital and schooling, see, furthermore, Coleman (1988, 1990), Pong (1997), Pong, Dronkers, and Hampden‐Thompson (2003), and Vanhuysse and Sabbagh (2004).

10. Yet another source of selection bias is reported by Moe (2006a, p. 10), namely that his sample includes only districts where unions were active enough to support candidates.

11. On incumbency effects, see Gelman and King (1990). On interaction models, see specifically Brambor, Clark, and Golder (2006, pp. 71–72). Moe (2006a) misinterprets his own data specifically on pages 12–13.

12. Moe (2006a, p. 12, fn. 6) himself acknowledges so much in a footnote. Furthermore, it should be noted that Moe estimates that the actual rate of cases in which the unions do not support the incumbent is ‘probably lower’ than 46%, since this figure pertains to contested elections only. Given the potentially important implications of this finding to Moe's argument, it would be helpful to see reported the proportion of contested elections as a share of all elections. Lastly, it would have been more conclusive methodologically if Moe could demonstrate that union support is associated with third terms of a candidate. After one term, knowledgeable voters can correct their initial adverse selection. But if they chose to support the incumbent, will they continue to support the next time?