Politics, management, and the allocation of arts funding: evidence from public support for the arts in the UK

Abstract

Studies of distributive public policy claim that electoral incentives shape the geographic distribution of government grants to individuals and organizations, such as those in arts and culture. Public management scholarship suggests that managers bring value to their communities and stakeholders within them through their capacity and skill. This study combines these literatures in a quantitative study of the geographic distribution of Grants for the Arts (GFA) in the UK between 2003 and 2006. Employing statistical regression techniques for count data, we find that GFA program in this period had a nonignorable distributive political character. Local authorities with swing voters for the governing party in Westminster received more GFA grants than did local authorities with its core supporters. We also find significant evidence that, at the same time, well-managed local authorities, as measured by performance assessment ratings, act as a magnet for GFA grants. Our conceptual discussion, quantitative modeling strategy, and results blend distributive politics and public management in a novel way for the study of cultural policy.

Keywords:

• arts funding
• public management
• distributive politics

Acknowledgments

We wish to thank Whitney Afonso for assistance in the data collection and staff at the Arts Council England for their insights into the Grants for the Arts program. Funding for this project was provided by the Undergraduate Research Associates Program at the University of Southern California. Mistakes remain our own.

Notes

1. We also considered that given its regional divisions, ACE would only be concerned with the geography of grant distribution at the regional level, rather than the local authority level. To test for this, we estimated the preferred specifications of our model in Table 2 with fixed effects for region. There is no change in the effects and Wald tests cannot reject the null hypothesis that all fixed effects are zero in the marginal indicator specification that corresponds to model 1 in Table 2 (χ² = 11.69, p = 0.17) or that using an extent of marginality measure as in model 2 in Table 2 (χ² = 7.65, p = 0.46). We thus find no support for regional influences, ceteris paribus.

2. This address information was retrieved from the University of Manchester’s GeoConvert, an online geocoding system available at http://geoconvert.mimas.ac.uk/. Weighting is performed multiplicatively for continuous variables, while dichotomous indicator variables take the values that prevail in the dyad at the time of measurement. That is, if local authority A has a population of 200,000 and 20% of the census addresses of that local authority are located in parliamentary constituency B, then dyad AB registers a population of 200,000 × 0.20 = 40,000. If local authority A has a council controlled by the Labor party, dyad AB shows a value of 1 for the Labor Local Council variable.

3. A total of 297 dyads are observed in 2003, 497 in 2004, 545 in 2005, and 246 in 2006 for a total of 1585 in the estimation sample. These observations fluctuate because, as Table 1 makes plain, a dyad is only included if it receives a GFA allocation in the year indicated.

4. Unreported specifications with clustering on the parliamentary constituency in the dyad provide slightly less conservative results with identical patterns of statistical significance.