Electoral Cycles And Local Government Debt Management

Abstract

The literature provides both theories and empirical assessments that link national electoral cycles and opportunistic incumbents' behaviour. However, at the subnational level the literature is scarce. Using a panel of 238 Spanish municipalities over the period 1992–2005, this paper investigates for the first time in Spain whether electoral events contribute to shape municipal debt policies. We show that the electoral cycle influences the municipal debt per capita. Furthermore, both weak (no-majority) and wealthier municipal governments have higher levels of debt per capita. Finally, our data show that the 2001 Spanish Budgetary Stability Law (stemming from the European Stability and Growth Pact) appears to have reduced the electoral effect on municipal debt per capita.

Key Words:

• Electoral cycle
• local government
• debt

Acknowledgements

This study has benefited from the financial support of the Spanish National R&D Plan through the research projects ECO2010-17463 and ECO2010-20522 (Ministry of Science and Innovation).

Notes

1. In Spain, the Constitution (art. 131) confers on the Central Government exclusive competence in income redistribution.

2. As a clear example of this convergence, the Spanish socialist National Government has eliminated in 2009 the equity tax, which is a ‘classical’ income-redistribution tax.

3. Blais and Nadeau (1992), Galli and Rossi (2002) and Binet and Pentecote (2004) apply the same specification, i.e., 1-year lagged dependent variable as regressor.

4. We have turned current euros into 1992 real euros in order to control for the effects of inflation (Cropf and Wendel, 1998).

5. We use the per capita debt to control for significant differences in size among the cities studied (Clingermayer and Wood, 1995; Cropf and Wendel, 1998).

6. Regarding the existence of unobserved heterogeneity in debt preferences, Clingermayer and Wood (1995) show that there are specific political-cultural factors that affect the propensity to debt-finance.

7. For more details, the interested reader may consult Wooldridge (2002) chapter 11, Arellano (2003) chapter 6 and Baltagi (2001) Chapter 8, among others.

8. A constant term has also been introduced in the model.

9. According to Pettersson-Lidbom (2001), LG economic level measures the local business cycle variations. Thus, this variable is a proxy that controls, among other factors, for the economic cycle.

10. Franzese (2000) builds indicators aimed at capturing the exact period in which the election is held. This index is useful when the date of election varies substantially from one election call to another. However, since Spanish municipal elections have all taken place in the same period (28 May 1995, 13 June 1999 and 25 May 2003) a binary 0–- 1 variable is accurate.

11. See note 10.

12. See note 10.

13. We have run estimations in which the explanatory cost variable was expressed in level instead of first differences (that is with RCOST alone, as well as with RCOST and RCOST₋₁ instead of ΔRCOST). However, the estimated model was either worse (especially with respect to the specification tests) or indicated that ΔRCOST was statistically better than the separate use of RCOST and RCOST₋₁.