Political cycles in Greece’s municipal employment

Abstract

We consider the politically motivated fluctuations in Greece’s municipal employment, constructing a data-set from primary data and focusing on the composition of municipal employment in terms of employment relationship forms. Our analysis produces strong evidence of pre-electoral manipulation through increases in the number of contract employees. Considering a number of control variables and robustness checks does not affect the key results. Such variables include whether mayors run for reelection, incumbents’ political alignment with central government, partisan shifts, general elections, mayors’ turnover rate, and timing patterns. Our evidence provides insights into Greece’s political economy in the run-up to the current economic crisis.

Keywords:

• political cycles
• employment
• municipalities
• Greece
• panel data

JEL Classification:

• D72
• H7
• C23

Notes

1. Presidential Decree 64/2004.

2. Coelho, Veiga, and Veiga (2006) consider employment across different activities (e.g. construction, transportation, social services, electricity) but not the permanent or temporary nature of employment contracts.

3. Sections 4.2 and 4.3 discuss in more detail the interaction terms used in our analysis.

4. See notes in Table 1.

5. For the whole time dimension of our panel Greece was ruled by single party governments.

6. A Wald test does not reject the hypothesis that the estimated coefficients on ELE*PolAlignment and ELE*NotPolAlignment are equal. This result is similar for the coefficients obtained when Total Employees and Permanent Employees are used as the dependent variables. When we include the variable PolAlignment as a separate control in the model the estimated coefficient is not significant, while all other coefficients remain the same as in Columns (5) to (8).

7. The variable Swing corresponds to partisan shifts and not in mayor changes as a right wing mayor can be succeeded by another one with the same affiliation in the case where the former mayor in no longer a candidate.

8. A Wald test does not reject the hypothesis that the estimated coefficients are equal.