Uncovering the effect of local government debt brakes in Germany using synthetic controls

ABSTRACT

Strategies for limiting public debt remain a constant issue in public policy and are set to become more salient considering the current high levels of public spending in response to the COVID-19 crisis. One noteworthy strategy has been the introduction of institutional debt brakes. Although public debt at the local level does not necessarily follow the same rules and trends as debt at the national level, debt brakes at the local level are not nearly as comprehensively covered by both public and scholarly debate. The adoption of voluntary debt brakes in several German municipalities in the last decade offers intra-state variance to assess their fiscal effectiveness. A generalised synthetic control analysis is applied to mitigate for the special challenges of causal attribution and rare treatment cases. This approach provides a clear result: on average debt brakes in their present form are unable to limit debt levels in the examined municipalities.

KEYWORDS:

• Local public finance
• debt brake
• debt limit
• local government debt
• synthetic control method

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. For further technical details see Burth (2015).

2. (Brooks, Halberstam, and Phillips 2016, 316) also failed to identify a single study about self-imposed or voluntary limits on debt.

3. This does not mean that there are no other debt relevant factors that cannot be influenced by local actors (e.g., economic pressure, fiscal transfers), however, these factors cannot be addressed by a local debt rule either.

4. Unfortunately, the remaining ten municipalities are part of states that do not provide comparable statistical information about their municipalities. So, a comparison across states is not feasible.

5. Unfortunately, the accounting rules in North Rhine-Westphalia have changed in recent years however, the impact of this remains minimal. Since this analysis concentrates solely on one state all units are affected in the same way. Additionally, the calculation basis for the short-term debt has not changed at all, while the change for the long-term debts has been marginal.

6. The data for all variables is publicly available on the website of the North Rhine-Westphalian Statistical Office (https://www.landesdatenbank.nrw.de/ldbnrw/online/).

7. Since the municipalities did not adopt their debt brakes in exactly the same year, i.e., the treatment does not begin at the same time, the figures are not structured along calendrical years but in relation to the years before and after treatment.

8. The results for the long-term debt are identical for all four estimated models (with/without predictors, restricted/unrestricted sample). Due to limited space the following results are based on an unrestricted sample without predictors.

9. The appendix contains the results of an alternative estimation strategy which is based on a Bayesian multilevel identification approach for synthetic controls (Pang, Liu, and Xu 2021). The results in Table A1 and A2 confirm that the debt brakes neither have an average significant effect on short-term nor on long-term debt.

Additional information

Notes on contributors

Steffen Zabler

Steffen Zabler is a postdoctoral research associate in the Department of Political Science at the German University of Administrative Sciences in Speyer, Germany. His research interests include public finance, local government, and administrative reforms, as well as methodology and statistics.