Soft budget constraints and strategic interactions in subnational borrowing: Evidence from the German States, 1975–2005

https://doi.org/10.1016/j.jue.2011.07.003Get rights and content

Abstract

Cooperative federations are usually characterized by the existence of bailout guarantees and intergovernmental transfer schemes. This paper explores whether such features of cooperative federations lead to subnational soft budget constraints using panel data from the German States covering the 1975–2005 period. The methodology is based on the premise that subnational governments’ borrowing will exhibit vertical and horizontal strategic interactions if they operate under soft budget constraints. Therefore, a test for strategic interactions in subnational borrowing can be used to infer whether a cooperative federation like Germany is susceptible to soft budget constraints. The results suggest that state borrowing in Germany exhibited horizontal but not vertical interactions during the time-frame of the analysis. This indicates (i) that German States faced soft budget constraints and (ii) that they were more concerned about the likelihood of a bailout than about its volume.

Introduction

The fiscal federalism literature distinguishes two types of federations. Countries where subnational governments possess substantial tax autonomy, where revenues are not shared, and little fiscal equalization takes place are called competitive federations. In cooperative federations, on the other hand, subnational tax autonomy tends to be low, revenues are usually shared, considerable fiscal equalization takes place, and im- or explicit bailout guarantees are often given (Schaltegger and Feld, 2009).

These two types of federations lead to different incentives for policy makers. Consequently, they will have different implications for macroeconomic and fiscal outcomes (Weingast, 2009). One criticism leveled against cooperative systems of federalism in recent years is that they are much more susceptible to subnational over-borrowing. This assertion is primarily founded on case studies which show that cooperative federations such as Argentina (Dillinger and Webb, 1999), Brazil (Samuels, 2003), and Germany (Seitz, 1999) have experienced considerably more problems with subnational debt than competitive federations like Switzerland (Feld and Kirchgässner, 2007) and the United States (Inman, 2003).

These case studies suggest that lower-level jurisdictions in cooperative federations have an incentive to over-borrow because of soft budget constraints. In federal fiscal relations, soft budget constraints can emerge if subnational governments expect that some of the costs of their borrowing will eventually be paid for by the federal government. It has been argued in the literature that federal governments in cooperative federations are more likely to bail out an indebted jurisdiction than those in competitive federations (Blankart and Klaiber, 2006). Indeed, it is a reasonable conjecture that once schools and police departments have to be closed due to subnational over-borrowing, calls for a federal intervention will be more pronounced in cooperative than in competitive federations because of differing expectations regarding the responsibilities of the federation. In particular, the existence of an extensive fiscal equalization scheme signals to all states that the federation has the institutional means and the political will to bail out an indebted jurisdiction.1

One methodological approach put forward in the literature to test whether the soft budget constraints problem and thus subnational over-borrowing is more prevalent in cooperative federations than in competitive ones is to simply regress a measure for the indebtedness of the public sector on a measure for the importance of contemporaneous intergovernmental transfers. The underlying idea behind this approach is that the importance of intergovernmental transfers is a reasonable proxy for the competitiveness of a federation. In general, studies using this approach, e.g. De Mello, 2000, Rodden, 2002, Baskaran, 2010, find that cooperative federations do not systematically borrow more than competitive ones.

However, one problem with the methodological approach advanced in these studies is that they do not test the soft budget constraints argument properly. According to the theoretical literature on soft budget constraints, the reason for their emergence are expected future transfers that are paid to alleviate the costs associated with a large debt burden, not contemporaneous transfers. Therefore, contemporaneous transfers are at best only an imperfect proxy for bailout expectations, and at worst an incorrect one.

Due to this methodological problem, a second strand of the empirical literature focuses on individual countries and explores whether specific institutional features within a country lead to soft budget constraints.2 In this paper, I follow this strand of the literature and explore whether specific features of the fiscal constitution prevailing in the Federal Republic of Germany have led to subnational soft budget constraints during the 1975–2005 period.

The methodology used in this paper is based on the premise that subnational borrowing in cooperative federations should be characterized by vertical and horizontal strategic interactions if subnational governments face soft budget constraints. In other words, if a state expects that it will receive a bailout once it faces a severe fiscal crisis, then its contemporaneous borrowing should react systematically to the borrowing of the federal government (vertical interactions) and/or to the borrowing of other member states of the federation (horizontal interactions).

Why should bailout guarantees lead to vertical strategic interactions in subnational borrowing? If the federal government cannot commit to a no-bailout policy, every subnational government has an incentive to increase borrowing. The magnitude of this incentive, however, will depend on the expected amount of resources the federal government will have at its disposal when a bailout becomes necessary. If there is reason to suspect that the federal government will have fewer resources in the future because of its own borrowing, subnational incentives to over-borrow will be, ceteris paribus, less pronounced, and vice versa.

Horizontal strategic interactions may emerge for two reasons. First, higher transfers to other subnational jurisdictions imply lower transfers for one’s own jurisdiction as long as federal resources are finite. This means that if other jurisdictions over-borrow, the available resources for a bailout for one’s own jurisdiction will be reduced. Second, the decision of whether or not to grant a bailout to a given subnational jurisdiction will be made not only based on its own debt burden but also based on that in other subnational jurisdictions. Usually, a jurisdiction has to prove that it faces extraordinary fiscal difficulties before it will receive a bailout. If all jurisdictions exhibit a high debt burden, it is difficult for any individual jurisdiction to argue that it is extraordinarily needy. Hence, each subnational government has to take the borrowing in other jurisdictions into account when optimally determining its own borrowing policy.3

For these reasons, it can be expected that subnational borrowing in cooperative federations will be characterized by vertical and horizontal interactions if subnational governments operate under soft budget constraints. Anecdotal evidence suggests that this line of reasoning applies to Germany, an archetypical cooperative federation. Decisions regarding whether or not a state should receive a federal bailout have been made in Germany by the federal constitutional court. Relying on a number of constitutional stipulations (in particular, Art. 72, Art. 106, and Art. 107) which demand that equal living conditions have to be guaranteed throughout the federation, the court granted two states, Saarland and Bremen, a bailout in 1992.4 In determining whether the debt levels in Bremen and Saarland necessitate a federal bailout, the constitutional court compared the debt to GDP ratio and the fiscal burden caused by the interest payments in these two states with that in all other German states.

This ruling by the constitutional court hence made explicit what subnational policy makers in Germany have in all likelihood already known implicitly: that the probability of a state receiving a bailout depends not only on its own borrowing, but also on the borrowing of all other states. The ruling also underscored that the likelihood of another state receiving a bailout depends on any given state’s borrowing insofar as it affects the average debt burden of the subnational tier. Moreover, it implicitly affirmed that the fiscal situation of the federal government will affect decisions regarding subnational bailouts. It is consequently a reasonable conjecture that the borrowing polices of the constituent members of the German federation, i.e. the states and the federal government, are interdependent and will exhibit strategic interactions if some or all states operate under soft budget constraints.

In this paper, therefore, I empirically study whether the borrowing in a given German State reacts to the borrowing of other member states of the federation and/or to the borrowing of the federal government in order to establish whether subnational governments in Germany operate under soft budget constraints. More specifically, I estimate linear models where the dependent variable is the deficit to GDP ratio of state i in year t and the main independent variables are (i) the weighted average5 deficit to GDP ratio of the “other” states in the federation and (ii) the deficit to GDP ratio of the federal government.

Since OLS does not produce consistent estimates in models with such fiscal policy interactions, I rely on the instrumental variable approach to identify the effect of the other states’ weighted average deficit to GDP ratio and the federal government’s deficit to GDP ratio on the deficit to GDP ratio of a given state i. Unfortunately, I am not aware of any natural-experiment type mechanism that induces truly exogenous variation in the endogenous variables. Therefore, I use a set of instruments for which I argue below that they are likely to be robust to certain sources of endogeneity. However, since I cannot rule out all possible sources of endogeneity ex-ante, I also present a set of robustness checks in order to explore to what extent the results are driven by each of the instruments.

I use the other states’ lagged weighted average deficit to GDP ratio and their weighted average contemporaneous population growth as instruments for their weighted average deficit to GDP ratio. As instruments for the deficit to GDP ratio of the federal government, I use a dummy variable indicating federal election years and a dummy variable indicating the ideology of the federal government.

The lagged and contemporaneous deficits of other states will be related if there is some persistence in deficits. The reason to expect a strong relationship between the other states’ population growth and their deficits is that population size is the most important determinant of the amount of equalization transfers a state receives during horizontal and vertical equalization in Germany. An increase in population size will therefore lead to higher revenues and consequently to smaller deficits in any given state. A relationship between federal deficits and federal election years can emerge if the federal government adjusts deficits in view of federal elections. Similarly, it is also a reasonable conjecture that there are ideological differences in the borrowing policies of the federal government.6

In addition to being strongly related to the endogenous variables, the instruments should also be uncorrelated with the error term in the second stage regressions in order to produce reliable estimates. One reason to suspect such a correlation in the current context is reverse causality between the deficit of a state i and the population growth in other states. Such reverse causality may emerge if inhabitants migrate in view of state deficits. However, the execution of a decision to migrate, even if it is based on deficits, will take time. Therefore, even though this instrument may not be strictly exogenous, it appears to be feasible to treat it as pre-determined. Nonetheless, I present in Section 4.3 a robustness check where the baseline models are re-estimated after omitting the other states’ weighted average population growth from the instrument set to explore to what extent the baseline results rely on the validity of this particular instrument.

This kind of reverse-causality is implausible for the lagged weighted average deficit to GDP ratio of other states since there cannot be an effect of the deficit of state i in period t on the deficits of other states in t  1. However, it is possible that this variable has a direct effect on the deficit of a state i if states react with some lag to each others’ borrowing. Since I cannot exclude this possibility ex-ante, I explore in Section 4.3 to what extent the baseline results change when this variable is explicitly included in the second stage regressions.

Reversed causality between state deficits and the federal election and ideology dummies is implausible as well because the deficit of any particular state i is unlikely to have an effect on the timing or on the outcome of federal elections. However, these variables may have a direct effect on state i’s deficits if some states receive fiscal benefits from a politically aligned federal government or if they adjust their fiscal policy in view of federal elections. Consequently, in line with the procedure for the other states’ lagged weighted average deficit to GDP ratio, I explore the robustness of the results to including these instruments in the second stage regressions in Section 4.3.

The remainder of this paper is structured as follows. In Section 2, I briefly discuss the federal system and the evolution of subnational debt in Germany. In Section 3, I derive the empirical specification for testing whether strategic interactions in subnational borrowing exist and describe the data. In Section 4, I report the results. They suggest the presence of positive horizontal strategic interactions. However, while this finding is consistent with the existence of soft budget constraints, subnational borrowing in Germany might exhibit horizontal strategic interactions for other reasons than soft budget constraints. In particular, if voters use the borrowing in other states as yardstick to evaluate the performance of their own state governments, every state government might have an incentive to react strategically to other states’ borrowing, a mechanism that is referred to as yardstick competition (Salmon, 1987). Thus, Section 5 explores whether the horizontal strategic interactions can be explained by yardstick competition or whether soft budget constraints are indeed the most likely explanation. Section 6 concludes with a discussion of the findings.

Section snippets

Fiscal federalism in Germany

The Federal Republic of Germany is founded on strong federalist principles. It consists of three tiers of government: federal (Bund), state (Länder), and local (Gemeinden). The localities, however, are with respect to fiscal matters subordinate to the states. The important protagonists for the purposes of this paper are therefore the federal and state governments.

Until the unification of West-Germany with the former GDR in October 1990, the German federation consisted of eleven states.

Empirical analysis

As argued in the introduction, a test for strategic interactions in borrowing can be used to infer whether a cooperative federation is characterized by soft budget constraints. In general, studies exploring fiscal interactions tend to focus on taxes (Devereux et al., 2008, Edmark and Agren, 2008) and expenditures (Case et al., 1993, Baicker, 2005); studies exploring fiscal interactions in borrowing policies are rarer. One exception is Landon and Smith (2000) who study for Canada how debt

Estimation and results

It is well known that estimating a model that includes a lagged dependent variable and fixed effects with the least squares dummy variable (LSDV) estimator leads to the Nickell-Bias because the within-transformed lagged dependent variable is correlated with the transformed error term (Nickell, 1981).

Soft budget constraints vs. yardstick competition

The estimates reported in the previous section indicate that the borrowing policy of German States displayed horizontal strategic interactions during the 1975–2005 period. Until now, it has been implicitly assumed that soft budget constraints are the cause for these interactions. This is indeed a likely explanation given the provisions in the German fiscal constitution. However, there is also an alternative explanation for the horizontal strategic interactions: that state governments engaged in

Conclusion

The goal of this paper was to explore whether the German States face soft budget constraints by testing for strategic interactions in state borrowing with panel data covering the 1975–2005 period. In the empirical analysis, I found evidence pointing toward the existence of horizontal strategic interactions, but no evidence for vertical strategic interactions. With regard to the horizontal interactions, tests indicated that they were not due to yardstick competition. Soft budget constraints were

Acknowledgments

I am grateful to two anonymous referees and the editor of the journal. Their comments and suggestions have improved this paper considerably.

References (36)

  • T. Baskaran

    On the link between fiscal decentralization and public debt in OECD countries

    Public Choice

    (2010)
  • T. Besley et al.

    Incumbent behavior: vote-seeking, tax setting, and yardstick competition

    American Economic Review

    (1995)
  • C.B. Blankart et al.

    Subnational government organization and public debt crises

    Economic Affairs

    (2006)
  • J. Brueckner

    Strategic interaction among governments: an overview of empirical studies

    International Regional Science Review

    (2003)
  • Crivelli, E., Staal, K., 2008. Size, spillovers and soft budget constraints. Max Planck Institute for Research on...
  • Dillinger, W.R., Webb, S.B., 1999. Fiscal Management in Federal Democracies: Argentina and Brazil. World Bank Policy...
  • Doi, T., Ihori, T., 2006. Soft budget constraints and local expenditures. CIRJE Discussion Paper No. F-442....
  • Federal Ministry of Finance, 2009. Federation/Länder financial relations on the basis of constitutional financial...
  • Cited by (46)

    • Decentralization, incentives, and local tax enforcement

      2020, Journal of Urban Economics
      Citation Excerpt :

      In particular, departing from full decentralization under which subnational governments possess autonomy to set both spending and taxes, many countries, especially developing countries, have pursued partial fiscal decentralization under which subnational tax autonomy tends to be low and local governments often have to rely on transfers from the central government to finance the provision of public goods.1 This variation in the setup of fiscal decentralization policy leads to different incentives for local governments and thus different implications for macroeconomic and fiscal outcomes (Baskaran, 2012; Weingast, 2009). One criticism leveled against a partial fiscal decentralization setting in recent years is that the lack of subnational tax autonomy and reliance on transfers have become sources of local fiscal indiscipline by encouraging local governments to over-spend/over-borrow and/or lowering tax effort (e.g., Borge and Rattsø, 2002; Eyraud and Lusinyan, 2013; Rodden et al., 2003).

    • Does fiscal oversight matter?

      2018, Journal of Urban Economics
      Citation Excerpt :

      Following Kornai (1986) , the term soft budget constraint has been coined for this problem. It has been extensively studied by Epple and Spatt (1986), von Hagen (1991), Qian and Roland (1998), Maskin (1999), Buettner and Wildasin (2006), Bordignon and Turati (2009), Pettersson-Lidbom (2010), and Baskaran (2012), among others, and is an important rationale for fiscal restraints, in particular for sub-national governments. As we discuss in detail in Section 8, our findings provide some support for this argument, given that the effects of a withdrawal of fiscal oversight are concentrated on municipalities that were effectively restricted by the oversight system.

    • On the effects of intergovernmental grants: a survey

      2024, International Tax and Public Finance
    View all citing articles on Scopus
    View full text