## 1. Features of Subnational Governments

## Source details

**Canonical URL:** [1. Features of Subnational Governments](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp1245.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp1245.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp1245.pdf.json)

---

### I. Introduction and research question
- Many European countries reassigned spending and revenue responsibilities from the center to subnational (local and regional) governments.
- Subnational spending in the European Union (EU) increased from 23 percent of general government spending in 1995 to 26 percent in 2009; the revenue share increased to a lesser extent.
- Economic rationales: better information and matching of policies to preferences (Oates 1972); jurisdictional competition (Brennan and Buchanan, 1980); increased accountability and transparency.
- Risks: “common pool” transfer financing can induce overspending and lower tax effort; bailout expectations can soften local budget constraints (Rodden et. al. 2003).
- Institutional arrangements—e.g., fiscal rules—could potentially correct incentives, enhance accountability, and anchor expectations.
- Purpose: assess empirically the impact of fiscal decentralization on fiscal performance in the EU, explicitly examining the role of institutional arrangements covering subnational governments.
- Summary highlighted finding:
  - Spending decentralization has been associated with stronger fiscal performance, especially when transfer dependency of subnational governments was low.
  - Subnational fiscal rules do not seem to play a role in ensuring better performance.

### II. Literature and theoretical background
- Empirical literature shows mixed results; cross-country econometric work is scarce and focuses mostly on OECD countries.
- Theoretical concerns: decentralization may complicate macro stabilization (Ter-Minassian, 1997a; Tanzi, 1995; Tanzi, 2000; IMF, 2009) and create coordination failures and “deficit bias” (Oates, 2006), especially with central bailouts.
- Empirical cross-country findings cited:
  - Rodden (2002): revenue decentralization deteriorates general government balance.
  - Neyapti (2010): revenue and spending decentralizations improve the balance.
  - Afonso and Hauptmeier (2009): higher spending decentralization worsens primary balance (for high debt levels); revenue decentralization does not matter.
  - Thornton (2009): no significant impact of revenue decentralization.
  - Baskaran (2010): expenditure decentralization reduces public indebtedness; tax decentralization insignificant.
- Three institutional features emphasized:
  - Transfer dependency: higher reliance on transfers may reduce overall balance; grants can be endogenous to deficits.
  - Subnational borrowing autonomy: unrestricted borrowing or “soft” financing can undermine discipline.
  - Fiscal rules: may offset coordination problems, but evidence is inconclusive; subnational rules often show no significant impact on general government fiscal balances.

### III. Institutional features and stylized institutional facts (EU)
- Large variation in subnational role: subnational spending share ranges from less than two per cent in Malta to almost two-thirds in Denmark.
- Federal vs unitary: only Austria, Belgium, and Germany are federal; some unitary countries (Denmark, Finland, Spain) are highly decentralized.
- More decentralized countries tend to have more tiers of subnational government.
- Main areas of subnational expenditure: education, health, and social welfare.
  - Between 1995 and 2008:
    - Subnational expenditure shares for education and social welfare rose.
    - Subnational expenditure share for health decreased in the majority of countries: eight countries reduced subnational health share; three—Estonia, Hungary, Romania—reduced it by 15-20 percentage points; Ireland’s subnational health share dropped from 95 percent to zero after 2005 reassignment to the center.
- Control mechanisms:
  - Fiscal rules (mainly borrowing or balanced budget rules) are common at subnational levels; number of fiscal rules increased at central and general government levels.
  - Majority of fiscal rules are applied at the local government level.
  - Budget balance rules more prevalent in EU15; debt/borrowing rules common in new member states (NMS); expenditure rules rare at subnational level.
  - Subnational fiscal rules are more prevalent in countries with higher decentralization and where subnational governments rely more on own revenues.
  - Fiscal rules indices derive from the European Commission Fiscal Rule Index database (European Commission, 2009); index measures legal basis, coverage, monitoring/enforcement strictness (including sanctions and escape clauses), and media visibility.
  - Strength of rules does not equal effectiveness: sanctions often weak; central governments retain discretion; bailouts occurred in at least nine EU countries; bailouts more frequent where more administrative tiers exist.
  - Less than one-third of countries have formal coordination arrangements; minority include subnational targets in medium-term budgetary frameworks.

### IV. Stylized empirical facts on decentralization and fiscal performance (1995–2009)
- Data: Eurostat fiscal data covering 1995-2008 (econometric panel extends to 1990-2008 in some analyses); subnational measures aggregate regional and local subsectors (S1312 and S1313 in ESA95).
- Definitions:
  - Spending decentralization: Subnational spending (excluding transfers paid) in percent of general government spending.
  - Revenue decentralization: Subnational revenue (excluding transfers received) in percent of general government revenue.
  - Transfer dependency: Subnational net transfers received (current and capital) in percent of subnational revenue.
- Stylized fact 1: Spending decentralization associated with better fiscal performance at general government level.
  - Over 1995–2008, cyclically adjusted general government fiscal balances were higher among more decentralized countries (e.g., Denmark, Sweden, Spain) and lower in less decentralized countries (e.g., Greece, Malta, Slovakia).
  - Increases in spending decentralization are not associated with increases in debt.
  - On average, overall fiscal balances in countries with medium and low decentralization are respectively 2 and 2½ percentage points of GDP below those of countries with high decentralization.
  - Subnational governments show close-to-balance fiscal positions irrespective of decentralization level; central government positions are relatively stronger in highly decentralized countries.
- Stylized fact 2: Expenditure decentralization outpaced decentralization of resources.
  - Between 1995 and 2009:
    - Subnational spending rose by 3¾ percentage points as a share of general government spending (Average = 3.79).
    - Increase in subnational own revenues averaged 1.09 percentage points.
    - Change in net transfers averaged 1.4 percentage points.
  - Vertical imbalances increased over time, suggesting possible unfunded mandates and rationing of resources to subnational governments; rationing could have forced expenditure savings at subnational level—especially if borrowing constrained—and contributed to improved central and general government fiscal positions.
- Stylized fact 3: Subnational fiscal rules do not appear to have an effect on fiscal performance.
  - Overall fiscal rule index shows a positive relationship with general government balance, but the subnational fiscal rule index does not show a clear relationship.
  - Possible interpretations: weak implementation of rules; bailouts; rules introduced where fiscal performance was weak.

### V. Econometric approach and key results
- Empirical approach: estimate a fiscal reaction function following Bohn (1998) and Debrun et al. (2008).
  - Dependent variable: primary balance to GDP (PB) or general government cyclically adjusted primary balance (CAPB).
  - Explanatory variables: lagged primary balance (PB_{it-1}), debt-to-GDP ratio (d), output gap (gap), decentralization measures (spending decentralization, revenue decentralization, transfer dependency), interactions, and control variables x (political and fiscal institutions). Country fixed effects η_i and error ε_it included.
- Output gap: actual GDP less potential GDP as percent of potential (positive values indicate operating above potential).
- Data panel: EU27 data for 1990-2008 unbalanced panel; typical specifications report Observations 322 and Number of countries 27 (other robustness columns show varying observations and country counts).
- Estimation method: bias-corrected Least Square Dummy Variable (LSDVC) estimator of Bruno (2005) to correct dynamic panel fixed-effects bias in narrow-T panels; LSDVC preferred over IV and GMM in small-T settings.
- Baseline econometric findings:
  - Spending decentralization improves fiscal performance across specifications; examples of estimated coefficients (from reported tables):
    - Spending decentralization coefficients include 0.0730**, 0.562***, 0.801*** (Table 2 panel); 0.529***, 0.567***, 0.583***, 0.567***, 0.562***, 0.560***, 0.534*** (Table 3); robustness splits include 0.948*** (column 1), 0.699*** (column 3), 0.761** (column 5), 0.844*** (column 6), 0.489*** (column 7).
  - Transfer dependency diminishes the positive impact of spending decentralization: interaction spending decentralization x Transfer dependency negative and significant (examples: -0.00485***, -0.00651***; -0.00462*** to -0.00504*** across specifications).
  - Revenue decentralization associated with deterioration in fiscal position in several specifications (negative significant coefficients reported: -0.203*** in one panel; -0.555*** to -0.685*** in Table 2; robustness hits include -0.901***, -0.728***, -0.678**, -0.881***, -0.502***).
  - CAPB displays significant persistence: lagged dependent variable coefficients positive and significant across specifications (examples: 0.575***, 0.552***, 0.504***, 0.336***; 0.498*** to 0.508*** in Table 2; 0.403***, 0.373***, 0.533***, 0.433***, 0.0617, 0.549***, 0.485*** in Table 4).
  - Lagged debt coefficients often positive and significant (examples: 0.0455***, 0.0448***, 0.0563***, 0.0083; 0.0585***, 0.0537***, 0.0565***, 0.0559***, 0.0563***, 0.0562***, 0.0580***).
  - Output gap coefficients negative and significant in many specifications indicating procyclical policy (examples: -0.263***, -0.270***, -0.304***, -0.282***; -0.320*** to -0.301*** in Table 2).
  - Parliamentary election variable generally negative and significant in many specifications (examples: -0.479**, -0.492**, -0.445**, -0.466**, -0.458**, -0.446**, -0.467**, -0.509***, -0.39, -0.671***, -0.421**).
- Institutional interaction tests:
  - Overall fiscal rule index included in baseline sometimes improves performance (examples: 0.349* in Table 3; 0.464** in Table 4 column 1), but the positive effect may not be robust.
  - Central and subnational fiscal rules considered separately do not matter; subnational fiscal rule index often not significant (example coefficient -0.212 with large standard error in Table 3).
  - No significant impact from stringency of rules constraining subnational borrowing or from subnational budget-balance rules in presented specifications.
  - Interaction between spending decentralization and formal coordination in medium-term budgetary framework not statistically significant.
- Robustness checks overview:
  - Replacing dependent variable with general government balance yields similar message.
  - Using changes in general government debt as alternative performance measure discussed.
  - Subsample analyses:
    - New Member States (NMS) vs EU15: decentralization variables lose significance in NMS sample; results hold for EU15; fiscal rules matter in NMS.
    - Split sample in 1999 (SGP): no major differences pre/post-1999 except fiscal rules lose significance before SGP.
    - Excluding 2008: results unchanged.
    - Leave-one-out: estimates stable except fiscal rules often not significant.

### VI. Stylized facts and institutional practices (Appendices summary)
- Expenditure assignments: main subnational responsibilities include preschool, primary and secondary education, vocational training, hospitals, primary care, welfare homes, care for the elderly, etc.
- Fiscal rules at subnational level:
  - Mostly budget balance rules (BBR) or borrowing constraints (BC); both nearly doubled between 1990 and 2005.
  - Expenditure limits (EL) and tax limits (TL) uncommon at subnational level.
  - Sanctions and escape clauses are frequently weak or not predefined:
    - For BBR: 0.26 impose financial sanctions, 0.04 sanction officials, 0.26 mandate actions, 0.35 constrain actions, 0.26 other, 0.22 not predefined.
    - For BC: 0.21 impose financial sanctions, 0.05 sanction officials, 0.37 mandate actions, 0.26 constrain actions, 0.21 other, 0.26 not predefined.
    - For EL: 0.27 impose financial sanctions, 0.00 sanction officials, 0.18 mandate actions, 0.27 constrain actions, 0.18 other, 0.55 not predefined.
    - For TL: 0.00 impose financial sanctions, 0.00 sanction officials, 0.11 mandate actions, 0.00 constrain actions, 0.22 other, 0.67 not predefined.
  - Escape clauses by rule type (percent of rules):
    - BBR: shock to local economy 0.13, shock to local revenues 0.09, natural/other disaster 0.30, no escape 0.30, other 0.13, not predefined 0.35.
    - BC: shock to local economy 0.05, shock to local revenues 0.05, natural/other disaster 0.26, no escape 0.26, other 0.05, not predefined 0.47.
    - EL: natural/other disaster 0.18, no escape 0.18, not predefined 0.64.
    - TL: not predefined 1.00.
- Subnational bailouts occurred in at least nine EU countries; bailouts more frequent where more administrative tiers exist (one tier ~ one quarter of countries with bailouts; two tiers 44.4 per cent; three tiers about two thirds).

### VII. Key empirical and policy-relevant takeaways
- Empirical results point to a positive association between spending decentralization and better general government fiscal performance in the EU.
- This positive association is attenuated when subnational governments are highly transfer-dependent.
- Subnational fiscal rules, as measured, do not show a clear beneficial effect on general government fiscal outcomes; overall fiscal rule strength shows some positive association but may not be robust.
- Subnational spending responsibilities increased faster than corresponding decentralization of revenue sources (own revenue and transfers), leading to larger vertical imbalances and possible unfunded mandates.
- Policy-relevant considerations emerging from findings:
  - Ensure that decentralization of spending is matched by commensurate resource decentralization (own revenue instruments or transfers) to avoid vertical imbalances.
  - Reducing transfer dependency of subnational governments may preserve fiscal-discipline benefits associated with spending decentralization.
  - Strengthen effectiveness (implementation, enforcement, sanctioning) of fiscal rules across levels of government; mere existence of subnational rules is not sufficient.
  - Improve formal coordination mechanisms between central and subnational levels (including inclusion of subnational targets in medium-term frameworks), although coordination dummy was not statistically significant in presented specifications.

*Content derived from the supplied excerpt of the IMF working paper (sections I–V, stylized facts, figures, econometric framework, and appendices) in the provided PDF content.*

### 1. Features of Subnational Governments .................................................................................

### _wp1245 - 1. Features of Subnational Governments .................................................................................

### Sections
- 1. Features of Subnational Governments .................................................................................18
- 2. Fiscal Decentralization and Fiscal Performance ..................................................................19
- 3. Do Fiscal Institutions Matter? ..............................................................................................20
- 4. Robustness Checks...............................................................................................................21

### Figures
- 1. Trends in Subnational Expenditure Shares by Number of Countries,1995–2008 .................6
- 2. Fiscal Rule Strength and Level of Decentralization ..............................................................8
- 3. Average Fiscal Balances by Level of Decentralization in the 
European Union, 1995–2009 ................................................................................................10
- 4. Fiscal Performance in the European Union, 1995–2008 .....................................................10
- 5. Increase in Spending and Revenue Decentralization ...........................................................12
- 6. Fiscal Rules Indices and Fiscal Performance, 2008.............................................................12

### Appendices
- 1. Expenditure Assignments to Subnational Governments in the European Union ................22
- 2. Fiscal Rules at the Subnational Level in the European Union .............................................23
- 3. Selected Episodes of Subnational Bailouts in the European Union ....................................25
- 4. Data Sources and Definitions ...............................................................................................26

### Appendix Tables and Figures
- 1. Fiscal Rules in the EU Member States by Type of Rule and 
Level of Government, 2008 ..................................................................................................23
- 2. Sanctions and Escape Clauses by Type of Fiscal Rule ........................................................24

*Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp1245.pdf*

### References .............................................................................................................

### _wp1245 - References .............................................................................................................

### I. Introduction and research question
- Many European countries have reassigned spending and revenue collection responsibilities from the center to subnational (local and regional) governments.
- Subnational spending in the European Union (EU) increased from 23 percent of general government spending in 1995 to 26 percent in 2009, with the revenue share increasing to a lesser extent.
- Economic rationales for decentralization: better information and matching of policies to citizens’ preferences (Oates 1972); jurisdictional competition limiting local tax burden (Brennan and Buchanan, 1980); increased accountability and transparency.
- Risks of decentralization: “common pool” transfer financing can induce overspending and lower tax effort; bailout expectations can soften local budget constraints (Rodden et. al. 2003).
- Institutional arrangements—e.g., fiscal rules—could potentially correct incentives, enhance accountability, and anchor expectations.
- Purpose of paper: assess empirically the impact of fiscal decentralization on fiscal performance in the EU, explicitly examining the role of institutional arrangements covering subnational governments.
- Summary finding highlighted in the introduction:
  - Spending decentralization has been associated with stronger fiscal performance, especially when transfer dependency of subnational governments was low.
  - Subnational fiscal rules do not seem to play a role in ensuring better performance.

### II. Literature and theoretical background
- Empirical literature on decentralization and fiscal performance is mixed; cross-country econometric work is scarce and focuses mostly on OECD countries.
- Theoretical concerns:
  - Decentralization complicates macroeconomic stabilization and may make countercyclical policies more difficult (Ter-Minassian, 1997a; Tanzi, 1995; Tanzi, 2000; IMF, 2009).
  - Decentralization can create coordination failures and “deficit bias” (Oates, 2006), especially when central bailouts of indebted subnational entities occur.
- Empirical cross-country findings cited:
  - Rodden (2002): revenue decentralization deteriorates the general government balance.
  - Neyapti (2010): revenue and spending decentralizations improve the balance.
  - Afonso and Hauptmeier (2009): higher spending decentralization worsens the primary balance (for high debt levels); revenue decentralization does not matter.
  - Thornton (2009): no significant impact of revenue decentralization.
  - Baskaran (2010): expenditure decentralization reduces public indebtedness; tax decentralization effect insignificant.
- Three institutional features emphasized as crucial:
  - Transfer dependency: higher reliance on transfers may reduce overall balance; grants can be additional to central spending and endogenous to deficits (Rodden 2002; Fornasari et al., 2000; De Mello, 2007).
  - Subnational borrowing autonomy: unrestricted borrowing (or “soft” financing) can undermine discipline; restricting borrowing associated with better performance (Rodden, 2002; Plekhanov and Singh, 2007).
  - Fiscal rules: may offset coordination problems, but empirical evidence is inconclusive—rules at subnational level often show no significant impact on general government fiscal balances (Debrun et al., 2008; Afonso and Hauptmeier, 2009).

### III. Institutional features of European subnational governments (stylized institutional facts)
- Wide variation in subnational role across EU: subnational spending share ranges from less than two per cent in Malta to almost two-thirds in Denmark.
- Federal vs unitary: only Austria, Belgium, and Germany are federal; some unitary countries (Denmark, Finland, Spain) are highly decentralized.
- More decentralized countries tend to have more tiers of subnational government.
- Main areas of subnational expenditure: education, health, and social welfare. Between 1995 and 2008:
  - Subnational expenditure shares for education and social welfare rose.
  - Subnational expenditure share for health decreased in the majority of countries (eight countries reduced the subnational expenditure share for health; three—Estonia, Hungary, Romania—reduced it by 15-20 percentage points; Ireland’s subnational health share dropped from 95 percent to zero after 2005 reassignment to the center).
- Control mechanisms:
  - Fiscal rules (mainly borrowing or balanced budget rules) are common at subnational levels; number of fiscal rules increased at central and general government levels.
  - Majority of fiscal rules are applied at the local government level.
  - Budget balance rules more prevalent in EU15; debt/borrowing rules common in new member states (NMS); expenditure rules rare at subnational level.
  - Subnational fiscal rules are more prevalent in countries with higher decentralization and where subnational governments rely more on own revenues.
  - Fiscal rules indices used come from the European Commission Fiscal Rule Index database (European Commission, 2009); index measures legal basis, coverage, monitoring/enforcement strictness (including sanctions and escape clauses), and media visibility.
  - Strength of rules does not necessarily equal effectiveness: sanctions are often weak; central governments retain discretion; bailouts have occurred in at least nine EU countries; bailouts more frequent where more administrative tiers exist.
  - Coordination in budgetary procedures is limited: less than one-third of countries have formal coordination arrangements; minority of countries include subnational targets in medium-term budgetary frameworks.

### IV. Stylized facts on decentralization and fiscal performance (empirical patterns)
- Data: Eurostat fiscal data covering 1995-2008 used to assess general government fiscal performance (balance and debt). Definitions noted:
  - Subnational spending defined as local and regional government spending excluding transfers paid.
  - Subnational revenue defined as revenues of local and regional governments excluding transfers received; transfers are net current and capital transfers received from other levels of government.
  - Revenue autonomy measured by share of subnational own revenues (adjusted for central transfers) in general government revenues.
  - Transfer dependency measured by share of transfers received by subnational governments in total subnational revenues.
- Stylized fact 1: Spending decentralization is associated with better fiscal performance at the general government level.
  - Over 1995–2008, cyclically adjusted general government fiscal balances were higher among more decentralized countries (e.g., Denmark, Sweden, Spain) and much lower in less decentralized countries (e.g., Greece, Malta, Slovakia).
  - Increases in spending decentralization are not associated with increases in debt.
  - On average, overall fiscal balances in countries with medium and low decentralization are respectively 2 and 2½ percentage points of GDP below those of countries with high decentralization.
  - Subnational governments have close-to-balance fiscal positions irrespective of degree of decentralization; central government positions are relatively stronger in highly decentralized countries.
- Stylized fact 2: Expenditure decentralization has outpaced decentralization of resources to subnational governments.
  - Between 1995 and 2009:
    - Subnational spending rose by 3¾ percentage points as a share of general government spending (Average = 3.79).
    - Increase in subnational own revenues averaged 1.09 percentage points.
    - Change in net transfers averaged 1.4 percentage points.
  - Vertical imbalances (gap between spending and revenue decentralization) increased over time, suggesting possible unfunded mandates and rationing of resources to subnational governments.
  - This rationing could have forced expenditure savings at subnational level—especially if borrowing constrained—and contributed to improved fiscal positions at the center and general government.
- Stylized fact 3: Subnational fiscal rules do not appear to have an effect on fiscal performance.
  - Overall fiscal rule index shows a positive relationship with general government balance, but the subnational fiscal rule index does not show a clear relationship.
  - Possible interpretations: weak implementation of rules; bailouts; rules introduced where fiscal performance was weak.

### V. Econometric analysis: specification, methodology, and key results
- Empirical approach: estimate a fiscal reaction function following Bohn (1998) and Debrun et al. (2008). General form (as presented):
  - Dependent variable: primary balance to GDP (PB).
  - Explanatory variables include: lagged primary balance (PB_{it-1}), debt-to-GDP ratio (d), output gap (gap), decentralization measures (Dec: spending decentralization, revenue decentralization, transfer dependency), interactions, and control variables x (political and fiscal institutions). Country fixed effects η_i and error ε_it included.
  - Output gap defined as actual GDP less potential GDP as a percent of the latter (positive gap values indicate operating above potential).
- Data panel: EU27 data for 1990-2008 constituting an unbalanced panel (reflecting data availability).
- Estimation method: bias-corrected Least Square Dummy Variable (LSDVC) estimator of Bruno (2005) to correct the dynamic panel fixed-effects bias in narrow-T panels; LSDVC preferred over IV and GMM in small-T settings.
- Baseline econometric findings:
  - Spending decentralization improves fiscal performance across specifications.
  - Significant persistence in the cyclically adjusted primary balance (CAPB).
  - Evidence of procyclical fiscal policy in the estimates (negative coefficient on output gap).
  - Electoral cycle evident: fiscal performance worsens on election years.
  - Transfer dependency diminishes the positive impact of spending decentralization (interaction effect): when subnational governments rely more on transfers, the improvement from spending decentralization is reduced.
  - Revenue decentralization appears associated with deterioration in the fiscal position in some specifications.
  - Interpretation: tight subnational resource constraints (restricted own revenue and transfers) may have been used as rationing mechanisms by the center, contributing to better overall fiscal performance; losing central control over subnational resources could undermine this channel.
  - Caution: the revenue decentralization measure (subnational own revenue as share of general government revenue) may capture other effects (e.g., decline in general government revenue raising the measured decentralization even if subnational revenue constant).
- Institutional interaction tests:
  - Overall fiscal rule index included in baseline: overall rules improve performance (consistent with literature), but this positive effect may not be robust.
  - Central and subnational fiscal rules considered separately do not matter (consistent with stylized fact 3).
  - Testing stringency of rules constraining subnational borrowing: no significant impact found.
  - Subnational budget-balance rules also do not show significant effects.
  - Interaction between spending decentralization and presence of formal coordination in the medium-term budgetary framework: not statistically significant.
- Robustness checks (overview):
  - Replacing dependent variable with general government balance yields similar message.
  - Using changes in general government debt as a measure of performance is discussed as an alternative because debt increases sometimes exceeded recorded deficits; additional robustness exercises are indicated (details beyond presented excerpt).

### Key empirical and policy-relevant takeaways (bulleted)
- Empirical results point to a positive association between spending decentralization and better general government fiscal performance in the EU.
- This positive association is attenuated when subnational governments are highly transfer-dependent.
- Subnational fiscal rules, as measured, do not show a clear beneficial effect on general government fiscal outcomes; overall fiscal rule strength shows some positive association but may not be robust.
- Subnational spending responsibilities increased faster than corresponding decentralization of revenue sources (own revenue and transfers), leading to larger vertical imbalances and possible unfunded mandates.
- Potential policy implications that emerge from the findings (presented as in-text considerations rather than formal prescriptions):
  - Ensuring that decentralization of spending is matched by commensurate resource decentralization (own revenue instruments or transfers) could be important to avoid vertical imbalances.
  - Reducing transfer dependency of subnational governments may preserve the fiscal-discipline benefits associated with spending decentralization.
  - Strengthening the effectiveness (implementation, enforcement, sanctioning) of fiscal rules across levels of government may be required if rules are to alter fiscal outcomes; mere existence of subnational rules is not sufficient.
  - Improving formal coordination mechanisms between central and subnational levels (including inclusion of subnational targets in medium-term frameworks) could be important but was not shown statistically significant in the specifications presented.

*Italicized source attribution: Content derived from the supplied excerpt of the IMF working paper (section headings I–V, stylized facts, figures, and econometric framework) in the provided PDF content.*

### introduction of the Stability and Growth Pact (Buti et al., 2007), possibly indicating the use

### _wp1245 - introduction of the Stability and Growth Pact (Buti et al., 2007), possibly indicating the use

### Research scope and caveats
- Results should be interpreted with caution since the analysis does not explicitly account for the potential endogeneity of fiscal rules (i.e. governments with stronger preference for fiscal discipline are more likely to adopt stronger fiscal rules).
- Debrun et al. (2008) finds negligible the potential estimator bias introduced by reverse causality in a similar model.
- Numerical fiscal institutions’ indicators used may not capture the complexities of interactions between the center and subnational levels of government.
- Some results are not reported in the source “in the interest of brevity.”

### Main empirical findings on decentralization and fiscal performance
- Spending decentralization:
  - Improves the fiscal position of the general government.
  - Shows positive coefficients across specifications (examples: 0.0730**, 0.562***, 0.801*** in Table 2 panel; 0.529***, 0.567***, 0.583***, 0.567***, 0.562***, 0.560***, 0.534*** in Table 3).
  - Effect is reduced when transfer dependency is high (interaction term spending decentralization x Transfer dependency negative and significant: -0.00485***, -0.00651***; -0.00462*** to -0.00504*** across specifications).
  - In Table 4 robustness splits, spending decentralization coefficients reported include 0.948*** (column 1), 0.699*** (column 3), 0.761** (column 5), 0.844*** (column 6), 0.489*** (column 7).
- Revenue decentralization:
  - Associated with weaker fiscal performance at the general government level (negative and significant coefficients: -0.203*** in one panel; -0.555*** to -0.685*** in Table 2; -0.901***, -0.728***, -0.678**, -0.881***, -0.502*** in robustness checks).
- Transfer dependency:
  - Reduces the positive effect of spending decentralization (see significant negative interaction terms).
- Elections:
  - Parliamentary election variable generally negative and significant in many specifications (examples: -0.479**, -0.492**, -0.445**, -0.466**, -0.458**, -0.446**, -0.467**, -0.509***, -0.39, -0.671***, -0.421**).
- Dynamic panel controls:
  - Lagged dependent variable coefficients are positive and significant across specifications (examples: 0.575***, 0.552***, 0.504***, 0.336***; 0.498*** to 0.508*** in Table 2; 0.403***, 0.373***, 0.533***, 0.433***, 0.0617, 0.549***, 0.485*** in Table 4).
  - Lagged government debt often positive and significant (examples: 0.0455***, 0.0448***, 0.0563***, 0.0083; 0.0585***, 0.0537***, 0.0565***, 0.0559***, 0.0563***, 0.0562***, 0.0580***; 0.0367***, -0.176***, 0.0415***, 0.0714***, 0.076, 0.0606***, 0.0583***).
  - Lagged output gap coefficients negative and significant in many specifications (examples: -0.263***, -0.270***, -0.304***, -0.282***; -0.320*** to -0.301*** in Table 2; -0.123***, 0.277***, -0.379***, -0.305***, -0.225, -0.288***, -0.296*** in Table 4).

### Role of fiscal rules and institutions
- Subnational fiscal rules:
  - Results suggest subnational fiscal rules have not played a material role on fiscal performance; subnational fiscal rule index not significant in many specifications (example coefficient -0.212 with large standard error in Table 3).
  - Possible explanations:
    - Fiscal rules in the EU may be relatively weak because the center has considerable discretion in addressing breaches.
    - Rules may be breached due to politically sensitive spending (e.g., health care), prompting central government compensation and rendering rules nonbinding.
    - Subnational fiscal rules may mainly address fiscal indiscipline and not unfunded mandates.
- Overall fiscal rule index:
  - Sometimes positive and statistically significant (examples: 0.349* in Table 3; 0.464** in Table 4 column 1; other columns show mixed significance and signs).
- Interactions between decentralization and rules:
  - Interactions tested (e.g., Spending decentralization x Subnational debt rule; Spending decentralization x Subnational budget balance rule; Spending decentralization x Central debt rule; Spending decentralization x Central budget balance rule; Spending decentralization x Coordination dummy) generally show nonsignificant coefficients in presented specifications.

### Robustness checks and subsample analysis
- Subsamples:
  - New Member States (NMS, countries that joined the EU in 2004 or thereafter) vs EU15:
    - Decentralization variables lose significance in the NMS sample, while results hold for the EU15.
    - Fiscal rules matter in NMS countries.
    - Results for NMS should be interpreted with caution because the NMS sample is short and has less variation in decentralization measures.
  - Split sample in 1999 (first year of the Stability and Growth Pact):
    - No major difference between pre-1999 and post-1999 sample periods except fiscal rules lose significance before the introduction of the SGP.
    - This pre-1999 result is driven by EU15 countries and could indicate fiscal discipline was observed even in the absence of rules in the run up to the euro.
  - Excluding 2008:
    - Excluding the year 2008 (to eliminate potential crisis effects) leaves results unchanged.
  - Leave-one-out:
    - Excluding one country at a time does not significantly alter estimates except for fiscal rules, which are not significant in most leave-one-out cases.

### Stylized facts on subnational fiscal institutions and practices (Appendices)
- Features of subnational governments (2009 decentralization classification examples):
  - Unitary low: BUL, CYP, LUX, MLT, SVN, GRC, IRL, PRT, SVK, etc.
  - Unitary medium: EST, LTU, LVA, CZE, HUN, ROM, ITA, POL, GBR, etc.
  - Unitary high / Federal high examples: FIN (Unitary high), BEL (Federal high), DEU (Federal high), ESP (Unitary high).
  - Number of subnational tiers examples: Three, One, Two (country-level mapping shown in table).
- Expenditure assignments to subnational governments:
  - Appendix 1 lists education, health, welfare responsibilities across countries (examples include preschool, primary and secondary education, vocational training, hospitals, primary care, welfare homes, care for the elderly, etc.). (Sources: World Bank Database of Qualitative Decentralization Indicators; Monasterio-Escudero and Suarez-Pandiello (2002); McLure and Martinez Vazquez (2000); Vigvari (2008); and OECD (2010).)
- Fiscal rules at the subnational level (Appendix 2):
  - Subnational fiscal rules are mainly budget balance rules or debt rules; both nearly doubled between 1990 and 2005.
  - Budget balance rules are the most common; expenditure rules are uncommon at subnational level.
  - Subnational rules more prevalent in countries with higher decentralization and lower transfer dependency.
  - Sanctions and escape clauses:
    - In more than one fifth of countries there are no predefined sanctions at all.
    - Infringement of a budget balance requirement is sanctioned by more than three quarters of countries; borrowing constraints are sanctioned in barely two thirds; expenditure or tax limitations penalize non-compliance in only about one third of cases.
    - Many sanctions are weak or discretionary (e.g., option to “recommend actions”).
    - Only a few countries have absolutely no “escape clauses”; escape clauses mainly apply for natural/other disasters or shocks to local revenues/economy.
  - Appendix Table 2 (percent of total rules in sample):
    - Sanctions by rule type (percent of total rules): BBR: 0.26 impose financial sanctions, 0.04 sanction officials, 0.26 mandate actions, 0.35 constrain actions, 0.26 other, 0.22 not predefined.
    - BC: 0.21 impose financial sanctions, 0.05 sanction officials, 0.37 mandate actions, 0.26 constrain actions, 0.21 other, 0.26 not predefined.
    - EL: 0.27 impose financial sanctions, 0.00 sanction officials, 0.18 mandate actions, 0.27 constrain actions, 0.18 other, 0.55 not predefined.
    - TL: 0.00 impose financial sanctions, 0.00 sanction officials, 0.11 mandate actions, 0.00 constrain actions, 0.22 other, 0.67 not predefined.
    - Escape clauses by rule type (percent): BBR: shock to local economy 0.13, shock to local revenues 0.09, natural/other disaster 0.30, no escape 0.30, other 0.13, not predefined 0.35.
    - BC: shock to local economy 0.05, shock to local revenues 0.05, natural/other disaster 0.26, no escape 0.26, other 0.05, not predefined 0.47.
    - EL: shock to local economy 0.00, shock to local revenues 0.00, natural/other disaster 0.18, no escape 0.18, other 0.00, not predefined 0.64.
    - TL: shock to local economy 0.00, shock to local revenues 0.00, natural/other disaster 0.00, no escape 0.00, other 0.00, not predefined 1.00.
    - Note: BBR = Budget Balance Rule; BC = Borrowing Constraint; EL = Expenditure Limit; TL = Tax Limit. (Tables apply to 26 and 24 countries as noted.)
- Subnational bailouts (Appendix 2 and 3):
  - Subnational governments had to be bailed out by higher levels of government in at least nine EU countries.
  - Bailouts more frequent in more decentralized countries with more administrative tiers:
    - Share of countries with subnational bailout by number of tiers: one tier ~ one quarter; two tiers 44.4 per cent; three tiers about two thirds.
  - Selected episodes:
    - DEU: 1992 Bremen and Saarland assistance; 1993 five-year contract with additional grants; target debt reduction from DM 16 billion to DM 11.5 billion missed; 1999 extension until 2004; further transfers after 2004 excluded; implication: no differences in credit risks of German states.
    - ESP: late 1970s overspending/over-borrowing after democracy restoration; 1980 center assumed 50% of local authorities' debt; 1983 central government covered current deficits by grants and granted local tax-setting freedom (later declared anti-constitutional); 1988 wide-ranging reform of local finances.
    - HUN: 1999 one-third of localities applied for deficit grants; deficit grants provided soft budget constraint; since 1996-1997 central government improved transparency and audit procedures.
    - ITA: 1977 transfers from center increased by 300% and fiscal rules introduced; 1978 center assumed responsibility for municipal debts accumulated before 1977; tight local expenditure control did not solve soft budget-constraint problem; 1990s public finance reforms reduced transfers and increased local revenue and spending autonomy.
    - SWE: 1992 Haninge city assistance for housing company debt; 1995 central government assumed debt and extended loan; extra grant given to city; city lost most company shares and had to raise local tax by one percentage point; by 1998, 87 of 288 municipal governments had applied at least once for financial assistance.

### Tables and estimation notes (selected)
- Estimation methods and notes:
  - Estimators include LSDVC, LSDVC FE, CLSDVC FE (noted as LSDVCLSDVCLSDVCFE across table header).
  - Robust standard errors in parentheses. Significance codes: *** p<0.01, ** p<0.05, * p<0.1.
  - Dependent variable in primary regressions: general government primary cyclically adjusted balance (CAPB).
  - Number of observations and countries reported across tables (examples: Observations 322, Number of countries 27 in multiple specifications; other robustness columns show Observations 300, 195, 127, 65, 235, 295 and Number of countries 27, 15, 12, 22, 27, 27).
  - Fixed effects F-test reported (example: Fixed effects (F-test) 8.48***).

*Italic: Source document content provided from the specified IMF working paper excerpt.*

### Appendix 4. Data Sources and Definitions

### Appendix 4. Data Sources and Definitions

### Fiscal data
- Data source: Eurostat covering the years 1990-2008.
- Sample: an unbalanced panel including all EU27 countries for which data for the period were available; this yielded a sample with about 12 observations per country on average.
- Note on coverage: The number of countries is substantially lower at the beginning of the sample, particularly for NMS as coverage for subnational fiscal statistics has only improved over time.
- Subnational measures: calculated by aggregating the regional and local government subsectors (S1312 and S1313 in ESA95).
- Variables used in the econometric analysis:
  - Spending decentralization: Subnational spending (excluding transfers paid) in percent of general government spending.
  - Revenue decentralization: Subnational revenue (excluding transfers received) in percent of general government revenue.
  - Transfer dependency: Subnational net transfers received (both current and capital) in percent of subnational revenue.
  - CAPB: General government cyclically adjusted primary balance calculated as follows:
    - CAPB = r - g + (1 + εg) / (1 + εr) * gap ? (original formula layout in source indicates CAPB constructed using r, g, gap, εr, εg; see definitions below)
    - Definitions of components:
      - r is primary revenue in percent of GDP;
      - g is the primary expenditure in percent of GDP;
      - gap is the output gap;
      - εr is the elasticity of revenue with respect to the output gap; and
      - εg is the elasticity of expenditure with respect to the output gap.
    - Footnote 18: Where available, elasticities are taken from Girouard and Andre (2005). In other cases, revenue elasticity is assumed to be 1 and expenditure elasticity is assumed to be 0.
  - Debt: Gross general government debt in percent of GDP.

### Macro data
- Output gap: Macroeconomic data needed to calculate the output gap were taken from the IMF’s World Economic Outlook database.

### Political data
- Data on political institutions are based on the World Bank Political Database (see, Thorsten et al., 2001).

### Fiscal institutions
- Data on fiscal rules and medium-term budget frameworks come from the European Commission and are available at:
  - http://ec.europa.eu/economy_finance/db_indicators/fiscal_governance/fiscal_rules/index_en.htm

### Fiscal rules
- Overall fiscal rule index: comes from the European Commission Fiscal Rule Index database (see, EC, 2009).
- Construction of strength index: based on legal basis, coverage, strictness of monitoring and enforcement (including through sanctions and escape clauses), and media visibility.
- Aggregation across government levels:
  - Indices available for central, social security, regional and local government level rules are used to construct two series:
    - central (combining central and social security); and
    - subnational (combining regional and local).
  - Aggregation technique: seeks to use the same methodological principle as Debrun et al. (2008), placing a higher weight of 1 on the strongest rule and a weight of ½ on the remaining rules.

### Medium-term budget framework
- For the econometric analysis a dummy variable is constructed that takes value 1 if there is coordination with subnational levels of government in the preparation and status of the Stability and Convergence Program.

*Source: Appendix 4. Data Sources and Definitions (from the supplied IMF content unit).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2012/_wp1245.pdf_