## _wp14192

## Source details

**Canonical URL:** [_wp14192](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14192.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14192.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14192.pdf.json)

---

### I. Introduction
- Goal: assess successes and failures of past public wage bill consolidations and draw lessons for post–Great Recession episodes.
- Coverage:
  - Historical episodes: 1979–2009.
  - Recent episodes: 2009–present.
  - Classification of measures by type: action on average wage and/or employment; one-off vs. structural; extent of consultations/social dialogue.
- Context and cross-country statistics referenced:
  - Wage bill shares (2010 and text references):
    - As a share of total government expenditures: about 20–25 percent in AEs and EMEs, almost 30 percent in LICs.
    - As a share of GDP: about 10 percent in AEs, between 5 and 10 percent in EMEs and LICs.
    - Public employees represent about 7–8 percent of the total population in AEs, and about 5 percent in EMEs and LICs.
  - Government wage bill in health and education (IMF, April 2014 Fiscal Monitor): about 30 and 60 percent of government spending in health and education, respectively, in advanced economies.
- Main findings (summary):
  - Structural measures have been associated with larger and more sustained reductions in the public wage bill as a share of GDP.
  - Prior to the Great Recession, countries that created consensus and/or negotiated reductions via social dialogue achieved larger and more permanent reductions.
  - Historical and recent episodes are similar in many respects (frequency of targeting average wage vs. employment); main difference: less social dialogue in recent episodes.
  - Early evidence for recent episodes: structural measures again appear to produce larger and more sustained reductions, though it is too early for definitive statements.
- Structure of paper: Section II (trends), III (literature), IV (data and episodes), V (results), VI (conclusions).

### II. Trends in Public Sector Compensation and Employment
- Cross-country patterns (selected points from figures/text):
  - Figure 1 (2010):
    - Public wage bill highest for advanced economies: around 10 percent of GDP in 2010.
    - Emerging market economies: about 6 percent of GDP.
    - Low income countries: about 7 percent of GDP.
    - Average wage ratio defined as average wage of general government divided by that of entire economy. General government employment figures refer to 2008 data.
  - Figure 2:
    - Low income countries: wage bill as share of GDP has been on the rise over the past decade.
    - Emerging market economies: relatively stable expenditure on the public wage bill.
    - As a share of general government expenditures: low income countries tend to spend more but with volatility; emerging and advanced economies show similar shares with a tendency to decline over the past decade.
  - Figure 3:
    - Public employment as a share of total population is quite stable over time in advanced and emerging economies.
    - For low income countries, insufficient data to assess trends in public employment over time.
- Data sources:
  - Wage bill data: IMF World Economic Outlook database, Eurostat, AMECO, and occasionally IMF country teams.
  - Employment data: International Labour Organization; World Development Indicators.

### III. Literature Review
- Civil service reforms (lessons from emerging/developing economies):
  - Emergency measures (temporary wage/hiring freezes) tend to have only short-term effects.
  - Long-term reforms yield substantial results but are politically difficult to implement.
  - Successful reforms often included targeting by skills and age plus compensation packages aiding reallocation; such reforms could be financially costly.
  - Coordination needed between wage bill reforms and reforms in labor-intensive health and education; misuse of “wage bill ceilings” without effective payroll management may impede hiring in social sectors.
- Public wage premium:
  - Significant public wage premium documented in many European countries, often higher at lower pay scales and for women; EMEs and LICs tend to spend less on the public wage bill as a share of GDP and have lower levels of public employment, implying typically higher average public wages compared to per-capita GDP (comparisons are suggestive).
- Public–private labor market interactions:
  - Public compensation affects private compensation; effect stronger with centralized wage bargaining and high union membership, weaker with high trade openness and wage indexation.
  - Increasing trade openness may reduce the effect of public-sector wages on private-sector wages.

### IV. Episodes of Public Employment and Compensation Reform
- Episode selection and scope:
  - Two sets: historical (1979–2009) and recent (post-2009). Dataset compiled from Article IV Staff Reports and other sources.
  - Identified episodes: 16 historical and 23 recent episodes of wage bill consolidation.
- Identification and classification dimensions:
  - Target of measures:
    - Average wage measures (most common: wage freeze, occasionally wage cut).
    - Employment measures (hiring freezes, attrition, abolishing positions, closing agencies).
  - Short-term vs. structural:
    - Short-term: one-off measures such as temporary wage freezes.
    - Structural: longer-term reforms (reforming bargaining model, compensation grid, wage indexation rules, introducing pay-for-performance, abolishing positions/closing agencies).
- Sources used for documentation in addition to Article IVs: OECD Public Sector Compensation in Times of Austerity (2012), IMF country teams, IMF Recent Economic Developments, Stability and Convergence Programmes, Eurostat employment and compensation data, ILO Laborsta when necessary.

### V. Key Empirical Results and Stylized Facts
- Main empirical regularities:
  - Structural measures are associated with larger and more sustained reductions in the public wage bill as a share of GDP (Figures 5, 6, 9 referenced).
  - Episodes with consensus-building/social dialogue achieved larger and more permanent reductions historically (Figures 7, 8, 10 referenced).
  - Historical vs. recent episodes: similar frequency of measures targeting average wage vs. employment; recent episodes show less social dialogue.
  - Early evidence in recent episodes: structural measures again appear to produce larger and more sustained reductions, but assessment is tentative because of limited time horizon.
- Specific stylized statistics and observations (preserve original numeric values):
  - Historical dataset: 16 episodes (1979–2009). Recent dataset: 23 episodes (2009–present).
  - In historical episodes, over 80 percent of episodes included wage freezes.
  - Between 50 and 60 percent of episodes also included measures on the employment level.
  - Average reduction in the public wage bill is similar for historical and recent episodes: about 1 percent of potential GDP.
    - For historical episodes the wage bill reduction is the cumulative reduction calculated over five years after the first year of measures.
    - For recent episodes the wage bill reduction is calculated as the cumulative reduction from 2009 to 2012.
  - Figure 5 (structural versus non-structural historical): difference in reduction of the public wage bill as a share of GDP is about 0.5 percent of GDP after four years; over seven to ten years the difference can be as large as about 0.7 percent of GDP.
  - Figure 7 (social dialogue versus little social dialogue historical): difference about 0.5 percent of GDP after about five years; over seven to ten years the difference can be as large as about 1 percent of GDP or more.
  - Recent episodes (Figure 9): structural measures associated with reductions of close to 1 percent of potential GDP already in the first year after measures were taken.
  - Recent episodes (Figure 10): difference in wage bill reductions between countries with and without social dialogue is about 0.25 percent of GDP over the three-year window (2009–2012).
- Decompositions and component results:
  - Wage bill components considered: (1) average public wage component (average wage divided by GDP per capita); (2) public employment component (public employment divided by total population).
  - Historical episodes with structural measures had a smaller increase in the public employment level compared to episodes with non-structural measures.
  - Average reduction in the average wage was not very different between episodes with and without structural measures; in years six to 10 the average wage (as a percentage of per capita GDP) appears to fall less for episodes with structural measures.
  - Episodes with social dialogue tend to have lower increases in public employment and similar reductions in the average wage; differences are not large.
- Data caveats and limitations:
  - Findings are correlations rather than causal relationships.
  - No information on expected savings from each measure; sample sizes are small and insufficient for statistical significance.
  - Public employment data limited (available for 9 out of 16 historical episodes); wage bill measures often use Eurostat Compensation of Employees.

### Box 1 — Structural Measures and Social Dialogue (summary of concrete measures and vehicles)
- Structural measures and examples:
  - Reassess employment adequacy: increase working hours (Portugal, 2009–13; Spain, 2011), reduce overtime (Greece, Ireland, 2009–12), close certain government agencies (Greece, 2009–12), reallocate positions across departments.
  - Outsource noncore functions where private provision is more efficient (United Kingdom, 1994): transport, mail, cleaning, maintenance.
  - Tighten link between pay and performance; simplify and decompress salary structure (Ireland, 2011; Latvia, 2008–10).
  - Spending reviews to audit functions and identify duplication (Canada, 1991; United Kingdom, 1994, 2010, 2013; Portugal, 2013).
- Vehicles to foster social dialogue:
  - Public information campaigns (Canada, 1991–92).
  - Consultation within public administration (Portugal, 2013).
  - Negotiations with unions to reach agreement on comprehensive reforms (Austria, 1996–97; Italy, 1993–95; the Netherlands, 1984–86; Ireland, 2010).
- Coding of social dialogue:
  - Episodes coded “high” or “low” social dialogue based on existence of actions intended to reach social consensus (open negotiations with unions, communication with private sector employers/trade unions, a public communication strategy, or other explicit efforts). Coding measures effort toward social consensus, not its achievement.

### Appendix Tables — Selected episode statistics (selected rows and key values preserved)
- Historical episodes — selected cumulative reductions over five years (share of potential GDP unless noted):
  - Austria, 1996–97: Wage bill reduction (% GDP): −2.35. Structural measure: Yes. Extent of social dialogue: High.
  - Belgium, 1982: Wage bill reduction (% GDP): −1.70. Structural measure: Yes. Extent of social dialogue: High.
  - Canada, 1991–92: Wage bill reduction (% GDP): −2.02. Structural measure: Yes. Extent of social dialogue: High.
  - Netherlands, 1981, 1983, 1984–86: Wage bill reduction (% GDP): −1.66. Structural measure: No. Extent of social dialogue: High.
  - Portugal, 2005–07: Wage bill reduction (% GDP): −1.37. Structural measure: Yes. Extent of social dialogue: Low.
  - United Kingdom, 1994: Wage bill reduction (% GDP): −0.92. Structural measure: Yes. Extent of social dialogue: Low.
  - Note: For Canada, Denmark, Spain, and Ireland (1982) the reduction is expressed as a share of nominal GDP because potential GDP was not available.
- Recent episodes — cumulative reductions from 2009 to 2012 (as a share of potential GDP):
  - Bosnia, 2012–13: Wage bill reduction (% pGDP): −1.02. Structural Measure: No. Extent of social dialogue: Low.
  - Bulgaria, 2010–13: Wage bill reduction (% pGDP): −0.95. Structural Measure: Minor. Extent of social dialogue: Low.
  - Greece, 2009–13: Wage bill reduction (% pGDP): −2.87. Structural Measure: Major. Extent of social dialogue: Low.
  - Ireland, 2009–13: Wage bill reduction (% pGDP): −1.12. Structural Measure: Major. Extent of social dialogue: High.
  - Latvia, 2009–12: Wage bill reduction (% pGDP): −1.62. Structural Measure: Major. Extent of social dialogue: Medium.
  - Portugal, 2011–13: Wage bill reduction (% pGDP): −3.02. Structural Measure: Major. Extent of social dialogue: High.
  - Romania, 2010–13: Wage bill reduction (% pGDP): −3.23. Structural Measure: Major. Extent of social dialogue: Medium.
  - United Kingdom, 2011–12: Wage bill reduction (% pGDP): −0.89. Structural Measure: No. Extent of social dialogue: Low.
  - United States, 2011–12: Wage bill reduction (% pGDP): −0.66. Structural Measure: No. Extent of social dialogue: Low.
  - Several entries list “...” where values are not reported in the source excerpt.

### VI. Conclusions and Policy Implications
- Methodological caveats:
  - Analysis based on a limited number of cases; coding of episodes entailed judgment.
  - Effectiveness assessed only in terms of reducing the public wage bill; effects on public sector efficiency or provision of public services were not addressed.
- Summary conclusions:
  - Structural measures and/or social dialogue support larger and more sustained reductions in the public wage bill.
  - Across-the-board measures, such as wage freezes, may bring temporary budget relief but are unlikely to be effective for longer-term consolidation needs.
  - Historical patterns (1979–2009) are broadly consistent with preliminary evidence from recent episodes (2009–present).
- Policy recommendations and implications:
  - Rely less on temporary and one-off measures and more on longer term structural changes guided by careful assessments of spending needs.
  - Bring public opinion on board and implement changes in a context of dialogue with important stakeholders to improve prospects for lasting consolidation.
  - Anticipate changes in expenditure composition (example: less expenditure on education and more on health due to aging) and plan structural reforms accordingly.

*Source: _wp14192 — References (PDF).*

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

### _wp14192 - References .............................................................................................................

### I. Introduction
- Goal: assess successes and failures of past public wage bill consolidations and draw lessons for post–Great Recession episodes.
- Coverage:
  - Historical episodes: 1979–2009.
  - Recent episodes: 2009–present.
  - Classification of measures by type: action on average wage and/or employment; one-off vs. structural; extent of consultations/social dialogue.
- Context:
  - Wage bill shares (2010 and text references):
    - As a share of total government expenditures: about 20–25 percent in AEs and EMEs, almost 30 percent in LICs.
    - As a share of GDP: about 10 percent in AEs, between 5 and 10 percent in EMEs and LICs.
    - Public employees represent about 7–8 percent of the total population in AEs, and about 5 percent in EMEs and LICs.
  - Government wage bill in health and education (IMF, April 2014 Fiscal Monitor): about 30 and 60 percent of government spending in health and education, respectively, in advanced economies.
- Main findings:
  - Structural measures vs. one-off measures:
    - Implementation of structural measures has been associated with larger and more sustained reductions in the public wage bill as a share of GDP.
  - Social dialogue:
    - Prior to the Great Recession, countries that created consensus and/or negotiated reductions via social dialogue achieved larger and more permanent reductions.
  - Historical vs. recent episodes:
    - Similar in many respects (frequency of targeting average wage vs. employment).
    - Main difference: less social dialogue in recent episodes.
  - Early evidence for recent episodes:
    - Structural measures again appear to produce larger and more sustained reductions, though it is too early for definitive statements.
- Structure of paper: Section II (trends), III (literature), IV (data and episodes), V (results), VI (conclusions).

### II. Trends in Public Sector Compensation and Employment
- Cross-country patterns (figures referenced):
  - Figure 1 (2010):
    - Public wage bill highest for advanced economies: around 10 percent of GDP in 2010.
    - Emerging market economies: about 6 percent of GDP.
    - Low income countries: about 7 percent of GDP.
    - Note: average wage ratio defined as average wage of general government divided by that of entire economy. Figures for general government employment refer to 2008 data.
  - Figure 2:
    - General government wage bill as a share of GDP and of general government expenditures:
      - Low income countries: wage bill as share of GDP has been on the rise over the past decade.
      - Emerging market economies: relatively stable expenditure on the public wage bill.
      - As a share of general government expenditures: low income countries tend to spend more but with volatility; emerging and advanced economies show similar shares with a tendency to decline over the past decade.
  - Figure 3:
    - Public employment as a share of total population is quite stable over time in advanced and emerging economies.
    - For low income countries, insufficient data to assess trends in public employment over time.
- Data sources:
  - Wage bill data: IMF World Economic Outlook database, Eurostat, AMECO, and occasionally IMF country teams.
  - Employment data: International Labour Organization; World Development Indicators.

### III. Literature Review (three broad strands)
- Civil service reforms (focus on emerging/developing economies):
  - Nunberg and Nellis (1995); World Bank and IMF (2002): comprehensive civil service reforms in 1980s–1990s with mixed success.
  - Three lessons:
    - Emergency measures (temporary wage/hiring freezes) tend to have only short-term effects.
    - Long-term reforms yield substantial results but are politically difficult to implement.
    - Successful reforms often included targeting by skills and age plus compensation packages aiding reallocation; such reforms could be financially costly.
  - Coordination needed between wage bill reforms and reforms in labor-intensive health and education; misuse of “wage bill ceilings” without effective payroll management may impede hiring in social sectors (Fedelino, Schwartz, and Verhoeven, 2006).
- Public wage premium:
  - Evidence of a significant public wage premium in many European countries, often higher at lower pay scales and for women; cross-country differences exist.
  - EMEs and LICs: they tend to spend less on the public wage bill as a share of GDP and have lower levels of public employment, leading to typically higher average public wages compared to per-capita GDP (comparisons are suggestive and do not control for composition/skills).
- Public–private labor market interactions:
  - Public compensation affects private compensation; effect stronger where wage bargaining centralization and union membership are high, weaker with high trade openness and wage indexation.
  - Openness to trade has increased over time, potentially reducing the effect of public-sector wages on private-sector wages.

### IV. Episodes of Public Employment and Compensation Reform
- Episode selection:
  - Two sets: historical (1979–2009) and recent (post-2009). Appendix Tables list country-year episodes.
  - Identified episodes: 16 historical and 23 recent episodes of wage bill consolidation.
- Identification and classification:
  - Episodes identified based on discretionary government measures aimed at reducing the public wage bill (primarily using IMF Article IV Staff Reports).
  - Possible outcomes:
    - Countries may take measures without reducing the wage bill (measures ineffective).
    - A reduction in wage bill share can occur for reasons outside discretionary measures (e.g., strong GDP growth).
- Classification dimensions:
  - Target of measures:
    - Average wage measures (most common: wage freeze, occasionally wage cut).
    - Employment measures (hiring freezes, attrition, abolishing positions, closing agencies).
  - Short-term vs. structural:
    - Short-term: one-off measures such as temporary wage freezes.
    - Structural: longer-term reforms aimed at efficiency (reforming bargaining model, compensation grid, wage indexation rules, introducing pay-for-performance, abolishing positions/closing agencies).
  - Sources used for episode documentation (in addition to Article IVs): OECD Public Sector Compensation in Times of Austerity (2012), IMF country teams, IMF Recent Economic Developments, Stability and Convergence Programmes, Eurostat employment and compensation data for selected episodes, ILO Laborsta for Canada and Ireland where necessary.

### V. Key empirical results (summary of findings reported)
- Structural measures are associated with:
  - Larger and more sustained reductions in the public wage bill as a share of GDP (text emphasis; figures in paper: Figures 5, 6, 9).
- Social dialogue:
  - Episodes with consensus-building/social dialogue achieved larger and more permanent reductions historically (Figures 7, 8, 10).
  - Recent episodes show less social dialogue than historical episodes.
- Historical vs. recent episodes:
  - Similarities in frequency of measures targeting average wage vs. employment.
  - Recent episodes: early evidence suggests structural measures again yield larger and sustained reductions, but conclusions are tentative given limited time horizon since measures were introduced.
- Data scope and limitations:
  - Wage bill measures refer to general government Compensations of Employees (Eurostat) for many European episodes.
  - General government employment data for some countries derived from Eurostat NACE Rev. 1.1 categories L, M, N; exceptions for Canada and Ireland using ILO Laborsta.

*Source: _wp14192 - References (PDF). *

### Box 1. Structural Measures and Social Dialogue

### Box 1. Structural Measures and Social Dialogue

### Structural measures on public wage formation and hiring
- Reassessing employment level adequacy might require increasing working hours (Portugal, 2009–13, Spain, 2011) or reducing overtime, depending on needs (Greece, Ireland, 2009–12); closing certain government agencies (Greece, 2009–12); or reallocating positions across departments.  
- Outsourcing noncore functions if these functions can be provided more efficiently by the private sector (United Kingdom, 1994). Noncore functions such as transport, mail, cleaning, and maintenance are potential candidates for outsourcing.  
- Tightening the link between pay and performance can help connect wage increases to individual and organizational performance and create incentives to improve efficiency and productivity (Ireland, 2011). Linking pay to performance might also require simplifying and decompressing the salary structure (Latvia, 2008–10).  
- Spending reviews are organizational audits of the functions of government agencies that attempt to determine the optimal organizational arrangement for government tasks. They provide the basis for defining specific measures to be implemented and can identify areas of duplication and overlap. Such reviews have been used in Canada in 1991; the United Kingdom in 1994, 2010, and 2013; and Portugal in 2013, among others.

### Vehicles to foster social dialogue
- Public information campaigns to inform employees early on of the detailed administrative and financial aspects of the reform and to make the general public aware of reform objectives (Canada, 1991–92).  
- Consultation within the public administration to allow staff from different parts of the administration to be involved in the reform effort (Portugal, 2013).  
- Negotiations with unions to help the government and social partners reach agreement on comprehensive reforms (Austria, 1996–97; Italy, 1993–95; the Netherlands, 1984–86; Ireland, 2010).

### Coding and interpretation of social dialogue
- Episodes are recorded as having “high” or “low” level of social dialogue based on the existence of actions intended to reach social consensus (open negotiations with unions, communication with private sector employers/trade unions, a public communication strategy, or other explicit efforts).  
- This coding is a measure of the existence of effort toward social consensus, not of actual social consensus achieved.

### Historical episodes (summary)
- Dataset: 16 historical episodes of wage bill consolidation identified from Devries and others (2011), covering 17 OECD economies from 1979 to 2009.  
- Contexts prompting measures included: rein in inflation (run-up to euro adoption in the mid 1990s: Austria, Belgium, Germany, Italy, Spain), response to the ERM 1992 crisis (Italy, the United Kingdom), and stagflation from the oil shocks in the late seventies and early eighties (Belgium, Germany, Denmark, the Netherlands).  
- Some episodes tied to country-specific imbalances (devaluation episodes in Ireland in the late 1980s).  
- Only in a few cases (Portugal and Ireland in the 2000s) were wage bill measures explicitly linked to the gap between public and private sector compensation.

### Recent episodes (summary)
- Dataset: 23 recent episodes of wage bill consolidation; all started in 2009 or after.  
- Sources: IMF staff information for European IMF member countries and OECD (Public Sector Compensation in Times of Austerity, 2012); structural measures and extent of social dialogue checked against IMF Staff Reports for Article IV consultations.  
- Variation across countries included initial size of the public sector; Greece, Ireland, Italy, Portugal, and Spain entered the Great Recession with significantly higher public wage premia than other European countries (see Giordano and others, 2011).

### Results and stylized facts: comparison of historical and recent episodes
- Historical episodes: 16. Recent episodes: 23.  
- Similarities: historical and recent episodes share many characteristics across measures taken.  
- Key statistics and observations:
  - The most common measure is a temporary wage freeze; all the recent episodes included a wage freeze, or a wage cut.  
  - In historical episodes, over 80 percent of episodes included wage freezes.  
  - Between 50 and 60 percent of episodes also included measures on the employment level (employment measures such as attrition rules or reduced replacement rates for retired workers typically take time to produce results).  
  - Example of sustained impact: the Netherlands in the 1980s achieved a substantial and sustained reduction in the wage bill (1.7 percent of GDP) primarily through negotiated and renewed wage freezes.  
  - Average reduction in the public wage bill is similar for historical and recent episodes: about 1 percent of potential GDP.  
    - For historical episodes the wage bill reduction is the cumulative reduction calculated over five years after the first year of measures.  
    - For recent episodes the wage bill reduction is calculated as the cumulative reduction from 2009 to 2012.  
  - The similar average reduction (about 1 percent of potential GDP) despite the shorter time period for recent episodes reflects substantial recent efforts and indicates the general scope for public wage bill reduction and its contribution to deficit reduction.

*Source: _wp14192 - Box 1. Structural Measures and Social Dialogue*

### Appendix Tables 1–2.

### Appendix Tables 1–2.

### B. What Types of Measures are Associated with More Sustained Wage Bill Reductions Among Historical Episodes?
- Main finding:
  - Countries that accomplished a sustained reduction in the wage bill tended to rely more on structural measures and/or extensive social dialogue.
- Table 1 (Structural Measures and Social Dialogue Across Episodes):
  - High social dialogue, Structural Yes: 4
  - High social dialogue, Structural No: 7
  - High social dialogue, Total: 11
  - Low social dialogue, Structural Yes: 2
  - Low social dialogue, Structural No: 3
  - Low social dialogue, Total: 5
  - Total Structural Yes: 6
  - Total Structural No: 10
  - Total episodes: 16
- Notes on wage bill measure:
  - The wage bill measure used is Compensation of Employees from Eurostat, which includes social contributions paid by the employer and the employees and the gross take home pay of the employee.
  - Example caveat: Ireland 2009 introduced an additional levy on the gross wage of public employees of seven percent on average (called Pension Related Deduction) that is not included in social contributions and not counted in the computation of pension benefits.
- Interpretation of Table 1:
  - There was substantial variation in the mix between structural wage bill measures and degree of social dialogue.
  - The share of episodes with structural measures is almost identical for high- and low-dialogue cases.
  - Episodes with extensive social dialogue sometimes achieved long-term wage bill reduction relying on repeated non-structural measures.
- Figure 5 (Structural Versus Non-Structural Measures: Cumulative Change in the Public Wage Bill Ten Years After the First Year of Measures, Percent of Potential GDP):
  - When adjustment relied on structural measures the containment of the public wage bill has been larger and more sustained.
  - Difference in reduction of the public wage bill as a share of GDP is about 0.5 percent of GDP after four years.
  - Over seven to ten years, the difference can be as large as about 0.7 percent of GDP.
  - These values come from simple averages across two groups; small sample sizes preclude precise confidence intervals—treated as stylized facts.
- Decomposition of wage bill into components:
  - Components: (1) average public wage component (average wage divided by GDP per capita); (2) public employment component (public employment divided by total population).
  - Figure 6 findings:
    - Episodes with structural measures had a smaller increase in the public employment level compared to episodes with non-structural measures.
    - The average reduction in the average wage was not very different between episodes with and without structural measures; in years six to 10 the average wage (as a percentage of per capita GDP) appears to fall less for episodes with structural measures.
    - Caution: data availability for public employment is limited (public employment data available for 9 out of the 16 episodes).
  - Structural measures can target average wage, employment level, or both; episodes coded as structural often included a combination.
- Figure 7 (Social Dialogue Versus Little Social Dialogue: Cumulative Change in the Public Wage Bill Ten Years after the First Year of Measures, Percent of Potential GDP):
  - Episodes with substantial social dialogue typically saw larger and more sustained reductions in the wage bill.
  - Difference across episodes with high versus little social dialogue is about 0.5 percent of GDP after about five years.
  - Over seven to ten years the difference can be as large as about 1 percent of GDP or more.
  - Examples:
    - Austria 1996/97: authorities consulted social partners extensively and secured a lasting agreement (contrast with failed 1995 attempt).
    - Canada: initial measures in early 1990s consolidated with a comprehensive review of federal spending culminating in “long-lasting structural change” and “fundamental change in how the government delivers programs and services” (1995 Budget).
    - Recent efforts: Portugal and Ireland aimed to deepen social dialogue and involve public employees and the public in public expenditure reviews.
- Decomposition for social dialogue episodes:
  - Episodes with social dialogue tend to have lower increases in public employment and similar reductions in the average wage; differences are not large.
- Overall caveats for Section B:
  - Findings are correlations rather than causal relationships.
  - No information on expected savings from each measure; countries relying more on wage bill reductions may have chosen more structural measures or dialogue.
  - Sample sizes are very small, insufficient for statistical significance.
  - Results are preliminary and intended to open avenues for future research.

### C. Preliminary Results on Recent Episodes
- Scope:
  - Recent episodes defined as countries that introduced measures in 2009 or 2010 (see Appendix Table 2).
  - Wage bill reductions calculated as percent of potential GDP from first year of measures; wage bill data available through 2013, so reductions analyzed over three years starting from 2009 or 2010.
- Figure 9 (Structural versus Non-Structural Measures, Recent Episodes):
  - Structural measures in recent episodes are associated with reductions of close to 1 percent of potential GDP already in the first year after measures were taken.
  - Interpretation: quicker reduction than in past episodes, possibly reflecting crisis-driven stronger and more urgent action.
- Figure 10 (Social Dialogue versus Little Social Dialogue, Recent Episodes):
  - Much less social dialogue in recent episodes, perhaps due to lack of time to develop it.
  - Difference in wage bill reductions between countries with and without social dialogue is about 0.25 percent of GDP over the three-year window.
  - More time is needed to make conclusive assessments of effectiveness and shortcomings of recent measures.

### VI. CONCLUSION (Key conclusions and policy implications)
- Methodological caveats:
  - Analysis based on a limited number of cases; coding of episodes entailed judgment.
  - Effectiveness assessed only in terms of reducing the public wage bill; effects on public sector efficiency or provision of public services were not addressed.
- Summary of results:
  - Structural measures and/or social dialogue support larger and more sustained reductions in the public wage bill.
  - Across-the-board measures, such as wage freezes, may bring temporary budget relief but are unlikely to be effective for longer-term consolidation needs.
  - This pattern held in historical episodes (1979–2009) and appears in recent episodes as well.
- Policy implications:
  - As fiscal consolidation progresses, governments will have to accommodate changes in composition of public expenditures (example: less expenditure on education and more on health due to aging).
  - Need to rely less on temporary and one-off measures and more on longer term structural changes guided by careful assessments of spending needs.
  - Bringing public opinion on board and implementing changes in a context of dialogue with important stakeholders might prove essential.

### Appendix Table 1. Historical Episodes of Wage Bill Reduction (1979–2009) — Selected data (cumulative reduction over five years following first year of measures, as a share of potential GDP unless noted)
- Austria, 1996–97: Reason: meet Maastricht targets; improve competitiveness. Wage measure: Yes. Staff measure: Yes. Structural measure: Yes. Extent of social dialogue: High. Wage bill reduction (% GDP): −2.35
- Belgium, 1982: Reason: improve competitiveness. Wage measure: Yes. Staff measure: No. Structural measure: Yes. Extent of social dialogue: High. Wage bill reduction (% GDP): −1.70
- Belgium, 1992, 1994: Reason: meet Maastricht targets. Wage measure: Yes. Staff measure: Yes. Structural measure: No. Extent of social dialogue: High. Wage bill reduction (% GDP): −0.23
- Canada, 1991–92: Reason: reduce fiscal deficit. Wage measure: Yes. Staff measure: No. Structural measure: Yes. Extent of social dialogue: High. Wage bill reduction (% GDP): −2.02
- Denmark, 1983–84: Reason: improve competitiveness. Wage measure: Yes. Staff measure: No. Structural measure: No. Extent of social dialogue: Medium. Wage bill reduction (% GDP): −0.80
- Germany, 1983–84: Reason: economy wide wage moderation. Wage measure: Yes. Staff measure: No. Structural measure: No. Extent of social dialogue: Medium. Wage bill reduction (% GDP): ...
- Germany, 1995, 1997, 2000: Reason: meet Maastricht targets; increase productivity. Wage measure: Yes. Staff measure: Yes. Structural measure: No. Extent of social dialogue: Medium. Wage bill reduction (% GDP): −0.33
- Ireland, 1982: Reason: reduce fiscal deficit. Wage measure: No. Staff measure: Yes. Structural measure: No. Extent of social dialogue: Low. Wage bill reduction (% GDP): −0.65
- Ireland, 1987–88: Reason: reduce fiscal deficit. Wage measure: Yes. Staff measure: Yes. Structural measure: No. Extent of social dialogue: Medium. Wage bill reduction (% GDP): 0.07
- Italy, 1993, 1995: Reason: reduce fiscal deficit; macro adjustment after exit from ERM. Wage measure: Yes. Staff measure: Yes. Structural measure: Yes. Extent of social dialogue: High. Wage bill reduction (% GDP): −1.16
- Netherlands, 1981, 1983, 1984–86: Reason: reduce fiscal deficits. Wage measure: Yes. Staff measure: No. Structural measure: No. Extent of social dialogue: High. Wage bill reduction (% GDP): −1.66
- Netherlands, 2005: Reason: reduce fiscal deficit; improve competitiveness. Wage measure: Yes. Staff measure: No. Structural measure: No. Extent of social dialogue: Low. Wage bill reduction (% GDP): 0.23
- Portugal, 2000, 2003: Reason: reduce deficit to SGP target; high public employment. Wage measure: No. Staff measure: Yes. Structural measure: No. Extent of social dialogue: Low. Wage bill reduction (% GDP): −0.20
- Portugal, 2005–07: Reason: reduce deficit to SGP's target; high pub. wage premium; high pub. employment. Wage measure: Yes. Staff measure: Yes. Structural measure: Yes. Extent of social dialogue: Low. Wage bill reduction (% GDP): −1.37
- Spain, 1997: Reason: meet Maastricht targets. Wage measure: Yes. Staff measure: Yes. Structural measure: No. Extent of social dialogue: Medium. Wage bill reduction (% GDP): −0.80
- United Kingdom, 1994: Reason: macro adjustment after exit from ERM. Wage measure: Yes. Staff measure: Yes. Structural measure: Yes. Extent of social dialogue: Low. Wage bill reduction (% GDP): −0.92
- Note: For Canada, Denmark, Spain, and Ireland (1982) the reduction is expressed as a share of nominal GDP because potential GDP was not available.

### Appendix Table 2. Recent Episodes of Wage Bill Reduction (2009–13) — Selected data (cumulative reduction from 2009 to 2012, as a share of potential GDP)
- Bosnia, 2012–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −1.02
- Bulgaria, 2010–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Minor. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.95
- Croatia, 2009–13: Wage measure: Yes. Staff measure: No. Structural Measure: Medium. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.06
- Cyprus, 2013: Wage measure: Yes. Staff measure: Yes. Structural Measure: Medium. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.66
- Czech Republic, 2010: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.57
- France, 2011–12: Wage measure: Yes. Staff measure: Yes. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.13
- Greece, 2009–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Major. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −2.87
- Hungary, 2010–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Medium. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −1.48
- Ireland, 2009–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Major. Extent of social dialogue: High. Wage bill reduction (% pGDP): −1.12
- Italy, 2009–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Minor. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.64
- Japan, 2012: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): ...
- Latvia, 2009–12: Wage measure: Yes. Staff measure: Yes. Structural Measure: Major. Extent of social dialogue: Medium. Wage bill reduction (% pGDP): −1.62
- Macedonia, 2009–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Minor. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.60
- Netherlands, 2012: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.38
- Poland, 2009–13: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.54
- Portugal, 2011–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Major. Extent of social dialogue: High. Wage bill reduction (% pGDP): −3.02
- Romania, 2010–13: Wage measure: Yes. Staff measure: Yes. Structural Measure: Major. Extent of social dialogue: Medium. Wage bill reduction (% pGDP): −3.23
- Serbia, 2009–10, 2012–13: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.09
- Slovakia, 2013: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.51
- Slovenia, 2010, 2012–13: Wage measure: Yes. Staff measure: No. Structural Measure: Minor. Extent of social dialogue: Medium. Wage bill reduction (% pGDP): −0.63
- Spain, 2010–12: Wage measure: Yes. Staff measure: No. Structural Measure: Minor. Extent of social dialogue: Medium. Wage bill reduction (% pGDP): −0.86
- United Kingdom, 2011–12: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.89
- United States, 2011–12: Wage measure: Yes. Staff measure: No. Structural Measure: No. Extent of social dialogue: Low. Wage bill reduction (% pGDP): −0.66

*Source: Author’s calculations and IMF staff reports as presented in the Appendix Tables 1–2.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2014/_wp14192.pdf_