## 1. Average Annual Domestic Inflation

## Source details

**Canonical URL:** [1. Average Annual Domestic Inflation](https://www.imf.org/-/media/files/publications/cr/2023/english/1luxea2023002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2023/english/1luxea2023002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2023/english/1luxea2023002.pdf.json)

---

### Introduction and context
- Automatic wage indexation (AWI, “Den Index”) has been universal in Luxembourg since 1975 (introduced for some categories in the 1920s).
- AWI became contested during the surge in inflation after Russia’s invasion of Ukraine in February 2022; the government postponed the index tranche in 2022 and the March 2022 Tripartite agreed on additional price controls, subsidies, and temporary/partial compensatory measures to firms.
- The note summarizes literature and evidence on AWI’s effects in Luxembourg and proposes reform avenues to increase resilience to supply shocks.

### Mechanics and coverage of Luxembourg’s AWI
- Trigger rule:
  - Wages are increased by 2.5 percent in the month after the 6-month moving average of the national CPI (NICP) exceeds by 2.5 percent the threshold (cote d’échéance).
- Two indexes:
  - cote d’échéance (signals when benchmark exceeded and tranche is due).
  - cote d’application (records when tranche is applied and wages adjusted).
- Coverage:
  - Applies to salaries and wages in both private and public sectors except workers posted to Luxembourg.
  - Pensions and minimum wages are automatically indexed and also separately increased to keep track of real wages.
  - Many social benefits are automatically indexed, notably family benefits (allocations familiales).
  - Law does not prescribe indexation for prices of goods/services or for hourly rates of professionals and self-employed, but many are indexed in practice (example: rates of medical consultations).

### Historical modulations and recent policy actions
- Modulations and postponements:
  - 1980s: some adjustments compensated less than fully for inflation or were skipped.
  - 2006–2009: indexation tranches were postponed, producing several months of delay between trigger and disbursement.
  - 2010–2013: only one indexation per year allowed, creating months of uncompensated delay.
  - March 2022 Tripartite agreement: provision to postpone the second indexation tranche for 2022 to April 2023.
- Practical separation of trigger and application (cote d’échéance vs cote d’application) allows government modulations (e.g., adjusting the cote d’application less than full compensation).

### Cross-country comparison and recent trends
- In the euro area:
  - Automatic indexation is most prevalent for pensions; less common for private and public sector wages.
  - Countries with some automatic indexation for wages/pensions include Luxembourg, Belgium, Cyprus, and Malta.
    - Luxembourg and Belgium: proportional increase fully compensates for inflation.
    - Cyprus: wages automatically increased by 50% of the rate of inflation.
    - Malta: inflation adjustment paid as a fixed Euro amount for all workers.
- Coverage of employees by automatic indexation across the euro area has fallen since the global financial crisis (GFC).
- Recent tendencies include exclusion of volatile components (e.g., energy) from benchmarks and moves away from backward-looking inflation benchmarks.
- Comparative inflation statistics (select observations):
  - Average NICP inflation 2002–2019: 1.9 percent; volatility: 1.0 percent.
  - NICP inflation 2020–2022: 3.2 percent; volatility: 2.5 percent.
- Note: these comparisons are presented as suggestive correlations, not causal claims.

### Potential drawbacks and macroeconomic effects
- Wage rigidity:
  - Microdata 2001–2006 show spikes of wage changes exactly at 2.5 percent in years with an indexation tranche and at 0 percent in years without, implying near-full downward real wage rigidity.
- Supply shocks and amplification:
  - Theoretical models indicate that indexation amplifies volatility from supply-side shocks (inflation and real wages move in opposite directions) and can reduce volatility from demand-side shocks (they move together).
  - AWI can increase volatility and competitiveness losses following supply shocks such as the 2022 energy crisis.
- Second-round effects and wage-price spiral risk:
  - Applying AWI during rapid inflation could lead to multiple indexation tranches in a short span, with potential harm to competitiveness.

### Empirical estimates and key quantitative findings
- IMF staff estimates (with lead and lagged effects):
  - Each indexation tranche generates an additional 0.7 percentage points in headline inflation.
  - Each indexation tranche generates an additional 1.2 percentage point for inflation in services.
  - If 4-5 tranches occurred, the additional impact on inflation would range between 2.8-3.5 percentage points.
- Alternative estimate (STATEC, 2017):
  - Each indexation tranche estimated impact of 0.2 percentage point on inflation.
- Caveats:
  - Impacts could be larger in a high inflation environment.
  - Pass-through from wages to prices is likely positively correlated with increases in firms’ personnel costs.

### Distributional considerations
- AWI protects purchasing power broadly (salaries, pensions, family benefits), supporting social peace, but effects are heterogeneous across sectors and income groups.
- Supply shocks concentrate on a few internationally traded commodities (oil, gas, food) and affect households differently across the income distribution.
- For the 2022 energy price shock, despite March 2022 Tripartite price control measures, lower income quintiles were disproportionately impacted.
- Consumption equity considerations:
  - High income individuals/households likely benefit relatively more from indexation than middle-class earners because of higher propensity to save (lower marginal propensity to consume).
  - Personal income tax rises relatively quickly to top marginal tax rates; for most middle-class households, the marginal tax rate is approximately the same as that applying to those at the very top of the income distribution.
  - After a few rounds of indexation, this may put further pressure on real estate and other assets.

- Table of before/after tax salary increase following an indexation event (as presented):
  - €   20,000.00 — Marginal tax rate 12% — Gross index €    500.00 — Net index € 440.00 — Effective index 2.20%
  - €   38,700.00 — Marginal tax rate 32% — Gross index €    967.50 — Net index € 657.90 — Effective index 1.70%
  - €   58,000.00 — Marginal tax rate 39% — Gross index € 1,450.00 — Net index € 884.50 — Effective index 1.53%
  - € 116,000.00 — Marginal tax rate 40% — Gross index € 2,900.00 — Net index € 1,740.00 — Effective index 1.50%
  - € 155,000.00 — Marginal tax rate 41% — Gross index € 3,875.00 — Net index € 2,286.25 — Effective index 1.48%
  - € 205,000.00 — Marginal tax rate 42% — Gross index € 5,125.00 — Net index € 2,972.50 — Effective index 1.45%

### Structural impacts: productivity and competitiveness
- Labor productivity:
  - Luxembourg has very high levels of labor productivity in some sectors, but apparent labor productivity has stagnated since the GFC relative to main trading partners (France, Germany, United States).
- Relative unit labor costs (ULC):
  - Relative ULCs compared to the euro area have worsened over the last three decades, particularly in financial services and in professional activities.
  - These trends accompanied an increase in the share of intermediate consumption of gross output, possibly signaling deeper integration of services in global value chains and outsourcing by some firms.
- Compensation vs productivity:
  - On aggregate, real hourly compensation broadly matched productivity during 1995-2021.
  - Over 2010-2021, compensation outpaced productivity growth, more pronounced in the services sector; manufacturing, construction, and transportation showed the reverse in some cases.

### Firm adjustments and sectoral responses to indexation
- Firms adjust along several margins: base pay change, variable pay change, employment change, and a component capturing indexation (same across sectors).
- Correlations (2010-2021 annual data):
  - Indexation events tend to be associated with lower employment or lower base/variable pay growth in general.
  - Manufacturing, construction, and administrative and support services tend to exhibit slower employment growth following indexation.
  - Domestically oriented sectors such as retail trade may see employment gains.
- Scenario:
  - Several index tranches expected over 2022-2024 could raise base pay by up to 12.5 percent, if left unmodulated.

### Costs and benefits of a flexible approach to AWI
- Modulations of AWI:
  - Lower the magnitude of second-round effects and the likelihood of a wage-price spiral.
  - If AWI operated fully automatically, it could raise firms’ inflation expectations and their incentive to raise prices, producing higher second-round effects.
  - Endogenous firm responses may mute headline inflation outcomes but not necessarily employment impacts.
- Lack of rules about suspension increases uncertainty since ad-hoc modulation or fiscal compensation is politically driven during supply-side shocks.
  - Since February 2022, three Tripartite meetings addressed the energy price shock and AWI; measures included untargeted price caps/controls for energy goods, postponement of the second round of indexation for 2022 to April 2023, and a subsidy to pay for the cost of a second indexation for 2023.
  - Upcoming election cycle in the second half of 2023 increased pressure for swift measures.
- Fiscal and political constraints:
  - Practical implementation of automatic indexation hinges on political will and could erode fiscal space.
  - Trend toward more costly fiscal measures to compensate households and firms for modulation increases fiscal risk, especially given large aging-related costs.

### Directions for reform — measures to increase resilience and equity
- Adopt a measure of core inflation as a more resilient benchmark for automatic indexation:
  - Many systems exclude volatile components from the inflation benchmark (example: Belgium’s Health Index excludes alcohol, tobacco, and energy; Italy uses an inflation projection excluding energy for wage negotiations).
  - Empirical stats for 2002M1-2023M1:
    - Headline inflation averaged 2.1 percent with a standard deviation of 1.4.
    - Core inflation averaged 1.9 percent with a standard deviation of 0.8.
  - Excluding Covid-19 and energy shocks (2002M1-2019M12):
    - Headline and core inflation averages about 1.9 and 1.8 percent, respectively.
    - Core’s standard deviation half that of headline (0.5 versus 1.0).
  - Indexing to core inflation:
    - Reduces likelihood of additional indexation following an energy price shock and produces correspondingly lower second-round effects.
    - DSGE model simulation (small open economy, downward nominal wage rigidity, indexation rule: all wages increased by 2.5 percent in nominal terms as soon as lagged cumulative inflation exceeds 2.5 percent since last indexation) shows that for the same shock size, indexing to core inflation does not lead to an index tranche while indexing to headline does.
    - Short-run: both core and headline inflation end up much lower because of the missing index tranche under core indexing.
    - Long-run: effects are much smaller and align with relatively small empirical estimates of STATEC (2017) for second-round effects.
  - Distributional ramifications:
    - Excluding components (energy, perhaps food) that weigh heavily on lower-income households can reduce volatility but may worsen distributional outcomes immediately after shocks.
    - Such distributional effects could be addressed by targeted, temporary support measures whose fiscal cost would be much lower than broad-based measures.

- Introduce progressivity in wage adjustments:
  - Practical design: set a threshold above which implied wage change less than fully matches inflation.
    - Example: individuals below threshold receive proportional compensation fully matching inflation (e.g., 2.5 percent); individuals above threshold receive a wage increase equal to 2.5 percent of the threshold itself (partial compensation).
  - Implementation issues:
    - Whether threshold applies to individual wages or total household income (many households file taxes jointly).
    - Political decision about where to set the threshold requires distributional analysis.
  - Concerns and mitigating factors:
    - Progressive system could lead to wage compression, but indexation is only one driver of wage growth.
    - Decomposition of annualized average wage growth (2010/Q1–2022/Q3) shows variable wage components routinely occur in addition to indexation and can be positive or negative.
    - Reduced automatic benefit for high-income earners may lead them to bargain for higher wages with employers, easing the cost of automatic indexation on firms.
  - Precedent: Malta’s cost-of-living allowance (COLA) is a fixed monetary benefit given equally to all workers; proposed Luxembourg system would mix Malta features with current AWI.

- Move toward a rule-based and sectoral approach to modulations:
  - Implement explicit quantitative criteria for triggering or suspending wage indexation to reduce uncertainty.
  - Possible rule designs:
    - Prescribe the maximum allowed number of indexations in a certain period, with excess tranches automatically postponed.
    - Track quantitative measures of cost-competitiveness (e.g., relative ULCs); when exceeded, automatically trigger suspension and/or talks between unions and employers (as in Belgium).

### Decentralization of wage-setting: rationale and mechanism
- Universal automatic indexation benefits all workers, pensioners, and recipients of public benefits that are automatically adjusted for inflation.
- Proposed reform: define competitiveness benchmarks (measured in neighboring countries and across sectors) and continually evaluate whether Luxembourg is losing competitiveness.
- Trigger mechanism:
  - If competitiveness benchmarks are crossed, suspend automatic wage indexation and trigger more decentralized bargaining.
- Expected benefits:
  - Lessen competitiveness concerns at the national level that often induce suspension of the system for all.
  - Give unions and employers more leeway to agree on the most sustainable way to split the costs of shocks, which could save jobs.
  - Reduce instances where system-wide suspension is used as the only instrument to address sector-specific pressures.

### Conclusions on resilience and distributional effects of automatic indexation
- AWI is a distinctive feature of Luxembourg’s labor market institutions, but recent events suggest its resilience to shocks could be improved.
- The system is routinely suspended during times of high inflation or economic slowdowns; most recently in 2022 when the second index tranche was postponed to April 2023.
- Common reasons for suspension:
  - Competitiveness grounds and concerns about wage-price spirals.
- Political economy responses have sometimes included increased fiscal spending to mitigate the cost of indexation to firms (examples: the 2022 and 2023 Tripartite agreements).
- Suspension of indexation may increase uncertainty and disproportionately hurt the most vulnerable parts of the population.

### Empirical analysis and model evidence
- Regression analysis (quarter-on-quarter, seasonally adjusted) for 2000–21 assessed effects of indexation events on headline, core, and services inflation.
  - Key variable: dummy = 1 if an indexation event took place in a quarter, 0 otherwise.
  - Controls included GDP, inflation of petroleum products, euro area inflation.
- Selected regression coefficient estimates (Table I.1):
  - Index tranche (+1): Headline 0.11**, Core 0.14***, Services 0.20***
  - Index tranche: Headline 0.20***, Core 0.25***, Services 0.42***
  - Index tranche (-1): Headline 0.25***, Core 0.21***, Services 0.31***
  - Index tranche (-2): Headline 0.16***, Core 0.14***, Services 0.29***
  - EA Inflation: Headline 0.52***, Core 0.19***, Services 0.25***
  - 2015Q1 dummy: Headline 0.52***, Core 0.74***, Services 0.65***
  - Petroleum Products: 0.03*** (headline)
  - Sample and fit statistics:
    - N: 88, 88, 88
    - R-squared: 0.88, 0.61, 0.48
    - DW: 2.00, 2.12, 2.15
  - Note: ***, **, * denote significance at 1 percent, 5 percent, and 10 percent respectively.

- Small open economy DSGE model simulations:
  - Medium-scale New Keynesian small open economy model with price and wage rigidity, imperfect passthrough of imported inflation, and downward nominal wage rigidity.
  - Imported good interpreted as a generic energy commodity whose price is the main shock.
  - Indexation rule modeled: wage adjusted by 2.5 percent only when lagged cumulated inflation rate exceeds 2.5 percent, then cumulated inflation variable is reset.
  - Distinction between headline and core indexation:
    - Headline: cumulate NICP index.
    - Core: cumulate price index of domestic goods.
  - Simulation outcomes:
    - For many shock sizes, no case triggered indexation when indexing to core but triggered when indexing to headline for the same shock size.
    - In a realistic shock (similar to 2022 energy shock), when indexation occurs under headline but not under core:
      - Additional wage increases weigh on short-term consumption and investment.
      - Core inflation spikes significantly within 8 quarters due to higher wage costs under headline indexing; medium-to-long term impact on core inflation is small.
      - Headline inflation increases significantly under both rules, somewhat more when headline indexing applied.
      - After about 6 quarters, investment recovers and inflation abates.

*Source: IMF staff note “Automatic wage indexation in Luxembourg: Limitations and Options for Reforms”, May 1, 2023.*

### 1. Average Annual Domestic Inflation  __________________________________________________ 4

### 1. Average Annual Domestic Inflation

### Introduction and context
- Automatic wage indexation (AWI, “Den Index”) is a long-standing labor market institution in Luxembourg; first introduced for some categories of workers in the 1920s and made universal in 1975.
- AWI became contested during the surge in inflation after Russia’s invasion of Ukraine in February 2022; the government postponed the index tranche in 2022 and the Tripartite in September 2022 agreed on additional price controls, subsidies, and temporary/partial compensatory measures to firms.
- The note summarizes literature and evidence on AWI’s effects in Luxembourg and proposes reform avenues to increase resilience to supply shocks.

### Mechanics and coverage of Luxembourg’s AWI
- Trigger rule: wages are increased by 2.5 percent in the month after the 6-month moving average of the national CPI (NICP) exceeds by 2.5 percent the threshold (cote d’échéance).
- Two indexes: cote d’échéance (signals when benchmark exceeded and tranche is due) and cote d’application (records when tranche is applied and wages adjusted).
- Coverage:
  - Applies to salaries and wages in both private and public sectors except workers posted to Luxembourg.
  - Pensions and minimum wages are automatically indexed and also separately increased to keep track of real wages.
  - Many social benefits are automatically indexed, notably family benefits (allocations familiales).
  - While law does not prescribe indexation for prices of goods/services or for hourly rates of professionals and self-employed, many of those are indexed in practice (example: rates of medical consultations).

### Historical modulations and recent policy actions
- Modulations and postponements have been used repeatedly:
  - 1980s: some adjustments compensated less than fully for inflation or were skipped.
  - 2006–2009: indexation tranches were postponed, producing several months of delay between trigger and disbursement.
  - 2010–2013: only one indexation per year allowed, creating months of uncompensated delay.
  - March 2022 Tripartite agreement: provision to postpone the second indexation tranche for 2022 to April 2023.
- The practical separation of trigger and application (cote d’échéance vs cote d’application) allows government modulations (e.g., adjusting the cote d’application less than full compensation).

### Cross-country comparison and recent trends
- In the euro area:
  - Automatic indexation is most prevalent for pensions; less common for private and public sector wages.
  - Countries with some automatic indexation for wages/pensions include Luxembourg, Belgium, Cyprus, and Malta.
    - Luxembourg and Belgium: proportional increase fully compensates for inflation.
    - Cyprus: wages automatically increased by 50% of the rate of inflation.
    - Malta: inflation adjustment paid as a fixed Euro amount for all workers.
- Coverage of employees by automatic indexation across the euro area has fallen since the global financial crisis (GFC); recent tendencies include exclusion of volatile components (e.g., energy) from benchmarks and moves away from backward-looking inflation benchmarks.
- Comparative inflation statistics (select observations from the note):
  - Average NICP inflation 2002–2019: 1.9 percent; volatility: 1.0 percent.
  - NICP inflation 2020–2022: 3.2 percent; volatility: 2.5 percent.
  - Note: these comparisons are presented as suggestive correlations, not causal claims.

### Potential drawbacks and macroeconomic effects
- Wage rigidity:
  - Evidence of a high degree of downward real wage rigidity in Luxembourg (microdata 2001–2006): spikes of wage changes exactly at 2.5 percent in years with an indexation tranche and at 0 percent in years without, implying near-full downward real wage rigidity.
- Supply shocks and amplification:
  - Theoretical models indicate that indexation amplifies volatility from supply-side shocks (where inflation and real wages move in opposite directions) and can reduce volatility from demand-side shocks (where they move together).
  - AWI can thus increase volatility and competitiveness losses following supply shocks such as the 2022 energy crisis.
- Second-round effects and wage-price spiral risk:
  - Applying AWI during rapid inflation would lead to multiple indexation tranches in a short span, with potential harm to competitiveness.

### Empirical estimates and key quantitative findings
- Estimated pass-through of indexation to inflation (IMF staff estimates, with lead and lagged effects):
  - Each indexation tranche generates an additional 0.7 percentage points in headline inflation.
  - Each indexation tranche generates an additional 1.2 percentage point for inflation in services.
  - If 4-5 tranches occurred (as could have happened if price signals had been allowed to fully operate in 2022–23), the additional impact on inflation would range between 2.8-3.5 percentage points.
- Alternative estimate (STATEC, 2017): each indexation tranche had a much smaller estimated impact of 0.2 percentage point on inflation.
- The note cautions that impacts could be larger in a high inflation environment and that pass-through from wages to prices is likely positively correlated with increases in firms’ personnel costs.

### Distributional considerations
- AWI protects purchasing power broadly (salaries, pensions, family benefits), supporting social peace, but can have heterogeneous effects across sectors and income groups (distributional analysis is a theme in the note and annexes).

### Proposed reform avenues (summary of options discussed)
- Change the inflation benchmark:
  - Move from headline inflation (NICP) to core inflation that excludes volatile components such as energy prices.
- Introduce progressive considerations:
  - Adjust indexation to incorporate progressivity so that lower-income households are relatively better protected while reducing overall inflationary pass-through.
- Adopt a rules-based suspension:
  - Replace ad-hoc suspensions with a transparent, rules-based suspension mechanism to improve predictability and credibility.

*Source: IMF staff note “Automatic wage indexation in Luxembourg: Limitations and Options for Reforms”, May 1, 2023.*

### 16.      While a key aim of indexation is to preserve purchasing power, this is not necessarily

### 16.      While a key aim of indexation is to preserve purchasing power, this is not necessarily 

### Distributional impacts of supply shocks and indexation
- Supply shocks (e.g., aftermath of the war in Ukraine) concentrate on a few internationally traded commodities, such as oil, gas, or food.
- The consumption basket of households varies along the income distribution, producing heterogeneous impacts across the income distribution (see Figure 5.A).
- For the specific energy price shock of 2022, despite price control measures taken by the March 2022 Tripartite, households in lower income quintiles have been disproportionately impacted (see Figure 5.B).
- From a consumption equity perspective:
  - High income individuals and households are likely to benefit relatively more from indexation than middle-class earners because of higher propensity to save (lower marginal propensity to consume).
  - Personal income tax rises relatively quickly to top marginal tax rates; for most middle-class households, the marginal tax rate is approximately the same as that applying to those at the very top of the income distribution (Table 1).
  - After a few rounds of indexation, this may put further pressure on real estate and other assets.

### Table 1: Before and After Tax Salary Increase Following an Indexation Event (as presented)
- €   20,000.00 — Marginal tax rate 12% — Gross index €    500.00 — Net index € 440.00 — Effective index 2.20%
- €   38,700.00 — Marginal tax rate 32% — Gross index €    967.50 — Net index € 657.90 — Effective index 1.70%
- €   58,000.00 — Marginal tax rate 39% — Gross index € 1,450.00 — Net index € 884.50 — Effective index 1.53%
- € 116,000.00 — Marginal tax rate 40% — Gross index € 2,900.00 — Net index € 1,740.00 — Effective index 1.50%
- € 155,000.00 — Marginal tax rate 41% — Gross index € 3,875.00 — Net index € 2,286.25 — Effective index 1.48%
- € 205,000.00 — Marginal tax rate 42% — Gross index € 5,125.00 — Net index € 2,972.50 — Effective index 1.45%

### Structural impacts: productivity and competitiveness
- Labor productivity:
  - Luxembourg has very high levels of labor productivity in some sectors, but apparent labor productivity has stagnated since the global financial crisis (GFC) relative to main trading partners (France, Germany, United States).
- Relative unit labor costs (ULC):
  - Relative ULCs compared to the euro area have worsened over the last three decades, particularly in financial services and in professional activities.
  - These trends accompanied an increase in the share of intermediate consumption of gross output, possibly signaling deeper integration of services in global value chains and outsourcing by some firms.
- Compensation vs productivity:
  - On aggregate, real hourly compensation broadly matched productivity during 1995-2021.
  - Over 2010-2021, compensation outpaced productivity growth, more pronounced in the services sector; manufacturing, construction, and transportation showed the reverse in some cases.
  - Evidence suggests a differential risk that a universal indexation system poses to different sectors.

### Firm adjustments and sectoral responses to indexation
- Firms adjust along other margins: base pay change, variable pay change, employment change, and a component capturing indexation (same across sectors).
- Correlations (2010-2021 annual data):
  - Indexation events tend to be associated with lower employment or lower base/variable pay growth in general.
  - Manufacturing, construction, and administrative and support services tend to exhibit slower employment growth following indexation.
  - Domestically oriented sectors such as retail trade may see employment gains.
- Contextual scenario:
  - Several index tranches expected over 2022-2024 could raise base pay by up to 12.5 percent, if left unmodulated.

### Costs and benefits of a flexible approach to AWI (Automatic Wage Indexation)
- Modulations of AWI in Luxembourg:
  - Lower the magnitude of second-round effects and the likelihood of a wage-price spiral.
  - If AWI operated fully automatically, it could raise firms’ inflation expectations and their incentive to raise prices, producing higher second-round effects.
  - Endogenous firm responses may mute headline inflation outcomes but not necessarily employment impacts.
- Lack of rules about suspension:
  - Increases uncertainty since ad-hoc modulation or fiscal compensation is politically driven during supply-side shocks.
  - Example: since February 2022, three Tripartite meetings were held addressing energy price shock and AWI; measures included untargeted price caps/controls for energy goods, postponement of the second round of indexation for 2022 to April 2023, and a subsidy to pay for the cost of a second indexation for 2023.
  - Upcoming election cycle in the second half of 2023 increased pressure for swift measures.
- Fiscal and political constraints:
  - Practical implementation of automatic indexation hinges on political will and could erode fiscal space.
  - Trend toward more costly fiscal measures to compensate households and firms for modulation increases fiscal risk, especially given large aging-related costs.

### Directions for reform — measures to increase resilience and equity

- Adopt a measure of core inflation as a more resilient benchmark for automatic indexation:
  - Many systems exclude volatile components from the inflation benchmark (example: Belgium’s Health Index excludes alcohol, tobacco, and energy; Italy uses an inflation projection excluding energy for wage negotiations).
  - Figure 9 (2002–2023): long-term averages of headline and core inflation are broadly the same, while volatility of core inflation is much lower (roughly half that of headline).
  - Empirical stats for 2002M1-2023M1:
    - Headline inflation averaged 2.1 percent with a standard deviation of 1.4.
    - Core inflation averaged 1.9 percent with a standard deviation of 0.8.
  - Excluding Covid-19 and energy shocks (2002M1-2019M12):
    - Headline and core inflation averages about 1.9 and 1.8 percent, respectively.
    - Core’s standard deviation half that of headline (0.5 versus 1.0).
  - Indexing to core inflation:
    - Reduces the likelihood of an additional indexation following an energy price shock and produces correspondingly lower second-round effects.
    - DSGE model simulation (small open economy, downward nominal wage rigidity, indexation rule: all wages increased by 2.5 percent in nominal terms as soon as lagged cumulative inflation exceeds 2.5 percent since last indexation) shows that for the same shock size, indexing to core inflation does not lead to an index tranche while indexing to headline does.
    - Short-run: both core and headline inflation end up much lower because of the missing index tranche under core indexing.
    - Long-run: effects are much smaller and align with relatively small empirical estimates of STATEC (2017) for second-round effects.
  - Distributional ramifications:
    - Excluding components (energy, perhaps food) that weigh heavily on lower-income households can reduce volatility but may worsen distributional outcomes immediately after shocks.
    - Such distributional effects could be addressed by targeted, temporary support measures whose fiscal cost would be much lower than broad-based measures.

- Introduce progressivity in wage adjustments:
  - Practical implementation: set a threshold above which implied wage change less than fully matches inflation.
    - Example design: individuals below threshold receive proportional compensation fully matching inflation (e.g., 2.5 percent), individuals above threshold receive a wage increase equal to 2.5 percent of the threshold itself (partial compensation).
  - Complexities to address:
    - Whether threshold applies to individual wages or total household income (many households file taxes jointly).
    - Political decision about where to set the threshold requires distributional analysis.
  - Concerns and mitigating factors:
    - Progressive system could lead to wage compression, but indexation is only one driver of wage growth.
    - Decomposition of annualized average wage growth (2010/Q1–2022/Q3) shows variable wage components routinely occur in addition to indexation and can be positive or negative.
    - Reduced automatic benefit for high-income earners may lead them to bargain for higher wages with employers, easing the cost of automatic indexation on firms.
  - Precedent: Malta’s cost-of-living allowance (COLA) is a fixed monetary benefit given equally to all workers; proposed Luxembourg system would mix Malta features with current AWI.

- Move toward a rule-based and sectoral approach to modulations:
  - Implement explicit quantitative criteria for triggering or suspending wage indexation to reduce uncertainty.
  - Possible rule designs:
    - Prescribe the maximum allowed number of indexations in a certain period, with excess tranches automatically postponed.
    - Track quantitative measures of cost-competitiveness (e.g., relative ULCs); when exceeded, automatically trigger suspension and/or talks between unions and employers (as in Belgium).

*Source: STATEC, Eurostat, and IMF Staff Calculations.*

### 32.      Given Luxembourg’s diversified economy, some decentralization of wage-setting

### 1luxea2023002 - 32.      Given Luxembourg’s diversified economy, some decentralization of wage-setting

### Decentralization of wage-setting: rationale and mechanism
- Universal automatic indexation currently benefits all workers, pensioners, and recipients of public benefits that are automatically adjusted for inflation.
- Proposed reform: define competitiveness benchmarks (measured, for instance, in neighboring countries and across several sectors) and continually evaluate whether Luxembourg is losing competitiveness.
- Trigger mechanism: if competitiveness benchmarks are crossed, suspend automatic wage indexation and trigger more decentralized bargaining.
- Expected benefits:
  - Lessen competitiveness concerns at the national level that often induce suspension of the system for all.
  - Give unions and employers more leeway to agree on the most sustainable way to split the costs of shocks, which could save jobs.
  - Reduce instances where system-wide suspension is used as the only instrument to address sector-specific pressures.

### Conclusions on resilience and distributional effects of automatic indexation
- Automatic wage indexation is a distinctive feature of Luxembourg’s labor market institutions, but recent events suggest its resilience to shocks could be improved.
- The system is routinely suspended during times of high inflation or economic slowdowns; most recently in 2022 when the second index tranche was postponed to April 2023.
- Common reasons for suspension:
  - Competitiveness grounds (e.g., to avoid multiple rounds of potentially costly wage increases for Luxembourgish firms).
  - Concerns about wage-price spirals.
- Political economy responses have sometimes included increased fiscal spending to mitigate the cost of indexation to firms (examples: the 2022 and 2023 Tripartite agreements).
- Suspension of the indexation may increase uncertainty and disproportionally hurt the most vulnerable parts of the population.
  - The recent energy price shock affected disproportionately households and individuals in lower income quantiles, given the higher weight of these goods in their consumption basket.

### Reform options to strengthen the system
- i) Indexing to core inflation rather than to headline.
  - Model-based analyses show that, in the face of a supply shock, changing the indexation rule to inflation excluding volatile components leads to a relatively lower probability of an indexation event taking place, with correspondingly lower second round effects and less volatility.
  - Targeted compensatory measures can be envisaged to ensure that the most vulnerable are not disproportionally affected by the relatively less frequent indexations.
- ii) Reforming the scheme to make it more progressive.
  - Options include tapering the wage increase or excluding higher quantiles of the income distribution from a proportional adjustment of the wage.
  - Expected effects: improve consumption equity and reduce cost impact for both the Government and for firms.
- iii) Greater decentralization of wage bargaining.
  - Implement explicit provisions to automatically suspend indexation in some sectors exposed to heightened pressures to make the system more resilient.

### Empirical analysis: Effect of indexation on inflation outcomes (Annex I)
- Methodology:
  - Regression analysis of headline, core, and services inflation (quarter-on-quarter, seasonally adjusted) for 2000–21.
  - Key variable: dummy = 1 if an indexation event took place in a quarter, 0 otherwise.
  - Controls included GDP, inflation of petroleum products, euro area inflation; all variables seasonally adjusted using Census X-12 multiplicative procedure.
- Table I.1. Luxembourg: Regression Analysis of the Effect of Indexation on Inflation (selected coefficient estimates)
  - Headline / Core / Services
    - Index tranche (+1): 0.11**, 0.14***, 0.20***
    - Index tranche: 0.20***, 0.25***, 0.42***
    - Index tranche (-1): 0.25***, 0.21***, 0.31***
    - Index tranche (-2): 0.16***, 0.14***, 0.29***
    - EA Inflation: 0.52***, 0.19***, 0.25***
    - Output Gap (-3): 0.01, 0.01, ...
    - 2015Q1: 0.52***, 0.74***, 0.65***
    - Petroleum Products: 0.03***, ..., ...
    - Petroleum Products (-4): ..., 0.00, ...
    - AR(1): ..., 0.07, 0.19*
    - AR(2): ..., 0.16**, ...
    - Constant: 0.10***, 0.13**, 0.14**
  - Sample and fit statistics:
    - N: 88, 88, 88
    - R-squared: 0.88, 0.61, 0.48
    - DW: 2.00, 2.12, 2.15
- Notes:
  - ***, **, * represent statistical significance at, respectively, 1 percent, 5 percent, and 10 percent level.
  - All variables, except dummies, have been defined in percentage terms.
  - A parenthesis close to the variable denotes the lead (if positive) or the lag (if negative), measured in quarters relative to any period t in the sample.

### Small open economy DSGE model and simulation results (Annex II)
- Model features:
  - Medium-scale New Keynesian small open economy model drawing on Gali and Monacelli (2016).
  - Features price and wage rigidity à la Calvo (1983) with indexation to past inflation, and imperfect passthrough of imported inflation.
  - Calibration adjusted to reflect key features of Luxembourg’s economy.
- Key modifications to better capture import-price shock dynamics:
  - Imported good interpreted as a generic energy commodity whose price is the main shock.
  - Incorporated downward nominal wage rigidity: nominal salaries cannot fall if not renegotiated.
  - Indexation rule adjusted to reflect practice in Luxembourg (wage adjusted by 2.5 percent only when the lagged cumulated inflation rate exceeds 2.5 percent, upon which the cumulated inflation variable is reset).
- Distinction between headline and core indexation:
  - Headline indexation: cumulate the price index of the final good consumed by the representative household (NICP index).
  - Core indexation: cumulate the price index of the good produced by domestic firms (measure of core inflation).
- Simulation exercise:
  - Perfect-foresight simulations following shocks to the price of the imported energy good.
  - Ran simulations for many shock sizes; in no case was an indexation triggered when indexing to core but not when indexing to headline for the same shock size.
  - Selected a realistic shock size (given the observed magnitude of the energy price shock of 2022) and calculated impulse response functions for the case where indexation happens when indexing to headline but not when indexing to core.
- Impulse response highlights:
  - Shock: increase in the error term of the AR(1) process for the price of the imported energy good.
  - Real income falls (present discounted value of income for the representative household declines), leading to a reduction in investment.
  - Short-term employment increases, as higher price of the domestic good spurs additional domestic production given relatively high capital stock.
  - Headline inflation increases significantly more than core inflation.
  - When indexing to headline (indexation event occurs) but not to core:
    - Additional wage increases weigh on short-term consumption and investment.
    - Core inflation spikes significantly within 8 quarters due to higher wage costs when indexing to headline, but over the medium-to-long term there is little impact of the indexation rule on core inflation.
    - Headline inflation increases significantly under both indexation rules, but somewhat more when headline inflation indexing is applied.
  - After about 6 quarters, the initial response gives way to medium-term adjustment: investment recovers and inflation abates.

*Source: STATEC, Eurostat, and IMF Staff Calculations.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2023/english/1luxea2023002.pdf_