## 1. Spain’s Revenue B oom

## Source details

**Canonical URL:** [1. Spain’s Revenue B oom](https://www.imf.org/-/media/files/publications/cr/2024/english/1espea2024002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2024/english/1espea2024002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2024/english/1espea2024002.pdf.json)

---

### Introduction and context
- Since the COVID-19 pandemic, Spain experienced one of the largest increases in general government revenues relative to GDP among euro area countries (May 15, 2024).
- By end-2023 the ratio of total revenues to GDP was about 3 percentage points above its 2019 level.
- In 2020–2021 the rise in revenues was driven mainly by social security contributions (SSCs) and income taxes (personal and corporate), accounting for a 2-percentage point increase in the overall revenue-to-GDP ratio by the end of 2021.
- In 2022 VAT increased by 0.5 percentage points; by 2023 VAT revenues declined, SSCs stabilized, and income taxes continued to grow, representing two-thirds of the revenue growth since 2019 by year-end.
- The four tax categories analyzed (PIT, SSCs, VAT, CIT) collectively accounted for 28 percent of GDP on average between 2000 and 2019 and 82 percent of total revenues.

### Objectives and questions addressed
- Determine whether post-2019 tax dynamics reflect regular cyclical dynamics, COVID-19–specific factors, or discretionary policy changes.
- Assess the permanence of the revenue boom and whether the level of revenues-to-GDP at end-2023 can be considered structural.
- Inform medium-term projections on structural tax-to-GDP ratios and design of a medium-term consolidation plan to rebuild fiscal space and reduce debt.

### Data and empirical approach (Tax buoyancy framework)
- Main econometric tool: Error Correction Models (ECMs) estimating short- and long-run buoyancies (γ short-run; θ long-run) between tax receipts and nominal GDP, the macroeconomic base, or the administrative base.
- Quarterly data at annualized values used; lags included for persistence (H = 1–2, K = 3–4 depending on specification).
- Macroeconomic bases constructed per García Miralles and Martínez Pagés (2023):
  - PIT macro base ≈ all sources of household income.
  - SSCs macro base ≈ households’ earnings from work.
  - VAT macro base ≈ sale of goods and services excluding public sector final consumption.
  - CIT macro base ≈ firms’ gross operating margins.
- Counterfactual revenue series for 2020–2023 constructed excluding key discretionary policy measures; policy impacts taken from authorities’ Stability Program Update, Draft Budgetary Plan, and AIReF reports.

### Key empirical findings and drivers of the 2020–2023 revenue boom
- Multiple factors contributed; principal drivers identified:
  - Large employment gains: small contraction in employment in 2020 followed by three years of sustained growth; employment reached its highest historical level in 2021–2023.
  - Fiscal drag from inflation beginning in late 2021: higher nominal income growth mechanically raises PIT revenues by pushing taxpayers into higher marginal brackets and reducing the relative value of fixed allowances/deductions.
  - Job retention schemes (e.g., ERTE) cushioned household incomes in 2020–early 2021, supporting PIT and SSCs bases.
  - VAT dynamics: historical estimations predicted a large fall in VAT in 2020–2021 which did not materialize; low VAT compliance gap during 2020–2023 and a possible partial shift from cash to electronic payments contributed.
- Tax-category concentration:
  - By end-2023 the remaining part of the post-2019 boom mostly reflected the PIT and SSCs components.

### Quantitative assessment of persistence and medium-term outlook
- A portion of the post-2019 rise in the revenue-to-GDP ratio, in the range of 2 – 2.5 percentage points, can be expected to persist in the medium term.
- Persistent components identified:
  - Increased effective PIT rate due to fiscal drag—expected to remain while central and regional governments do not adjust marginal rate schedules.
  - SSCs revenues expected to continue growing gradually over the next decade, by about 1 percent of GDP, as the 2021–2023 pensions reforms are phased in.
  - VAT revenues expected to remain at high levels as anti-inflationary measures introduced in 2022–2023 are phased out.
- Short-term buoyancy (γ) explains 2020–2023 dynamics; long-run buoyancy (θ) included to control for trends affecting estimation and longer-term trajectories.

### Personal Income Tax (PIT): outturn, fiscal drag, and persistence
- Outturn and drivers:
  - PIT macroeconomic base, tax base, and revenues recovered to 2019 levels by end-2020 and grew steadily over 2021-2023.
  - Starting from late 2021, revenues grew faster than both macroeconomic and administrative bases.
- Fiscal-drag decomposition and quantitative findings:
  - A decomposition suggests fiscal drag accounts for most of the revenue boom as of end-2023.
  - Counterfactual: fixing the base to its 2019 level while letting the effective PIT rate evolve explains a rise in PIT revenues of about 0.7 percentage points of GDP by end of 2023.
  - Applying a constant growth of 11.2 percent to 2019 incomes (cumulative CPI 2019-2022) explains most of the observed upward shift for salaries above the minimum wage.
- Elasticity estimates:
  - Bank of Spain (2024) ERTI estimate: 1.85 using microdata and 2019 tax parameters.
  - Revenue elasticity with respect to the base over 2019-2023: 1.57 for total PIT; 1.48 for PIT revenue accrued on income from workers’ salaries.
  - Adjusting for social security affiliation yields an ERTI of 1.68.
- Policy context and persistence:
  - As of 2024, the central government had not updated marginal rates for the national portion of PIT; some autonomous communities adjusted regional brackets.
  - 2022–2023 central government parameter updates:
    - Earned-income tax allowance now fully phases out at €19,747.5 instead of €16,825.
    - Minimum income level for automatic PIT withholding raised from €14,000 to €15,000.
  - These measures targeted low-income workers and did not remove fiscal drag across the whole distribution; absent further adjustments, elevated effective PIT burden will mechanically persist.

### Social Security Contributions (SSCs): drivers and reform impacts
- 2020 dynamics: macroeconomic base contracted more deeply than revenues and the administrative base, likely reflecting furloughs and job protection schemes.
- 2021-2023: macroeconomic and administrative bases grew steadily in parallel; revenues by end-2023 grew by a larger extent than both bases in relative terms.
- Pension reform impacts:
  - Main reform components: Intergenerational Equality Mechanism; Solidarity contribution above the maximum contributory base; progressive rise in the upper bound of the contributory base; obligation for self-employed to make contributions proportional to income; incentives for delayed retirement.
  - Quantitative estimates (Ministry of Social Security, Migration, and Inclusion; AIReF; FEDEA): revenue increase around 1 percent of GDP on average over 2022-2050.
  - As reforms are phased in gradually, the impact by 2029 is likely to be slightly below 1 percent of GDP.

### Value Added Tax (VAT): behavior, compliance, and outlook
- Pandemic impact and recovery:
  - VAT revenues and base fell sharply at pandemic onset and recovered steadily; VAT collection returned to pre-pandemic levels by mid-2022.
  - By end-2023 growth of VAT revenues, administrative base, and macroeconomic base relative to 2019 levels was relatively aligned.
- Compliance and model evidence:
  - During the GFC, VAT “compliance gap” spiked in 2008-2009; during 2020-2023 the VAT compliance gap remained low.
  - Adding the compliance gap to ECMs markedly improves historical fit over 2020-2023 and reduces the role of macroeconomic contraction in predicting post-COVID VAT revenues.
- 2024 outlook:
  - Revenues are expected to grow slightly in 2024 as anti-inflationary support measures are phased out, and stabilize above their 2019 values by approximately 0.3-0.4 percentage points of GDP.
  - Temporary VAT rate reductions on electricity, gas, and essential foods—amounting to approximately  0.2 percent of GDP—are set to expire by the of 2024.
  - Potential structural effect from increased electronic payments noted but magnitude difficult to estimate.

### Corporate Income Tax (CIT): volatility, recent measures, and medium-term outlook
- Historical volatility:
  - CIT revenues fell by more than 50 percent in Q2 and Q3 2020 relative to 2019 Q4; during the GFC CIT fell by over two-thirds.
  - CIT revenues fully recovered by 2021 and continued to grow in 2022 and 2023.
- 2023 policy changes:
  - Cap in deductions for losses in the consolidated-firm regime and a 15 percent marginal tax rate on profits above €20 million.
  - Combined impact in 2023 estimated at €1.4 billion (about 0.1 percent of GDP) per the government’s October 2023 Draft Budgetary Plans.
- Forecasts:
  - Nominal GDP and the macroeconomic base predict the 2020 fall and ensuing recovery; with no new measures the CIT revenues-to-GDP ratio is expected to remain broadly at its current level in the medium term.

### Implications for fiscal policy, projections, and reform priorities
- Persistence assessment:
  - A non-negligible share of the revenue boom is likely to be structural rather than purely cyclical if fiscal drag persists and reforms phase in as anticipated.
  - The employment-driven PIT component is likely to persist if moderate positive employment growth continues through 2029 (IMF staff projection).
- Policy levers:
  - Adjusting marginal PIT rate schedules or tax bases would alter the persistence of the PIT-driven component; absent such adjustments, the elevated effective PIT burden will mechanically persist.
- Fiscal consolidation context:
  - The revenue boom closed almost half of Spain’s revenues-to-GDP gap relative to peer countries—from 7.1 percentage points in 2019 to 4 in 2023.
  - Given the need for sizeable and sustained fiscal consolidation to rebuild fiscal buffers and set debt on a downward trajectory, continued efforts to boost revenues will be needed.
- Potential revenue-raising reforms highlighted:
  - Environmental taxation to raise additional resources while addressing externalities and supporting the green transition.
  - Enhancing tax efficiency by broadening bases and harmonizing rates, primarily in the VAT, to raise additional revenues while minimizing distortions.

### 1. Spain’s Revenue B oom ______________________________________________________________ 4

### 1. Spain’s Revenue B oom

### Introduction and context
- Since the COVID-19 pandemic, Spain experienced one of the largest increases in general government revenues relative to GDP among euro area countries (May 15, 2024).
- By end-2023 the ratio of total revenues to GDP was about 3 percentage points above its 2019 level.
- In 2020–2021 the rise in revenues was driven mainly by social security contributions (SSCs) and income taxes (personal and corporate), accounting for a 2-percentage point increase in the overall revenue-to-GDP ratio by the end of 2021.
- In 2022 VAT increased by 0.5 percentage points, alongside continued growth in income taxes and "other taxes". By 2023 VAT revenues declined, SSCs stabilized, and income taxes continued to grow, representing two-thirds of the revenue growth since 2019 by year-end.
- The four tax categories analyzed (PIT, SSCs, VAT, CIT) collectively accounted for 28 percent of GDP on average between 2000 and 2019 and 82 percent of total revenues.

### Objectives and questions addressed
- Determine whether post-2019 tax dynamics reflect regular cyclical dynamics, COVID-19–specific factors, or discretionary policy changes.
- Assess the permanence of the revenue boom and whether the level of revenues-to-GDP at end-2023 can be considered structural.
- Inform medium-term projections on structural tax-to-GDP ratios and design of a medium-term consolidation plan to rebuild fiscal space and reduce debt.

### Data and empirical approach (Tax buoyancy framework)
- Main econometric tool: Error Correction Models (ECMs) estimating short- and long-run buoyancies (γ short-run; θ long-run) between tax receipts and nominal GDP, the macroeconomic base, or the administrative base.
- Quarterly data at annualized values used to increase sample size and capture sharp COVID-19 dynamics; lags included for persistence (H = 1–2, K = 3–4 depending on specification).
- Macroeconomic bases constructed per García Miralles and Martínez Pagés (2023): 
  - PIT macro base ≈ all sources of household income.
  - SSCs macro base ≈ households’ earnings from work.
  - VAT macro base ≈ sale of goods and services excluding public sector final consumption.
  - CIT macro base ≈ firms’ gross operating margins.
- Counterfactual revenue series for 2020–2023 constructed excluding key discretionary policy measures; policy impacts taken from authorities’ Stability Program Update, Draft Budgetary Plan, and AIReF reports. Overall, these policies were quantitatively small and do not affect main results.

### Key empirical findings and drivers of the 2020–2023 revenue boom
- Multiple factors contributed to the boom, notably large employment gains and "fiscal drag".
- Labor market:
  - Small contraction in employment in 2020 followed by three years of sustained growth; employment reached its highest historical level in 2021–2023.
  - Robust labor market performance was a key driver of PIT and SSCs revenues.
  - Extended national job retention schemes (e.g., ERTE) cushioned household incomes in 2020–early 2021, supporting PIT and SSCs bases.
- Inflation and fiscal drag:
  - Starting in late 2021, rising inflation was the main driver of growing PIT revenues via fiscal drag: higher nominal income growth mechanically raises revenues, including by pushing workers into unchanged higher marginal tax brackets and to the upper bounds of deductions and allowances.
- VAT dynamics:
  - Historical estimations predicted a large fall in VAT revenues during 2020–2021, which did not materialize.
  - The discrepancy reflects pronounced VAT dynamics in the post-GFC years when sluggish aggregate demand and filing compliance issues drove a persistent fall in VAT revenue.
  - A partial shift from cash to electronic payments might have contributed to bringing a larger share of transactions into the formal economy (not definitively quantified here).
- Corporate income tax (CIT) and other factors are analyzed in the full chapter (not reproduced in detail here).

### Quantitative assessment of persistence and medium-term outlook
- A portion of the post-2019 rise in the revenue-to-GDP ratio, in the range of 2 – 2.5 percentage points, can be expected to persist in the medium term.
- Persistent components identified:
  - Increased effective PIT rate due to fiscal drag—expected to remain while central and regional governments do not adjust marginal rate schedules.
  - SSCs revenues expected to continue growing gradually over the next decade, by about 1 percent of GDP, as the 2021–2023 pensions reforms are phased in.
  - VAT revenues expected to remain at high levels as anti-inflationary measures introduced in 2022–2023 are phased out.
- The short-term buoyancy concept (γ) is most relevant for explaining 2020–2023 dynamics, but long-run buoyancy (θ) is included to control for trends that affect estimation and longer-term trajectories.

### Implications for fiscal policy and projections
- Understanding drivers (cyclical, pandemic-specific, discretionary) is essential for realistic revenue projections and debt dynamics.
- If fiscal drag effects persist and structural reforms phase in as anticipated (pensions changes, phasing out of temporary VAT measures), a non-negligible share of the revenue boom is likely to be structural rather than purely cyclical.
- Policy adjustments to marginal PIT schedules or tax bases would alter the persistence of the PIT-driven component of the revenue boom; absent such adjustments, the elevated effective PIT burden will mechanically persist.
- The combined analysis of historically estimated buoyancy and tax-specific developments informs calibrated medium-term paths for each tax and the aggregate revenue-to-GDP profile.

*Prepared by Carlo Pizzinelli (EUR); Chapter: "DISSECTING SPAIN’S SOARING TAX REVENUES AND THEIR IMPLICATIONS FOR PROJECTIONS" (May 15, 2024).*

### 14.      Since the onset of the pandemic, the PIT macroeconomic base, the tax base, and

### 14.      Since the onset of the pandemic, the PIT macroeconomic base, the tax base, and

### Personal Income Tax (PIT): outturn and drivers
- PIT macroeconomic base, the tax base, and revenues all recovered to 2019 levels by the end of 2020 and grew steadily over 2021-2023.
- Starting from late 2021, revenues grew at a faster pace than both the macroeconomic and administrative bases.
- Contrasts with the GFC: during the GFC the sharp and persistent contraction in PIT revenues was closely aligned with the drop and sluggish recovery in the PIT base.
- Dynamic forecasts:
  - Over 2020-2021, the macroeconomic and administrative bases predicted PIT outturns better than nominal GDP.
  - No indicator explains well the continued growth in revenues during 2022-2023.
  - Extending the ECM with the 4-quarter lag of the effective PIT rate substantially improves forecast fit; adding consumer price inflation provides only modest improvement.

### Fiscal drag and decomposition for PIT
- Fiscal drag related to the inflationary wave beginning in late 2021 is a likely explanation for mounting forecasting gaps over 2022-2023.
- Mechanisms of fiscal drag (as described):
  - With a progressive marginal PIT rate schedule, increases in income are taxed at marginal rates higher than average rates, raising the latter.
  - Income growth can move taxpayers into higher marginal tax brackets.
  - Many tax benefits (allowances, deductions, credits) have fixed amounts or upper-income limits, reducing their relative value as nominal incomes rise.
- Quantitative findings:
  - A simple decomposition suggests fiscal drag accounts for most of the revenue boom as of end-2023.
  - Counterfactual: fixing the base to its 2019 level while letting the effective PIT rate evolve explains a rise in PIT revenues of about 0.7 percentage points of GDP by end of 2023.
  - Alternative microdata/tabulations support this picture: workers’ income distribution shifted upward between 2019 and 2022.
  - Applying a constant growth of 11.2 percent to 2019 incomes (cumulative CPI 2019-2022) explains most of the observed upward shift for salaries above the minimum wage.
- Elasticity estimates and comparisons:
  - Bank of Spain (2024) estimates the PIT elasticity of revenues to taxable income (ERTI) at 1.85 using microdata and 2019 tax parameters.
  - Revenue elasticity with respect to the base computed using these series over 2019-2023 is 1.57 for total PIT and 1.48 when considering only PIT revenue accrued on income from workers’ salaries.
  - Using social security affiliation to proxy for growth in workers filing PIT adjusts the latter to compute an ERTI of 1.68, close to 1.85.

### PIT: near-term persistence and policy context
- Outlook:
  - The PIT revenue boom is likely to persist over the medium term.
  - Growth in revenues accounted for by a higher base is underpinned by sustained employment growth over the post-COVID period; IMF staff project moderate positive employment growth until 2029, making this component permanent in the baseline.
  - The component related to the rise in the average effective tax rate is likely to persist but may not be fully permanent.
- Policy actions to date:
  - As of 2024, the central government had not updated marginal rates profile for the national portion of PIT; some autonomous communities have adjusted regional brackets.
  - In 2022 and 2023, the central government updated parameters relating to one tax allowance and its automatic withholding:
    - Earned-income tax allowance now fully phases out at €19,747.5 instead of €16,825.
    - Minimum income level for automatic PIT withholding raised from €14,000 to €15,000.
  - These measures were targeted to low-income workers and did not address the rise in average tax rates across the entire income distribution.
- Implication: future regional and possible central adjustments to PIT schedules may halt further increases in effective PIT rates but are unlikely to fully revert the rise already realized.

### Social Security Contributions (SSCs)
- 2020 dynamics:
  - The macroeconomic base experienced a deeper contraction than revenues and the administrative base during 2020.
  - Likely reason: macroeconomic base focuses on compensation of individuals at work and may not capture certain dynamics (e.g., furloughs, job protection schemes).
- 2021-2023:
  - Macroeconomic and administrative bases grew steadily and in parallel over 2021-2023.
  - Revenues by end-2023 had grown by a larger extent than both bases in relative terms.
  - Dynamic forecasts: macroeconomic and administrative bases track SSC outturns over 2020-2023 better than GDP; macroeconomic base continues to capture outturn well over 2022-2023 while the administrative base overpredicts SSCs in 2022-2023.
  - Reasons for administrative base divergence: administrative proxy limitations (e.g., upper bound of retirement pension contributory base excludes higher salary growth; self-employed could choose minimum contributions until 2023).
- Policy and reform impacts:
  - 2021-2023 pension reforms will further increase total SSC revenues over time.
  - Main revenue-generating components from reforms include:
    - Intergenerational Equality Mechanism.
    - Solidarity contribution above the maximum contributory base.
    - Progressive rise in the upper bound of the contributory base.
    - Obligation for self-employed workers to make contributions proportional to their income.
    - Incentives for delayed retirement increasing total contributors.
  - Quantitative estimates (Ministry of Social Security, Migration, and Inclusion; AIReF; FEDEA): revenue increase around 1 percent of GDP on average over 2022-2050.
  - As reforms are phased in gradually, the impact by 2029 is likely to be slightly below 1 percent of GDP.

### Value Added Tax (VAT)
- Pandemic impact and recovery:
  - VAT revenues and base fell sharply at onset of the pandemic and subsequently recovered steadily.
  - VAT collection returned to pre-pandemic levels by mid-2022; administrative and macroeconomic bases recouped losses over the following two quarters.
  - By end-2023, growth of VAT revenues, administrative base, and macroeconomic base relative to 2019 levels was relatively aligned.
- Dynamic forecasts and compliance:
  - All dynamic forecasts predicted a large contraction in VAT revenues as a share of GDP over 2020-2022, while actual outturn remained stable; forecasts and outturn broadly consistent by late 2023.
  - The predicted drop was likely influenced by the GFC years in the model estimation, when VAT revenues fell more than GDP and the tax base.
  - Compliance:
    - During the GFC, a sharp spike in the VAT “compliance gap” occurred in 2008-2009, contributing to large revenue losses; the gap then gradually declined.
    - The compliance gap spike was tied to changes in the filing system that coincided with the recession.
    - During the pandemic (2020-2023), the VAT compliance gap remained low and did not seem to drive revenue dynamics.
  - Model extensions:
    - Adding the compliance gap to the ECM markedly improves historical fit over 2020-2023 and reduces the role of macroeconomic contraction in predicting post-COVID VAT revenues.
    - Controlling for the one-year lag of the effective VAT rate does not improve prediction in the early horizon and produces larger over-prediction in the last two years.
    - Effective VAT rate evolves little over the business cycle unless statutory rates or regime coverage change; thus it contains limited short-run predictive power.

*Source: 1espea2024002 - 14.      Since the onset of the pandemic, the PIT macroeconomic base, the tax base, and*

### 26.      In 2024, revenues are expected to grow slightly, as the anti-inflationary support

### 26.      In 2024, revenues are expected to grow slightly, as the anti-inflationary support

### Revenues outlook for 2024 and drivers
- Revenues are expected to grow slightly in 2024 as the anti-inflationary support measures are phased out, and stabilize above their 2019 values by approximately 0.3-0.4 percentage points of GDP.
- The temporary VAT rate reductions on electricity, gas, and essential foods—amounting to approximately  0.2 percent of GDP—are set to expire by the of 2024.
- A slight increase in 2023 revenues relative to 2019 may be linked to a permanent increase in electronic payments and the consequent fall in cash transactions during the pandemic (Ferrando and Posada, 2022; Barreiro et al., 2023). Because electronic payments are more likely to be recorded for tax purposes, this behavioral switch could entail a structural increase in revenues, but:
  - The magnitude of this change is difficult to estimate.
  - Further increases in the use of cash transactions in future years remain challenging to predict.

### Corporate Income Taxes (CIT): historical volatility and outlook
- Historical behavior:
  - CIT revenues fell by more than 50 percent in the second and third quarters of 2020 relative to 2019 Q4.
  - During the GFC, CIT revenues had fallen by over two-thirds.
  - CIT revenues fully recovered by 2021 and continued to grow in 2022 and 2023.
- Bases behavior:
  - The macroeconomic base (firms’ operating margins rather than profits) also fell in 2020 but not to the same extent as revenues.
  - The administrative base did not contract but showed high quarter-on-quarter volatility.
- Policy changes in 2023:
  - Two material revenue-raising measures were introduced in 2023: a cap in the deductions for losses in the consolidated-firm regime and a 15 percent marginal tax rate on profits above €20 million.
  - The combined impact of the two measures in 2023 was estimated at €1.4 billion (about 0.1 percent of GDP) in the government’s October 2023 Draft Budgetary Plans.
- Forecasts and explanation of forecast errors:
  - Nominal GDP and the macroeconomic base both predict the fall in the CIT revenues-to-GDP ratio over 2020 and its ensuing recovery to the current level, which is expected to remain broadly stable in the medium term.
  - The administrative base does not capture the initial fall in 2020; historical precedent from the GFC (2007–2012) suggests administrative base dynamics can align with revenue contractions when downturns are prolonged.
  - COVID-19 was a shorter-lived downturn than the GFC; the sudden initial slowdown in economic activity may not have been fully reflected in firms’ reported profits.
  - A moratorium for tax payments at the onset of the pandemic may have temporarily lowered quarterly outturns in 2020.
  - With no new measures, the CIT revenues-to-GDP ratio is expected to remain broadly at its current level in the medium term.

### Conclusions on the 2020–2023 revenue boom and medium-term prospects
- Drivers of the tax revenue boom of 2020-2023:
  - Strong labor market performance with only a minor contraction in employment followed by sustained growth avoided a persistent fall in the PIT and SSCs bases.
  - Job protection schemes at the onset of the pandemic provided workers income even when not effectively working, sustaining PIT and SSCs tax bases.
  - In 2022–2023, the rise in nominal wages in response to high inflation generated a fiscal drag effect in the PIT as tax brackets and most tax benefits were not indexed.
  - Unlike in the GFC, VAT revenues did not fall sharply during COVID-19 due to sustained household consumption when allowed and strong collection compliance.
  - By end-2023, the remaining part of the post-2019 boom in the revenue-to-GDP ratio mostly reflected the PIT and SSCs components.
- Medium-term persistence:
  - The component of PIT revenue increase associated with employment growth is likely to persist as long as the labor market remains robust.
  - The bracket drift component should largely persist for now, though it may wane partially if the central government and more regional governments update PIT schedules.
  - Social security revenues are expected to rise further through the progressive phasing-in of the contribution increases implied by the 2021-2023 pension reforms.
- Fiscal context and reform priorities:
  - The revenue boom closed almost half of Spain’s revenues-to-GDP gap relative to peer countries—from 7.1 percentage points in 2019 to 4 in 2023.
  - Given the need for sizeable and sustained fiscal consolidation to rebuild fiscal buffers and set debt on a downward trajectory, continued efforts to boost revenues will be needed.
  - Potential revenue-raising reforms highlighted:
    - Environmental taxation could raise additional resources while addressing externalities and supporting the green transition.
    - Enhancing the efficiency of taxes by broadening their base and harmonizing rates, primarily in the VAT, could raise additional revenues while minimizing distortions.

### Annex: data definitions and estimation details (selected)
- Personal Income Taxes:
  - Macroeconomic base comprises employees’ compensation net of social security contributions, pensions, other benefits, households’ gross operating surplus, and income from interests, dividends, and other sources.
  - Administrative tax base includes income reported by households in tax filings from work earnings, pensions, social transfers, business income, interests and dividends, rent, and gains from asset sales.
  - Time series: revenues and administrative tax base start in 1995 Q1; macroeconomic base starts in 1999Q1.
  - ECM estimation: 1 lag for the explanatory variable and 3 for the dependent variable.
- Social Security Contributions:
  - Macroeconomic base: employee compensation and number of self-employed persons multiplied by compensation per employee in the private economy.
  - Administrative base proxied through PIT income base from salaries and other non-pension non-capital income.
  - Time series: revenues, administrative tax base, and macroeconomic base all start in 1995 Q1.
  - ECM estimation: 2 lags for the explanatory variable and 4 for the dependent variable.
- Value Added Taxes:
  - VAT macroeconomic base: household consumption, government intermediate consumption, general government investment, housing investment, and tourism exports.
  - Time series start in 1995 Q1.
  - ECM estimation: 2 lags for the explanatory variable and 4 for the dependent variable.
- Corporate Income Taxes:
  - Macroeconomic base: firms’ gross operating surplus from the national accounts (time series start in 1999Q1).
  - Administrative base and revenue series from the tributary agency begin in 1995 Q1.
  - ECM estimation: 1 lag for the explanatory variable and 3 for the dependent variable.

*Source: IMF staff analysis as presented in the chapter.*

### 8.      Spain's performance in non-tourism services exports has been on an upward trajectory

### 8.      Spain's performance in non-tourism services exports has been on an upward trajectory

### Recent export performance and decomposition
- Spain’s non-tourism services exports have been on an upward trajectory since before the pandemic.
- Since the GFC, Spain’s non-tourism services exports growth has been converging with the global average, contrasting with euro area peers whose market shares declined.
- Decomposition (shift-share) findings:
  - Geographic specialization effect had a detrimental impact, especially during 2012-15, reflecting Spain’s concentration of non-tourism services exports within Europe (in contrast to Germany or France which exported a substantial portion of their services outside of the EU).
  - Sectoral specialization effect, though smaller, also negatively influenced Spain’s non-tourism services export performance—primarily due to global downturns in transportation activities and wholesale and retail trade (sectors heavily hit during the GFC and the euro area crisis).
  - Country-specific export performance (Spain’s own export performance) has been positive since the euro area crisis and suggests Spain could have surpassed the global average if not for unfavorable sectoral specialization and export market effects during that period.
- Sub-periods of positive country-specific export performance typically coincided with periods of REER depreciation, tentatively suggesting a contribution from cost competitiveness improvements.

### Econometric evidence on cost competitiveness (error-correction model)
- The error-correction model (sample: euro area countries) identifies a statistically significant negative relationship between the index of ULC-based REER and country-specific non-tourism services export performance in both the long term and the short term.
- Long-term elasticity (stage 1): a 10-percent REER depreciation—indicative of more cost-competitive export activities—is associated with a 10 percent relative (vis-à-vis the world average) increase in exports.
- Short-term elasticity (stage 2): reported in text as -0.24, implying a 10 percent REER depreciation would lead to an additional 2.4-percentage-point increase in non-tourism services export growth (relative to the world average) during the same period.
- Contribution to observed improvements: the substantial depreciation in REER observed in Spain since the GFC explained about three-fourths of the actual improvements from 2008-12 to 2016-19.

Table 1 (select econometric estimates as reported)
- Stage 1 — Log (Performance Index) on Log (ULC-based REER)
  - Coefficient: -1.01**  (0.24)
  - Country FE: Yes
  - No. of Obs.: 52
  - Adj. R-squared: 0.48
- Stage 2 — Δlog (Performance Index) on Δlog (ULC-based REER) and stage-1 residual, lagged
  - Δlog (ULC-based REER) coefficient: -0.28*  (0.16)
  - Stage-1 residual, lagged: -0.85***  (0.12)
  - Time FE: Yes
  - No. of Obs.: 39
  - Adj. R-squared: 0.24

(Note: both the textual short-term elasticity -0.24 and the table entry -0.28* appear in the source.)

### Role of non-cost competitiveness (product market and services trade regulation)
- Alternative second-stage specifications incorporate non-cost competitiveness measures and their interaction with the ULC-based REER.
- Interaction results:
  - Using either product market regulation (PMR) or services trade restrictiveness indices (STRI) as non-cost competitiveness indicators, the interaction term with REER depreciation is negative and statistically significant.
  - Interpretation: for a given extent of REER depreciation, countries with less regulated product markets or less restrictive services trade regulations experience greater improvements in export performance.
- Spain’s reform context:
  - Since the GFC, Spain undertook significant product market deregulation; by 2018, the overall PMR index and indices for various sub-categories stood below those of the majority of other EU countries.
  - This deregulation, together with REER depreciation, likely played a key role in elevating Spain’s non-tourism services export growth.
- Identification caveat: the exact roles are hard to pin down given limitations of the identification strategy, including potential omitted variable bias and other interactions with non-cost-competitiveness factors.

### Post-pandemic export performance
- Overall services export growth in 2021-22 was way below the global average, primarily due to the pandemic’s large impact on tourism (tourism accounted for around 52 percent of the country’s total services exports pre-pandemic).
- Focusing on non-tourism services exports:
  - Spain’s export growth since the pandemic onset has aligned closely with the global aggregate growth rate and outperformed euro area peers.
  - The geographic effect remained adverse (due to predominance of Europe in Spain’s services exports).
  - Sectoral effect was only mildly positive, driven by sectors such as ICT and professional services—areas where Spain had begun specializing even before the pandemic—yet overall non-tourism performance has been notably robust.

### Productivity trends in services and links to export performance
- Productivity growth in services can be both a driver and an outcome of strong services export performance:
  - High productivity growth in services can lower ULCs and thereby depreciate the ULC-based REER (cost competitiveness), while expanding variety and quality of services (non-cost competitiveness).
  - Higher services exports may feed back into productivity growth through scale or learning-by-doing effects.
- Empirical patterns (pre- and post-pandemic):
  - Using EU-KLEMS, sectors such as wholesale and retail trade and repairs, telecommunications, and professional services (including legal and accounting services) have seen positive labor productivity growth since 2012—driven by labor composition changes, accumulation of intangible capital, and improvements in TFP.
  - This productivity growth performance was typically stronger than in other sectors of the Spanish economy.
  - Labor productivity in non-tourism services sectors has continued to grow steadily since the pandemic; Spain’s labor productivity growth performance in non-tourism services has been among the strongest within the euro area.
  - Terms-of-trade for non-tourism services (ratio of the export price index to the import price index) have experienced a significant decline in recent years, contrasting with the relatively stable ratio in goods trade and tourism sectors.

### Conclusions and policy recommendations
- Summary conclusion:
  - Since the GFC, Spain’s non-tourism services exports have been on an upward trajectory and have played an increasingly important role in improving the current account balance.
  - The positive trend accelerated during and after the pandemic, supporting economic recovery despite weak domestic demand and sluggish goods exports.
  - Shift-share analysis shows robust country-specific performance since 2012, accelerating post-pandemic. Econometric analysis attributes improvements to both cost competitiveness (ULC-based REER depreciation) and non-cost competitiveness (product market and services trade deregulation). Sustained productivity growth in key tradable services sectors likely reinforced cost competitiveness and may also have benefited from enhanced export performance.
- Policy recommendations to sustain performance:
  - Maintain hard-earned cost competitiveness gains.
  - Continue addressing product market inefficiencies and seize remaining opportunities to enhance efficiency in product markets, including:
    - Reevaluating the extensive array of size-related rules and regulations.
    - Persisting in efforts to harmonize regional regulatory frameworks to build a fully integrated Spanish internal market in which the most productive firms can scale up more easily.
    - Fostering product market competition.
  - Support ongoing productivity growth in key tradable services sectors by facilitating innovation and talent acquisition so wages can grow steadily without putting significant upward pressure on ULCs.

*Source: IMF staff analysis in the chapter “Spain’s performance in non-tourism services exports has been on an upward trajectory.”*

### 4.      The approach taken to reduce labor market duality was to strengthen employment

### 4. The approach taken to reduce labor market duality was to strengthen employment protection legislation (EPL) for current temporary workers

### Reform design and immediate regulatory changes
- The reform limited the set of circumstances that warrant the use of temporary contracts (the temporary contract for work or service was eliminated), strengthened enforcement, and imposed higher penalties on very short durations and fraudulent contracting.
- Employers of temporary workers on a contract shorter than 30 days must pay an additional social security fee when that contract is terminated, except for workers under special regimes (e.g., carbon mining, housekeeping, and self-employed in agriculture) as well as temporary contracts for substitution.
- Penalties for the fraudulent use of temporary contracts are now established per contract and not per firm, setting the maximum penalty at €10,000 per fraudulent contract.
- Temporary contracts are automatically converted to permanent if: (i) the temporary hiring does not comply with the regulation, or (ii) the same worker is hired repeatedly with a temporary contract, within the same firm, for 18 months in a period of 24 months.
- The regulation promoted the use of seasonal open-ended contracts (“fixed-discontinuous”, FD) for seasonal activities (e.g., agriculture or retail around holidays) and created a new permanent contract for the construction sector.

### Research objectives and empirical strategy
- The paper analyzes whether the reform:
  - (i) had a positive impact on aggregate employment (after controlling for the economy’s position in the cycle);
  - (ii) increased job quality by promoting a shift from temporary to permanent hiring;
  - (iii) enhanced job stability, including by reducing overall turnover.
- Identification: differences-in-differences comparing outcomes across region-sector-occupation (rso) cells before and after the reform, exploiting that an EPL reform should disproportionately affect rso cells with a higher “natural” turnover rate.
- Natural turnover rate proxied by Spain’s pre-COVID historical average (2015-19) of the actual turnover rate for each rso cell.
- Estimated equation used:
  Δyy_Pss = α + β θ_Pss + δ_P + δ_s + δ_s + ε_Pss
  where yy_Pss is the labor market outcome; θ_Pss is degree of exposure; Δyy_Pss is computed as the (log-)difference between 2021Q4 and 2023Q4. Region, sector and occupation fixed effects (δ) included.

### Data, sample design and limitations
- Main micro datasets: Encuesta de Población Activa (EPA) used for empirical analysis; Muestra Continua de Vidas Laborales (MCVL) and social security administrative data used in related literature.
- EPA flows: 1/6 of the sample renewed every quarter; 5/6 maintained between consecutive quarters. Flows used to compute natural turnover rate.
- Analysis restricted to salaried workers in the private sector (public sector not affected by reform).
- When computing turnover rates at rso level, minimum acceptable number of observations per rso cell set at 20 to avoid extreme values; with this cutoff distribution has a large mass between 1 and 7 percent.
- Sample-size limitation: EPA flows available only at 1-digit sectors and occupations, limiting observations. Ongoing work will exploit U.S. data disaggregated at 3-digit level to increase observations and address endogeneity.

### Stylized facts and aggregate patterns
- The 2021 labor reform was followed by a reduction in the incidence of temporary employment of nearly 10 percentage points.
- Heterogeneous impacts by group:
  - Young workers (15-24 years old): decline in the temporary share of over 20 percentage points between end-2021 and end-2023.
  - Migrants: decrease of about 16 percentage points over the same period.
  - Larger decline for low-skilled (educational attainment up to lower secondary education) compared to high-skilled.
  - Women did not seem to experience a differential impact relative to men.
- Sectoral composition:
  - Temporary employment in the public sector has not been affected by the reform so far; composition remains biased toward contracts lasting 1 year or more.
  - Decline in private-sector temporary employment largely driven by reduction in number of workers with contracts lasting 1-11 months.
- Fixed-discontinuous (FD) contracts:
  - Cumulative number of new FD contracts reached 4.8 million in January 2024.
  - Average monthly social security affiliation of FD workers remained broadly stable at about 900 thousand persons.
  - Interpretation: large FD-job creation accompanied by large FD-job destruction (many FD workers shift into inactivity or are dismissed), implying increased FD-job turnover.

### Job stability, transitions, and disenrollment
- Transitions from employment to unemployment (E→U):
  - Increased with the pandemic, though not as much as during the Global Financial Crisis (GFC).
  - Since GFC, E→U flows on a downward path; after a temporary rise during the pandemic they are back to pre-COVID downward trend with no straightforward impact of the reform.
- By contract type:
  - Transitions from employment to non-employment have increased for FD workers, and to a lesser extent for regular permanent workers since the reform.
- Disenrollment from social security:
  - Disenrollment rates of FD workers increased significantly since the reform, driven by a large increase in shifts to inactivity and a non-negligible increase in quits/voluntary disenrollments.
  - Unfinished trial periods have increased but have not materially increased overall disenrollment for permanent workers.
- Disenrollment rates are computed as the ratio of average monthly disenrollments over average number of workers affiliated to the General Regime of social security.

### Findings from related literature (microdata-based studies)
- Conde-Ruiz et al (2023) (administrative social security data): reduction of contractual temporary employment, but employment stability in terms of duration and job transitions has not improved significantly.
- Verd et al (2024) (EPA up to end-2022): temporary share declined substantially, but temporary employment is now more concentrated among the most vulnerable young people and other precariousness (e.g., involuntary part-time) emerged.
- Martínez Jorge and Victoria (2022) (EPA): segments more exposed to the reform experienced relatively larger increases in permanent employment; effect on temporary and overall employment not statistically different across exposure segments.
- Preliminary Ministry of Economy work (MCVL, survival analysis): some increase in overall survival rates of job episodes after the reform, despite some decline in survival rate of open-ended episodes due to partial substitution of temporary contracts by open-ended ones.
- Anghel et al (2023) (micro-simulations, Household Budget Survey & Survey of Living Conditions): shift from temporary to permanent contracts in 2022 may have led to an increase in the spending-to-income ratio between 0.18 and 0.24 percentage points.
- Martínez-Carrascal (2023) (Consumer Expectations Survey): rising share of permanent contracts reduced perceived probability of job loss, lowered precautionary savings, and raised household spending levels.

### Empirical results (EPA rso-level analysis)
- Main finding: the reform was successful in lowering the share of temporary employment, although its impact on broader employment stability is less clear.
- Rso cells with historically high turnover rates (more exposed to reform) experienced:
  - Relative declines in the temporary employment share, especially for the young and migrants.
  - Increases in the shares of permanent and fixed-discontinuous contracts.
- Inconclusive evidence regarding impact on key transitions (e.g., employment to unemployment) and overall employment stability.
- Baseline regression results (At least 20 observations per cell; dependent variable: Change in Labor Market Outcome (2023Q4 vs 2021Q4); controls: region, sector and occupation fixed effects; robust standard errors):
  - Column (1) Change temporary share: Turnover Rate (E-->U) coefficient = -1.083*** (standard error 0.318)
  - Column (2) Change FD share: Turnover Rate (E-->U) coefficient = 0.385* (standard error 0.194)
  - Column (3) Change permanent share: Turnover Rate (E-->U) coefficient = 0.698* (standard error 0.354)
  - Column (4) Growth total employment: Turnover Rate (E-->U) coefficient = -0.776 (standard error 0.919) — not statistically significant
  - Column (5) Growth temporary employment: Turnover Rate (E-->U) coefficient = 0.137 (standard error 2.438) — not statistically significant
  - Column (6) Growth FD employment: Turnover Rate (E-->U) coefficient = -2.549 (standard error 14.93) — not statistically significant
  - Column (7) Growth permanent employment: Turnover Rate (E-->U) coefficient = 3.266 (standard error 2.212) — not statistically significant
  - Column (8) Growth average effective hours worked: Turnover Rate (E-->U) coefficient = 0.528 (standard error 0.443) — not statistically significant
- Interpretation:
  - Negative and statistically significant β on exposure variable for temporary share implies more exposed rso cells experienced larger relative decreases in temporary employment share.
  - Positive and significant coefficients for FD and permanent shares indicate relative increases in those shares in more exposed rso cells.
  - No conclusive evidence of effects on growth of employment (total or by contract type) or hours worked (coefficients not significant).

### Areas for further work and robustness checks
- Additional analysis is needed to make a more definitive assessment:
  - Replace natural turnover rates computed from historical Spanish data with natural turnover rates in the U.S., where EPL is virtually non-existent, to further address endogeneity concerns and increase observations by exploiting sector-occupation at the 3-digit level.
  - Consider alternative specifications including dynamic regressions to examine how reform effects evolved throughout the post-reform period.
  - Robustness exercises with different minimum-observation cutoffs showed unchanged key findings.

*Source: 1espea2024002 - 4.      The approach taken to reduce labor market duality was to strengthen employment (IMF staff analysis and EPA-based empirical results).*

### 24.      The effect of the reform on the temporary employment share has been heterogeneous

### The effect of the reform on the temporary employment share has been heterogeneous

### Heterogeneous effects across groups
- Rso cells more exposed to the reform show statistically significant declines in temporary employment shares for the young and the low skilled (columns 1 and 4 of Table 2).
- Coefficients on the reform exposure variable in the other regressions in Table 2 are insignificant:
  - negative sign for the change in the temporary share for women (column 2);
  - positive sign for migrants (column 5).

### Effects on employment stability and transition rates
- The broader effect of the reform on employment stability is unclear.
- Transition-rate analysis (Table 3) finds a single statistically significant coefficient:
  - positive impact on transitions from temporary employment to FD employment (column 6).
- For other flows there is no statistical evidence supporting a clear effect; the lack of significance may reflect limited statistical power given the small sample size.

### Robustness of the decline in temporary employment share
- The estimated decline in the temporary share is robust to several alternative specifications:
  - Different combinations of fixed effects, including pairs: region-sector, region-occupation, and sector-occupation (Table A.1).
  - Different cutoffs for observations per cell: 10, 15, and 20 observations per cell (Figures A.1, A.2; Table A.2, columns 1-3).
  - Broader measure of turnover from employment to non-employment (includes unemployment and inactivity) (Table A.2, column 5).
  - Alternative standard-error calculations, including clustering at different levels (Table A.3).
  - Alternative calculation of post-reform vs pre-reform change (difference of averages in 2023 and 2021 rather than 2023Q4 vs 2021Q4) (Tables A.2 column 4, A.4, A.5, A.6).
- The decline in the temporary employment share persists in almost all robustness exercises; statistical significance of other coefficients varies by specification.
  - Example: in Table A.4, when using average-2023 vs average-2021 changes, the positive impact on the share of permanent employment is no longer significant, while there is a significant effect on hours worked.

### Concluding evidence, planned extensions, and limitations
- Preliminary evidence suggests the 2021 reform reduced the temporary employment share, with less clear effects on overall employment stability; results remain preliminary.
- Planned extensions to reach a more definitive assessment:
  - Use natural turnover rates derived from US occupational data to address endogeneity concerns and exploit sector-occupation at the 3-digit level.
  - Run dynamic regressions to quantify how reform effects evolved through the post-reform period.
  - Use difference-in-differences estimates to produce back-of-the-envelope aggregate impact estimates (temporary share, permanent share, etc.).
- Current limitations include small sample sizes and restricted statistical power; a full paper with the above extensions will be published later.

### Potential positive side effects beyond measured outcomes
- By increasing contractual stability, FD contracts may reduce job and income uncertainty for FD workers (recall after inactivity; access to unemployment benefits).
- FD contracts may improve access to training (stronger job attachment), and enhance access to housing and credit if FD incomes are perceived as more stable than temporary incomes.
- These potential benefits require further analysis beyond the scope of the present paper.

### Other reform dimensions and policy implications
- Non-EPL elements of the 2021 reform (short-time work schemes, collective bargaining) will affect firms’ ability to respond to shocks and will be tested in downturns:
  - Few workers currently participate in short-time work schemes (ERTE and RED); RED implementing regulation exists but the financing fund has not been set up.
  - The reform restored the priority of sectoral agreements over firm agreements on wages and the indefinite extension of prior agreements (ultra-activity).
  - Going forward, preserving some wage flexibility at the firm level is important to allow wages and/or hours to respond to changing economic conditions.
- Policy priorities to complement contractual stability gains:
  - Boost active labor market policies and enhance coordination with unemployment insurance and assistance.
  - Advance the unemployment assistance reform to ensure protection while strengthening incentives for job take-up.
  - Continue increasing incentives for employers to create regular permanent contracts.

*Source: 1espea2024002 - 24.      The effect of the reform on the temporary employment share has been heterogeneous*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2024/english/1espea2024002.pdf_