## 1. Selected High-Frequency Indicators

## Source details

**Canonical URL:** [1. Selected High-Frequency Indicators](https://www.imf.org/-/media/files/publications/cr/2026/english/1andea2026002-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2026/english/1andea2026002-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2026/english/1andea2026002-source-pdf.pdf.json)

---

### Introduction and economic context
- Timely high-frequency indicators (tourism flows, mobility, energy usage, external demand proxies, sectoral activity measures) supplement delayed national accounts for monitoring the business cycle.
- Main GVA drivers: tourism, construction, real estate, and retail.
- Tourism and retail: number of overnight tourists increased in 2025 and retail sales in 2025 to date are broadly in line with past data.
- Imports related to retail trade and hospitality point to continued strength in GVA through the end of 2025; imports of construction goods are not highly correlated with GVA growth in the second half of 2025 but energy consumption in the sector suggests robust activity.
- Andorran GDP growth is highly correlated with growth in Spain and France (correlation coefficients of around 0.9).

### Methodology and data
- Framework: mixed-frequency dynamic factor model (DFM) estimated in state-space form, with initial indicator screening using Bayesian Model Averaging (BMA).
- COVID-aware adjustments:
  - Measurement equations augmented with pandemic dummies.
  - “Normal-times” uncertainty estimated from pre-COVID-19 periods.
- Estimation sample: 2015m01 to 2025m11.
- Data sources: Haver Analytics and Andorran Department of Statistics.
- Dataset composition:
  - 20 quarterly series and 90 monthly series (tourist flows, employment, prices, vehicle registrations, electricity consumption, external factors).
  - Quarterly series represented in monthly form via shape-preserving piecewise cubic interpolation (PCHIP) consistent with mixed-frequency framework.
- Transformations and state-space:
  - For levels: Δ12 log(z_t) = log(z_t) − log(z_{t−12}).
  - For rates/percent units: Δ12(z_t) = z_t − z_{t−12}.
  - Two latent factors F_t = (f_{1t}, f_{2t})′; baseline: AR(1) factors (A = diag(φ1, φ2)).
- PIP threshold: around 0.4 (slightly more inclusive than common 0.5) to retain indicators such as one-day visitors flows.

### Selection of leading indicators (BMA)
- BMA-selected indicators emphasize external demand and tourism dynamics plus domestic price and activity conditions.
- Posterior Inclusion Probabilities (PIP) examples:
  - Spain GDP growth: 1.00000
  - CPI Health services: 0.75125
  - Tourists: 0.68273
  - Hotel and restaurants electricity consumption: 0.52927
  - One-Day Visitors: 0.39211
- Spain GDP growth is consistently selected; France GDP growth is not picked by BMA.
- Indicator selection is vintage-dependent; BMA screening is diagnostic, not a fixed model choice.

### Nowcasting GDP growth — baseline and short-horizon forecasts
- Baseline nowcast for 2025Q4: 2.75 percent, year-on-year (yoy) (point estimate, mean).
  - 68 percent confidence bands: between 1.5 and 3.7.
  - 95 percent confidence bands: between 0 and 5 percent.
- Short-horizon forecast: 2026Q1 projected at around 2.6 percent yoy.
- Machine-learning NowHub comparison:
  - ML-based nowcast around 3 percent yoy for 2025Q4; RMSE of selected ML model is 1.40.
  - Results broadly consistent with baseline despite methodological differences.
- Robustness check:
  - Re-running with tourism data revised in February 2026 yields 2025Q4 nowcast at 2.8 percent yoy (these revised series were not available in the November 2025 vintage and thus are a robustness check, not real-time input).

### Validation and model comparison (horse race)
- Alternative checks carried out (examples summarized):
  - Check #1: include energy consumption of financial services.
  - Check #2a: include imports of materials for construction.
  - Check #2b: include construction imports of inputs and employment.
  - Check #2c: include only employment in construction.
  - Check #3: include employment in retail.
  - Check #4a: include Spain and France GDP growth.
  - Check #4b: include France GDP growth only.
  - Check #5: remove Spain GDP growth.
- Range of 2025Q4 nowcasts across checks: between 1.4 percent and around 3 percent yoy; baseline at 2.75 percent yoy.
  - Lowest (~1.4 percent) when excluding Spain GDP growth (Check #5) or with France GDP growth only (Check #4b).
  - Highest (~3 percent) when including construction or retail employment.
- RMSEs (monthly valuation) — horse race results (Model: RMSE (rank)):
  - Baseline (from BMA selection): 0.71 (1)
  - Check #1 (+energy consumption of financial services): 0.74 (3)
  - Check #2a (+only construction imports of inputs): 0.78 (4)
  - Check #2b (+construction imports of inputs and employment): 1.99 (6)
  - Check #2c (+only employment in construction): 0.74 (3)
  - Check #3 (+employment in retail): 0.73 (2)
  - Check #4a (+France GDP growth): 0.98 (5)
  - Check #4b (+France GDP growth – Spain GDP growth): 2.06 (7)
  - Check #5 (-Spain GDP growth): 2.06 (7)
- Key validation findings:
  - Baseline (BMA-selected set) yields best in-sample fit.
  - Including multiple construction proxies induces multicollinearity and worsens fit; current high-frequency indicators may underestimate construction’s short-term contribution.
  - Once Spain GDP growth is included, France adds little independent information and can worsen fit due to multicollinearity.

### Scenario analysis: Road closure at Pas de la Casa
- Scenario: temporary disruption closing the road at Pas de la Casa between January 31 and March 9, 2026.
- Assumed shock: cumulative decline of 10 percent in visitors in January and February 2026.
- Model implication: real GDP growth in 2026Q1 declines from 2.6 percent to 2.4 percent yoy (indicative; based on short-horizon forecast rather than a nowcast).

### Conclusions and actionable recommendations (nowcasting framework)
- A mixed-frequency nowcasting framework combining high-frequency indicators and historical national accounts:
  - Delivers timely, interpretable nowcasts of quarterly GDP growth with uncertainty bands.
  - Supports near-term scenario analysis.
- Spain GDP growth is a robust external anchor for Andorra’s short-term outlook and materially improves nowcast fit.
- Important domestic signals: tourist flows, hotel and restaurant electricity consumption, one-day visitors, and CPI for health services.
- Construction activity may be underrepresented by current high-frequency proxies; recommendation: improve high-frequency measurement for construction (labor and output proxies) to better capture sectoral dynamics.
- Nowcasts improve as monthly data are released; reasonably precise estimates are attainable 2–3 months ahead of official GDP releases.
- Scenario-analysis capability is useful for assessing short-run shocks (e.g., road closures) but results based on short-horizon forecasts should be interpreted as indicative.
- Replication package (annotated Stata code) available upon request to authorities to operationalize and update the framework.

*IMF staff analysis from "High-Frequency Indicators and GDP Nowcasting in Andorra" (selected chapter).*

### Andorra — Healthcare system: fiscal findings, projections, and policy priorities

### System overview and financing structure
- The Andorran public healthcare system delivers broad coverage, high-quality care, low household out-of-pocket spending, and low maternal and infant mortality.
- Key institutions: SAAS (Servei Andorrà d'Atenció Sanitària), CASS (Caixa Andorrana de Seguretat Social), Ministry of Health.
- Total current health spending: 8.3 percent of GDP.
- Financing composition:
  - One quarter of health financing from the central government budget.
  - 18 percent from employee and self-employed contributions.
  - 29 percent from employer contributions.
  - 29 percent from voluntary health insurance and household out-of-pocket payments.
- CASS General Fund financing:
  - Employee payroll contribution: 3 percent.
  - Employer payroll contribution: 7 percent.
  - CASS General Fund deficits are covered by the central government budget.
  - Tax financing to cover the deficit made up around one fifth of CASS health expenditure financing in 2024.
- Covered treatments subsidized at rates from 65 to 100 percent of total costs.
- Significant cross-border spending: €24.3m directly from CASS in 2023 for high complexity and high-cost care.

### Performance, workforce, and service delivery
- Life expectancy: 84 years.
- HAQ Index score: 89.1 out of 100 in 2019, ranking ninth in Europe.
- Older age group (65–74 years) HAQ score: 79.3.
- Number of doctors per capita close to peers; number of nurses lower (scope to enhance community and primary care staffing or task shifting).
- Preferred route since 2020: primary care doctors and nurses main entry point with lower co-payments than direct specialist access.
- E-health project supports better quality and coordination.

### Recent spending composition and drivers (2014–25)
- Government health spending doubled in nominal terms from €119.8 million in 2014 to €236.0 million in 2025.
- Comparison over same period: no change in real GDP per capita and a 26 percent increase in population.
- Decomposition of spending increase (approximate shares of total increase):
  - Inflation: 34 percent of total increase.
  - Excess cost growth: 38 percent of total increase.
- Major category increases 2014-25:
  - €37 million increase in payments to domestic providers.
  - €28 million increase in wages and salaries of SAAS employees.
  - €23 million increase in spending on pharmaceuticals and medical devices, including through SAAS.
- SAAS wage bill:
  - Grew from €34.8 million in 2016 to €62.5 million in 2025.
  - Contributors to increase:
    - Population growth: €8.2 million.
    - Inflation: €12.2 million.
    - Increase in health sector employment to population ratio: €5.3 million.
    - Increase in real average wages: €2.0 million.
  - Collective bargaining agreement for 2026-30 will raise SAAS wages by €11.5 million over 2026-30.
  - If SAAS wage bill had grown at population growth plus inflation, it would be spending €7 million less per year in 2025.
- Year-specific spikes:
  - 2020: 13.4 percent annual growth due to the COVID-19 pandemic.
  - 2024: 13.8 percent growth due in part to higher payments to domestic health providers from CASS.
- Pharmaceuticals and medical devices public spending: around €46 million in 2023.
- Sickness and disability payments: €77 million in 2024 (nearly one quarter of total General Branch of CASS spending).
- Sickness benefit recipients increased by more than 5 percent a year since 2016; sickness benefit spending grew similarly.

### Financing evolution 2014-25 (key figures)
- Central government budget financing for health:
  - €51.8 million in 2014 (2.1 percent of GDP) → €101.1 million in 2025 (2.6 percent of GDP).
  - Driven by a €28.5 million increase in the transfer to SAAS and a €17.5 million transfer to cover the CASS deficit.
- CASS General Fund spending:
  - €137.1 million in 2014 → €244.7 million in 2025.
  - Mostly financed from an increase in social security contributions for the General Fund rising from 4.2 percent to 5.3 percent of GDP (an increase of €102.3 million over the 11 years).

### Medium-term and long-term projections and fiscal implications
- Medium-term (2025-30) scenario assuming average annual growth 2014-25 persists:
  - Health spending rises from €236 million in 2025 to €312 million in 2030.
  - Government health spending would rise from 6.0 percent to 6.8 percent of GDP in 2030 (using IMF WEO nominal GDP projections).
- Immediate pressures 2026-30:
  - Expansion of SAAS hospital (capital and recurrent costs).
  - 2026-30 collective agreement: raises average remuneration for doctors by 25 percent and for nurses by 15 percent, increasing annual SAAS salary bill by €11.5 million by 2030.
- Long-term demographic projections (2025-50) and drivers:
  - Share of population aged over 65: 15 percent in 2025 → 25 percent by 2035.
  - On average, population over 65 have health costs around 3.8 times the population average.
  - IMF Fiscal Monitor methodology projections:
    - Government health spending projected to rise from 6.0 percent in 2025 to 7.4 percent in 2035.
    - Drivers:
      - Aging: adds 0.9 percentage points between 2025 and 2035, and a further 1 percentage point by 2045.
      - Excess cost growth: adds 0.5 percentage points over 2025-35 and 0.6 percentage points over 2035-45.
- Strategic choices to accommodate a 0.8 percent of GDP increase (6.0 → 6.8 percent between 2025–30):
  - Higher revenue, lower spending elsewhere, or higher deficits.
  - If financed entirely by social contributions: contributions rise from 10 percent of payroll in 2025 to 11.5 percent in 2030.
  - If prioritized over other government spending: implies €32 million less in spending in other areas (in 2025 GDP-equivalent terms).
  - Potential to increase private household financing in some areas (example: reduce subsidy to non-registered CASS providers).

### Efficiency opportunities, reform priorities, and implementation guidance
- Areas with room for efficiency savings:
  - Hospitals (hospitalization expenditures cited at €110 million+).
  - Pharmaceuticals and medical products (€46 million).
  - Sickness and disability payments (€66.2 million).
- Suggested actions:
  - Integrate the draft Health Sustainability Action Plan into budgeting; review provider payments and move away from fee-for-service models.
  - Update health nomenclature, tariff and treatment codes; revise referral codes for overseas treatments.
  - Promote treatment outside hospitals and improve chronic condition management.
  - Optimize workforce roles and manage wage/employment growth expectations.
  - Reduce waste, increase use of generics, and monitor hospital pharmaceuticals and medical devices use.
  - Improve information on cost and effectiveness of high-cost treatments before adding them to publicly covered schedules.
  - Tighten and earlier work capacity assessments and encourage return to work for sickness and disability cases; consider experience-rated employer premiums and third-party medical assessments as international lessons suggest.
- Budgetary practice recommendations:
  - Use a top-down anchor (e.g., align spending growth with GDP) to discipline expenditure and meet additional needs via efficiency gains.
  - Robust budget formulation with clear total budget allocation, multi-year budgeting, spending reviews, and corrective mechanisms (examples: alert committees and thresholds).
  - Complementary controls: Health Technology Assessments, price regulation, volume controls, and incentive mechanisms.
- Reform sequencing and stakeholder engagement are critical to durable success; avoid large year-to-year swings in health spending.

*Source: IMF staff chapter on Andorra’s health system and fiscal implications, extracted from the provided content unit.*

### 1. Selected High-Frequency Indicators __________________________________________________ 5

### 1. Selected High-Frequency Indicators

### Introduction
- Timely information on economic activity is essential for monitoring the business cycle; national accounts are released with a delay while many high-frequency indicators (tourism flows, mobility, energy usage, external demand proxies, sectoral activity measures) are available at higher frequency and shorter lags.
- Main drivers of Andorra’s economy on a gross value-added (GVA) basis include tourism, construction, real estate, and retail.
- Tourism and retail: number of overnight tourists increased in 2025 and retail sales in 2025 to date are broadly in line with past data.
- Imports related to retail trade and hospitality point to continued strength in GVA through the end of 2025; imports of construction goods are not highly correlated with GVA growth in the second half of 2025 but energy consumption in the sector suggests robust activity.
- Andorran GDP growth is highly correlated with growth in Spain and France (correlation coefficients of around 0.9).

### Methodology
- Framework: mixed-frequency dynamic factor model (DFM) estimated in state-space form, with initial indicator screening using Bayesian Model Averaging (BMA).
- COVID-aware adjustments: measurement equations augmented with pandemic dummies and estimation of “normal-times” uncertainty restricted to pre-COVID-19 periods.
- Estimation sample: 2015m01 to 2025m11.
- Data sources: Haver Analytics and Andorran Department of Statistics.
- Dataset composition: 20 quarterly series and 90 monthly series (tourist flows, employment, prices, vehicle registrations, electricity consumption, external factors).
- Quarterly series are represented in monthly form consistent with the mixed-frequency framework.
- PIP threshold: around 0.4 (slightly more inclusive than the common 0.5) to retain indicators such as one-day visitors flows.

### Selection of Leading Indicators
- BMA-selected indicators emphasize external demand and tourism dynamics plus domestic price and activity conditions.
- Spain GDP growth is consistently selected in all specifications; France GDP growth is not picked by BMA.
- Electricity consumption in hotels and restaurants and CPI for health services enter among selected series.
- Table of Posterior Inclusion Probabilities (PIP):
  - Spain GDP growth: 1.00000
  - CPI Health services: 0.75125
  - Tourists: 0.68273
  - Hotel and restaurants electricity consumption: 0.52927
  - One-Day Visitors: 0.39211
- Indicator selection is vintage-dependent; BMA screening is diagnostic, not a fixed model choice.

### Nowcasting GDP Growth
- Baseline nowcast for 2025Q4: 2.75 percent, year-on-year (yoy) (point estimate, mean).
  - 68 percent confidence bands: between 1.5 and 3.7.
  - 95 percent confidence bands: between 0 and 5 percent.
- Short-horizon forecast: 2026Q1 projected at around 2.6 percent yoy.
- Machine-learning NowHub comparison: ML-based nowcast around 3 percent yoy for 2025Q4; RMSE of selected ML model is 1.40. Results broadly consistent with baseline despite methodological differences.
- Robustness check: re-running with tourism data revised in February 2026 yields 2025Q4 nowcast at 2.8 percent yoy (these revised series were not available in the November 2025 vintage and thus are a robustness check, not real-time input).

### Validation and Model Comparison (Horse Race)
- Alternative checks undertaken:
  - Check #1: include energy consumption of financial services.
  - Check #2a: include imports of materials for construction.
  - Check #2b: include construction imports of inputs and employment.
  - Check #2c: include only employment in construction.
  - Check #3: include employment in retail.
  - Check #4a: include Spain and France GDP growth.
  - Check #4b: include France GDP growth only.
  - Check #5: remove Spain GDP growth.
- Range of 2025Q4 nowcasts across checks: between 1.4 percent and around 3 percent yoy; baseline at 2.75 percent yoy.
  - Lowest estimates (~1.4 percent) occur in models excluding Spain GDP growth (Check #5) or with France GDP growth only (Check #4b).
  - Highest estimates (~3 percent) when including construction or retail employment.
- RMSEs (monthly valuation) — horse race results (Model followed by RMSE and rank in parentheses):
  - Baseline (from BMA selection): 0.71 (1)
  - Check #1 (+energy consumption of financial services): 0.74 (3)
  - Check #2a (+only construction imports of inputs): 0.78 (4)
  - Check #2b (+construction imports of inputs and employment): 1.99 (6)
  - Check #2c (+only employment in construction): 0.74 (3)
  - Check #3 (+employment in retail): 0.73 (2)
  - Check #4a (+France GDP growth): 0.98 (5)
  - Check #4b (+France GDP growth – Spain GDP growth): 2.06 (7)
  - Check #5 (-Spain GDP growth): 2.06 (7)
- Key validation findings:
  - Baseline (BMA-selected set) yields best in-sample fit.
  - Including multiple construction proxies induces multicollinearity and worsens fit; current high-frequency indicators may underestimate construction’s short-term contribution.
  - Once Spain GDP growth is included, France adds little independent information and can worsen fit due to multicollinearity.

### Scenario Analysis: Road Closure at Pas de la Casa
- Scenario: temporary disruption closing the road at Pas de la Casa between January 31 and March 9, 2026.
- Assumed shock: cumulative decline of 10 percent in visitors in January and February 2026.
- Model implication: real GDP growth in 2026Q1 declines from 2.6 percent to 2.4 percent yoy (indicative; based on short-horizon forecast rather than a nowcast).

### Conclusions and Recommendations (summary of actionable points)
- A mixed-frequency nowcasting framework combining high-frequency indicators and historical national accounts delivers timely, interpretable nowcasts of quarterly GDP growth with uncertainty bands and supports near-term scenario analysis.
- Spain GDP growth is a robust external anchor for Andorra’s short-term outlook and materially improves nowcast fit.
- Tourism-related indicators (tourist flows, hotel and restaurant electricity consumption, one-day visitors) and CPI for health services are important domestic signals for short-term activity.
- Construction activity may be underrepresented by current high-frequency proxies; consider improving high-frequency measurement for construction (labor and output proxies) to better capture sectoral dynamics.
- Nowcasts improve as monthly data are released; reasonably precise estimates are attainable 2–3 months ahead of official GDP releases.
- Scenario-analysis capability is useful for assessing short-run shocks (e.g., road closures) but results based on short-horizon forecasts should be interpreted as indicative.

*IMF staff analysis from "High-Frequency Indicators and GDP Nowcasting in Andorra" (selected chapter).*

### 16.      This paper develops a mixed-frequency nowcasting framework to provide a timely

### This paper develops a mixed-frequency nowcasting framework to provide a timely estimate of quarterly GDP growth in Andorra using high-frequency indicators

### Nowcasting framework and main findings
- The framework provides a transparent real-time estimate of quarterly GDP ahead of its release.
- Results confirm the central role of tourism-related variables and external conditions—particularly Spain’s GDP growth—in explaining short-term fluctuations in activity.
- The framework delivers nowcasts by combining a monthly GDP proxy (constructed via shape-preserving piecewise cubic interpolation, PCHIP) with selected monthly indicators in a mixed-frequency dynamic factor model (DFM).

### Data limitations and sectoral coverage
- Important data limitations are highlighted, especially in capturing construction activity at high frequency.
- Available proxies only partially reflect sectoral dynamics: they may not fully capture labor and imported inputs or may induce multicollinearity when combined, implying the construction sector’s contribution to growth may be underestimated.
- Recommendation: Improving the availability and coverage of high-frequency indicators for the construction sector would help better reflect its contribution to short-term GDP dynamics and strengthen real-time monitoring.

### Replication and operationalization
- A replication package is available upon request and will be made available to the authorities.
- The package includes an annotated Stata code covering data preparation, the BMA indicators selection, and the estimation of the mixed-frequency dynamic factor model, allowing the framework to be easily updated as new data becomes available and adapted by the authorities for ongoing monitoring.

### Econometric methodology — overview (four main steps)
- 1) Data preparation
  - All series aligned to a monthly calendar t = 1, . . . , T and cleaned to (i) ensure stationarity, (ii) harmonize units across indicators, and (iii) accommodate mixed-frequency information in a single monthly state-space representation.
  - Transformations:
    - For indicators in levels: year-on-year (yoy) log differences: Δ12 log(z_t) = log(z_t) − log(z_{t−12}).
    - For indicators in rates or percentage units: yoy simple differences: Δ12(z_t) = z_t − z_{t−12}.
  - Quarterly-to-monthly mapping for GDP: quarterly GDP yoy growth is mapped to the monthly grid using PCHIP to produce a smooth monthly series that preserves quarter-to-quarter profiles implied by observed quarterly values.
- 2) Bayesian Model Averaging (BMA)
  - BMA addresses model uncertainty by averaging over a collection of regression models formed by different combinations of candidate indicators using posterior model probabilities (PIP).
  - Let y_t denote GDP growth and X_t = (x_{1t}, ..., x_{Kt}) the set of K candidate indicators. Each model M_j corresponds to a subset X_{j,t} and is written as y_t = α_j + X_{j,t}′ β_j + ε_{j,t}, ε_{j,t} ∼ N(0, σ_j^2).
  - Indicators with PIPs exceeding a pre-specified threshold are retained and used as inputs to the DFM to promote parsimony and mitigate overfitting.
- 3) Nowcasting
  - Let I_t denote the information set at month t. The h-step-ahead nowcast/forecast for GDP growth yoy is ŷ_{t+h|t} = E(y_{t+h} | I_t) computed via the Kalman filter and smoother implied by the state-space system.
  - Observed monthly indicators vector Y_t (l×1) and latent common factors F_t = (f_{1t}, f_{2t})′ (2×1) are modeled with a two-factor specification (chosen for best nowcasting performance while maintaining parsimony given the limited time dimension).
  - State-space form:
    - Y_t = Λ F_t + ε_t, ε_t ∼ (0, Σ_ε)
    - F_t = A F_{t−1} + η_t, η_t ∼ (0, Q)
    - Baseline: each factor follows an AR(1) so A = diag(φ1, φ2).
- 4) COVID-19 correction and uncertainty
  - A dummy D_t is included in the measurement equation: Y_t = Λ F_t + Γ D_t + ε_t to account for COVID-19 distortions.
  - Confidence bands are calibrated from pre-COVID volatility to avoid pandemic-driven inflation of forecast-error dispersion.
  - For “normal-times” uncertainty, the model is (re-)estimated excluding a COVID-19 window.

### Model implementation details and choices
- Kalman filter advantages: accommodates missing observations typical of real-time datasets and updates the GDP nowcast as new monthly releases arrive.
- Two-factor DFM chosen because:
  - Delivers best nowcasting performance compared to a single factor.
  - Additional factors do not materially improve nowcasting performance but substantially increase parameter uncertainty (reference: Bańbura and Modugno, 2014).

### Dataset summary
- Quarterly series: 20 series including EA, Spain, France GDP and HICP; business confidence by sectors; value of completed real estate transactions and transfers.
- Monthly series: 90 series including merchandise imports by main sectors, retail trade, employment by sectors, CPI by sectors, salary by sectors, vehicle registrations, electricity consumption by sectors, number of overnight tourists and visitors, number of nights spent by overnight tourists.
- Selected dataset items (examples preserved exactly as listed):
  - Andorra: Real GDP growth — Quarterly
  - Andorra Hotel Occupancy — Monthly
  - Andorra: Merchandise imports, aggregate construction — Monthly
  - Andorra: Retail trade Index general (nominal) — Monthly
  - Andorra: Mortgages granted — Monthly-semiannual
  - Andorra: Employees: Construction (NSA, Persons) — Monthly
  - Andorra: CPI (NSA, 2021=100) — Monthly
  - Andorra: Vehicle Registrations (NSA, Units) — Monthly
  - Andorra: Average Tourist Nights Spent (Nights) — Monthly
  - Andorra: Real Estate Trans: Flats, Detached Houses & Buildings (NSA, Number) — Quarterly
  - External data examples:
    - EA20: HICP: Monetary Union Index: Consumer Prices (NSA, 2015=100) — Monthly
    - Spain: Gross Domestic Product (NSAWDA, Mil.Chn.2020.EUR) — Quarterly
    - France: Gross Domestic Product (NSWDA, Mil.Ch.20.EUR) — Quarterly

### Healthcare system — key fiscal findings and policy priorities (introductory summary)
- Andorra has a health system delivering strong outcomes with lower public spending than average EU countries; life expectancy is among the highest globally and the system responded effectively to the pandemic.
- Government health spending increased from 3.5 percent of GDP in 2003 to 6.0 percent in 2025.
- Without reform, spending is projected to continue increasing as a share of GDP over the medium-term, placing increasing strain on public finances.
- Policy priority: balance accommodating health spending growth with tighter management of high-cost areas.
  - Aim: slow health spending growth to around the rate of population growth and inflation to limit increases in social contributions and household financing and preserve fiscal space for other priorities (such as housing and infrastructure).
  - Suggested reforms include: strengthening administrative controls, setting clearer medium-term targets, identifying concrete savings measures, implementing payment reforms to improve efficiency, and tightening eligibility and oversight of sickness benefits.

*International Monetary Fund — material from the provided content unit*

### 4.      The Andorran public healthcare system provides broad coverage. The system delivers

### 4.      The Andorran public healthcare system provides broad coverage. The system delivers 

### Overview and institutional roles
- The system delivers high-quality care to the population with extensive access, low household out-of-pocket spending, and low maternal and infant mortality.
- Key institutions that finance, manage and deliver healthcare:
  - National Health Service (Servei Andorrà d'Atenció Sanitària or SAAS)
  - Social Security Agency (Caixa Andorrana de Seguretat Social or CASS)
  - Ministry of Health
- Strong coordination across budgeting and financing, management, and policy functions is required.

### Financing structure and key shares
- Total current health spending is 8.3 percent of GDP financed from taxation, social contributions and private financing.
- Financing composition:
  - One quarter of health financing is from the central government budget.
  - 18 percent from employee and self-employed contributions.
  - 29 percent from employer contributions.
  - 29 percent from voluntary health insurance and household out-of-pocket payments.
- Household out-of-pocket costs are low compared to European peers (Litvinova and Rechel, 2026).
- Central government budget: four main health spending components:
  1. Transfer to the SAAS, particularly for the national hospital.
  2. Transfers to the CASS, including transfers to cover the health fund deficit.
  3. Transfers to the University of Andorra.
  4. Health spending directly by the Ministry of Health, including public health programs on primary and community care.

### CASS role, benefits and financing mechanics
- CASS General Fund covers:
  1. Direct transfers to the national hospital for services.
  2. Payments to domestic providers for services.
  3. Payments to foreign providers.
  4. Health related benefits and pensions.
  5. Reimbursements for health-related goods.
- In general, covered treatments are subsidized at rates from 65 to 100 percent of total costs.
- Low-income households can receive government support to cover additional costs; most other residents opt for voluntary health insurance to cover remaining costs.
- CASS structure and contributions:
  - Two parts: the Pension Fund and the General Fund (the latter includes health spending) which operate financially independently.
  - CASS General Fund is financed by a 3 percent payroll contribution from employees and a 7 percent payroll contribution by employers (as well as self-employed and government contributions).
  - CASS General Fund deficits are covered by the central government budget.
  - In practice, tax financing to cover the deficit made up around one fifth of CASS health expenditure financing in 2024.

### Service delivery, external provider relationships, and subsidies
- Health services are delivered through the SAAS, Ministry of Health, and the National Health Service, including the national hospital, health system stewardship and capacity building.
- Financing is provided to the Andorra University for training for nurses.
- Providers (physicians, pharmacists and other health providers) are a combination of public and private provision.
- Most providers are registered with CASS and so subject to a higher subsidy rate; subsidies for non-registered CASS providers are subsidized at 20 percent of costs.
- Andorra has agreements with neighboring countries to provide specialized care; significant health spending goes to neighboring Spanish and French health systems including on high complexity and high-cost care (€24.3m directly from CASS in 2023).

### International comparisons and performance indicators
- Government health spending in Andorra is comparable to or below peer countries:
  - Comparable to the 6.1 percent average in advanced economies and 6.9 percent average in the EU in 2023 (staff calculations based on budget estimates, WEO data sources and WHO data).
  - Government Finance Statistics may show somewhat higher spending due to inclusion of some privately financed but publicly managed health spending.
- Health outcomes and efficiency:
  - Andorra has one of the highest life expectancies in the world at 84 years.
  - HAQ Index score: 89.1 out of 100 in 2019, ranking ninth in Europe.
  - Older age group (65–74 years) HAQ score: 79.3.
  - Rising burden of disease associated with alcohol consumption, smoking and increasing levels of obesity (Litvinova and Rechel, 2026).
- Financing mix:
  - Andorra relies more on social insurance and private pre-paid financing than its peers; social contributions managed by the social insurance fund are more significant.
- Health workforce:
  - Number of doctors per capita is close to peers; number of nurses is lower.
  - Low number of nurses per capita suggests scope to enhance resourcing of community and primary healthcare and/or task shifting towards nurses and community workers.
  - Preferred route for accessing health services introduced in 2020: primary care doctors and nurses are main entry point and offer patients lower co-payments compared to accessing specialist care directly.
  - Andorra’s e-health project supports better quality and more coordinated care.

### Spending composition and drivers (recent years)
- Pharmaceuticals and medical devices:
  - Public spending on pharmaceuticals and medical products including hospital pharmacy and medical device reimbursement was around €46 million in 2023.
  - Andorra sources pharmaceuticals and medical goods from France and Spain.
- Sickness and disability payments:
  - Spending was €77 million in 2024 or nearly one quarter of total spending by the General Branch of CASS.
  - Spending on sickness and disability benefits in Andorra is higher than in Portugal and Italy, but lower than in Spain and France.
- Growth 2014-25:
  - Government health spending doubled in nominal terms from €119.8 million in 2014 to €236.0 million in 2025.
  - Comparison over same period: no change in real GDP per capita and a 26 percent increase in population.
  - Approximately half of the remaining increase in health spending can be attributed to inflation (34 percent of total increase) and another half to excess cost growth (38 percent).
  - Major category increases 2014-25:
    - €37 million increase in payments to domestic providers.
    - €28 million increase in wages and salaries of SAAS employees.
    - €23 million increase in spending on pharmaceuticals and medical devices, including through SAAS.
- SAAS wage bill:
  - SAAS salaries grew from €34.8 million in 2016 to €62.5 million in 2025.
  - Contributors to this increase:
    - Population growth: €8.2 million.
    - Inflation: €12.2 million.
    - Increase in health sector employment to population ratio: €5.3 million.
    - Increase in real average wages: €2.0 million.
  - Collective bargaining agreement for 2026-30 will raise SAAS wages by €11.5 million over 2026-30.
  - If SAAS wage bill had grown at population growth plus inflation, it would be spending €7 million less per year in 2025.
- Year-specific spikes:
  - 2020: 13.4 percent annual growth due to the COVID-19 pandemic.
  - 2024: 13.8 percent growth due in part to higher payments to domestic health providers from CASS.

### Sickness benefits trend and fiscal risk
- Number of sickness benefit recipients increased by more than 5 percent a year since 2016.
- Sickness benefit spending grew by a similar amount as spending per beneficiary grew close to inflation.
- Sickness and disability spending is a key driver of the General Fund deficit, which is tax-financed.
- Historical note: 2015 reform to the disability pension changed the pension point calculation, lowering the level of disability benefits and reducing incentives for take-up.

### Financing evolution 2014-25
- Central government budget financing for health:
  - Rose from €51.8 million in 2014 (2.1 percent of GDP) to €101.1 million in 2025 (2.6 percent of GDP).
  - Driven by a €28.5 million increase in the transfer to SAAS and a €17.5 million transfer to cover the CASS deficit.
- CASS General Fund spending:
  - Increased from €137.1 million in 2014 to €244.7 million in 2025.
  - Mostly financed from an increase in social security contributions for the General Fund rising from 4.2 percent to 5.3 percent of GDP (an increase of €102.3 million over the 11 years).

### Medium-term what-if scenario (2025-30) and drivers
- Scenario assumption: average annual growth over 2014-25 continues through 2025-30.
- Under this assumption:
  - Health spending would rise from €236 million in 2025 to €312 million in 2030.
  - Using IMF WEO projections for nominal GDP, government health spending would rise from 6.0 percent to 6.8 percent of GDP in 2030.
- Immediate pressures in projection period:
  - Expansion of the SAAS hospital (potential significant capital expenditure and higher recurrent spending).
  - Rising SAAS wages: 2026-30 collective agreement will raise average remuneration for doctors by 25 percent and for nurses by 15 percent, increasing the annual SAAS salary bill by €11.5 million by 2030.

### Long-term demographic pressures and projections (2025-50)
- Population aging:
  - Share of population aged over 65 years expected to rise from 15 percent in 2025 to 25 percent by 2035.
  - On average, population over 65 years have health costs around 3.8 times the population average.
- Longer-term projections using IMF Fiscal Monitor methodology:
  - Government health spending projected to rise from 6.0 percent in 2025 to 7.4 percent in 2035.
  - Drivers:
    - Aging: a 0.9 percentage point increase between 2025 and 2035, and a further 1 percentage point by 2045.
    - Excess cost growth: a 0.5 percentage point increase over 2025-35 and an additional 0.6 percentage points over 2035-45.
- These projections separate demographic effects from excess cost growth and are illustrative rather than deterministic.

### System adaptation needs and reform context
- An aging population will increase demand for geriatric care, dementia care and support for daily tasks.
- Importance of promoting healthy aging and preventative, primary and community care.
- Andorra has started reforms focusing on aging well and preventative, primary and community care (see Bibolov, 2025).

*Source: IMF staff chapter on Andorra’s health system and fiscal implications, extracted from the provided content unit.*

### 26.      Government health spending will face continued pressure from aging and other cost

### 26.      Government health spending will face continued pressure from aging and other cost

### Projections and fiscal implications
- Government health spending could rise from 6.0 percent of GDP in 2025 to 6.8 percent in 2030 (a 0.8 percent of GDP increase).
- Strategic choice on accommodating the 0.8 percent of GDP growth: higher revenue, lower spending elsewhere, or higher deficits.
- If additional government health spending were entirely financed by social contributions:
  - Contributions would rise from 10 percent of payroll in 2025 to 11.5 percent in 2030.
- If government health spending were prioritized over other government spending:
  - This would imply €32 million less in spending in other areas (in 2025 GDP-equivalent terms).
- There may be scope to increase private household financing in some areas (example: reducing the subsidy to health providers not registered with CASS).

### Medium-term budgeting and fiscal sustainability
- Health spending is budgeted using a bottom-up assessment of sector needs, but must be aligned with underlying cost drivers and available sustainable financing.
- A top-down anchor—such as aligning spending growth with GDP—could help discipline expenditure, with additional needs met through efficiency gains.
- General public financial management practices can support health spending management (see Box 1 in source).

### Efficiency opportunities and targeted areas for savings
- Areas with room for efficiency savings:
  - Hospitals (hospitalization expenditures cited at €110 million+).
  - Pharmaceuticals and medical products (€46 million).
  - Sickness and disability payments (€66.2 million).
- Suggested actions to improve efficiency:
  - Integrate and implement the draft Health Sustainability Action Plan into the budget; actions include reviewing provider payments and moving away from fee-for-service payment models, updating health nomenclature and tariff and treatment codes, and revising referral codes for overseas treatments.
  - Address hospitalization expenditures through:
    - Promoting treatment outside hospitals.
    - Improving management of chronic conditions.
    - Optimizing workforce roles.
    - Managing wage and employment growth expectations.
    - Reducing waste.
  - Improve information on cost and effectiveness of high-cost treatments (including considering affordability) when deciding additions to the publicly covered treatments schedule.
  - Continue initiatives to reduce waste and improve use of generics and increase monitoring of hospital pharmaceuticals and medical devices use.
  - Stricter and earlier work capacity assessments and encouragement to return to work for sickness and disability cases; lessons noted from other countries (Box 2).

### Budgetary practices and controls (Box 1 highlights)
- Robust budget formulation sets a clear total budget allocation for public spending on health to provide predictability and cost control.
- Total budgets for health typically derive from baseline estimates for maintaining existing services; costs of new initiatives estimated separately.
- Efficiency dividends can be based on top-down constraints or bottom-up policy changes; spending reviews help identify savings.
- Multi-year budgeting improves certainty but may reduce Ministry of Finance flexibility.
- Corrective mechanisms detect overspending and identify actions (example: Alert Committee and thresholds—France alerts when evolution of health expenditures risks exceeding the ceiling by more than 0.5 percent).
- Complementary spending controls and incentives include Health Technology Assessments, price regulation, volume controls and incentive mechanisms.

### Reform design, implementation, and lessons (Box 2 highlights)
- Durable success depends on sequencing, stakeholder engagement, and avoiding large year-to-year swings in health spending.
- International lessons on sickness and disability benefits:
  - Sweden and Switzerland: mandatory payout initially set at 80 percent of previous salary.
  - Netherlands: payout set at 70 percent of previous salary for two years.
  - Sweden’s reforms indicate benefit level and focus on work participation influence take-up; small changes in payment rate affect outcomes.
  - The Netherlands: employer experience-rated premiums led to drastic falls in new benefit claims.
  - Switzerland: early identification and early intervention reduce disability claims; shifting assessment from general practitioners to a public medical authority and third-party medical assessments are feasible and effective.

*Source: Chapter text from the provided IMF content.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2026/english/1andea2026002-source-pdf.pdf_