## 3. MULTI-SPEED SECTORAL RECOVERY AND REALLOCATION POTENTIAL

## Source details

**Canonical URL:** [3. MULTI-SPEED SECTORAL RECOVERY AND REALLOCATION POTENTIAL](https://www.imf.org/-/media/files/publications/reo/eur/2021/october/english/ch3annex.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/reo/eur/2021/october/english/ch3annex.pdf.md)
- [Structured JSON version](/-/media/files/publications/reo/eur/2021/october/english/ch3annex.pdf.json)

---

### Online Annex 3.1. Expected Earnings and Sectoral GVA Growth Projection
- Data
  - Expected earnings: Institutional Brokers Estimate System (IBES) database; institutional brokers’ forecasts for listed firms from the current year to five years in the future.
  - Sector-level expected earnings: aggregated across all available European firms in IBES (more than 9,000 in total), using market capital weighted medians.
  - Sectoral GVA: annualized from quarterly GVA from Eurostat and national statistical agencies (for most countries, only the 2020 quarterly outturns are available).
  - Forecasts produced for two country groups: Advanced Europe and Emerging Market Europe (to ensure sample size and representativeness). Countries with relatively large numbers of listed firms—CHE, DEU, ESP, FRA, GBR, ITA, NOR, POL, ROU, RUS, SWE, and TUR—are also estimated individually.
  - Annualized expected earnings growth is presented relative to the reported 2019 level.
- Empirical strategy
  - Sectoral GVA growth is regressed on earnings for each of the 11 sectors, s:
    - Growth_st = α_s + β_s Earnings_st + ε_st
  - Earnings are detrended using HP filtering.
  - Using the estimated historical relationship, the medium-term path of GVA growth is projected based on the five-year IBES earnings forecasts.
- Main findings
  - Constructed sectoral earnings (HP filter detrended) are closely correlated with GVA growth (example sectors: manufacturing and contact-intensive sectors).
  - Multi-speed recovery: divergence in sectoral earnings prospects implies contact-intensive sectors (trade/transport/hospitality) are expected to lag and lose GVA share, while less-contact intensive sectors are likely to revert to pre-COVID-19 tracks.
  - ICT is expected to gain significant share compared to its pre-COVID trend, benefiting from pandemic-triggered technology advancement.
  - Over the medium-term, dispersion in growth rates across sectors may remain notably above the pre-COVID level.
- Robustness checks
  - Results robust to: using option-implied expected rate of return one-year ahead instead of IBES earnings forecasts, and using nominal GVA forecasts from Fitch Analysis.

### Online Annex 3.2. Sectoral Employment Projections
- Data
  - Quarterly sectoral GVA and employment (employed persons) from Eurostat and national statistical agencies.
  - Analysis covers 10 NACE sectors: agriculture (A); mining, utility, and energy (B, D, E); manufacturing (C); construction (F); trade, transportation, food and accommodation (G, H, I); information and communication (J); finance and insurance (K); real estate (L); professionals and administrative supports (M, N); public services (O, P, Q); and arts, entertainment and other services (R, S, T, U).
  - Countries included: AUT, BEL, BGR, CHE, CYP, CZE, DEU, DNK, EST, ESP, FIN, FRA, GBR, IRL, HRV, HUN, ITA, LTU, LVA, LUX, MLT, NLD, NOR, POL, PRT, SWE, SVK, SVN, and ROU.
  - Skill groups: low-, medium-, and high-skilled based on ISCO occupational classification (managers and professionals (ISCO 1 and 2) as high skill; laborers/elementary (ISCO 9) as low skill; other skilled workers (ISCO 3–8, 10) as medium skill).
- Empirical strategy: sectoral Okun’s law (dynamic specification)
  - Dynamic specification estimated at country-sector level for the 20-year period prior to COVID-19 (1999-2019).
  - Equation (as presented):
    - logEmpl_qsc̃ = α_sc + Σ_j= q−2^q β_jsc logGVA_jsc̃_q + Σ_j= q−2^q γ_jsc logEmpl_jsc̃_q−1 + η_qsc
  - cyclical components (logX̃) are HP-filtered logarithms.
  - Dynamic coefficient:
    - β_scdyn = (Σ_j= q−2^q β_jsc) / (1 − Σ_j= q−2^q γ_jsc)
  - Long-term employment elasticities are high in manufacturing and construction activities, as well as trade and other hospitality-related services.
  - Results are broadly robust across countries in terms of cross-sector relativity.
- Alternative scenarios for labor reallocation needs
  - Scenario 1: Larger sectoral demand shocks from permanent shifts in consumer and worker preferences (towards at-home or online activities). Additional final demand changes mapped to sectoral GVA through country-specific input-output tables; resulting sectoral employment changes calculated.
  - Scenario 2: Acceleration of automation altering output-employment relationships. Occupational probabilities of being computerized (Frey and Osborne, 2017) are used to compute the share of jobs in each sector that could be replaced by robots/computers based on occupation-specific probabilities and occupational composition by sector.
- Robustness
  - Results robust to specification changes, including first differences and percentage change estimations.

### Online Annex 3.3. Labor Market Dynamics and Reallocation
- Data and reallocation index
  - Reallocation measured by dispersion in sectoral stock returns (consistent with literature).
  - Sector stock returns: FTSE indices for around 40 sectors (Industry Classification Benchmark (ICB)), across 22 European countries.
  - Countries included: AUT, BEL, CHE, CZE, DEU, DNK, ESP, FIN, FRA, GBR, GRC, HUN, IRL, ISR, ITA, NLD, NOR, POL, PRT, RUS, SWE, and TUR.
  - Dispersion proxied by standard deviation of return rates across sectors, weighted by market capitalization at daily frequency and collapsed to quarterly frequency (excluding outliers).
  - Real wage measured by compensation per employee deflated by CPI inflation.
- Empirical strategy (two steps)
  - Step 1: Purge changes in labor market dynamics explainable by standard macro relationships.
    - For employment: Okun’s Law; for wage: Phillips Curve. Regressions estimated country-by-country.
    - Example formulations:
      - ΔlnE_it ~ GDP growth, 2 lags of (ΔlnE_it, GDP growth)
      - ΔlnW_it ~ unemployment gap, 2 lags of (ΔlnW_it, unemployment gap)
  - Step 2: Residualized employment and wage growth (parts unexplained by standard macro variables) are regressed on the reallocation index in a panel regression with country fixed effects, for horizons up to 20 quarters:
    - ε_it+h = α_i + β_h × Reallocation_it + δ_it+h,  h = 0,1,...,20
- Robustness checks
  - Variations include: different number of lags in step 1; allowing Okun/Phillips coefficients to differ across normal and recession years; purging stock returns movements explainable by CAPM before calculating sectoral dispersion; controlling for global uncertainty; excluding COVID-19 crisis quarters. None change qualitative results.
- Sample period
  - Starts from 1995Q1 or earliest quarter available for each country.

### Online Annex 3.4. Reallocation-Enhancing Policies in Response to the COVID-19 Crisis
- Overview
  - European countries are implementing a wide range of measures to facilitate worker reallocation and prepare the workforce for a rapidly changing technological and green transition environment, reflected in Recovery and Resilience Plans and other announcements.
  - Policy focus is shifting from lifelines to households and firms toward facilitating resource reallocation and easing adjustment costs as the recovery strengthens.
- Policy categories (examples and uptake)
  - i) Hiring subsidies and wage-loss insurance
    - Hiring subsidies implemented by 35 of the 50 European countries surveyed; some target vulnerable groups (e.g., youth workers in France and the U.K.) and/or vulnerable sectors.
  - ii) Incentives to invest
    - Investment incentives focus on green technology, digitalization, healthcare, and R&D (examples: Climate Fund in Switzerland; “Future Package” in Germany).
  - iii) ALMP: training and reskilling
    - Emphasis on reskilling for future needs: enhancing digital preparedness and technologies in training (Czech Republic, Italy); identifying future occupations (Turkey); skill and training needs (Slovenia); expanding training programs to support digital and green transition (Spain).
  - iv) ALMP: education
    - Broader education curriculum reforms to create a more adaptable workforce (example: Slovak Republic).
  - v) ALMP: job search
    - Job search support examples: career planning services, improving employment centers’ information systems, strengthening coordination between regional and national systems (Slovenia, Italy).
  - vi) Other policies
    - Labor and product market reforms; insolvency regime reforms to facilitate early restructuring and ease access to judicial reorganization (examples of countries implementing insolvency reforms: Belgium, Estonia, France, Hungary, Latvia, Malta, the Slovak Republic, and Spain).
- Implementation note
  - Proper implementation, careful monitoring, and evaluation of these initiatives could yield important lessons for policymakers to continue supporting reallocation and strengthen the recovery.

*International Monetary Fund, Online Annexes to Chapter 3, “Multi-Speed Sectoral Recovery and Reallocation Potential.”*

---


_Source: https://www.imf.org/-/media/files/publications/reo/eur/2021/october/english/ch3annex.pdf_