## 1. Productivity and Employment in Market Services

## Source details

**Canonical URL:** [1. Productivity and Employment in Market Services](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020039-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020039-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020039-print-pdf.pdf.json)

---

### Key facts and sectoral importance
- Market services defined as nonfinancial and nonreal estate market services: wholesale and retail trade, transportation, accommodation and food services, information and communication technologies (ICT), professional and administrative services, arts and entertainment, and other market services.
- 2015 shares:
  - 36 percent of total value-added,
  - 47 percent of total hours worked,
  - 46 percent of employment.
- Major employers:
  - Wholesale and retail trade: 16 percent of economy-wide employment,
  - Professional services: 12 percent of economy-wide employment.

### Recent productivity trends (2000–2015)
- Aggregate labor productivity of market services in Italy fell by over 5 percent between 2000 and 2015; remained relatively stable in the euro area.
- Sub-sectoral productivity declines between 2000 and 2015:
  - Professional services: 30 percent decline,
  - Accommodation and food services: 20 percent decline,
  - Other services: 20 percent decline.

### Integration with manufacturing
- Services in Italy are more tightly integrated to the manufacturing sector than in other European countries.

### Theoretical framework and model properties
- Model type and references:
  - Monopolistic competition model with heterogeneous firms and a trans-log demand function; draws on Felix and Maggi (2019) and Bergin and Feenstra (2000).
- Core assumptions:
  - Firms heterogeneous in productivity; labor is the only input.
  - Aggregate labor supply increasing in real wages; lower prices increase aggregate labor supply and aggregate output.
  - Two equilibrium properties of the trans-log demand:
    - Average markups are decreasing in the number of active firms.
    - Firm-level markups are increasing in a firm’s productivity.
  - Model includes a fixed production cost, entrants with below-average productivity, and reduced-form entry/exit barrier specifications.
- Model solution highlights:
  - Demand for firm i: q_i = [1/M − γ (ln p_i − ln p̄)]^{−1/γ} E / p_i.
  - Aggregate price index: ln P = 1/(2 γ M) + ln p̄ − 1/2 γ M V(ln p_i).
  - Firm pricing: p_i = (1 + s_i/γ) (1/a_i), where s_i = max{0, 1/M − γ (ln p_i − ln p̄)}.

### Exit barriers: definition, modelling, and effects
- Definition:
  - Exit barriers: distortions preventing displacement of low-productivity incumbents by higher-productivity firms or limiting expansion of high-productivity incumbents.
- Modelling:
  - Introduced as a lump-sum transfer t ≥ 0 to active firms, financed by lump-sum taxes on consumers; allows firms with negative net profits to survive.
  - Examples captured: protracted insolvency procedures (average length about 7 years in 2018), policies favoring firms below a certain size, government subsidies preserving employment (e.g., payroll subsidy mechanism "CIG").
- Comparative statics when reducing exit barriers:
  - Productivity distribution:
    - Reducing the transfer increases the minimum productivity threshold among active firms; low-productivity firms exit; average firm productivity rises.
  - Average productivity and markups:
    - As transfers fall and low-productivity firms exit, average markups increase because:
      - Number of active firms falls (relaxing competitive pressure),
      - Average productivity of remaining firms is higher and firm-level markups increase with productivity.
  - Real aggregate output:
    - Real aggregate output equals sum of firm-level sales net of fixed costs, divided by the price index.
    - Lower exit barriers generally increase real aggregate output because higher average productivity dominates loss of variety and higher markups.
    - Alternative calibration: increasing the relative mass of firms below the maximum entry productivity threshold by 20 percent yields stronger positive effects from removing exit barriers.
  - Welfare note:
    - Assuming no disutility from labor, welfare is monotonically increasing in real aggregate output.
    - Exit barriers are not a first-best policy; directly encouraging competition (e.g., removing entry barriers) is preferable.
    - Negative exit barriers (a lump-sum tax on active firms) are never optimal in the model.

### Entry barriers: definition, modelling, and effects
- Definition and modelling:
  - Entry barriers: distortions limiting incentives of new firms to start production (e.g., quotas, stringent regulations).
  - Modeled as an exogenous increase in the mass of entrants M; entrants’ productivity drawn from incumbents’ c.d.f. truncated above at ā.
  - Entrants have lower average productivity than incumbents.
- Short-run effects:
  - Immediate effect of removing entry barriers is a reduction in average productivity.
  - More entry increases the number of firms, reduces average markups, and lower-productivity entrants charge lower markups.
  - Figure 8 note: "Productivity is measured in logs. Markups are measured as the ratio of price to marginal cost minus one. Higher values for the share of entrants imply a lower minimum productivity across active firms."
- Output effects and robustness:
  - Removing entry barriers is always optimal in the model, causing a monotonical increase in real aggregate output.
  - Gains from reduced markups and increased variety dominate the loss in average productivity.
  - Alternative calibration: modeling high initial markups as a 20 percent reduction in the initial mass of firms makes reducing entry barriers particularly beneficial.
  - Figure 9 note: "the orange line corresponds to an economy with high initial markups, calibrated as a 20 percent smaller initial mass of firms. Real aggregate output is normalized to one when there is zero entry."
- Model abstractions and caveats:
  - Model abstracts from potential endogenous increase in entry after reduction in exit barriers.
  - Calibration ensures no incumbents are pushed to exit by new entrants.
  - Footnote 10: "This is a static or short-run effect. In a long-run steady state, entrants would grow to be as productive as the average incumbent."

### Policy implications and sectoral priorities
- Model predictions:
  - Removing entry barriers most efficient in sectors with high markups.
  - Removing exit barriers most efficient in sectors with a large number of low-productivity firms.
  - Where both distortions present, policies should facilitate both entry and exit.
- Sectoral prescriptions (based on firm-level markup and mass of unproductive firms):
  - High-markup sectors (e.g., administrative and professional services): remove entry barriers such as abolishing quotas for regulated professions, repealing requirement for a manager or owner to be a professional, and eliminating minimum tariffs.
  - Sectors with a large mass of unproductive firms (e.g., retail): foster consolidation; remove impediments to factor reallocation; lower exit barriers through modernization of the insolvency framework, more efficient active labor policies, and better-designed social safety nets.

### Evidence from literature on reform payoffs and timing
- Reported estimates:
  - Forni et al. (2010): reducing markups in services to euro-area levels induces an 11 percent increase in Italian GDP in the long run and a 3.5 percent increase in welfare; half of the GDP increase realized in the first three years.
  - OECD (2009): Italy’s labor productivity could increase by about 14 percent over 10 years if product market regulation in professional services aligned to international best practice.
  - Lusinyan and Muir (2013): closing roughly half the gap in product and labor markets with the rest of the euro area and best-practice OECD could raise real GDP by 5¾ percent after 5 years and by 10½ percent in the long run.
  - Andrle et al. (2018): estimate a 4 percent output gain in the steady-state from a package of product market deregulation (closing half the gap with the frontier) and other structural reforms.
  - González-Torres (2016): cutting firm startup procedures boosts aggregate productivity and output.
- Timing:
  - Bouis et al. (2019): positive dividends typically become apparent two to three years after reforms; prices drop and productivity and output increase significantly, with no evidence of short-term costs even under adverse macroeconomic conditions.
- Complementarities:
  - Reducing barriers in upstream services (e.g., professional services and network industries) boosts downstream industry productivity.

### III. Empirical results — data and productivity distribution findings
- Data:
  - Italian firm-level data from Orbis for 2006–2016.
  - After exclusions, sample around 400,000 firms per year for 2011–2016.
  - Cleaned sample covers ⅔ of non-financial corporate value added.
  - Aggregate GDP deflators from IStat; industry-level deflators from Eurostat.
- Productivity measurement:
  - Marginal Product of Labor in revenues (MPLR) approximated by value added per worker multiplied by labor share (employee compensation over value added at sector level, averaged over time).
- Empirical findings:
  - Market services display a substantially thicker mass of firms at the bottom of the productivity distribution compared to manufacturing.
  - Conditioning on sectoral means: share of firms with labor productivity below ½ of their sector’s average remains higher in market service sectors, except ICT.
  - Within market services, sectors with particularly high mass at the bottom: entertainment; professional and administrative services; followed by other services, transport, accommodation and food, construction, retail and wholesale trade.
- Data coverage and robustness:
  - Footnote 11: Smaller firms in services are underreported in Orbis; including them would make the gap between services and manufacturing even larger.
  - The thicker left tail in market services holds both in the North-Center and in the South, though MPLR distributions in the South are shifted left for both manufacturing and market services.

### Appendix II — labor productivity distributions and markups (methods and findings)
- Labor productivity distributions:
  - MPL of the firm = 훼훼 y_i / l_i with labor share 훼훼 calculated at sector level (NACE main sectors).
  - Firms with MPL below €100 or above €120,000 dropped for kernel density estimates.
  - Mass of firms below ½ sectoral mean: firms with labor productivity below €100 or above €200,000 are dropped; values are yearly averages for 2011–2016.
  - Appendix II: choice of productivity threshold not critical; similar sector ranking when using share below ¾ of the mean.
- Markup estimation:
  - Production function estimated at sector level following Ackerberg et al. (2015); assumed Cobb-Douglas in capital and a variable input (sum of employee compensation and cost of goods sold).
  - Markups identified as in De Loecker and Warzynski (2012): divide production function exponent on variable input by firm’s share of variable input over sales.
  - Deflation: variable inputs deflated with GDP deflator; capital deflated with manufacturing deflator; sales deflated at industry level; NACE-4 missing deflators use next superior aggregation level.
- Markup findings and caveats:
  - Average markup by sector, 2005–2016: most market service sectors feature higher-than-average markups, notably administrative and professional services and ICT.
  - Network sector tops ranking but may reflect natural monopoly properties.
  - Caveats: high-share of self-employed not captured in Orbis; estimated markups for one-employee firms are higher than average; potential misreporting of managers’ labor income as capital income may overestimate markups in small firms.
  - Appendix II: ranking similar when including only firms with more than 10 employees; administrative and professional services remain significantly above average.
  - Methodological note: methodology does not disentangle pure monopolistic rents from markups required to compensate for fixed costs.

### Diagnostic framework and reform priorities (sectoral mapping)
- Figure 13 diagnostic:
  - Merges mass of unproductive firms (share below ½ mean) and average markups into a sectoral scatter plot.
  - Interpretation:
    - High-markup sectors (right-hand side): lowering entry barriers yields more benefits.
    - Thick-tail sectors (upper side): reducing exit barriers yields more benefits.
  - Largest sectors by employment:
    - High-markup sectors: professional and administrative services, networks.
    - Thick-tail sectors: wholesale and retail trade.
  - Manufacturing shows comparatively less evidence of frictions.

### Evidence on regulation and reform history in Italy
- Three waves of PMR reforms (Bugamelli et al., 2018):
  1. late 1990s–early 2000s: liberalizations in network industries (electricity, gas, telecom);
  2. 2006–07 (Bersani’s decrees): professional services and retail sector, adjustments in telecom, insurance, banking;
  3. 2012–14 (Monti reforms): further liberalization of network industries, professional services, retail, transport, water, postal services.
- August 2017: adoption of first Annual Competition Law (ACL) since 2009 requirement; emphasis on consumer protection with subsequent weakening, delay, or partial implementation of pro-competition measures.
- Comparative indicators:
  - Italy lags in several competition indicators relative to OECD peers.
  - Economy-wide PMR score slightly lower than OECD average in 2018; service sectors featured much higher regulatory impediments.
  - Professional services and retail trade are major outliers.
  - European Commission (2017): restrictiveness for professions in Italy higher than EU weighted average for six out of seven professions analyzed; most regulated professions include accountants, estate agents, architects, and engineers.

### Sectoral reform recommendations (selected)
- Professional services:
  - Eliminate minimum tariffs ("equo compenso"); note Decree Law 148/2017 and the 2018 Budget Law context.
  - Abolish quantitative restrictions for regulated professions (examples noted for notaries, pharmacists).
  - Repeal requirement that manager or owner be a professional (partial liberalizations noted; limits remain).
  - Reform professional orders to limit conflicts of interest; suggested separation of administrative, education and disciplinary functions; remove restrictions on advertising.
  - Evidence: regulated occupations characterized by a wage premium of about 9 percent; rises to 18 percent for those regulated by a professional order (Mocetti et al., 2019).
- Retail:
  - Relax strict conditions on opening new retail outlets; Italy is among most restrictive EU countries per EC retail restrictiveness indicator.
  - Remove restrictions on sales promotions and allow distribution of some products (including non-prescription drugs) in retail stores.
  - Note: opening hours liberalized in 2011 (Decree Law 201/2011) but severe restrictions remain; a proposed law on re-regulating stores’ opening hours could have sizeable employment impact (Rizzica et al., 2019).
- Transportation and energy:
  - Open up local transport services to competition; ACL envisaged framework for non-public transportation services including platform-based ride services, but implementing measures not adopted in delegation timeframes.
  - Reduce vertical integration in railways; encourage regions to open tenders for train and road services.
  - Continue liberalizing postal services; ACL abolished some Poste Italiane monopolies but 2018 Budget Law imposed licensing conditions and reserved items up to 5kg.
  - Phase out regulated tariffs in retail gas and electricity markets; transition to free market postponed to 2022 by Law Decree 162/2019.
  - Eliminate asymmetric obligations for new entrants in fuel distribution; exclusive contracts hinder entry and protect dealers via fixed margins.
- Economy-wide policies:
  - Eliminate barriers to exit via well-functioning social safety net, modern insolvency regime, active labor market policies, and retraining.
  - Cassa Integrazione Guadagni (CIG) details:
    - Entitled workers receive 80 percent of their average gross earnings of the previous 3 months, reduced by 20 percent after the first year.
    - Support can last up to 24 months over a period of 5 years.
    - Jobs Act of 2015 cancelled CIG use for firms that have ended or are about to end production; subsequent delays and extensions noted through 2020 under "Emergency Decree".
  - Modernize insolvency regime after new insolvency code adopted February 2019; reform enters into force in August 2020; authorities working on secondary legislation.
  - Complementary measures: improve court functioning; ensure qualified insolvency administrators; reform civil procedures; facilitate collateral sales; incentivize courts to reduce backlogs; develop uniform practices nationwide.
  - Financial policies: continue strengthening bank balance sheets to improve credit allocation and facilitate efficient exit of firms.
  - Strengthen institutional framework: enhance Competition Authority enforcement powers; invest in evidence-based regulatory policies; introduce benchmarking among regions.
  - Secure political adhesion: stakeholder consultations, publish consultation results, use impact analyses and benchmarking; address multi-level governance challenges where regional jurisdiction hampers unified legislation.

### Appendix I — model calibration and parameter values
- Key calibration choices:
  - Demand dispersion parameter γ set to 1; M set to maximum such that all firms have strictly positive market share.
  - Labor supply elasticity ν = 0.5.
  - Productivity distribution F(a) specified as log-normal LLN(μ_a, σ_a), bounded between €1,000 and €200,000; μ_a and σ_a estimated from 2016 distribution and normalized by average wage of €20,000.
  - Comparative statics for exit barriers: transfers considered such that lower bound of productivity distribution increased by 0 to 4 percent of productivity range.
  - Comparative statics for entry barriers: relative mass of entrants over incumbents ranges from 0 to 0.2; entrants’ maximum productivity threshold ā calibrated so average productivity of entrants (firms < 5 years) relative to incumbents is 68 percent.
  - Calibration ensures no incumbent pushed to exit by new entrants.
- Table A1. Parameter Values:
  - Markup level parameter γ: 1
  - Labor supply elasticity ν: 0.5
  - Productivity log-normal mean μ_a: 0.61
  - Productivity log-normal std. deviation σ_a: 0.78
  - Max productivity threshold entrants ā: 1.38
  - Fixed cost K: 0.37
  - Mass of firms M: 0.6

*Source: wpiea2020039-print-pdf*

### 1. Productivity and Employment in Market Services ________________________________3

### 1. Productivity and Employment in Market Services

### Key facts and sectoral importance
- Market services (defined as nonfinancial and nonreal estate market services: wholesale and retail trade, transportation, accommodation and food services, information and communication technologies (ICT), professional and administrative services, arts and entertainment, and other market services) accounted in 2015 for:
  - 36 percent of total value-added,
  - 47 percent of total hours worked,
  - 46 percent of employment.
- Major employers within market services:
  - Wholesale and retail trade: 16 percent of economy-wide employment,
  - Professional services: 12 percent of economy-wide employment.

### Recent productivity trends (2000–2015)
- Aggregate labor productivity of market services in Italy fell by over 5 percent between 2000 and 2015, while remaining relatively stable in the euro area.
- Sub-sectoral productivity declines between 2000 and 2015:
  - Professional services: 30 percent decline,
  - Accommodation and food services: 20 percent decline,
  - Other services: 20 percent decline.

### Integration with manufacturing
- Services in Italy are more tightly integrated to the manufacturing sector than in other European countries (see "Share of Services in Manufacturing Input, 2014" analysis).

### Theoretical framework and model properties
- A monopolistic competition model with heterogeneous firms and a trans-log demand function is used to study entry and exit barriers (model draws on Felix and Maggi (2019) and Bergin and Feenstra (2000); Appendix I contains full model and calibration).
- Model assumptions and key implications:
  - Firms are heterogeneous in productivity and use labor as the only input.
  - Aggregate labor supply is increasing in real wages; lower prices increase aggregate labor supply and aggregate output.
  - Two equilibrium properties of the trans-log demand:
    - Average markups are decreasing in the number of active firms.
    - Firm-level markups are increasing in a firm’s productivity.
  - The model includes: a fixed production cost (so firms may exit), entrants with below-average productivity, and reduced-form entry/exit barrier specifications.

### Exit barriers: definition and modelling
- Exit barriers are any distortion preventing displacement of low-productivity incumbents by higher-productivity incumbents or limiting expansion of high-productivity incumbents into markets of low-productivity firms.
- In the model, exit barriers are introduced as a lump-sum transfer to active firms, allowing firms with negative net profits (the least productive) to survive; the transfer is financed by lump-sum taxes on consumers.
- Examples of economic distortions captured by this specification:
  - Protracted insolvency procedures (the average length of insolvency proceedings was about 7 years in 2018),
  - Policies favoring firms below a certain size or penalizing firm growth,
  - Government subsidies that preserve employment (e.g., payroll subsidy mechanism "CIG").

### Effects of reducing exit barriers (model comparative statics)
- Productivity distribution:
  - Reducing the transfer (exit barrier) increases the minimum productivity threshold among active firms, pushing low-productivity firms to exit and raising average firm productivity.
- Average productivity and markups:
  - As transfers fall and low-productivity firms exit, average markups increase for two reasons:
    - The number of active firms falls (relaxing competitive pressure),
    - Average productivity of remaining firms is higher, and firm-level markups are increasing in productivity.
- Real aggregate output:
  - Real aggregate output equals sum of firm-level sales net of fixed costs, divided by the price index. Lower exit barriers generally increase real aggregate output because the positive effect of higher average productivity dominates the negative effect of reduced variety and higher markups.
  - Alternative calibration with a higher initial mass of low-productivity firms (relative mass of firms below the maximum entry productivity threshold for entrants increased by 20 percent) shows stronger positive effects from removing exit barriers—more inefficiently-subsidized firms exit, accelerating average productivity gains.
- Notes on welfare:
  - Assuming no disutility from labor, welfare is monotonically increasing in real aggregate output.
  - Although parameter choices can yield cases where a small positive exit barrier increases output (via pro-competition effects), exit barriers are not a first-best policy; directly encouraging competition (e.g., removing entry barriers) is preferable.
  - Negative exit barriers (a lump-sum tax on active firms) are never optimal in the model.

### Policy implications and sectoral priorities
- The model predicts:
  - Removing entry barriers is most efficient in sectors with high markups.
  - Removing exit barriers is most efficient in sectors with a large number of low-productivity firms.
- Sectoral policy prescriptions (based on sector classification using firm-level markup and mass of unproductive firms):
  - For sectors with high markups (e.g., administrative and professional services): remove entry barriers such as abolishing quotas for regulated professions, repealing requirement for a manager or owner to be a professional, and eliminating minimum tariffs.
  - For sectors with a large mass of unproductive firms (e.g., retail): foster consolidation; remove impediments to factor reallocation; lower exit barriers through modernization of the insolvency framework, more efficient active labor policies, and better-designed social safety nets.

### Evidence from literature on reform payoffs
- Examples of estimated payoffs from prior studies (as reported in the source):
  - Forni et al. (2010): reducing markups in services to euro-area levels induces an 11 percent increase in Italian GDP in the long run and a 3.5 percent increase in welfare; half of the GDP increase realized in the first three years.
  - OECD (2009): Italy’s labor productivity could increase by about 14 percent over 10 years if product market regulation in professional services was aligned to international best practice.
  - Lusinyan and Muir (2013): closing roughly half the gap in product and labor markets with the rest of the euro area and best-practice OECD could raise real GDP by 5¾ percent after 5 years and by 10½ percent in the long run.
  - Andrle et al. (2018): estimate a 4 percent output gain in the steady-state from a package of product market deregulation (closing half the gap with the frontier) and other structural reforms.
  - González-Torres (2016): cutting firm startup procedures boosts aggregate productivity and output.
  - Timing of payoffs: Bouis et al. (2019) find positive dividends typically become apparent two to three years after reforms; prices drop and productivity and output increase significantly, with no evidence of short-term costs even under adverse macroeconomic conditions.
- Complementarity across sectors:
  - Additional gains accrue because some services (e.g., professional services and network industries) are inputs into other industries; reducing barriers in upstream services boosts downstream industry productivity.

*Source: wpiea2020039-print-pdf - 1. Productivity and Employment in Market Services*

### Section III) by 0.3 percentage points.

### Section III) by 0.3 percentage points.

### C. Entry Barriers
- Definition and modeling:
  - Entry barriers include any distortions limiting incentives of new firms to start production, such as quotas or excessively stringent regulations.
  - In the model, a reduction of entry barriers is introduced as an increase in the mass of entrants; the model is agnostic on the exact source of entry barriers and focuses on the impact of adding more entrants on the market equilibrium.
- Productivity distribution of entrants:
  - Productivity of entrants is drawn from the distribution of existing firms, bounded above by a maximum productivity threshold.
  - The threshold is calibrated such that the relative productivity of entrants is similar to that observed in the data.
  - Entrants feature lower average productivity than incumbents.
- Short-run effects:
  - Immediate effect of removing entry barriers is a reduction in average productivity (Figure 8).
  - More entry implies more firms, which reduces average markups; lower-productivity firms charge lower markups, compounding this effect.
  - Figure 8 note: "Productivity is measured in logs. Markups are measured as the ratio of price to marginal cost minus one. Higher values for the share of entrants imply a lower minimum productivity across active firms."
- Output effects and robustness:
  - Removing entry barriers is always optimal in the model, causing a monotonical increase in real aggregate output (Figure 9).
  - Gains from reduced markups and increased variety dominate the loss in average productivity, in contrast to lowering exit barriers.
  - Alternative calibration: higher initial markups modeled as a 20 percent reduction in the initial mass of firms. Reducing entry barriers is particularly beneficial when initial markups are high.
  - Figure 9 note: "the orange line corresponds to an economy with high initial markups, calibrated as a 20 percent smaller initial mass of firms. Real aggregate output is normalized to one when there is zero entry."
- Model abstractions and caveats:
  - The model abstracts from potential endogenous increase in entry after a reduction in exit barriers.
  - Calibration ensures no incumbents are endogenously pushed to exit after a reduction in entry barriers.
  - These assumptions aim to transparently illustrate direct effects; including indirect channels would magnify output gains of removing entry and exit barriers.
- Footnote on dynamics:
  - Footnote 10: "This is a static or short-run effect. In a long-run steady state, entrants would grow to be as productive as the average incumbent."

### D. Policy Implications
- Summary of priorities:
  - Removing exit barriers is predicted to be most effective in economies or sectors with a high mass of unproductive firms.
  - Removing entry barriers is predicted to be most effective in sectors with high average markups.
  - Where both distortions are present, policies should facilitate both entry and exit.
- Additional note:
  - Table 1 displays policy priorities, while the model shows that removing entry and exit barriers is optimal for any sector.
  - Table 1 note: "the table entries show the policies that would have a higher impact as a function of the productivity distribution and average markups in an economy/sector."

### III. EMPIRICAL RESULTS
#### A. Data
- Data source and coverage:
  - Italian firm-level data are sourced from Orbis for the period 2006–2016.
  - After excluding non-market sectors, one-employee firms, and extreme outliers, a sample of around 400,000 firms per year remains for the period 2011–2016, where sample size is largest and more stable.
  - The cleaned sample covers ⅔ of non-financial corporate value added, although coverage is more limited in a few service subsectors.
  - Aggregate GDP deflators are obtained from IStat and industry-level deflators from Eurostat.
  - Appendix II describes in detail the sample selection procedure.
- Use of data:
  - The data are used to classify sectors along two dimensions signaling exit and entry frictions: the tail of low-productivity firms and average markups.

#### B. Productivity Distribution
- Measurement approach:
  - Comparing labor productivity distributions across sectors requires controlling for differences in production function labor intensity.
  - The marginal product of labor in revenues (MPLR) is used: MPLR approximated by value added per worker multiplied by the labor share (employee compensation over value added at the sector level, averaged over time).
- Empirical findings:
  - In a frictionless model, MPLR should be equalized across firms and sectors; Figure 10 shows this is not the case in the data.
  - Market services display a substantially thicker mass of firms at the bottom of the distribution compared to manufacturing.11
  - Conditioning on sectoral means (Figure 11): share of firms with labor productivity below ½ of their sector’s average remains higher in market service sectors, except ICT.
  - Within market services, entertainment and professional and administrative sectors have a particularly high mass of firms at the bottom, followed by other services, transport, accommodation and food, construction, and retail and wholesale trade.
- Data coverage and robustness notes:
  - Footnote 11: Smaller firms in services are underreported in Orbis; including them would make the gap between services and manufacturing even larger because those firms tend to be at the bottom of the productivity distribution.
  - Footnote 11 (continued): The gap is not purely due to geographical factors; both in the North-Center and in the South, market services feature a thicker left tail of firms with low MPLR. The main difference is that MPLR distributions in the South are shifted to the left for both manufacturing and market services.

*Source: wpiea2020039-print-pdf - Section III) by 0.3 percentage points.*

### Appendix II underlines that the choice of the productivity threshold is not critical, as a very similar ranking of

### wpiea2020039-print-pdf - Appendix II underlines that the choice of the productivity threshold is not critical, as a very similar ranking of

### Labor productivity distributions and firm masses
- Kernel density estimates of the Marginal Product of Labor (MPL) across Italian firms by sector, 2016.
  - MPL of the firm = 훼훼 y푖푖 / l푖푖 (Labor share 훼훼 calculated at the sector level (NACE main sectors)).
  - Firms with an MPL below €100 or above €120,000 are dropped.
- Mass of firms with labor productivity below ½ of their sectoral mean, 2011–2016.
  - Labor productivity defined as value-added per worker.
  - Firms with labor productivity below €100 or above €200,000 are dropped.
  - The values displayed are yearly share averages for the period from 2011 to 2016.
- Appendix II: choice of the productivity threshold is not critical; a very similar ranking of sectors is obtained for the share of firms below ¾ of the mean.

### Markups (Section C)
- Estimation method:
  - Production function estimated at the sector level following Ackerberg et al. (2015); assumed Cobb-Douglas in capital and a variable input (sum of employee compensation and cost of goods sold).
  - Markups identified as in De Loecker and Warzynski (2012): divide the production function exponent on the variable input by the firm’s share of the variable input over sales.
  - Intuition: higher markups imply a smaller share of variable input over sales in equilibrium, yielding a higher ratio.
- Deflation and data notes:
  - Markup calculations are based on real variables.
  - Variable inputs deflated with the GDP deflator.
  - Capital (fixed assets) deflated with the manufacturing deflator.
  - Sales deflated at the industry level.
  - For NACE-4 sectors with missing deflator data, the next available superior level of aggregation is used.
- Empirical findings:
  - Average markup by sector, 2005–2016: most market service sectors feature higher-than-average markups, notably administrative and professional services and ICT.
  - The network sector tops the ranking but may reflect natural monopoly properties rather than policy distortions.
- Caveats:
  - Some high-markup sectors include a high share of self-employed workers not captured in Orbis.
    - Estimated markups for one-employee firms are higher than average, suggesting excluding self-employed workers does not overestimate markups.
  - Potential misreporting of managers’ labor income as capital income may overestimate markups in small firms.
  - Appendix II: ranking of sectors is quite similar when including only firms with more than 10 employees.
    - Markups in administrative and professional services remain significantly above average.
- Methodological note:
  - While allowing for heterogeneous production function exponents across sectors, the methodology does not disentangle pure monopolistic rents from markups required to compensate for fixed costs.

### Policy implications (Section D)
- Diagnostic framework:
  - Figure 13 merges mass of unproductive firms (share below ½ mean) and average markups into a sectoral scatter plot.
  - Interpretation:
    - Sectors with high markups (right-hand side): lowering entry barriers would yield more benefits.
    - Sectors with a thick tail of unproductive firms (upper side): reducing exit barriers would yield more benefits.
  - Largest sectors in terms of employment:
    - High-markup sectors: professional and administrative services, networks.
    - Thick-tail (unproductive firms): wholesale and retail trade.
  - Overall: manufacturing shows comparatively less evidence of frictions.

### Evidence on the state of regulation in Italy (Section IV)
- Reform history:
  - Three waves of PMR reforms (Bugamelli et al., 2018):
    1. late 1990s–early 2000s: liberalizations in network industries (electricity, gas, telecom);
    2. 2006–07 (Bersani’s decrees): professional services and retail sector, adjustments in telecom, insurance, banking;
    3. 2012–14 (Monti reforms): further liberalization of network industries, professional services, retail, transport, water, postal services.
  - August 2017: adoption of the first Annual Competition Law (ACL) since the 2009 legislative requirement; emphasis on consumer protection with subsequent weakening, delay, or partial implementation of pro-competition measures.
- Comparative indicators:
  - Italy still lags in several competition indicators relative to other OECD countries.
  - Economy-wide PMR score slightly lower than the OECD average in 2018, while service sectors featured much higher regulatory impediments than the average.
  - Professional services and retail trade are major outliers compared to the rest of the OECD.
  - European Commission (2017): overall level of restrictiveness for professions in Italy is higher than the EU weighted average for six out of seven professions analyzed.
  - Most regulated professions include accountants, estate agents, architects, and engineers.

### Reform priorities in highly regulated sectors (Section V)
- Overall approach:
  - Empirical evidence and regulatory impediments justify targeted reforms for professional services, retail, transportation and energy, plus economy-wide measures.
  - Recommendations draw on past advocacy by the Italian Competition Authority (AGCM), the European Commission, and the OECD.

- A. Professional Services — recommended measures:
  - Eliminating minimum tariffs (equo compenso).
    - 2018 Budget Law introduced “fair compensation” rules for professionals contracting with general government, banks, insurance companies and large firms.
    - Fair compensation legislation introduced via Decree Law 148/2017.
  - Abolishing quantitative restrictions for regulated professions.
    - ACL increased number of notaries from 1 per 7,000 inhabitants to 1 per 5,000 inhabitants and allowed a second office within the same region; subsequent amendments allowed local orders to limit competition.
    - Some professions remain subject to binding quotas such as pharmacists.
  - Repealing requirement for a manager or owner to be a professional.
    - ACL allowed limited liability companies to own pharmacies and removal of the limit to own more than four pharmacies, but implementation hampered by interpretation issues.
    - Law firms partially liberalized with non-professionals allowed to own only up to ⅓ of shares.
  - Reform professional orders to limit conflict of interest.
    - Governance often leads to conflict of interests where members are both competitors and overseers.
    - Evidence: regulated occupations characterized by a wage premium of about 9 percent; rises to 18 percent for those regulated by a professional order (Mocetti et al., 2019).
    - Recommendations: separate administrative, education and disciplinary functions; remove restrictions on advertising.

- B. Retail Sector — recommended measures:
  - Relax strict conditions on opening new retail outlets.
    - EC retail restrictiveness indicator: Italy among most restrictive EU countries.
    - Restrictions mainly in authorization process and high number of permits, impact assessments, and location-specific requirements.
  - Remove restrictions on sales promotions and allow distribution of some products in retail stores.
    - Opening hours liberalized in 2011 (Decree Law 201/2011), but severe restrictions remain on sales promotions, discounts, and below-cost sales.
    - Distribution of some products, including non-prescription drugs, is not allowed.
    - Recently proposed law on re-regulating stores’ opening hours could have a sizeable impact on employment (see Rizzica et al., 2019).

- C. Transportation and Energy — recommended measures:
  - Opening up local transport services to competition.
    - ACL envisaged a new regulatory framework to foster competition in non-public transportation services (including platform-based ride services), but implementing measures were not adopted within delegation timeframes.
  - Reduce vertical integration in railways.
    - Ferrovie dello Stato owns most infrastructure and major operators, benefits from directly awarded contracts; intercity service run exclusively by Trenitalia.
    - Encourage regions to open tenders for train and road services following liberalization of high speed train services.
  - Continue liberalizing postal services.
    - ACL abolished Poste Italiane's monopoly on delivery of legal notifications and traffic fines, but 2018 Budget Law conditioned licensing on storage characteristics and reserved postal items up to 5kg to Poste Italiane.
  - Phase out regulated tariffs in retail gas and electricity markets.
    - Transition to free market postponed to 2022 by Law Decree 162/2019.
    - Electricity prices remain high due to deficient transmission infrastructure and lack of competition in final sale of electricity.
    - Italian retail power market composition: (i) households and small enterprises purchase electricity under the “Maggio Tutela” standard offer; (ii) customers with turnover greater than €10 million purchase electricity at prices negotiated with retailers.
    - Standard offer price set at the beginning of each regulatory quarter; acts as a reference value that discourages competition among retailers.
  - Eliminate asymmetric obligations for new entrants in fuel distribution.
    - Exclusive contracts between fuel distributors and suppliers hinder entry; contractual relationships protect dealers via fixed margins.

- D. Economy-wide policies — recommended measures:
  - Eliminating barriers to exit.
    - Need: well-functioning social safety net, modern and efficient insolvency regime, active labor market policies to foster labor mobility and retraining.
    - Example: Cassa Integrazione Guadagni (CIG) provides cash payments to employed workers after reduced hours or redundancies.
      - Entitled workers receive 80 percent of their average gross earnings of the previous 3 months, reduced by 20 percent after the first year.
      - Support can last up to 24 months over a period of 5 years.
    - Jobs Act of 2015 cancelled CIG use for firms that have ended or are about to end production; 2016 inter-ministerial Decree No. 950575 delayed activation until 2018 conditional on sale prospects; Council of Ministers extended further until 2020 under the “Emergency Decree”.
    - Pursue modernization of the insolvency regime after adoption of new insolvency code in February 2019.
      - Reform will enter into force in August 2020.
      - Authorities working on pending secondary legislation (early warning mechanisms, liquidation procedures).
      - Special insolvency regime for large enterprises is excluded and advisable to fold into modernized framework.
    - Complementary efforts: improve court functioning, ensure qualified insolvency administrators, reform civil procedures, facilitate collateral sales, incentivize courts to reduce backlogs, develop uniform practices nationwide.
    - Financial policies: continue strengthening bank balance sheets to improve credit allocation and facilitate efficient exit of firms.
  - Strengthening the institutional framework.
    - Enhance Competition Authority’s enforcement powers; invest in evidence-based regulatory policies; introduce benchmarking among regions; stimulate diffusion of norms and best practices.
  - Securing political adhesion to reforms.
    - Build consensus via stakeholder consultations, publish consultation results, reveal parties consulted, use impact analyses and benchmarking to communicate need and benefits of reforms.
    - Address multi-level governance challenges where regional jurisdiction (e.g., opening of new retail outlets) hampers unified legislation and allows leeway for restrictions.

*Source: wpiea2020039-print-pdf (selected sections: Appendix II, Figures 10–17, Sections C–V).*

### APPENDIX I. MODEL

### APPENDIX I. MODEL

### A. Set Up

- Consumers
  - Representative consumer with a symmetric trans-log demand function (as in Feenstra (2003)).
  - Consumer chooses quantities q_i for each variety i ∈ [0, M], taking variety prices p_i and income E as given.
  - Income from supplying labor competitively and owning firms.
  - Consumer maximization yields an inverse demand function p_i(q_i).

- Firms
  - Each variety produced by one monopolistically competitive firm.
  - Firm productivity a_i heterogeneous with cdf F(a).
  - Production: q_i = a_i l_i.
  - Fixed cost K.
  - Government pays lump-sum transfer t ≥ 0 to each active firm, financed with lump-sum taxes on consumers.
  - Firm profit maximization:
    - π_i = max_{l_i} { p_i(q_i) q_i − W l_i + t − K, 0 }, (1)
    - π_i = 0 if firm does not produce (l_i = 0).

- Labor market
  - Aggregate labor supply: L_S = ξ(W/P)^ν, where P is aggregate price level.
  - Nominal wages normalized to the numeraire: W = 1.

- Goods market
  - Consumer expenditure equals total income:
    - E = W L_S + ∫_0^M π_i di − M t.
  - Aggregate real output: E / P.

- Entry and exit barriers
  - Exit barriers introduced by setting transfer t > 0, preventing some firms with negative net profits from exiting.
  - Reduction of entry barriers modeled as an exogenous increase in the mass of firms M.
  - Entrant productivity drawn from incumbents’ c.d.f., censored above at productivity ā.

### B. Equilibrium

- Definition
  - Equilibrium defined by aggregates {M, L_S, E}, firm-level {q_i, l_i}, aggregate price P, and firm-level prices {p_i} such that:
    - Consumers maximize utility.
    - Firms maximize profits (equation (1)).
    - Labor market clears: L_S = ∫_0^M l_i di.
    - Goods market clears: E = ∫_0^M p_i q_i di.

- Solution (as in Feenstra (2003))
  - Demand for firm i:
    - q_i = [1/M − γ (ln p_i − ln p̄)]^{−1/γ} E / p_i,     (2)
    - where γ governs level and dispersion of markups across firms.
  - Average log price:
    - ln p̄ = (1/M) ∫_0^M ln p_i di.
  - Aggregate price index:
    - ln P = 1/(2 γ M) + ln p̄ − 1/2 γ M V(ln p_i),     (3)
    - where V(·) is the variance operator.
  - Firm profit maximization implies:
    - p_i = (1 + s_i/γ) (1/a_i),
    - expenditure share s_i = max{0, 1/M − γ (ln p_i − ln p̄)},
    - price p_i equals markup τ_i = 1 + s_i/γ times marginal unit cost 1/a_i.
  - Explicit solution obtained numerically after discretizing the productivity distribution across firms.

### C. Calibration

- Productivity distribution
  - F(a) specified as log-normal LLN(μ_a, σ_a), bounded between €1,000 and €200,000 to avoid extremes.
  - μ_a and σ_a estimated from 2016 Italian labor productivity distribution with bounds, normalized by approximate average wage of €20,000 (nominal wages normalized to one in model).

- Parameter choices and calibration details
  - γ cannot be separately identified from M; set γ = 1 and M set to the maximum such that all firms have strictly positive market share (s_i > 0).
  - Labor supply elasticity ν = 0.5 (lower range of literature to reflect higher labor rigidity in Italy).
  - Comparative statics for exit barriers:
    - Transfers considered such that the lower bound of the productivity distribution is increased by 0 to 4 percent of the productivity range.
    - The transfer t corresponding to each point is not explicitly calculated (lump-sum reallocation).
    - Entry cost K chosen such that the lowest-productivity firm at the maximum truncation level (4 percent) has zero profits.
    - Results shown for baseline calibration and for an economy where relative density of firms below ā is increased by 20 percent (high density of low-productivity firms).
  - Comparative statics for entry barriers:
    - Relative mass of entrants over incumbents ranges from 0 to 0.2.
    - Maximum productivity threshold for entrants ā calibrated so average productivity of entrants (firms < 5 years) relative to incumbents is 68 percent (as in Italian data).
    - Results shown for baseline and for an economy where initial mass of firms M is 20 percent lower (case with high initial markups).
    - Calibration ensures no incumbent is pushed to exit by new entrants.

- Table A1. Parameter Values
  - Markup level parameter γ: 1
  - Labor supply elasticity ν: 0.5
  - Productivity log-normal mean μ_a: 0.61
  - Productivity log-normal std. deviation σ_a: 0.78
  - Max productivity threshold entrants ā: 1.38
  - Fixed cost K: 0.37
  - Mass of firms M: 0.6

*Source: APPENDIX I. MODEL, wpiea2020039-print-pdf.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020039-print-pdf.pdf_