## _wp1067 - References

## Source details

**Canonical URL:** [_wp1067 - References](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1067.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1067.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1067.pdf.json)

---

### Introduction: research question and main findings
- Research objective: investigate empirically the impact of competition on firm productivity using firm observations from the World Bank Enterprise Survey database.
- Main empirical findings:
  - A positive and robust causal relationship between proxies for competition at the firm level and measures of firm productivity.
  - Firm productivity growth is higher in countries with more substantial product-market reforms.
  - Countries that reformed during the period experienced a more pronounced increase in competition, providing a channel by which productivity is affected by reform.
- Context and motivation:
  - Product market regulations vary substantially across countries; example comparative start-up costs in 2003 (Doing Business database): 0.7 percent of GDP per capita in the United States; 16.8 percent in Italy; 1316.4 percent in Angola.
  - Theoretical ambiguity: standard endogenous growth models predict competition may reduce innovation (“rent dissipation effect” and shortened durability of innovations), while extensions (Aghion et al.) show a possible positive “escape competition” incentive to innovate depending on distance to the technological frontier.
- Contribution of this study:
  - Uses firm-level measures of competition and exogenous variation in product-market reforms across countries to draw causal inferences while controlling for firm-level institutional, financial, and labor market frictions.
  - Leverages detailed firm employee education data to improve human-capital-adjusted productivity measures.
  - Focus on emerging-market countries where productivity improvements were the main source of growth from 1999–2005 and where a wave of product market deregulations occurred in some, but not all, countries.

### Data and empirical methodology
- Data coverage and timing:
  - Firm data for 27 countries in 2004.
  - Use of World Bank Enterprise Survey data for 2004 (and 2001 for panel/growth measures where available).
- Productivity measurement (four approaches):
  - First: firm-level total factor productivity (TFP) as a residual from an augmented Cobb–Douglas production function, estimating input elasticities using the Olley and Pakes (1996) two-step procedure.
    - Country-level input elasticity regression (in logs): y_ji = θ^k_i k_ji + θ^h_i h_ji + η_s + ε_ji.
    - Physical capital measured as net book value of fixed assets after depreciation; output = value added (2004).
    - Human capital H measured as L × exp^{φ(s)} with φ(s) linear slope 0.10.
    - Olley–Pakes correction: human capital coefficients move in predicted direction in 18 out of 19 cases; capital coefficient increases in 15 out of 19 cases.
  - Second: TFP index relative to the industry median (Caves, Christensen, and Diewert (1982)), using observed firm factor shares and including intermediates.
  - Third: firm labor productivity measured as sales per human capital.
  - Fourth: natural experiment TFP growth (2001–2004) measured as a Solow residual; methodology detailed in Appendix C.
- Measurement and estimation details:
  - All output and capital series deflated using aggregate price levels from the Penn World Tables.
  - Industry-specific effects based on one-digit NACE codes.
  - Olley and Pakes (1996) two-step procedure used to address input demand endogeneity; lagged investment and human capital from the World Bank Surveys used as instruments.
  - Tables referenced: Table 1 (Olley–Pakes versus OLS input elasticities), Table 2 (correlations across productivity measures), Table 10 (input elasticities per industry per country).
  - Productivity distributions: skewed right; high dispersion across firms.
- Competition measurement and controls:
  - Firm-level competition proxies from World Bank Enterprise Survey:
    - ln-markup: mark-ups measured as sales over operating costs.
    - comp-cost: discrete 1–4 pressure from domestic competitors to reduce production costs (1 = low importance to 4 = high importance).
    - comp-import: qualitative external competition from imports, values 1 to 6 (1 = low import competition; 6 = high import competition).
  - Controls:
    - Firm characteristics: age; size categories (small <50 employees, medium 50–249, large ≥250); foreign (headquarters abroad); exports (exporter dummy); privatized; new techno (introduced substantially new technology).
    - Financing/input-market: ret-earn (percentage of investments financed from retained earnings); stock (listed dummy); hire (manager’s estimate of percent workforce increase absent restrictions).
    - Institutional: audit; pow-outages (number of power outages per year); juris (qualitative 1–6, 1 = “never affordable”, 6 = “always affordable”).
- Panel and subgroup considerations:
  - Sector-level estimation feasible mainly for manufacturing and wholesale (Table 9 reports manufacturing results).
  - Natural experiment leverages cross-country variation in reform timing (Section IV).
- Data limitations:
  - Intermediate input information available in 2004 but not 2001.
  - Inventories information not available; assume inventory-to-sales ratio uncorrelated with firm capital or human capital.
  - Disaggregated price-level deflators not available; use country fixed effects and firm characteristics.

### C. Reforms — data and classification
- Data source: Economic Freedom of the World Index (Fraser Institute) — annual index on “ease of starting a new business” for 1995 and annually from 2000–2004.
- Index construction (three variables averaged): (i) number of days to comply with regulations, (ii) money costs of fees paid to regulatory authorities measured as a share of per-capita income, (iii) minimum capital requirements measured as a share of per-capita income.
- Index range: 1 to 10 (decreasing order of entry costs).
- Reformer definition: country classified as a reformer if the Fraser index improved by at least 40 percent from 2001 to 2004.
- Empirical grouping and summary statistics:
  - Reformer group: 10 countries; index increased by an average of 69 percent (largest increase 147 percent in Romania).
  - Non-reformer group: index declined on average by 1.4 percent; exceptions Poland and the Czech Republic where index increased by 22.3 percent and 10.6 percent respectively (classified as non-reformers based on 1995–2004 performance).
  - Group average index levels: 2004 = 5.8 and 2001 = 5.6.
- Complementary regulatory indices (trade, financial, labor) from Fraser: no important changes other than entry costs and no differential changes between reformer and non-reformer groups.
- Median TFP growth: period average median TFP growth for reformer countries is about double corresponding growth for non-reformer countries.

### III. Empirical relation between TFP and competition — model and measures
- Productivity measures (log):
  - lnTFP-OP (Olley and Pakes, 1996)
  - lnTFP-Caves (Caves, Christensen, and Diewert, 1982)
  - ln-laborprod
- Competition measures:
  - lnmarkup
  - comp-cost
- Estimation equation (pooled, country and industry fixed effects):
  - y_jit = β_c C_jit + β_x X_jit + μ_i + η_s + ε_jit.
- Sample/time coverage:
  - TFP dependent variable: observations from 2004 only.
  - Labor productivity: observations from 2001 and 2004 with time dummy.

### III. Empirical relation — main empirical findings and non-linearity
- Markups (OLS and quantile analysis):
  - Markups are negatively correlated with all three productivity measures after controls; coefficients significant at the 5 percent level in all three cases.
  - Firms with 20 percent higher markups have, on average:
    - 1.2 percent lower corporate TFP levels (Olley and Pakes or Caves).
    - 8 percent lower labor productivity.
  - Example OLS coefficient: ln-markup on lnTFP-OP = −.062; LAD (median) = −.038.
  - Quantile regressions (18 quantiles): the positive competition–productivity correlation increases with firm productivity level (larger impact for most productive firms).
- Cost-competition (comp-cost) results:
  - OLS β_c positive, but only significant for labor productivity.
  - comp-cost positive and significant for TFP-OP up to the 40 percent quantile; not significant thereafter.
  - comp-cost indicator limited (1–4), constraining resolution for non-linear patterns; labor productivity vs. cost competition appears hump-shaped.
- Additional firm-level controls:
  - Exporting and foreign firms more productive (consistent with Melitz (2003); Helpman, Melitz, and Yeaple (2004)).
  - Firms that apply international audit standards are more productive.
  - Larger and younger firms associated with higher TFP.
  - TFP-Caves index: not significantly correlated at 5 percent level with foreign ownership, exporting, or auditing status; firms facing stronger hiring restrictions have higher TFP index (Scarpetta et al. (2002)).
- Overall conclusion of section III:
  - Positive correlation between competition (measured by markups) and productivity in 27 Eastern European and Central Asian emerging economies.
  - Corroboration for two of three productivity variables when using subjective cost-competition indicator.
  - Average OLS coefficients mask a systematic non-linear pattern: positive correlation between competition and productivity increases with firm productivity.

### IV. Direction of causality — A. Instrumental Variable (IV) approach
- Internal instruments:
  - Changes in markups between 2001 and 2004 (for panel subset).
  - Lagged values of qualitative competition indicators.
  - Example qualitative indicator: pressure from domestic competitors on decisions about developing new products or services and markets.
- External instruments:
  - Country-level index on costs of starting a business (Fraser database).
  - Firm-level qualitative indicator (0 to 4): “How problematic are anti-competitive practices of other producers for the operation and growth of your business?”
- Justification:
  - Aggregate entry-cost index varies at country-level and is assumed not affected by firm-level productivity.
  - Subjective anti-competitive practices measure approximates firm’s exposure to industry competition, affecting productivity only indirectly via competition and controls.
- First-stage IV diagnostics:
  - Both external instruments significantly correlated with cost competition in 2004; coefficients significant at the 1 percent level.
  - Wald-test: reject null that both instruments jointly equal zero at the 1 percent level.
  - Correlation between the two instruments = .073.
- IV estimation results:
  - IV regression (external instruments): positive causal impact of cost competition on TFP-OP; coefficient significant at the 5 percent level.
  - IV regressions including internal instruments (284 firms with data in both periods): instruments significantly correlated with cost-competition; Wald-test rejects null at the 1 percent level; results confirm causal effect of cost-competition on Olley–Pakes TFP.
  - TFP-Caves index: ambiguous — internal IV finds positive causal effect; external IV finds negative effect.
  - Labor productivity: positive causal impact of competition in both IV specifications.
  - Hansen test statistics: cannot reject validity of instruments in all cases where positive causal impact from competition on productivity is found.
- IV approach summary:
  - Evidence of causal relation from competition to productivity for Olley and Pakes (1996) TFP measure and for labor productivity; ambiguous for the Caves TFP index.

### IV. Direction of causality — B. Natural experiment (difference-in-difference from entry deregulation)
- Rationale:
  - Several Eastern European and Central Asian economies implemented entry deregulation in early 2000s with varying timing; deregulation treated as exogenous shock to competition.
  - Comparable non-reformer control group enables difference-in-difference estimation.
- Difference-in-difference specification:
  - ∆y_jit = β_1 R_ji + β_2 X_jit + β_3 F_it + η_s + ε_jit (j = 1,...,N; i = 1,...,16; t = 2001, 2004).
  - ∆y = TFP growth; R_ji = 1 if firm j operates in a reformer country.
  - F_i includes aggregate GDP (Penn World Tables) and aggregate reform indices (Fraser).
- TFP growth measurement and correction:
  - Firm-level factor shares of human and physical capital and firm sales as output.
  - Adjustment following Kim (2000) to correct for bias from imperfect competition and non-constant returns to scale.
- Difference-in-difference results:
  - Estimate for β_1 between .19 percent and .25 percent: TFP growth in reformer countries higher by between .19 percent and .25 percent (2001–2004).
  - Robust to inclusion of regulation differences, industry and firm controls.
  - Coefficients significantly different from zero at the 1 percent level.
  - Benchmark: average growth in TFP for all countries between 2001 and 2004 = 1.64 percent (median = 2 percent).
  - Interpretation: reformer country productivity growth would be between one seventh and one ninth higher than non-reformer on average — i.e., between 12 and 15 percent of productivity growth explained by competition spurred by reforms (or between 10 and 13 percent if relative to median).
- Difference-in-difference on competition measures:
  - Interaction d_reform04 = 1 if country is reformer and year is 2004.
  - Estimated interaction coefficient for ln-markup = −.097, significant at the 5 percent level.
  - Conclusion: increase in competition more pronounced in reformer countries relative to non-reformers between 2001 and 2004.
- Natural experiment summary:
  - Reformer countries experienced higher TFP growth and stronger increases in competition following entry deregulation.
  - Given exogenous, politically-motivated nature of reforms and comparable trade/financial/labor regulation changes, results point to causal effect of entry deregulation on firm-level TFP growth.

### Key quantitative results (preserved exactly)
- Reformer group index increase: average 69 percent; largest increase 147 percent in Romania.
- Non-reformer group index change: decline on average by 1.4 percent.
- Poland index increase: 22.3 percent; Czech Republic index increase: 10.6 percent.
- Group average index levels: 2004 = 5.8; 2001 = 5.6.
- Firms with 20 percent higher markups → 1.2 percent lower corporate TFP and 8 percent lower labor productivity.
- OLS ln-markup coefficient on lnTFP-OP example: −.062; LAD median = −.038.
- Correlation between external instruments = .073.
- IV sample with internal instruments: 284 firms with data in both periods.
- Difference-in-difference β_1 estimate range: between .19 percent and .25 percent.
- Average TFP growth across sample (2001–2004): 1.64 percent; median = 2 percent.
- Reform × time interaction on ln-markup = −.097 (significant at 5 percent).

### Conclusion and overall finding
- The study of Eastern Europe and Central Asia firms during widespread product-market reforms finds a positive causal relationship from competition to productivity.
- Representative magnitudes:
  - Firms with 20 percent higher markups have, on average, 1.2 percent lower TFP levels and 8 percent lower labor productivity.
  - Competition induced by entry deregulation explains around 12–15 percent of productivity growth in reformer countries over 2001–2004.
- Results robust across pooled OLS, quantile regressions, IV estimations, manufacturing subsample, and a natural experiment difference-in-difference design.

### Appendix summaries
- Appendix A — Survey overview:
  - Data source: World Bank Enterprise Surveys covering firms in 71 countries.
  - Standardized sample covers an average of 434 establishments per country/time period; minimal coverage 170 establishments (Albania 2002); maximum 4,234 (India 2006).
  - Panel availability: standardized instruments available for multiple periods for 39 countries; regional composition: 27 Europe and Central Asia, 7 Latin America, 3 African, 2 South Asian.
  - Industry coverage: manufacturing (group D), construction (group F), services (groups G and H), transport/storage/communications (group I).
  - Firm-size stratification: small (5–19 employees), medium (20–99), large (100 or more); targeted establishments: at least five full-time employees.
  - Survey administration: face-to-face interviews; interviewer training and exams; confidentiality; core survey completion ~45 minutes.
- Appendix B — Estimation of production functions under simultaneity:
  - Problem: OLS biased if inputs contemporaneously correlated with technology shock.
  - Chosen approach: Olley and Pakes (1996) two-step procedure; assumptions listed (labor sole contemporaneous responder, capital predetermined, investments monotone in productivity shock).
  - Two-stage outline: first-stage series estimator for φ_t(i,k) yields α_h; second stage uses Markov assumption on ω_jt and non-linear least squares to estimate α_k.
  - Practical note: Olley and Pakes requirements may eliminate firms due to data availability; Levinsohn and Petrin (2003) motivate alternative with intermediate inputs.
- Appendix C — Computation of TFP growth and the Kim (2000) correction:
  - Production function: Y_jt = A_jt F(K_jt, H_jt, M_jt).
  - Log-differenced TFP growth: dTFP_it = dy_jt / y_jt − α_kj dk_jt / k_jt + α_hj dh_jt / h_jt + α_mj dm_jt / m_jt.
  - α_mj and α_hj measured by firm cost shares; α_kj = 1 − α_hj − α_mj.
  - Bias from imperfect competition and non-constant returns to scale characterized via Hall (1988) decomposition.
  - Kim (2000) correction implemented via regression (equation (10)) including dz_ijt = α_hj dln(H_jt/K_jt) + α_mj dln(M_jt/K_jt); β_1 measures relation between reform and an unbiased TFP growth.

*Source: _wp1067 - References*

### References .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .

### _wp1067 - References .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .

### Introduction: research question and main findings
- Research objective: investigate empirically the impact of competition on firm productivity using firm observations from the World Bank Enterprise Survey database.
- Main empirical findings:
  - A positive and robust causal relationship between proxies for competition at the firm level and measures of firm productivity.
  - Firm productivity growth is higher in countries with more substantial product-market reforms.
  - Countries that reformed during the period experienced a more pronounced increase in competition, providing a channel by which productivity is affected by reform.
- Context and motivation:
  - Product market regulations vary substantially across countries; example comparative start-up costs in 2003 (Doing Business database): 0.7 percent of GDP per capita in the United States; 16.8 percent in Italy; 1316.4 percent in Angola.
  - Theoretical ambiguity: standard endogenous growth models predict competition may reduce innovation (“rent dissipation effect” and shortened durability of innovations), while extensions (Aghion et al.) show a possible positive “escape competition” incentive to innovate depending on distance to the technological frontier.
- Empirical literature overview:
  - Mixed evidence on the effect of product-market deregulation on productivity and growth; cited results include positive effects (Loayza, Oviedo, and Serven (2002); Barseghyan (2008)) and negative relationships between deregulation and innovation/R&D in some studies.
  - Microdata approaches improve identification but face measurement challenges for product-market regulations and endogeneity of industry-level regulatory measures.
- Contribution of this study:
  - Uses firm-level measures of competition and exogenous variation in product-market reforms across countries to draw causal inferences while controlling for firm-level institutional, financial, and labor market frictions.
  - Leverages detailed firm employee education data to improve human-capital-adjusted productivity measures.
  - Focus on emerging-market countries where productivity improvements were the main source of growth from 1999–2005 and where a wave of product market deregulations occurred in some, but not all, countries.

### Data and empirical methodology
- Data coverage and timing:
  - Firm data for 27 countries in 2004.
  - Use of World Bank Enterprise Survey data for 2004 (and 2001 for panel/growth measures where available).
- Productivity measurement (four different approaches described):
  - First: firm-level total factor productivity (TFP) as a residual from an augmented Cobb–Douglas production function, estimating input elasticities using the Olley and Pakes (1996) two-step procedure to correct endogeneity rather than assuming factor shares equal elasticities.
    - Country-level input elasticity regression (in logs): y_ji = θ^k_i k_ji + θ^h_i h_ji + η_s + ε_ji (equation (1)), where y_ji is firm j’s value added in country i, k_ji physical capital, h_ji human-capital-adjusted labor, η_s industry fixed effects, and θ_i = (θ_ik, θ_ih).
    - Physical capital measured as net book value of fixed assets after depreciation; output = value added (2004).
    - Human capital H measured as L × exp^{φ(s)} with φ(s) linear slope 0.10 (average returns on education for an additional year of schooling in emerging economies).
    - Olley–Pakes correction performance: human capital coefficients move in the predicted direction in 18 out of 19 cases; capital coefficient increases in 15 out of 19 cases.
  - Second: TFP index relative to the industry median following Caves, Christensen, and Diewert (1982), using observed firm factor shares (assumes constant returns to scale) and including intermediates; TFP expressed as equation (2) (index form) using sales, intermediate inputs (m), physical and human capital relative to industry median values.
  - Third: firm labor productivity measured as sales per human capital (used as a benchmark).
  - Fourth: in the natural experiment reported in Section IV.B, the growth rate of total factor productivity (TFP growth from 2001 to 2004 measured as a Solow residual; methodology detailed in Appendix C).
- Measurement and estimation details:
  - All output and capital series deflated using aggregate price levels from the Penn World Tables.
  - Industry-specific effects based on one-digit NACE codes.
  - Use of Olley and Pakes (1996) two-step procedure to address input demand endogeneity; lagged investment and human capital from the World Bank Surveys (2004 and 2001) used as instruments in the procedure.
  - Table 1 compares Olley–Pakes versus OLS input elasticities; Table 10 provides input elasticities per industry per country using the Caves methodology.
  - Correlations across productivity measures reported in Table 2 (positive and relatively high but not uniform); Figure 1 shows distributions of the three productivity measures (skewed right, heterogeneity across firms).
- Competition measurement and controls:
  - Firm-level competition proxies from World Bank Enterprise Survey:
    - ln-markup: mark-ups measured as sales over operating costs.
    - comp-cost: cost competition (discrete 1–4) from question on pressure from domestic competitors to reduce production costs (1 = low importance to 4 = high importance).
    - comp-import: qualitative external competition from imports, values 1 to 6 (1 = low import competition; 6 = high import competition).
  - Controls for corporate and market frictions:
    - Firm characteristics: age (age), size (size) categories: small <50 employees, medium 50–249, large ≥250; foreign (headquarters abroad), exports (exporter dummy), privatized (established via privatization), new techno (introduced substantially new technology).
    - Financing and input-market controls: ret-earn (percentage of investments financed from retained earnings), stock (listed dummy), hire (extent of hiring restrictions measured as manager’s estimate of percent workforce increase absent restrictions).
    - Institutional controls: audit (audit variable), pow-outages (number of power outages per year), juris (affordable jurisdiction qualitative 1–6, 1 = “never affordable”, 6 = “always affordable”).
  - Panel and subgroup considerations:
    - Manufacturing and wholesale sectors discussed; sector-level estimation limited by observation counts except manufacturing and wholesale (Table 9 reports manufacturing results).
    - The study includes a natural experiment leveraging cross-country variation in reform timing to identify causal links (Section IV).
- Data limitations noted:
  - Intermediate input information available in 2004 but not 2001, constraining some measures to 2004.
  - Inventories information not available; assumption that inventory-to-sales ratio is uncorrelated with firm capital or human capital.
  - Disaggregated price-level deflators not available; country fixed effects and firm characteristics used to account for vertical shifts across countries and industries.

*Source: _wp1067 - References .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .*

### 2004. From these statistics we can infer a high degree of dispersion in firm productivity, as was

### _wp1067 - 2004. From these statistics we can infer a high degree of dispersion in firm productivity, as was

### C. Reforms — data and classification
- Data source for product market regulation changes: Economic Freedom of the World Index (Fraser Institute) — annual index on “ease of starting a new business” for 1995 and annually from 2000–2004.
- Index construction (three variables averaged): (i) number of days to comply with regulations, (ii) money costs of fees paid to regulatory authorities measured as a share of per-capita income, (iii) minimum capital requirements measured as a share of per-capita income.
- Index range: 1 to 10 (decreasing order of entry costs).
- Reformer definition: country classified as a reformer if the Fraser index improved by at least 40 percent from 2001 to 2004.
- Empirical grouping and summary statistics:
  - Reformer group: 10 countries; index increased by an average of 69 percent (largest increase 147 percent in Romania).
  - Non-reformer group: index declined on average by 1.4 percent (i.e., increased entry costs); exceptions Poland and the Czech Republic where index increased by 22.3 percent and 10.6 percent respectively (classified as non-reformers based on 1995–2004 performance).
  - Group average index levels: 2004 = 5.8 and 2001 = 5.6 (reflecting similar index levels overall).
- Complementary regulatory indices (trade, financial, labor) from Fraser: do not reveal important changes (other than entry costs) and changes do not differ between reformer and non-reformer groups.
- Median TFP growth: period average median TFP growth for reformer countries is about double corresponding growth for non-reformer countries.

### III. Empirical relation between TFP and competition — model and measures
- Productivity measures (log):
  - lnTFP-OP (Olley and Pakes, 1996)
  - lnTFP-Caves (Caves, Christensen, and Diewert, 1982)
  - ln-laborprod
- Competition measures:
  - lnmarkup
  - comp-cost (perceived pressure to reduce costs from domestic competitors, qualitative)
- Estimation equation (pooled, country and industry fixed effects):
  - y_jit = β_c C_jit + β_x X_jit + μ_i + η_s + ε_jit, j = 1,2,...,N and i = 1,...,27. (3)
- Sample/time coverage:
  - When dependent variable is TFP, observations from 2004 only.
  - For labor productivity, observations from 2001 and 2004 with time dummy.

### III. Empirical relation — main empirical findings and non-linearity
- Markups (OLS results from Table 6 and quantile analysis):
  - Markups are negatively correlated with all three productivity measures after controls; coefficients significant at the 5 percent level in all three cases.
  - Firms with 20 percent higher markups have, on average:
    - 1.2 percent lower corporate TFP levels (Olley and Pakes (1996) or Caves, Christensen, and Diewert (1982))
    - 8 percent lower labor productivity
  - Example OLS coefficient cited: ln-markup on lnTFP-OP = −.062 (dashed line in figure); LAD (median) = −.038.
  - Quantile regressions (18 quantiles, 5 percent increments to the 95th): for the most productive firms (higher quantiles), quantile coefficients reflect a larger impact from competition via markups — i.e., the positive competition-productivity correlation increases with firm productivity level (consistent with Aghion et al. (2001)).
- Cost-competition (comp-cost) results:
  - OLS β_c positive, but only the estimator for labor productivity is significant.
  - Cost-competition coefficient is positive and significant for TFP-OP up to the 40 percent quantile, not significant thereafter.
  - comp-cost indicator varies only from 1 to 4 (limiting resolution for non-linear patterns).
  - Relationship between labor productivity and cost competition appears hump-shaped.
- Additional firm-level controls:
  - Exporting and foreign firms are more productive (consistent with Melitz (2003); Helpman, Melitz, and Yeaple (2004)).
  - Firms that apply international audit standards (better access to finance) are more productive.
  - Larger and younger firms associated with higher TFP.
  - TFP-Caves index: not significantly correlated at 5 percent level with foreign ownership, exporting, or auditing status; firms facing stronger hiring restrictions have higher TFP index (Scarpetta et al. (2002)).
- Overall conclusion of section III:
  - Positive correlation between competition (measured by markups) and productivity in a sample of 27 Eastern European and Central Asian emerging economies.
  - Corroboration for two of three productivity variables when using subjective cost-competition indicator.
  - Average OLS coefficients conceal systematic non-linear pattern: positive correlation between competition and productivity increases with firm productivity.

### IV. Direction of causality — A. Instrumental Variable (IV) approach
- Internal instruments:
  - Changes in markups between 2001 and 2004 (for panel subset).
  - Lagged values of qualitative competition indicators.
  - Qualitative indicator example: response to “How important is pressure from domestic competitors on key decisions about your business with respect to developing new products or services and markets?”
- External instruments:
  - Country-level index on costs of starting a business (Fraser database; Table 4).
  - Firm-level qualitative indicator (0 to 4 scale): response to “How problematic are anti-competitive practices of other producers for the operation and growth of your business?”
- Justification:
  - Aggregate entry-cost index varies at country-level and is assumed not affected by firm-level productivity.
  - Subjective anti-competitive practices measure assumed to approximate firm’s exposure to industry competition, affecting productivity only indirectly via competition and controls.
- First-stage IV diagnostics (Table 7 summary):
  - Both external instruments significantly correlated with measure of cost competition in 2004; coefficients significant at the 1 percent level.
  - Wald-test: null that both instruments jointly equal zero is rejected at the 1 percent level.
  - Correlation coefficient between the two instruments = .073 (indicating they provide distinct information).
- IV estimation results (Table 7 summary):
  - IV regression (external instruments): positive causal impact of cost competition on TFP-OP; coefficient significant at the 5 percent level.
  - IV regressions including internal instruments (lagged markups and competition in technologies) using 284 firms with data in both periods: instruments significantly correlated with cost-competition; Wald-test rejecting null at the 1 percent level; results confirm causal effect of cost-competition on Olley–Pakes TFP.
  - TFP-Caves index: ambiguous results — internal IV regression finds positive causal effect; external IV regression finds negative effect.
  - Labor productivity: positive causal impact of competition on labor productivity in both IV specifications.
  - Hansen test statistics: cannot reject validity of instruments in all cases where positive causal impact from competition on productivity is found.
- Summary conclusion of IV approach:
  - Evidence of causal relation from competition to productivity for Olley and Pakes (1996) TFP measure and for labor productivity; ambiguous evidence for the Caves, Christensen, and Diewert (1982) TFP index.

### IV. Direction of causality — B. Natural experiment (difference-in-difference from entry deregulation)
- Natural experiment rationale:
  - Several Eastern European and Central Asian economies implemented entry deregulation in early 2000s with varying timing (e.g., Slovak Republic 2002/2003; Czech Republic later).
  - Deregulation viewed as politically-motivated and not endogenously determined by country productivity performance — treated as exogenous shock to competition.
  - Existence of comparable non-reformer control group enables difference-in-difference estimation.
- Difference-in-difference specification:
  - ∆y_jit = β_1 R_ji + β_2 X_jit + β_3 F_it + η_s + ε_jit (4)
  - j = 1,...,N; i = 1,...,16; t = 2001, 2004
  - ∆y = TFP growth; R_ji = 1 if firm j operates in a reformer country, 0 otherwise
  - F_i includes aggregate GDP (Penn World Tables) and aggregate reform indices (Fraser)
- TFP growth measurement:
  - Firm-level factor shares of human and physical capital and firm sales as output.
  - Adjustment following Kim (2000) to correct for bias arising from imperfect competition and non-constant returns to scale.
- Difference-in-difference results (Table 8 summary):
  - Estimate for β_1 between .19 percent and .25 percent. That is, TFP growth in reformer countries is higher than in non-reformer economies by between .19 percent and .25 percent (2001–2004).
  - Robust to inclusion of regulation differences, industry and firm controls.
  - Statistical significance: corresponding coefficients significantly different from zero at the 1 percent level.
  - Benchmark context:
    - Average growth in TFP for all countries in sample between 2001 and 2004 = 1.64 percent (median = 2 percent).
    - Interpretation: reformer country productivity growth would be between one seventh and one ninth higher than non-reformer on average — i.e., between 12 and 15 percent of productivity growth explained by competition spurred by reforms (or between 10 and 13 percent if relative to median).
- Difference-in-difference on competition measures:
  - Interaction variable d_reform04 = 1 if country is reformer and year is 2004, 0 otherwise.
  - Expectation: interaction coefficient negative for ln-markup and positive for cost-comp.
  - Estimated interaction coefficient for ln-markup = −.097, significant at the 5 percent level (last two columns of Table 8).
  - Conclusion: increase in level of competition more pronounced in reformer countries relative to non-reformers between 2001 and 2004.
- Overall conclusion of natural experiment:
  - Reformer countries experienced higher TFP growth and stronger increase in degree of competition following entry deregulation.
  - Given the exogenous, politically-motivated nature of product market reforms and relative homogeneity of reformer and non-reformer groups on trade, financial, and labor regulations, results point to a causal effect of entry deregulation on firm-level TFP growth in the Eastern European and Central Asian countries.

### Key quantitative results (exact figures preserved)
- Reformer group index increase: average 69 percent; largest increase 147 percent in Romania.
- Non-reformer group index change: decline on average by 1.4 percent.
- Poland index increase: 22.3 percent; Czech Republic index increase: 10.6 percent (both classified as non-reformers based on 1995–2004 performance).
- Group average index levels: 2004 = 5.8; 2001 = 5.6.
- Firms with 20 percent higher markups → 1.2 percent lower corporate TFP (Olley–Pakes or Caves) and 8 percent lower labor productivity.
- OLS ln-markup coefficient on lnTFP-OP example: −.062; LAD median = −.038.
- Correlation between external instruments = .073.
- IV sample with internal instruments: 284 firms with data in both periods.
- Difference-in-difference β_1 estimate range: between .19 percent and .25 percent.
- Average TFP growth across sample (2001–2004): 1.64 percent; median = 2 percent.
- Reform × time interaction on ln-markup = −.097 (significant at 5 percent).

*Source: _wp1067 (2004) — extracted content unit as supplied.*

### Section III.

### Section III.

### Conclusion and Overall Finding
- The study uses firm-level data for countries in Eastern Europe and Central Asia during a period of widespread product-market reforms aimed at EU accession.
- There is a positive causal relationship from competition to productivity.
- Firms that have 20 percent higher markups have, on average:
  - 1.2 percent lower TFP levels.
  - 8 percent lower labor productivity.
- Countries that reformed experienced a more pronounced increase in competition; the contribution to productivity growth due to competition spurred by these reforms is around 12–15 percent.

### Natural-Experiment Comparison: Reformers vs Non-Reformers
- Because reform timing and intensity varied, the paper exploits a natural experiment comparing reformers and non-reformers.
- Reformer countries (listed in the study) show larger increases in competition measures and higher associated productivity gains than non-reformers.
- Table-based summaries in the source provide cross-country comparisons of changes in entry costs, trade, credit, labor market regulations, GDP, and TFP growth for reformers and non-reformers (2001–2004).

### Key Econometric Results (selected coefficients and findings)
- Pooled estimations (Table 6, estimation period: 2004; heteroscedasticity robust s.e.; ***,**, * denote significance at 1%, 5%, 10%):
  - ln-markup coefficients: -.062 (TFP-OP), -.0591 (TFP-OP), -.0397 (TFP-Caves) with significance noted in table.
  - comp-cost reported as .0275 in one specification.
  - Firm characteristics with notable coefficients: size (.1967, .2507, .1266, .1432 across columns), for (.1127), exp (.0859, .1181), privatized negative coefficients (e.g., -.0499, -.2238).
  - Estimation sample sizes reported (e.g., Countr./Firms 19/2054; 19/2143; 26/2400; 26/2498; 27/6163; 27/6535) and Industry FE: yes in all columns; Country FE: varies.
- IV estimations (Table 7, estimation period: 2004; heteroscedasticity robust s.e.; ***,**, * significance):
  - comp-cost coefficient examples: .3038 ∗∗; .6666 ∗∗; -.1055 ∗; .2766 ∗; .6251 ∗∗∗; 1.05 ∗∗∗ (t-statistics shown in table).
  - anti-comp04 coefficient .1708 ∗∗∗; .1695 ∗∗∗; .1744 ∗∗∗ (t-statistics indicate strong significance).
  - aggr. entry04 coefficients .0871 ∗∗∗; .1574 ∗∗; .0921 ∗∗∗; .0749 ∗∗∗.
  - ∆ lnmarkup01-04 coefficients around -.1014 ∗; -.1034 ∗; -.0807 ∗ (negative and marginally significant).
  - comp-techno01 positive and significant (.2113 ∗∗∗; .1818 ∗∗∗; .1375 ∗∗∗).
  - Country/Firms sample sizes reported across columns (e.g., 16/1707; 19/284; 26/438; 27/4122; 27/543).
  - Reported Wald-test and Hansen test values in table; correlation between anti-comp04 and aggr. entry04 is .073.
- Reformers vs Non-Reformers panel (Table 8, estimation period: 2001 and 2004):
  - dreform coefficient .2483 ∗∗∗; .1937 ∗∗∗; .2517 ∗∗∗ in different specifications.
  - d04 coefficient .2822 ∗∗∗; .1097 ∗∗∗.
  - interaction dreform04 coefficient -.0969 ∗∗ (negative in one specification) and .0470 (insignificant in another).
  - In the reformer vs non-reformer split, Reformer countries identified: Bulgaria, Latvia, Lithuania, Romania, Russia, Slovak Republic, Turkey, Ukraine, Georgia and Macedonia. Non-reformers: Croatia, Czech Republic, Estonia, Hungary, Poland, Slovenia.
- Manufacturing-specific pooled estimations (Table 9, estimation period: 2004):
  - ln-markup coefficients near -.019, -.018, 1.0198 in different columns (t-statistics reported).
  - comp-cost .1322 ∗∗∗; .0013; .166 ∗∗∗ in columns shown.
  - Other firm controls (age, size, for, exp, stock, privatized, import comp., power-out) show mixed significant effects; sample sizes and FE structure listed in table.

### Additional Empirical Details and Data Notes
- The production-function estimation is based on the log production function (2004): y_jsi = θ_k_i k_jsi + θ_h_i h_jsi + η_s + ε_jsi, where y is measured by value added.
- Table 1 and Table 10 in the source report estimated input elasticities (Olley–Pakes and Caves methods) by country and by ISIC sector; differences across countries/sectors are tabulated in the source.
- Correlation coefficients for productivity measures (2004, Table 2):
  - lnTFP-OP with itself 1.0000.
  - lnTFP-Caves correlation with lnTFP-OP is 0.42601.
  - Ln-laborprod correlations: 0.45130 with lnTFP-OP; 0.21461 with lnTFP-Caves; 1.0000 with itself.
- Descriptive statistics (selected, Table 3): sample sizes and means/SDs reported for variables including lnTFP-OP, lnTFP-Caves, ln-laborprod, ln-markup, comp-cost, age, size, for, exp, stock, privatized, comp-import, power-outages, ret-earn, audit, juris, hire, new techno for years 2001 and 2004.

### Major Quantitative Takeaways
- A 20 percent higher markup is associated with:
  - 1.2 percent lower TFP.
  - 8 percent lower labor productivity.
- Productivity contribution from competition induced by reforms: around 12–15 percent.
- Several competition-related instruments and controls (comp-cost, anti-comp04, aggr. entry04, comp-techno01, ∆ lnmarkup01-04) show statistically significant associations with productivity measures in IV and pooled specifications.
- Results are robust across specifications including pooled, IV, industry-specific (manufacturing), and reformer/non-reformer splits; standard errors are heteroscedasticity robust and significance levels are reported in tables.

*Italic: Source: _wp1067 - Section III.*

### REFERENCES

### REFERENCES

### Bibliographic entries
- Aghion, P., R. Burgess, S. Redding, and F. Zilibotti, 2005, “Entry Liberalization and Inequality in Industrial Performance,” Journal of the European Economic Association, Papers and Proceedings, vol. 3, no. 2–3, pp. 291–302.
- Aghion, P., C. Harris, P. Howitt, and J. Vickers, 2001, “Competition, Imitation and Growth with Step-by-Step Innovation,” Review of Economic Studies, vol. 68, pp. 467–492.
- Aghion, P., and P. Howitt, 1992, “A Model of Growth Through Creative Destruction,” Econometrica, vol. 2, pp. 323–351.
- Barseghyan, L., 2008, “Entry Costs and Cross-Country Differences in Productivity and Output,” Journal of Economic Growth, vol. 13, pp. 145–167.
- Bartelsman, E., and M. Doms, 2000, “Understanding Productivity: Lessons from Longitudinal Microdata,” Journal of Economic Literature, vol. 38, pp. 569–594.
- Beck, T., and R. Levine, 1999, “A new database on financial development and structure,” World Bank Policy Research Working Paper 2146.
- Bergoeing, R., N. Loayza, and A. Repetto, 2004, “Slow recoveries,” Journal of Economic Development, vol. 2, no. 75, pp. 473–506.
- Blundell, R., and S. Bond, 2000, “GMM Estimation with persistent panel data: an application to production functions,” Econometric Reviews, vol. 19, pp. 321–340.
- Calderón, C., and L. Servén, 2005, “The Effects of Infrastructure Development on Growth and Income Distribution,” World Bank Working Paper, WPS3400.
- Caselli, Francesco, 2005, “Accounting for Cross-Country Income Differences,” in Philippe Aghion and Steven N. Durlauf (Eds.), Handbook of Economic Growth, Volume 1A, Handbooks in Economics 22, chap. 9, pp. 679–741 (Amsterdam: North Holland).
- Caves, D., L. Christensen, and W. Diewert, 1982, “The Economic Theory of Index Numbers And the Measurement of Input, Output, and Productivity,” Econometrica, vol. 6, no. 50, pp. 1393–1414.
- Christopoulos, D., and E. Tsionas, 2004, “Financial development and economic growth: evidence from panel unit root and cointegration tests,” Journal of Development Economics, vol. 73, no. 1, pp. 55–74.
- Cincera, M., and O. Galgau, 2005, “Impact of Market Entry and Exit on EU Productivity and Growth Performance,” Economic Papers 222, European Economy, European Commission, Directorate-General for Economic and Financial Affairs.
- Cole, H., L. Ohanian, A. Riascos, and J Schmitz, 2005, “Latin America in the Rearview Mirror,” Journal of Monetary Economics, vol. 52, no. 1, pp. 69–107.
- Griffith, R., and R. Harrison, 2004, “The Link Between Product Market Reform and Macroeconomic Performance,” , no. 209.
- Hall, R., 1988, “The Relation between Price and Marginal Cost in U.S. Industry,” Journal of Political Economy, vol. 96, no. 5, pp. 921–947.
- Hall, R., and C. Jones, 1999, “Why do some countries produce so much more output than others?” Quarterly Journal of Economics, vol. 114, no. 1, pp. 83–116.
- Hallward-Driemeier, M., S. Wallsten, and L. Xu, 2003, “The Investment Climate and the Firm: Firm-Level Evidence from China,” World Bank Policy Research Working Paper No. 3003.
- Helpman, E., M. Melitz, and S. Yeaple, 2004, “Exports versus FDI with Heterogeneous Firms,” American Economic Review, vol. 94, pp. 300–316.
- Kaplan, D., E. Piedra, and E. Seira, 2007, “Entry regulations and Business Start-Ups: Evidence from Mexico,” Policy Research Working Paper.
- Kim, E., 2000, “Trade liberalization and productivity growth in Korean manufacturing industries: price protection, market power, and scale efficiency,” Journal of Development Economics, vol. 62, pp. 58–83.
- Koenker, R., and G. Bassett, 1978, “Regression Quantiles,” Econometrica, vol. 46, pp. 33–50.
- Levinsohn, J., and A. Petrin, 2003, “Estimating Production Functions Using Inputs to Control for Unobservables,” Review of Economic Studies, vol. 70, pp. 317–341.
- Loayza, N., Oviedo, and L. Serven, 2002, “The Role of Policy and Institutions for Productivity and Firm Dynamics: Evidence from Micro and Industry Data,” Working Papers 329, OECD Economics Department.
- Maudos, J., J. Pastora, and L. Serrano, 1999, “Total Factor Productivity Measurement and Human Capital in OECD Countries,” Economic Letters, vol. 63, pp. 39–44.
- Melitz, M., 2003, “The Impact of Trade on intra-industry Reallocations and Aggregate Productivity,” Econometrica, vol. 71, pp. 1695–1725.
- Nickell, S., 1996, “Competition and Corporate Performance,” Journal of Political Economy, vol. 104, pp. 724–746.
- Nicoletti, G., and S. Scarpetta, 2003, “Regulation, Productivity and Growth,” Economic Policy, vol. 36, no. 18, pp. 11–72.
- Olley, G., and A. Pakes, 1996, “The Dynamics of Productivity in the Telecommunications Equipment Industry,” Econometrica, vol. 64, pp. 1263–1279.
- Poschke, M., 2009, “The regulation of entry and aggregate productivity,” EUI Working Paper.
- Ramaswamy, K. V., 1999, “Productivity Growth, Protection and Plant Entry in a Deregulating Economy: The Case of India,” Small Business Economics, vol. 13, pp. 131–139.
- Romer, P., 1986, “Increasing Returns and Long-run Growth,” Journal of Political Economy, vol. 94, no. 5, pp. 1002–1037.
- Scarpetta, S., P. Hemmings, T. Tressel, and J. Woo, 2002, “The Role of Policy and Institutions for Productivity and Firm Dynamics,” Working Paper 329, OECD Economics Department.
- Schiantarelli, F., 2005, “Product Market Regulation and Macroeconomic Performance: A Review of Cross Country Evidence,” , no. 1791.
- Srivastava, V., 1996, Liberalization, Productivity and Competition (Delhi: Oxford University Press).
- Van Biesebroeck, J., 2003, “Revisiting Some Productivity Debates,” Working Paper Series 10065, NBER.
- Van Biesebroeck, J., 2005, “Firm Size Matters: Growth and Productivity Growth in African Manufacturing,” Economic Development and Cultural Change, vol. 53, no. 3, pp. 545–583.

### Page numbering notes
- Pages referenced in this unit: 29, 30, 31.

---

### APPENDIX A. SURVEY OVERVIEW

- Data source: The World Bank’s Enterprise Surveys covering firms in 71 countries.
- Core features:
  - Questionnaires are country specific and changed over time, but contain core questions asked in all countries, for all industries, and in all time periods to generate internationally comparable data.
  - Sampling methodology is the same in all countries to provide panel data sets with minimal measurement errors.
- Coverage of topics: balance sheet data and manager’s assessments on obstacles to enterprise growth, including infrastructure services, access to international markets, degree of competition and innovation, corruption, crime, business-government relations, finance, quality of the workforce.
- Panel availability:
  - Standardized instruments available for multiple time periods for a subset of 39 countries.
  - Same core questions asked in 3 years intervals in these countries.
  - Regional composition of the 39 countries: 27 Europe and Central Asia, 7 Latin America, 3 African, and 2 South Asian countries.
- Industry and firm selection:
  - Industries considered (ISIC codes): all manufacturing sectors (group D), construction (group F), services (groups G and H), and transport, storage, and communications (group I).
  - Firm-size stratification: small (5–19 employees), medium (20–99 employees), large (100 or more employees).
  - Targeted establishments: at least five full-time employees (Full-time refers to a minimum of eight working hours per day).
- Sample size and precision:
  - Standardized sample covers an average of 434 establishments for each country and time period.
  - Minimal coverage: 170 establishments reported in Albania in 2002.
  - Maximum coverage: 4,234 establishments reported in India in 2006.
  - Sample size ensures a minimum precision of 7.5% for the 90% confidence interval about estimates of (i) the population proportion and (ii) the mean of log sales for key industries.
- Survey administration and data quality safeguards:
  - Face-to-face interviews with managing directors, accountants, or other relevant staff.
  - Interviewers and staff thoroughly trained; World Bank experts supervise training.
  - Interviewers must pass an exam to qualify.
  - Strict confidentiality: neither respondent name nor firm name is used in any document.
  - Publicity for survey launch to improve response incentives.
  - Core survey completion time approximately 45 minutes.
  - Missing data or inconsistencies are checked immediately by interviewer and field supervisor after interview and after filing of the data.
- Non-response handling:
  - Randomly selected firms that refuse or do not respond are replaced by willing participants to ensure sufficient sample size.
  - A field-work report lists reasons for non-responds, including refusals, by country, industry, and class of firm size.

---

### APPENDIX B. ESTIMATION OF PRODUCTION FUNCTIONS UNDER SIMULTANEITY

- Problem statement:
  - OLS estimation of model (1) may be biased if contemporaneous correlation exists between inputs (k_jt, h_jt) and the unobservable technology shock ε_jt.
  - Estimates ˆθ_k and ˆθ_h would be biased under simultaneity.
- Alternative approaches referenced:
  - Olley and Pakes (1996): use changes in firm’s investment decision as a proxy for the productivity shock; requires non-missing data for 2002 and 2003 on output, capital, labor, and investment.
  - Levinsohn and Petrin (2003): use intermediate inputs instead of investment as the proxy.
  - Blundell and Bond (2000): GMM system estimator; requires a long time-series of cross-sections.
- Chosen approach and assumptions:
  - Follow Olley and Pakes (1996) two-step procedure to estimate (1) consistently.
  - Required assumptions:
    - (i) labor is the only input which contemporaneously responds to a technology shock,
    - (ii) capital stock is predetermined and hence uncorrelated with a contemporary technology shock,
    - (iii) changes in corporate investment decisions depend on the contemporaneous technology shock and the capital stock of a firm,
    - (iv) investments are monotonically increasing in the technology shock for a given value of capital.
  - Under these assumptions, the technology shock can be instrumented as a function of capital and investment; estimation uses a series estimator.
- Model decomposition and two-stage estimation outline:
  - Decompose ε_jt into ω_jt + ε_jt where ω_jt is the productivity shock and ε_jt is the true error term.
  - Define age a_jt of a firm; write Y_jt,Olley–Pakes = α_0 + α_k k_jt + α_h h_jt + ω_jt + ε_jt (equation (5)).
  - Define investment function ι_t such that i_jt = ι_t(ω_jt, k_jt, a_jt); if investments monotone in ω_jt given k_jt, invert to ω_jt = m_t(i_jt, k_jt).
  - Rewrite as Y_jt = α_0 + α_k k_jt + α_h h_jt + m_t(i_jt, k_jt) + ε_jt and define φ_t(i_jt, k_jt) := α_0 + α_k k_jt + m_t(i_jt, k_jt).
  - Approximate φ_t by a third order polynomial series in k and i.
  - First-stage regression: Y_jt = α_h h_jt + φ_t(i_jt, k_jt) + ε_jt (equation (6)); OLS yields consistent α_h.
  - Second stage: assume ω_jt is first-order Markov: ω_jt = E(ω_jt | ω_j,t−1) + ξ_jt where ξ_jt uncorrelated with capital in period t.
  - Define v_jt as output net of the contributions of labor and substitute h_t−1(i_j,t−1, k_j,t−1) into g(φ_j,t−1 − α_k k_j,t−1), yielding v_jt = α_0 + α_k k_jt + g(·) + ξ_jt + ε_jt (equation (7)).
  - Restrict capital and lagged capital to have the same coefficient; estimate α_k efficiently and consistently by applying a non-linear least squares procedure to (7).
- Practical notes:
  - Olley and Pakes requirements may eliminate firms due to data availability (example: non-missing 2002 and 2003 variables).
  - Levinsohn and Petrin motivate their choice by weaker data requirements and argue intermediate inputs may be a better instrument than investment.
  - No correction for sample attrition is necessary because the database consists of two independent random samples of firms in 2001 and 2004.

---

### APPENDIX C. COMPUTATION OF TFP GROWTH AND THE KIM (2000) CORRECTION

- Production function and TFP growth definition:
  - Firm-level production function: Y_jt = A_jt F(K_jt, H_jt, M_jt) (equation (8)); all variables in real terms.
  - Log-differenced TFP growth (dTFP_it = da_jt / a_jt) expressed as:
    - dTFP_it = dy_jt / y_jt − α_kj dk_jt / k_jt + α_hj dh_jt / h_jt + α_mj dm_jt / m_jt
  - Input elasticities α_mj and α_hj measured by factor-level factor shares:
    - Factor shares measured by ratio of firm’s cost of employees over sales and firm’s expenses for intermediates over sales, respectively.
    - α_kj computed as 1 − α_hj − α_mj.
- Bias sources:
  - The above TFP growth measure assumes perfect competition and constant returns to scale and therefore is biased by imperfect competition and non-constant returns to scale.
- Hall (1988) decomposition:
  - Relationship between the standard biased TFP growth measure and the true TFP growth dTFP*_jt:
    - dTFP_jt = (μ_jt−1)[α_hj dln(H_jt/K_jt) + α_mj dln(M_jt/K_jt)] + (φ_jt−1) ln K_jt + dTFP*_jt (equation (9))
  - Definitions:
    - (μ_jt−1) = δ_dz_ij + δ_Rz × R_ij
    - (φ_jt−1) = δ_dK_ij + δ_RK × R_ij
    - δ_dz_ij + δ_Rz × R_ij captures effects of imperfect competition on TFP growth.
    - δ_dK_ij + δ_RK × R_ij captures effect of non-constant returns to scale.
- Kim (2000) correction and empirical specification:
  - Kim (2000) shows the following equation corrects for imperfect competition and non-constant returns to scale in the standard TFP growth measure:
    - dTFP_ijt = β_1 R_ij + β_2 X_ij,t−1 + β_3 F_i,t−1 + δ_dz_ij dz_ijt + δ_Rz (R_ij × dz_ijt) + δ_dK_ij dlnK_ijt + δ_RK (R_ij × dlnK_ijt) + ε_ijt (equation (10))
  - Where dz_ijt = α_hj dln(H_jt/K_jt) + α_mj dln(M_jt/K_jt).
  - Interpretation: coefficient β_1 measures the relation between a reform (R_ij) and an unbiased or “true” measure of TFP growth.

*Source: _wp1067 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1067.pdf_