## 1. State-level TFP Gains from Eliminating Resource Misallocation (in %)

## Source details

**Canonical URL:** [1. State-level TFP Gains from Eliminating Resource Misallocation (in %)](https://www.imf.org/-/media/files/publications/wp/2018/wp18112.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18112.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18112.pdf.json)

---

### I. Introduction
- Objective: provide evidence that resource misallocation rather than access to technology could lie at the heart of Mexico’s low productivity, using the Hsieh and Klenow (2009) indirect approach and firm-level Mexican Economic Census data.
- Data scope: universe of urban formal and informal firms with fixed establishments (Mexican Economic Census, 2013).
- Key headline: fully eliminating distortions that give rise to resource misallocation would increase aggregate TFP by almost 125 percent.

### II. Conceptual framework
- Framework: Hsieh and Klenow (2009) applied at the state level with industry j in state s having N_js monopolistically competitive firms; industry aggregation via Cobb-Douglas and CES structures (Equations (1)–(11)).
- Distortions: firm-specific wedges τ^y_ijs (output wedge) and τ^k_ijs (capital wedge) that affect first-order conditions (Equations (4)–(7)).
- Revenue productivity: TFPR_ijs = p_ijs A_ijs; dispersion of TFPR and MRPL/MRPK indicates misallocation.
- TFP gain measure: TFPgain_s = 100 × (TFP_s / TFP*_s − 1) (Equation (11)), where TFP*_s is TFP under efficient allocation.

### III. Data
- Source: 2013 Mexican Economic Census (INEGI), ~3.5 million observations; final sample ~3 million establishments and 3,139 industry-state pairs after exclusions.
- Sector coverage: manufacturing, retail and wholesale trade, transportation and warehousing, accommodation and food services, information and other services.
- Exclusions: financial services, construction, utilities, real estate, professional/technical services, management of shell companies, health, education, arts and culture.
- Inputs and parameters:
  - Production function inputs: firm-level employment (baseline) and capital and sales from data.
  - Output elasticities approximated by U.S. cost shares.
  - Rental price of capital set to 0.1 (assumes real interest and depreciation rates of 5 percent).
  - Elasticity of substitution σ = 3.
- Robustness alternative: wage bill used instead of employment in robustness check (reduces sample).

### IV. Stylized facts — Aggregate and State-level TFP gains
- Aggregate TFP gains:
  - Baseline full-elimination estimate: almost 125 percent aggregate TFP gain.
  - Robustness checks:
    - Use firm-level wage bill → estimate 116 percent.
    - Expand trimmed tails to 2.5 percent → estimate remains around 125 percent.
  - Comparisons:
    - Busso et al. (2012) estimate ≈ 200 percent.
    - Mayorga Garrido-Cortes (2017) ≈ 180 percent.
  - Benchmarks:
    - Reducing distortions to Sweden-level (90th percentile benchmark) → TFP gains almost 70 percent.
    - Bringing distortions down to level of Mexico’s best-performing state (Nuevo León) → TFP gains about 25 percent.
- State-level TFP gains (range and summary statistics):
  - Range across states: around 80 to 190 percent.
  - Interquartile range (uncorrected) = 73.5 percentage points.
  - Table 1 summary statistics (state-level TFP gains, in percent):
    - Aggregate, uncorrected: Min 78.1; 10th perc. 89.3; Med 116.5; Mean 123.3; 90th perc. 162.8; Max 192.2; SD 29.0; IQR 73.5.
    - Aggregate, corrected for Industry FE: Min 95.7; 10th perc. 103.6; Med 126.1; Mean 127.1; 90th perc. 148.7; Max 159.6; SD 16.7; IQR 45.1.
  - Correlation with incomes:
    - After omitting Mexico City, correlation coefficient between state-level resource misallocation and GDP per capita = -0.84.
    - Using TFP gains corrected for industry effects, correlation = -0.83.

### V. Econometric results
- Estimation setup:
  - Unit: industry-state level TFP gain (dependent variable).
  - Sample: baseline excludes 10th and 90th percentile of TFP gains → 2,443 industry-state observations across 32 states.
  - Regressors: informality, crime, access to finance, internet use, distance to population centers (constructed from firm/municipal/state data and other sources).
  - Specification: industry and state fixed effects; regressors entered as quartile dummies (first/second/third quartiles with fourth quartile as omitted category).
  - Estimation: OLS with heteroscedasticity-consistent standard errors (robustness to clustering reported).
- Baseline findings (magnitudes and signs):
  - Informality:
    - Moving from 4th quartile to 3rd quartile → TFP gain falls by 11 percentage points.
    - Further moves to 2nd and 1st quartiles reduce misallocation by an additional 3 and 5 percentage points, respectively.
    - Interpretation: higher informality associated with higher resource misallocation.
  - No Financial Access (share without bank accounts):
    - Moving from 4th to 1st quartile associated with total reduction in TFP gain ≈ 18 percentage points.
  - Distance to population centers:
    - Moving from 4th to 1st quartile associated with total reduction in TFP gain ≈ 12 percentage points.
  - Crime and No Internet Use:
    - Both matter, but impacts are non-linear; largest reductions occur when moving to the first quartile (Crime total reduction ≈ 14 percentage points; Internet total reduction ≈ 9 percentage points).
- Median-state reform experiment:
  - Simulation: lower each distortion in each industry-state pair to the first quartile (domestic frontier).
  - Median Mexican state would see TFP rise by some 13 percent under this reform scenario.
  - Example state simulations (Figure 3):
    - Oaxaca (max state): reform scenario TFP gain would exceed 20 percentage points.
    - San Luis Potosí (median state): intermediate gain (shown in Figure 3).
    - Nuevo León (min state): gains ≈ 5 percentage points.
  - Excluding Distance from the simulation → Oaxaca impact still around 15 percent of GDP.
- Robustness checks (high-level):
  - Alternative informality definition → informality dummies remain highly significant and coefficients similar.
  - Use homicide incidence instead of robberies → similar results.
  - Alternative financial access measure (absence of bank credit) → weaker/significant only for “very low” category.
  - Replace No Internet Use with share of employees not using computers → similar results.
  - Weighting by sales, clustering errors by state and industry, using continuous regressors, and including additional controls → main results qualitatively robust.
- Market concentration extension:
  - Herfindahl index computed by employment and by sales.
  - Unconditional coefficient on Herfindahl (by employment) negative and significant (higher concentration → lower misallocation).
  - Interaction of Herfindahl and informality:
    - Concentration has positive effect on misallocation in more formal industries (consistent with market power), but effect switches sign (becomes negative) in industries with high informality.
  - Results robust to alternative concentration measures and to replacing concentration with average firm size.
- Corruption extension:
  - Corruption indicators from ENCIG 2013 (varying only across states).
  - Identification via difference-in-differences: interaction between state-level corruption measures and industry dummy for industries with government procurement in the upper quartile.
  - Interaction coefficients:
    - Interaction Procurement/Corruption Experience = 1.050**.
    - Interaction Procurement/Corruption Very Frequent = 0.295**.
    - Interaction Procurement/Corruption Heard About = 0.347*.
  - Interpretation: higher corruption raises resource misallocation especially in industries with large government procurement exposure.

### VI. Conclusion and policy implications
- Main conclusions:
  - Resource misallocation is a quantitatively important contributor to low productivity in Mexico; fully eliminating misallocation implies very large TFP gains (almost 125 percent).
  - Large subnational heterogeneity: state-level TFP gains range from around 80 to 190 percent; variation correlates closely with state GDP per capita (correlation ≈ -0.84).
  - Observable distortions explaining misallocation: informality, crime, corruption (especially in government-procurement-exposed industries), market concentration (conditional on informality), limited financial and internet access, and geographic isolation.
- Policy recommendations:
  - Strengthen competition policy and reduce market-power distortions.
  - Enhance access to financial services (bank accounts) and telecommunications (internet/computer use).
  - Strengthen the rule of law to reduce corruption and crime.
  - Reduce labor informality through targeted reforms.
  - Address geographic isolation and factor mobility limitations via targeted physical or transportation infrastructure investments.
- Reform significance:
  - Moving distortions to the domestic frontier (first quartile) yields economically meaningful TFP gains at the state level (median state ≈ 13 percent).
  - Reform effects differ across states, implying reforms would also reduce spatial disparities in misallocation and productivity.

### Annexes — Variable definitions and summary statistics (selected)
- Baseline Regressors (definitions):
  - Informality: Share of firms paying no social security and no VAT. — 2013 Economic Census.
  - Crime: Share of firms in high crime municipalities (robberies per capita in upper quartile). — 2010 SIMBAD; 2010 Population Census.
  - No Bank Account: Avg. across firms of dummy that takes value 1 when firm has a bank account. — 2013 Economic Census.
  - No Internet Use: Average share of employees not using the internet for their work. — 2013 Economic Census.
  - Distance: Average distance between the locality of the firm and the closest population center (population > 500,000). — 2013 Economic Census and own computation.
- Regressors in Extensions (definitions):
  - Herfindahl by Employment and by Sales — 2013 Economic Census.
  - Procurement: Share of firms reporting the government as their most important client. — 2013 Economic Census.
  - Corruption Experience / Very Frequent / Heard About / Top 3 — Encuesta Nacional de Calidad e Impacto Gubernamental 2013.
- Selected summary statistics (preserve values exactly):
  - Informality: Mean 0.634; Minimum 0; Maximum 1.
  - Crime: Mean 0.327; Minimum 0; Maximum 1.
  - No Bank Account: Mean 0.622; Minimum 0; Maximum 1.
  - No Internet Use: Mean 0.967; Minimum 0; Maximum 1.
  - Distance (in meters): Mean 85879.9; Minimum 114; Maximum 370799.4.
  - Herfindahl by Employment: Mean 0.057; Minimum 0.000; Maximum 0.859.
  - Herfindahl by Sales: Mean 0.075; Minimum 0.000; Maximum 0.717.
  - Firm Size: Mean 11.950; Minimum 1.244; Maximum 1000.164.
  - Procurement: Mean 0.005; Minimum 000.364; Maximum 1.
  - Corruption Experience: Mean 9.345; Minimum 4.484; Maximum 16.596.
  - Corruption Very Frequent: Mean 44.206; Minimum 18.833; Maximum 62.164.
  - Corruption Heard About: Mean 0.309; Minimum 0.173; Maximum 0.501.
  - Corruption Top 3: Mean 460.34; Minimum 33.987; Maximum 54.606.
  - High Population: Mean 0.251; Minimum 0; Maximum 1.
  - Firms per Capita: Mean 0.043; Minimum 0.023; Maximum 0.130.
  - Number of Firms: Mean 880.9; Minimum 10; Maximum 127602.
  - FDI: Mean 0.072; Minimum 0; Maximum 1.
  - GDP per Capita: Mean 72.067; Minimum 3.720; Maximum 1500.0031.

*Source: wp18112 - 1. State-level TFP Gains from Eliminating Resource Misallocation (in %) — wp18112.pdf*

### 1. State-level TFP Gains from Eliminating Resource Misallocation (in %) ................................................

### State-level TFP Gains from Eliminating Resource Misallocation (in %)

### Main sections
- 1. State-level TFP Gains from Eliminating Resource Misallocation (in %) ................................................ 12
- 2. Baseline Regressions ................................................................................................................................................ 23
- 3. Robustness Checks .................................................................................................................................................... 24
- 4. Market Concentration Regressions .................................................................................................................... 25
- 5. Corruption Regressions ........................................................................................................................................... 26

### Figures listed
- 1. TFP Gains from Efficient Allocation of Resources ......................................................................................... 10
- 2. Correlation between State-level per Capita Incomes and Resource  
    Misallocation, 2013 ................................................................................................................................................... 13
- 3. TFP Gains in Reform Scenario Across States ................................................................................................... 17

### Annexes
- 1. Resource Misallocation by Mexican State ........................................................................................................ 29
- 2. Variable Definitions and Sources ......................................................................................................................... 30

*Source: wp18112 - 1. State-level TFP Gains from Eliminating Resource Misallocation (in %) — wp18112.pdf*

### 3. Summary Statistics ..................................................................................................

### 3. Summary Statistics

### I. Introduction
- Objective: provide evidence that resource misallocation rather than access to technology could lie at the heart of Mexico’s low productivity, using the Hsieh and Klenow (2009) indirect approach and firm-level Mexican Economic Census data.
- Data scope: universe of urban formal and informal firms with fixed establishments (Mexican Economic Census, 2013).
- Key headline: fully eliminating distortions that give rise to resource misallocation would increase aggregate TFP by almost 125 percent.

### II. Conceptual framework
- Framework: Hsieh and Klenow (2009) applied at the state level with industry j in state s having N_js monopolistically competitive firms; industry aggregation via Cobb-Douglas and CES structures (Equations (1)–(11)).
- Distortions modeled as firm-specific wedges: τ^y_ijs (output wedge) and τ^k_ijs (capital wedge) that affect first-order conditions (Equations (4)–(7)).
- Revenue productivity defined as TFPR_ijs = p_ijs A_ijs; dispersion of TFPR and MRPL/MRPK indicates misallocation.
- TFP gain measure: TFPgain_s = 100 × (TFP_s / TFP*_s − 1) (Equation (11)), where TFP*_s is TFP under efficient allocation.

### III. Data
- Source: 2013 Mexican Economic Census (INEGI), ~3.5 million observations; final sample ~3 million establishments and 3,139 industry-state pairs after exclusions.
- Sector coverage: manufacturing, retail and wholesale trade, transportation and warehousing, accommodation and food services, information and other services. Excluded: financial services, construction, utilities, real estate, professional/technical services, management of shell companies, health, education, arts and culture.
- Production function inputs: firm-level employment (baseline) and capital and sales from data. Output elasticities approximated by U.S. cost shares; rental price of capital set to 0.1 (assumes real interest and depreciation rates of 5 percent); elasticity of substitution σ = 3.
- Robustness alternative: wage bill used instead of employment in robustness check (reduces sample).

### IV. Stylized facts
- Aggregate TFP gains
  - Baseline full-elimination estimate: almost 125 percent aggregate TFP gain.
  - Robustness checks:
    - Use firm-level wage bill → estimate 116 percent.
    - Expand trimmed tails to 2.5 percent → estimate remains around 125 percent.
  - Comparisons:
    - Busso et al. (2012) estimate ≈ 200 percent (differences due to years, industries, data cleaning, imputations).
    - Mayorga Garrido-Cortes (2017) ≈ 180 percent (uses imputations).
  - Benchmarks:
    - Reducing distortions to Sweden-level (90th percentile benchmark) → TFP gains almost 70 percent.
    - Bringing distortions down to level of Mexico’s best-performing state (Nuevo León) → TFP gains about 25 percent.
- State-level TFP gains
  - Range across states: around 80 to 190 percent.
  - Interquartile range (uncorrected) = 73.5 percentage points.
  - Table 1 summary statistics (state-level TFP gains, in percent):
    - Aggregate, uncorrected: Min 78.1; 10th perc. 89.3; Med 116.5; Mean 123.3; 90th perc. 162.8; Max 192.2; SD 29.0; IQR 73.5.
    - Aggregate, corrected for Industry FE: Min 95.7; 10th perc. 103.6; Med 126.1; Mean 127.1; 90th perc. 148.7; Max 159.6; SD 16.7; IQR 45.1.
  - Correlation with incomes:
    - After omitting Mexico City (outlier), correlation coefficient between state-level resource misallocation and GDP per capita = -0.84.
    - Using TFP gains corrected for industry effects, correlation = -0.83.

### V. Econometric results
- Estimation setup
  - Unit: industry-state level TFP gain (dependent variable).
  - Sample: baseline excludes 10th and 90th percentile of TFP gains → 2,443 industry-state observations across 32 states.
  - Baseline regressors (all constructed from firm/municipal/state data and other sources): informality, crime, access to finance, internet use, distance to population centers.
  - Specification: industry and state fixed effects; regressors entered as quartile dummies (first/second/third quartiles with fourth quartile as omitted category).
  - Estimation: OLS with heteroscedasticity-consistent standard errors (robustness to clustering reported).
- Baseline findings (summary of magnitudes and signs)
  - Informality:
    - Moving from 4th quartile (most severe) to 3rd quartile → TFP gain falls by 11 percentage points.
    - Further moves to 2nd and 1st quartiles reduce misallocation by an additional 3 and 5 percentage points, respectively (total reductions reported in text).
    - Interpretation: higher informality associated with higher resource misallocation.
  - No Financial Access (share without bank accounts):
    - Moving from 4th to 1st quartile associated with total reduction in TFP gain ≈ 18 percentage points.
  - Distance to population centers:
    - Moving from 4th to 1st quartile associated with total reduction in TFP gain ≈ 12 percentage points.
  - Crime and No Internet Use:
    - Both matter, but impacts are non-linear; largest reductions occur when moving to the first quartile (Crime total reduction ≈ 14 percentage points; Internet total reduction ≈ 9 percentage points).
  - Median-state reform experiment:
    - Baseline regression used to simulate hypothetical reform: lowering each distortion in each industry-state pair to the first quartile (domestic frontier).
    - Median Mexican state would see TFP rise by some 13 percent under this reform scenario.
    - Example state simulations (Figure 3):
      - Oaxaca (max state): reform scenario TFP gain would exceed 20 percentage points.
      - San Luis Potosí (median state): shown in Figure 3 (intermediate gain).
      - Nuevo León (min state): gains ≈ 5 percentage points.
    - Excluding Distance from the simulation → Oaxaca impact still around 15 percent of GDP.
- Robustness checks (high-level)
  - Alternative informality definition (narrower) → informality dummies remain highly significant and coefficients similar.
  - Use homicide incidence instead of robberies → similar results.
  - Alternative financial access measure (absence of bank credit) → weaker/significant only for “very low” category; suggests bank account access matters more than credit access for allocation efficiency.
  - Replace No Internet Use with share of employees not using computers → similar results.
  - Weighting by sales, clustering errors by state and industry, using continuous regressors, and including additional controls (population density, firms per capita, number of firms, FDI share, municipal GDP per capita) → main results qualitatively robust.
- Market concentration extension
  - Herfindahl index computed by employment and by sales.
  - Unconditional coefficient on Herfindahl (by employment) negative and significant (higher concentration → lower misallocation).
  - Hypothesis: concentration may reflect productive firm dominance rather than market power in many Mexican industries where many small informal firms exist.
  - Interaction results:
    - Interaction of Herfindahl and informality: concentration has positive effect on misallocation in more formal industries (consistent with market power), but effect switches sign (becomes negative) in industries with high informality.
    - Results robust to alternative concentration measures and to replacing concentration with average firm size (with similar interaction patterns).
- Corruption extension
  - Corruption indicators from ENCIG 2013 (perception and experience measures), varying only across states.
  - Identification via difference-in-differences: interaction between state-level corruption measures and industry dummy for industries with government procurement in the upper quartile.
  - Interaction coefficients:
    - Interaction Procurement/Corruption Experience = 1.050** (positive and significant).
    - Interaction Procurement/Corruption Very Frequent = 0.295**.
    - Interaction Procurement/Corruption Heard About = 0.347*.
    - Interpretation: higher corruption raises resource misallocation especially in industries with large government procurement exposure.

### VI. Conclusion and policy implications
- Main conclusions:
  - Resource misallocation is a quantitatively important contributor to low productivity in Mexico; fully eliminating misallocation implies very large TFP gains (almost 125 percent).
  - There is large subnational heterogeneity: state-level TFP gains range from around 80 to 190 percent, and variation correlates closely with state GDP per capita (correlation ≈ -0.84).
  - Observable distortions explaining misallocation (across industries and states) include informality, crime, corruption (especially in government-procurement-exposed industries), market concentration (conditional on informality), limited financial and internet access, and geographic isolation.
- Policy recommendations (derived from empirical findings):
  - Strengthen competition policy and reduce market-power distortions.
  - Enhance access to financial services (bank accounts) and telecommunications (internet/computer use).
  - Strengthen the rule of law to reduce corruption and crime.
  - Reduce labor informality through targeted reforms.
  - Address geographic isolation and factor mobility limitations via targeted physical or transportation infrastructure investments.
- Reform significance:
  - Moving distortions to the domestic frontier (first quartile) yields economically meaningful TFP gains at the state level (median state ≈ 13 percent).
  - Reform effects differ across states, implying reforms would also reduce spatial disparities in misallocation and productivity.

*Source: 3. Summary Statistics, wp18112 (2013 Mexican Economic Census analyses) — IMF Working Paper content provided in the source PDF.*

### References

### wp18112 - References

### References
- Busso, M., Fazio, M. and Levy, S., 2012, “(In)Formal and (Un)Productive: The Productivity Costs of Excessive Informality in Mexico,” IDB Working Paper 341, Inter-American Development Bank.
- Busso, M., and Madrigal, L., 2013, “Productivity and resource misallocation in Latin America,” The BE Journal of Macroeconomics, Vol.13(1), 903–32.
- Calligaris, S., Del Gatto, M., Hassan, F., Ottaviano, G. and Schivardi, F., 2016, “Italy’s Productivity Conundrum. A Study on Resource Misallocation in Italy,” European Economy - Discussion Papers 030, European Commission.
- Dias, D., Richmond, C. and Robalo Marques, C., 2016, “A Tale of Two Sectors; Why is Misallocation Higher in Services than in Manufacturing?,” IMF Working Paper 16/220 (International Monetary Fund, Washington, D.C.).
- Haltiwanger, J., Kulick, R and Syverson, C., 2018, “Misallocation Measures: The Distortion That Ate the Residual,” NBER Working Papers 24199.
- Hanson, G., 2010, “Why Isn’t Mexico Rich?”, Journal of Economic Literature, Vol. 48(4), 987–1004.
- Hopenhayn, H., 2014, “Firms, Misallocation, and Aggregate Productivity: A Review,” Annual Review of Economics, Vol. 6(1), 735–70.
- Hsieh, C. T., and Klenow, P. J., 2009, “Misallocation and manufacturing TFP in China and India,” The Quarterly Journal of Economics, 124(4), 1403–48.
- IMF, 2017. “IMF Fiscal Monitor: Achieving More with Less,” International Monetary Fund, Washington, D.C.
- Inklaar, R., Lashitew, A. and Timmer, M., 2017, “The Role Of Resource Misallocation In Cross-Country Differences In Manufacturing Productivity,” Macroeconomic Dynamics, Vol. 21(03), 733–56.
- Keller, W., 2004, “International Technology Diffusion,” Journal of Economic Literature, Vol. 42(3), 752–82.
- Levy, S. and Rodrik, D. 2017, “The Mexican Paradox,” available at http://www.project-syndicate.org.
- Mayorga Garrido-Cortés, J. 2017, “Resource misallocation in Mexico,” Tesis de maestría, Centro de Investigación y Docencia Económicas, available at http://repositorio-digital.cide.edu.
- Rajan, R. Zingales. L., 1998, “Financial Dependence and Growth.” The American Economic Review, vol. 88(3), 559-586.
- Restuccia, D., and Rogerson, R., 2008, “Policy Distortions and Aggregate Productivity with Heterogeneous Plants,” Review of Economic Dynamics, Vol. 11(4), 707–20.
- Restuccia, D., and Rogerson, R., 2013, “Misallocation and productivity,” Review of Economic Dynamics, Vol. 16(1), 1–10.
- Restuccia, D., and Rogerson, R., 2017, “The Causes and Costs of Misallocation,” Journal of Economic Perspectives, Vol. 31(3), 151–74.
- Saborowski, C., 2017, “Mexico’s Structural Reform Agenda: Early Signs of Success,” Selected Issues Paper, 2017 IMF Art. IV Consultation for Mexico.

### Annex 1. Resource Misallocation by Mexican State
- Notes: The map shows the level of resource misallocation (expressed in quartiles) by state which corresponds to the TFP gains if resources were allocated optimally across firms within all sectors under consideration within each state.
- Sources: 2013 Mexican Economic Census and own compilation.

### Annex 2. Variable Definitions and Sources
- Baseline Regressors
  - Informality: Share of firms paying no social security and no VAT. — 2013 Economic Census
  - alternative proxy: Share of firms paying no social security, VAT, income or excise tax.
  - Crime: Share of firms in high crime municipalities (in which robberies per capita is in the upper quartile of the distribution). — 2010 SIMBAD; 2010 Population Census
  - alternative proxy: Share of firms in high crime municipalities (in which homicides per capita is in the upper quartile of the distribution).
  - No Bank Account: Avg. across firms of dummy that takes value 1 when firm has a bank account. — 2013 Economic Census
  - alternative proxy: Avg. across firms of dummy that takes value 1 when firm has bank credit.
  - No Internet Use: Average share of employees not using the internet for their work. — 2013 Economic Census
  - alternative proxy: Average share of employees not using computers for their work.
  - Distance: Average distance between the locality (e.g., city / town) of the firm and the closest population center (population > 500,000). — 2013 Economic Census and own computation
- Regressors in Extensions
  - Herfindahl by Employment: Herfindahl index calculated based on each firm's number of employees. — 2013 Economic Census
  - Herfindahl by Sales: Herfindahl index calculated based on each firm's total sales. — 2013 Economic Census
  - Firm Size: Average sales by firm. — 2013 Economic Census
  - Procurement: Share of firms reporting the government as their most important client. — 2013 Economic Census
  - Corruption Experience: Proportion of respondents who say that they experienced corruption in dealing with the government. — Encuesta Nacional de Calidad e Impacto Gubernamental 2013
  - Corruption Very Frequent: Proportion of respondents who answer that corruption is very frequent. — Encuesta Nacional de Calidad e Impacto Gubernamental 2013
  - Corruption Heard About: Proportion of respondents who heard from relatives/friends that there are people who had to pay bribes. — Encuesta Nacional de Calidad e Impacto Gubernamental 2013
  - Corruption Top 3: Proportions of respondents who consider corruption one of the top three issues in the state government. — Encuesta Nacional de Calidad e Impacto Gubernamental 2013
- Other Regressors
  - High Population: Share of firms in municipalities within 4th quartile of population density. — 2010 SIMBAD
  - Firms per Capita: Number of firms per capita. — 2013 Economic Census; 2010 Population Census
  - Number of Firms: Number of firms. — 2013 Economic Census
  - FDI: Share of firms engaged in FDI relationships. — 2013 Economic Census
  - GDP per Capita: Average GDP per capita across firms of municipalities the firms are located in. — INEGI’s National Account Statistics

### Annex 3. Summary Statistics
- Baseline Regressors
  - Informality: Mean 0.634; Minimum 0; Maximum 1
  - alternative proxy: Mean 0.628; Minimum 0; Maximum 1
  - Crime: Mean 0.327; Minimum 0; Maximum 1
  - alternative proxy: Mean 0.334; Minimum 0; Maximum 1
  - No Bank Account: Mean 0.622; Minimum 0; Maximum 1
  - alternative proxy: Mean 0.815; Minimum 0; Maximum 1
  - No Internet Use: Mean 0.967; Minimum 0; Maximum 1
  - alternative proxy: Mean 0.942; Minimum 0; Maximum 1
  - Distance (in meters): Mean 85879.9; Minimum 114; Maximum 370799.4
- Regressors in Extensions
  - Herfindahl by Employment: Mean 0.057; Minimum 0.000; Maximum 0.859
  - Herfindahl by Sales: Mean 0.075; Minimum 0.000; Maximum 0.717
  - Firm Size: Mean 11.950; Minimum 1.244; Maximum 1000.164
  - Procurement: Mean 0.005; Minimum 000.364; Maximum 1
  - Corruption Experience: Mean 9.345; Minimum 4.484; Maximum 16.596
  - Corruption Very Frequent: Mean 44.206; Minimum 18.833; Maximum 62.164
  - Corruption Heard About: Mean 0.309; Minimum 0.173; Maximum 0.501
  - Corruption Top 3: Mean 460.34; Minimum 33.987; Maximum 54.606
- Other Regressors
  - High Population: Mean 0.251; Minimum 0; Maximum 1
  - Firms per Capita: Mean 0.043; Minimum 0.023; Maximum 0.130
  - Number of Firms: Mean 880.9; Minimum 10; Maximum 127602
  - FDI: Mean 0.072; Minimum 0; Maximum 1
  - GDP per Capita: Mean 72.067; Minimum 3.720; Maximum 1500.0031

*Content derived from wp18112 - References (IMF).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18112.pdf_