Accessible Unlicensed Requires Authentication Published by De Gruyter September 1, 2017

Crime and Establishment Size: Evidence from South America

Umut Oguzoglu ORCID logo and Ashantha Ranasinghe

Abstract

Establishment exposure to crime is a frequent occurrence and a severe impediment to business operation in developing economies. We present a simple theory for the frequency and severity of crime across establishment size and evaluate its central predictions using micro-data. We find that high expectation of crime at the establishment level is strongly associated with lower sales, labor and capital. Consistent with our theory, crime has a differential role across size and is less prevalent among the smallest and largest establishments. When evaluated relative to major distortions to production highlighted in developing economies, crime remains important for explaining establishment size.

JEL Classification: O1; O4; D2

Acknowledgements

We are grateful to Pedro Bento, Ryan Compton, Josh Lewis, Wayne Simpson, Trevor Tombe, Julia Witt, a referee and participants at various conferences and seminars for insightful conversations and helpful comments.

Appendix

A Correlations and Regression Results with Continuous Measure for Crime Expectation

Table 7 presents correlations for protection spending and crime expectation on establishment size (sales, employees, annual cost of labor and annual capital expenditure). Spending on protection is positively correlated with establishment size. For crime expectation and size there are no clear patterns. Table 8 reports correlations between crime expectation and whether an establishment is a victim of crime.

Table 7:

Correlations –- protection and crime expectation on establishment size.

SalesEmployeesAnnual cost of LaborAnnual capital Expenditure
Protection spending0.71520.49760.28070.3404
Protection spending (0/1)0.22970.30230.16590.013
Crime expectation (high/low)0.01520.09290.10310.1670
Crime expectation (04)0.05490.06630.10540.1984

  1. Column variables are in logs. Annual Capital Expenditure is restricted to establishments that have positive investment. Protection spending is in logs, and (0/1) is a dummy variable for whether an establishment buys protection. Crime expectation (04) represents an establishment’s response to whether crime, theft and disorder are not an obstacle, a minor, moderate, major and severe obstacle to business operation. An establishment has ‘high’ crime expectation if they report major or severe, and ‘low’ otherwise. N=5610.

Table 8:

Crime expectation and indicator of crime.

Crime expectationCrime expectation
(high/low)(04)
Victim of Crime0.160.25

  1. Crime Expectation is defined as in Table 7. Victim of crime is whether an establishment faced crime in the past year. N=6034.

In Section 2 we report mean descriptive statistics for crime across establishment size based on the definition for size used in the WBES. Table 9 reports these same statistics based on the definition for size used in Section 6.2.

Table 9:

Crime across establishment size: mean statistics.

All establishmentsSmall establishmentsMedium establishmentsLarge establishments
Incidence of Crime31.6%24.3%31.6%43.3%
Crime is a major obstacle33.8%38.2%34.0%25.5%
Crime Losses (as % of sales)3.83%4.28%4.09%1.89%
Bought Security76.7%54.8%78.8%94.9%
Security Expenditure (as % of sales)2.89%3.87%2.96%1.65%
Observations60349354510589

  1. Notes: Small, medium and large establishments are defined as having less than 10, between 10 to 249, and 250 or more employees, in line with the definition used in Section 6.2. Incidence of Crime and Bought Security are averages based on indicators for whether an establishment faced crime and bought private security in the respective year.

Table 10 reports results when crime expectation is a continuous measure taking five possible values, where both OLS and RF estimates are presented. These results are analogous to the ones presented in Table 3 and Table 4 where crime expectation is a dummy variable (high or low).

Table 10:

Continuous measure of crime expectation on establishment size.

OLSReduced Form
Dependent Variable(1)(2)(3)(4)(5)(6)
Sales0.0703***0.0702***0.0462**0.1360.149*0.156**
(0.0257)(0.0244)(0.0217)(0.0933)(0.0851)(0.0725)
Employees0.0420***0.0329**0.01760.172***0.105*0.105**
(0.0137)(0.0134)(0.0110)(0.0593)(0.0548)(0.0449)
Annual Cost of Labor0.0828***0.0696***0.0463**0.249***0.146*0.146*
(0.0253)(0.0251)(0.0233)(0.0908)(0.0886)(0.0758)
Annual Capital Expenditure0.0606***0.04500.01760.208*0.1290.133
(0.0301)(0.0290)(0.0271)(0.107)(0.0985)(0.0888)
Country-level fixed effectsXXXXXX
Industry and city fixed effectsXXXX
Establishment specific controlsXX

  1. Each cell reports point estimates from a separate regression. Crime expectation is a continuous measure taking five possible values: non-obstacle, minor, moderate, major or severe obstacle to business operation. Dependent variables are in logs and Annual Capital Expenditure is restricted to establishments that have positive investment. Number of observations for Sales, Employees, Annual Cost of Labor and Annual Capital Expenditure are 6034, 6034, 5610 and 3827. Robust standard errors, clustered at the country-industry-city level, are in parenthesis. , , denote significance at the 1, 5 and 10 percent level.

B Robustness Checks

To evaluate whether our central findings are robust to alternate specifications we estimate four separate models by OLS and RF with a full set of controls. Table 11 reports the results. In our estimates we have used crime expectation as an independent variable, the rationale being that it accounts for the wider role of crime, including establishments that did not face crime. In our sample about one-third of establishments report facing crime and so it is possible the negative associations we found in Section 6.1 are driven by these establishments. To address this, Column (1) reports estimates for crime expectation on measures related to establishment size when the sample is restricted to establishments that did not face crime. The coefficients are negative and significant implying that even among establishments that are not victims of crime, high crime expectation is associated with lower measures of establishment size.

Table 11:

Robustness checks.

(1)(2)(3)(4)
Dependent variableDid not face crimeComplainersExclude BrazilInclude Mexico
OLS results
Sales0.277***0.315***0.281***0.228***
(0.0745)(0.0555)(0.0539)(0.0511)
Employees0.159***0.152***0.136***0.118***
(0.0386)(0.0306)(0.0366)(0.0280)
Annual cost of labor0.316***0.275***0.270***0.188***
(0.0714)(0.0596)(0.0590)(0.0555)
Annual capital expenditure0.1090.126*0.1070.0340
(0.0877)(0.0709)(0.0747)(0.0664)
RF results
Sales0.660**0.819***0.3970.823***
(0.290)(0.262)(0.254)(0.251)
Employees0.508***0.587***0.2560.538***
(0.189)(0.173)(0.173)(0.166)
Annual cost of labor0.603*0.687**0.1960.644**
(0.307)(0.279)(0.258)(0.265)
Annual capital expenditure0.758**0.653**0.3130.574**
(0.361)(0.291)(0.267)(0.272)

  1. Each cell reports point estimates from a separate regression where all controls are included. Column (1) restricts the sample to establishments that did not face crime in the previous year, column (2) includes controls for whether a manager is a complainer, column (3) excludes observations for Brazil and column (4) adds observations for Mexico. OLS results are based on whether crime expectation is a major or severe obstacle to business operation (indicator variable). RF results is when average crime expectation at the country-industry-city level is used as the independent variable. Dependent variables are in logs and Annual Capital Expenditure is restricted to establishments that have positive investment. Robust standard errors, clustered at the country-industry-city level, are in parenthesis. , , denote significance at the 1, 5 and 10 percent level.

Another concern is that crime expectation is a subjective measure and our estimates may reflect manager biases. It is possible that managers who report high crime expectation may be overly pessimistic or low quality managers such that the role of crime expectation on establishment size reflects weak management more than anything else. For instance, a low quality manager might report high crime expectation to justify poor performance, or alternatively, not fully understand the environment and over or understate the severity of crime. In our regressions we attempt to control for manager quality by including the top manager’s experience in industry. We now attempt to control for whether a manager’s pessimism is driving our results by including a measure for whether a manager is a complainer, as in Gaviria (2002). Specifically, we define a manager as a complainer if their response to whether transportation and political instability are major obstacles to business operation both exceed the averages of what all managers in their country-industry-city report. The rationale is that political instability and transportation are ubiquitous issues, and should be fairly common across establishments within a country-industry-city and not specific to an establishment. Thus, if a manager’s response to both of these questions exceed the country-industry-city average we define them as a ‘complainer’. Column (2) reports our results when we include a dummy variable for whether a manager is a complainer. The results for crime expectation remain negative and significant.

We also re-estimate crime expectation excluding observations for Brazil (Column 3). Brazil accounts for about 25 percent of our sample and close to 70 percent of establishments in Brazil report high crime expectation. Hence, it is possible that our results are driven by Brazilian establishments. We find that all coefficients retain the expected sign when Brazil is excluded, however, due to the drop in the size of RF estimates, only OLS estimates remain statistically significant. This may imply that establishment level performance in Brazil is more closely linked to the crime expectation of counterpart establishments (proxied by our industry-city-country groupings) than other South American establishments in the sample, which affects the magnitude, but not overall direction, of the RF estimates. Lastly, in Column 4 we include observations for Mexico in our regressions given that crime is fairly prevalent in many regions in this country. Estimates of crime expectation on measures related to size remain negative and statistically significant when Mexico is included. Taken together, our results show that crime expectation is negatively associated with sales, number of full-time employees, annual cost of labor and capital expenditure.

C Crime and Size Interaction –- Alternate Definitions for Size

In Section 6.2 we evaluate the role of crime across small, medium and large establishments when we define a medium establishment as one that has between 10 to 249 employees. Table 12, Table 13 and Table 14 consider alternate definitions for a medium size establishment.

Table 12:

Crime expectation and size interaction –- Medium size based on 10–75 employees.

SalesAnnual Cost of Labor
OLSRFOLSRF
Crime expectation0.02470.5490.1090.386
(0.0895)(0.356)(0.0937)(0.328)
Crime expectation × Medium Size0.1611.223***0.252**0.855***
(0.105)(0.389)(0.113)(0.323)
Crime expectation × Large size0.08550.778*0.303**0.503
(0.117)(0.423)(0.125)(0.378)
Country-level fixed effectsXXXX
Industry and city fixed effectsXXXX
Establishment specific controlsXXXX

  1. A medium size establishment is defined as having between 10 to 75 full-time employees. Each column reports estimates from a separate regression. OLS results report when crime expectation is an indicator for whether an establishment reports crime is a major or severe obstacle to business operation. RF results report when average crime expectation at the country-industry-city level is the independent variable. All dependent variables are in logs. Number of observations for Sales and Annual Cost of Labor are 6034 and 5610. Robust standard errors, clustered at the country-industry-city level, are in parenthesis. , , denote significance at the 1, 5 and 10 percent level.

Table 13:

Crime expectation and size interaction –- Medium size based on 10–150 employees.

SalesAnnual cost of labor
OLSRFOLSRF
Crime expectation0.01460.621*0.1200.501
(0.0900)(0.0188)(0.0967)(0.338)
Crime expectation × Medium size0.226**1.228***0.330***0.835**
(0.0991)(0.372)(0.103)(0.326)
Crime expectation × Large size0.1030.970*0.341*0.854*
(0.151)(0.535)(0.174)(0.441)
Country-level fixed effectsXXXX
Industry and city fixed effectsXXXX
Establishment specific controlsXXXX

  1. A medium size establishment is defined as having between 10 to 150 full-time employees. Each column reports estimates from a separate regression. OLS results report when crime expectation is an indicator for whether an establishment reports crime is a major or severe obstacle to business operation. RF results report when average crime expectation at the country-industry-city level is the independent variable. All dependent variables are in logs. Number of observations for Sales and Annual Cost of Labor are 6034 and 5610. Robust standard errors, clustered at the country-industry-city level, are in parenthesis. , , denote significance at the 1, 5 and 10 percent level.

Table 14:

Crime expectation and size interaction –- medium size based on 20–99 employees.

SalesAnnual cost of labor
OLSRFOLSRF
Crime expectation0.02460.3630.001240.385
(0.0705)(0.257)(0.0681)(0.265)
Crime expectation × Medium size0.264**0.1190.1730.595**
(0.112)(0.283)(0.111)(0.269)
Crime expectation × Large size0.02690.2970.1320.322
(0.119)(0.417)(0.127)(0.359)
Country-level fixed effectsXXXX
Industry and city fixed effectsXXXX
Establishment specific controlsXXXX

  1. A medium size establishment is defined as having between 20 to 99 full-time employees (as per the WBES). Each column reports estimates from a separate regression. OLS results report when crime expectation is an indicator for whether an establishment reports crime is a major or severe obstacle to business operation. RF results report when average crime expectation at the country-industry-city level is the independent variable. All dependent variables are in logs. Number of observations for Sales and Annual Cost of Labor are 6034 and 5610. Robust standard errors, clustered at the country-industry-city level, are in parenthesis. , , denote significance at the 1, 5 and 10 percent level.

References

Amin, Mohammad. 2009. Who Suffers More From Crime?. Mimeo.Search in Google Scholar

Aterido, Reyes, Mary Hallward-Driemeier, and Carmen Pages. 2011. “Big Constraints to Small Firms’ Growth? Business Environment and Employment Growth Across Firms.” Economic Development and Cultural Change 59: 609–47.10.1086/658349Search in Google Scholar

Aw, Bee Yan, Mark J. Roberts, and Daniel Yi Xu. 2011. “R & D Investment, Exporting, and Productivity Dynamics.” American Economic Review 101: 1312–44.10.1257/aer.101.4.1312Search in Google Scholar

Banerjee, Ahbijit, and Esther Duflo. 2005. “Growth Theory Through the Lens of Development Economics.” Handbook of Economic Growth 1a: 473–52.Search in Google Scholar

Ben-Yishay, Airel, and Sarah Pearlman. 2014. “Crime and Microenterprise Growth: Evidence from Mexico.” World Development 56: 139–52.10.1016/j.worlddev.2013.10.020Search in Google Scholar

Bloom, Nicholas, Benn Eifert, Aprajit Mahajan, David McKenzie, and John Roberts. 2013. “Does Management Matter? Evidence from India.” The Quarterly Journal of Economics 128: 1–51.10.1093/qje/qjs044Search in Google Scholar

Buera, Francisco, Joseph Kaboski, and Yongseok Shin. 2011. “Finance and Development: A Tale of Two Sectors.” American Economic Review 101: 1964–2002.10.1257/aer.101.5.1964Search in Google Scholar

D’Erasmo, Pablo, and Hernan Moscoso Boedo. 2012. “Financial Structure, Informality and Development.” Journal of Monetary Economics 59: 386–02. Journal of Monetary Economics.Search in Google Scholar

Fajnzylber, Pablo, Daniel Lederman, and Norma Loayza. 1998. “Determinants of Crime Rates in Latin America and the World: An Empirical Assessment”.Search in Google Scholar

Fajnzylber, Pablo, Daniel Lederman, and Norman Loayza. 1998. Determinants of Crime Rates in Latin America and the World: An Empirical Assessment. World Bank Publications World Bank Latin American and Caribbean Studies.Search in Google Scholar

Fisman, Raymond, and Jakob Svensson. 2007. “Are Corruption and Taxation Harmful to Growth? Firm Level Evidence.” Journal of Development Economics 83: 63–75.10.1016/j.jdeveco.2005.09.009Search in Google Scholar

Gaviria, Alejandro. 2002. “Assessing the Effects of Corruption and Crime on Firm Performance: Evidence from Latin America.” Emerging Markets Review 3: 245–68.10.1016/S1566-0141(02)00024-9Search in Google Scholar

Glaeser, Edward, and Bruce Sacerdote. 1999. “Why Is There More Crime in Cities?” Journal of Political Economy 107:. S225–S258.Search in Google Scholar

Guner, Nezih, Gustavo Ventura, and Daniel Yi Xu. 2008. “Macroeconomic Implications of Size Dependent Policies.” Review of Economic Dynamics 11: 721–44.10.1016/j.red.2008.01.005Search in Google Scholar

Hsieh, Chang-Tai, and Peter J. Klenow. 2009. “Misallocation and Manufacturing Productivity in China and India.” Quarterly Journal of Economics 124: 1403–48.10.1162/qjec.2009.124.4.1403Search in Google Scholar

Johnson, Simon, John McMillan, and Christopher Woodruff. 2002. “Property Rights and Finance.” American Economic Review 92: 1335–56.10.1257/000282802762024539Search in Google Scholar

Kalemli-Ozcan, Sebnem, and Bent Sorensen. 2016. “Misallocation, Property Rights, and Access to Finance: Evidence from Within and Across Africa.”. In Africa Successes: Modernization and Development. University of Chicago Press A.Search in Google Scholar

Kisunko, Gregory, Aymo Brunetti, and Beatrice Weder. 1999. “Institutional Obstacles to Doing Business: Region-by-Region Results from a Worldwide Survey of the Private Sector.” The World Bank.Search in Google Scholar

LaPorta, Rafael, and Andrei Shleifer. 2014. “Informality and Development.” Journal of Economic Perspectives 28: 109–26.10.1257/jep.28.3.109Search in Google Scholar

Pearlman, Sarah. 2014. “Officials Versus Thieves: Is Public or Private Expropriation More Harmful to Small Firms?” Journal of Development Studies 50: 773–88.10.1080/00220388.2014.895819Search in Google Scholar

Ranasinghe, Ashantha. 2017. “Property Rights, Extortion and the Misallocation of Talent.” European Economic Review 98: 86–110.10.1016/j.euroecorev.2017.06.010Search in Google Scholar

Restuccia, Diego, and Richard Rogerson. 2008. “Policy Distortions and Aggregate Productivity with Heterogeneous Establishments.” Review of Economic Dynamics 11: 707–720.10.1016/j.red.2008.05.002Search in Google Scholar

Shleifer, Andrei, and Robert. W. Vishny. 1998. The Grabbing Hand: Government Pathologies and Their Cures. Harvard University Press.Search in Google Scholar

Svensson, Jakob. 1998. “Investment, Property Rights and Political Instability: Theory and Evidence.” European Economic Review 42: 1317–41.10.1016/S0014-2921(97)00081-0Search in Google Scholar

Treisman, Daniel. 2000. “The Causes of Corruption: A Cross-National Study.” Journal of Public Economics 76: 399–457.10.1016/S0047-2727(99)00092-4Search in Google Scholar

Published Online: 2017-9-1

© 2017 Walter de Gruyter GmbH, Berlin/Boston