Diffusion of crime control benefits: forced eradication and coca crops in Colombia

ABSTRACT

One explanation for the increasing number of hectares of coca is that eradication strategies displace coca crops but fail to completely clear affected areas. In the drug policy literature, that dynamic shifting is commonly known as the balloon effect. This study integrates georeferenced agricultural data through spatially explicit econometric models to test the hypothesis that forced eradication displaces coca crops. Using annual data for 1116 contiguous municipalities in Colombia, we estimate a spatial Durbin error model. Our results suggest that, on average, aerial fumigation in a municipality diffuses the benefits of this crime control strategy to neighbouring municipalities.

KEYWORDS:

• coca crops
• eradication
• drug policy
• spatial dependence
• spatial Durbin error model
• spillover

JEL:

• K42
• R12
• R14

ACKNOWLEDGEMENTS

We thank SIMCI at the United Nations Office of Drugs and Crime in Colombia for making data available for this study early on. We thank Liliana M. Dávalos for insightful discussions on illicit crops and feedback on multiple drafts of this paper. We also thank the editorial board and two anonymous referees for their constructive comments and suggestions. All remaining errors are our own.

DATA AVAILABILITY

The data that support the findings of this study are openly available at https://github.com/eleodavalos/balloon_effect.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors. The opinions contained in this document are the sole responsibility of the authors and do not commit Banco de la República or its Board of Directors.

Notes

1 For model specification and spatial weights matrices selection, we use traditional and modern diagnostics tools based on Lagrange multiplier tests, Wald tests and a Bayesian comparison approach.

2 As of 2014, 65% of agricultural units in Colombia used family labour in their production process and 73% had fewer than 5 ha (Torres Franco et al., 2021).

3 Where the mark-up of the licit crop is denoted by $M_L=P_L-C_L$.

4 This is explained by the continuity of the π function assumption and the fact the only source of farmers' heterogeneity in the model is location and its characteristics.

5 ${\displaystyle \frac{\mathrm{\partial }{I}_j^{\ast }}{\mathrm{\partial }{\pi }_j}=A\alpha {\left[{\left({\displaystyle \frac{(1-{\pi }_j)P_i-C_i}{M_L}}\right)}^{{\displaystyle \frac{1}{\alpha -1}}}+1\right]}^{-\alpha -1}\cdot \left[{\displaystyle \frac{1}{\alpha -1}}\right]\cdot \left[{\left({\displaystyle \frac{(1-{\pi }_j)P_i-C_i}{M_L}}\right)}^{{\displaystyle \frac{2-\alpha }{\alpha -1}}}{\displaystyle \frac{P_i}{M_L}}\right]}$.

The expressions inside the first and third brackets are positive, but since we assume decreasing marginal returns to labour ($0<\alpha <1$) the second bracket is negative, and then the whole expression is negative.

6 See note 5.

7 Contiguity means that two municipalities share a common border of non-zero length. Queen criterion of contiguity defines neighbours as municipalities sharing a common edge or common vertex, and first-order neighbours refers to neighbouring municipalities adjacent to the municipality analysed.

8 For more details about the methodology, see the Ajustes y estimaciones área sembrada section in UNODC-SIMCI (2016).

9 1 ha = 2.54 acres.

10 Because we estimate a model of fixed effects, the $x_{jt}$ vector of covariates cannot contain variables fixed in time. A poverty index is a variable important in the determination of illicit crops variation. Unfortunately, in Colombia there is no measure of poverty at the municipality level that varies through time. Therefore, we included some time-varying proxy variables describing the level of wealthiness and development of a municipality.

11 As 1 ha = 10,000 m², a reduction of 0.14 ha means a reduction of 1,400 m², or 14% of 1 ha.

12 The Colombian police reported 855 felonies related with coca cultivation in 2009 (Espino-Duque, 2010), but only 162 people were prosecuted (Tiempo, 2009). However, for the same period, the UNODC estimated that 56,910 families were involved in coca cultivation (UNODC-SIMCI, 2010).