The spatio-temporal dynamics of pesticide use: a global exploratory analysis

Abstract

We utilize exploratory techniques to identify patterns in pesticide use spatial and temporal dimensions. Using data on 149 countries from 1993 to 2020, we find that some countries with low pesticide use are “catching-up” slightly. From a Markov chain analysis, we also note that the distribution of pesticide use per hectare is characterized by a low inter-class mobility, more pronounced in the head and tail of the distribution. This suggests a lock-in phenomenon once “unsustainable” agricultural systems are adopted. We complement this temporal analysis by exploring the role of “proximity” between countries in the use of pesticides and its dynamics. Three definitions of proximity are considered. The first is geographical, the second and third are based on crops produced at the country level and on countries’ GDP per capita respectively. Our results suggest that countries’ economic development explains more pesticide use than their geographic proximity or their crop similarity.

Keywords:

• pesticides use
• spatio-temporal dynamics
• crop proximity
• geographic proximity
• economic proximity

JEL codes:

• Q01
• Q15
• C21
• R11
• O57

Disclosure statement

No potential conflict of interest was reported by the author(s).

Code and data availability

Data used are public and can be accessed at http://www.fao.org/faostat/en/#data. The code for reproducing all of the results in this paper can be found at the following GitHub repository: https://github.com/ayoubakassoum/pesticides_trends.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/09640568.2023.2274284.

Notes

1 The concept of technological lock-in explains why certain technologies are widespread and continue to be used despite the existence of more advanced or socially desirable alternatives. The dominance of one technology over others may result from the economies of scale it provides, or because it reduces uncertainty (Arthur 1989). It may also be due to the fact that the adoption of alternatives requires complex, and time-consuming skills.

2 The low Moran I obtained may be, in part, explained by the construction of this neighborhood structure. Indeed, a country $A$ with a very small area devoted to a highly pesticide-consuming crop (e.g., grapes) may be considered as neighbors of another country $B$ which has a close distribution on all $A$ remaining crops. These two countries, although considered as neighbors in terms of crop distribution, may not end up with similar pesticide use.

3 Regarding the $l\left(1\right)$ transition matrix, we should note that there is only one observed transition of a high use of pesticide per hectare ($Q5$) to a medium use ($Q3$) in our data. There are also only four observed transitions from high-medium ($Q4$) use: two from $Q4$ to $Q3,$ one remaining in $Q4,$ and one from $Q4$ to $Q5.$ Given the infrequency of information (i.e., transitions) associated with high and high-medium levels of pesticide use, it is possible to conclude that the general behavior of countries whose neighbors use low levels of pesticides is to also use low levels.

4 Each year, 149 movements are possible corresponding to the number of countries. As we report all intra-year movements, a total of 4,023 (149$\times$ (2020-1993)) movements are reported.