The Plant Protection Products (PPP) Sector in the European Union: A Special View on Herbicides

Abstract

The policy debates on plant protection products (PPPs) in the European Union (EU) are dominated by the environmental implications of crop protection (in particular, the use of herbicides) and the concentration of the herbicide industry. This article aims at presenting an overview of the patterns herbicide usage over time between and within European countries, and an overview of the industry structure. Potential determinants driving some of these differences are discussed, such as the recent PPP policies adopted by the EU. Results show that herbicides are the most important input used in crop protection, but regional differences are substantial. The concentration of the industry is high, but below levels that would raise concerns by EU regulators. The sector is also highly regulated, which contributes to a high concentration and a consequent decline in innovations. This finding raises the possibility of substituting bans of active ingredients in herbicides with alternative solutions.

Les débats de politique sur les produits de protection des plantes (PPP) dans l’Union Européenne sont dominés par les implications environnementales des protections offerts aux cultives (en particulier, l’utilisation des herbicides), et par la concentration de l’industrie des herbicides. Cet étude vise à présenter une vue d’ensemble de l’utilisation des herbicides dans les pays européens, et un aperçu de la structure du secteur des herbicides. Les déterminants de ces différences sont examinés (par exemple, les politiques récentes de PPP adoptes par l’Union Européenne). Les résultats indiquent que les herbicides sont l’apport le plus important en matière de phytoprotection, mais qu’il y a des différences régionales considérables. La concentration de l’industrie des herbicides est haute, mais en dessus du niveau qui feraient surgir des préoccupations chez les régulateurs UE. Le secteur est aussi très règlementé, ce qui implique une haute concentration et une conséquente diminution des innovations; on se pose donc la question si, à la place d’interdire des ingrédients actifs dans les herbicides, on pourrait trouver des solutions alternatives.

Appendix

Details of the Regression Model

The FADN data allow us also to study changes in PPP cost shares distinguishing for EU-15 Member States, EU-27 Member States, and EU PEMS over time. To do so, we estimate the following regression model:

$$\frac{{{\text{PPP}}\;{\text{costs}}}}{{{\text{Total}}\;{\text{farm}}\;{\text{costs}}}} = \alpha_{0} + \alpha_{1} t + \alpha_{2} t^{2} + \alpha_{3} D_{2004} + \sum\nolimits_{i = 1}^{26} {\gamma_{i} } D_{i} + \varepsilon_{i} ,$$

(1)

where t denotes a time trend, normalized to zero for the first year in each of the two periods (i.e. 1989 = 0 and 2004 = 0, respectively) and D ₂₀₀₄ denotes an indicator variable equal to one for observations starting in 2004 and zero otherwise, to capture the effect of the EU enlargement on the farm cost of PPPs. To capture differences in country-specific utilization of PPPs, we include country-specific dummy variables, D _i, one for each of the i-th EU Member State, with Luxemburg as the excluded one.⁵ The α’s and γ _i are parameters to be estimated and ɛ _i is the error term.

The model was estimated using FADN data at both the National and at the NUTS2-level, and for different samples of countries: EU-27, EU-15, and PEMS. The model at the NUTS2-level has the same structure as specified in Eq. (1), but with NUTS2-specific fixed-effects instead of Member State ones (keeping Luxembourg as a reference region as it is both an EU Member State and a single NUTS2 region). We estimate the regression model using ordinary least squares with the heteroscedasticity-robust standard errors procedure in STATA v.13.

The estimated coefficients and R ² values are reported in Table 1. The estimates in the third and fourth column of the table are obtained with the 1989–2009 FADN data. The results for EU-27 show statistically significant time trend coefficients and an inverted U-shaped trend of the cost shares. The cost shares of PPPs increased (positive linear trend coefficient) for the first twelve years in the data but at declining rates (negative quadratic trend coefficient). Thus, the initial positive upwards trend is followed by a stagnation period and then a decline, reaching rates as low as −0.13 per cent per year. The regression coefficients patterns are similar for the country-level and with NUTS2-level models. The coefficients for the post-enlargement indicator variable are negative and statistically significant in both models, suggesting that farms in the new Member States have, on average, a lower cost share of PPPs. Estimates for the EU-15 show the same inverted U-shaped relationship and statistically significant coefficients, resulting in a market growth of 0.27 for the year 1989 to a decline rate of −0.2 in 2009. For PEMS, none of the estimated trend coefficients is statistically different from zero. The estimated coefficients capturing changes of PPPs’ cost shares in the period 2004–2012 are not statistically different from zero, suggesting that the cost share of PPP on farming cost has not changed significantly in this period. The estimated year-specific trends differ across Member State subsamples: the share of PPP cost in the EU-15 declined slightly, reaching annual changes of −0.15 per cent per year, whereas the cost share increased for PEMS (in particular, after 2006), reaching values as high as 0.6 per cent per year (Table A1).

Table A1 Definition of the Market segments used to derive crop protection agents sale

Calculation of HHI_max

We calculate the HHI as the sum of the squared market shares of all the M firms in an industry as ${\text{HHI}} = \sum\nolimits_{i = 1}^{M} {S_{i}^{2} }$. As the HHI of the PPP sector could not be calculated as the market share of all firms is not known, we calculate a proxy for the HHI, that is, HHI_max as follows. First, note that the HHI can be divided into a portion (HHI _k ) which can be calculated from the shares of the largest K firms (available) and another (HHI _r ) attributable to the remaining firms, r = K + 1, …, M whose individual shares are not observed, or ${\text{HHI}} = {\text{HHI}}_{K} + {\text{HHI}}_{r} = \sum\nolimits_{i = 1}^{K} {S_{i}^{2} } + \sum\nolimits_{K + 1}^{M} {S_{r}^{2} }$. The HHI _max is calculated as

$${\text{HHI}}_{\hbox{max} } = {\text{HHI}}_{K} + \left\lfloor {\frac{X}{{\hbox{min} \{ x_{K} \} }}} \right\rfloor \hbox{min} \{ S_{K}^{2} \} + \frac{{X_{\text{L}} }}{X},$$

(3)

where we assume that the residual revenue is produced by a number of $\left\lfloor {\frac{X}{{\hbox{min} \{ x_{K} \} }}} \right\rfloor$ firms that have the same market share, min{s _k } of the smallest of the K firms, and X _L (“leftover” sales) are $X_{\text{L}} = X - \sum\nolimits_{i = 1}^{K} {x_{i} } - \left\lfloor {\frac{X}{{\hbox{min} \{ x_{K} \} }}} \right\rfloor \hbox{min} \{ x_{K} \}$ is assumed to be produced by a single firm.