Agricultural and Forest Meteorology 201S (2015) 693
Contents lists available at ScienceDirect
Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet
Peer Review Report
Peer review report 2 On “Extending and improving regionalized winter wheat and silage maize yield regression models for Germany: enhancing the predictive skill by panel definition through cluster analysis”
1. Original Submission 1.1. Recommendation Minor Revision 2. Comments to Author: The authors present a method to improve the predictive skills of yield regression models. They present separate time series models (STSM) for county yield of winter wheat and silage maize and extend them to regional cluster based panel data models (PDM). The paper is clearly structured and well written and the results are presented illustratively. I find the paper very interesting and and well elaborated. In my opinion the paper provides a relevant contribution to regional climate-yield modelling research and is suitable to be published. However, the authors might consider to answer/address/consider some minor comments/questions I have, in order to avoid the questions as they appeared to me. (1) The sense of considering the economic variables (line 27 ff): To cluster only parameter from the same drivers (e.g. weather drivers: evapotranspiration, rainfall, temperature) seems plausible to me, since the parameter value should in regional clusters. Therefore, it is as well plausible to exclude parameters, which are not necessarily similar for the same clusters (e.g. economic variables, depending on soil quality, like fertilizer demand). Therefore, I am wondering why in the starting point authors consider at all the the economic variables, if their objective is a regional clustering. (2) Selection of the economic variables (line 191-192): The multi variate regressions consider as economic variables fertilizer price and acreages, which are available only at the national level. 2.1) To my knowledge for the acreages there are data available at county level. However, I am wondering in which sense the acreage represents an economic parameter. It might be useful to consider the relative acreages as indicator to represent the relative competitiveness in the cluster. The absolut acreage might not be a useful measure, since it varies depending on the size of the county. 2.2) I am wondering how suitable the fertilizer price is a suitable economic variable in Germany. Several reasons would let me expect the fertilizer price not be a relevant variable to explain yields.
DOI of published article: http://dx.doi.org/10.1016/j.agrformet.2015.10.003. 0168-1923/$ – see front matter http://dx.doi.org/10.1016/j.agrformet.2015.12.018
(a) In Germany farmers should apply purchased minearal fertilizer according to the optimal specific intensity. An increase of yields should not be expected, with changes in fertilizer quantity due to fertilizer prices. (b) The price response to changes in fertilizer is supposed to be quite inelastic. Thus, only extreme price increases might have an impact on the quantity of applied fertilizer. (c) Farmers are allowed to apply fertilzer a with security net. Thus, they apply more fertilizer than necessary. If fertilizer prices increase they probably would reduce first this safety net, without impacting the yield. (d) Furthermore, the regional impact of fertilizer price depends on the regional availability of manure, which is e.g. high in NorthWest and small in the NorthEast of Germany. Thus, the NorthEast might react more to fertilizer price changes than the NorthWest. Thus, to consider the fertilizer price as an independing parameter is difficult. 2.3) If economic variables are considered in the model, I am wondering why the producer price (for, wheat, energy biomass, or beef or milk) is not considered. The aspects 2.1 to 2.3, need in my opinion more clarification. Maybe it is a possibility to delete the aspect of considering and dropping economic variables. Then a lots of question would not appear. (3) Discussion and Conclusion:It seems to me, that the value added for research is described as suitable intermediated step for further work. However, in my opinion already in this stage of development the model can be used to improve climate yield modelling in research. Maybe the authors could give potential applications as examples. Freelance Researcher Thünen-Institute Thünen Institute of Rural Studies 2 Rue Bellot Le Mans, 76600 FRANCE + 33658526467 Martin Henseler Available online 18 December 2015