Hazards of Policy Modelling: India and the world markets for groundnuts and groundnut products

Major advances in modelling techniques and computing power that have occurred over the past 20 years or so now make it relatively easy to quantify the results of policy experiments when large numbers of parameters and variables are involved. The most important and difficult part of this kind of empirical research is the description of a base scenario which is then perturbed by changing one or more policy variables, for example a tax, a subsidy, a tariff, or a non-price constraint of some kind. One of the principal difficulties is deciding on the parameters which are likely to dominate the adjustment of the system and which need to be specified with special care, and on the ot her hand making judgments on parameters which are likely to be less significant and which can be safely ignored or treated in an aggregate way or as residuals. These difficulties can be especially acute in commodity-specific multi- country trade models, since — despite the increasi ng availability of national and international databases — it will seldom be possible to accu rately describe the policies followed and the adjustment mechanisms in each of a large numb er of countries, owing to lack of relevant up- to-date published information of the required sp ecificity, and limits on the time and resources available to the researchers. In this regard, because substitution and complementarity relationships between commodities on both the supply and demand sides can vary substantially across countries — especially as between developed and low income countries — decisions on whether commodities other than the commodity of initial interest should be modelled explicitly, are especially important.