Predictive output validation of almond prices

Abstract

The almond market is especially important in Castellón area and a large part of the farmland in the province is used for its production. The new consumer trends and new production techniques have meant an improvement in the sector in recent years, but taking into account that it is a fairly traditional sector there is a resistance to change. There are not specific quantitative studies about Spanish almond market, so this paper aims to analyse the prices of almonds at the national level. To carry out this study, we have used the techniques “in sample” and “out sample” calibration, consisting on the development of several models. In the first test we are going to take the whole data set (in sample) and in the second calibration we will take data out of the sample with "t-1" to "t-5". The object of the study is to analyse two of the most representative varieties of almonds nationwide: Marcona and Comuna. We have created two models for each one: on the one hand OLS and on the other AR (1) using the same explanatory variables for both. The results after both calibrations show that, as a general rule, in this market, AR models have a better adjustment to reality, and in terms of the difference between varieties, in both calibrations the Marcona extracts more robust models.