Spline functions: assessing their forecasting consistency with changes in the type of model and choice of joint points


The possibility of the type of spline function and joint points selected affecting the consistency of the ex-post and ex-ante forecasts were tested using cereal production (1961-2006) and percent contribution of agriculture to GDP (1961-2004) in Nigeria. Three types of model, that is, Linear-Quadratic-Linear, Quadratic-Quadratic-Linear and Linear-Quadratic-Quadratic, were used. The result indicated that there is no universality as to which model is appropriate, rather all possible models should be tried and the one that gives most consistent result when compared to observed data and other factors should be used.

10th Annual Conference of the Nigerian Association of Agricultural Economists, 750 Seater Lecture Theatre, University of Abuja Permanent Site (Opposite Nmandi Azikiwe International Airport), Abuja, October 7th–10th
Job Nmadu
Professor of Agricultural Economics and Dean, School of Agriculture and Agricultural Technology

Research interests are economic efficiencies of small scale farming and welfare effects of agricultural interventions.