Kenya Climate Change Interdependence Modelling using Archimedean Copulas
Abstract
Using Kenya historical rainfall and temperature climate data from 1991 to 2016, we conduct an analysis of the interdependence structure for climate change risk analysis. The Archimedean copula approach to dependence modelling has been used by various researchers in modelling interdependent rates. In the present study, we apply the Archimedean copulas to address interdependence effects on climate change data from a developing country. Our research encompasses several key objectives: identifying the interdependence structure of rainfall and temperature data, construction of Clayton, Frank and Gumbel copulas, fitting and selecting the optimum copulas model using maximum log-likelihood criteria and predicting future rainfall and temperature values. The method of moment procedure is used to calibrate the copula dependence parameter and maximum likelihood estimation for the marginal distribution specifications. Subsequently, the performance of the marginal distribution is compared following the AIC, BIC and Log-likelihood values. The findings show that the Frank copula with gamma marginal distribution appropriately accounts for the interdependence effects. Assuming independence of climate data leads to an underestimation of the climate change risk. This dependence structure is utilized to predict future weather patterns. The research findings have a significant effect on Kenya's economy since agriculture is a major economic activity heavily dependent on accurate weather predictions
