dc.description.abstract |
In this thesis, we aim at modeling daily rainfall and pricing rainfall derivatives in Ethiopia.
To this end, the standardized two part modeling approach which is known as "occurrence amount process" model is used to model daily rainfall in Ethiopia. A daily rainfall con sists of two components: the occurrence process of daily rainfall and the amount of rain fall on wet days. Therefore, these two components of the daily rainfall are modeled sepa rately. In order to model the occurrence process, we used Markov chain(MC) and Markov
chain analogue year(MCAY) model, that is Markov chain with the AY component which
is a new component and the main contribution of this dissertation. The performance of
these two models are compared and the result shows that the MCAY model describes the
occurrence process in an excellent way compared to Markov chain (MC without the AY
component).
In the second part of the model, that is, in the amount of the daily rainfall modeling
part, four different nonnegative continuous probability distributions are used: exponen tial, mixed exponential, gamma and Weibull distributions. The parameters of these distri butions are estimated using maximum likelihood estimation (MLE). Then by combing the
two component models (the occurrence and amount model), we developed four different
stochastic daily rainfall models: Markov chain analogue year exponential model(MCAYEM),
Markov chain analogue year mixed exponential model(MCAYMEM), Markov chain ana logue year gamma model(MCAYGM) and Markov chain analogue year Weibull model
(MCAYWBM).
The performance of these models are assessed by taking daily rainfall data from 21
weather stations from different parts of Ethiopia. The result revealed that the analoguev
year(AY) component included in this study show great effect on describing the occur rence process of daily rainfall. Also it is observed that all the four distributions in
combination with the MCAY that is, MCAYEM, MCAYMEM, MCAYGM and MCAY WBM have nearly the same performance with only slight difference between them. Since
MCAYMEM has slightly better overall performance compared to the other three models,
the MCAYMEM is used to simulate 10,000 paths via Monte Carlo simulation to obtain
rainfall index. Then the simulated index is fitted to Gamma distribution and transformed
using Esscher transform to obtain risk neutral price. Finally, we calculate option price
for Teff and wheat crops in their growing season because these crops are among the most
common crops in production and area coverage in Ethiopia |
en_US |