Authors: Alexander M. Goulielmos
Addresses: Department of Maritime Studies, University of Pireaus, 80 Karaoli and Dimitriou Street, Piraeus 18534, Greece
Abstract: This paper uses historic data for the prices of second-hand tankers to predict a value for the following 12 months. This then was compared with the actual. Five non-linear forecast methods were tested in a preliminary test and the three more efficient used. Technical analysis was carried out using softwares NLTSA (2000) and MATLAB 5.3. The methods were tested for predicting accuracy inside the sample on the criterion to achieve deviations less than 1%. Then, forecasting outside the sample was attempted. The main results are that second-hand prices time series are not random as believed; they have a long memory and exhibit persistence. There are also out there non-periodic cycles in values, of 48 and 96 months, being more risky for the first nine years (total 22 years). Older analyses used random walk with heavy skewness and heavy kurtosis defeating, thus, their conclusions. They did not realise that this denied normality. The asset play among Greeks for 1992-2004 is also presented.
Keywords: tanker prices; nonlinear forecasting; timing; V-statistic; cycles; risk; asset play; second-hand ships; second-hand maritime market.
International Journal of Shipping and Transport Logistics, 2010 Vol.2 No.4, pp.383 - 410
Available online: 30 Sep 2010 *Full-text access for editors Access for subscribers Purchase this article Comment on this article