The paper titles An econometric model of tanker spot freight, applies econometric time-series models to model and forecast Baltic Exchange Dirty Tanker Index routes in Worldscale units. The focal point is the crude oil tanker segment; Very Large Crude-, Suezmax- and Aframax carriers, in regards to Intra-European, Asian and North American crude oil trade routes. The data was sourced in co-operation with the World Maritime University, Malmö, Sweden and the world’s leading maritime data provider Clarksons Research Shipping Intelligence Network. A relatively limited amount of studies has been so far subject to econometrics modelling in the spot freight tanker market context, thus providing exceptional aspiration to fill such gap within the academia. The appliance of spot rates strive from the notion that these tend to reflect a uniform worldwide accessible current price in a marketplace at which a commodity can be sold or bought for immediate delivery. Regarding, seaborne trade spot freight rates, these depict the price charged for the carriage of cargo. Companies linked to global supply chains can enhance their competitiveness, in relation to accurately evaluating and forecasting these spot rates.
‘Can crude tanker spot freight rates be modelled and forecasted by an econometric model?’
The first section applies the univariate autoregressive moving average (ARMA) model, based on the Box-Jenkins framework. Further, the concept of stationarity, relation to the Dicky Fuller test is thoroughly discussed. Based on the acceptance of the null hypothesis of a present unit root, the time-series are transformed to log-returns, thus achieving satisfactory estimates. Additionally, in the second section, crude oil prices and –production, in regards to allocated BDTI routes are introduced. Frist, the vectorised AR (VAR) model is utilised, which takes into account the dynamic relationship between the routes in Worldscale units and the crude oil price benchmarks Brent and West Texas Intermediate (WTI). The Granger causality test confirms that crude oil prices have predictive causality on BDTI rates. Cointegration is rejected by the majority of the models, which included crude oil prices, BDTI routes and respective geographical linked crude oil production, with the exception of two Intra-European routes. In the latter section, the WTI-related, Brent-related VAR- and ARMA models are subject to dynamic and static forecasting methods. Benchmarking is conducted via a random walk model. The criteria for evaluating forecast performance, in particular, the mean absolute error (MAE) and root mean squared error (RMSE), are applied. The WTI-related VAR model seems to be on average superior across the Baltic Dirty Tanker Index routes while showcasing sufficient forecasting performance for the South East Asia to East Coast Australia route.
|Educations||MSc in Supply Chain Management , (Graduate Programme) Final Thesis|
|Number of pages||82|