Panamax 4TC Rate Forecasting: A Kalman-Filter Approach

22 Pages Posted: 9 Dec 2012

Date Written: May 20, 2010

Abstract

This paper compares a series of Vector Autoregressive Models, Vector Error Coerrection Models and ARIMA to State space models using the Kalman filter routine to Forecast Spot Panamax 4TC rates. The null hypothesis of this investigation is that VAR and ARIMA models with lagged spot and Forward Freight Agreements (FFA) returns provide the best forecast for spot rates. The alternative hypothesis is that they do not. In that respect, if the null the hypothesis holds then VAR or ARIMA models calculated through a Maximum likelihood OLS will be the best estimates of expected spot prices. The second Null hypothesis is that the FFA returns can be used to provide better forecasts than regressing the spot returns to its lagged first differences i.e. either through OLS or not. Although it is found that FFAs do contain information about the spot, as indicated by past literature, linear dynamic models of lagged spot returns exhibit a lower Root Mean Squared error and directional forecasting Accuracy than all other models for daily forecasts. However, results of weekly and monthly forecasts are mixed, indicating that incremental reduction in RMSE is higher when including forward contracts of longer maturities as explanatory variables. In addition, it is found that chained timeseries of contracts with longer maturity contain more information about the spot.

Keywords: Forward Freight Agreement, Panamax 4TC, Econometrics, Time-series Forecasting, Kalman Filter

Suggested Citation

Grammatikopoulos, Alexandros, Panamax 4TC Rate Forecasting: A Kalman-Filter Approach (May 20, 2010). Available at SSRN: https://ssrn.com/abstract=2186664 or http://dx.doi.org/10.2139/ssrn.2186664

Alexandros Grammatikopoulos (Contact Author)

City University London - The Business School ( email )

106 Bunhill Row
London, EC1Y 8TZ
United Kingdom

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
223
Abstract Views
1,150
Rank
248,552
PlumX Metrics