Duration analysis for recurrent ship accidents

ABSTRACT

About 63% of the world’s shipping accidents are recurrent—they occur to ships that have already experienced at least one prior accident. Therefore, reducing recurrent accidents can contribute significantly to maritime safety. We study the factors affecting both first and recurrent accidents, by focusing on the duration between two accidents. Cox proportional hazard models are applied to ship accident data from 1996 to 2015, and the results identify which ships have a high risk of recurrent accidents, based on ship attributes, ship supply and market conditions, shipbuilding country, previous accident type, and ship type. The recurrent rate is high when the ship involved in the accident is old, small, flies a flag of convenience, and has no detention record. In addition, the accident risk increases when the shipping market faces a high bunker price, overcapacity in supply, a high time charter rate, or low newbuilding price. On the other hand, ships built in China and Japan have lower recurrent accident rates than those built elsewhere, although ships built in China have earlier first accidents than do others. General cargo ships have the highest recurrent accident rate, followed by dry bulkers, container ships, and tankers, in that order.

KEYWORDS:

• Recurrent ship accidents
• duration analysis
• extended Cox proportional hazard models
• maritime safety
• statistical analysis

Acknowledgment

This work is partly supported by The Hong Kong Polytechnic University Central Research Grant (Grant No: PolyU 155056/15E).

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The 34 countries that have been declared FOCs by the International Transport Worker’s Federation (ITF) are as follows: Antigua and Barbuda, Bahamas, Barbados, Belize, Bermuda (the U.K.), Bolivia, Burma, Cambodia, Cayman Islands, Comoros, Cyprus, Equatorial Guinea, Faroe Islands, French International Ship Register, German International Ship Register, Georgia, Gibraltar (the U.K.), Honduras, Jamaica, Lebanon, Liberia, Malta, Marshall Islands (the U.S.A.), Mauritius, Moldova, Mongolia, Netherlands Antilles, North Korea, Panama, Sao Tome and Príncipe, St Vincent, Sri Lanka, Tonga, and Vanuatu.

2. In this study, European countries include Austria, Belgium, Denmark, Finland, France, Germany, Germany (West), Greece, Iceland, Ireland, Netherlands, Norway, Poland, Portugal, Turkey, the U.K.

Additional information

Funding

This work is partly supported by The Hong Kong Polytechnic University Central Research Grant (grant number PolyU 155056/15E) .