Econometric Analysis of Ship Life Cycles – are safety inspections effective?
and changes thereof during its economic life. One approach forinference could be to apply a logit or probit model on the probability of a ship incident.In such a model, the dynamics are discarded because it only considers whether a shiphad an accident at fixed points in time. The choice of these fixed points has a bigimpact on the estimation results. It is also not straightforward to include time-varyingcovariates into such a model. We have information on daily basis and therefore useduration analysis on the length of the ship’s economic life is the natural approach forthis dynamic framework. It enables us to measure the effect of inspections on theincidence rate of a vessel.Table 1 lists some of the major maritime incidents for oil tankers and passenger vesselsstarting in 1912 with the
claiming 1,517 lives to one of the most recent passenger vessel accident claiming 1,000 lives (
2006).According to the International Oil Pollution Claim Fund (IOPCF) and the InternationalTanker Owners Pollution Federation (ITOPF), the associated costs can vary from USD9.5 billion (
, 1989) to USD 37 million (
, 1996) where the sizeof the vessel or oil spill is not directly related with the associated economic costs(ITOPF, 2007). According to Grey (1999), accident costs translated to USD/tons of oilspilled reveals a wide range from as little as USD 667/tons of oil spilled (
, 1991)to USD 180,000/tons of oil spilled (
Shinryu Maru No 8
, 1995) This furtherdemonstrates the difficulty in estimating the true economic cost associated with an oilspill. For most incidents, an estimated cost figure could not be found, especially forsome of the older incidents.Very often the shipping industry triggers legislative reactions after incidents. For theUS, the
incident triggered the creation of the Oil Pollution Act (OPA90)while for the European Union (EU), the two latest incidents at the coast of France andSpain, the
(2002) triggered a full revision of the EU maritimelegislative framework dealing with all aspects of the shipping industry – the so calledEU Third Maritime Safety Package.Despite the development of a complex legislative framework in the shipping industry, parts of the enforcement remain weak. This is due to the fact that the industry is veryglobal and its regulatory framework is based on international law. The loopholes in theenforcement create a market for substandard ships which is estimated by the OECD tocomprise 5-10 percent
of the world fleet. These substandard ships distort competitionto prudent ship owners since costs are saved to cut the edges on what would otherwise be acceptable. Following a series of major oil tanker accidents in the 70’s, the conceptof port state control (PSC) evolved which allows port states to conduct safetyinspections on foreign flagged vessels entering its ports. The countries groupedthemselves into PSC regimes based on Memoranda of understanding and today, thereare currently 10 such regimes in force covering most of the port states.From a public perspective, the desired situation is to promote safe, secure andenvironmentally friendly maritime transportation and to decrease the number of
Ship particulars is a standard term used in the shipping industry and contain the description of physicalcharacteristics of a vessel such as ship type, hull type, gross tonnage and operational items such as ownership, safetymanagement or the registry of a vessel.
Peijs, K. (2003). Ménage a trois. Speech at
(November 2003: Amsterdam)
4substandard vessels in order to prevent the likelihood of a maritime incident bearingsubstantial economic costs.
Table 1: Major Shipping Incidents of Oil Tankers and Passenger Ships
Ship Name Year LocationSpill Size(tonnes)Economic Costs
Titanic 1912 North Atlantic 1,517 livesTorrey Canyon 1968 Scilly Isles, UK 119,000 No estimateSea Star 1972 Gulf of Oman 115,000 No estimateMetula 1974 Magellan Street, Chile 47,000 No estimateJakob Maersk 1975 Oporto, Portugal 88,000 No estimateUrquiola 1976 La Coruna, Spain 100,000 No estimateArgo Merchant 1976 Nantucket Sound, USA 28,000 No estimateHawaiian Patriot 1977 300 nautical miles off Honolulu 95,000 No estimateAmoco Cadiz 1978 Off Brittany, France 223,000 US$ 282 millionIndependenta 1979 Bosphorus, Turkey 95,000 No estimateAtlantic Empress 1979 Off Tobago, West Indies 287,000 No estimateIrenes Serenade 1980 Navarino Bay, Greece 100,000 No estimateCastillo de Bellver 1983 Off Saldanha Bay, South Africa 252,000 No estimate Nova 1985 Off Kharg Island, Gulf of Iran 70,000 No estimateHerald of Free Enterpr. 1987 Off coast of Belgium 193 livesDona Pax 1987 Philippines 4,000 livesOdyssey 1988 Off Nova Scotia, Canada 132,000 No estimateKhark 5 1989 Off Atlantic coast of Morocco 80,000 No estimateExxon Valdez 1989 Prince William Sound, USA 37,000 US$ 9.5 billionScandinavian Star 1990 Baltic Sea 158 livesABT Summer 1991 700 nautical miles off Angola 260,000 No estimateHaven 1991 Genoa, Italy 144,000 US$ 96 millionAegean Sea 1992 La Coruna, Spain 74,000 US$ 60 millionKatina P 1992 Off Maputo, Mozambique 72,000 No estimateBraer 1993 Shetland Islands, UK 84,700 US$ 83 millionEstonia 1994 Baltic Sea 852 livesSea Empress 1996 Milford Haven, UK 72,000 US$ 62 million Nakhodkha 1997 Japan 17,500 US$ 219 millionErika 1999 Off Coast of France 20,000 US$ 180 millionMV Joola 2002 West Africa 1,863 livesPrestige 2002 Off the Spanish coast 77,000 Euro 778 millionTasman Spirit 2003 Pakistan 30,000 US$ 291 millionAl Salam Boccachio 98 2006 Red Sea 1,000 lives
Source: compiled by authors from various sources, (spill size is in tonnes of oil spilled)
On a regional scale and based on data from one country, Cariou et all (2007a,b) touchupon the topic of effectiveness of PSC inspections and concludes that some of the shipcharacteristics appear to be significant predictors for risk. Talley et al (2005) look at the probability of a vessel being inspected by the United State Coast Guard for a safetyinspection versus a pollution inspection. They recommend a revision of the targeting ofships for inspection in order to enhance their effectiveness. Knapp and Franses(2007a,b,c) look at various aspects of safety inspections using binary logistic regressionon combined datasets. They estimate that a port state control inspection leads to a 5 to10 percent decrease in the probability of a very serious casualty. They recommend thattargeting of ships for inspections can be enhanced by using data from various port state
5control regimes and industry inspections and by taking inspections of other regimes intoaccount. Another policy recommendation is to revise the release of a vessel fromdetention. Furthermore, their analysis shows the importance of ownership versusregistry to define the risk profile of a vessel.The paper is organized as follows. In Section I we explain the combination of datasetsand variables used in this article. To obtain a general impression of the data somedescriptive statistics are given. Section II explains the construction of the variables andmodels used for duration analysis. It also presents its results including the visualizationthereof by taking policy implications into account. In Section III we discuss the resultsand present our conclusions.
I. The Datasets and VariablesA. Combination of Datasets to create ship life cycles
The dataset used in this analysis is compiled from several sources which will beexplained in this section in detail. Particular care was placed on the creation of the datawith respect to the choice of the data sources which are limited in the shipping industrydue to its secretive and political nature. The final data comprises of information on52,130 ships over 100 gross tonnage and contains 748,621 events, ships incidents,inspections and changes in ship particulars, over a 29 year time period (1978-2007).The information in the data can be split into four main groups:1.
ship particular data
(e.g. ship type, ship yard country, beneficial ownership of a vessel, DoC Company
, flag and classificationsociety
representing the shipping markets influencing the shippingeconomic cycles (on a monthly basis)
information on safety inspections4.
data on the timing of incidents and regular death of a vessel (demolition of avessel)
To create the dataset, we combine data from the three major data providers
in theshipping industry. Due to the nature of the shipping industry, it is difficult to obtain rawdata on ship inspections, in particular from port state control regimes (PSC), or industryinspections. We have to our disposal information on safety inspections from six PSCregimes. However, the data could not be obtained from all regimes for the whole time period. The data was further complemented by data from industry inspection systems
and safety management audits (ISM
audits). For each vessel we have a number ofship related information: the flag state, the classification society, the ship yard country