1 modelling the impact of opa 90 and double hull technology on oil spill numbers david glen london...
TRANSCRIPT
1
Modelling the impact of OPA 90 and double hull technology
on oil spill numbers
David GlenLondon Metropolitan University
Presented to IMSF Annual ConferenceGdansk, April 2008
2
Outline
• Literature Review
• Model
• Estimation & Results
• Conclusion
3
“It is difficult to identify any one factor which contributes to the decline in overall volume and frequency of spills, rather it is considered to be the result of a range of initiatives taken by governments and the shipping industry.” (Hujier, 2005 p1.)
5
Literature Review
• Talley & Anderson (1995),Talley Jin, and Kite Powell (2001)
• Spills as a function of vessel size, vessel damage severity, and 'regulatory effort'
• Homan and Steiner (2008)
• Spills as a function of traffic volume, nos. of tankers, average tanker size, real oil price, and repair costs
6
Literature Review (2)
• HS model – Poisson Count model of spill numbers pre and post 1991 (OPA introduction)
• Model applied to US Coastguard Data • Found – variables gave reasonable fit.• Post 1991 –the OPA effect
– % of tankers with double hull technology – Dummy variable measuring OPA introduction
• Found both these to be significant in explaining the reduction in spill numbers post 1991
7
ITOPF Oil Spill Numbers, 1970 - 2005
0
20
40
60
80
100
120
14019
7019
7119
7219
7319
7419
7519
7619
7719
7819
7919
8019
8119
8219
8319
8419
8519
8619
8719
8819
8919
9019
9119
9219
9319
9419
9519
9619
9719
9819
9920
0020
0120
0220
0320
0420
05
Year
Nu
mb
ers
SmallSpills LargeSpills AllSpills
Source: www.itopf.org – Oil spill statistics - accessed January 2008
So can we model the ITOPF data in the same way?
8
Data and Model
• 1970 - 1990• Tonne mile data • Real oil price • Average tanker size• Tanker lay up %
• 1970 - 2005• As left hand column• plus• Double hull fleet %• Dummies for
– OPA– MarPol– ISM
9
Model Estimation
• Poisson Count Model• Tested for 'overdispersion' (mean and variance
of spill numbers should be the same value in a PC model)
• Used negative binomial model when overdispersion present
• (Models used because dependent variable takes positive or zero values)
• Applied to All Spills, 'Small Spills' and 'Large Spills' as per ITOPF definitions
10
ResultsModels1970-1990
All S pills C oefficient G L M R obust P - ValueS t E rror
C onstant 2.89400 4.49E -02 0.000S eabtrade 0.00193 3.25E -05 0.000R loilpr 0.00725 3.97E -03 0.068S ize -0.01190 6.04E -03 0.049L ayUp 0.0249 1.11E -02 0.024
N 21 R -squared 0.720L R Index 0.980
Model Type Negative B inomialOverdispers ion tes ts t -s tatisticC -T 0.016 2.59 0.006Wooldridge 0.019 3.20 0.020
11
Results 1970 - 2005
All S pills C oefficient G L M R obust P - ValueS . E rror
C onstant 3.1793 3.661E -01 0.000S eabtrade 1.710E -04 2.910E -05 0.000R loilpr 0.0096 2.692E -03 0.000S ize -0.0141 4.293E -03 0.000L ayUp 0.0209 1.003E -02 0.037Dhull -2.2167 0.3007 0.000N 36 R -squared 0.819
L R Index 0.930Model Type Negative B inomialOverdispers ion tes ts t -s tatisticC -T 0.016 2.99Wooldridge 0.023 2.52
12
Principal Results
• Model variables all 'statistically significant'
• Real oil price differs in sign from HS
• Double Hull % highly significant
• Dummy variables – MarPol/ISM/OPA all insignificant so not reported
• Increase in recent accidents driven by increase in world trade tonne miles
13
Actual spills and forecasts from Pre and Post Dhull models
0
20
40
60
80
100
120
140
1970
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
Year
Sp
ill n
um
ber
s
Actual Pre Dhull Post Dhull
14
Large Spills: Actual, pre dhull and post dhull forecasts
0
510
1520
25
3035
40
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
Year
Sp
ill
nu
mb
ers
Actual Pre Dhull Post Dhull
15
Small Spills: Actual, Pre and Post Dhull forecasts
0
20
40
60
80
10019
70
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
Year
Sp
ill
nu
mb
ers
Actual Pre Dhull Post Dhull
16
Summed Up
Average Spill numbers 2000 - 2005 Actual Model Forecasts D Hull No D Hull All Spills 20 21 72 Small Spills 17 17 52 Large Spills 4 3 22
Note: Spill totals rounded to nearest whole spill, hence discrepancies in totals
17
Conclusion
• Oil spills significantly reduced by double hull technology
• No evidence found for other effects
• Trade is a key driver
• Model based on relatively few observations, so may not be robust
18Instead of this, then perhaps we might get....
19
• Thank you for your attention!
ships as green as this one in the future