5-th International ASPO Conference, San Rossore, 18-19 July 2006, Italy World Oil Depletion: Diffusion Models, Price Effects, Strategic and Technological Interventions Renato Guseo Department of Statistical Sciences Renato Guseo University of Padova, Italy ASPO-5 San Rossore 18-19 July 2006, Italy 1 World Crude Oil Production Thousand barrels daily 80000 Global 70000 60000 50000 OPEC 40000 FSU 30000 20000 CSI USA (NGL) 10000 0 1900 10 Renato Guseo 20 30 40 50 60 ASPO-5 San Rossore 18-19 July 2006, Italy 70 80 90 2000 2 World Production and Prices Crude Oil Production and Prices per Barrel (X 10000) 8 80 Variables Production (barrel x 1000) Price (dollars, 2002) 6 60 4 40 2 20 0 0 1900 1920 1940 1960 1980 2000 2020 year Fonte: BP Statistical Review of World Energy, 2004 Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 3 Growth and Development after World War II -1• Cohen, J.E. (2003) Human Population: The next Half Century, Science, 203, 1172-1175; • Exceptional demographic expansion; • Rural population peak in rich countries: 1950; • Increse of world average life: 30 years in 1900 and 65 years in 2000; • Population: 6.3 billion in 2004; United Nations Population Division: 8.9 billion in 2050 (ex medium variant scenario forecasting); • Regional population contraction in 2050: Japan -24%; Italy -22%; FSU -29%; Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 4 Growth and Development after World War II -2• USA dominance in oil estraction and refining: shock 1918 (positive with local memory); • Decisive military advantage. Competitive advantage after World War II; • American way of life; • Energetic Surplus based on cheap crude oil; • Structural change in sustainable economic evolution based on non-renewable resources; Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 5 Growth and Development after World War II -3• Today risk: physical restrictions towards expansion in oil production due to emergent demand by China and India; • Risk of a late migration towards renewable energetic resources; • Emerging technologies are not completely sustainable and efficient (fuel-cells, hydrogen, photovoltaic systems, solar thermal systems, eolic systems, etc.). Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 6 Recent strategic researches 1. Morse, E.L. e Jaffe, A.M. (2001). “Strategic Energy Policy Challenges for the 21st Century”; (2000 - aprile 2001); James A. Baker III Institute for Public Policy of Rice University, Texas; Council of Foreign Relations of USA 2. National Energy Policy Development Group, (2001, Task Force supervised by D. Cheney). • • • • USA energetic policy since 1940 – security -; Global Economic growth based on a production surplus with cheap prices; New emerging world oil demand; Supply dependence (Midle East). Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 7 Non-Renewable Resources Depletion: Hubbert and recent developtments • Hubbert, M.K. (1949). Energy from fossil fuels, Science, 4, 103-109. • In 1956 Hubbert forecasts the peak of annual production within 48-lower states in USA by year 1970; • Campbell, C. e Laherrère, J. (1998). The End of Cheap Oil, Scientific American, March 1998. • Laherrère, J. (2003). Modelling future oil production, population and the economy, ASPO 2nd international workshop on oil and gas, Paris, 26-27. • ASPO (Association for the Study of Peak Oil and Gas); Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 8 Economic and financial “estimation” of oil reserves: inflated figures • Warranties on international longterm loans; • Investments on production plants; • OPEC restrictions overcoming: export is proportional to the “declared” reserves. Renato Guseo BP - Statistical Review of World Energy Source: BP Statistical Review of World Energy ASPO-5 San Rossore 18-19 July 2006, Italy 9 Ultimate Recoverable Resouce: URR • URR: total amount of a finite resource obtainable at the end of extraction process; • Geologic estimates of Oil URR during the life- cycle of resource extraction: 1. Production to date. Oil heterogeneity. (weight 100%); 2. Proven reserves. Ex-post recovery factor 35% is a median and today’s variability is very large. Revision principles and “reserves growth” (USGS) enlarge uncertainties; 3. Probable and possible reserves: undiscoverd petroleum based on subjective assessments (probability, e.g. 5%). •Could we do a simple weighted sum of such components? Probably not Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 10 Recent statistical modeling • Guseo, R. (2004) Interventi strategici e aspetti competitivi nel ciclo di vita di innovazioni, Working Paper Series, 11, Department of Statistical Sciences, University of Padua. • Guidolin, M. (2004) Cicli energetici e diffusione delle innovazioni. Il ruolo dei modelli di Marchetti e di Bass, Thesis, University of Padua. • Guseo, R and Dalla Valle, A. (2005) Oil and Gas Depletion: Diffusion Models and Forecasting under Strategic Intervention, Statistical Methods and Applications, 14(3), 375-387; • Guseo, R., Dalla Valle, A. and Guidolin, M. (2006) World Oil Depletion Models: Price effects compared with strategic or technological interventions, Technological Forecasting and Social Change (in press); • Guseo, R. (2006) Bass-let Detection of Automobile Successive Generations: Evidence from the Italian Case (submitted). Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 11 Crude Oil Production: Diffusion of an Innovation • Oil production is modulated by the dynamics of international demand; • Oil demand is a function of the diffusion processes related to basical technologies (transport, chemical industries, heating, etc.); • Diffusion of technological innovations is conditioned by social communication structure: innovators and imitators (word-of-mouth) Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 12 Bass Equation: BM • z’(t) = mf(t) = m[p+qF(t)][1-F(t)] or • z’(t) = pm+(q-p)z(t) - (q/m) z(t)2 (Riccati) • z’(t)=mf(t) (instantaneous adoptions); • f(t)=F’(t) • z(t)=m F(t) (cumulative adoptions); F(t)=z(t)/m • f(t)/[1-F(t)]=p+qF(t) Bass Hazard rate • m=potential market; carrying capacity; URR • p=innovation coefficient, p>=0 • q=imitation coefficient, q>=0 Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 13 Normalized Bass models: BM and GBM BM: f(t)/[1-F(t)]=[p+qF(t)] GBM: f(t)/[1-F(t)]=[p+qF(t)] x(t) “Standard” “GBM” x(t) is a quite general intervention function: integrable, positive and “centered” around unitary “neutral pole”: 1. Representation of temporal price variations, of advertising pressure, of political, strategic, legal, environmental interventions. Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 14 Equation Solution: GBM z z ' (t ) ( p q )( m z ) x (t ) m t ( p q ) x ( ) d 0 1 e z (t ) m t q ( p q ) x ( ) d 0 1 pe b1(t a1) b2(t a2 ) Exp. shocks x(t ) 1 c1 e Rect. shocks x(t ) 1 c1 I (t a ) I (t b ) c2 I (t a I (t a 1 Mixed shocks x(t ) 1 Renato Guseo b1(t a1) ce c2 e ) 1 1 I (t a ASPO-5 San Rossore 18-19 July 1 2006, Italy ) 1 c2 I (t I 2) (t I a2 ) ( t b2 ) a2) I (t b152) Great Britain: GBM, 2 mixed sh. Estimation method: Marquardt Estimation stopped after maximum iterations reached. Number of iterations: 31 Number of function calls: 330 Estimation Results GBbpC Asymptotic 95,0% Asymptotic Confidence Interval Parameter Estimate Standard Error Lower Upper ---------------------------------------------------------------------------m 4513,39 154,806 4196,77 4830,0 p 0,0000708436 0,0000324773 0,00000441993 0,000137267 q 0,111872 0,00516425 0,10131 0,122434 c1 8,54019 1,02935 6,43493 10,6454 b1 -0,250721 0,0114596 -0,274159 -0,227284 a1 10,7677 0,458356 9,83028 11,7052 c2 -0,331417 0,0175489 -0,367309 -0,295526 a2 23,4341 0,190843 23,0438 23,8245 b2 28,6819 0,164258 28,3459 29,0178 ---------------------------------------------------------------------------Analysis of Variance Source Sum of Squares Df Mean Square Model 6,52091E7 9 7,24545E6 Residual 657,546 29 22,674 (X 1000) ----------------------------------------------------3 Total 6,52097E7 38 Total (Corr.) 3,31712E7 37 R-Squared = 99,998 percent 2,5 R-Squared (adjusted for d.f.) = 99,9975 percent Standard Error of Est. = 4,76172 Mean absolute error = 3,10566 2 Durbin-Watson statistic = 0,889298 Positive Shock with local memory Plot of Fitted Model 1,5 1 0,5 0 0 Renato Guseo 20 ASPO-5 San Rossore 18-19 July 2006, Italy 40 t 60 80 16 Great Britain: analysis • The “saddle” 19871991-1999 is perfectly absorbed by a rectangular shock: • a) Petroleum Reven Tax modification; • b) pipelines restructuring 19861991; symmetric behaviour confirms ordinary regime; • c) partial production stall due to the reduction of new discoveries. Renato Guseo Multiple X-Y Plot 150 Variables GBbp DIFF(PRED) DIFF(FOR) 120 90 60 30 0 0 20 40 60 80 t ASPO-5 San Rossore 18-19 July 2006, Italy 17 USA: 48 lower States and Alaska, one exponential shock Estimation Results ---------------------------------------------------------------------------Asymptotic 95,0% Asymptotic Confidence Interval Parameter Estimate Standard Error Lower Upper ---------------------------------------------------------------------------m 224,885 0,784401 223,328 226,442 p 0,000445866 0,0000177788 0,000410571 0,000481162 q 0,0571941 0,000403937 0,0563922 0,057996 c1 0,682617 0,0735348 0,536632 0,828602 b1 -0,0852885 0,00948373 -0,104116 -0,0664609 a1 18,0477 0,981086 16,1 19,9954 ---------------------------------------------------------------------------- Plot of Fitted Model 200 cum 160 120 80 40 0 Analysis of Variance ----------------------------------------------------Source Sum of Squares Df Mean Square ----------------------------------------------------Model 735809,0 6 122635,0 Residual 7,39124 95 0,0778026 ----------------------------------------------------Total 735817,0 101 Total (Corr.) 352880,0 100 R-Squared = 99,9979 percent R-Squared (adjusted for d.f.) = 99,9978 percent Standard Error of Est. = 0,278931 Mean absolute error = 0,207909 Durbin-Watson statistic = 0,173839 Positive shock with local memory Renato Guseo 0 20 40 60 80 100 120 t Multiple X-Y Plot 4 Variables barili DIFF(PREDb1) DIFF(PREDbe1) 3 2 1 0 0 20 ASPO-5 San Rossore 18-19 July 2006, Italy 40 60 t 80 100 120 18 USA: 48 lower States and Alaska, ARMAX(4,0,2) sharpening Time Sequence Plot for barili ARIMA(4,0,2) + 1 regressor 4 actual forecast 95,0% limits 3 barili Forecasting - barili Analysis Summary Data variable: barili Number of observations = 101 Start index = 1,0 Sampling interval = 1,0 2 1 0 Forecast Summary ---------------Forecast model selected: ARIMA(4,0,2) + 1 regressor Number of forecasts generated: 40 Number of periods withheld for validation: 0 -1 0 ARIMA Model Summary Parameter Estimate Stnd. Error t P-value ---------------------------------------------------------------------------AR(1) 1,21416 0,691695 1,75534 0,082426 AR(2) -0,140994 1,11031 -0,126986 0,899220 AR(3) -0,146337 0,49692 -0,294488 0,769028 AR(4) -0,132467 0,0891259 -1,48629 0,140514 MA(1) 0,591549 0,68527 0,863235 0,390183 MA(2) 0,299352 0,650254 0,460362 0,646308 DIFF(PREDbe1) 0,20426 0,0890786 2,29303 0,024052 ---------------------------------------------------------------------------Backforecasting: yes Estimated white noise variance = 0,00495321 with 95 degrees of freedom Estimated white noise standard deviation = 0,0703791 Number of iterations: 17 Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 30 60 90 120 150 Shock: 1918 19 Alaska: ARMAX(2,0,1) sharpening ARIMA Model Summary Parameter Estimate Stnd. Error t P-value ---------------------------------------------------------------------------AR(1) 0,323713 0,100318 3,22686 0,002667 AR(2) -0,172177 0,054492 -3,15967 0,003195 MA(1) -0,818595 0,102552 -7,98224 0,000000 PREDbme1 0,847508 0,0501864 16,8872 0,000000 Mean -0,0143829 0,0282678 -0,508809 0,613990 Constant -0,0122034 ---------------------------------------------------------------------------Backforecasting: yes Estimated white noise variance = 0,00281514 with 36 degrees of freedom Estimated white noise standard deviation = 0,0530579 Number of iterations: 20 Multiple X-Y Plot Residual Autocorrelations for alaskac 2,4 2 1,6 1,2 0,8 ARIMA(2,0,1) with constant + 1 regressor Autocorrelations 1 0,6 0,2 -0,2 Variables alaskap DIFF(FORar2ma1 ) 0,4 0 -0,6 0 -1 0 3 6 9 12 20 15 40 60 80 t lag Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 20 World Oil data: Daily Production Sources: • Industriedatenbank 2001 (1900 – 1986) • BP Statistical Review of World Energy (19872002) Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 21 GBM: x(t) pure prices control Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 22 GBM exp shocks + price effect Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 23 Guidolin (2004): GBM, 2 exp shocks World Oil Depletion Models (X 10000) 8 Variables Production data gbm2e+for 6 4 2 Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 2010 2000 year 1990 1980 1970 1960 1950 0 24 GBM: 3 exp shocks (memory persistence) World Oil Depletion Models (X 10000) 8 Variables Production data gbm3e+for 6 4 2 Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 2010 2000 year 1990 1980 1970 1960 1950 0 25 GBM 3 exp shocks: estimates (memory persistence) q/p = 608 Qp=1%; R2 0,999994708 F ( m. parz / 1951) 17,06 Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 26 World Oil Depletion: GBM with three shocks vs Hubbert-Bass Oil Peak: 2007 URR=1524 Gbo Depletion time 90% : 2019 Renato Guseo Depletion time 95% : 2023 ASPO-5 San Rossore 18-19 July 2006, Italy 27 Oil market Operators: prices growth • Demand and supply self-control similar to 1973 and 1979-’83 behaviour. Extension of crude oil economic life cycle; • Limited techonological efficiency margins due to the improvementes of 70’s , 80’s and 90’s; • Savings through “life styles” modification: this is the central dilemma in industrialized countries; • Sluggishness of middle class whose life style is based on an expected irreversible and indefinite growth and development. Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 28 New Emergent Economies • Increase in crude oil requirements by recent emergent economies: China, India, and other Asiatic countries; • US EIA (Energy Information Administration) “forecasts” a world oil demand of 40 Gbo/year (or 109.6 milion daily barrels) in 2020; • Guseo, Dalla Valle, Guidolin (2006) and Bakhtiari (2004) forecast, in 2019-20, only 55 milion daily barrels. Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 29 Crude Oil: Area consumption Consumi medi giornalieri (in barrel X 1000) (X 10000) 3 2,5 2 Variables North Ame 1,5 Europe Eurasia Asia Pacific 1 South Central Ame 0,5 0 1960 Renato Guseo Middle East Africa 1970 1980 1990 Anno ASPO-5 San Rossore 18-19 July 2006, Italy 2000 2010 30 Outlook • Nuclear fission perspective is probably a tardy strategy with known collateral externalities. A new plant in Italy can be launched after 13-15 years; • Technological, political and economic efforts must be distributed in different areas: photovoltaic, solar thermal, bio-fuel, biomass, eolic, hydrogen, etc.; • Electric sector: distributed investments (photovoltaic, micro-cogeneration, etc.). • Individual and collective mobility. Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 31 World Oil Depletion: GBM with three shocks vs Hubbert-Bass vs five shocks scenario Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 32 World Oil Depletion: GBM with three shocks vs five shocks vs four shocks scenarios Shock 2008 (sim. 1951) Depletion time 90% : 2017 Renato Guseo ASPO-5 San Rossore 18-19 July 2006, Italy 33