Malte Schwoon Learning-by-doing, Learning Spillovers and the Diffusion of Fuel Cell Vehicles Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences University of Hamburg International Max Planck Research School on EARTH SYSTEM MODELLING Presentation at the International Conference on Computational Management Science May 17-18, 2006, Amsterdam
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 2 Introduction Why fuel cell vehicles (FCVs)? Agent based technology diffusion model Learning by doing (LBD) in fuel cell technologies LBD in energy technologies Calibration/scenarios Diffusion of FCVs depends on learning rate Learning spillovers Increase speed of diffusion Asymmetric impact on car producers Conclusion Outline
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 3 Why FCVs? No local emissions, low noise Long term potential: Individual transport with low CO 2 emissions (depending on energy mix of hydrogen production) Reduced dependency on oil New design options (low floor, low center of gravity) Introduction
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 4 Mercedes Benz: NECAR 2 (1996) Fuel: CGH 2 Two 25 kWe PEMFC (Ballard) Cont. 33 kW, max. 45 kW Range: 250km Max speed: 110km/h Acceleration: quite good Introduction Fuel: CGH kWe PEMFC (Honda) 80 kW front + 2x 25 kW rear Regenerative braking Range > 500 km Max speed: 160 km/h (limited) Honda FCX (2005)
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 5 Can we switch to an H 2 -economy? (1) Technological problems basically solved (RECENTLY!) : Fuel cell technology, H 2 -on-board storage, etc. We will never switch! We can switch soon! The - problem of H 2 -infrastructure (2) Economic start up problem for large scale introduction: No H 2 -infrastructure nobody buys FCV Nobody buys FCV no H 2 -infrastructure Introduction
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 6 or vice versa Introduction Scenarios/Projections of the diffusion of FCVs and/or H 2 -infrastructure: Schlecht (2003), Thomas et al. (1998), Moore and Raman (1998), Ogden (1999, 2002), Stromberger (2003), Mercuri et al. (2002), Sørensen et al. (2004), Oi and Wada (2004), Hart (2005), etc. Common approach 1. Develop scenarios of the number of hydrogen vehicles 2. Derive implied H 2 -demand/H 2 -infrastructure Implied assumption: smooth and successful introduction of both technologies Studies ignore dynamic interactions Technology driven studies ignore impact on producers/consumers
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 7 Introduction Tax Filling station owners: Increase share of stations with H 2 -outlet Government: Sets taxes and increases number of H 2 -outlets Producers: Production and price decisions Consumers: Buying decisions Credit availability Producers capital R&D funds Investment decisions Market sales Profits Savings Car characteristics (Expected) LBD cost reductions Refueling worries Driving patterns Neighbors Kwasnicki (1996) Janssen and Jager (2002)
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 8 Learning by doing Electric Technologies in EU, Source: Wene (2000) Progress ratio Learning rate = 1 – Progress ratio
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 9 Learning by doing Energy technologies (25 obs.) Various industries (>100 obs.) Observed learning rates McDonald and Schrattenholzer (2001)Dutton and Thomas (1984) Learning rate for fuel cell technologies?
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 10 Calibration/scenario Central case parameterization German compact car segment (1 mio sales per year) - 12 producers different representative consumers Initial fuel cell cost of per unit for (mass) production of 1000 units Learning rate (LR) 15% (sens %) Fuel cell cost Internal combustion engine 5% tax increase every year (tax 40%) Introduction
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 11 Learning by doing Percentage share of FCVs within newly registered cars in the German compact car segment
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 12 Learning spillovers due to Reverse engineering Inter-firm mobility of workers Proximity (industry clusters) Weak patent rights (government control) Joint research projects Learning on sub-contractor level (Ballard Power Systems, International Fuel Cells) (Opposite: proprietary learning) Learning spillovers
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 13 10% spillover: 10 FCVs produced at competitor's plant equivalent to 1 produced at own plant Learning spillovers Percentage share of FCVs within newly registered cars in the German compact car segment
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 14 Learning spillovers Change of NPV of profits ( ) relative to no spillover case
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 15 Conclusion Hydrogen/FCV individual transport system: Technological option, but requires governmental commitment Multi-agent simulation model helps understanding of dynamics (Standard sim-problems apply: parameters, functional forms, random events…) Modeling results High learning rates High spillovers High spillovers 2nd/3rd mover advantage Spillover policies?Environmentally concerned government: High spillover policy fast diffusion Asymmetric impact on producers Resistance/appreciation of producers depends on their position in the switching-chain fast diffusion
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 16 Thank you!
Research Unit Sustainability and Global Change Centre for Marine and Atmospheric Sciences 17 Learning by doing Percentage share of FCVs within newly registered cars: Different lengths of the producers' decision horizons