Research Rtports . which reco rd research conducted at IIASA, are ind ependently re vi ewed before publication. Views or opinions expressed herein do not necessarily re prese nt those o f th e Institute , its Na.tiona.l Member Orga.nizations, or other organizations s upporting th e wo rk.

MESSAGE-MACRO is the result of linking a macroeconomic model with a detailed energy supply model. The purpose of the linkage is to consistently reflect the influence of energy supply costs as calculated by the energy supply model in the optimal mix of production factors included in the macroeconomic model. In this article, we describe an automated link of two independently running models. The advantages of this setup over a single, fully integrated model are twofold: First, it is more flexible, leaving the constituent models intact for independent runs , thus making further model development an easier task. Second, the decomposed model solution benefits numerically from having the most non-linearities concentrated in the smaller of the two modules. The emphasis of the paper is on methodology, but we also include an example demonstrating the feedback mechanisms of MESSAGE-MACRO by applying it to two global economicenergy-environment scenarios. The two scenarios are a reference scenario and a scenario that limits the global atmospheric carbon concentration to 550 ppmv. The scenarios are compared in terms of GDP, energy su pply and demand. and energy prices.

We use a coupled carbon-cycle and energy systems engineering model to analyze the future time path of carbon emissions under an illustrative C0 2 concentration stabilization limit of 550 ppm. Our findings confirm the emission pattern as found by WRE: global emissions rise initially, pass through stabilization, in order to decline in the second half of the 21st century. We show that for a given C0 2 concentration target, emission trajectories within an intertemporal optimization framework depend mainly on two factors: the discount rate, and the representation of technological change as either static or dynamic. We obtain a similar near-term emission time path as WRE when using a model with static technology and a discount rate of 7%. We obtain a trajectory with lower emissions in the near-term when using a lower discount rate and/or treating technology dynamics endogenously in the model. We briefly outline a model that endogenizes technological change through learning curves. We then compare differences in emission trajectories between alternative model formulations of technological change. They are sufficiently small as to be of secondary importance when compared to treating C0 2 concentration stabilization as an inter-temporal optimization problem or not. Whereas our results confirm the computational results of WRE, we arrive nonetheless at different policy conclusions. If long-term emission reduction is the goal, we cannot follow 'business as usual' even in the short-term. Action needs to start now. Action does not necessarily mean aggressive short-term emission reductions but rather enhanced R & D and technology demonstration efforts that stimulate technological learning. These are the necessary preconditions that long-term reduction targets can be met with improved technology and at costs lower than today. We close by pointing out two further *Corresponding author. Tel.: +43 2236 807470; fax: +43 2236 71313; e-mail: gruebler@iiasa.ac.at 0140-9883/98/$ -see front matter© 1998 Elsevier Science B.V. All rights reserved. P/IS0140-9883(98)00010-3 496A . Griibler, S. Messner/ Energy Economics 20 (1998) [495][496][497][498][499][500][501][502][503][504][505][506][507][508][509][510][511][512] critical issues: uncertainty, and the possible mismatch between the world of economic models and that of climate policy.

All rights resen ecl. No pa.rt of this publica ti o n m a.v be re pro duced o r tra ns mitted in a.ny form or by a n~' mea.ns. elec t ronic o r mec ha ni ca l. includin g photoco py, recordin g, or any inform a.t.ion s to rage o r retrie val syste m , \\·it ho ut pe rmissio n in writin g fr om t he co py right hold er. Abstract-This paper introduces an approach to modeling the uncertainties concerning future characteristics of energy technologies within the framework of long-term dynamic linear programming models. The approach chosen explicitly incorporates the uncertainties in the model, endogenizing interactions between decision structure and uncertainties involved. The use of this approach for future investment costs of electricity generation technologies in the framework of very long-term energy scenarios shows improvements in model behavior and more robust solutions with respect to technology choices made.

No abstract