1. Rob Dellink — Modelling the costs of environmental policy 1 Dynamic CGE Modelling for Analyzing Environmental Policies Ekko van Ierland and Rob Dellink Ekko.vanIerland@wur.nl Rob.Dellink@wur.nl or: www.enr.wur.nl/uk/staff/dellink/

2. Rob Dellink — Modelling the costs of environmental policy 2 Set-up of the presentation Aim:assessing the costs of Dutch environmental policy by developing a dynamic AGE model with special attention to pollution and abatement (DEAN)  Introduction  Overview of the model  Data and policy scenarios  Main results  Concluding remarks

3. Rob Dellink — Modelling the costs of environmental policy 3 Part I: Model description

4. Rob Dellink — Modelling the costs of environmental policy 4 Overview of the DEAN model  Multi-sector dynamic Applied General Equilibrium model – perfect-foresight behaviour: Ramsey-type model  Environmental module: pollution and abatement – pollution and abatement are present in the benchmark  No impact from environment to economy – no amenity value of environmental quality – no damages from environment on economy – no efficiency analysis, just cost-effectiveness  Model specified in GAMS / MPSGE & available on website

5. Rob Dellink — Modelling the costs of environmental policy 5 Specification of economic activity  Multi-sector Applied General Equilibrium model – description of the national economy – producers: profit maximisation under perfect competition – consumers: utility maximisation under budget balance & LES structure – equilibrium on all markets (Walras’ Law) – individual agents are price takers; no money illusion  International trade – small open economy – domestic and foreign goods are imperfect substitutes (Armington) – no international co-ordination of environmental policy

6. Rob Dellink — Modelling the costs of environmental policy 6 Specification of economic growth  Dynamic model – perfect-foresight behaviour: Ramsey-type model with finite horizon – exogenous increase in labour supply – endogenous accumulation of capital and greenhouse gasses  Comparison of dynamic behaviour in Chapter 3 – comparative-static specification – recursive-dynamic specification – perfect-foresight speciciation – comparison uses small version of the model

7. Rob Dellink — Modelling the costs of environmental policy 7 Specification of pollution  Environmental themes –individual pollutants aggregated using ‘theme equivalents’ –interactions within theme fully taken into account  Polluters need pollution (permits) for their activities –necessary input of production process / utility formation –tradable permit system implemented in the benchmark –autonomous pollution efficiency improvements  Government auctions pollution permits –environmental policy implemented as restriction of number of permits –revenues are recycled lumpsum to private households

8. Rob Dellink — Modelling the costs of environmental policy 8 Specification of abatement  Using bottom-up technical abatement information – costs and effects of end-of-pipe and process-integrated options: discrete modelling of all available options is practically infeasible – measures ordered by increasing marginal abatement costs – technical potential: in the short run not all pollution can be abated – ‘spending effects’: inputs in Abatement production function  Endogenous choice between (i) paying for pollution permits or (ii) investing in abatement or (iii) reducing activity level  Estimation of “Pollution - Abatement Substitution” (PAS) curves: limited substitution between pollution and abatement

9. Rob Dellink — Modelling the costs of environmental policy 9 From MAC to PAS 0 20 40 60 80 100 120 020406080100120 Emissions (in % of current level) Cumulative abatement costs (in % of maximum) Data abatement costs PAS curve Technical potential Current pollution level Sustainability estimate Short-term policy target

10. Rob Dellink — Modelling the costs of environmental policy 10 Abatement as an economic good  Abatement modelled like ‘normal’ production sector – abatement goods are demanded by all polluters (on a perfect market) – decisions on ratio between pollution and abatement are reversible  The ‘Abatement sector’ production function – nested CES production function – labour, capital and produced goods are inputs in abatement sector production function (the ‘spending effects’) – changes in input costs leads to changes in marginal abatement costs (mainly changes in labour productivity)  Autonomous pollution efficiency improvements

11. Rob Dellink — Modelling the costs of environmental policy 11 Structure of the production function Output Environmental Services 0 Production Intermediate deliveries Labour Capital KLKL YY  ID Pollution permits - unabatabl e part Abatement 0  PAS Pollution permits - abatable part

12. Rob Dellink — Modelling the costs of environmental policy 12 Part II: Calibration

13. Rob Dellink — Modelling the costs of environmental policy 13 Calibration of the model  Environmental themes –Climate change, Acidification, Eutrophication, Smog formation, Dispersion of fine dust, Desiccation, Soil contamination  Benchmark projection –model calibrated to the Netherlands, accounting matrix for 1990 –balanced growth of 2% per year –theme-specific autonomous pollution efficiency improvements –27 production sectors –1 representative consumer for all private households –1 government sector: existing distortionary taxes

14. Rob Dellink — Modelling the costs of environmental policy 14 Data sources  Description of initial situation in 1990 –Social Accounting Matrix: Statistics Netherlands (National accounts) –emissions: Statistics Netherlands / RIVM –abatement cost curves: own compilation based on various sources, including RIVM and ICARUS  Growth rates –own calculations based on data for 1995 and 2000  Parameters –elasticities: extended Keller model / SNI-AGE model –other parameters: existing literature

15. Rob Dellink — Modelling the costs of environmental policy 15 Policy scenarios  Policy scenario NEPP2030 –emission targets for 2030 based on NEPP4 (+expert judgements): Climate -50%; Acid. -85%; Eutroph. -75%; Smog -85%; PM10 - 90% –linear path to target from 2000 - 2030 –stabilisation of emissions from 2030 onwards  Policy scenario Delay –targets for 2030 postponed to 2040  Policy scenario NEPP2010 –additional targets for 2010 based on NEPP3 (+expert judgements)

16. Rob Dellink — Modelling the costs of environmental policy 16 Policy impulse for Acidification

17. Rob Dellink — Modelling the costs of environmental policy 17 Part III: Main results

18. Rob Dellink — Modelling the costs of environmental policy 18 Impact on GDP

19. Rob Dellink — Modelling the costs of environmental policy 19 Impact on GDP

20. Rob Dellink — Modelling the costs of environmental policy 20 Sectoral results  Indirect effects are important – most dirty sectors not necessarily most heavily impacted  Impacts on production sectors very diverse – in long run large reductions in energy sectors and heavy industry – small reductions (or even small increases) in services sectors – combination of shift and shrink  Impacts on consumption more evenly spread – impacts depend crucially on environmental policy abroad – in short run increase in consumption

21. Rob Dellink — Modelling the costs of environmental policy 21 Grouped sectoral results Sectoral effects of NEPP2030 policy 1990201020302050 Private consumption Agriculture0.44-0.08-6.88-9.30 Private consumption Industry0.890.91-8.80-12.05 Private consumption Services1.061.34-3.23-8.57 Sectoral production Agriculture-1.09-7.46-32.64-34.58 Sectoral production Industry-0.60-3.25-35.05-30.64 Sectoral production Services0.09-0.640.49-3.74 Sectoral production Abatement services-0.034.2316.5915.81

22. Rob Dellink — Modelling the costs of environmental policy 22 Emission reductions (year 2030)

23. Rob Dellink — Modelling the costs of environmental policy 23 Technically abatable emissions

24. Rob Dellink — Modelling the costs of environmental policy 24 Gross environmental expenditures

25. Rob Dellink — Modelling the costs of environmental policy 25 Part IV: Final remarks

26. Rob Dellink — Modelling the costs of environmental policy 26 Sensitivity analysis  Specification of technical potential – results highly sensitive to technical potential Smog formation – higher technical potential means lower costs and more abatement  Specification of PAS-elasticity – small impact, as all VOC measures will be implemented anyway – higher elasticity means lower costs and less abatement expenditure  Specification of endogenous environmental innovation – endogenous innovation (read: learning by doing) is likely to occur – any excessive economic costs of environmental policy can be prevented

27. Rob Dellink — Modelling the costs of environmental policy 27 Impact of model variants on welfare Equivalent variation Base specification-5.8 GHG emission policy-7.4 Endogenous innovation-3.2 Labour tax recycling-5.6 Multilateral policy-11.7 High technical potential Smog formation-4.1

28. Rob Dellink — Modelling the costs of environmental policy 28 Future research / room for improvement  Better modelling of energy carriers and fuel switch options – linking emissions of GHGs to input of energy where appropriate – top-down modelling of fuel switch options – ay suggestions on modelling national climate policy?  Add more empirical details on abatement options – sectoral specification of potential options (if possible) – differentiate production function abatement sector – improve modelling of negative cost options  Add feedback effects from environment to economy (benefits)

29. Rob Dellink — Modelling the costs of environmental policy 29 Conclusions  Major (bottom-up) characteristics of abatement options can be integrated in a (top-down) CGE framework  Macro-economic impact ‘modest’  10 percent / 5 years delay / 80 bn Euro net / 145 bn Euro gross  Environmental policy creates both threats and opportunities for production sectors  Technical measures and economic restructuring are both essential  Interactions between environmental problems have substantial influence on results