1. Prices and Quantities in a Climate Policy Setting Svante Mandell

2. Observations and aim In practice (the EU): ● Overarching quantitative target for CO 2 ● A dual regulation; CaT and emission taxes Under uncertainty, emissions taxes outperform CaT for handling GHG Q: (When) is a dual regulation justifiable?

3. The model Starts in a classic Weitzman (-74) setting ● Linear MAC- and MAB-functions ● Uncertainty (additative, symmetric round zero) ● Aggregate abatement benefits relevant Answers if CaT or emission tax is preferable CO 2 causes a stock externality  A flat MAB  Variation in emissions ‘better’ than variation in price  Use a tax

4. The model, cont. Mandell (2008), JEEM: Allow for dual regulation ● Tax a subset of emitters, the rest CaT Outcome closer to optimum, but not cost effective Full CaT never optimal, full tax optimal for (sufficiently) flat MAB-functions

5. The model, cont. This paper: ● Flat (horizontal) MAB-function ● A global cap that may never be exceeded ● Two periods Intuition: ● The global cap may require high tax to be met  full tax may not be optimal

6. Timing of the model STAGE 0 Policy maker decides on share to tax and tax level STAGE 1 Emitters choose emission volumes Uncertainty 1 is resolvedPossible surplus is bankedUncertainty 2 is resolved STAGE 2 Emitters choose emission volumes Tax level may be changed

7. Policy goal Policy maker strives to ● Minimize present value of expected efficiency loss ● S.t. the global cap must not be exceeded Thus, we need an expression for E{DWL tot }

8. Two sources of eff. loss Volume error Actual emissions differ from efficient amount Allocation error Abatement efforts not distributed in a cost effective manner Period 1 Period 2

9. The taxes As low as global cap permits, but never below the MAB Less stringent global cap  lower taxes Period 1 is ”sunk” when setting T 2 ● T 2 typically lower than T 1, due to surplus Taxes increase in share of taxed emitters

10.  n * / N  = 0  = 0.5  = 1 ”Strict” global caps, i.e., a cap below expected efficient level Optimal share to tax (n*) ”Lenient” global caps, i.e., a cap above expected efficient level  = discount factor At  =0 the model becomes a one-period model (outcomes in period 2 are given zero weight)

11. Some intuition for n* Start in a situation where ● Global cap = expected efficient level ● All emitters are in CaT At low MAC realizations – too high emissions At high MAC realizations – too low ● Thus, an expected volume error ● But no allocation error

12. Some intuition for n* (cont.) Move some emitters to taxed sector ● At low realizations; decreased error ● At highest realization; emissions equal global cap ● Other high realizations; increased error ● And also an allocation error Motivates a small taxed sector

13. Some intuition for n* (cont.) Now, consider a higher global cap A larger set of realizations will yield a decrease in efficiency loss Motivates taxing a larger share Thus, n* increases in the global cap

14. The role of the discount factor Most likely a surplus in period 1 ● Policy maker may not destroy permits – increased cap period 2 Lenient cap period 1  even more so period 2  risk for large efficiency loss Stringent cap period 1  less stringent period 2  may decrease efficiency loss

15. The role of the discount factor  n * / N  = 0  = 0.5  = 1 More weight on period 2 calls for a lower n* under leninet global cap… …but a higher n* under stricter global cap

16. Conclusions Often, a dual regime is better than full emissions tax or CaT Even accounting for not cost effective This depends on ● The global cap vs. E{eff. emissions} ● Indirectly the slope of the MAC vs MAB ● The discount factor

17. Actual EU policy Contains both crucial elements ● A quantitative target and a flat MAB, but: The ’global cap’ is not entirely fixed, e.g., CDM ● Suggests the model underestimates the optimal share to tax Trading firms may bank ’individually’ ● Suggests the model overestimates the optimal share to tax