Baseline choice and performance implications for REDD

Abstract

Reducing Emissions from Deforestation and forest Degradation (REDD) projects are being designed and implemented across tropical countries, intending to curb the contribution of deforestation to greenhouse gas emissions. An important aspect of REDD implementation is the baseline against which reductions are measured. The baseline estimates the business-as-usual emissions from deforestation and forest degradation. We solve a dynamic model of land conversion from forest to agriculture in the presence of REDD, and assess the performance of four baselines. We show that none of the analysed baselines dominates in all performance aspects, and that the final baseline choice needs to maximise the trade-off between the effectiveness to reduce deforestation, cost-efficiency, and changes in income. The frequently used historical average baseline could be improved by using a forward-looking one, which is shown to better account for the opportunity costs faced by landowners. This result hinges on the ability of the baseline to predict deforestation rates without significant underestimations. We advocate the switch from a single-threshold baseline to a corridor methodology, which would provide continued incentives to reduce deforestation, even during periods of high opportunity costs. We finally show how the selection of certain baseline attributes, such as corridor bandwidth and symmetry, can enhance performance.

Keywords:

• deforestation
• REDD
• baselines
• effectiveness and efficiency

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. While REDD gives priority to reducing emissions from deforestation, REDD+ targets additionally the sustainable management of forests and the enhancement of carbon stocks.

2. These emitters could be found among the European polluting companies that are regulated by the EU ETS and need to comply with emission reduction targets.

3. We thank an anonymous referee for the input.

4. We are thankful to an anonymous reviewer for raising this important point.

5. In the paper of Busch et al. (2009, 2011), the owner will ‘opt in’ if the REDD revenues are higher than the agricultural rental price and ‘opt out’ otherwise.

6. Reducing the complexity of the harvesting function, which is not central for comparing the reference levels, allows us to focus on the dynamic choice between maintaining the forest cover and harvesting.

7. Using deterministic processes simplifies the solution to the model, but leaves outside of the scope of our analysis the role and influence of risk on the optimal land allocation decision. Under the hypothesis of a risk-neutral landowner, the presence of risk would have no specific effects. However, in the presence of risk aversion, the decision between preservation and deforestation will be significantly impacted by the relative volatility of the two prices and would favour the strategy giving the smallest cash flow variability.

8. We consider this setting to be in line with the reality of many landowners’ decision processes in tropical countries. In Latin America, ownership rights tend to be concentrated in the hands of a few proprietors (Brockett 1990; Borras Jr et al. 2012).

9. This time frame is expected to be aligned with the phases of the EU ETS or the successor of the Kyoto protocol.

10 Out of the 6 baseline methodologies reviewed by Griscom et al. (2009), 5 rely partially or totally on historical reference levels.

11. According to Angelsen (2007), ‘The FT describes a sequence where a forested region goes through four stages: (1) initially high forest cover and low deforestation, (2) accelerating and high deforestation, (3) slow-down of deforestation and forest cover stabilisation, and (4) a period of reforestation.’

12. According to Huettner et al. (2009), prospective (forward-looking) methods attempt to model land-use change taking into account the various market drivers. The forecasting can be done by using either analytical, regression or simulation models.

13. The Corridor 1 method proposes that deforestation rates within the corridor accrue credits that would only be eligible for sale once emissions go below the lower boundary of the corridor (Joanneum Research Institute 2006).

14. The analytical results can be provided by the authors upon request.

15. With about 68 million hectares of tropical forest covering nearly 53% of its territory, Peru is fourth in the global ranking, after Brazil, the Democratic Republic of Congo, and Indonesia. About 89% of the total classifies as primary forest (FAO 2010).

16. According to Diaz et al. (2011), the Peruvian and Brazilian Amazon dominate the forest carbon market, with Latin America accountable for about 60% of the 2010 total primary market volume.

17. The annual change in forest area was −0.22% for 2005--2010 (FAO 2010).

18. Hajek et al. (2011) compare 12 local REDD+ projects in south-eastern Peru, 5 of which were at feasibility and 7 at an early implementation stage at the time of writing.

19. Due to the lengthy decision horizon (100 years), we are constrained to select a low value for the discount rate; otherwise, the discounted value of incomes at later periods of time would be very close to zero, rendering irrelevant the decisions further away in the future. This approach is consistent, for instance, with the work of Gollier (2002).

20. Stern (2008) suggests the evaluation of REDD design proposals with the help of three criteria: effectiveness, efficiency, and equity and co-benefits.

21. In the Appendix, we analyse the case of decreasing deforestation paths, where the growth rate of the agricultural composite commodity is low (δ = 0) (see Figure D.9). We find that the ranking of baselines is robust across regions with different trajectories in the deforestation path.

22. For a more detailed illustration of deforestation paths for each period, see Figure B.3 in the Appendix.

23. d(t) is bounded from above by d_BaU(t), due to extraction cost constraints.

24. For the full demonstration, see the Appendix.

25. For a demonstration, see Table E.1 in the Appendix.

26. The score allocated to each baseline takes values from 1 to 4 (4 is the number of baselines considered for comparison: historical, model-implied, upward-biased fixed corridor, and upward-biased variable corridor), such that for each indicator, a score of 4 is awarded to the baseline believed to be most likely to fulfil the criterion, and a score of 1 to the baseline least likely.

27. This might not always be the case; some proposed forward-looking baselines rely on the historical deforestation average as a starting point for predicting future deforestation rates.

28. Misestimations could also occur in the computation of the historical deforestation rate, impacting the financial incentives provided by the historical and fixed-corridor baselines. In this section, we focus only on estimation errors concerning the BaU deforestation path. In our model, this has no impact on the incentives provided by the historical and fixed-corridor methodologies.

29. This refers to relying on less emission intensive sources of energy, such as renewables or the more common switch from coal to gas for the generation of electricity.

30. We thank an anonymous referee for raising this point.

31. We allow for all possible switching points in the range [0, T].

32. Data source is the OSIRIS v.3-4 spreadsheet, available online at http://sp10.conservation.org/osiris/Pages/overview.aspx.

33. Of course, not the entire territory of Peru is covered with forest and eligible for REDD projects. As mentioned in Section 2.3, Peru has about 68 million hectares of tropical forest, covering nearly 53% of its territory.

34. Various international standards have emerged to distinguish between different forest projects, such as the Panda Standard in China.